1 systemd System and Service Manager
4 * "systemctl help <unit>" now shows the man page if one is
7 * Several new man pages have been added.
9 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=, MaxLevelConsole=
10 can now be specified in journald.conf.
12 * TimerSlackNSec= can now be specified in system.conf for PID1.
14 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
15 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
20 * logind is now capable of (optionally) handling power and
21 sleep keys as well as the lid switch.
23 * journalctl now understands the syntax "journalctl
24 /usr/bin/avahi-daemon" to get all log output of a specific
27 * CapabilityBoundingSet= in system.conf now also influences
28 the capability bound set of usermode helpers of the kernel.
30 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
31 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
32 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
33 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
36 * Note that we skipped 139 releases here in order to set the
37 new version to something that is greater than both udev's
38 and systemd's most recent version number.
40 * udev: all udev sources are merged into the systemd source tree now.
41 All future udev development will happen in the systemd tree. It
42 is still fully supported to use the udev daemon and tools without
43 systemd running, like in initramfs or other init systems. Building
44 udev though, will require the *build* of the systemd tree, but
45 udev can be properly *run* without systemd.
47 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
48 should be used to create dead device nodes as workarounds for broken
51 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
52 no longer supported. udev_monitor_new_from_netlink() needs to be
53 used to subscribe to events.
55 * udev: when udevd is started by systemd, processes which are left
56 behind by forking them off of udev rules, are unconditionally cleaned
57 up and killed now after the event handling has finished. Services or
58 daemons must be started as systemd services. Services can be
59 pulled-in by udev to get started, but they can no longer be directly
62 * udev: the daemon binary is called systemd-udevd now and installed
63 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
64 to adapt to that, create symlink, or rename the binary after building
67 * libudev no longer provides these symbols:
68 udev_monitor_from_socket()
69 udev_queue_get_failed_list_entry()
70 udev_get_{dev,sys,run}_path()
71 The versions number was bumped and symbol versioning introduced.
73 * systemd-loginctl and systemd-journalctl have been renamed
74 to loginctl and journalctl to match systemctl.
76 * The config files: /etc/systemd/systemd-logind.conf and
77 /etc/systemd/systemd-journald.conf have been renamed to
78 logind.conf and journald.conf. Package updates should rename
79 the files to the new names on upgrade.
81 * For almost all files the license is now LGPL2.1+, changed
82 from the previous GPL2.0+. Exceptions are some minor stuff
83 of udev (which will be changed to LGPL2.1 eventually, too),
84 and the MIT licensed sd-daemon.[ch] library that is suitable
85 to be used as drop-in files.
87 * systemd and logind now handle system sleep states, in
88 particulary suspending and hibernating.
90 * logind now implements a sleep/shutdown/idle inhibiting logic
91 suitable for a variety of uses. Soonishly Lennart will blog
92 about this in more detail.
94 * var-run.mount and var-lock.mount are no longer provided
95 (which prevously bind mounted these directories to their new
96 places). Distributions which have not converted these
97 directories to symlinks should consider stealing these files
98 from git history and add them downstream.
100 * We introduced the Documentation= field for units and added
101 this to all our shipped units. This is useful to make it
102 easier to explore the boot and the purpose of the various
105 * All smaller setup units (such as
106 systemd-vconsole-setup.service) now detect properly if they
107 are run in a container and are skipped when
108 appropriate. This guarantees an entirely noise-free boot in
109 Linux container environments such as systemd-nspawn.
111 * A framework for implementing offline system updates is now
112 integrated, for details see:
113 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
115 * A new service type Type=idle is available now which helps us
116 avoiding ugly interleaving of getty output and boot status
119 * There's now a system-wide CapabilityBoundingSet= option to
120 globally reduce the set of capabilities for the
121 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
122 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
123 even CAP_NET_ADMIN system-wide for secure systems.
125 * There are now system-wide DefaultLimitXXX= options to
126 globally change the defaults of the various resource limits
127 for all units started by PID 1.
129 * Harald Hoyer's systemd test suite has been integrated into
130 systemd which allows easy testing of systemd builds in qemu
131 and nspawn. (This is really awesome! Ask us for details!)
133 * The fstab parser is now implemented as generator, not inside
136 * systemctl will now warn you if .mount units generated from
137 /etc/fstab are out of date due to changes in fstab that
138 haven't been read by systemd yet.
140 * systemd is now suitable for usage in initrds. Dracut has
141 already been updated to make use of this. With this in place
142 initrds get a slight bit faster but primarily are much
143 easier to introspect and debug since "systemctl status" in
144 the host system can be used to introspect initrd services,
145 and the journal from the initrd is kept around too.
147 * systemd-delta has been added, a tool to explore differences
148 between user/admin configuration and vendor defaults.
150 * PrivateTmp= now affects both /tmp and /var/tmp.
152 * Boot time status messages are now much prettier and feature
153 proper english language. Booting up systemd has never been
156 * Read-ahead pack files now include the inode number of all
157 files to pre-cache. When the inode changes the pre-caching
158 is not attempted. This should be nicer to deal with updated
159 packages which might result in changes of read-ahead
162 * We now temporaritly lower the kernel's read_ahead_kb variable
163 when collecting read-ahead data to ensure the kernel's
164 built-in read-ahead does not add noise to our measurements
165 of necessary blocks to pre-cache.
167 * There's now RequiresMountsFor= to add automatic dependencies
168 for all mounts necessary for a specific file system path.
170 * MountAuto= and SwapAuto= have been removed from
171 system.conf. Mounting file systems at boot has to take place
174 * nspawn now learned a new switch --uuid= to set the machine
175 ID on the command line.
177 * nspawn now learned the -b switch to automatically search
180 * vt102 is now the default TERM for serial TTYs, upgraded from
183 * systemd-logind now works on VT-less systems.
185 * The build tree has been reorganized. The individual
186 components now have directories of their own.
188 * A new condition type ConditionPathIsReadWrite= is now available.
190 * nspawn learned the new -C switch to create cgroups for the
191 container in other hierarchies.
193 * We now have support for hardware watchdogs, configurable in
196 * The scheduled shutdown logic now has a public API.
198 * We now mount /tmp as tmpfs by default, but this can be
199 masked and /etc/fstab can override it.
201 * Since udisks doesn't make use of /media anymore we are not
202 mounting a tmpfs on it anymore.
204 * journalctl gained a new --local switch to only interleave
205 locally generated journal files.
207 * We can now load the IMA policy at boot automatically.
209 * The GTK tools have been split off into a systemd-ui.
211 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
212 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
213 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
214 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
215 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
216 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
217 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
218 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
219 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
223 * This is mostly a bugfix release
225 * Support optional initialization of the machine ID from the
226 KVM or container configured UUID.
228 * Support immediate reboots with "systemctl reboot -ff"
230 * Show /etc/os-release data in systemd-analyze output
232 * Many bugfixes for the journal, including endianess fixes and
233 ensuring that disk space enforcement works
235 * sd-login.h is C++ comptaible again
237 * Extend the /etc/os-release format on request of the Debian
240 * We now refuse non-UTF8 strings used in various configuration
241 and unit files. This is done to ensure we don't pass invalid
242 data over D-Bus or expose it elsewhere.
244 * Register Mimo USB Screens as suitable for automatic seat
247 * Read SELinux client context from journal clients in a race
250 * Reorder configuration file lookup order. /etc now always
251 overrides /run in order to allow the administrator to always
252 and unconditionally override vendor supplied or
253 automatically generated data.
255 * The various user visible bits of the journal now have man
256 pages. We still lack man pages for the journal API calls
259 * We now ship all man pages in HTML format again in the
262 Contributions from: Dave Reisner, Dirk Eibach, Frederic
263 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
264 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
268 * This is mostly a bugfix release
270 * systems lacking /etc/os-release are no longer supported.
272 * Various functionality updates to libsystemd-login.so
274 * Track class of PAM logins to distuingish greeters from
277 Contributions from: Kay Sievers, Lennart Poettering, Michael
281 * This is an important bugfix release for v41.
283 * Building man pages is now optional which should be useful
284 for those building systemd from git but unwilling to install
287 * Watchdog support for supervising services is now usable. In
288 a future release support for hardware watchdogs
289 (i.e. /dev/watchdog) will be added building on this.
291 * Service start rate limiting is now configurable and can be
292 turned off per service. When a start rate limit is hit a
293 reboot can automatically be triggered.
295 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
297 Contributions from: Benjamin Franzke, Bill Nottingham,
298 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
299 Schmidt, Michał Górny, Piotr Drąg
302 * The systemd binary is installed /usr/lib/systemd/systemd now;
303 An existing /sbin/init symlink needs to be adapted with the
306 * The code that loads kernel modules has been ported to invoke
307 libkmod directly, instead of modprobe. This means we do not
308 support systems with module-init-tools anymore.
310 * Watchdog support is now already useful, but still not
313 * A new kernel command line option systemd.setenv= is
314 understood to set system wide environment variables
317 * We now limit the set of capabilities of systemd-journald.
319 * We now set SIGPIPE to ignore by default, since it only is
320 useful in shell pipelines, and has little use in general
321 code. This can be disabled with IgnoreSIPIPE=no in unit
324 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
325 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
329 * This is mostly a bugfix release
331 * We now expose the reason why a service failed in the
332 "Result" D-Bus property.
334 * Rudimentary service watchdog support (will be completed over
335 the next few releases.)
337 * When systemd forks off in order execute some service we will
338 now immediately changes its argv[0] to reflect which process
339 it will execute. This is useful to minimize the time window
340 with a generic argv[0], which makes bootcharts more useful
342 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
343 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
344 Mike Kazantsev, Ray Strode
347 * This is mostly a test release, but incorporates many
350 * New systemd-cgtop tool to show control groups by their
353 * Linking against libacl for ACLs is optional again. If
354 disabled, support tracking device access for active logins
355 goes becomes unavailable, and so does access to the user
356 journals by the respective users.
358 * If a group "adm" exists, journal files are automatically
359 owned by them, thus allow members of this group full access
360 to the system journal as well as all user journals.
362 * The journal now stores the SELinux context of the logging
363 client for all entries.
365 * Add C++ inclusion guards to all public headers
367 * New output mode "cat" in the journal to print only text
368 messages, without any meta data like date or time.
370 * Include tiny X server wrapper as a temporary stop-gap to
371 teach XOrg udev display enumeration. This is used by display
372 managers such as gdm, and will go away as soon as XOrg
373 learned native udev hotplugging for display devices.
375 * Add new systemd-cat tool for executing arbitrary programs
376 with STDERR/STDOUT connected to the journal. Can also act as
377 BSD logger replacement, and does so by default.
379 * Optionally store all locally generated coredumps in the
380 journal along with meta data.
382 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
383 writing short strings to files (for usage for /sys), and for
384 creating symlinks, character and block device nodes.
386 * New unit file option ControlGroupPersistent= to make cgroups
387 persistent, following the mechanisms outlined in
388 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
390 * Support multiple local RTCs in a sane way
392 * No longer monopolize IO when replaying readahead data on
393 rotating disks, since we might starve non-file-system IO to
394 death, since fanotify() will not see accesses done by blkid,
397 * Don't show kernel threads in systemd-cgls anymore, unless
398 requested with new -k switch.
400 Contributions from: Dan Horák, Kay Sievers, Lennart
401 Poettering, Michal Schmidt
404 * This is mostly a test release, but incorporates many
407 * The git repository moved to:
408 git://anongit.freedesktop.org/systemd/systemd
409 ssh://git.freedesktop.org/git/systemd/systemd
411 * First release with the journal
412 http://0pointer.de/blog/projects/the-journal.html
414 * The journal replaces both systemd-kmsg-syslogd and
415 systemd-stdout-bridge.
417 * New sd_pid_get_unit() API call in libsystemd-logind
419 * Many systemadm clean-ups
421 * Introduce remote-fs-pre.target which is ordered before all
422 remote mounts and may be used to start services before all
425 * Added Mageia support
427 * Add bash completion for systemd-loginctl
429 * Actively monitor PID file creation for daemons which exit in
430 the parent process before having finished writing the PID
431 file in the daemon process. Daemons which do this need to be
432 fixed (i.e. PID file creation must have finished before the
433 parent exits), but we now react a bit more gracefully to them.
435 * Add colourful boot output, mimicking the well-known output
436 of existing distributions.
438 * New option PassCredentials= for socket units, for
439 compatibility with a recent kernel ABI breakage.
441 * /etc/rc.local is now hooked in via a generator binary, and
442 thus will no longer act as synchronization point during
445 * systemctl list-unit-files now supports --root=.
447 * systemd-tmpfiles now understands two new commands: z, Z for
448 relabelling files according to the SELinux database. This is
449 useful to apply SELinux labels to specific files in /sys,
452 * Output of SysV services is now forwarded to both the console
453 and the journal by default, not only just the console.
455 * New man pages for all APIs from libsystemd-login.
457 * The build tree got reorganized and a the build system is a
458 lot more modular allowing embedded setups to specifically
459 select the components of systemd they are interested in.
461 * Support for Linux systems lacking the kernel VT subsystem is
464 * configure's --with-rootdir= got renamed to
465 --with-rootprefix= to follow the naming used by udev and
468 * Unless specified otherwise we'll now install to /usr instead
469 of /usr/local by default.
471 * Processes with '@' in argv[0][0] are now excluded from the
472 final shut-down killing spree, following the logic explained
474 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
476 * All processes remaining in a service cgroup when we enter
477 the START or START_PRE states are now killed with
478 SIGKILL. That means it is no longer possible to spawn
479 background processes from ExecStart= lines (which was never
480 supported anyway, and bad style).
482 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
483 reloading of units together.
485 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
486 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
487 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
488 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
489 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek