1 systemd System and Service Manager
5 * Several tools now understand kernel command line arguments,
6 which are only read when run in an initial RAM disk. They
7 usually follow closely their normal counterparts, but are
10 * There's a new tool to analyze the readahead files that are
11 automatically generated at boot. Use:
13 /usr/lib/systemd/systemd-readahead analyze /.readahead
15 * We now provide an early debug shell on tty9 if this enabled. Use:
17 systemctl enable debug-shell.service
19 * All plymouth related units have been moved into the Plymouth
20 package. Please make sure to upgrade your Plymouth version
23 * systemd-tmpfiles now supports getting passed the basename of
24 a configuration file only, in which case it will look for it
25 in all appropriate directories automatically.
27 * udevadm info now takes a /dev or /sys path as argument, and
28 does the right thing. Example:
31 udevadm info /sys/class/block/sda
33 * systemctl now prints a warning if a unit is stopped but a
34 unit that might trigger it continues to run. Example: a
35 service is stopped but the socket that activates it is left
38 * "systemctl status" will now mention if the log output was
39 shortened due to rotation since a service has been started.
41 * The journal API now exposes functions to determine the
42 "cutoff" times due to rotation.
44 * journald now understands SIGUSR1 and SIGUSR2 for triggering
45 immediately flushing of runtime logs to /var if possible,
46 resp. for triggering immediate rotation of the journal
49 * It is now considered an error if a service is attempted to
50 be stopped that is not loaded.
52 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
54 * systemd-analyze now supports Python 3
56 * tmpfiles now supports cleaning up directories via aging
57 where the first level dirs are always kept around but
58 directories beneath it automatically aged. This is enabled
59 by prefixing the age field with '~'.
61 * Seat objects now expose CanGraphical, CanTTY properties
62 which is required to deal with very fast bootups where the
63 display manager might be running before the graphics drivers
64 completed initialization.
66 * Seat objects now expose a State property.
68 * We now include RPM macros for service enabling/disabling
69 based on the preset logic. We recommend RPM based
70 distributions to make use of these macros if possible. This
71 makes it simpler to reuse RPM spec files across
74 * We now make sure that the collected systemd unit name is
75 always valid when services log to the journal via
78 * There's a new man page kernel-command-line(7) detailing all
79 command line options we understand.
81 * The fstab generator may now be disabled at boot by passing
82 fstab=0 on the kernel command line.
84 * A new kernel command line option modules-load= is now understood
85 to load a specific kernel module statically, early at boot.
87 * Unit names specified on the systemctl command line are now
88 automatically escaped as needed. Also, if file system or
89 device paths are specified they are automatically turned
90 into the appropriate mount or device unit names. Example:
92 systemctl status /home
93 systemctl status /dev/sda
95 * The SysVConsole= configuration option has been removed from
98 * The SysV search path is no longer exported on the D-Bus
101 * The Names= option is been removed from unit file parsing.
103 * There's a new man page bootup(7) detailing the boot process.
105 * Every unit and every generator we ship with systemd now
106 comes with full documentation. The self-explanatory boot is
109 * A couple of services gained "systemd-" prefixes in their
110 name if they wrap systemd code, rather than only external
111 code. Among them fsck@.service which is now
112 systemd-fsck@.service.
114 * The HaveWatchdog property has been removed from the D-Bus
117 * systemd.confirm_spawn= on the kernel command line should now
120 * There's a new man page crypttab(5) which details all options
121 we actually understand.
123 * systemd-nspawn gained a new --capability= switch to pass
124 additional capabilities to the container.
126 * timedated will now read known NTP implementation unit names
127 from /usr/share/systemd/ntp-units.d/*.list,
128 systemd-timedated-ntp.target has been removed.
130 * journalctl gained a new switch "-b" that lists log data of
131 the current boot only.
133 * The notify socket is in the abstract namespace again, in
134 order to support daemons which chroot() at start-up.
136 * There is a new Storage= configuration option for journald
137 which allows configuration of where log data should go. This
138 also provides a way to disable journal logging entirely, so
139 that data collected is only forwarded to the console, the
140 kernel log buffer or another syslog implementation.
142 * Many bugfixes and optimizations.
144 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
145 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
146 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
147 Shawn Landden, Tom Gundersen
150 * "systemctl help <unit>" now shows the man page if one is
153 * Several new man pages have been added.
155 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
156 MaxLevelConsole= can now be specified in
157 journald.conf. These options allow reducing the amount of
158 data stored on disk or forwarded by the log level.
160 * TimerSlackNSec= can now be specified in system.conf for
161 PID1. This allows system-wide power savings.
163 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
164 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
168 * logind is now capable of (optionally) handling power and
169 sleep keys as well as the lid switch.
171 * journalctl now understands the syntax "journalctl
172 /usr/bin/avahi-daemon" to get all log output of a specific
175 * CapabilityBoundingSet= in system.conf now also influences
176 the capability bound set of usermode helpers of the kernel.
178 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
179 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
180 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
181 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
184 * Note that we skipped 139 releases here in order to set the
185 new version to something that is greater than both udev's
186 and systemd's most recent version number.
188 * udev: all udev sources are merged into the systemd source tree now.
189 All future udev development will happen in the systemd tree. It
190 is still fully supported to use the udev daemon and tools without
191 systemd running, like in initramfs or other init systems. Building
192 udev though, will require the *build* of the systemd tree, but
193 udev can be properly *run* without systemd.
195 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
196 should be used to create dead device nodes as workarounds for broken
199 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
200 no longer supported. udev_monitor_new_from_netlink() needs to be
201 used to subscribe to events.
203 * udev: when udevd is started by systemd, processes which are left
204 behind by forking them off of udev rules, are unconditionally cleaned
205 up and killed now after the event handling has finished. Services or
206 daemons must be started as systemd services. Services can be
207 pulled-in by udev to get started, but they can no longer be directly
208 forked by udev rules.
210 * udev: the daemon binary is called systemd-udevd now and installed
211 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
212 to adapt to that, create symlink, or rename the binary after building
215 * libudev no longer provides these symbols:
216 udev_monitor_from_socket()
217 udev_queue_get_failed_list_entry()
218 udev_get_{dev,sys,run}_path()
219 The versions number was bumped and symbol versioning introduced.
221 * systemd-loginctl and systemd-journalctl have been renamed
222 to loginctl and journalctl to match systemctl.
224 * The config files: /etc/systemd/systemd-logind.conf and
225 /etc/systemd/systemd-journald.conf have been renamed to
226 logind.conf and journald.conf. Package updates should rename
227 the files to the new names on upgrade.
229 * For almost all files the license is now LGPL2.1+, changed
230 from the previous GPL2.0+. Exceptions are some minor stuff
231 of udev (which will be changed to LGPL2.1 eventually, too),
232 and the MIT licensed sd-daemon.[ch] library that is suitable
233 to be used as drop-in files.
235 * systemd and logind now handle system sleep states, in
236 particulary suspending and hibernating.
238 * logind now implements a sleep/shutdown/idle inhibiting logic
239 suitable for a variety of uses. Soonishly Lennart will blog
240 about this in more detail.
242 * var-run.mount and var-lock.mount are no longer provided
243 (which prevously bind mounted these directories to their new
244 places). Distributions which have not converted these
245 directories to symlinks should consider stealing these files
246 from git history and add them downstream.
248 * We introduced the Documentation= field for units and added
249 this to all our shipped units. This is useful to make it
250 easier to explore the boot and the purpose of the various
253 * All smaller setup units (such as
254 systemd-vconsole-setup.service) now detect properly if they
255 are run in a container and are skipped when
256 appropriate. This guarantees an entirely noise-free boot in
257 Linux container environments such as systemd-nspawn.
259 * A framework for implementing offline system updates is now
260 integrated, for details see:
261 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
263 * A new service type Type=idle is available now which helps us
264 avoiding ugly interleaving of getty output and boot status
267 * There's now a system-wide CapabilityBoundingSet= option to
268 globally reduce the set of capabilities for the
269 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
270 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
271 even CAP_NET_ADMIN system-wide for secure systems.
273 * There are now system-wide DefaultLimitXXX= options to
274 globally change the defaults of the various resource limits
275 for all units started by PID 1.
277 * Harald Hoyer's systemd test suite has been integrated into
278 systemd which allows easy testing of systemd builds in qemu
279 and nspawn. (This is really awesome! Ask us for details!)
281 * The fstab parser is now implemented as generator, not inside
284 * systemctl will now warn you if .mount units generated from
285 /etc/fstab are out of date due to changes in fstab that
286 haven't been read by systemd yet.
288 * systemd is now suitable for usage in initrds. Dracut has
289 already been updated to make use of this. With this in place
290 initrds get a slight bit faster but primarily are much
291 easier to introspect and debug since "systemctl status" in
292 the host system can be used to introspect initrd services,
293 and the journal from the initrd is kept around too.
295 * systemd-delta has been added, a tool to explore differences
296 between user/admin configuration and vendor defaults.
298 * PrivateTmp= now affects both /tmp and /var/tmp.
300 * Boot time status messages are now much prettier and feature
301 proper english language. Booting up systemd has never been
304 * Read-ahead pack files now include the inode number of all
305 files to pre-cache. When the inode changes the pre-caching
306 is not attempted. This should be nicer to deal with updated
307 packages which might result in changes of read-ahead
310 * We now temporaritly lower the kernel's read_ahead_kb variable
311 when collecting read-ahead data to ensure the kernel's
312 built-in read-ahead does not add noise to our measurements
313 of necessary blocks to pre-cache.
315 * There's now RequiresMountsFor= to add automatic dependencies
316 for all mounts necessary for a specific file system path.
318 * MountAuto= and SwapAuto= have been removed from
319 system.conf. Mounting file systems at boot has to take place
322 * nspawn now learned a new switch --uuid= to set the machine
323 ID on the command line.
325 * nspawn now learned the -b switch to automatically search
328 * vt102 is now the default TERM for serial TTYs, upgraded from
331 * systemd-logind now works on VT-less systems.
333 * The build tree has been reorganized. The individual
334 components now have directories of their own.
336 * A new condition type ConditionPathIsReadWrite= is now available.
338 * nspawn learned the new -C switch to create cgroups for the
339 container in other hierarchies.
341 * We now have support for hardware watchdogs, configurable in
344 * The scheduled shutdown logic now has a public API.
346 * We now mount /tmp as tmpfs by default, but this can be
347 masked and /etc/fstab can override it.
349 * Since udisks doesn't make use of /media anymore we are not
350 mounting a tmpfs on it anymore.
352 * journalctl gained a new --local switch to only interleave
353 locally generated journal files.
355 * We can now load the IMA policy at boot automatically.
357 * The GTK tools have been split off into a systemd-ui.
359 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
360 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
361 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
362 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
363 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
364 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
365 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
366 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
367 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
371 * This is mostly a bugfix release
373 * Support optional initialization of the machine ID from the
374 KVM or container configured UUID.
376 * Support immediate reboots with "systemctl reboot -ff"
378 * Show /etc/os-release data in systemd-analyze output
380 * Many bugfixes for the journal, including endianess fixes and
381 ensuring that disk space enforcement works
383 * sd-login.h is C++ comptaible again
385 * Extend the /etc/os-release format on request of the Debian
388 * We now refuse non-UTF8 strings used in various configuration
389 and unit files. This is done to ensure we don't pass invalid
390 data over D-Bus or expose it elsewhere.
392 * Register Mimo USB Screens as suitable for automatic seat
395 * Read SELinux client context from journal clients in a race
398 * Reorder configuration file lookup order. /etc now always
399 overrides /run in order to allow the administrator to always
400 and unconditionally override vendor supplied or
401 automatically generated data.
403 * The various user visible bits of the journal now have man
404 pages. We still lack man pages for the journal API calls
407 * We now ship all man pages in HTML format again in the
410 Contributions from: Dave Reisner, Dirk Eibach, Frederic
411 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
412 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
416 * This is mostly a bugfix release
418 * systems lacking /etc/os-release are no longer supported.
420 * Various functionality updates to libsystemd-login.so
422 * Track class of PAM logins to distuingish greeters from
425 Contributions from: Kay Sievers, Lennart Poettering, Michael
429 * This is an important bugfix release for v41.
431 * Building man pages is now optional which should be useful
432 for those building systemd from git but unwilling to install
435 * Watchdog support for supervising services is now usable. In
436 a future release support for hardware watchdogs
437 (i.e. /dev/watchdog) will be added building on this.
439 * Service start rate limiting is now configurable and can be
440 turned off per service. When a start rate limit is hit a
441 reboot can automatically be triggered.
443 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
445 Contributions from: Benjamin Franzke, Bill Nottingham,
446 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
447 Schmidt, Michał Górny, Piotr Drąg
450 * The systemd binary is installed /usr/lib/systemd/systemd now;
451 An existing /sbin/init symlink needs to be adapted with the
454 * The code that loads kernel modules has been ported to invoke
455 libkmod directly, instead of modprobe. This means we do not
456 support systems with module-init-tools anymore.
458 * Watchdog support is now already useful, but still not
461 * A new kernel command line option systemd.setenv= is
462 understood to set system wide environment variables
465 * We now limit the set of capabilities of systemd-journald.
467 * We now set SIGPIPE to ignore by default, since it only is
468 useful in shell pipelines, and has little use in general
469 code. This can be disabled with IgnoreSIPIPE=no in unit
472 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
473 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
477 * This is mostly a bugfix release
479 * We now expose the reason why a service failed in the
480 "Result" D-Bus property.
482 * Rudimentary service watchdog support (will be completed over
483 the next few releases.)
485 * When systemd forks off in order execute some service we will
486 now immediately changes its argv[0] to reflect which process
487 it will execute. This is useful to minimize the time window
488 with a generic argv[0], which makes bootcharts more useful
490 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
491 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
492 Mike Kazantsev, Ray Strode
495 * This is mostly a test release, but incorporates many
498 * New systemd-cgtop tool to show control groups by their
501 * Linking against libacl for ACLs is optional again. If
502 disabled, support tracking device access for active logins
503 goes becomes unavailable, and so does access to the user
504 journals by the respective users.
506 * If a group "adm" exists, journal files are automatically
507 owned by them, thus allow members of this group full access
508 to the system journal as well as all user journals.
510 * The journal now stores the SELinux context of the logging
511 client for all entries.
513 * Add C++ inclusion guards to all public headers
515 * New output mode "cat" in the journal to print only text
516 messages, without any meta data like date or time.
518 * Include tiny X server wrapper as a temporary stop-gap to
519 teach XOrg udev display enumeration. This is used by display
520 managers such as gdm, and will go away as soon as XOrg
521 learned native udev hotplugging for display devices.
523 * Add new systemd-cat tool for executing arbitrary programs
524 with STDERR/STDOUT connected to the journal. Can also act as
525 BSD logger replacement, and does so by default.
527 * Optionally store all locally generated coredumps in the
528 journal along with meta data.
530 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
531 writing short strings to files (for usage for /sys), and for
532 creating symlinks, character and block device nodes.
534 * New unit file option ControlGroupPersistent= to make cgroups
535 persistent, following the mechanisms outlined in
536 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
538 * Support multiple local RTCs in a sane way
540 * No longer monopolize IO when replaying readahead data on
541 rotating disks, since we might starve non-file-system IO to
542 death, since fanotify() will not see accesses done by blkid,
545 * Don't show kernel threads in systemd-cgls anymore, unless
546 requested with new -k switch.
548 Contributions from: Dan Horák, Kay Sievers, Lennart
549 Poettering, Michal Schmidt
552 * This is mostly a test release, but incorporates many
555 * The git repository moved to:
556 git://anongit.freedesktop.org/systemd/systemd
557 ssh://git.freedesktop.org/git/systemd/systemd
559 * First release with the journal
560 http://0pointer.de/blog/projects/the-journal.html
562 * The journal replaces both systemd-kmsg-syslogd and
563 systemd-stdout-bridge.
565 * New sd_pid_get_unit() API call in libsystemd-logind
567 * Many systemadm clean-ups
569 * Introduce remote-fs-pre.target which is ordered before all
570 remote mounts and may be used to start services before all
573 * Added Mageia support
575 * Add bash completion for systemd-loginctl
577 * Actively monitor PID file creation for daemons which exit in
578 the parent process before having finished writing the PID
579 file in the daemon process. Daemons which do this need to be
580 fixed (i.e. PID file creation must have finished before the
581 parent exits), but we now react a bit more gracefully to them.
583 * Add colourful boot output, mimicking the well-known output
584 of existing distributions.
586 * New option PassCredentials= for socket units, for
587 compatibility with a recent kernel ABI breakage.
589 * /etc/rc.local is now hooked in via a generator binary, and
590 thus will no longer act as synchronization point during
593 * systemctl list-unit-files now supports --root=.
595 * systemd-tmpfiles now understands two new commands: z, Z for
596 relabelling files according to the SELinux database. This is
597 useful to apply SELinux labels to specific files in /sys,
600 * Output of SysV services is now forwarded to both the console
601 and the journal by default, not only just the console.
603 * New man pages for all APIs from libsystemd-login.
605 * The build tree got reorganized and a the build system is a
606 lot more modular allowing embedded setups to specifically
607 select the components of systemd they are interested in.
609 * Support for Linux systems lacking the kernel VT subsystem is
612 * configure's --with-rootdir= got renamed to
613 --with-rootprefix= to follow the naming used by udev and
616 * Unless specified otherwise we'll now install to /usr instead
617 of /usr/local by default.
619 * Processes with '@' in argv[0][0] are now excluded from the
620 final shut-down killing spree, following the logic explained
622 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
624 * All processes remaining in a service cgroup when we enter
625 the START or START_PRE states are now killed with
626 SIGKILL. That means it is no longer possible to spawn
627 background processes from ExecStart= lines (which was never
628 supported anyway, and bad style).
630 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
631 reloading of units together.
633 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
634 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
635 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
636 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
637 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek