1 systemd System and Service Manager
4 * Note that we skipped 139 releases here in order to set the
5 new version to something that is greater than both udev's
6 and systemd's most recent version number.
8 * udev: all udev sources are merged into the systemd source tree now.
9 All future udev development will happen in the systemd tree. It
10 is still fully supported to use the udev daemon and tools without
11 systemd running, like in initramfs or other init systems. Building
12 udev though, will require the *build* of the systemd tree, but
13 udev can be properly *run* without systemd.
15 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
16 should be used to create dead device nodes as workarounds for broken
19 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
20 no longer supported. udev_monitor_new_from_netlink() needs to be
21 used to subscribe to events.
23 * udev: when udevd is started by systemd, processes which are left
24 behind by forking them off of udev rules, are unconditionally cleaned
25 up and killed now after the event handling has finished. Services or
26 daemons must be started as systemd services. Services can be
27 pulled-in by udev to get started, but they can no longer be directly
30 * udev: the daemon binary is called systemd-udevd now and installed
31 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
32 to adapt to that, create symlink, or rename the binary after building
35 * libudev no longer provides these symbols:
36 udev_monitor_from_socket()
37 udev_queue_get_failed_list_entry()
38 udev_get_{dev,sys,run}_path()
39 The versions number was bumped and symbol versioning introduced.
41 * systemd-loginctl and systemd-journalctl have been renamed
42 to loginctl and journalctl to match systemctl.
44 * The config files: /etc/systemd/systemd-logind.conf and
45 /etc/systemd/systemd-journald.conf have been renamed to
46 logind.conf and journald.conf. Package updates should rename
47 the files to the new names on upgrade.
49 * For almost all files the license is now LGPL2.1+, changed
50 from the previous GPL2.0+. Exceptions are some minor stuff
51 of udev (which will be changed to LGPL2.1 eventually, too),
52 and the MIT licensed sd-daemon.[ch] library that is suitable
53 to be used as drop-in files.
55 * systemd and logind now handle system sleep states, in
56 particulary suspending and hibernating.
58 * logind now implements a sleep/shutdown/idle inhibiting logic
59 suitable for a variety of uses. Soonishly Lennart will blog
60 about this in more detail.
62 * var-run.mount and var-lock.mount are no longer provided
63 (which prevously bind mounted these directories to their new
64 places). Distributions which have not converted these
65 directories to symlinks should consider stealing these files
66 from git history and add them downstream.
68 * We introduced the Documentation= field for units and added
69 this to all our shipped units. This is useful to make it
70 easier to explore the boot and the purpose of the various
73 * All smaller setup units (such as
74 systemd-vconsole-setup.service) now detect properly if they
75 are run in a container and are skipped when
76 appropriate. This guarantees an entirely noise-free boot in
77 Linux container environments such as systemd-nspawn.
79 * A framework for implementing offline system updates is now
80 integrated, for details see:
81 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
83 * A new service type Type=idle is available now which helps us
84 avoiding ugly interleaving of getty output and boot status
87 * There's now a system-wide CapabilityBoundingSet= option to
88 globally reduce the set of capabilities for the
89 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
90 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
91 even CAP_NET_ADMIN system-wide for secure systems.
93 * There are now system-wide DefaultLimitXXX= options to
94 globally change the defaults of the various resource limits
95 for all units started by PID 1.
97 * Harald Hoyer's systemd test suite has been integrated into
98 systemd which allows easy testing of systemd builds in qemu
99 and nspawn. (This is really awesome! Ask us for details!)
101 * The fstab parser is now implemented as generator, not inside
104 * systemctl will now warn you if .mount units generated from
105 /etc/fstab are out of date due to changes in fstab that
106 haven't been read by systemd yet.
108 * systemd is now suitable for usage in initrds. Dracut has
109 already been updated to make use of this. With this in place
110 initrds get a slight bit faster but primarily are much
111 easier to introspect and debug since "systemctl status" in
112 the host system can be used to introspect initrd services,
113 and the journal from the initrd is kept around too.
115 * systemd-delta has been added, a tool to explore differences
116 between user/admin configuration and vendor defaults.
118 * PrivateTmp= now affects both /tmp and /var/tmp.
120 * Boot time status messages are now much prettier and feature
121 proper english language. Booting up systemd has never been
124 * Read-ahead pack files now include the inode number of all
125 files to pre-cache. When the inode changes the pre-caching
126 is not attempted. This should be nicer to deal with updated
127 packages which might result in changes of read-ahead
130 * We now temporaritly lower the kernel's read_ahead_kb variable
131 when collecting read-ahead data to ensure the kernel's
132 built-in read-ahead does not add noise to our measurements
133 of necessary blocks to pre-cache.
135 * There's now RequiresMountsFor= to add automatic dependencies
136 for all mounts necessary for a specific file system path.
138 * MountAuto= and SwapAuto= have been removed from
139 system.conf. Mounting file systems at boot has to take place
142 * nspawn now learned a new switch --uuid= to set the machine
143 ID on the command line.
145 * nspawn now learned the -b switch to automatically search
148 * vt102 is now the default TERM for serial TTYs, upgraded from
151 * systemd-logind now works on VT-less systems.
153 * The build tree has been reorganized. The individual
154 components now have directories of their own.
156 * A new condition type ConditionPathIsReadWrite= is now available.
158 * nspawn learned the new -C switch to create cgroups for the
159 container in other hierarchies.
161 * We now have support for hardware watchdogs, configurable in
164 * The scheduled shutdown logic now has a public API.
166 * We now mount /tmp as tmpfs by default, but this can be
167 masked and /etc/fstab can override it.
169 * Since udisks doesn't make use of /media anymore we are not
170 mounting a tmpfs on it anymore.
172 * journalctl gained a new --local switch to only interleave
173 locally generated journal files.
175 * We can now load the IMA policy at boot automatically.
177 * The GTK tools have been split off into a systemd-ui.
180 * This is mostly a bugfix release
182 * Support optional initialization of the machine ID from the
183 KVM or container configured UUID.
185 * Support immediate reboots with "systemctl reboot -ff"
187 * Show /etc/os-release data in systemd-analyze output
189 * Many bugfixes for the journal, including endianess fixes and
190 ensuring that disk space enforcement works
192 * sd-login.h is C++ comptaible again
194 * Extend the /etc/os-release format on request of the Debian
197 * We now refuse non-UTF8 strings used in various configuration
198 and unit files. This is done to ensure we don't pass invalid
199 data over D-Bus or expose it elsewhere.
201 * Register Mimo USB Screens as suitable for automatic seat
204 * Read SELinux client context from journal clients in a race
207 * Reorder configuration file lookup order. /etc now always
208 overrides /run in order to allow the administrator to always
209 and unconditionally override vendor supplied or
210 automatically generated data.
212 * The various user visible bits of the journal now have man
213 pages. We still lack man pages for the journal API calls
216 * We now ship all man pages in HTML format again in the
219 Contributions from: Dave Reisner, Dirk Eibach, Frederic
220 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
221 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
225 * This is mostly a bugfix release
227 * systems lacking /etc/os-release are no longer supported.
229 * Various functionality updates to libsystemd-login.so
231 * Track class of PAM logins to distuingish greeters from
234 Contributions from: Kay Sievers, Lennart Poettering, Michael
238 * This is an important bugfix release for v41.
240 * Building man pages is now optional which should be useful
241 for those building systemd from git but unwilling to install
244 * Watchdog support for supervising services is now usable. In
245 a future release support for hardware watchdogs
246 (i.e. /dev/watchdog) will be added building on this.
248 * Service start rate limiting is now configurable and can be
249 turned off per service. When a start rate limit is hit a
250 reboot can automatically be triggered.
252 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
254 Contributions from: Benjamin Franzke, Bill Nottingham,
255 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
256 Schmidt, Michał Górny, Piotr Drąg
259 * The systemd binary is installed /usr/lib/systemd/systemd now;
260 An existing /sbin/init symlink needs to be adapted with the
263 * The code that loads kernel modules has been ported to invoke
264 libkmod directly, instead of modprobe. This means we do not
265 support systems with module-init-tools anymore.
267 * Watchdog support is now already useful, but still not
270 * A new kernel command line option systemd.setenv= is
271 understood to set system wide environment variables
274 * We now limit the set of capabilities of systemd-journald.
276 * We now set SIGPIPE to ignore by default, since it only is
277 useful in shell pipelines, and has little use in general
278 code. This can be disabled with IgnoreSIPIPE=no in unit
281 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
282 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
286 * This is mostly a bugfix release
288 * We now expose the reason why a service failed in the
289 "Result" D-Bus property.
291 * Rudimentary service watchdog support (will be completed over
292 the next few releases.)
294 * When systemd forks off in order execute some service we will
295 now immediately changes its argv[0] to reflect which process
296 it will execute. This is useful to minimize the time window
297 with a generic argv[0], which makes bootcharts more useful
299 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
300 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
301 Mike Kazantsev, Ray Strode
304 * This is mostly a test release, but incorporates many
307 * New systemd-cgtop tool to show control groups by their
310 * Linking against libacl for ACLs is optional again. If
311 disabled, support tracking device access for active logins
312 goes becomes unavailable, and so does access to the user
313 journals by the respective users.
315 * If a group "adm" exists, journal files are automatically
316 owned by them, thus allow members of this group full access
317 to the system journal as well as all user journals.
319 * The journal now stores the SELinux context of the logging
320 client for all entries.
322 * Add C++ inclusion guards to all public headers
324 * New output mode "cat" in the journal to print only text
325 messages, without any meta data like date or time.
327 * Include tiny X server wrapper as a temporary stop-gap to
328 teach XOrg udev display enumeration. This is used by display
329 managers such as gdm, and will go away as soon as XOrg
330 learned native udev hotplugging for display devices.
332 * Add new systemd-cat tool for executing arbitrary programs
333 with STDERR/STDOUT connected to the journal. Can also act as
334 BSD logger replacement, and does so by default.
336 * Optionally store all locally generated coredumps in the
337 journal along with meta data.
339 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
340 writing short strings to files (for usage for /sys), and for
341 creating symlinks, character and block device nodes.
343 * New unit file option ControlGroupPersistent= to make cgroups
344 persistent, following the mechanisms outlined in
345 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
347 * Support multiple local RTCs in a sane way
349 * No longer monopolize IO when replaying readahead data on
350 rotating disks, since we might starve non-file-system IO to
351 death, since fanotify() will not see accesses done by blkid,
354 * Don't show kernel threads in systemd-cgls anymore, unless
355 requested with new -k switch.
357 Contributions from: Dan Horák, Kay Sievers, Lennart
358 Poettering, Michal Schmidt
361 * This is mostly a test release, but incorporates many
364 * The git repository moved to:
365 git://anongit.freedesktop.org/systemd/systemd
366 ssh://git.freedesktop.org/git/systemd/systemd
368 * First release with the journal
369 http://0pointer.de/blog/projects/the-journal.html
371 * The journal replaces both systemd-kmsg-syslogd and
372 systemd-stdout-bridge.
374 * New sd_pid_get_unit() API call in libsystemd-logind
376 * Many systemadm clean-ups
378 * Introduce remote-fs-pre.target which is ordered before all
379 remote mounts and may be used to start services before all
382 * Added Mageia support
384 * Add bash completion for systemd-loginctl
386 * Actively monitor PID file creation for daemons which exit in
387 the parent process before having finished writing the PID
388 file in the daemon process. Daemons which do this need to be
389 fixed (i.e. PID file creation must have finished before the
390 parent exits), but we now react a bit more gracefully to them.
392 * Add colourful boot output, mimicking the well-known output
393 of existing distributions.
395 * New option PassCredentials= for socket units, for
396 compatibility with a recent kernel ABI breakage.
398 * /etc/rc.local is now hooked in via a generator binary, and
399 thus will no longer act as synchronization point during
402 * systemctl list-unit-files now supports --root=.
404 * systemd-tmpfiles now understands two new commands: z, Z for
405 relabelling files according to the SELinux database. This is
406 useful to apply SELinux labels to specific files in /sys,
409 * Output of SysV services is now forwarded to both the console
410 and the journal by default, not only just the console.
412 * New man pages for all APIs from libsystemd-login.
414 * The build tree got reorganized and a the build system is a
415 lot more modular allowing embedded setups to specifically
416 select the components of systemd they are interested in.
418 * Support for Linux systems lacking the kernel VT subsystem is
421 * configure's --with-rootdir= got renamed to
422 --with-rootprefix= to follow the naming used by udev and
425 * Unless specified otherwise we'll now install to /usr instead
426 of /usr/local by default.
428 * Processes with '@' in argv[0][0] are now excluded from the
429 final shut-down killing spree, following the logic explained
431 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
433 * All processes remaining in a service cgroup when we enter
434 the START or START_PRE states are now killed with
435 SIGKILL. That means it is no longer possible to spawn
436 background processes from ExecStart= lines (which was never
437 supported anyway, and bad style).
439 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
440 reloading of units together.
442 Contributions from: Bill Nottingham, Daniel Walsh, Dave
443 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
444 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
445 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
446 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek