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