1 systemd System and Service Manager
5 * Whenever a unit changes state we'll now log this to the
6 journal and show along the unit's own log output in
9 * ConditionPathIsMountPoint= can now properly detect bind
10 mount points too. (Previously, a bind mount of one file
11 system to another place in the same file system could not be
12 detected as mount, since they shared struct stat's st_dev
15 * We will now mount the cgroup controllers cpu, cpuacct,
16 cpuset and the controllers net_cls, net_prio together by
19 * nspawn containers will now have a virtualized boot
20 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
21 over with a randomized ID at container initialization). This
22 has the effect of making "journalctl -b" do the right thing
25 * The JSON output journal serialization has been updated not
26 to generate "endless" list objects anymore, but rather one
27 JSON object per line. This is more in line how most JSON
28 parsers expect JSON objects. The new output mode
29 "json-pretty" has been added to provide similar output, but
30 neatly aligned for readability by humans.
32 * We dropped all explicit sync() invocations in the shutdown
33 code. The kernel does this implicitly anyway in the kernel
34 reboot() syscall. halt(8)'s -n option is now a compatibility
37 * We now support virtualized reboot() in containers, as
38 supported by newer kernels. We will fall back to exit() if
39 CAP_SYS_REBOOT is not available to the container. Also,
40 nspawn makes use of this now and will actually reboot the
41 container if the containerized OS asks for that.
43 * journalctl will only show local log output by default
44 now. Use --merge (-m) to show remote log output, too.
46 * libsystemd-journal gained the new sd_journal_get_usage()
47 call to determine the current disk usage of all journal
48 files. This is exposed in the new "journalctl --disk-usage"
51 * journald gained a new configuration setting SplitMode= in
52 journald.conf which may be used to control how user journals
53 are split off. See journald.conf(5) for details.
55 * A new condition type ConditionFileNotEmpty= has been added.
57 * tmpfiles' "w" lines now support file globbing, to write
58 multiple files at once.
60 * We added Python bindings for the journal submission
61 APIs. More Python APIs for a number of selected APIs will
62 likely follow. Note that we intend to add native bindings
63 only for the Python language, as we consider it common
64 enough to deserve bindings shipped within systemd. There are
65 various projects outside of systemd that provide bindings
66 for languages such as PHP or Lua.
68 * Many conditions will now resolve specifiers such as %i. In
69 addition, PathChanged= and related directives of .path units
70 now support specifiers as well.
72 * There's now a new RPM macro definition for the system preset
75 * journald will now warn if it can't foward a message to the
76 syslog daemon because it's socket is full.
78 * timedated will no longer write or process /etc/timezone,
79 except on Debian. As we do not support late mounted /usr
80 anymore /etc/localtime always being a symlink is now safe,
81 and hence the information in /etc/timezone is not necessary
84 * logind will now always reserve one VT for a text getty (VT6
85 by default). Previously if more than 6 X sessions where
86 started they took up all the VTs with auto-spawned gettys,
87 so that no text gettys were available anymore.
89 * udev will now automatically inform the btrfs kernel logic
90 about btrfs RAID components showing up. This should make
91 simple hotplug based btrfs RAID assembly work.
93 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
94 (but not for its children which will stay at the kernel
95 default). This should allow setups with a lot more listening
98 * systemd will now always pass the configured timezone to the
99 kernel at boot. timedated will do the same when the timezone
102 * logind's inhibition logic has been updated. By default,
103 logind will now handle the lid switch, the power and sleep
104 keys all the time, even in graphical sessions. If DEs want
105 to handle these events on their own they should take the new
106 handle-power-key, handle-sleep-key and handle-lid-switch
107 inhibitors during their runtime. A simple way to achiveve
108 that is to invoke the DE wrapped in an invocation of:
110 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
112 * Access to unit operations is now checked via SELinux taking
113 the unit file label and client process label into account.
115 * systemd will now notify the administrator in the journal
116 when he over-mounts a non-empty directory.
118 * There are new specifiers that are resolved in unit files,
119 for the host name (%H), the machine ID (%m) and the boot ID
122 Contributions from Allin Cottrell, Auke Kok, Brandon Philips,
123 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
124 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
125 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
126 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
127 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
128 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
132 * Support for reading structured kernel messages from
133 /dev/kmsg has now been added and is enabled by default.
135 * Support for reading kernel messages from /proc/kmsg has now
136 been removed. If you want kernel messages in the journal
137 make sure to run a recent kernel (>= 3.5) that supports
138 reading structured messages from /dev/kmsg (see
139 above). /proc/kmsg is now exclusive property of classic
140 syslog daemons again.
142 * The libudev API gained the new
143 udev_device_new_from_device_id() call.
145 * The logic for file system namespace (ReadOnlyDirectory=,
146 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
147 require pivot_root() anymore. This means fewer temporary
148 directories are created below /tmp for this feature.
150 * nspawn containers will now see and receive all submounts
151 made on the host OS below the root file system of the
154 * Forward Secure Sealing is now supported for Journal files,
155 which provide cryptographical sealing of journal files so
156 that attackers cannot alter log history anymore without this
157 being detectable. Lennart will soon post a blog story about
158 this explaining it in more detail.
160 * There are two new service settings RestartPreventExitStatus=
161 and SuccessExitStatus= which allow configuration of exit
162 status (exit code or signal) which will be excepted from the
163 restart logic, resp. consider successful.
165 * journalctl gained the new --verify switch that can be used
166 to check the integrity of the structure of journal files and
167 (if Forward Secure Sealing is enabled) the contents of
170 * nspawn containers will now be run with /dev/stdin, /dev/fd/
171 and similar symlinks pre-created. This makes running shells
172 as container init process a lot more fun.
174 * The fstab support can now handle PARTUUID= and PARTLABEL=
177 * A new ConditionHost= condition has been added to match
178 against the hostname (with globs) and machine ID. This is
179 useful for clusters where a single OS image is used to
180 provision a large number of hosts which shall run slightly
181 different sets of services.
183 * Services which hit the restart limit will now be placed in a
186 Contributions from Bertram Poettering, Dave Reisner, Huang
187 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
188 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
192 * When running in --user mode systemd will now become a
193 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
194 tree a lot more organized.
196 * A new PartOf= unit dependency type has been introduced that
197 may be used to group services in a natural way.
199 * "systemctl enable" may now be used to enable instances of
202 * journalctl now prints error log levels in red, and
203 warning/notice log levels in bright white. It also supports
204 filtering by log level now.
206 * cgtop gained a new -n switch (similar to top), to configure
207 the maximum number of iterations to run for. It also gained
208 -b, to run in batch mode (accepting no input).
210 * The suffix ".service" may now be ommited on most systemctl
211 command lines involving service unit names.
213 * There's a new bus call in logind to lock all sessions, as
214 well as a loginctl verb for it "lock-sessions".
216 * libsystemd-logind.so gained a new call sd_journal_perror()
217 that works similar to libc perror() but logs to the journal
218 and encodes structured information about the error number.
220 * /etc/crypttab entries now understand the new keyfile-size=
223 * shutdown(8) now can send a (configurable) wall message when
224 a shutdown is cancelled.
226 * The mount propagation mode for the root file system will now
227 default to "shared", which is useful to make containers work
228 nicely out-of-the-box so that they receive new mounts from
229 the host. This can be undone locally by running "mount
230 --make-rprivate /" if needed.
232 * The prefdm.service file has been removed. Distributions
233 should maintain this unit downstream if they intend to keep
234 it around. However, we recommend writing normal unit files
235 for display managers instead.
237 * Since systemd is a crucial part of the OS we will now
238 default to a number of compiler switches that improve
239 security (hardening) such as read-only relocations, stack
240 protection, and suchlike.
242 * The TimeoutSec= setting for services is now split into
243 TimeoutStartSec= and TimeoutStopSec= to allow configuration
244 of individual time outs for the start and the stop phase of
247 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
248 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
249 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
250 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
251 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
252 Gundersen, Zbigniew Jędrzejewski-Szmek
256 * The journal and id128 C APIs are now fully documented as man
259 * Extra safety checks have been added when transitioning from
260 the initial RAM disk to the main system to avoid accidental
263 * /etc/crypttab entries now understand the new keyfile-offset=
266 * systemctl -t can now be used to filter by unit load state.
268 * The journal C API gained the new sd_journal_wait() call to
269 make writing synchronous journal clients easier.
271 * journalctl gained the new -D switch to show journals from a
274 * journalctl now displays a special marker between log
275 messages of two different boots.
277 * The journal is now explicitly flushed to /var via a service
278 systemd-journal-flush.service, rather than implicitly simply
279 by seeing /var/log/journal to be writable.
281 * journalctl (and the journal C APIs) can now match for much
282 more complex expressions, with alternatives and
285 * When transitioning from the initial RAM disk to the main
286 system we will now kill all processes in a killing spree to
287 ensure no processes stay around by accident.
289 * Three new specifiers may be used in unit files: %u, %h, %s
290 resolve to the user name, user home directory resp. user
291 shell. This is useful for running systemd user instances.
293 * We now automatically rotate journal files if their data
294 object hash table gets a fill level > 75%. We also size the
295 hash table based on the configured maximum file size. This
296 together should lower hash collisions drastically and thus
297 speed things up a bit.
299 * journalctl gained the new "--header" switch to introspect
300 header data of journal files.
302 * A new setting SystemCallFilters= has been added to services
303 which may be used to apply blacklists or whitelists to
304 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
306 * nspawn gained a new --link-journal= switch (and quicker: -j)
307 to link the container journal with the host. This makes it
308 very easy to centralize log viewing on the host for all
309 guests while still keeping the journal files separated.
311 * Many bugfixes and optimizations
313 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
314 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
315 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
320 * Several tools now understand kernel command line arguments,
321 which are only read when run in an initial RAM disk. They
322 usually follow closely their normal counterparts, but are
325 * There's a new tool to analyze the readahead files that are
326 automatically generated at boot. Use:
328 /usr/lib/systemd/systemd-readahead analyze /.readahead
330 * We now provide an early debug shell on tty9 if this enabled. Use:
332 systemctl enable debug-shell.service
334 * All plymouth related units have been moved into the Plymouth
335 package. Please make sure to upgrade your Plymouth version
338 * systemd-tmpfiles now supports getting passed the basename of
339 a configuration file only, in which case it will look for it
340 in all appropriate directories automatically.
342 * udevadm info now takes a /dev or /sys path as argument, and
343 does the right thing. Example:
345 udevadm info /dev/sda
346 udevadm info /sys/class/block/sda
348 * systemctl now prints a warning if a unit is stopped but a
349 unit that might trigger it continues to run. Example: a
350 service is stopped but the socket that activates it is left
353 * "systemctl status" will now mention if the log output was
354 shortened due to rotation since a service has been started.
356 * The journal API now exposes functions to determine the
357 "cutoff" times due to rotation.
359 * journald now understands SIGUSR1 and SIGUSR2 for triggering
360 immediately flushing of runtime logs to /var if possible,
361 resp. for triggering immediate rotation of the journal
364 * It is now considered an error if a service is attempted to
365 be stopped that is not loaded.
367 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
369 * systemd-analyze now supports Python 3
371 * tmpfiles now supports cleaning up directories via aging
372 where the first level dirs are always kept around but
373 directories beneath it automatically aged. This is enabled
374 by prefixing the age field with '~'.
376 * Seat objects now expose CanGraphical, CanTTY properties
377 which is required to deal with very fast bootups where the
378 display manager might be running before the graphics drivers
379 completed initialization.
381 * Seat objects now expose a State property.
383 * We now include RPM macros for service enabling/disabling
384 based on the preset logic. We recommend RPM based
385 distributions to make use of these macros if possible. This
386 makes it simpler to reuse RPM spec files across
389 * We now make sure that the collected systemd unit name is
390 always valid when services log to the journal via
393 * There's a new man page kernel-command-line(7) detailing all
394 command line options we understand.
396 * The fstab generator may now be disabled at boot by passing
397 fstab=0 on the kernel command line.
399 * A new kernel command line option modules-load= is now understood
400 to load a specific kernel module statically, early at boot.
402 * Unit names specified on the systemctl command line are now
403 automatically escaped as needed. Also, if file system or
404 device paths are specified they are automatically turned
405 into the appropriate mount or device unit names. Example:
407 systemctl status /home
408 systemctl status /dev/sda
410 * The SysVConsole= configuration option has been removed from
413 * The SysV search path is no longer exported on the D-Bus
416 * The Names= option is been removed from unit file parsing.
418 * There's a new man page bootup(7) detailing the boot process.
420 * Every unit and every generator we ship with systemd now
421 comes with full documentation. The self-explanatory boot is
424 * A couple of services gained "systemd-" prefixes in their
425 name if they wrap systemd code, rather than only external
426 code. Among them fsck@.service which is now
427 systemd-fsck@.service.
429 * The HaveWatchdog property has been removed from the D-Bus
432 * systemd.confirm_spawn= on the kernel command line should now
435 * There's a new man page crypttab(5) which details all options
436 we actually understand.
438 * systemd-nspawn gained a new --capability= switch to pass
439 additional capabilities to the container.
441 * timedated will now read known NTP implementation unit names
442 from /usr/lib/systemd/ntp-units.d/*.list,
443 systemd-timedated-ntp.target has been removed.
445 * journalctl gained a new switch "-b" that lists log data of
446 the current boot only.
448 * The notify socket is in the abstract namespace again, in
449 order to support daemons which chroot() at start-up.
451 * There is a new Storage= configuration option for journald
452 which allows configuration of where log data should go. This
453 also provides a way to disable journal logging entirely, so
454 that data collected is only forwarded to the console, the
455 kernel log buffer or another syslog implementation.
457 * Many bugfixes and optimizations
459 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
460 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
461 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
462 Shawn Landden, Tom Gundersen
465 * "systemctl help <unit>" now shows the man page if one is
468 * Several new man pages have been added.
470 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
471 MaxLevelConsole= can now be specified in
472 journald.conf. These options allow reducing the amount of
473 data stored on disk or forwarded by the log level.
475 * TimerSlackNSec= can now be specified in system.conf for
476 PID1. This allows system-wide power savings.
478 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
479 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
483 * logind is now capable of (optionally) handling power and
484 sleep keys as well as the lid switch.
486 * journalctl now understands the syntax "journalctl
487 /usr/bin/avahi-daemon" to get all log output of a specific
490 * CapabilityBoundingSet= in system.conf now also influences
491 the capability bound set of usermode helpers of the kernel.
493 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
494 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
495 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
496 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
499 * Note that we skipped 139 releases here in order to set the
500 new version to something that is greater than both udev's
501 and systemd's most recent version number.
503 * udev: all udev sources are merged into the systemd source tree now.
504 All future udev development will happen in the systemd tree. It
505 is still fully supported to use the udev daemon and tools without
506 systemd running, like in initramfs or other init systems. Building
507 udev though, will require the *build* of the systemd tree, but
508 udev can be properly *run* without systemd.
510 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
511 should be used to create dead device nodes as workarounds for broken
514 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
515 no longer supported. udev_monitor_new_from_netlink() needs to be
516 used to subscribe to events.
518 * udev: when udevd is started by systemd, processes which are left
519 behind by forking them off of udev rules, are unconditionally cleaned
520 up and killed now after the event handling has finished. Services or
521 daemons must be started as systemd services. Services can be
522 pulled-in by udev to get started, but they can no longer be directly
523 forked by udev rules.
525 * udev: the daemon binary is called systemd-udevd now and installed
526 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
527 to adapt to that, create symlink, or rename the binary after building
530 * libudev no longer provides these symbols:
531 udev_monitor_from_socket()
532 udev_queue_get_failed_list_entry()
533 udev_get_{dev,sys,run}_path()
534 The versions number was bumped and symbol versioning introduced.
536 * systemd-loginctl and systemd-journalctl have been renamed
537 to loginctl and journalctl to match systemctl.
539 * The config files: /etc/systemd/systemd-logind.conf and
540 /etc/systemd/systemd-journald.conf have been renamed to
541 logind.conf and journald.conf. Package updates should rename
542 the files to the new names on upgrade.
544 * For almost all files the license is now LGPL2.1+, changed
545 from the previous GPL2.0+. Exceptions are some minor stuff
546 of udev (which will be changed to LGPL2.1 eventually, too),
547 and the MIT licensed sd-daemon.[ch] library that is suitable
548 to be used as drop-in files.
550 * systemd and logind now handle system sleep states, in
551 particular suspending and hibernating.
553 * logind now implements a sleep/shutdown/idle inhibiting logic
554 suitable for a variety of uses. Soonishly Lennart will blog
555 about this in more detail.
557 * var-run.mount and var-lock.mount are no longer provided
558 (which prevously bind mounted these directories to their new
559 places). Distributions which have not converted these
560 directories to symlinks should consider stealing these files
561 from git history and add them downstream.
563 * We introduced the Documentation= field for units and added
564 this to all our shipped units. This is useful to make it
565 easier to explore the boot and the purpose of the various
568 * All smaller setup units (such as
569 systemd-vconsole-setup.service) now detect properly if they
570 are run in a container and are skipped when
571 appropriate. This guarantees an entirely noise-free boot in
572 Linux container environments such as systemd-nspawn.
574 * A framework for implementing offline system updates is now
575 integrated, for details see:
576 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
578 * A new service type Type=idle is available now which helps us
579 avoiding ugly interleaving of getty output and boot status
582 * There's now a system-wide CapabilityBoundingSet= option to
583 globally reduce the set of capabilities for the
584 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
585 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
586 even CAP_NET_ADMIN system-wide for secure systems.
588 * There are now system-wide DefaultLimitXXX= options to
589 globally change the defaults of the various resource limits
590 for all units started by PID 1.
592 * Harald Hoyer's systemd test suite has been integrated into
593 systemd which allows easy testing of systemd builds in qemu
594 and nspawn. (This is really awesome! Ask us for details!)
596 * The fstab parser is now implemented as generator, not inside
599 * systemctl will now warn you if .mount units generated from
600 /etc/fstab are out of date due to changes in fstab that
601 haven't been read by systemd yet.
603 * systemd is now suitable for usage in initrds. Dracut has
604 already been updated to make use of this. With this in place
605 initrds get a slight bit faster but primarily are much
606 easier to introspect and debug since "systemctl status" in
607 the host system can be used to introspect initrd services,
608 and the journal from the initrd is kept around too.
610 * systemd-delta has been added, a tool to explore differences
611 between user/admin configuration and vendor defaults.
613 * PrivateTmp= now affects both /tmp and /var/tmp.
615 * Boot time status messages are now much prettier and feature
616 proper english language. Booting up systemd has never been
619 * Read-ahead pack files now include the inode number of all
620 files to pre-cache. When the inode changes the pre-caching
621 is not attempted. This should be nicer to deal with updated
622 packages which might result in changes of read-ahead
625 * We now temporaritly lower the kernel's read_ahead_kb variable
626 when collecting read-ahead data to ensure the kernel's
627 built-in read-ahead does not add noise to our measurements
628 of necessary blocks to pre-cache.
630 * There's now RequiresMountsFor= to add automatic dependencies
631 for all mounts necessary for a specific file system path.
633 * MountAuto= and SwapAuto= have been removed from
634 system.conf. Mounting file systems at boot has to take place
637 * nspawn now learned a new switch --uuid= to set the machine
638 ID on the command line.
640 * nspawn now learned the -b switch to automatically search
643 * vt102 is now the default TERM for serial TTYs, upgraded from
646 * systemd-logind now works on VT-less systems.
648 * The build tree has been reorganized. The individual
649 components now have directories of their own.
651 * A new condition type ConditionPathIsReadWrite= is now available.
653 * nspawn learned the new -C switch to create cgroups for the
654 container in other hierarchies.
656 * We now have support for hardware watchdogs, configurable in
659 * The scheduled shutdown logic now has a public API.
661 * We now mount /tmp as tmpfs by default, but this can be
662 masked and /etc/fstab can override it.
664 * Since udisks doesn't make use of /media anymore we are not
665 mounting a tmpfs on it anymore.
667 * journalctl gained a new --local switch to only interleave
668 locally generated journal files.
670 * We can now load the IMA policy at boot automatically.
672 * The GTK tools have been split off into a systemd-ui.
674 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
675 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
676 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
677 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
678 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
679 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
680 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
681 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
682 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
686 * This is mostly a bugfix release
688 * Support optional initialization of the machine ID from the
689 KVM or container configured UUID.
691 * Support immediate reboots with "systemctl reboot -ff"
693 * Show /etc/os-release data in systemd-analyze output
695 * Many bugfixes for the journal, including endianess fixes and
696 ensuring that disk space enforcement works
698 * sd-login.h is C++ comptaible again
700 * Extend the /etc/os-release format on request of the Debian
703 * We now refuse non-UTF8 strings used in various configuration
704 and unit files. This is done to ensure we don't pass invalid
705 data over D-Bus or expose it elsewhere.
707 * Register Mimo USB Screens as suitable for automatic seat
710 * Read SELinux client context from journal clients in a race
713 * Reorder configuration file lookup order. /etc now always
714 overrides /run in order to allow the administrator to always
715 and unconditionally override vendor supplied or
716 automatically generated data.
718 * The various user visible bits of the journal now have man
719 pages. We still lack man pages for the journal API calls
722 * We now ship all man pages in HTML format again in the
725 Contributions from: Dave Reisner, Dirk Eibach, Frederic
726 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
727 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
731 * This is mostly a bugfix release
733 * systems lacking /etc/os-release are no longer supported.
735 * Various functionality updates to libsystemd-login.so
737 * Track class of PAM logins to distuingish greeters from
740 Contributions from: Kay Sievers, Lennart Poettering, Michael
744 * This is an important bugfix release for v41.
746 * Building man pages is now optional which should be useful
747 for those building systemd from git but unwilling to install
750 * Watchdog support for supervising services is now usable. In
751 a future release support for hardware watchdogs
752 (i.e. /dev/watchdog) will be added building on this.
754 * Service start rate limiting is now configurable and can be
755 turned off per service. When a start rate limit is hit a
756 reboot can automatically be triggered.
758 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
760 Contributions from: Benjamin Franzke, Bill Nottingham,
761 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
762 Schmidt, Michał Górny, Piotr Drąg
765 * The systemd binary is installed /usr/lib/systemd/systemd now;
766 An existing /sbin/init symlink needs to be adapted with the
769 * The code that loads kernel modules has been ported to invoke
770 libkmod directly, instead of modprobe. This means we do not
771 support systems with module-init-tools anymore.
773 * Watchdog support is now already useful, but still not
776 * A new kernel command line option systemd.setenv= is
777 understood to set system wide environment variables
780 * We now limit the set of capabilities of systemd-journald.
782 * We now set SIGPIPE to ignore by default, since it only is
783 useful in shell pipelines, and has little use in general
784 code. This can be disabled with IgnoreSIPIPE=no in unit
787 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
788 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
792 * This is mostly a bugfix release
794 * We now expose the reason why a service failed in the
795 "Result" D-Bus property.
797 * Rudimentary service watchdog support (will be completed over
798 the next few releases.)
800 * When systemd forks off in order execute some service we will
801 now immediately changes its argv[0] to reflect which process
802 it will execute. This is useful to minimize the time window
803 with a generic argv[0], which makes bootcharts more useful
805 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
806 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
807 Mike Kazantsev, Ray Strode
810 * This is mostly a test release, but incorporates many
813 * New systemd-cgtop tool to show control groups by their
816 * Linking against libacl for ACLs is optional again. If
817 disabled, support tracking device access for active logins
818 goes becomes unavailable, and so does access to the user
819 journals by the respective users.
821 * If a group "adm" exists, journal files are automatically
822 owned by them, thus allow members of this group full access
823 to the system journal as well as all user journals.
825 * The journal now stores the SELinux context of the logging
826 client for all entries.
828 * Add C++ inclusion guards to all public headers
830 * New output mode "cat" in the journal to print only text
831 messages, without any meta data like date or time.
833 * Include tiny X server wrapper as a temporary stop-gap to
834 teach XOrg udev display enumeration. This is used by display
835 managers such as gdm, and will go away as soon as XOrg
836 learned native udev hotplugging for display devices.
838 * Add new systemd-cat tool for executing arbitrary programs
839 with STDERR/STDOUT connected to the journal. Can also act as
840 BSD logger replacement, and does so by default.
842 * Optionally store all locally generated coredumps in the
843 journal along with meta data.
845 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
846 writing short strings to files (for usage for /sys), and for
847 creating symlinks, character and block device nodes.
849 * New unit file option ControlGroupPersistent= to make cgroups
850 persistent, following the mechanisms outlined in
851 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
853 * Support multiple local RTCs in a sane way
855 * No longer monopolize IO when replaying readahead data on
856 rotating disks, since we might starve non-file-system IO to
857 death, since fanotify() will not see accesses done by blkid,
860 * Don't show kernel threads in systemd-cgls anymore, unless
861 requested with new -k switch.
863 Contributions from: Dan Horák, Kay Sievers, Lennart
864 Poettering, Michal Schmidt
867 * This is mostly a test release, but incorporates many
870 * The git repository moved to:
871 git://anongit.freedesktop.org/systemd/systemd
872 ssh://git.freedesktop.org/git/systemd/systemd
874 * First release with the journal
875 http://0pointer.de/blog/projects/the-journal.html
877 * The journal replaces both systemd-kmsg-syslogd and
878 systemd-stdout-bridge.
880 * New sd_pid_get_unit() API call in libsystemd-logind
882 * Many systemadm clean-ups
884 * Introduce remote-fs-pre.target which is ordered before all
885 remote mounts and may be used to start services before all
888 * Added Mageia support
890 * Add bash completion for systemd-loginctl
892 * Actively monitor PID file creation for daemons which exit in
893 the parent process before having finished writing the PID
894 file in the daemon process. Daemons which do this need to be
895 fixed (i.e. PID file creation must have finished before the
896 parent exits), but we now react a bit more gracefully to them.
898 * Add colourful boot output, mimicking the well-known output
899 of existing distributions.
901 * New option PassCredentials= for socket units, for
902 compatibility with a recent kernel ABI breakage.
904 * /etc/rc.local is now hooked in via a generator binary, and
905 thus will no longer act as synchronization point during
908 * systemctl list-unit-files now supports --root=.
910 * systemd-tmpfiles now understands two new commands: z, Z for
911 relabelling files according to the SELinux database. This is
912 useful to apply SELinux labels to specific files in /sys,
915 * Output of SysV services is now forwarded to both the console
916 and the journal by default, not only just the console.
918 * New man pages for all APIs from libsystemd-login.
920 * The build tree got reorganized and a the build system is a
921 lot more modular allowing embedded setups to specifically
922 select the components of systemd they are interested in.
924 * Support for Linux systems lacking the kernel VT subsystem is
927 * configure's --with-rootdir= got renamed to
928 --with-rootprefix= to follow the naming used by udev and
931 * Unless specified otherwise we'll now install to /usr instead
932 of /usr/local by default.
934 * Processes with '@' in argv[0][0] are now excluded from the
935 final shut-down killing spree, following the logic explained
937 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
939 * All processes remaining in a service cgroup when we enter
940 the START or START_PRE states are now killed with
941 SIGKILL. That means it is no longer possible to spawn
942 background processes from ExecStart= lines (which was never
943 supported anyway, and bad style).
945 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
946 reloading of units together.
948 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
949 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
950 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
951 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
952 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek