1 systemd System and Service Manager
5 * journalctl --update-catalog now understands a new --root=
6 option to operate on catalogs found in a different root
9 * During shutdown after systemd has terminated all running
10 services a final killing loop kills all remaining left-over
11 processes. We will now print the name of these processes
12 when we send SIGKILL to them, since this usually indicates a
15 * If /etc/crypttab refers to password files stored on
16 configured mount points automatic dependencies will now be
17 generated to ensure the specific mount is established first
18 before the key file is attempted to be read.
20 * 'systemctl status' will now show information about the
21 network sockets a socket unit is listening on.
23 * 'systemctl status' will also shown information about any
24 drop-in configuration file for units. (Drop-In configuration
25 files in this context are files such as
26 /etc/systemd/systemd/foobar.service.d/*.conf)
28 * systemd-cgtop now optionally shows summed up CPU times of
29 cgroups. Press '%' while running cgtop to switch between
30 percentage and absolute mode. This is useful to determine
31 which cgroups use up the most CPU time over the entire
32 runtime of the system. systemd-cgtop has also been updated
33 to be 'pipeable' for processing with further shell tools.
35 * 'hostnamectl set-hostname' will now allow setting of FQDN
38 * The formatting and parsing of time span values has been
39 changed. The parser now understands fractional expressions
40 such as "5.5h". The formatter will now output fractional
41 expressions for all time spans under 1min, i.e. "5.123456s"
42 rather than "5s 123ms 456us". For time spans under 1s
43 millisecond values are shown, for those under 1ms
44 microsecond values are shown. This should greatly improve
45 all time-related output of systemd.
47 * libsystemd-login and libsystemd-journal gained new
48 functions for querying the poll() events mask and poll()
49 timeout value for integration into arbitrary event
52 * localectl gained the ability to list available X11 keymaps
53 (models, layouts, variants, options).
55 * 'systemd-analyze dot' gained the ability to filter for
56 specific units via shell-style globs, to create smaller,
57 more useful graphs. I.e. it's now possible to create simple
58 graphs of all the dependencies between only target units, or
59 of all units that Avahi has dependencies with.
61 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
62 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
63 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
64 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
65 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
66 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
67 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
71 * The boot-time readahead implementation for rotating media
72 will now read the read-ahead data in multiple passes which
73 consist of all read requests made in equidistant time
74 intervals. This means instead of strictly reading read-ahead
75 data in its physical order on disk we now try to find a
76 middle ground between physical and access time order.
78 * /etc/os-release files gained a new BUILD_ID= field for usage
79 on operating systems that provide continuous builds of OS
82 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
83 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
84 William Douglas, Zbigniew Jędrzejewski-Szmek
88 * systemd-python gained an API exposing libsystemd-daemon.
90 * The SMACK setup logic gained support for uploading CIPSO
93 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
94 ReadOnlyDirectories= and InaccessibleDirectories= has
95 changed. The private /tmp and /var/tmp directories are now
96 shared by all processes of a service (which means
97 ExecStartPre= may now leave data in /tmp that ExecStart= of
98 the same service can still access). When a service is
99 stopped its temporary directories are immediately deleted
100 (normal clean-up with tmpfiles is still done in addition to
103 * By default, systemd will now set a couple of sysctl
104 variables in the kernel: the safe sysrq options are turned
105 on, IP route verification is turned on, and source routing
106 disabled. The recently added hardlink and softlink
107 protection of the kernel is turned on. These settings should
108 be reasonably safe, and good defaults for all new systems.
110 * The predictable network naming logic may now be turned off
111 with a new kernel command line switch: net.ifnames=0.
113 * A new libsystemd-bus module has been added that implements a
114 pretty complete D-Bus client library. For details see:
116 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
118 * journald will now explicitly flush the journal files to disk
119 at the latest 5min after each write. The file will then also
120 be marked offline until the next write. This should increase
121 reliability in case of a crash. The synchronization delay
122 can be configured via SyncIntervalSec= in journald.conf.
124 * There's a new remote-fs-setup.target unit that can be used
125 to pull in specific services when at least one remote file
126 system is to be mounted.
128 * There are new targets timers.target and paths.target as
129 canonical targets to pull user timer and path units in
130 from. This complements sockets.target with a similar
131 purpose for socket units.
133 * libudev gained a new call udev_device_set_attribute_value()
134 to set sysfs attributes of a device.
136 * The udev daemon now sets the default number of worker
137 processes executed in parallel based on the number of available
138 CPUs instead of the amount of available RAM. This is supposed
139 to provide a more reliable default and limit a too aggressive
140 paralellism for setups with 1000s of devices connected.
142 Contributions from: Auke Kok, Colin Walters, Cristian
143 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
144 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
145 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
146 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
147 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
148 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
149 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
150 Zbigniew Jędrzejewski-Szmek
154 * Configuration of unit files may now be extended via drop-in
155 files without having to edit/override the unit files
156 themselves. More specifically, if the administrator wants to
157 change one value for a service file foobar.service he can
158 now do so by dropping in a configuration snippet into
159 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
160 will load all these snippets and apply them on top of the
161 main unit configuration file, possibly extending or
162 overriding its settings. Using these drop-in snippets is
163 generally nicer than the two earlier options for changing
164 unit files locally: copying the files from
165 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
166 them there; or creating a new file in /etc/systemd/system/
167 that incorporates the original one via ".include". Drop-in
168 snippets into these .d/ directories can be placed in any
169 directory systemd looks for units in, and the usual
170 overriding semantics between /usr/lib, /etc and /run apply
173 * Most unit file settings which take lists of items can now be
174 reset by assigning the empty string to them. For example,
175 normally, settings such as Environment=FOO=BAR append a new
176 environment variable assignment to the environment block,
177 each time they are used. By assigning Environment= the empty
178 string the environment block can be reset to empty. This is
179 particularly useful with the .d/*.conf drop-in snippets
180 mentioned above, since this adds the ability to reset list
181 settings from vendor unit files via these drop-ins.
183 * systemctl gained a new "list-dependencies" command for
184 listing the dependencies of a unit recursively.
186 * Inhibitors are now honored and listed by "systemctl
187 suspend", "systemctl poweroff" (and similar) too, not only
188 GNOME. These commands will also list active sessions by
191 * Resource limits (as exposed by the various control group
192 controllers) can now be controlled dynamically at runtime
193 for all units. More specifically, you can now use a command
194 like "systemctl set-cgroup-attr foobar.service cpu.shares
195 2000" to alter the CPU shares a specific service gets. These
196 settings are stored persistently on disk, and thus allow the
197 administrator to easily adjust the resource usage of
198 services with a few simple commands. This dynamic resource
199 management logic is also available to other programs via the
200 bus. Almost any kernel cgroup attribute and controller is
203 * systemd-vconsole-setup will now copy all font settings to
204 all allocated VTs, where it previously applied them only to
207 * libsystemd-login gained the new sd_session_get_tty() API
210 * This release drops support for a few legacy or
211 distribution-specific LSB facility names when parsing init
212 scripts: $x-display-manager, $mail-transfer-agent,
213 $mail-transport-agent, $mail-transfer-agent, $smtp,
214 $null. Also, the mail-transfer-agent.target unit backing
215 this has been removed. Distributions which want to retain
216 compatibility with this should carry the burden for
217 supporting this themselves and patch support for these back
218 in, if they really need to. Also, the facilities $syslog and
219 $local_fs are now ignored, since systemd does not support
220 early-boot LSB init scripts anymore, and these facilities
221 are implied anyway for normal services. syslog.target has
224 * There are new bus calls on PID1's Manager object for
225 cancelling jobs, and removing snapshot units. Previously,
226 both calls were only available on the Job and Snapshot
229 * systemd-journal-gatewayd gained SSL support.
231 * The various "environment" files, such as /etc/locale.conf
232 now support continuation lines with a backslash ("\") as
233 last character in the line, similar in style (but different)
234 to how this is supported in shells.
236 * For normal user processes the _SYSTEMD_USER_UNIT= field is
237 now implicitly appended to every log entry logged. systemctl
238 has been updated to filter by this field when operating on a
239 user systemd instance.
241 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
242 CAP_AUDIT_CONTROL capabilities to the capabilities set for
243 the container. This makes it easier to boot unmodified
244 Fedora systems in a container, which however still requires
245 audit=0 to be passed on the kernel command line. Auditing in
246 kernel and userspace is unfortunately still too broken in
247 context of containers, hence we recommend compiling it out
248 of the kernel or using audit=0. Hopefully this will be fixed
249 one day for good in the kernel.
251 * nspawn gained the new --bind= and --bind-ro= parameters to
252 bind mount specific directories from the host into the
255 * nspawn will now mount its own devpts file system instance
256 into the container, in order not to leak pty devices from
257 the host into the container.
259 * systemd will now read the firmware boot time performance
260 information from the EFI variables, if the used boot loader
261 supports this, and takes it into account for boot performance
262 analysis via "systemd-analyze". This is currently supported
263 only in conjunction with Gummiboot, but could be supported
264 by other boot loaders too. For details see:
266 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
268 * A new generator has been added that automatically mounts the
269 EFI System Partition (ESP) to /boot, if that directory
270 exists, is empty, and no other file system has been
271 configured to be mounted there.
273 * logind will now send out PrepareForSleep(false) out
274 unconditionally, after coming back from suspend. This may be
275 used by applications as asynchronous notification for
276 system resume events.
278 * "systemctl unlock-sessions" has been added, that allows
279 unlocking the screens of all user sessions at once, similar
280 how "systemctl lock-sessions" already locked all users
281 sessions. This is backed by a new D-Bus call UnlockSessions().
283 * "loginctl seat-status" will now show the master device of a
284 seat. (i.e. the device of a seat that needs to be around for
285 the seat to be considered available, usually the graphics
288 * tmpfiles gained a new "X" line type, that allows
289 configuration of files and directories (with wildcards) that
290 shall be excluded from automatic cleanup ("aging").
292 * udev default rules set the device node permissions now only
293 at "add" events, and do not change them any longer with a
294 later "change" event.
296 * The log messages for lid events and power/sleep keypresses
297 now carry a message ID.
299 * We now have a substantially larger unit test suite, but this
300 continues to be work in progress.
302 * udevadm hwdb gained a new --root= parameter to change the
303 root directory to operate relative to.
305 * logind will now issue a background sync() request to the kernel
306 early at shutdown, so that dirty buffers are flushed to disk early
307 instead of at the last moment, in order to optimize shutdown
310 * A new bootctl tool has been added that is an interface for
311 certain boot loader operations. This is currently a preview
312 and is likely to be extended into a small mechanism daemon
313 like timedated, localed, hostnamed, and can be used by
314 graphical UIs to enumerate available boot options, and
315 request boot into firmware operations.
317 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
318 the rest of the package. It also has been updated to work
319 correctly in initrds.
321 * Policykit previously has been runtime optional, and is now
322 also compile time optional via a configure switch.
324 * systemd-analyze has been reimplemented in C. Also "systemctl
325 dot" has moved into systemd-analyze.
327 * "systemctl status" with no further parameters will now print
328 the status of all active or failed units.
330 * Operations such as "systemctl start" can now be executed
331 with a new mode "--irreversible" which may be used to queue
332 operations that cannot accidentally be reversed by a later
333 job queuing. This is by default used to make shutdown
334 requests more robust.
336 * The Python API of systemd now gained a new module for
337 reading journal files.
339 * A new tool kernel-install has been added that can install
340 kernel images according to the Boot Loader Specification:
342 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
344 * Boot time console output has been improved to provide
345 animated boot time output for hanging jobs.
347 * A new tool systemd-activate has been added which can be used
348 to test socket activation with, directly from the command
349 line. This should make it much easier to test and debug
350 socket activation in daemons.
352 * journalctl gained a new "--reverse" (or -r) option to show
353 journal output in reverse order (i.e. newest line first).
355 * journalctl gained a new "--pager-end" (or -e) option to jump
356 to immediately jump to the end of the journal in the
357 pager. This is only supported in conjunction with "less".
359 * journalctl gained a new "--user-unit=" option, that works
360 similar to "--unit=" but filters for user units rather than
363 * A number of unit files to ease adoption of systemd in
364 initrds has been added. This moves some minimal logic from
365 the various initrd implementations into systemd proper.
367 * The journal files are now owned by a new group
368 "systemd-journal", which exists specifically to allow access
369 to the journal, and nothing else. Previously, we used the
370 "adm" group for that, which however possibly covers more
371 than just journal/log file access. This new group is now
372 already used by systemd-journal-gatewayd to ensure this
373 daemon gets access to the journal files and as little else
374 as possible. Note that "make install" will also set FS ACLs
375 up for /var/log/journal to give "adm" and "wheel" read
376 access to it, in addition to "systemd-journal" which owns
377 the journal files. We recommend that packaging scripts also
378 add read access to "adm" + "wheel" to /var/log/journal, and
379 all existing/future journal files. To normal users and
380 administrators little changes, however packagers need to
381 ensure to create the "systemd-journal" system group at
382 package installation time.
384 * The systemd-journal-gatewayd now runs as unprivileged user
385 systemd-journal-gateway:systemd-journal-gateway. Packaging
386 scripts need to create these system user/group at
389 * timedated now exposes a new boolean property CanNTP that
390 indicates whether a local NTP service is available or not.
392 * systemd-detect-virt will now also detect xen PVs
394 * The pstore file system is now mounted by default, if it is
397 * In addition to the SELinux and IMA policies we will now also
398 load SMACK policies at early boot.
400 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
401 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
402 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
403 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
404 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
405 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
406 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
407 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
408 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
409 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
410 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
411 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
412 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
413 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
417 * Timer units now support calendar time events in addition to
418 monotonic time events. That means you can now trigger a unit
419 based on a calendar time specification such as "Thu,Fri
420 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
421 or fifth day of any month of the year 2013, given that it is
422 a thursday or friday. This brings timer event support
423 considerably closer to cron's capabilities. For details on
424 the supported calendar time specification language see
427 * udev now supports a number of different naming policies for
428 network interfaces for predictable names, and a combination
429 of these policies is now the default. Please see this wiki
430 document for details:
432 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
434 * Auke Kok's bootchart implementation has been added to the
435 systemd tree. It's an optional component that can graph the
436 boot in quite some detail. It's one of the best bootchart
437 implementations around and minimal in its code and
440 * nss-myhostname has been integrated into the systemd source
441 tree. nss-myhostname guarantees that the local hostname
442 always stays resolvable via NSS. It has been a weak
443 requirement of systemd-hostnamed since a long time, and
444 since its code is actually trivial we decided to just
445 include it in systemd's source tree. It can be turned off
446 with a configure switch.
448 * The read-ahead logic is now capable of properly detecting
449 whether a btrfs file system is on SSD or rotating media, in
450 order to optimize the read-ahead scheme. Previously, it was
451 only capable of detecting this on traditional file systems
454 * In udev, additional device properties are now read from the
455 IAB in addition to the OUI database. Also, Bluetooth company
456 identities are attached to the devices as well.
458 * In service files %U may be used as specifier that is
459 replaced by the configured user name of the service.
461 * nspawn may now be invoked without a controlling TTY. This
462 makes it suitable for invocation as its own service. This
463 may be used to set up a simple containerized server system
464 using only core OS tools.
466 * systemd and nspawn can now accept socket file descriptors
467 when they are started for socket activation. This enables
468 implementation of socket activated nspawn
469 containers. i.e. think about autospawning an entire OS image
470 when the first SSH or HTTP connection is received. We expect
471 that similar functionality will also be added to libvirt-lxc
474 * journalctl will now suppress ANSI color codes when
477 * systemctl will no longer show control group information for
478 a unit if a the control group is empty anyway.
480 * logind can now automatically suspend/hibernate/shutdown the
483 * /etc/machine-info and hostnamed now also expose the chassis
484 type of the system. This can be used to determine whether
485 the local system is a laptop, desktop, handset or
486 tablet. This information may either be configured by the
487 user/vendor or is automatically determined from ACPI and DMI
488 information if possible.
490 * A number of PolicyKit actions are now bound together with
491 "imply" rules. This should simplify creating UIs because
492 many actions will now authenticate similar ones as well.
494 * Unit files learnt a new condition ConditionACPower= which
495 may be used to conditionalize a unit depending on whether an
496 AC power source is connected or not, of whether the system
497 is running on battery power.
499 * systemctl gained a new "is-failed" verb that may be used in
500 shell scripts and suchlike to check whether a specific unit
501 is in the "failed" state.
503 * The EnvironmentFile= setting in unit files now supports file
504 globbing, and can hence be used to easily read a number of
505 environment files at once.
507 * systemd will no longer detect and recognize specific
508 distributions. All distribution-specific #ifdeffery has been
509 removed, systemd is now fully generic and
510 distribution-agnostic. Effectively, not too much is lost as
511 a lot of the code is still accessible via explicit configure
512 switches. However, support for some distribution specific
513 legacy configuration file formats has been dropped. We
514 recommend distributions to simply adopt the configuration
515 files everybody else uses now and convert the old
516 configuration from packaging scripts. Most distributions
517 already did that. If that's not possible or desirable,
518 distributions are welcome to forward port the specific
519 pieces of code locally from the git history.
521 * When logging a message about a unit systemd will now always
522 log the unit name in the message meta data.
524 * localectl will now also discover system locale data that is
525 not stored in locale archives, but directly unpacked.
527 * logind will no longer unconditionally use framebuffer
528 devices as seat masters, i.e. as devices that are required
529 to be existing before a seat is considered preset. Instead,
530 it will now look for all devices that are tagged as
531 "seat-master" in udev. By default framebuffer devices will
532 be marked as such, but depending on local systems other
533 devices might be marked as well. This may be used to
534 integrate graphics cards using closed source drivers (such
535 as NVidia ones) more nicely into logind. Note however, that
536 we recommend using the open source NVidia drivers instead,
537 and no udev rules for the closed-source drivers will be
538 shipped from us upstream.
540 Contributions from: Adam Williamson, Alessandro Crismani, Auke
541 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
542 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
543 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
544 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
545 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
546 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
547 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
548 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
549 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
550 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
551 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
556 * udev gained support for loading additional device properties
557 from an indexed database that is keyed by vendor/product IDs
558 and similar device identifiers. For the beginning this
559 "hwdb" is populated with data from the well-known PCI and
560 USB database, but also includes PNP, ACPI and OID data. In
561 the longer run this indexed database shall grow into
562 becoming the one central database for non-essential
563 userspace device metadata. Previously, data from the PCI/USB
564 database was only attached to select devices, since the
565 lookup was a relatively expensive operation due to O(n) time
566 complexity (with n being the number of entries in the
567 database). Since this is now O(1), we decided to add in this
568 data for all devices where this is available, by
569 default. Note that the indexed database needs to be rebuilt
570 when new data files are installed. To achieve this you need
571 to update your packaging scripts to invoke "udevadm hwdb
572 --update" after installation of hwdb data files. For
573 RPM-based distributions we introduced the new
574 %udev_hwdb_update macro for this purpose.
576 * The Journal gained support for the "Message Catalog", an
577 indexed database to link up additional information with
578 journal entries. For further details please check:
580 http://www.freedesktop.org/wiki/Software/systemd/catalog
582 The indexed message catalog database also needs to be
583 rebuilt after installation of message catalog files. Use
584 "journalctl --update-catalog" for this. For RPM-based
585 distributions we introduced the %journal_catalog_update
586 macro for this purpose.
588 * The Python Journal bindings gained support for the standard
589 Python logging framework.
591 * The Journal API gained new functions for checking whether
592 the underlying file system of a journal file is capable of
593 properly reporting file change notifications, or whether
594 applications that want to reflect journal changes "live"
595 need to recheck journal files continuously in appropriate
598 * It is now possible to set the "age" field for tmpfiles
599 entries to 0, indicating that files matching this entry
600 shall always be removed when the directories are cleaned up.
602 * coredumpctl gained a new "gdb" verb which invokes gdb
603 right-away on the selected coredump.
605 * There's now support for "hybrid sleep" on kernels that
606 support this, in addition to "suspend" and "hibernate". Use
607 "systemctl hybrid-sleep" to make use of this.
609 * logind's HandleSuspendKey= setting (and related settings)
610 now gained support for a new "lock" setting to simply
611 request the screen lock on all local sessions, instead of
612 actually executing a suspend or hibernation.
614 * systemd will now mount the EFI variables file system by
617 * Socket units now gained support for configuration of the
618 SMACK security label.
620 * timedatectl will now output the time of the last and next
621 daylight saving change.
623 * We dropped support for various legacy and distro-specific
624 concepts, such as insserv, early-boot SysV services
625 (i.e. those for non-standard runlevels such as 'b' or 'S')
626 or ArchLinux /etc/rc.conf support. We recommend the
627 distributions who still need support this to either continue
628 to maintain the necessary patches downstream, or find a
629 different solution. (Talk to us if you have questions!)
631 * Various systemd components will now bypass PolicyKit checks
632 for root and otherwise handle properly if PolicyKit is not
633 found to be around. This should fix most issues for
634 PolicyKit-less systems. Quite frankly this should have been
635 this way since day one. It is absolutely our intention to
636 make systemd work fine on PolicyKit-less systems, and we
637 consider it a bug if something doesn't work as it should if
638 PolicyKit is not around.
640 * For embedded systems it is now possible to build udev and
641 systemd without blkid and/or kmod support.
643 * "systemctl switch-root" is now capable of switching root
644 more than once. I.e. in addition to transitions from the
645 initrd to the host OS it is now possible to transition to
646 further OS images from the host. This is useful to implement
647 offline updating tools.
649 * Various other additions have been made to the RPM macros
650 shipped with systemd. Use %udev_rules_update() after
651 installing new udev rules files. %_udevhwdbdir,
652 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
653 %_sysctldir are now available which resolve to the right
654 directories for packages to place various data files in.
656 * journalctl gained the new --full switch (in addition to
657 --all, to disable ellipsation for long messages.
659 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
660 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
661 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
662 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
663 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
664 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
665 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
666 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
667 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
671 * journalctl gained new --since= and --until= switches to
672 filter by time. It also now supports nice filtering for
673 units via --unit=/-u.
675 * Type=oneshot services may use ExecReload= and do the
678 * The journal daemon now supports time-based rotation and
679 vacuuming, in addition to the usual disk-space based
682 * The journal will now index the available field values for
683 each field name. This enables clients to show pretty drop
684 downs of available match values when filtering. The bash
685 completion of journalctl has been updated
686 accordingly. journalctl gained a new switch -F to list all
687 values a certain field takes in the journal database.
689 * More service events are now written as structured messages
690 to the journal, and made recognizable via message IDs.
692 * The timedated, localed and hostnamed mini-services which
693 previously only provided support for changing time, locale
694 and hostname settings from graphical DEs such as GNOME now
695 also have a minimal (but very useful) text-based client
696 utility each. This is probably the nicest way to changing
697 these settings from the command line now, especially since
698 it lists available options and is fully integrated with bash
701 * There's now a new tool "systemd-coredumpctl" to list and
702 extract coredumps from the journal.
704 * We now install a README each in /var/log/ and
705 /etc/rc.d/init.d explaining where the system logs and init
706 scripts went. This hopefully should help folks who go to
707 that dirs and look into the otherwise now empty void and
710 * When user-services are invoked (by systemd --user) the
711 $MANAGERPID env var is set to the PID of systemd.
713 * SIGRTMIN+24 when sent to a --user instance will now result
714 in immediate termination of systemd.
716 * gatewayd received numerous feature additions such as a
717 "follow" mode, for live syncing and filtering.
719 * browse.html now allows filtering and showing detailed
720 information on specific entries. Keyboard navigation and
721 mouse screen support has been added.
723 * gatewayd/journalctl now supports HTML5/JSON
724 Server-Sent-Events as output.
726 * The SysV init script compatibility logic will now
727 heuristically determine whether a script supports the
728 "reload" verb, and only then make this available as
731 * "systemctl status --follow" has been removed, use "journalctl
734 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
735 have been removed since they are hardly useful to be
738 * And I'd like to take the opportunity to specifically mention
739 Zbigniew for his great contributions. Zbigniew, you rock!
741 Contributions from: Andrew Eikum, Christian Hesse, Colin
742 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
743 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
744 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
745 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
746 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
747 Jędrzejewski-Szmek, Сковорода Никита Андреевич
751 * If /etc/vconsole.conf is non-existent or empty we will no
752 longer load any console font or key map at boot by
753 default. Instead the kernel defaults will be left
754 intact. This is definitely the right thing to do, as no
755 configuration should mean no configuration, and hard-coding
756 font names that are different on all archs is probably a bad
757 idea. Also, the kernel default key map and font should be
758 good enough for most cases anyway, and mostly identical to
759 the userspace fonts/key maps we previously overloaded them
760 with. If distributions want to continue to default to a
761 non-kernel font or key map they should ship a default
762 /etc/vconsole.conf with the appropriate contents.
764 Contributions from: Colin Walters, Daniel J Walsh, Dave
765 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
766 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
770 * journalctl gained a new --cursor= switch to show entries
771 starting from the specified location in the journal.
773 * We now enforce a size limit on journal entry fields exported
774 with "-o json" in journalctl. Fields larger than 4K will be
775 assigned null. This can be turned off with --all.
777 * An (optional) journal gateway daemon is now available as
778 "systemd-journal-gatewayd.service". This service provides
779 access to the journal via HTTP and JSON. This functionality
780 will be used to implement live log synchronization in both
781 pull and push modes, but has various other users too, such
782 as easy log access for debugging of embedded devices. Right
783 now it is already useful to retrieve the journal via HTTP:
785 # systemctl start systemd-journal-gatewayd.service
786 # wget http://localhost:19531/entries
788 This will download the journal contents in a
789 /var/log/messages compatible format. The same as JSON:
791 # curl -H"Accept: application/json" http://localhost:19531/entries
793 This service is also accessible via a web browser where a
794 single static HTML5 app is served that uses the JSON logic
795 to enable the user to do some basic browsing of the
796 journal. This will be extended later on. Here's an example
797 screenshot of this app in its current state:
799 http://0pointer.de/public/journal-gatewayd
801 Contributions from: Kay Sievers, Lennart Poettering, Robert
802 Milasan, Tom Gundersen
806 * The bash completion logic is now available for journalctl
809 * We don't mount the "cpuset" controller anymore together with
810 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
811 started if no parameters are assigned to it. "cpuset" hence
812 broke code that assumed it it could create "cpu" groups and
815 * journalctl -f will now subscribe to terminal size changes,
816 and line break accordingly.
818 Contributions from: Dave Reisner, Kay Sievers, Lennart
819 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
823 * nspawn will now create a symlink /etc/localtime in the
824 container environment, copying the host's timezone
825 setting. Previously this has been done via a bind mount, but
826 since symlinks cannot be bind mounted this has now been
827 changed to create/update the appropriate symlink.
829 * journalctl -n's line number argument is now optional, and
830 will default to 10 if omitted.
832 * journald will now log the maximum size the journal files may
833 take up on disk. This is particularly useful if the default
834 built-in logic of determining this parameter from the file
835 system size is used. Use "systemctl status
836 systemd-journald.service" to see this information.
838 * The multi-seat X wrapper tool has been stripped down. As X
839 is now capable of enumerating graphics devices via udev in a
840 seat-aware way the wrapper is not strictly necessary
841 anymore. A stripped down temporary stop-gap is still shipped
842 until the upstream display managers have been updated to
843 fully support the new X logic. Expect this wrapper to be
844 removed entirely in one of the next releases.
846 * HandleSleepKey= in logind.conf has been split up into
847 HandleSuspendKey= and HandleHibernateKey=. The old setting
848 is not available anymore. X11 and the kernel are
849 distuingishing between these keys and we should too. This
850 also means the inhibition lock for these keys has been split
853 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
854 Poettering, Lukas Nykryn, Václav Pavlín
858 * Whenever a unit changes state we'll now log this to the
859 journal and show along the unit's own log output in
862 * ConditionPathIsMountPoint= can now properly detect bind
863 mount points too. (Previously, a bind mount of one file
864 system to another place in the same file system could not be
865 detected as mount, since they shared struct stat's st_dev
868 * We will now mount the cgroup controllers cpu, cpuacct,
869 cpuset and the controllers net_cls, net_prio together by
872 * nspawn containers will now have a virtualized boot
873 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
874 over with a randomized ID at container initialization). This
875 has the effect of making "journalctl -b" do the right thing
878 * The JSON output journal serialization has been updated not
879 to generate "endless" list objects anymore, but rather one
880 JSON object per line. This is more in line how most JSON
881 parsers expect JSON objects. The new output mode
882 "json-pretty" has been added to provide similar output, but
883 neatly aligned for readability by humans.
885 * We dropped all explicit sync() invocations in the shutdown
886 code. The kernel does this implicitly anyway in the kernel
887 reboot() syscall. halt(8)'s -n option is now a compatibility
890 * We now support virtualized reboot() in containers, as
891 supported by newer kernels. We will fall back to exit() if
892 CAP_SYS_REBOOT is not available to the container. Also,
893 nspawn makes use of this now and will actually reboot the
894 container if the containerized OS asks for that.
896 * journalctl will only show local log output by default
897 now. Use --merge (-m) to show remote log output, too.
899 * libsystemd-journal gained the new sd_journal_get_usage()
900 call to determine the current disk usage of all journal
901 files. This is exposed in the new "journalctl --disk-usage"
904 * journald gained a new configuration setting SplitMode= in
905 journald.conf which may be used to control how user journals
906 are split off. See journald.conf(5) for details.
908 * A new condition type ConditionFileNotEmpty= has been added.
910 * tmpfiles' "w" lines now support file globbing, to write
911 multiple files at once.
913 * We added Python bindings for the journal submission
914 APIs. More Python APIs for a number of selected APIs will
915 likely follow. Note that we intend to add native bindings
916 only for the Python language, as we consider it common
917 enough to deserve bindings shipped within systemd. There are
918 various projects outside of systemd that provide bindings
919 for languages such as PHP or Lua.
921 * Many conditions will now resolve specifiers such as %i. In
922 addition, PathChanged= and related directives of .path units
923 now support specifiers as well.
925 * There's now a new RPM macro definition for the system preset
928 * journald will now warn if it can't forward a message to the
929 syslog daemon because it's socket is full.
931 * timedated will no longer write or process /etc/timezone,
932 except on Debian. As we do not support late mounted /usr
933 anymore /etc/localtime always being a symlink is now safe,
934 and hence the information in /etc/timezone is not necessary
937 * logind will now always reserve one VT for a text getty (VT6
938 by default). Previously if more than 6 X sessions where
939 started they took up all the VTs with auto-spawned gettys,
940 so that no text gettys were available anymore.
942 * udev will now automatically inform the btrfs kernel logic
943 about btrfs RAID components showing up. This should make
944 simple hotplug based btrfs RAID assembly work.
946 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
947 (but not for its children which will stay at the kernel
948 default). This should allow setups with a lot more listening
951 * systemd will now always pass the configured timezone to the
952 kernel at boot. timedated will do the same when the timezone
955 * logind's inhibition logic has been updated. By default,
956 logind will now handle the lid switch, the power and sleep
957 keys all the time, even in graphical sessions. If DEs want
958 to handle these events on their own they should take the new
959 handle-power-key, handle-sleep-key and handle-lid-switch
960 inhibitors during their runtime. A simple way to achiveve
961 that is to invoke the DE wrapped in an invocation of:
963 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
965 * Access to unit operations is now checked via SELinux taking
966 the unit file label and client process label into account.
968 * systemd will now notify the administrator in the journal
969 when he over-mounts a non-empty directory.
971 * There are new specifiers that are resolved in unit files,
972 for the host name (%H), the machine ID (%m) and the boot ID
975 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
976 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
977 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
978 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
979 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
980 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
981 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
985 * Support for reading structured kernel messages from
986 /dev/kmsg has now been added and is enabled by default.
988 * Support for reading kernel messages from /proc/kmsg has now
989 been removed. If you want kernel messages in the journal
990 make sure to run a recent kernel (>= 3.5) that supports
991 reading structured messages from /dev/kmsg (see
992 above). /proc/kmsg is now exclusive property of classic
993 syslog daemons again.
995 * The libudev API gained the new
996 udev_device_new_from_device_id() call.
998 * The logic for file system namespace (ReadOnlyDirectory=,
999 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1000 require pivot_root() anymore. This means fewer temporary
1001 directories are created below /tmp for this feature.
1003 * nspawn containers will now see and receive all submounts
1004 made on the host OS below the root file system of the
1007 * Forward Secure Sealing is now supported for Journal files,
1008 which provide cryptographical sealing of journal files so
1009 that attackers cannot alter log history anymore without this
1010 being detectable. Lennart will soon post a blog story about
1011 this explaining it in more detail.
1013 * There are two new service settings RestartPreventExitStatus=
1014 and SuccessExitStatus= which allow configuration of exit
1015 status (exit code or signal) which will be excepted from the
1016 restart logic, resp. consider successful.
1018 * journalctl gained the new --verify switch that can be used
1019 to check the integrity of the structure of journal files and
1020 (if Forward Secure Sealing is enabled) the contents of
1023 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1024 and similar symlinks pre-created. This makes running shells
1025 as container init process a lot more fun.
1027 * The fstab support can now handle PARTUUID= and PARTLABEL=
1030 * A new ConditionHost= condition has been added to match
1031 against the hostname (with globs) and machine ID. This is
1032 useful for clusters where a single OS image is used to
1033 provision a large number of hosts which shall run slightly
1034 different sets of services.
1036 * Services which hit the restart limit will now be placed in a
1039 Contributions from: Bertram Poettering, Dave Reisner, Huang
1040 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1041 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1045 * When running in --user mode systemd will now become a
1046 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1047 tree a lot more organized.
1049 * A new PartOf= unit dependency type has been introduced that
1050 may be used to group services in a natural way.
1052 * "systemctl enable" may now be used to enable instances of
1055 * journalctl now prints error log levels in red, and
1056 warning/notice log levels in bright white. It also supports
1057 filtering by log level now.
1059 * cgtop gained a new -n switch (similar to top), to configure
1060 the maximum number of iterations to run for. It also gained
1061 -b, to run in batch mode (accepting no input).
1063 * The suffix ".service" may now be omitted on most systemctl
1064 command lines involving service unit names.
1066 * There's a new bus call in logind to lock all sessions, as
1067 well as a loginctl verb for it "lock-sessions".
1069 * libsystemd-logind.so gained a new call sd_journal_perror()
1070 that works similar to libc perror() but logs to the journal
1071 and encodes structured information about the error number.
1073 * /etc/crypttab entries now understand the new keyfile-size=
1076 * shutdown(8) now can send a (configurable) wall message when
1077 a shutdown is cancelled.
1079 * The mount propagation mode for the root file system will now
1080 default to "shared", which is useful to make containers work
1081 nicely out-of-the-box so that they receive new mounts from
1082 the host. This can be undone locally by running "mount
1083 --make-rprivate /" if needed.
1085 * The prefdm.service file has been removed. Distributions
1086 should maintain this unit downstream if they intend to keep
1087 it around. However, we recommend writing normal unit files
1088 for display managers instead.
1090 * Since systemd is a crucial part of the OS we will now
1091 default to a number of compiler switches that improve
1092 security (hardening) such as read-only relocations, stack
1093 protection, and suchlike.
1095 * The TimeoutSec= setting for services is now split into
1096 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1097 of individual time outs for the start and the stop phase of
1100 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1101 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1102 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1103 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1104 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1105 Gundersen, Zbigniew Jędrzejewski-Szmek
1109 * The journal and id128 C APIs are now fully documented as man
1112 * Extra safety checks have been added when transitioning from
1113 the initial RAM disk to the main system to avoid accidental
1116 * /etc/crypttab entries now understand the new keyfile-offset=
1119 * systemctl -t can now be used to filter by unit load state.
1121 * The journal C API gained the new sd_journal_wait() call to
1122 make writing synchronous journal clients easier.
1124 * journalctl gained the new -D switch to show journals from a
1127 * journalctl now displays a special marker between log
1128 messages of two different boots.
1130 * The journal is now explicitly flushed to /var via a service
1131 systemd-journal-flush.service, rather than implicitly simply
1132 by seeing /var/log/journal to be writable.
1134 * journalctl (and the journal C APIs) can now match for much
1135 more complex expressions, with alternatives and
1138 * When transitioning from the initial RAM disk to the main
1139 system we will now kill all processes in a killing spree to
1140 ensure no processes stay around by accident.
1142 * Three new specifiers may be used in unit files: %u, %h, %s
1143 resolve to the user name, user home directory resp. user
1144 shell. This is useful for running systemd user instances.
1146 * We now automatically rotate journal files if their data
1147 object hash table gets a fill level > 75%. We also size the
1148 hash table based on the configured maximum file size. This
1149 together should lower hash collisions drastically and thus
1150 speed things up a bit.
1152 * journalctl gained the new "--header" switch to introspect
1153 header data of journal files.
1155 * A new setting SystemCallFilters= has been added to services
1156 which may be used to apply blacklists or whitelists to
1157 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1159 * nspawn gained a new --link-journal= switch (and quicker: -j)
1160 to link the container journal with the host. This makes it
1161 very easy to centralize log viewing on the host for all
1162 guests while still keeping the journal files separated.
1164 * Many bugfixes and optimizations
1166 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1167 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1168 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1173 * Several tools now understand kernel command line arguments,
1174 which are only read when run in an initial RAM disk. They
1175 usually follow closely their normal counterparts, but are
1178 * There's a new tool to analyze the readahead files that are
1179 automatically generated at boot. Use:
1181 /usr/lib/systemd/systemd-readahead analyze /.readahead
1183 * We now provide an early debug shell on tty9 if this enabled. Use:
1185 systemctl enable debug-shell.service
1187 * All plymouth related units have been moved into the Plymouth
1188 package. Please make sure to upgrade your Plymouth version
1191 * systemd-tmpfiles now supports getting passed the basename of
1192 a configuration file only, in which case it will look for it
1193 in all appropriate directories automatically.
1195 * udevadm info now takes a /dev or /sys path as argument, and
1196 does the right thing. Example:
1198 udevadm info /dev/sda
1199 udevadm info /sys/class/block/sda
1201 * systemctl now prints a warning if a unit is stopped but a
1202 unit that might trigger it continues to run. Example: a
1203 service is stopped but the socket that activates it is left
1206 * "systemctl status" will now mention if the log output was
1207 shortened due to rotation since a service has been started.
1209 * The journal API now exposes functions to determine the
1210 "cutoff" times due to rotation.
1212 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1213 immediately flushing of runtime logs to /var if possible,
1214 resp. for triggering immediate rotation of the journal
1217 * It is now considered an error if a service is attempted to
1218 be stopped that is not loaded.
1220 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1222 * systemd-analyze now supports Python 3
1224 * tmpfiles now supports cleaning up directories via aging
1225 where the first level dirs are always kept around but
1226 directories beneath it automatically aged. This is enabled
1227 by prefixing the age field with '~'.
1229 * Seat objects now expose CanGraphical, CanTTY properties
1230 which is required to deal with very fast bootups where the
1231 display manager might be running before the graphics drivers
1232 completed initialization.
1234 * Seat objects now expose a State property.
1236 * We now include RPM macros for service enabling/disabling
1237 based on the preset logic. We recommend RPM based
1238 distributions to make use of these macros if possible. This
1239 makes it simpler to reuse RPM spec files across
1242 * We now make sure that the collected systemd unit name is
1243 always valid when services log to the journal via
1246 * There's a new man page kernel-command-line(7) detailing all
1247 command line options we understand.
1249 * The fstab generator may now be disabled at boot by passing
1250 fstab=0 on the kernel command line.
1252 * A new kernel command line option modules-load= is now understood
1253 to load a specific kernel module statically, early at boot.
1255 * Unit names specified on the systemctl command line are now
1256 automatically escaped as needed. Also, if file system or
1257 device paths are specified they are automatically turned
1258 into the appropriate mount or device unit names. Example:
1260 systemctl status /home
1261 systemctl status /dev/sda
1263 * The SysVConsole= configuration option has been removed from
1264 system.conf parsing.
1266 * The SysV search path is no longer exported on the D-Bus
1269 * The Names= option is been removed from unit file parsing.
1271 * There's a new man page bootup(7) detailing the boot process.
1273 * Every unit and every generator we ship with systemd now
1274 comes with full documentation. The self-explanatory boot is
1277 * A couple of services gained "systemd-" prefixes in their
1278 name if they wrap systemd code, rather than only external
1279 code. Among them fsck@.service which is now
1280 systemd-fsck@.service.
1282 * The HaveWatchdog property has been removed from the D-Bus
1285 * systemd.confirm_spawn= on the kernel command line should now
1288 * There's a new man page crypttab(5) which details all options
1289 we actually understand.
1291 * systemd-nspawn gained a new --capability= switch to pass
1292 additional capabilities to the container.
1294 * timedated will now read known NTP implementation unit names
1295 from /usr/lib/systemd/ntp-units.d/*.list,
1296 systemd-timedated-ntp.target has been removed.
1298 * journalctl gained a new switch "-b" that lists log data of
1299 the current boot only.
1301 * The notify socket is in the abstract namespace again, in
1302 order to support daemons which chroot() at start-up.
1304 * There is a new Storage= configuration option for journald
1305 which allows configuration of where log data should go. This
1306 also provides a way to disable journal logging entirely, so
1307 that data collected is only forwarded to the console, the
1308 kernel log buffer or another syslog implementation.
1310 * Many bugfixes and optimizations
1312 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1313 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1314 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1315 Shawn Landden, Tom Gundersen
1319 * "systemctl help <unit>" now shows the man page if one is
1322 * Several new man pages have been added.
1324 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1325 MaxLevelConsole= can now be specified in
1326 journald.conf. These options allow reducing the amount of
1327 data stored on disk or forwarded by the log level.
1329 * TimerSlackNSec= can now be specified in system.conf for
1330 PID1. This allows system-wide power savings.
1332 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1333 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1338 * logind is now capable of (optionally) handling power and
1339 sleep keys as well as the lid switch.
1341 * journalctl now understands the syntax "journalctl
1342 /usr/bin/avahi-daemon" to get all log output of a specific
1345 * CapabilityBoundingSet= in system.conf now also influences
1346 the capability bound set of usermode helpers of the kernel.
1348 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1349 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1350 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1351 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1355 * Note that we skipped 139 releases here in order to set the
1356 new version to something that is greater than both udev's
1357 and systemd's most recent version number.
1359 * udev: all udev sources are merged into the systemd source tree now.
1360 All future udev development will happen in the systemd tree. It
1361 is still fully supported to use the udev daemon and tools without
1362 systemd running, like in initramfs or other init systems. Building
1363 udev though, will require the *build* of the systemd tree, but
1364 udev can be properly *run* without systemd.
1366 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1367 should be used to create dead device nodes as workarounds for broken
1370 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1371 no longer supported. udev_monitor_new_from_netlink() needs to be
1372 used to subscribe to events.
1374 * udev: when udevd is started by systemd, processes which are left
1375 behind by forking them off of udev rules, are unconditionally cleaned
1376 up and killed now after the event handling has finished. Services or
1377 daemons must be started as systemd services. Services can be
1378 pulled-in by udev to get started, but they can no longer be directly
1379 forked by udev rules.
1381 * udev: the daemon binary is called systemd-udevd now and installed
1382 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1383 to adapt to that, create symlink, or rename the binary after building
1386 * libudev no longer provides these symbols:
1387 udev_monitor_from_socket()
1388 udev_queue_get_failed_list_entry()
1389 udev_get_{dev,sys,run}_path()
1390 The versions number was bumped and symbol versioning introduced.
1392 * systemd-loginctl and systemd-journalctl have been renamed
1393 to loginctl and journalctl to match systemctl.
1395 * The config files: /etc/systemd/systemd-logind.conf and
1396 /etc/systemd/systemd-journald.conf have been renamed to
1397 logind.conf and journald.conf. Package updates should rename
1398 the files to the new names on upgrade.
1400 * For almost all files the license is now LGPL2.1+, changed
1401 from the previous GPL2.0+. Exceptions are some minor stuff
1402 of udev (which will be changed to LGPL2.1 eventually, too),
1403 and the MIT licensed sd-daemon.[ch] library that is suitable
1404 to be used as drop-in files.
1406 * systemd and logind now handle system sleep states, in
1407 particular suspending and hibernating.
1409 * logind now implements a sleep/shutdown/idle inhibiting logic
1410 suitable for a variety of uses. Soonishly Lennart will blog
1411 about this in more detail.
1413 * var-run.mount and var-lock.mount are no longer provided
1414 (which prevously bind mounted these directories to their new
1415 places). Distributions which have not converted these
1416 directories to symlinks should consider stealing these files
1417 from git history and add them downstream.
1419 * We introduced the Documentation= field for units and added
1420 this to all our shipped units. This is useful to make it
1421 easier to explore the boot and the purpose of the various
1424 * All smaller setup units (such as
1425 systemd-vconsole-setup.service) now detect properly if they
1426 are run in a container and are skipped when
1427 appropriate. This guarantees an entirely noise-free boot in
1428 Linux container environments such as systemd-nspawn.
1430 * A framework for implementing offline system updates is now
1431 integrated, for details see:
1432 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1434 * A new service type Type=idle is available now which helps us
1435 avoiding ugly interleaving of getty output and boot status
1438 * There's now a system-wide CapabilityBoundingSet= option to
1439 globally reduce the set of capabilities for the
1440 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1441 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1442 even CAP_NET_ADMIN system-wide for secure systems.
1444 * There are now system-wide DefaultLimitXXX= options to
1445 globally change the defaults of the various resource limits
1446 for all units started by PID 1.
1448 * Harald Hoyer's systemd test suite has been integrated into
1449 systemd which allows easy testing of systemd builds in qemu
1450 and nspawn. (This is really awesome! Ask us for details!)
1452 * The fstab parser is now implemented as generator, not inside
1455 * systemctl will now warn you if .mount units generated from
1456 /etc/fstab are out of date due to changes in fstab that
1457 haven't been read by systemd yet.
1459 * systemd is now suitable for usage in initrds. Dracut has
1460 already been updated to make use of this. With this in place
1461 initrds get a slight bit faster but primarily are much
1462 easier to introspect and debug since "systemctl status" in
1463 the host system can be used to introspect initrd services,
1464 and the journal from the initrd is kept around too.
1466 * systemd-delta has been added, a tool to explore differences
1467 between user/admin configuration and vendor defaults.
1469 * PrivateTmp= now affects both /tmp and /var/tmp.
1471 * Boot time status messages are now much prettier and feature
1472 proper english language. Booting up systemd has never been
1475 * Read-ahead pack files now include the inode number of all
1476 files to pre-cache. When the inode changes the pre-caching
1477 is not attempted. This should be nicer to deal with updated
1478 packages which might result in changes of read-ahead
1481 * We now temporaritly lower the kernel's read_ahead_kb variable
1482 when collecting read-ahead data to ensure the kernel's
1483 built-in read-ahead does not add noise to our measurements
1484 of necessary blocks to pre-cache.
1486 * There's now RequiresMountsFor= to add automatic dependencies
1487 for all mounts necessary for a specific file system path.
1489 * MountAuto= and SwapAuto= have been removed from
1490 system.conf. Mounting file systems at boot has to take place
1493 * nspawn now learned a new switch --uuid= to set the machine
1494 ID on the command line.
1496 * nspawn now learned the -b switch to automatically search
1499 * vt102 is now the default TERM for serial TTYs, upgraded from
1502 * systemd-logind now works on VT-less systems.
1504 * The build tree has been reorganized. The individual
1505 components now have directories of their own.
1507 * A new condition type ConditionPathIsReadWrite= is now available.
1509 * nspawn learned the new -C switch to create cgroups for the
1510 container in other hierarchies.
1512 * We now have support for hardware watchdogs, configurable in
1515 * The scheduled shutdown logic now has a public API.
1517 * We now mount /tmp as tmpfs by default, but this can be
1518 masked and /etc/fstab can override it.
1520 * Since udisks doesn't make use of /media anymore we are not
1521 mounting a tmpfs on it anymore.
1523 * journalctl gained a new --local switch to only interleave
1524 locally generated journal files.
1526 * We can now load the IMA policy at boot automatically.
1528 * The GTK tools have been split off into a systemd-ui.
1530 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1531 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1532 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1533 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1534 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1535 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1536 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1537 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1538 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1543 * This is mostly a bugfix release
1545 * Support optional initialization of the machine ID from the
1546 KVM or container configured UUID.
1548 * Support immediate reboots with "systemctl reboot -ff"
1550 * Show /etc/os-release data in systemd-analyze output
1552 * Many bugfixes for the journal, including endianness fixes and
1553 ensuring that disk space enforcement works
1555 * sd-login.h is C++ comptaible again
1557 * Extend the /etc/os-release format on request of the Debian
1560 * We now refuse non-UTF8 strings used in various configuration
1561 and unit files. This is done to ensure we don't pass invalid
1562 data over D-Bus or expose it elsewhere.
1564 * Register Mimo USB Screens as suitable for automatic seat
1567 * Read SELinux client context from journal clients in a race
1570 * Reorder configuration file lookup order. /etc now always
1571 overrides /run in order to allow the administrator to always
1572 and unconditionally override vendor supplied or
1573 automatically generated data.
1575 * The various user visible bits of the journal now have man
1576 pages. We still lack man pages for the journal API calls
1579 * We now ship all man pages in HTML format again in the
1582 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1583 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1584 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1589 * This is mostly a bugfix release
1591 * systems lacking /etc/os-release are no longer supported.
1593 * Various functionality updates to libsystemd-login.so
1595 * Track class of PAM logins to distuingish greeters from
1598 Contributions from: Kay Sievers, Lennart Poettering, Michael
1603 * This is an important bugfix release for v41.
1605 * Building man pages is now optional which should be useful
1606 for those building systemd from git but unwilling to install
1609 * Watchdog support for supervising services is now usable. In
1610 a future release support for hardware watchdogs
1611 (i.e. /dev/watchdog) will be added building on this.
1613 * Service start rate limiting is now configurable and can be
1614 turned off per service. When a start rate limit is hit a
1615 reboot can automatically be triggered.
1617 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1619 Contributions from: Benjamin Franzke, Bill Nottingham,
1620 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1621 Schmidt, Michał Górny, Piotr Drąg
1625 * The systemd binary is installed /usr/lib/systemd/systemd now;
1626 An existing /sbin/init symlink needs to be adapted with the
1629 * The code that loads kernel modules has been ported to invoke
1630 libkmod directly, instead of modprobe. This means we do not
1631 support systems with module-init-tools anymore.
1633 * Watchdog support is now already useful, but still not
1636 * A new kernel command line option systemd.setenv= is
1637 understood to set system wide environment variables
1638 dynamically at boot.
1640 * We now limit the set of capabilities of systemd-journald.
1642 * We now set SIGPIPE to ignore by default, since it only is
1643 useful in shell pipelines, and has little use in general
1644 code. This can be disabled with IgnoreSIPIPE=no in unit
1647 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1648 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1653 * This is mostly a bugfix release
1655 * We now expose the reason why a service failed in the
1656 "Result" D-Bus property.
1658 * Rudimentary service watchdog support (will be completed over
1659 the next few releases.)
1661 * When systemd forks off in order execute some service we will
1662 now immediately changes its argv[0] to reflect which process
1663 it will execute. This is useful to minimize the time window
1664 with a generic argv[0], which makes bootcharts more useful
1666 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1667 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1668 Mike Kazantsev, Ray Strode
1672 * This is mostly a test release, but incorporates many
1675 * New systemd-cgtop tool to show control groups by their
1678 * Linking against libacl for ACLs is optional again. If
1679 disabled, support tracking device access for active logins
1680 goes becomes unavailable, and so does access to the user
1681 journals by the respective users.
1683 * If a group "adm" exists, journal files are automatically
1684 owned by them, thus allow members of this group full access
1685 to the system journal as well as all user journals.
1687 * The journal now stores the SELinux context of the logging
1688 client for all entries.
1690 * Add C++ inclusion guards to all public headers
1692 * New output mode "cat" in the journal to print only text
1693 messages, without any meta data like date or time.
1695 * Include tiny X server wrapper as a temporary stop-gap to
1696 teach XOrg udev display enumeration. This is used by display
1697 managers such as gdm, and will go away as soon as XOrg
1698 learned native udev hotplugging for display devices.
1700 * Add new systemd-cat tool for executing arbitrary programs
1701 with STDERR/STDOUT connected to the journal. Can also act as
1702 BSD logger replacement, and does so by default.
1704 * Optionally store all locally generated coredumps in the
1705 journal along with meta data.
1707 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
1708 writing short strings to files (for usage for /sys), and for
1709 creating symlinks, character and block device nodes.
1711 * New unit file option ControlGroupPersistent= to make cgroups
1712 persistent, following the mechanisms outlined in
1713 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
1715 * Support multiple local RTCs in a sane way
1717 * No longer monopolize IO when replaying readahead data on
1718 rotating disks, since we might starve non-file-system IO to
1719 death, since fanotify() will not see accesses done by blkid,
1722 * Don't show kernel threads in systemd-cgls anymore, unless
1723 requested with new -k switch.
1725 Contributions from: Dan Horák, Kay Sievers, Lennart
1726 Poettering, Michal Schmidt
1730 * This is mostly a test release, but incorporates many
1733 * The git repository moved to:
1734 git://anongit.freedesktop.org/systemd/systemd
1735 ssh://git.freedesktop.org/git/systemd/systemd
1737 * First release with the journal
1738 http://0pointer.de/blog/projects/the-journal.html
1740 * The journal replaces both systemd-kmsg-syslogd and
1741 systemd-stdout-bridge.
1743 * New sd_pid_get_unit() API call in libsystemd-logind
1745 * Many systemadm clean-ups
1747 * Introduce remote-fs-pre.target which is ordered before all
1748 remote mounts and may be used to start services before all
1751 * Added Mageia support
1753 * Add bash completion for systemd-loginctl
1755 * Actively monitor PID file creation for daemons which exit in
1756 the parent process before having finished writing the PID
1757 file in the daemon process. Daemons which do this need to be
1758 fixed (i.e. PID file creation must have finished before the
1759 parent exits), but we now react a bit more gracefully to them.
1761 * Add colourful boot output, mimicking the well-known output
1762 of existing distributions.
1764 * New option PassCredentials= for socket units, for
1765 compatibility with a recent kernel ABI breakage.
1767 * /etc/rc.local is now hooked in via a generator binary, and
1768 thus will no longer act as synchronization point during
1771 * systemctl list-unit-files now supports --root=.
1773 * systemd-tmpfiles now understands two new commands: z, Z for
1774 relabelling files according to the SELinux database. This is
1775 useful to apply SELinux labels to specific files in /sys,
1778 * Output of SysV services is now forwarded to both the console
1779 and the journal by default, not only just the console.
1781 * New man pages for all APIs from libsystemd-login.
1783 * The build tree got reorganized and a the build system is a
1784 lot more modular allowing embedded setups to specifically
1785 select the components of systemd they are interested in.
1787 * Support for Linux systems lacking the kernel VT subsystem is
1790 * configure's --with-rootdir= got renamed to
1791 --with-rootprefix= to follow the naming used by udev and
1794 * Unless specified otherwise we'll now install to /usr instead
1795 of /usr/local by default.
1797 * Processes with '@' in argv[0][0] are now excluded from the
1798 final shut-down killing spree, following the logic explained
1800 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
1802 * All processes remaining in a service cgroup when we enter
1803 the START or START_PRE states are now killed with
1804 SIGKILL. That means it is no longer possible to spawn
1805 background processes from ExecStart= lines (which was never
1806 supported anyway, and bad style).
1808 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
1809 reloading of units together.
1811 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
1812 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
1813 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
1814 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
1815 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob