1 systemd System and Service Manager
5 * The systemd-resolved DNS resolver service has gained a substantial
6 set of new features, most prominently it may now act as a DNSSEC
7 validating stub resolver. DNSSEC mode is currently turned off by
8 default, but it is expected that this is turned on by default in one
9 of the next releases. For now, we invite everybody to test the DNSSEC
10 logic by setting DNSSEC=allow-downgrade in
11 /etc/systemd/resolved.conf. The service also gained a full set of
12 D-Bus interfaces, including calls to configure DNS and DNSSEC
13 settings per link (for consumption by external network management
14 software). systemd-resolved (and systemd-networkd along with it) now
15 know to distinguish between "search" and "routing" domains. The
16 former are used to qualify single-label names, the latter are purely
17 used for routing lookups within certain domains to specific
18 links. resolved will now also synthesize RRs for all entries from
21 * The systemd-resolve tool (which is a client utility for
22 systemd-resolved, and previously experimental) has been improved
23 considerably and is now fully supported and documented. Hence it has
24 moved from /usr/lib/systemd to /usr/bin.
26 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
29 * The coredump collection logic has been reworked: when a coredump is
30 collected it is now written to disk, compressed and processed
31 (including stacktrace extraction) from a new instantiated service
32 systemd-coredump@.service, instead of directly from the
33 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
34 processing large coredumps can take up a substantial amount of
35 resources and time, and this previously happened entirely outside of
36 systemd's service supervision. With the new logic the core_pattern
37 hook only does minimal metadata collection before passing off control
38 to the new instantiated service, which is configured with a time
39 limit, a nice level and other settings to minimize negative impact on
40 the rest of the system. Also note that the new logic will honour the
41 RLIMIT_CORE setting of the crashed process, which now allows users
42 and processes to turn off coredumping for their processes by setting
45 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
46 and all forked processes by default. Previously, PID 1 would leave
47 the setting at "0" for all processes, as set by the kernel. Note that
48 the resource limit traditionally has no effect on the generated
49 coredumps on the system if the /proc/sys/kernel/core_pattern hook
50 logic is used. Since the limit is now honoured (see above) its
51 default has been changed so that the coredumping logic is enabled by
52 default for all processes, while allowing specific opt-out.
54 * When the stacktrace is extracted from processes of system users, this
55 is now done as "systemd-coredump" user, in order to sandbox this
56 potentially security sensitive parsing operation. (Note that when
57 processing coredumps of normal users this is done under the user ID
58 of process that crashed, as before.) Packagers should take notice
59 that it is now necessary to create the "systemd-coredump" system user
60 and group at package installation time.
62 * The systemd-activate socket activation testing tool gained support
63 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
64 and --seqpacket switches. It also has been extended to support both
65 new-style and inetd-style file descriptor passing. Use the new
66 --inetd switch to request inetd-style file descriptor passing.
68 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
69 variable, which takes a boolean value. If set to false, ANSI color
70 output is disabled in the tools even when run on a terminal that
73 * The VXLAN support in networkd now supports two new settings
74 DestinationPort= and PortRange=.
76 * A new systemd.machine_id= kernel command line switch has been added,
77 that may be used to set the machine ID in /etc/machine-id if it is
78 not initialized yet. This command line option has no effect if the
79 file is already initialized.
81 * systemd-nspawn gained a new --as-pid2 switch that invokes any
82 specified command line as PID 2 rather than PID 1 in the
83 container. In this mode PID 1 will be a minimal stub init process
84 that implements the special POSIX and Linux semantics of PID 1
85 regarding signal and child process management. Note that this stub
86 init process is implemented in nspawn itself and requires no support
87 from the container image. This new logic is useful to support running
88 arbitrary command lines in the container, as normal processes are
89 generally not prepared to run as PID 1.
91 * systemd-nspawn gained a new --chdir= switch for setting the current
92 working directory for the process started in the container.
94 * "journalctl /dev/sda" will now output all kernel log messages from
95 the specified device, in addition to all devices that are parents of
96 it. This should make log output about devices pretty useful, as long
97 as kernel drivers attach enough metadata to the log messages. (The
98 usual SATA drivers do.)
100 * The sd-journal API gained two new calls
101 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
102 that report whether log data from /run or /var has been found.
104 * journalctl gained a new switch "--fields" that prints all journal
105 record field names currently in use in the journal. This is backed
106 by two new sd-journal API calls sd_journal_enumerate_fields() and
107 sd_journal_restart_fields().
109 * Most configurable timeouts in systemd now expect an argument of
110 "infinity" to turn them off, instead of "0" as before. The semantics
111 from now on is that a timeout of "0" means "now", and "infinity"
112 means "never". To maintain backwards compatibility, "0" continues to
113 turn off previously existing timeout settings.
115 * "systemctl reload-or-try-restart" has been renamed to "systemctl
116 try-reload-or-restart" to clarify what it actually does: the "try"
117 logic applies to both reloading and restarting, not just restarting.
118 The old name continues to be accepted for compatibility.
120 * On boot-up, when PID 1 detects that the system clock is behind the
121 release date of the systemd version in use, the clock is now set
122 to the latter. Previously, this was already done in timesyncd, in order
123 to avoid running with clocks set to the various clock epochs such as
124 1902, 1938 or 1970. With this change the logic is now done in PID 1
125 in addition to timesyncd during early boot-up, so that it is enforced
126 before the first process is spawned by systemd. Note that the logic
127 in timesyncd remains, as it is more comprehensive and ensures
128 montonic clocks by maintaining a persistant timestamp file in
129 /var. Since /var is generally not available in earliest boot or the
130 initrd, this part of the logic remains in timesyncd, and is not done
133 * Support for tweaking details in net_cls.class_id through the
134 NetClass= configuration directive has been removed, as the kernel
135 people have decided to deprecate that controller in cgroup v2.
136 Userspace tools such as nftables are moving over to setting rules
137 that are specific to the full cgroup path of a task, which obsoletes
138 these controllers anyway. The NetClass= directive is kept around for
139 legacy compatibility reasons. For a more in-depth description of the
140 kernel change, please refer to the respective upstream commit:
142 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
144 * A new service setting RuntimeMaxSec= has been added that may be used
145 to specify a maximum runtime for a service. If the timeout is hit, the
146 service is terminated and put into a failure state.
148 * A new service setting AmbientCapabilities= has been added. It allows
149 configuration of additional Linux process capabilities that are
150 passed to the activated processes. This is only available on very
153 * The process resource limit settings in service units may now be used
154 to configure hard and soft limits individually.
156 * The various libsystemd APIs such as sd-bus or sd-event now publicly
157 expose support for gcc's __attribute__((cleanup())) C
158 extension. Specifically, for many object destructor functions
159 alternative versions whose names are suffixed with "p" have been
160 added, which take a pointer to a pointer to the object to destroy,
161 instead of just a pointer to the object itself. This is useful because
162 these destructor functions may be used directly as parameters to the
163 cleanup construct. Internally, systemd has been a heavy user of the
164 GCC extension since a long time, and with this change similar support
165 is now available to consumers of the library outside of systemd. Note
166 that by using this extension in your sources compatibility with old
167 and strictly ANSI compatible C compilers is lost. However, any gcc or
168 LLVM version of recent years have supported this extension.
170 * Timer units gained support for a new setting RandomizedDelaySec= that
171 allows configuring some additional randomized delay to the configured
172 time. This is useful to spread out timer events to avoid load peaks in
173 clusters or larger setups.
175 * Calendar time specifications now support sub-second accuracy.
177 * Socket units now support listening on SCTP and UDP-lite protocol
180 * The sd-event API now comes with a full set of man pages.
182 * Older versions of systemd contained experimental support for
183 compressing journal files and coredumps with the LZ4 compressor that
184 was not compatible with the lz4 binary (due to API limitations of the
185 lz4 library). This support has been removed; only support for files
186 compatible with the lz4 binary remains. This LZ4 logic is now
187 officially supported and no longer considered experimental.
189 * The dkr image import logic has been removed again from importd. dkr's
190 micro-services focus doesn't fit into the machine image focus of
191 importd, and quickly got out of date with the upstream dkr API.
193 * Creation of the /run/lock/lockdev/ directory was dropped from
194 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
195 been available for many years. If you still need this, you need to
196 create your own tmpfiles.d config file with:
198 d /run/lock/lockdev 0775 root lock -
200 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
201 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
202 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
203 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
204 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
205 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
206 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
207 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
208 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
209 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
210 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
211 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
212 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
213 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
214 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
215 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
216 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
217 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
218 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
220 -- Berlin, 2016-02-11
224 * A number of properties previously only settable in unit
225 files are now also available as properties to set when
226 creating transient units programmatically via the bus, as it
227 is exposed with systemd-run's --property=
228 setting. Specifically, these are: SyslogIdentifier=,
229 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
230 EnvironmentFile=, ReadWriteDirectories=,
231 ReadOnlyDirectories=, InaccessibleDirectories=,
232 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
234 * When creating transient services via the bus API it is now
235 possible to pass in a set of file descriptors to use as
236 STDIN/STDOUT/STDERR for the invoked process.
238 * Slice units may now be created transiently via the bus APIs,
239 similar to the way service and scope units may already be
242 * Wherever systemd expects a calendar timestamp specification
243 (like in journalctl's --since= and --until= switches) UTC
244 timestamps are now supported. Timestamps suffixed with "UTC"
245 are now considered to be in Universal Time Coordinated
246 instead of the local timezone. Also, timestamps may now
247 optionally be specified with sub-second accuracy. Both of
248 these additions also apply to recurring calendar event
249 specification, such as OnCalendar= in timer units.
251 * journalctl gained a new "--sync" switch that asks the
252 journal daemon to write all so far unwritten log messages to
253 disk and sync the files, before returning.
255 * systemd-tmpfiles learned two new line types "q" and "Q" that
256 operate like "v", but also set up a basic btrfs quota
257 hierarchy when used on a btrfs file system with quota
260 * tmpfiles' "v", "q" and "Q" will now create a plain directory
261 instead of a subvolume (even on a btrfs file system) if the
262 root directory is a plain directory, and not a
263 subvolume. This should simplify things with certain chroot()
264 environments which are not aware of the concept of btrfs
267 * systemd-detect-virt gained a new --chroot switch to detect
268 whether execution takes place in a chroot() environment.
270 * CPUAffinity= now takes CPU index ranges in addition to
273 * The various memory-related resource limit settings (such as
274 LimitAS=) now understand the usual K, M, G, ... suffixes to
275 the base of 1024 (IEC). Similar, the time-related resource
276 limit settings understand the usual min, h, day, ...
279 * There's a new system.conf setting DefaultTasksMax= to
280 control the default TasksMax= setting for services and
281 scopes running on the system. (TasksMax= is the primary
282 setting that exposes the "pids" cgroup controller on systemd
283 and was introduced in the previous systemd release.) The
284 setting now defaults to 512, which means services that are
285 not explicitly configured otherwise will only be able to
286 create 512 processes or threads at maximum, from this
287 version on. Note that this means that thread- or
288 process-heavy services might need to be reconfigured to set
289 TasksMax= to a higher value. It is sufficient to set
290 TasksMax= in these specific unit files to a higher value, or
291 even "infinity". Similar, there's now a logind.conf setting
292 UserTasksMax= that defaults to 4096 and limits the total
293 number of processes or tasks each user may own
294 concurrently. nspawn containers also have the TasksMax=
295 value set by default now, to 8192. Note that all of this
296 only has an effect if the "pids" cgroup controller is
297 enabled in the kernel. The general benefit of these changes
298 should be a more robust and safer system, that provides a
299 certain amount of per-service fork() bomb protection.
301 * systemd-nspawn gained the new --network-veth-extra= switch
302 to define additional and arbitrarily-named virtual Ethernet
303 links between the host and the container.
305 * A new service execution setting PassEnvironment= has been
306 added that allows importing select environment variables
307 from PID1's environment block into the environment block of
310 * Timer units gained support for a new RemainAfterElapse=
311 setting which takes a boolean argument. It defaults on on,
312 exposing behaviour unchanged to previous releases. If set to
313 off, timer units are unloaded after they elapsed if they
314 cannot elapse again. This is particularly useful for
315 transient timer units, which shall not stay around longer
316 than until they first elapse.
318 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
319 default now (the kernel default is 16). This is beneficial
320 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
321 allows substantially larger numbers of queued
322 datagrams. This should increase the capability of systemd to
323 parallelize boot-up, as logging and sd_notify() are unlikely
324 to stall execution anymore. If you need to change the value
325 from the new defaults, use the usual sysctl.d/ snippets.
327 * The compression framing format used by the journal or
328 coredump processing has changed to be in line with what the
329 official LZ4 tools generate. LZ4 compression support in
330 systemd was considered unsupported previously, as the format
331 was not compatible with the normal tools. With this release
332 this has changed now, and it is hence safe for downstream
333 distributions to turn it on. While not compressing as well
334 as the XZ, LZ4 is substantially faster, which makes
335 it a good default choice for the compression logic in the
336 journal and in coredump handling.
338 * Any reference to /etc/mtab has been dropped from
339 systemd. The file has been obsolete since a while, but
340 systemd refused to work on systems where it was incorrectly
341 set up (it should be a symlink or non-existent). Please make
342 sure to update to util-linux 2.27.1 or newer in conjunction
343 with this systemd release, which also drops any reference to
344 /etc/mtab. If you maintain a distribution make sure that no
345 software you package still references it, as this is a
346 likely source of bugs. There's also a glibc bug pending,
347 asking for removal of any reference to this obsolete file:
349 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
351 * Support for the ".snapshot" unit type has been removed. This
352 feature turned out to be little useful and little used, and
353 has now been removed from the core and from systemctl.
355 * The dependency types RequiresOverridable= and
356 RequisiteOverridable= have been removed from systemd. They
357 have been used only very sparingly to our knowledge and
358 other options that provide a similar effect (such as
359 systemctl --mode=ignore-dependencies) are much more useful
360 and commonly used. Moreover, they were only half-way
361 implemented as the option to control behaviour regarding
362 these dependencies was never added to systemctl. By removing
363 these dependency types the execution engine becomes a bit
364 simpler. Unit files that use these dependencies should be
365 changed to use the non-Overridable dependency types
366 instead. In fact, when parsing unit files with these
367 options, that's what systemd will automatically convert them
368 too, but it will also warn, asking users to fix the unit
369 files accordingly. Removal of these dependency types should
370 only affect a negligible number of unit files in the wild.
372 * Behaviour of networkd's IPForward= option changed
373 (again). It will no longer maintain a per-interface setting,
374 but propagate one way from interfaces where this is enabled
375 to the global kernel setting. The global setting will be
376 enabled when requested by a network that is set up, but
377 never be disabled again. This change was made to make sure
378 IPv4 and IPv6 behaviour regarding packet forwarding is
379 similar (as the Linux IPv6 stack does not support
380 per-interface control of this setting) and to minimize
383 * In unit files the behaviour of %u, %U, %h, %s has
384 changed. These specifiers will now unconditionally resolve
385 to the various user database fields of the user that the
386 systemd instance is running as, instead of the user
387 configured in the specific unit via User=. Note that this
388 effectively doesn't change much, as resolving of these
389 specifiers was already turned off in the --system instance
390 of systemd, as we cannot do NSS lookups from PID 1. In the
391 --user instance of systemd these specifiers where correctly
392 resolved, but hardly made any sense, since the user instance
393 lacks privileges to do user switches anyway, and User= is
394 hence useless. Morever, even in the --user instance of
395 systemd behaviour was awkward as it would only take settings
396 from User= assignment placed before the specifier into
397 account. In order to unify and simplify the logic around
398 this the specifiers will now always resolve to the
399 credentials of the user invoking the manager (which in case
400 of PID 1 is the root user).
402 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
403 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
404 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
405 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
406 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
407 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
408 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
409 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
410 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
411 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
412 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
413 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
414 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
415 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
418 -- Berlin, 2015-11-18
422 * systemd now depends on util-linux v2.27. More specifically,
423 the newly added mount monitor feature in libmount now
424 replaces systemd's former own implementation.
426 * libmount mandates /etc/mtab not to be regular file, and
427 systemd now enforces this condition at early boot.
428 /etc/mtab has been deprecated and warned about for a very
429 long time, so systems running systemd should already have
430 stopped having this file around as anything else than a
431 symlink to /proc/self/mounts.
433 * Support for the "pids" cgroup controller has been added. It
434 allows accounting the number of tasks in a cgroup and
435 enforcing limits on it. This adds two new setting
436 TasksAccounting= and TasksMax= to each unit, as well as a
437 global option DefaultTasksAccounting=.
439 * Support for the "net_cls" cgroup controller has been added.
440 It allows assigning a net class ID to each task in the
441 cgroup, which can then be used in firewall rules and traffic
442 shaping configurations. Note that the kernel netfilter net
443 class code does not currently work reliably for ingress
444 packets on unestablished sockets.
446 This adds a new config directive called NetClass= to CGroup
447 enabled units. Allowed values are positive numbers for fixed
448 assignments and "auto" for picking a free value
451 * 'systemctl is-system-running' now returns 'offline' if the
452 system is not booted with systemd. This command can now be
453 used as a substitute for 'systemd-notify --booted'.
455 * Watchdog timeouts have been increased to 3 minutes for all
456 in-tree service files. Apparently, disk IO issues are more
457 frequent than we hoped, and user reported >1 minute waiting
460 * 'machine-id-commit' functionality has been merged into
461 'machine-id-setup --commit'. The separate binary has been
464 * The WorkingDirectory= directive in unit files may now be set
465 to the special value '~'. In this case, the working
466 directory is set to the home directory of the user
469 * "machinectl shell" will now open the shell in the home
470 directory of the selected user by default.
472 * The CrashChVT= configuration file setting is renamed to
473 CrashChangeVT=, following our usual logic of not
474 abbreviating unnecessarily. The old directive is still
475 supported for compat reasons. Also, this directive now takes
476 an integer value between 1 and 63, or a boolean value. The
477 formerly supported '-1' value for disabling stays around for
480 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
481 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
482 RootDirectory= properties can now be set for transient
485 * The systemd-analyze tool gained a new "set-log-target" verb
486 to change the logging target the system manager logs to
487 dynamically during runtime. This is similar to how
488 "systemd-analyze set-log-level" already changes the log
491 * In nspawn /sys is now mounted as tmpfs, with only a selected
492 set of subdirectories mounted in from the real sysfs. This
493 enhances security slightly, and is useful for ensuring user
494 namespaces work correctly.
496 * Support for USB FunctionFS activation has been added. This
497 allows implementation of USB gadget services that are
498 activated as soon as they are requested, so that they don't
499 have to run continously, similar to classic socket
502 * The "systemctl exit" command now optionally takes an
503 additional parameter that sets the exit code to return from
504 the systemd manager when exiting. This is only relevant when
505 running the systemd user instance, or when running the
506 system instance in a container.
508 * sd-bus gained the new API calls sd_bus_path_encode_many()
509 and sd_bus_path_decode_many() that allow easy encoding and
510 decoding of multiple identifier strings inside a D-Bus
511 object path. Another new call sd_bus_default_flush_close()
512 has been added to flush and close per-thread default
515 * systemd-cgtop gained support for a -M/--machine= switch to
516 show the control groups within a certain container only.
518 * "systemctl kill" gained support for an optional --fail
519 switch. If specified the requested operation will fail of no
520 processes have been killed, because the unit had no
521 processes attached, or similar.
523 * A new systemd.crash_reboot=1 kernel command line option has
524 been added that triggers a reboot after crashing. This can
525 also be set through CrashReboot= in systemd.conf.
527 * The RuntimeDirectory= setting now understands unit
528 specifiers like %i or %f.
530 * A new (still internal) libary API sd-ipv4acd has been added,
531 that implements address conflict detection for IPv4. It's
532 based on code from sd-ipv4ll, and will be useful for
533 detecting DHCP address conflicts.
535 * File descriptors passed during socket activation may now be
536 named. A new API sd_listen_fds_with_names() is added to
537 access the names. The default names may be overridden,
538 either in the .socket file using the FileDescriptorName=
539 parameter, or by passing FDNAME= when storing the file
540 descriptors using sd_notify().
542 * systemd-networkd gained support for:
544 - Setting the IPv6 Router Advertisment settings via
545 IPv6AcceptRouterAdvertisements= in .network files.
547 - Configuring the HelloTimeSec=, MaxAgeSec= and
548 ForwardDelaySec= bridge parameters in .netdev files.
550 - Configuring PreferredSource= for static routes in
553 * The "ask-password" framework used to query for LUKS harddisk
554 passwords or SSL passwords during boot gained support for
555 caching passwords in the kernel keyring, if it is
556 available. This makes sure that the user only has to type in
557 a passphrase once if there are multiple objects to unlock
558 with the same one. Previously, such password caching was
559 available only when Plymouth was used; this moves the
560 caching logic into the systemd codebase itself. The
561 "systemd-ask-password" utility gained a new --keyname=
562 switch to control which kernel keyring key to use for
563 caching a password in. This functionality is also useful for
564 enabling display managers such as gdm to automatically
565 unlock the user's GNOME keyring if its passphrase, the
566 user's password and the harddisk password are the same, if
567 gdm-autologin is used.
569 * When downloading tar or raw images using "machinectl
570 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
571 file is now also downloaded, if it is available and stored
572 next to the image file.
574 * Units of type ".socket" gained a new boolean setting
575 Writable= which is only useful in conjunction with
576 ListenSpecial=. If true, enables opening the specified
577 special file in O_RDWR mode rather than O_RDONLY mode.
579 * systemd-rfkill has been reworked to become a singleton
580 service that is activated through /dev/rfkill on each rfkill
581 state change and saves the settings to disk. This way,
582 systemd-rfkill is now compatible with devices that exist
583 only intermittendly, and even restores state if the previous
584 system shutdown was abrupt rather than clean.
586 * The journal daemon gained support for vacuuming old journal
587 files controlled by the number of files that shall remain,
588 in addition to the already existing control by size and by
589 date. This is useful as journal interleaving performance
590 degrades with too many seperate journal files, and allows
591 putting an effective limit on them. The new setting defaults
592 to 100, but this may be changed by setting SystemMaxFiles=
593 and RuntimeMaxFiles= in journald.conf. Also, the
594 "journalctl" tool gained the new --vacuum-files= switch to
595 manually vacuum journal files to leave only the specified
596 number of files in place.
598 * udev will now create /dev/disk/by-path links for ATA devices
599 on kernels where that is supported.
601 * Galician, Serbian, Turkish and Korean translations were added.
603 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
604 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
605 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
606 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
607 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
608 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
609 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
610 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
611 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
612 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
613 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
614 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
615 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
616 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
617 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
618 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
619 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
620 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
622 -- Berlin, 2015-10-07
626 * The DHCP implementation of systemd-networkd gained a set of
629 - The DHCP server now supports emitting DNS and NTP
630 information. It may be enabled and configured via
631 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
632 and NTP information is enabled, but no servers are
633 configured, the corresponding uplink information (if there
634 is any) is propagated.
636 - Server and client now support transmission and reception
637 of timezone information. It can be configured via the
638 newly introduced network options UseTimezone=,
639 EmitTimezone=, and Timezone=. Transmission of timezone
640 information is enabled between host and containers by
641 default now: the container will change its local timezone
642 to what the host has set.
644 - Lease timeouts can now be configured via
645 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
647 - The DHCP server improved on the stability of
648 leases. Clients are more likely to get the same lease
649 information back, even if the server loses state.
651 - The DHCP server supports two new configuration options to
652 control the lease address pool metrics, PoolOffset= and
655 * The encapsulation limit of tunnels in systemd-networkd may
656 now be configured via 'EncapsulationLimit='. It allows
657 modifying the maximum additional levels of encapsulation
658 that are permitted to be prepended to a packet.
660 * systemd now supports the concept of user buses replacing
661 session buses, if used with dbus-1.10 (and enabled via dbus
662 --enable-user-session). It previously only supported this on
663 kdbus-enabled systems, and this release expands this to
664 'dbus-daemon' systems.
666 * systemd-networkd now supports predictable interface names
669 * systemd now optionally supports the new Linux kernel
670 "unified" control group hierarchy. If enabled via the kernel
671 command-line option 'systemd.unified_cgroup_hierarchy=1',
672 systemd will try to mount the unified cgroup hierarchy
673 directly on /sys/fs/cgroup. If not enabled, or not
674 available, systemd will fall back to the legacy cgroup
675 hierarchy setup, as before. Host system and containers can
676 mix and match legacy and unified hierarchies as they
677 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
678 environment variable to individually select the hierarchy to
679 use for executed containers. By default, nspawn will use the
680 unified hierarchy for the containers if the host uses the
681 unified hierarchy, and the legacy hierarchy otherwise.
682 Please note that at this point the unified hierarchy is an
683 experimental kernel feature and is likely to change in one
684 of the next kernel releases. Therefore, it should not be
685 enabled by default in downstream distributions yet. The
686 minimum required kernel version for the unified hierarchy to
687 work is 4.2. Note that when the unified hierarchy is used
688 for the first time delegated access to controllers is
689 safe. Because of this systemd-nspawn containers will get
690 access to controllers now, as will systemd user
691 sessions. This means containers and user sessions may now
692 manage their own resources, partitioning up what the system
695 * A new special scope unit "init.scope" has been introduced
696 that encapsulates PID 1 of the system. It may be used to
697 determine resource usage and enforce resource limits on PID
698 1 itself. PID 1 hence moved out of the root of the control
701 * The cgtop tool gained support for filtering out kernel
702 threads when counting tasks in a control group. Also, the
703 count of processes is now recursively summed up by
704 default. Two options -k and --recursive= have been added to
705 revert to old behaviour. The tool has also been updated to
706 work correctly in containers now.
708 * systemd-nspawn's --bind= and --bind-ro= options have been
709 extended to allow creation of non-recursive bind mounts.
711 * libsystemd gained two new calls sd_pid_get_cgroup() and
712 sd_peer_get_cgroup() which return the control group path of
713 a process or peer of a connected AF_UNIX socket. This
714 function call is particularly useful when implementing
715 delegated subtrees support in the control group hierarchy.
717 * The "sd-event" event loop API of libsystemd now supports
718 correct dequeuing of real-time signals, without losing
721 * When systemd requests a PolicyKit decision when managing
722 units it will now add additional fields to the request,
723 including unit name and desired operation. This enables more
724 powerful PolicyKit policies, that make decisions depending
727 * nspawn learnt support for .nspawn settings files, that may
728 accompany the image files or directories of containers, and
729 may contain additional settings for the container. This is
730 an alternative to configuring container parameters via the
733 Contributions from: Cristian Rodríguez, Daniel Mack, David
734 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
735 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
736 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
737 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
738 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
739 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
740 Andersen, Tom Gundersen, Torstein Husebø
742 -- Berlin, 2015-09-08
746 * machinectl gained a new verb 'shell' which opens a fresh
747 shell on the target container or the host. It is similar to
748 the existing 'login' command of machinectl, but spawns the
749 shell directly without prompting for username or
750 password. The pseudo machine '.host' now refers to the local
751 host and is used by default. Hence, 'machinectl shell' can
752 be used as replacement for 'su -' which spawns a session as
753 a fresh systemd unit in a way that is fully isolated from
754 the originating session.
756 * systemd-networkd learned to cope with private-zone DHCP
757 options and allows other programs to query the values.
759 * SELinux access control when enabling/disabling units is no
760 longer enforced with this release. The previous
761 implementation was incorrect, and a new corrected
762 implementation is not yet available. As unit file operations
763 are still protected via PolicyKit and D-Bus policy this is
764 not a security problem. Yet, distributions which care about
765 optimal SELinux support should probably not stabilize on
768 * sd-bus gained support for matches of type "arg0has=", that
769 test for membership of strings in string arrays sent in bus
772 * systemd-resolved now dumps the contents of its DNS and LLMNR
773 caches to the logs on reception of the SIGUSR1 signal. This
774 is useful to debug DNS behaviour.
776 * The coredumpctl tool gained a new --directory= option to
777 operate on journal files in a specific directory.
779 * "systemctl reboot" and related commands gained a new
780 "--message=" option which may be used to set a free-text
781 wall message when shutting down or rebooting the
782 system. This message is also logged, which is useful for
783 figuring out the reason for a reboot or shutdown a
786 * The "systemd-resolve-host" tool's -i switch now takes
787 network interface numbers as alternative to interface names.
789 * A new unit file setting for services has been introduced:
790 UtmpMode= allows configuration of how precisely systemd
791 handles utmp and wtmp entries for the service if this is
792 enabled. This allows writing services that appear similar to
793 user sessions in the output of the "w", "who", "last" and
796 * systemd-resolved will now locally synthesize DNS resource
797 records for the "localhost" and "gateway" domains as well as
798 the local hostname. This should ensure that clients querying
799 RRs via resolved will get similar results as those going via
800 NSS, if nss-myhostname is enabled.
802 Contributions from: Alastair Hughes, Alex Crawford, Daniel
803 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
804 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
805 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
806 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
807 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
808 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
809 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
810 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
811 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
812 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
813 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
815 -- Berlin, 2015-08-27
819 * The systemd-efi-boot-generator functionality was merged into
820 systemd-gpt-auto-generator.
822 * systemd-networkd now supports Group Policy for vxlan
823 devices. It can be enabled via the new boolean configuration
824 option called 'GroupPolicyExtension='.
826 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
827 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
828 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
830 -- Berlin, 2015-07-31
834 * The python-systemd code has been removed from the systemd repository.
835 A new repository has been created which accommodates the code from
836 now on, and we kindly ask distributions to create a separate package
837 for this: https://github.com/systemd/python-systemd
839 * The systemd daemon will now reload its main configuration
840 (/etc/systemd/system.conf) on daemon-reload.
842 * sd-dhcp now exposes vendor specific extensions via
843 sd_dhcp_lease_get_vendor_specific().
845 * systemd-networkd gained a number of new configuration options.
847 - A new boolean configuration option for TAP devices called
848 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
849 device, thus allowing to send and receive GSO packets.
851 - A new tunnel configuration option called 'CopyDSCP='.
852 If enabled, the DSCP field of ip6 tunnels is copied into the
855 - A set of boolean bridge configuration options were added.
856 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
857 and 'UnicastFlood=' are now parsed by networkd and applied to the
858 respective bridge link device via the respective IFLA_BRPORT_*
861 - A new string configuration option to override the hostname sent
862 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
863 is true, networkd will use the configured hostname instead of the
864 system hostname when sending DHCP requests.
866 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
867 networkd will configure the IPv6 flow-label of the tunnel device
868 according to RFC2460.
870 - The 'macvtap' virtual network devices are now supported, similar to
871 the already supported 'macvlan' devices.
873 * systemd-resolved now implements RFC5452 to improve resilience against
874 cache poisoning. Additionally, source port randomization is enabled
875 by default to further protect against DNS spoofing attacks.
877 * nss-mymachines now supports translating UIDs and GIDs of running
878 containers with user-namespaces enabled. If a container 'foo'
879 translates a host uid 'UID' to the container uid 'TUID', then
880 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
881 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
882 mapped as 'vg-foo-TGID'.
884 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
885 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
886 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
887 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
888 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
889 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
890 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
891 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
892 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
893 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
895 -- Berlin, 2015-07-29
899 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
900 There are no known issues with current sysfs, and udev does not need
901 or should be used to work around such bugs.
903 * udev does no longer enable USB HID power management. Several reports
904 indicate, that some devices cannot handle that setting.
906 * The udev accelerometer helper was removed. The functionality
907 is now fully included in iio-sensor-proxy. But this means,
908 older iio-sensor-proxy versions will no longer provide
909 accelerometer/orientation data with this systemd version.
910 Please upgrade iio-sensor-proxy to version 1.0.
912 * networkd gained a new configuration option IPv6PrivacyExtensions=
913 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
914 for Stateless Address") on selected networks.
916 * For the sake of fewer build-time dependencies and less code in the
917 main repository, the python bindings are about to be removed in the
918 next release. A new repository has been created which accommodates
919 the code from now on, and we kindly ask distributions to create a
920 separate package for this. The removal will take place in v223.
922 https://github.com/systemd/python-systemd
924 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
925 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
926 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
927 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
928 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
929 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
930 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
931 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
932 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
933 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
935 -- Berlin, 2015-07-07
939 * The sd-bus.h and sd-event.h APIs have now been declared
940 stable and have been added to the official interface of
941 libsystemd.so. sd-bus implements an alternative D-Bus client
942 library, that is relatively easy to use, very efficient and
943 supports both classic D-Bus as well as kdbus as transport
944 backend. sd-event is a generic event loop abstraction that
945 is built around Linux epoll, but adds features such as event
946 prioritization or efficient timer handling. Both APIs are good
947 choices for C programs looking for a bus and/or event loop
948 implementation that is minimal and does not have to be
949 portable to other kernels.
951 * kdbus support is no longer compile-time optional. It is now
952 always built-in. However, it can still be disabled at
953 runtime using the kdbus=0 kernel command line setting, and
954 that setting may be changed to default to off, by specifying
955 --disable-kdbus at build-time. Note though that the kernel
956 command line setting has no effect if the kdbus.ko kernel
957 module is not installed, in which case kdbus is (obviously)
958 also disabled. We encourage all downstream distributions to
959 begin testing kdbus by adding it to the kernel images in the
960 development distributions, and leaving kdbus support in
963 * The minimal required util-linux version has been bumped to
966 * Support for chkconfig (--enable-chkconfig) was removed in
967 favor of calling an abstraction tool
968 /lib/systemd/systemd-sysv-install. This needs to be
969 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
970 in README for details.
972 * If there's a systemd unit and a SysV init script for the
973 same service name, and the user executes "systemctl enable"
974 for it (or a related call), then this will now enable both
975 (or execute the related operation on both), not just the
978 * The libudev API documentation has been converted from gtkdoc
981 * gudev has been removed from the systemd tree, it is now an
984 * The systemd-cgtop tool learnt a new --raw switch to generate
985 "raw" (machine parsable) output.
987 * networkd's IPForwarding= .network file setting learnt the
988 new setting "kernel", which ensures that networkd does not
989 change the IP forwarding sysctl from the default kernel
992 * The systemd-logind bus API now exposes a new boolean
993 property "Docked" that reports whether logind considers the
994 system "docked", i.e. connected to a docking station or not.
996 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
997 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
998 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
999 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1000 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1001 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1002 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1003 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1004 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1005 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1006 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1007 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1008 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1009 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1010 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1011 Fink, Zbigniew Jędrzejewski-Szmek
1013 -- Berlin, 2015-06-19
1017 * The gudev library has been extracted into a separate repository
1018 available at: https://git.gnome.org/browse/libgudev/
1019 It is now managed as part of the Gnome project. Distributions
1020 are recommended to pass --disable-gudev to systemd and use
1021 gudev from the Gnome project instead. gudev is still included
1022 in systemd, for now. It will be removed soon, though. Please
1023 also see the announcement-thread on systemd-devel:
1024 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1026 * systemd now exposes a CPUUsageNSec= property for each
1027 service unit on the bus, that contains the overall consumed
1028 CPU time of a service (the sum of what each process of the
1029 service consumed). This value is only available if
1030 CPUAccounting= is turned on for a service, and is then shown
1031 in the "systemctl status" output.
1033 * Support for configuring alternative mappings of the old SysV
1034 runlevels to systemd targets has been removed. They are now
1035 hardcoded in a way that runlevels 2, 3, 4 all map to
1036 multi-user.target and 5 to graphical.target (which
1037 previously was already the default behaviour).
1039 * The auto-mounter logic gained support for mount point
1040 expiry, using a new TimeoutIdleSec= setting in .automount
1041 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1043 * The EFI System Partition (ESP) as mounted to /boot by
1044 systemd-efi-boot-generator will now be unmounted
1045 automatically after 2 minutes of not being used. This should
1046 minimize the risk of ESP corruptions.
1048 * New /etc/fstab options x-systemd.requires= and
1049 x-systemd.requires-mounts-for= are now supported to express
1050 additional dependencies for mounts. This is useful for
1051 journalling file systems that support external journal
1052 devices or overlay file systems that require underlying file
1053 systems to be mounted.
1055 * systemd does not support direct live-upgrades (via systemctl
1056 daemon-reexec) from versions older than v44 anymore. As no
1057 distribution we are aware of shipped such old versions in a
1058 stable release this should not be problematic.
1060 * When systemd forks off a new per-connection service instance
1061 it will now set the $REMOTE_ADDR environment variable to the
1062 remote IP address, and $REMOTE_PORT environment variable to
1063 the remote IP port. This behaviour is similar to the
1064 corresponding environment variables defined by CGI.
1066 * systemd-networkd gained support for uplink failure
1067 detection. The BindCarrier= option allows binding interface
1068 configuration dynamically to the link sense of other
1069 interfaces. This is useful to achieve behaviour like in
1072 * systemd-networkd gained support for configuring the DHCP
1073 client identifier to use when requesting leases.
1075 * systemd-networkd now has a per-network UseNTP= option to
1076 configure whether NTP server information acquired via DHCP
1077 is passed on to services like systemd-timesyncd.
1079 * systemd-networkd gained support for vti6 tunnels.
1081 * Note that systemd-networkd manages the sysctl variable
1082 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1083 it is configured for since v219. The variable controls IP
1084 forwarding, and is a per-interface alternative to the global
1085 /proc/sys/net/ipv[46]/ip_forward. This setting is
1086 configurable in the IPForward= option, which defaults to
1087 "no". This means if networkd is used for an interface it is
1088 no longer sufficient to set the global sysctl option to turn
1089 on IP forwarding! Instead, the .network file option
1090 IPForward= needs to be turned on! Note that the
1091 implementation of this behaviour was broken in v219 and has
1094 * Many bonding and vxlan options are now configurable in
1097 * systemd-nspawn gained a new --property= setting to set unit
1098 properties for the container scope. This is useful for
1099 setting resource parameters (e.g "CPUShares=500") on
1100 containers started from the command line.
1102 * systemd-nspawn gained a new --private-users= switch to make
1103 use of user namespacing available on recent Linux kernels.
1105 * systemd-nspawn may now be called as part of a shell pipeline
1106 in which case the pipes used for stdin and stdout are passed
1107 directly to the process invoked in the container, without
1108 indirection via a pseudo tty.
1110 * systemd-nspawn gained a new switch to control the UNIX
1111 signal to use when killing the init process of the container
1114 * systemd-nspawn gained a new --overlay= switch for mounting
1115 overlay file systems into the container using the new kernel
1118 * When a container image is imported via systemd-importd and
1119 the host file system is not btrfs, a loopback block device
1120 file is created in /var/lib/machines.raw with a btrfs file
1121 system inside. It is then mounted to /var/lib/machines to
1122 enable btrfs features for container management. The loopback
1123 file and btrfs file system is grown as needed when container
1124 images are imported via systemd-importd.
1126 * systemd-machined/systemd-importd gained support for btrfs
1127 quota, to enforce container disk space limits on disk. This
1128 is exposed in "machinectl set-limit".
1130 * systemd-importd now can import containers from local .tar,
1131 .raw and .qcow2 images, and export them to .tar and .raw. It
1132 can also import dkr v2 images now from the network (on top
1135 * systemd-importd gained support for verifying downloaded
1136 images with gpg2 (previously only gpg1 was supported).
1138 * systemd-machined, systemd-logind, systemd: most bus calls
1139 are now accessible to unprivileged processes via
1140 PolicyKit. Also, systemd-logind will now allow users to kill
1141 their own sessions without further privileges or
1144 * systemd-shutdownd has been removed. This service was
1145 previously responsible for implementing scheduled shutdowns
1146 as exposed in /usr/bin/shutdown's time parameter. This
1147 functionality has now been moved into systemd-logind and is
1148 accessible via a bus interface.
1150 * "systemctl reboot" gained a new switch --firmware-setup that
1151 can be used to reboot into the EFI firmware setup, if that
1152 is available. systemd-logind now exposes an API on the bus
1153 to trigger such reboots, in case graphical desktop UIs want
1154 to cover this functionality.
1156 * "systemctl enable", "systemctl disable" and "systemctl mask"
1157 now support a new "--now" switch. If specified the units
1158 that are enabled will also be started, and the ones
1159 disabled/masked also stopped.
1161 * The Gummiboot EFI boot loader tool has been merged into
1162 systemd, and renamed to "systemd-boot". The bootctl tool has been
1163 updated to support systemd-boot.
1165 * An EFI kernel stub has been added that may be used to create
1166 kernel EFI binaries that contain not only the actual kernel,
1167 but also an initrd, boot splash, command line and OS release
1168 information. This combined binary can then be signed as a
1169 single image, so that the firmware can verify it all in one
1170 step. systemd-boot has special support for EFI binaries created
1171 like this and can extract OS release information from them
1172 and show them in the boot menu. This functionality is useful
1173 to implement cryptographically verified boot schemes.
1175 * Optional support has been added to systemd-fsck to pass
1176 fsck's progress report to an AF_UNIX socket in the file
1179 * udev will no longer create device symlinks for all block
1180 devices by default. A blacklist for excluding special block
1181 devices from this logic has been turned into a whitelist
1182 that requires picking block devices explicitly that require
1185 * A new (currently still internal) API sd-device.h has been
1186 added to libsystemd. This modernized API is supposed to
1187 replace libudev eventually. In fact, already much of libudev
1188 is now just a wrapper around sd-device.h.
1190 * A new hwdb database for storing metadata about pointing
1191 stick devices has been added.
1193 * systemd-tmpfiles gained support for setting file attributes
1194 similar to the "chattr" tool with new 'h' and 'H' lines.
1196 * systemd-journald will no longer unconditionally set the
1197 btrfs NOCOW flag on new journal files. This is instead done
1198 with tmpfiles snippet using the new 'h' line type. This
1199 allows easy disabling of this logic, by masking the
1200 journal-nocow.conf tmpfiles file.
1202 * systemd-journald will now translate audit message types to
1203 human readable identifiers when writing them to the
1204 journal. This should improve readability of audit messages.
1206 * The LUKS logic gained support for the offset= and skip=
1207 options in /etc/crypttab, as previously implemented by
1210 * /usr/lib/os-release gained a new optional field VARIANT= for
1211 distributions that support multiple variants (such as a
1212 desktop edition, a server edition, ...)
1214 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1215 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1216 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1217 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1218 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1219 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1220 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1221 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1222 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1223 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1224 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1225 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1226 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1227 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1228 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1229 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1230 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1231 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1232 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1233 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1234 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1235 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1236 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1237 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1238 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1239 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1240 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1242 -- Berlin, 2015-05-22
1246 * Introduce a new API "sd-hwdb.h" for querying the hardware
1247 metadata database. With this minimal interface one can query
1248 and enumerate the udev hwdb, decoupled from the old libudev
1249 library. libudev's interface for this is now only a wrapper
1250 around sd-hwdb. A new tool systemd-hwdb has been added to
1251 interface with and update the database.
1253 * When any of systemd's tools copies files (for example due to
1254 tmpfiles' C lines) a btrfs reflink will attempted first,
1255 before bytewise copying is done.
1257 * systemd-nspawn gained a new --ephemeral switch. When
1258 specified a btrfs snapshot is taken of the container's root
1259 directory, and immediately removed when the container
1260 terminates again. Thus, a container can be started whose
1261 changes never alter the container's root directory, and are
1262 lost on container termination. This switch can also be used
1263 for starting a container off the root file system of the
1264 host without affecting the host OS. This switch is only
1265 available on btrfs file systems.
1267 * systemd-nspawn gained a new --template= switch. It takes the
1268 path to a container tree to use as template for the tree
1269 specified via --directory=, should that directory be
1270 missing. This allows instantiating containers dynamically,
1271 on first run. This switch is only available on btrfs file
1274 * When a .mount unit refers to a mount point on which multiple
1275 mounts are stacked, and the .mount unit is stopped all of
1276 the stacked mount points will now be unmounted until no
1277 mount point remains.
1279 * systemd now has an explicit notion of supported and
1280 unsupported unit types. Jobs enqueued for unsupported unit
1281 types will now fail with an "unsupported" error code. More
1282 specifically .swap, .automount and .device units are not
1283 supported in containers, .busname units are not supported on
1284 non-kdbus systems. .swap and .automount are also not
1285 supported if their respective kernel compile time options
1288 * machinectl gained support for two new "copy-from" and
1289 "copy-to" commands for copying files from a running
1290 container to the host or vice versa.
1292 * machinectl gained support for a new "bind" command to bind
1293 mount host directories into local containers. This is
1294 currently only supported for nspawn containers.
1296 * networkd gained support for configuring bridge forwarding
1297 database entries (fdb) from .network files.
1299 * A new tiny daemon "systemd-importd" has been added that can
1300 download container images in tar, raw, qcow2 or dkr formats,
1301 and make them available locally in /var/lib/machines, so
1302 that they can run as nspawn containers. The daemon can GPG
1303 verify the downloads (not supported for dkr, since it has no
1304 provisions for verifying downloads). It will transparently
1305 decompress bz2, xz, gzip compressed downloads if necessary,
1306 and restore sparse files on disk. The daemon uses privilege
1307 separation to ensure the actual download logic runs with
1308 fewer privileges than the daemon itself. machinectl has
1309 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1310 make the functionality of importd available to the
1311 user. With this in place the Fedora and Ubuntu "Cloud"
1312 images can be downloaded and booted as containers unmodified
1313 (the Fedora images lack the appropriate GPG signature files
1314 currently, so they cannot be verified, but this will change
1315 soon, hopefully). Note that downloading images is currently
1316 only fully supported on btrfs.
1318 * machinectl is now able to list container images found in
1319 /var/lib/machines, along with some metadata about sizes of
1320 disk and similar. If the directory is located on btrfs and
1321 quota is enabled, this includes quota display. A new command
1322 "image-status" has been added that shows additional
1323 information about images.
1325 * machinectl is now able to clone container images
1326 efficiently, if the underlying file system (btrfs) supports
1327 it, with the new "machinectl list-images" command. It also
1328 gained commands for renaming and removing images, as well as
1329 marking them read-only or read-write (supported also on
1330 legacy file systems).
1332 * networkd gained support for collecting LLDP network
1333 announcements, from hardware that supports this. This is
1334 shown in networkctl output.
1336 * systemd-run gained support for a new -t (--pty) switch for
1337 invoking a binary on a pty whose input and output is
1338 connected to the invoking terminal. This allows executing
1339 processes as system services while interactively
1340 communicating with them via the terminal. Most interestingly
1341 this is supported across container boundaries. Invoking
1342 "systemd-run -t /bin/bash" is an alternative to running a
1343 full login session, the difference being that the former
1344 will not register a session, nor go through the PAM session
1347 * tmpfiles gained support for a new "v" line type for creating
1348 btrfs subvolumes. If the underlying file system is a legacy
1349 file system, this automatically degrades to creating a
1350 normal directory. Among others /var/lib/machines is now
1351 created like this at boot, should it be missing.
1353 * The directory /var/lib/containers/ has been deprecated and
1354 been replaced by /var/lib/machines. The term "machines" has
1355 been used in the systemd context as generic term for both
1356 VMs and containers, and hence appears more appropriate for
1357 this, as the directory can also contain raw images bootable
1360 * systemd-nspawn when invoked with -M but without --directory=
1361 or --image= is now capable of searching for the container
1362 root directory, subvolume or disk image automatically, in
1363 /var/lib/machines. systemd-nspawn@.service has been updated
1364 to make use of this, thus allowing it to be used for raw
1367 * A new machines.target unit has been introduced that is
1368 supposed to group all containers/VMs invoked as services on
1369 the system. systemd-nspawn@.service has been updated to
1370 integrate with that.
1372 * machinectl gained a new "start" command, for invoking a
1373 container as a service. "machinectl start foo" is mostly
1374 equivalent to "systemctl start systemd-nspawn@foo.service",
1375 but handles escaping in a nicer way.
1377 * systemd-nspawn will now mount most of the cgroupfs tree
1378 read-only into each container, with the exception of the
1379 container's own subtree in the name=systemd hierarchy.
1381 * journald now sets the special FS_NOCOW file flag for its
1382 journal files. This should improve performance on btrfs, by
1383 avoiding heavy fragmentation when journald's write-pattern
1384 is used on COW file systems. It degrades btrfs' data
1385 integrity guarantees for the files to the same levels as for
1386 ext3/ext4 however. This should be OK though as journald does
1387 its own data integrity checks and all its objects are
1388 checksummed on disk. Also, journald should handle btrfs disk
1389 full events a lot more gracefully now, by processing SIGBUS
1390 errors, and not relying on fallocate() anymore.
1392 * When journald detects that journal files it is writing to
1393 have been deleted it will immediately start new journal
1396 * systemd now provides a way to store file descriptors
1397 per-service in PID 1.This is useful for daemons to ensure
1398 that fds they require are not lost during a daemon
1399 restart. The fds are passed to the daemon on the next
1400 invocation in the same way socket activation fds are
1401 passed. This is now used by journald to ensure that the
1402 various sockets connected to all the system's stdout/stderr
1403 are not lost when journald is restarted. File descriptors
1404 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1405 an extension to sd_notify(). Note that a limit is enforced
1406 on the number of fds a service can store in PID 1, and it
1407 defaults to 0, so that no fds may be stored, unless this is
1408 explicitly turned on.
1410 * The default TERM variable to use for units connected to a
1411 terminal, when no other value is explicitly is set is now
1412 vt220 rather than vt102. This should be fairly safe still,
1413 but allows PgUp/PgDn work.
1415 * The /etc/crypttab option header= as known from Debian is now
1418 * "loginctl user-status" and "loginctl session-status" will
1419 now show the last 10 lines of log messages of the
1420 user/session following the status output. Similar,
1421 "machinectl status" will show the last 10 log lines
1422 associated with a virtual machine or container
1423 service. (Note that this is usually not the log messages
1424 done in the VM/container itself, but simply what the
1425 container manager logs. For nspawn this includes all console
1428 * "loginctl session-status" without further argument will now
1429 show the status of the session of the caller. Similar,
1430 "lock-session", "unlock-session", "activate",
1431 "enable-linger", "disable-linger" may now be called without
1432 session/user parameter in which case they apply to the
1433 caller's session/user.
1435 * An X11 session scriptlet is now shipped that uploads
1436 $DISPLAY and $XAUTHORITY into the environment of the systemd
1437 --user daemon if a session begins. This should improve
1438 compatibility with X11 enabled applications run as systemd
1441 * Generators are now subject to masking via /etc and /run, the
1442 same way as unit files.
1444 * networkd .network files gained support for configuring
1445 per-link IPv4/IPv6 packet forwarding as well as IPv4
1446 masquerading. This is by default turned on for veth links to
1447 containers, as registered by systemd-nspawn. This means that
1448 nspawn containers run with --network-veth will now get
1449 automatic routed access to the host's networks without any
1450 further configuration or setup, as long as networkd runs on
1453 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1454 or UDP posts of a container on the host. With this in place
1455 it is possible to run containers with private veth links
1456 (--network-veth), and have their functionality exposed on
1457 the host as if their services were running directly on the
1460 * systemd-nspawn's --network-veth switch now gained a short
1461 version "-n", since with the changes above it is now truly
1462 useful out-of-the-box. The systemd-nspawn@.service has been
1463 updated to make use of it too by default.
1465 * systemd-nspawn will now maintain a per-image R/W lock, to
1466 ensure that the same image is not started more than once
1467 writable. (It's OK to run an image multiple times
1468 simultaneously in read-only mode.)
1470 * systemd-nspawn's --image= option is now capable of
1471 dissecting and booting MBR and GPT disk images that contain
1472 only a single active Linux partition. Previously it
1473 supported only GPT disk images with proper GPT type
1474 IDs. This allows running cloud images from major
1475 distributions directly with systemd-nspawn, without
1478 * In addition to collecting mouse dpi data in the udev
1479 hardware database, there's now support for collecting angle
1480 information for mouse scroll wheels. The database is
1481 supposed to guarantee similar scrolling behavior on mice
1482 that it knows about. There's also support for collecting
1483 information about Touchpad types.
1485 * udev's input_id built-in will now also collect touch screen
1486 dimension data and attach it to probed devices.
1488 * /etc/os-release gained support for a Distribution Privacy
1491 * networkd gained support for creating "ipvlan", "gretap",
1492 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1494 * systemd-tmpfiles gained support for "a" lines for setting
1497 * systemd-nspawn will now mount /tmp in the container to
1498 tmpfs, automatically.
1500 * systemd now exposes the memory.usage_in_bytes cgroup
1501 attribute and shows it for each service in the "systemctl
1502 status" output, if available.
1504 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1505 immediate reboot is triggered. This useful if shutdown is
1506 hung and is unable to complete, to expedite the
1507 operation. Note that this kind of reboot will still unmount
1508 all file systems, and hence should not result in fsck being
1511 * A .device unit for an optical block device will now be
1512 considered active only when a medium is in the drive. Also,
1513 mount units are now bound to their backing devices thus
1514 triggering automatic unmounting when devices become
1515 unavailable. With this in place systemd will now
1516 automatically unmount left-over mounts when a CD-ROM is
1517 ejected or an USB stick is yanked from the system.
1519 * networkd-wait-online now has support for waiting for
1520 specific interfaces only (with globbing), and for giving up
1521 after a configurable timeout.
1523 * networkd now exits when idle. It will be automatically
1524 restarted as soon as interfaces show up, are removed or
1525 change state. networkd will stay around as long as there is
1526 at least one DHCP state machine or similar around, that keep
1529 * networkd may now configure IPv6 link-local addressing in
1530 addition to IPv4 link-local addressing.
1532 * The IPv6 "token" for use in SLAAC may now be configured for
1533 each .network interface in networkd.
1535 * Routes configured with networkd may now be assigned a scope
1538 * networkd's [Match] sections now support globbing and lists
1539 of multiple space-separated matches per item.
1541 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1542 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1543 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1544 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1545 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1546 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1547 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1548 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1549 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1550 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1551 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1552 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1553 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1554 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1555 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1556 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1557 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1558 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1559 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1560 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1561 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1562 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1563 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1564 Hoffmann, Zbigniew Jędrzejewski-Szmek
1566 -- Berlin, 2015-02-16
1570 * When querying unit file enablement status (for example via
1571 "systemctl is-enabled"), a new state "indirect" is now known
1572 which indicates that a unit might not be enabled itself, but
1573 another unit listed in its Also= setting might be.
1575 * Similar to the various existing ConditionXYZ= settings for
1576 units, there are now matching AssertXYZ= settings. While
1577 failing conditions cause a unit to be skipped, but its job
1578 to succeed, failing assertions declared like this will cause
1579 a unit start operation and its job to fail.
1581 * hostnamed now knows a new chassis type "embedded".
1583 * systemctl gained a new "edit" command. When used on a unit
1584 file, this allows extending unit files with .d/ drop-in
1585 configuration snippets or editing the full file (after
1586 copying it from /usr/lib to /etc). This will invoke the
1587 user's editor (as configured with $EDITOR), and reload the
1588 modified configuration after editing.
1590 * "systemctl status" now shows the suggested enablement state
1591 for a unit, as declared in the (usually vendor-supplied)
1592 system preset files.
1594 * nss-myhostname will now resolve the single-label host name
1595 "gateway" to the locally configured default IP routing
1596 gateways, ordered by their metrics. This assigns a stable
1597 name to the used gateways, regardless which ones are
1598 currently configured. Note that the name will only be
1599 resolved after all other name sources (if nss-myhostname is
1600 configured properly) and should hence not negatively impact
1601 systems that use the single-label host name "gateway" in
1604 * systemd-inhibit now allows filtering by mode when listing
1607 * Scope and service units gained a new "Delegate" boolean
1608 property, which, when set, allows processes running inside the
1609 unit to further partition resources. This is primarily
1610 useful for systemd user instances as well as container
1613 * journald will now pick up audit messages directly from
1614 the kernel, and log them like any other log message. The
1615 audit fields are split up and fully indexed. This means that
1616 journalctl in many ways is now a (nicer!) alternative to
1617 ausearch, the traditional audit client. Note that this
1618 implements only a minimal audit client. If you want the
1619 special audit modes like reboot-on-log-overflow, please use
1620 the traditional auditd instead, which can be used in
1621 parallel to journald.
1623 * The ConditionSecurity= unit file option now understands the
1624 special string "audit" to check whether auditing is
1627 * journalctl gained two new commands --vacuum-size= and
1628 --vacuum-time= to delete old journal files until the
1629 remaining ones take up no more than the specified size on disk,
1630 or are not older than the specified time.
1632 * A new, native PPPoE library has been added to sd-network,
1633 systemd's library of light-weight networking protocols. This
1634 library will be used in a future version of networkd to
1635 enable PPPoE communication without an external pppd daemon.
1637 * The busctl tool now understands a new "capture" verb that
1638 works similar to "monitor", but writes a packet capture
1639 trace to STDOUT that can be redirected to a file which is
1640 compatible with libcap's capture file format. This can then
1641 be loaded in Wireshark and similar tools to inspect bus
1644 * The busctl tool now understands a new "tree" verb that shows
1645 the object trees of a specific service on the bus, or of all
1648 * The busctl tool now understands a new "introspect" verb that
1649 shows all interfaces and members of objects on the bus,
1650 including their signature and values. This is particularly
1651 useful to get more information about bus objects shown by
1652 the new "busctl tree" command.
1654 * The busctl tool now understands new verbs "call",
1655 "set-property" and "get-property" for invoking bus method
1656 calls, setting and getting bus object properties in a
1659 * busctl gained a new --augment-creds= argument that controls
1660 whether the tool shall augment credential information it
1661 gets from the bus with data from /proc, in a possibly
1664 * nspawn's --link-journal= switch gained two new values
1665 "try-guest" and "try-host" that work like "guest" and
1666 "host", but do not fail if the host has no persistent
1667 journalling enabled. -j is now equivalent to
1668 --link-journal=try-guest.
1670 * macvlan network devices created by nspawn will now have
1671 stable MAC addresses.
1673 * A new SmackProcessLabel= unit setting has been added, which
1674 controls the SMACK security label processes forked off by
1675 the respective unit shall use.
1677 * If compiled with --enable-xkbcommon, systemd-localed will
1678 verify x11 keymap settings by compiling the given keymap. It
1679 will spew out warnings if the compilation fails. This
1680 requires libxkbcommon to be installed.
1682 * When a coredump is collected, a larger number of metadata
1683 fields is now collected and included in the journal records
1684 created for it. More specifically, control group membership,
1685 environment variables, memory maps, working directory,
1686 chroot directory, /proc/$PID/status, and a list of open file
1687 descriptors is now stored in the log entry.
1689 * The udev hwdb now contains DPI information for mice. For
1692 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1694 * All systemd programs that read standalone configuration
1695 files in /etc now also support a corresponding series of
1696 .conf.d configuration directories in /etc/, /run/,
1697 /usr/local/lib/, /usr/lib/, and (if configured with
1698 --enable-split-usr) /lib/. In particular, the following
1699 configuration files now have corresponding configuration
1700 directories: system.conf user.conf, logind.conf,
1701 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1702 resolved.conf, timesyncd.conf, journal-remote.conf, and
1703 journal-upload.conf. Note that distributions should use the
1704 configuration directories in /usr/lib/; the directories in
1705 /etc/ are reserved for the system administrator.
1707 * systemd-rfkill will no longer take the rfkill device name
1708 into account when storing rfkill state on disk, as the name
1709 might be dynamically assigned and not stable. Instead, the
1710 ID_PATH udev variable combined with the rfkill type (wlan,
1711 bluetooth, ...) is used.
1713 * A new service systemd-machine-id-commit.service has been
1714 added. When used on systems where /etc is read-only during
1715 boot, and /etc/machine-id is not initialized (but an empty
1716 file), this service will copy the temporary machine ID
1717 created as replacement into /etc after the system is fully
1718 booted up. This is useful for systems that are freshly
1719 installed with a non-initialized machine ID, but should get
1720 a fixed machine ID for subsequent boots.
1722 * networkd's .netdev files now provide a large set of
1723 configuration parameters for VXLAN devices. Similarly, the
1724 bridge port cost parameter is now configurable in .network
1725 files. There's also new support for configuring IP source
1726 routing. networkd .link files gained support for a new
1727 OriginalName= match that is useful to match against the
1728 original interface name the kernel assigned. .network files
1729 may include MTU= and MACAddress= fields for altering the MTU
1730 and MAC address while being connected to a specific network
1733 * The LUKS logic gained supported for configuring
1734 UUID-specific key files. There's also new support for naming
1735 LUKS device from the kernel command line, using the new
1736 luks.name= argument.
1738 * Timer units may now be transiently created via the bus API
1739 (this was previously already available for scope and service
1740 units). In addition it is now possible to create multiple
1741 transient units at the same time with a single bus call. The
1742 "systemd-run" tool has been updated to make use of this for
1743 running commands on a specified time, in at(1)-style.
1745 * tmpfiles gained support for "t" lines, for assigning
1746 extended attributes to files. Among other uses this may be
1747 used to assign SMACK labels to files.
1749 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1750 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1751 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1752 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1753 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1754 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1755 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1756 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1757 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1758 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1759 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1760 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1761 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1762 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1763 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1764 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1765 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1766 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1768 -- Berlin, 2014-12-10
1772 * journalctl gained the new options -t/--identifier= to match
1773 on the syslog identifier (aka "tag"), as well as --utc to
1774 show log timestamps in the UTC timezone. journalctl now also
1775 accepts -n/--lines=all to disable line capping in a pager.
1777 * journalctl gained a new switch, --flush, that synchronously
1778 flushes logs from /run/log/journal to /var/log/journal if
1779 persistent storage is enabled. systemd-journal-flush.service
1780 now waits until the operation is complete.
1782 * Services can notify the manager before they start a reload
1783 (by sending RELOADING=1) or shutdown (by sending
1784 STOPPING=1). This allows the manager to track and show the
1785 internal state of daemons and closes a race condition when
1786 the process is still running but has closed its D-Bus
1789 * Services with Type=oneshot do not have to have any ExecStart
1792 * User units are now loaded also from
1793 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1794 /run/systemd/user directory that was already previously
1795 supported, but is under the control of the user.
1797 * Job timeouts (i.e. time-outs on the time a job that is
1798 queued stays in the run queue) can now optionally result in
1799 immediate reboot or power-off actions (JobTimeoutAction= and
1800 JobTimeoutRebootArgument=). This is useful on ".target"
1801 units, to limit the maximum time a target remains
1802 undispatched in the run queue, and to trigger an emergency
1803 operation in such a case. This is now used by default to
1804 turn off the system if boot-up (as defined by everything in
1805 basic.target) hangs and does not complete for at least
1806 15min. Also, if power-off or reboot hang for at least 30min
1807 an immediate power-off/reboot operation is triggered. This
1808 functionality is particularly useful to increase reliability
1809 on embedded devices, but also on laptops which might
1810 accidentally get powered on when carried in a backpack and
1811 whose boot stays stuck in a hard disk encryption passphrase
1814 * systemd-logind can be configured to also handle lid switch
1815 events even when the machine is docked or multiple displays
1816 are attached (HandleLidSwitchDocked= option).
1818 * A helper binary and a service have been added which can be
1819 used to resume from hibernation in the initramfs. A
1820 generator will parse the resume= option on the kernel
1821 command line to trigger resume.
1823 * A user console daemon systemd-consoled has been
1824 added. Currently, it is a preview, and will so far open a
1825 single terminal on each session of the user marked as
1826 Desktop=systemd-console.
1828 * Route metrics can be specified for DHCP routes added by
1831 * The SELinux context of socket-activated services can be set
1832 from the information provided by the networking stack
1833 (SELinuxContextFromNet= option).
1835 * Userspace firmware loading support has been removed and
1836 the minimum supported kernel version is thus bumped to 3.7.
1838 * Timeout for udev workers has been increased from 1 to 3
1839 minutes, but a warning will be printed after 1 minute to
1840 help diagnose kernel modules that take a long time to load.
1842 * Udev rules can now remove tags on devices with TAG-="foobar".
1844 * systemd's readahead implementation has been removed. In many
1845 circumstances it didn't give expected benefits even for
1846 rotational disk drives and was becoming less relevant in the
1847 age of SSDs. As none of the developers has been using
1848 rotating media anymore, and nobody stepped up to actively
1849 maintain this component of systemd it has now been removed.
1851 * Swap units can use Options= to specify discard options.
1852 Discard options specified for swaps in /etc/fstab are now
1855 * Docker containers are now detected as a separate type of
1858 * The Password Agent protocol gained support for queries where
1859 the user input is shown, useful e.g. for user names.
1860 systemd-ask-password gained a new --echo option to turn that
1863 * The default sysctl.d/ snippets will now set:
1865 net.core.default_qdisc = fq_codel
1867 This selects Fair Queuing Controlled Delay as the default
1868 queuing discipline for network interfaces. fq_codel helps
1869 fight the network bufferbloat problem. It is believed to be
1870 a good default with no tuning required for most workloads.
1871 Downstream distributions may override this choice. On 10Gbit
1872 servers that do not do forwarding, "fq" may perform better.
1873 Systems without a good clocksource should use "pfifo_fast".
1875 * If kdbus is enabled during build a new option BusPolicy= is
1876 available for service units, that allows locking all service
1877 processes into a stricter bus policy, in order to limit
1878 access to various bus services, or even hide most of them
1879 from the service's view entirely.
1881 * networkctl will now show the .network and .link file
1882 networkd has applied to a specific interface.
1884 * sd-login gained a new API call sd_session_get_desktop() to
1885 query which desktop environment has been selected for a
1888 * UNIX utmp support is now compile-time optional to support
1889 legacy-free systems.
1891 * systemctl gained two new commands "add-wants" and
1892 "add-requires" for pulling in units from specific targets
1895 * If the word "rescue" is specified on the kernel command line
1896 the system will now boot into rescue mode (aka
1897 rescue.target), which was previously available only by
1898 specifying "1" or "systemd.unit=rescue.target" on the kernel
1899 command line. This new kernel command line option nicely
1900 mirrors the already existing "emergency" kernel command line
1903 * New kernel command line options mount.usr=, mount.usrflags=,
1904 mount.usrfstype= have been added that match root=, rootflags=,
1905 rootfstype= but allow mounting a specific file system to
1908 * The $NOTIFY_SOCKET is now also passed to control processes of
1909 services, not only the main process.
1911 * This version reenables support for fsck's -l switch. This
1912 means at least version v2.25 of util-linux is required for
1913 operation, otherwise dead-locks on device nodes may
1914 occur. Again: you need to update util-linux to at least
1915 v2.25 when updating systemd to v217.
1917 * The "multi-seat-x" tool has been removed from systemd, as
1918 its functionality has been integrated into X servers 1.16,
1919 and the tool is hence redundant. It is recommended to update
1920 display managers invoking this tool to simply invoke X
1921 directly from now on, again.
1923 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1924 message flag has been added for all of systemd's PolicyKit
1925 authenticated method calls has been added. In particular
1926 this now allows optional interactive authorization via
1927 PolicyKit for many of PID1's privileged operations such as
1928 unit file enabling and disabling.
1930 * "udevadm hwdb --update" learnt a new switch "--usr" for
1931 placing the rebuilt hardware database in /usr instead of
1932 /etc. When used only hardware database entries stored in
1933 /usr will be used, and any user database entries in /etc are
1934 ignored. This functionality is useful for vendors to ship a
1935 pre-built database on systems where local configuration is
1936 unnecessary or unlikely.
1938 * Calendar time specifications in .timer units now also
1939 understand the strings "semi-annually", "quarterly" and
1940 "minutely" as shortcuts (in addition to the preexisting
1941 "anually", "hourly", ...).
1943 * systemd-tmpfiles will now correctly create files in /dev
1944 at boot which are marked for creation only at boot. It is
1945 recommended to always create static device nodes with 'c!'
1946 and 'b!', so that they are created only at boot and not
1947 overwritten at runtime.
1949 * When the watchdog logic is used for a service (WatchdogSec=)
1950 and the watchdog timeout is hit the service will now be
1951 terminated with SIGABRT (instead of just SIGTERM), in order
1952 to make sure a proper coredump and backtrace is
1953 generated. This ensures that hanging services will result in
1954 similar coredump/backtrace behaviour as services that hit a
1957 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1958 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1959 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1960 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1961 Herrmann, David Sommerseth, David Strauss, Emil Renner
1962 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1963 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1964 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1965 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1966 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1967 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1968 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1969 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1970 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1971 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1972 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1973 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1974 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1975 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1976 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1977 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1980 -- Berlin, 2014-10-28
1984 * timedated no longer reads NTP implementation unit names from
1985 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1986 implementations should add a
1988 Conflicts=systemd-timesyncd.service
1990 to their unit files to take over and replace systemd's NTP
1991 default functionality.
1993 * systemd-sysusers gained a new line type "r" for configuring
1994 which UID/GID ranges to allocate system users/groups
1995 from. Lines of type "u" may now add an additional column
1996 that specifies the home directory for the system user to be
1997 created. Also, systemd-sysusers may now optionally read user
1998 information from STDIN instead of a file. This is useful for
1999 invoking it from RPM preinst scriptlets that need to create
2000 users before the first RPM file is installed since these
2001 files might need to be owned by them. A new
2002 %sysusers_create_inline RPM macro has been introduced to do
2003 just that. systemd-sysusers now updates the shadow files as
2004 well as the user/group databases, which should enhance
2005 compatibility with certain tools like grpck.
2007 * A number of bus APIs of PID 1 now optionally consult
2008 PolicyKit to permit access for otherwise unprivileged
2009 clients under certain conditions. Note that this currently
2010 doesn't support interactive authentication yet, but this is
2011 expected to be added eventually, too.
2013 * /etc/machine-info now has new fields for configuring the
2014 deployment environment of the machine, as well as the
2015 location of the machine. hostnamectl has been updated with
2016 new command to update these fields.
2018 * systemd-timesyncd has been updated to automatically acquire
2019 NTP server information from systemd-networkd, which might
2020 have been discovered via DHCP.
2022 * systemd-resolved now includes a caching DNS stub resolver
2023 and a complete LLMNR name resolution implementation. A new
2024 NSS module "nss-resolve" has been added which can be used
2025 instead of glibc's own "nss-dns" to resolve hostnames via
2026 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2027 be resolved via systemd-resolved D-Bus APIs. In contrast to
2028 the glibc internal resolver systemd-resolved is aware of
2029 multi-homed system, and keeps DNS server and caches separate
2030 and per-interface. Queries are sent simultaneously on all
2031 interfaces that have DNS servers configured, in order to
2032 properly handle VPNs and local LANs which might resolve
2033 separate sets of domain names. systemd-resolved may acquire
2034 DNS server information from systemd-networkd automatically,
2035 which in turn might have discovered them via DHCP. A tool
2036 "systemd-resolve-host" has been added that may be used to
2037 query the DNS logic in resolved. systemd-resolved implements
2038 IDNA and automatically uses IDNA or UTF-8 encoding depending
2039 on whether classic DNS or LLMNR is used as transport. In the
2040 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2041 implementation to systemd-resolved.
2043 * A new NSS module nss-mymachines has been added, that
2044 automatically resolves the names of all local registered
2045 containers to their respective IP addresses.
2047 * A new client tool "networkctl" for systemd-networkd has been
2048 added. It currently is entirely passive and will query
2049 networking configuration from udev, rtnetlink and networkd,
2050 and present it to the user in a very friendly
2051 way. Eventually, we hope to extend it to become a full
2052 control utility for networkd.
2054 * .socket units gained a new DeferAcceptSec= setting that
2055 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2056 TCP. Similarly, support for controlling TCP keep-alive
2057 settings has been added (KeepAliveTimeSec=,
2058 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2059 turning off Nagle's algorithm on TCP has been added
2062 * logind learned a new session type "web", for use in projects
2063 like Cockpit which register web clients as PAM sessions.
2065 * timer units with at least one OnCalendar= setting will now
2066 be started only after timer-sync.target has been
2067 reached. This way they will not elapse before the system
2068 clock has been corrected by a local NTP client or
2069 similar. This is particular useful on RTC-less embedded
2070 machines, that come up with an invalid system clock.
2072 * systemd-nspawn's --network-veth= switch should now result in
2073 stable MAC addresses for both the outer and the inner side
2076 * systemd-nspawn gained a new --volatile= switch for running
2077 container instances with /etc or /var unpopulated.
2079 * The kdbus client code has been updated to use the new Linux
2080 3.17 memfd subsystem instead of the old kdbus-specific one.
2082 * systemd-networkd's DHCP client and server now support
2083 FORCERENEW. There are also new configuration options to
2084 configure the vendor client identifier and broadcast mode
2087 * systemd will no longer inform the kernel about the current
2088 timezone, as this is necessarily incorrect and racy as the
2089 kernel has no understanding of DST and similar
2090 concepts. This hence means FAT timestamps will be always
2091 considered UTC, similar to what Android is already
2092 doing. Also, when the RTC is configured to the local time
2093 (rather than UTC) systemd will never synchronize back to it,
2094 as this might confuse Windows at a later boot.
2096 * systemd-analyze gained a new command "verify" for offline
2097 validation of unit files.
2099 * systemd-networkd gained support for a couple of additional
2100 settings for bonding networking setups. Also, the metric for
2101 statically configured routes may now be configured. For
2102 network interfaces where this is appropriate the peer IP
2103 address may now be configured.
2105 * systemd-networkd's DHCP client will no longer request
2106 broadcasting by default, as this tripped up some networks.
2107 For hardware where broadcast is required the feature should
2108 be switched back on using RequestBroadcast=yes.
2110 * systemd-networkd will now set up IPv4LL addresses (when
2111 enabled) even if DHCP is configured successfully.
2113 * udev will now default to respect network device names given
2114 by the kernel when the kernel indicates that these are
2115 predictable. This behavior can be tweaked by changing
2116 NamePolicy= in the relevant .link file.
2118 * A new library systemd-terminal has been added that
2119 implements full TTY stream parsing and rendering. This
2120 library is supposed to be used later on for implementing a
2121 full userspace VT subsystem, replacing the current kernel
2124 * A new tool systemd-journal-upload has been added to push
2125 journal data to a remote system running
2126 systemd-journal-remote.
2128 * journald will no longer forward all local data to another
2129 running syslog daemon. This change has been made because
2130 rsyslog (which appears to be the most commonly used syslog
2131 implementation these days) no longer makes use of this, and
2132 instead pulls the data out of the journal on its own. Since
2133 forwarding the messages to a non-existent syslog server is
2134 more expensive than we assumed we have now turned this
2135 off. If you run a syslog server that is not a recent rsyslog
2136 version, you have to turn this option on again
2137 (ForwardToSyslog= in journald.conf).
2139 * journald now optionally supports the LZ4 compressor for
2140 larger journal fields. This compressor should perform much
2141 better than XZ which was the previous default.
2143 * machinectl now shows the IP addresses of local containers,
2144 if it knows them, plus the interface name of the container.
2146 * A new tool "systemd-escape" has been added that makes it
2147 easy to escape strings to build unit names and similar.
2149 * sd_notify() messages may now include a new ERRNO= field
2150 which is parsed and collected by systemd and shown among the
2151 "systemctl status" output for a service.
2153 * A new component "systemd-firstboot" has been added that
2154 queries the most basic systemd information (timezone,
2155 hostname, root password) interactively on first
2156 boot. Alternatively it may also be used to provision these
2157 things offline on OS images installed into directories.
2159 * The default sysctl.d/ snippets will now set
2161 net.ipv4.conf.default.promote_secondaries=1
2163 This has the benefit of no flushing secondary IP addresses
2164 when primary addresses are removed.
2166 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2167 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2168 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2169 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2170 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2171 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2172 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2173 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2174 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2175 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2176 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2177 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2178 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2179 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2180 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2182 -- Berlin, 2014-08-19
2186 * A new tool systemd-sysusers has been added. This tool
2187 creates system users and groups in /etc/passwd and
2188 /etc/group, based on static declarative system user/group
2189 definitions in /usr/lib/sysusers.d/. This is useful to
2190 enable factory resets and volatile systems that boot up with
2191 an empty /etc directory, and thus need system users and
2192 groups created during early boot. systemd now also ships
2193 with two default sysusers.d/ files for the most basic
2194 users and groups systemd and the core operating system
2197 * A new tmpfiles snippet has been added that rebuilds the
2198 essential files in /etc on boot, should they be missing.
2200 * A directive for ensuring automatic clean-up of
2201 /var/cache/man/ has been removed from the default
2202 configuration. This line should now be shipped by the man
2203 implementation. The necessary change has been made to the
2204 man-db implementation. Note that you need to update your man
2205 implementation to one that ships this line, otherwise no
2206 automatic clean-up of /var/cache/man will take place.
2208 * A new condition ConditionNeedsUpdate= has been added that
2209 may conditionalize services to only run when /etc or /var
2210 are "older" than the vendor operating system resources in
2211 /usr. This is useful for reconstructing or updating /etc
2212 after an offline update of /usr or a factory reset, on the
2213 next reboot. Services that want to run once after such an
2214 update or reset should use this condition and order
2215 themselves before the new systemd-update-done.service, which
2216 will mark the two directories as fully updated. A number of
2217 service files have been added making use of this, to rebuild
2218 the udev hardware database, the journald message catalog and
2219 dynamic loader cache (ldconfig). The systemd-sysusers tool
2220 described above also makes use of this now. With this in
2221 place it is now possible to start up a minimal operating
2222 system with /etc empty cleanly. For more information on the
2223 concepts involved see this recent blog story:
2225 http://0pointer.de/blog/projects/stateless.html
2227 * A new system group "input" has been introduced, and all
2228 input device nodes get this group assigned. This is useful
2229 for system-level software to get access to input devices. It
2230 complements what is already done for "audio" and "video".
2232 * systemd-networkd learnt minimal DHCPv4 server support in
2233 addition to the existing DHCPv4 client support. It also
2234 learnt DHCPv6 client and IPv6 Router Solicitation client
2235 support. The DHCPv4 client gained support for static routes
2236 passed in from the server. Note that the [DHCPv4] section
2237 known in older systemd-networkd versions has been renamed to
2238 [DHCP] and is now also used by the DHCPv6 client. Existing
2239 .network files using settings of this section should be
2240 updated, though compatibility is maintained. Optionally, the
2241 client hostname may now be sent to the DHCP server.
2243 * networkd gained support for vxlan virtual networks as well
2244 as tun/tap and dummy devices.
2246 * networkd gained support for automatic allocation of address
2247 ranges for interfaces from a system-wide pool of
2248 addresses. This is useful for dynamically managing a large
2249 number of interfaces with a single network configuration
2250 file. In particular this is useful to easily assign
2251 appropriate IP addresses to the veth links of a large number
2252 of nspawn instances.
2254 * RPM macros for processing sysusers, sysctl and binfmt
2255 drop-in snippets at package installation time have been
2258 * The /etc/os-release file should now be placed in
2259 /usr/lib/os-release. The old location is automatically
2260 created as symlink. /usr/lib is the more appropriate
2261 location of this file, since it shall actually describe the
2262 vendor operating system shipped in /usr, and not the
2263 configuration stored in /etc.
2265 * .mount units gained a new boolean SloppyOptions= setting
2266 that maps to mount(8)'s -s option which enables permissive
2267 parsing of unknown mount options.
2269 * tmpfiles learnt a new "L+" directive which creates a symlink
2270 but (unlike "L") deletes a pre-existing file first, should
2271 it already exist and not already be the correct
2272 symlink. Similarly, "b+", "c+" and "p+" directives have been
2273 added as well, which create block and character devices, as
2274 well as fifos in the filesystem, possibly removing any
2275 pre-existing files of different types.
2277 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2278 'argument' field (which so far specified the source to
2279 symlink/copy the files from) is now optional. If omitted the
2280 same file os copied from /usr/share/factory/ suffixed by the
2281 full destination path. This is useful for populating /etc
2282 with essential files, by copying them from vendor defaults
2283 shipped in /usr/share/factory/etc.
2285 * A new command "systemctl preset-all" has been added that
2286 applies the service preset settings to all installed unit
2287 files. A new switch --preset-mode= has been added that
2288 controls whether only enable or only disable operations
2291 * A new command "systemctl is-system-running" has been added
2292 that allows checking the overall state of the system, for
2293 example whether it is fully up and running.
2295 * When the system boots up with an empty /etc, the equivalent
2296 to "systemctl preset-all" is executed during early boot, to
2297 make sure all default services are enabled after a factory
2300 * systemd now contains a minimal preset file that enables the
2301 most basic services systemd ships by default.
2303 * Unit files' [Install] section gained a new DefaultInstance=
2304 field for defining the default instance to create if a
2305 template unit is enabled with no instance specified.
2307 * A new passive target cryptsetup-pre.target has been added
2308 that may be used by services that need to make they run and
2309 finish before the first LUKS cryptographic device is set up.
2311 * The /dev/loop-control and /dev/btrfs-control device nodes
2312 are now owned by the "disk" group by default, opening up
2313 access to this group.
2315 * systemd-coredump will now automatically generate a
2316 stack trace of all core dumps taking place on the system,
2317 based on elfutils' libdw library. This stack trace is logged
2320 * systemd-coredump may now optionally store coredumps directly
2321 on disk (in /var/lib/systemd/coredump, possibly compressed),
2322 instead of storing them unconditionally in the journal. This
2323 mode is the new default. A new configuration file
2324 /etc/systemd/coredump.conf has been added to configure this
2325 and other parameters of systemd-coredump.
2327 * coredumpctl gained a new "info" verb to show details about a
2328 specific coredump. A new switch "-1" has also been added
2329 that makes sure to only show information about the most
2330 recent entry instead of all entries. Also, as the tool is
2331 generally useful now the "systemd-" prefix of the binary
2332 name has been removed. Distributions that want to maintain
2333 compatibility with the old name should add a symlink from
2334 the old name to the new name.
2336 * journald's SplitMode= now defaults to "uid". This makes sure
2337 that unprivileged users can access their own coredumps with
2338 coredumpctl without restrictions.
2340 * New kernel command line options "systemd.wants=" (for
2341 pulling an additional unit during boot), "systemd.mask="
2342 (for masking a specific unit for the boot), and
2343 "systemd.debug-shell" (for enabling the debug shell on tty9)
2344 have been added. This is implemented in the new generator
2345 "systemd-debug-generator".
2347 * systemd-nspawn will now by default filter a couple of
2348 syscalls for containers, among them those required for
2349 kernel module loading, direct x86 IO port access, swap
2350 management, and kexec. Most importantly though
2351 open_by_handle_at() is now prohibited for containers,
2352 closing a hole similar to a recently discussed vulnerability
2353 in docker regarding access to files on file hierarchies the
2354 container should normally not have access to. Note that, for
2355 nspawn, we generally make no security claims anyway (and
2356 this is explicitly documented in the man page), so this is
2357 just a fix for one of the most obvious problems.
2359 * A new man page file-hierarchy(7) has been added that
2360 contains a minimized, modernized version of the file system
2361 layout systemd expects, similar in style to the FHS
2362 specification or hier(5). A new tool systemd-path(1) has
2363 been added to query many of these paths for the local
2366 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2367 longer done. Since the directory now has a per-user size
2368 limit, and is cleaned on logout this appears unnecessary,
2369 in particular since this now brings the lifecycle of this
2370 directory closer in line with how IPC objects are handled.
2372 * systemd.pc now exports a number of additional directories,
2373 including $libdir (which is useful to identify the library
2374 path for the primary architecture of the system), and a
2375 couple of drop-in directories.
2377 * udev's predictable network interface names now use the dev_port
2378 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2379 distinguish between ports of the same PCI function. dev_id should
2380 only be used for ports using the same HW address, hence the need
2383 * machined has been updated to export the OS version of a
2384 container (read from /etc/os-release and
2385 /usr/lib/os-release) on the bus. This is now shown in
2386 "machinectl status" for a machine.
2388 * A new service setting RestartForceExitStatus= has been
2389 added. If configured to a set of exit signals or process
2390 return values, the service will be restarted when the main
2391 daemon process exits with any of them, regardless of the
2394 * systemctl's -H switch for connecting to remote systemd
2395 machines has been extended so that it may be used to
2396 directly connect to a specific container on the
2397 host. "systemctl -H root@foobar:waldi" will now connect as
2398 user "root" to host "foobar", and then proceed directly to
2399 the container named "waldi". Note that currently you have to
2400 authenticate as user "root" for this to work, as entering
2401 containers is a privileged operation.
2403 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2404 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2405 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2406 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2407 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2408 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2409 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2410 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2411 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2412 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2413 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2414 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2416 -- Berlin, 2014-07-03
2420 * As an experimental feature, udev now tries to lock the
2421 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2422 executes events for the disk or any of its partitions.
2423 Applications like partitioning programs can lock the
2424 disk device node (flock(LOCK_EX)) and claim temporary
2425 device ownership that way; udev will entirely skip all event
2426 handling for this disk and its partitions. If the disk
2427 was opened for writing, the close will trigger a partition
2428 table rescan in udev's "watch" facility, and if needed
2429 synthesize "change" events for the disk and all its partitions.
2430 This is now unconditionally enabled, and if it turns out to
2431 cause major problems, we might turn it on only for specific
2432 devices, or might need to disable it entirely. Device Mapper
2433 devices are excluded from this logic.
2435 * We temporarily dropped the "-l" switch for fsck invocations,
2436 since they collide with the flock() logic above. util-linux
2437 upstream has been changed already to avoid this conflict,
2438 and we will readd "-l" as soon as util-linux with this
2439 change has been released.
2441 * The dependency on libattr has been removed. Since a long
2442 time, the extended attribute calls have moved to glibc, and
2443 libattr is thus unnecessary.
2445 * Virtualization detection works without priviliges now. This
2446 means the systemd-detect-virt binary no longer requires
2447 CAP_SYS_PTRACE file capabilities, and our daemons can run
2448 with fewer privileges.
2450 * systemd-networkd now runs under its own "systemd-network"
2451 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2452 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2453 loses the ability to write to files owned by root this way.
2455 * Similarly, systemd-resolved now runs under its own
2456 "systemd-resolve" user with no capabilities remaining.
2458 * Similarly, systemd-bus-proxyd now runs under its own
2459 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2461 * systemd-networkd gained support for setting up "veth"
2462 virtual Ethernet devices for container connectivity, as well
2463 as GRE and VTI tunnels.
2465 * systemd-networkd will no longer automatically attempt to
2466 manually load kernel modules necessary for certain tunnel
2467 transports. Instead, it is assumed the kernel loads them
2468 automatically when required. This only works correctly on
2469 very new kernels. On older kernels, please consider adding
2470 the kernel modules to /etc/modules-load.d/ as a work-around.
2472 * The resolv.conf file systemd-resolved generates has been
2473 moved to /run/systemd/resolve/. If you have a symlink from
2474 /etc/resolv.conf, it might be necessary to correct it.
2476 * Two new service settings, ProtectHome= and ProtectSystem=,
2477 have been added. When enabled, they will make the user data
2478 (such as /home) inaccessible or read-only and the system
2479 (such as /usr) read-only, for specific services. This allows
2480 very light-weight per-service sandboxing to avoid
2481 modifications of user data or system files from
2482 services. These two new switches have been enabled for all
2483 of systemd's long-running services, where appropriate.
2485 * Socket units gained new SocketUser= and SocketGroup=
2486 settings to set the owner user and group of AF_UNIX sockets
2487 and FIFOs in the file system.
2489 * Socket units gained a new RemoveOnStop= setting. If enabled,
2490 all FIFOS and sockets in the file system will be removed
2491 when the specific socket unit is stopped.
2493 * Socket units gained a new Symlinks= setting. It takes a list
2494 of symlinks to create to file system sockets or FIFOs
2495 created by the specific Unix sockets. This is useful to
2496 manage symlinks to socket nodes with the same life-cycle as
2499 * The /dev/log socket and /dev/initctl FIFO have been moved to
2500 /run, and have been replaced by symlinks. This allows
2501 connecting to these facilities even if PrivateDevices=yes is
2502 used for a service (which makes /dev/log itself unavailable,
2503 but /run is left). This also has the benefit of ensuring
2504 that /dev only contains device nodes, directories and
2505 symlinks, and nothing else.
2507 * sd-daemon gained two new calls sd_pid_notify() and
2508 sd_pid_notifyf(). They are similar to sd_notify() and
2509 sd_notifyf(), but allow overriding of the source PID of
2510 notification messages if permissions permit this. This is
2511 useful to send notify messages on behalf of a different
2512 process (for example, the parent process). The
2513 systemd-notify tool has been updated to make use of this
2514 when sending messages (so that notification messages now
2515 originate from the shell script invoking systemd-notify and
2516 not the systemd-notify process itself. This should minimize
2517 a race where systemd fails to associate notification
2518 messages to services when the originating process already
2521 * A new "on-abnormal" setting for Restart= has been added. If
2522 set, it will result in automatic restarts on all "abnormal"
2523 reasons for a process to exit, which includes unclean
2524 signals, core dumps, timeouts and watchdog timeouts, but
2525 does not include clean and unclean exit codes or clean
2526 signals. Restart=on-abnormal is an alternative for
2527 Restart=on-failure for services that shall be able to
2528 terminate and avoid restarts on certain errors, by
2529 indicating so with an unclean exit code. Restart=on-failure
2530 or Restart=on-abnormal is now the recommended setting for
2531 all long-running services.
2533 * If the InaccessibleDirectories= service setting points to a
2534 mount point (or if there are any submounts contained within
2535 it), it is now attempted to completely unmount it, to make
2536 the file systems truly unavailable for the respective
2539 * The ReadOnlyDirectories= service setting and
2540 systemd-nspawn's --read-only parameter are now recursively
2541 applied to all submounts, too.
2543 * Mount units may now be created transiently via the bus APIs.
2545 * The support for SysV and LSB init scripts has been removed
2546 from the systemd daemon itself. Instead, it is now
2547 implemented as a generator that creates native systemd units
2548 from these scripts when needed. This enables us to remove a
2549 substantial amount of legacy code from PID 1, following the
2550 fact that many distributions only ship a very small number
2551 of LSB/SysV init scripts nowadays.
2553 * Privileged Xen (dom0) domains are not considered
2554 virtualization anymore by the virtualization detection
2555 logic. After all, they generally have unrestricted access to
2556 the hardware and usually are used to manage the unprivileged
2559 * systemd-tmpfiles gained a new "C" line type, for copying
2560 files or entire directories.
2562 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2563 lines. So far, they have been non-globbing versions of the
2564 latter, and have thus been redundant. In future, it is
2565 recommended to only use "z". "m" has hence been removed
2566 from the documentation, even though it stays supported.
2568 * A tmpfiles snippet to recreate the most basic structure in
2569 /var has been added. This is enough to create the /var/run →
2570 /run symlink and create a couple of structural
2571 directories. This allows systems to boot up with an empty or
2572 volatile /var. Of course, while with this change, the core OS
2573 now is capable with dealing with a volatile /var, not all
2574 user services are ready for it. However, we hope that sooner
2575 or later, many service daemons will be changed upstream so
2576 that they are able to automatically create their necessary
2577 directories in /var at boot, should they be missing. This is
2578 the first step to allow state-less systems that only require
2579 the vendor image for /usr to boot.
2581 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2582 empty tmpfs instance to a specific directory. This is
2583 particularly useful for making use of the automatic
2584 reconstruction of /var (see above), by passing --tmpfs=/var.
2586 * Access modes specified in tmpfiles snippets may now be
2587 prefixed with "~", which indicates that they shall be masked
2588 by whether the existing file or directory is currently
2589 writable, readable or executable at all. Also, if specified,
2590 the sgid/suid/sticky bits will be masked for all
2593 * A new passive target unit "network-pre.target" has been
2594 added which is useful for services that shall run before any
2595 network is configured, for example firewall scripts.
2597 * The "floppy" group that previously owned the /dev/fd*
2598 devices is no longer used. The "disk" group is now used
2599 instead. Distributions should probably deprecate usage of
2602 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2603 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2604 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2605 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2606 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2607 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2610 -- Berlin, 2014-06-11
2614 * A new "systemd-timesyncd" daemon has been added for
2615 synchronizing the system clock across the network. It
2616 implements an SNTP client. In contrast to NTP
2617 implementations such as chrony or the NTP reference server,
2618 this only implements a client side, and does not bother with
2619 the full NTP complexity, focusing only on querying time from
2620 one remote server and synchronizing the local clock to
2621 it. Unless you intend to serve NTP to networked clients or
2622 want to connect to local hardware clocks, this simple NTP
2623 client should be more than appropriate for most
2624 installations. The daemon runs with minimal privileges, and
2625 has been hooked up with networkd to only operate when
2626 network connectivity is available. The daemon saves the
2627 current clock to disk every time a new NTP sync has been
2628 acquired, and uses this to possibly correct the system clock
2629 early at bootup, in order to accommodate for systems that
2630 lack an RTC such as the Raspberry Pi and embedded devices,
2631 and to make sure that time monotonically progresses on these
2632 systems, even if it is not always correct. To make use of
2633 this daemon, a new system user and group "systemd-timesync"
2634 needs to be created on installation of systemd.
2636 * The queue "seqnum" interface of libudev has been disabled, as
2637 it was generally incompatible with device namespacing as
2638 sequence numbers of devices go "missing" if the devices are
2639 part of a different namespace.
2641 * "systemctl list-timers" and "systemctl list-sockets" gained
2642 a --recursive switch for showing units of these types also
2643 for all local containers, similar in style to the already
2644 supported --recursive switch for "systemctl list-units".
2646 * A new RebootArgument= setting has been added for service
2647 units, which may be used to specify a kernel reboot argument
2648 to use when triggering reboots with StartLimitAction=.
2650 * A new FailureAction= setting has been added for service
2651 units which may be used to specify an operation to trigger
2652 when a service fails. This works similarly to
2653 StartLimitAction=, but unlike it, controls what is done
2654 immediately rather than only after several attempts to
2655 restart the service in question.
2657 * hostnamed got updated to also expose the kernel name,
2658 release, and version on the bus. This is useful for
2659 executing commands like hostnamectl with the -H switch.
2660 systemd-analyze makes use of this to properly display
2661 details when running non-locally.
2663 * The bootchart tool can now show cgroup information in the
2664 graphs it generates.
2666 * The CFS CPU quota cgroup attribute is now exposed for
2667 services. The new CPUQuota= switch has been added for this
2668 which takes a percentage value. Setting this will have the
2669 result that a service may never get more CPU time than the
2670 specified percentage, even if the machine is otherwise idle.
2672 * systemd-networkd learned IPIP and SIT tunnel support.
2674 * LSB init scripts exposing a dependency on $network will now
2675 get a dependency on network-online.target rather than simply
2676 network.target. This should bring LSB handling closer to
2677 what it was on SysV systems.
2679 * A new fsck.repair= kernel option has been added to control
2680 how fsck shall deal with unclean file systems at boot.
2682 * The (.ini) configuration file parser will now silently
2683 ignore sections whose name begins with "X-". This may be
2684 used to maintain application-specific extension sections in unit
2687 * machined gained a new API to query the IP addresses of
2688 registered containers. "machinectl status" has been updated
2689 to show these addresses in its output.
2691 * A new call sd_uid_get_display() has been added to the
2692 sd-login APIs for querying the "primary" session of a
2693 user. The "primary" session of the user is elected from the
2694 user's sessions and generally a graphical session is
2695 preferred over a text one.
2697 * A minimal systemd-resolved daemon has been added. It
2698 currently simply acts as a companion to systemd-networkd and
2699 manages resolv.conf based on per-interface DNS
2700 configuration, possibly supplied via DHCP. In the long run
2701 we hope to extend this into a local DNSSEC enabled DNS and
2704 * The systemd-networkd-wait-online tool is now enabled by
2705 default. It will delay network-online.target until a network
2706 connection has been configured. The tool primarily integrates
2707 with networkd, but will also make a best effort to make sense
2708 of network configuration performed in some other way.
2710 * Two new service options StartupCPUShares= and
2711 StartupBlockIOWeight= have been added that work similarly to
2712 CPUShares= and BlockIOWeight= however only apply during
2713 system startup. This is useful to prioritize certain services
2714 differently during bootup than during normal runtime.
2716 * hostnamed has been changed to prefer the statically
2717 configured hostname in /etc/hostname (unless set to
2718 'localhost' or empty) over any dynamic one supplied by
2719 dhcp. With this change, the rules for picking the hostname
2720 match more closely the rules of other configuration settings
2721 where the local administrator's configuration in /etc always
2722 overrides any other settings.
2724 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2725 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2726 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2727 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2728 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2729 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2730 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2731 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2732 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2733 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2734 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2735 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2736 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2737 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2738 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2739 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2742 -- Beijing, 2014-05-28
2746 * When restoring the screen brightness at boot, stay away from
2747 the darkest setting or from the lowest 5% of the available
2748 range, depending on which is the larger value of both. This
2749 should effectively protect the user from rebooting into a
2750 black screen, should the brightness have been set to minimum
2753 * sd-login gained a new sd_machine_get_class() call to
2754 determine the class ("vm" or "container") of a machine
2755 registered with machined.
2757 * sd-login gained new calls
2758 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2759 to query the identity of the peer of a local AF_UNIX
2760 connection. They operate similarly to their sd_pid_get_xyz()
2763 * PID 1 will now maintain a system-wide system state engine
2764 with the states "starting", "running", "degraded",
2765 "maintenance", "stopping". These states are bound to system
2766 startup, normal runtime, runtime with at least one failed
2767 service, rescue/emergency mode and system shutdown. This
2768 state is shown in the "systemctl status" output when no unit
2769 name is passed. It is useful to determine system state, in
2770 particularly when doing so for many systems or containers at
2773 * A new command "list-machines" has been added to "systemctl"
2774 that lists all local OS containers and shows their system
2775 state (see above), if systemd runs inside of them.
2777 * systemctl gained a new "-r" switch to recursively enumerate
2778 units on all local containers, when used with the
2779 "list-unit" command (which is the default one that is
2780 executed when no parameters are specified).
2782 * The GPT automatic partition discovery logic will now honour
2783 two GPT partition flags: one may be set on a partition to
2784 cause it to be mounted read-only, and the other may be set
2785 on a partition to ignore it during automatic discovery.
2787 * Two new GPT type UUIDs have been added for automatic root
2788 partition discovery, for 32-bit and 64-bit ARM. This is not
2789 particularly useful for discovering the root directory on
2790 these architectures during bare-metal boots (since UEFI is
2791 not common there), but still very useful to allow booting of
2792 ARM disk images in nspawn with the -i option.
2794 * MAC addresses of interfaces created with nspawn's
2795 --network-interface= switch will now be generated from the
2796 machine name, and thus be stable between multiple invocations
2799 * logind will now automatically remove all IPC objects owned
2800 by a user if she or he fully logs out. This makes sure that
2801 users who are logged out cannot continue to consume IPC
2802 resources. This covers SysV memory, semaphores and message
2803 queues as well as POSIX shared memory and message
2804 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2805 limits. With this functionality, that is corrected. This may
2806 be turned off by using the RemoveIPC= switch of logind.conf.
2808 * The systemd-machine-id-setup and tmpfiles tools gained a
2809 --root= switch to operate on a specific root directory,
2812 * journald can now forward logged messages to the TTYs of all
2813 logged in users ("wall"). This is the default for all
2814 emergency messages now.
2816 * A new tool systemd-journal-remote has been added to stream
2817 journal log messages across the network.
2819 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2820 controller trees are mounted into it. Note that the
2821 directories mounted beneath it are not read-only. This is a
2822 security measure and is particularly useful because glibc
2823 actually includes a search logic to pick any tmpfs it can
2824 find to implement shm_open() if /dev/shm is not available
2825 (which it might very well be in namespaced setups).
2827 * machinectl gained a new "poweroff" command to cleanly power
2828 down a local OS container.
2830 * The PrivateDevices= unit file setting will now also drop the
2831 CAP_MKNOD capability from the capability bound set, and
2832 imply DevicePolicy=closed.
2834 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2835 comprehensively on all long-running systemd services where
2836 this is appropriate.
2838 * systemd-udevd will now run in a disassociated mount
2839 namespace. To mount directories from udev rules, make sure to
2840 pull in mount units via SYSTEMD_WANTS properties.
2842 * The kdbus support gained support for uploading policy into
2843 the kernel. sd-bus gained support for creating "monitoring"
2844 connections that can eavesdrop into all bus communication
2845 for debugging purposes.
2847 * Timestamps may now be specified in seconds since the UNIX
2848 epoch Jan 1st, 1970 by specifying "@" followed by the value
2851 * Native tcpwrap support in systemd has been removed. tcpwrap
2852 is old code, not really maintained anymore and has serious
2853 shortcomings, and better options such as firewalls
2854 exist. For setups that require tcpwrap usage, please
2855 consider invoking your socket-activated service via tcpd,
2856 like on traditional inetd.
2858 * A new system.conf configuration option
2859 DefaultTimerAccuracySec= has been added that controls the
2860 default AccuracySec= setting of .timer units.
2862 * Timer units gained a new WakeSystem= switch. If enabled,
2863 timers configured this way will cause the system to resume
2864 from system suspend (if the system supports that, which most
2867 * Timer units gained a new Persistent= switch. If enabled,
2868 timers configured this way will save to disk when they have
2869 been last triggered. This information is then used on next
2870 reboot to possible execute overdue timer events, that
2871 could not take place because the system was powered off.
2872 This enables simple anacron-like behaviour for timer units.
2874 * systemctl's "list-timers" will now also list the time a
2875 timer unit was last triggered in addition to the next time
2876 it will be triggered.
2878 * systemd-networkd will now assign predictable IPv4LL
2879 addresses to its local interfaces.
2881 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2882 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2883 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2884 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2885 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2886 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2887 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2888 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2891 -- Berlin, 2014-03-25
2895 * A new unit file setting RestrictAddressFamilies= has been
2896 added to restrict which socket address families unit
2897 processes gain access to. This takes address family names
2898 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2899 attack surface of services via exotic protocol stacks. This
2900 is built on seccomp system call filters.
2902 * Two new unit file settings RuntimeDirectory= and
2903 RuntimeDirectoryMode= have been added that may be used to
2904 manage a per-daemon runtime directories below /run. This is
2905 an alternative for setting up directory permissions with
2906 tmpfiles snippets, and has the advantage that the runtime
2907 directory's lifetime is bound to the daemon runtime and that
2908 the daemon starts up with an empty directory each time. This
2909 is particularly useful when writing services that drop
2910 privileges using the User= or Group= setting.
2912 * The DeviceAllow= unit setting now supports globbing for
2913 matching against device group names.
2915 * The systemd configuration file system.conf gained new
2916 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2917 DefaultMemoryAccounting= to globally turn on/off accounting
2918 for specific resources (cgroups) for all units. These
2919 settings may still be overridden individually in each unit
2922 * systemd-gpt-auto-generator is now able to discover /srv and
2923 root partitions in addition to /home and swap partitions. It
2924 also supports LUKS-encrypted partitions now. With this in
2925 place, automatic discovery of partitions to mount following
2926 the Discoverable Partitions Specification
2927 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2928 is now a lot more complete. This allows booting without
2929 /etc/fstab and without root= on the kernel command line on
2930 systems prepared appropriately.
2932 * systemd-nspawn gained a new --image= switch which allows
2933 booting up disk images and Linux installations on any block
2934 device that follow the Discoverable Partitions Specification
2935 (see above). This means that installations made with
2936 appropriately updated installers may now be started and
2937 deployed using container managers, completely
2938 unmodified. (We hope that libvirt-lxc will add support for
2939 this feature soon, too.)
2941 * systemd-nspawn gained a new --network-macvlan= setting to
2942 set up a private macvlan interface for the
2943 container. Similarly, systemd-networkd gained a new
2944 Kind=macvlan setting in .netdev files.
2946 * systemd-networkd now supports configuring local addresses
2949 * A new tool systemd-network-wait-online has been added to
2950 synchronously wait for network connectivity using
2953 * The sd-bus.h bus API gained a new sd_bus_track object for
2954 tracking the life-cycle of bus peers. Note that sd-bus.h is
2955 still not a public API though (unless you specify
2956 --enable-kdbus on the configure command line, which however
2957 voids your warranty and you get no API stability guarantee).
2959 * The $XDG_RUNTIME_DIR runtime directories for each user are
2960 now individual tmpfs instances, which has the benefit of
2961 introducing separate pools for each user, with individual
2962 size limits, and thus making sure that unprivileged clients
2963 can no longer negatively impact the system or other users by
2964 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2965 RuntimeDirectorySize= has been introduced that allows
2966 controlling the default size limit for all users. It
2967 defaults to 10% of the available physical memory. This is no
2968 replacement for quotas on tmpfs though (which the kernel
2969 still does not support), as /dev/shm and /tmp are still
2970 shared resources used by both the system and unprivileged
2973 * logind will now automatically turn off automatic suspending
2974 on laptop lid close when more than one display is
2975 connected. This was previously expected to be implemented
2976 individually in desktop environments (such as GNOME),
2977 however has been added to logind now, in order to fix a
2978 boot-time race where a desktop environment might not have
2979 been started yet and thus not been able to take an inhibitor
2980 lock at the time where logind already suspends the system
2981 due to a closed lid.
2983 * logind will now wait at least 30s after each system
2984 suspend/resume cycle, and 3min after system boot before
2985 suspending the system due to a closed laptop lid. This
2986 should give USB docking stations and similar enough time to
2987 be probed and configured after system resume and boot in
2988 order to then act as suspend blocker.
2990 * systemd-run gained a new --property= setting which allows
2991 initialization of resource control properties (and others)
2992 for the created scope or service unit. Example: "systemd-run
2993 --property=BlockIOWeight=10 updatedb" may be used to run
2994 updatedb at a low block IO scheduling weight.
2996 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2997 now also work in --scope mode.
2999 * When systemd is compiled with kdbus support, basic support
3000 for enforced policies is now in place. (Note that enabling
3001 kdbus still voids your warranty and no API compatibility
3004 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3005 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3006 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3007 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3008 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3009 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3010 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3011 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3012 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3013 Zbigniew Jędrzejewski-Szmek
3015 -- Berlin, 2014-03-12
3019 * systemd will now relabel /dev after loading the SMACK policy
3020 according to SMACK rules.
3022 * A new unit file option AppArmorProfile= has been added to
3023 set the AppArmor profile for the processes of a unit.
3025 * A new condition check ConditionArchitecture= has been added
3026 to conditionalize units based on the system architecture, as
3027 reported by uname()'s "machine" field.
3029 * systemd-networkd now supports matching on the system
3030 virtualization, architecture, kernel command line, host name
3033 * logind is now a lot more aggressive when suspending the
3034 machine due to a closed laptop lid. Instead of acting only
3035 on the lid close action, it will continuously watch the lid
3036 status and act on it. This is useful for laptops where the
3037 power button is on the outside of the chassis so that it can
3038 be reached without opening the lid (such as the Lenovo
3039 Yoga). On those machines, logind will now immediately
3040 re-suspend the machine if the power button has been
3041 accidentally pressed while the laptop was suspended and in a
3042 backpack or similar.
3044 * logind will now watch SW_DOCK switches and inhibit reaction
3045 to the lid switch if it is pressed. This means that logind
3046 will not suspend the machine anymore if the lid is closed
3047 and the system is docked, if the laptop supports SW_DOCK
3048 notifications via the input layer. Note that ACPI docking
3049 stations do not generate this currently. Also note that this
3050 logic is usually not fully sufficient and Desktop
3051 Environments should take a lid switch inhibitor lock when an
3052 external display is connected, as systemd will not watch
3055 * nspawn will now make use of the devices cgroup controller by
3056 default, and only permit creation of and access to the usual
3057 API device nodes like /dev/null or /dev/random, as well as
3058 access to (but not creation of) the pty devices.
3060 * We will now ship a default .network file for
3061 systemd-networkd that automatically configures DHCP for
3062 network interfaces created by nspawn's --network-veth or
3063 --network-bridge= switches.
3065 * systemd will now understand the usual M, K, G, T suffixes
3066 according to SI conventions (i.e. to the base 1000) when
3067 referring to throughput and hardware metrics. It will stay
3068 with IEC conventions (i.e. to the base 1024) for software
3069 metrics, according to what is customary according to
3070 Wikipedia. We explicitly document which base applies for
3071 each configuration option.
3073 * The DeviceAllow= setting in unit files now supports a syntax
3074 to whitelist an entire group of devices node majors at once,
3075 based on the /proc/devices listing. For example, with the
3076 string "char-pts", it is now possible to whitelist all
3077 current and future pseudo-TTYs at once.
3079 * sd-event learned a new "post" event source. Event sources of
3080 this type are triggered by the dispatching of any event
3081 source of a type that is not "post". This is useful for
3082 implementing clean-up and check event sources that are
3083 triggered by other work being done in the program.
3085 * systemd-networkd is no longer statically enabled, but uses
3086 the usual [Install] sections so that it can be
3087 enabled/disabled using systemctl. It still is enabled by
3090 * When creating a veth interface pair with systemd-nspawn, the
3091 host side will now be prefixed with "vb-" if
3092 --network-bridge= is used, and with "ve-" if --network-veth
3093 is used. This way, it is easy to distinguish these cases on
3094 the host, for example to apply different configuration to
3095 them with systemd-networkd.
3097 * The compatibility libraries for libsystemd-journal.so,
3098 libsystem-id128.so, libsystemd-login.so and
3099 libsystemd-daemon.so do not make use of IFUNC
3100 anymore. Instead, we now build libsystemd.so multiple times
3101 under these alternative names. This means that the footprint
3102 is drastically increased, but given that these are
3103 transitional compatibility libraries, this should not matter
3104 much. This change has been made necessary to support the ARM
3105 platform for these compatibility libraries, as the ARM
3106 toolchain is not really at the same level as the toolchain
3107 for other architectures like x86 and does not support
3108 IFUNC. Please make sure to use --enable-compat-libs only
3109 during a transitional period!
3111 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3112 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3113 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3114 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3115 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3116 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3117 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3118 Zbigniew Jędrzejewski-Szmek
3120 -- Berlin, 2014-02-24
3124 * A new component "systemd-networkd" has been added that can
3125 be used to configure local network interfaces statically or
3126 via DHCP. It is capable of bringing up bridges, VLANs, and
3127 bonding. Currently, no hook-ups for interactive network
3128 configuration are provided. Use this for your initrd,
3129 container, embedded, or server setup if you need a simple,
3130 yet powerful, network configuration solution. This
3131 configuration subsystem is quite nifty, as it allows wildcard
3132 hotplug matching in interfaces. For example, with a single
3133 configuration snippet, you can configure that all Ethernet
3134 interfaces showing up are automatically added to a bridge,
3135 or similar. It supports link-sensing and more.
3137 * A new tool "systemd-socket-proxyd" has been added which can
3138 act as a bidirectional proxy for TCP sockets. This is
3139 useful for adding socket activation support to services that
3140 do not actually support socket activation, including virtual
3141 machines and the like.
3143 * Add a new tool to save/restore rfkill state on
3146 * Save/restore state of keyboard backlights in addition to
3147 display backlights on shutdown/boot.
3149 * udev learned a new SECLABEL{} construct to label device
3150 nodes with a specific security label when they appear. For
3151 now, only SECLABEL{selinux} is supported, but the syntax is
3152 prepared for additional security frameworks.
3154 * udev gained a new scheme to configure link-level attributes
3155 from files in /etc/systemd/network/*.link. These files can
3156 match against MAC address, device path, driver name and type,
3157 and will apply attributes like the naming policy, link speed,
3158 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3159 address assignment policy (randomized, ...).
3161 * The configuration of network interface naming rules for
3162 "permanent interface names" has changed: a new NamePolicy=
3163 setting in the [Link] section of .link files determines the
3164 priority of possible naming schemes (onboard, slot, MAC,
3165 path). The default value of this setting is determined by
3166 /usr/lib/net/links/99-default.link. Old
3167 80-net-name-slot.rules udev configuration file has been
3168 removed, so local configuration overriding this file should
3169 be adapated to override 99-default.link instead.
3171 * When the User= switch is used in a unit file, also
3172 initialize $SHELL= based on the user database entry.
3174 * systemd no longer depends on libdbus. All communication is
3175 now done with sd-bus, systemd's low-level bus library
3178 * kdbus support has been added to PID 1 itself. When kdbus is
3179 enabled, this causes PID 1 to set up the system bus and
3180 enable support for a new ".busname" unit type that
3181 encapsulates bus name activation on kdbus. It works a little
3182 bit like ".socket" units, except for bus names. A new
3183 generator has been added that converts classic dbus1 service
3184 activation files automatically into native systemd .busname
3187 * sd-bus: add a light-weight vtable implementation that allows
3188 defining objects on the bus with a simple static const
3189 vtable array of its methods, signals and properties.
3191 * systemd will not generate or install static dbus
3192 introspection data anymore to /usr/share/dbus-1/interfaces,
3193 as the precise format of these files is unclear, and
3194 nothing makes use of it.
3196 * A proxy daemon is now provided to proxy clients connecting
3197 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3198 compatibility with classic D-Bus.
3200 * A bus driver implementation has been added that supports the
3201 classic D-Bus bus driver calls on kdbus, also for
3202 compatibility purposes.
3204 * A new API "sd-event.h" has been added that implements a
3205 minimal event loop API built around epoll. It provides a
3206 couple of features that direct epoll usage is lacking:
3207 prioritization of events, scales to large numbers of timer
3208 events, per-event timer slack (accuracy), system-wide
3209 coalescing of timer events, exit handlers, watchdog
3210 supervision support using systemd's sd_notify() API, child
3213 * A new API "sd-rntl.h" has been added that provides an API
3214 around the route netlink interface of the kernel, similar in
3215 style to "sd-bus.h".
3217 * A new API "sd-dhcp-client.h" has been added that provides a
3218 small DHCPv4 client-side implementation. This is used by
3221 * There is a new kernel command line option
3222 "systemd.restore_state=0|1". When set to "0", none of the
3223 systemd tools will restore saved runtime state to hardware
3224 devices. More specifically, the rfkill and backlight states
3227 * The FsckPassNo= compatibility option in mount/service units
3228 has been removed. The fstab generator will now add the
3229 necessary dependencies automatically, and does not require
3230 PID1's support for that anymore.
3232 * journalctl gained a new switch, --list-boots, that lists
3233 recent boots with their times and boot IDs.
3235 * The various tools like systemctl, loginctl, timedatectl,
3236 busctl, systemd-run, ... have gained a new switch "-M" to
3237 connect to a specific, local OS container (as direct
3238 connection, without requiring SSH). This works on any
3239 container that is registered with machined, such as those
3240 created by libvirt-lxc or nspawn.
3242 * systemd-run and systemd-analyze also gained support for "-H"
3243 to connect to remote hosts via SSH. This is particularly
3244 useful for systemd-run because it enables queuing of jobs
3245 onto remote systems.
3247 * machinectl gained a new command "login" to open a getty
3248 login in any local container. This works with any container
3249 that is registered with machined (such as those created by
3250 libvirt-lxc or nspawn), and which runs systemd inside.
3252 * machinectl gained a new "reboot" command that may be used to
3253 trigger a reboot on a specific container that is registered
3254 with machined. This works on any container that runs an init
3255 system of some kind.
3257 * systemctl gained a new "list-timers" command to print a nice
3258 listing of installed timer units with the times they elapse
3261 * Alternative reboot() parameters may now be specified on the
3262 "systemctl reboot" command line and are passed to the
3263 reboot() system call.
3265 * systemctl gained a new --job-mode= switch to configure the
3266 mode to queue a job with. This is a more generic version of
3267 --fail, --irreversible, and --ignore-dependencies, which are
3268 still available but not advertised anymore.
3270 * /etc/systemd/system.conf gained new settings to configure
3271 various default timeouts of units, as well as the default
3272 start limit interval and burst. These may still be overridden
3275 * PID1 will now export on the bus profile data of the security
3276 policy upload process (such as the SELinux policy upload to
3279 * journald: when forwarding logs to the console, include
3280 timestamps (following the setting in
3281 /sys/module/printk/parameters/time).
3283 * OnCalendar= in timer units now understands the special
3284 strings "yearly" and "annually". (Both are equivalent)
3286 * The accuracy of timer units is now configurable with the new
3287 AccuracySec= setting. It defaults to 1min.
3289 * A new dependency type JoinsNamespaceOf= has been added that
3290 allows running two services within the same /tmp and network
3291 namespace, if PrivateNetwork= or PrivateTmp= are used.
3293 * A new command "cat" has been added to systemctl. It outputs
3294 the original unit file of a unit, and concatenates the
3295 contents of additional "drop-in" unit file snippets, so that
3296 the full configuration is shown.
3298 * systemctl now supports globbing on the various "list-xyz"
3299 commands, like "list-units" or "list-sockets", as well as on
3300 those commands which take multiple unit names.
3302 * journalctl's --unit= switch gained support for globbing.
3304 * All systemd daemons now make use of the watchdog logic so
3305 that systemd automatically notices when they hang.
3307 * If the $container_ttys environment variable is set,
3308 getty-generator will automatically spawn a getty for each
3309 listed tty. This is useful for container managers to request
3310 login gettys to be spawned on as many ttys as needed.
3312 * %h, %s, %U specifier support is not available anymore when
3313 used in unit files for PID 1. This is because NSS calls are
3314 not safe from PID 1. They stay available for --user
3315 instances of systemd, and as special case for the root user.
3317 * loginctl gained a new "--no-legend" switch to turn off output
3320 * The "sd-login.h" API gained three new calls:
3321 sd_session_is_remote(), sd_session_get_remote_user(),
3322 sd_session_get_remote_host() to query information about
3325 * The udev hardware database now also carries vendor/product
3326 information of SDIO devices.
3328 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3329 determine whether watchdog notifications are requested by
3332 * Socket-activated per-connection services now include a
3333 short description of the connection parameters in the
3336 * tmpfiles gained a new "--boot" option. When this is not used,
3337 only lines where the command character is not suffixed with
3338 "!" are executed. When this option is specified, those
3339 options are executed too. This partitions tmpfiles
3340 directives into those that can be safely executed at any
3341 time, and those which should be run only at boot (for
3342 example, a line that creates /run/nologin).
3344 * A new API "sd-resolve.h" has been added which provides a simple
3345 asynchronous wrapper around glibc NSS host name resolution
3346 calls, such as getaddrinfo(). In contrast to glibc's
3347 getaddrinfo_a(), it does not use signals. In contrast to most
3348 other asynchronous name resolution libraries, this one does
3349 not reimplement DNS, but reuses NSS, so that alternate
3350 host name resolution systems continue to work, such as mDNS,
3351 LDAP, etc. This API is based on libasyncns, but it has been
3352 cleaned up for inclusion in systemd.
3354 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3355 "sd-daemon.h" are no longer found in individual libraries
3356 libsystemd-journal.so, libsystemd-login.so,
3357 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3358 merged them into a single library, libsystemd.so, which
3359 provides all symbols. The reason for this is cyclic
3360 dependencies, as these libraries tend to use each other's
3361 symbols. So far, we have managed to workaround that by linking
3362 a copy of a good part of our code into each of these
3363 libraries again and again, which, however, makes certain
3364 things hard to do, like sharing static variables. Also, it
3365 substantially increases footprint. With this change, there
3366 is only one library for the basic APIs systemd
3367 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3368 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3369 library as well, however are subject to the --enable-kdbus
3370 switch (see below). Note that "sd-dhcp-client.h" is not part
3371 of this library (this is because it only consumes, never
3372 provides, services of/to other APIs). To make the transition
3373 easy from the separate libraries to the unified one, we
3374 provide the --enable-compat-libs compile-time switch which
3375 will generate stub libraries that are compatible with the
3376 old ones but redirect all calls to the new one.
3378 * All of the kdbus logic and the new APIs "sd-bus.h",
3379 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3380 and "sd-utf8.h" are compile-time optional via the
3381 "--enable-kdbus" switch, and they are not compiled in by
3382 default. To make use of kdbus, you have to explicitly enable
3383 the switch. Note however, that neither the kernel nor the
3384 userspace API for all of this is considered stable yet. We
3385 want to maintain the freedom to still change the APIs for
3386 now. By specifying this build-time switch, you acknowledge
3387 that you are aware of the instability of the current
3390 * Also, note that while kdbus is pretty much complete,
3391 it lacks one thing: proper policy support. This means you
3392 can build a fully working system with all features; however,
3393 it will be highly insecure. Policy support will be added in
3394 one of the next releases, at the same time that we will
3395 declare the APIs stable.
3397 * When the kernel command line argument "kdbus" is specified,
3398 systemd will automatically load the kdbus.ko kernel module. At
3399 this stage of development, it is only useful for testing kdbus
3400 and should not be used in production. Note: if "--enable-kdbus"
3401 is specified, and the kdbus.ko kernel module is available, and
3402 "kdbus" is added to the kernel command line, the entire system
3403 runs with kdbus instead of dbus-daemon, with the above mentioned
3404 problem of missing the system policy enforcement. Also a future
3405 version of kdbus.ko or a newer systemd will not be compatible with
3406 each other, and will unlikely be able to boot the machine if only
3407 one of them is updated.
3409 * systemctl gained a new "import-environment" command which
3410 uploads the caller's environment (or parts thereof) into the
3411 service manager so that it is inherited by services started
3412 by the manager. This is useful to upload variables like
3413 $DISPLAY into the user service manager.
3415 * A new PrivateDevices= switch has been added to service units
3416 which allows running a service with a namespaced /dev
3417 directory that does not contain any device nodes for
3418 physical devices. More specifically, it only includes devices
3419 such as /dev/null, /dev/urandom, and /dev/zero which are API
3422 * logind has been extended to support behaviour like VT
3423 switching on seats that do not support a VT. This makes
3424 multi-session available on seats that are not the first seat
3425 (seat0), and on systems where kernel support for VTs has
3426 been disabled at compile-time.
3428 * If a process holds a delay lock for system sleep or shutdown
3429 and fails to release it in time, we will now log its
3430 identity. This makes it easier to identify processes that
3431 cause slow suspends or power-offs.
3433 * When parsing /etc/crypttab, support for a new key-slot=
3434 option as supported by Debian is added. It allows indicating
3435 which LUKS slot to use on disk, speeding up key loading.
3437 * The sd_journald_sendv() API call has been checked and
3438 officially declared to be async-signal-safe so that it may
3439 be invoked from signal handlers for logging purposes.
3441 * Boot-time status output is now enabled automatically after a
3442 short timeout if boot does not progress, in order to give
3443 the user an indication what she or he is waiting for.
3445 * The boot-time output has been improved to show how much time
3446 remains until jobs expire.
3448 * The KillMode= switch in service units gained a new possible
3449 value "mixed". If set, and the unit is shut down, then the
3450 initial SIGTERM signal is sent only to the main daemon
3451 process, while the following SIGKILL signal is sent to
3452 all remaining processes of the service.
3454 * When a scope unit is registered, a new property "Controller"
3455 may be set. If set to a valid bus name, systemd will send a
3456 RequestStop() signal to this name when it would like to shut
3457 down the scope. This may be used to hook manager logic into
3458 the shutdown logic of scope units. Also, scope units may now
3459 be put in a special "abandoned" state, in which case the
3460 manager process which created them takes no further
3461 responsibilities for it.
3463 * When reading unit files, systemd will now verify
3464 the access mode of these files, and warn about certain
3465 suspicious combinations. This has been added to make it
3466 easier to track down packaging bugs where unit files are
3467 marked executable or world-writable.
3469 * systemd-nspawn gained a new "--setenv=" switch to set
3470 container-wide environment variables. The similar option in
3471 systemd-activate was renamed from "--environment=" to
3472 "--setenv=" for consistency.
3474 * systemd-nspawn has been updated to create a new kdbus domain
3475 for each container that is invoked, thus allowing each
3476 container to have its own set of system and user buses,
3477 independent of the host.
3479 * systemd-nspawn gained a new --drop-capability= switch to run
3480 the container with less capabilities than the default. Both
3481 --drop-capability= and --capability= now take the special
3482 string "all" for dropping or keeping all capabilities.
3484 * systemd-nspawn gained new switches for executing containers
3485 with specific SELinux labels set.
3487 * systemd-nspawn gained a new --quiet switch to not generate
3488 any additional output but the container's own console
3491 * systemd-nspawn gained a new --share-system switch to run a
3492 container without PID namespacing enabled.
3494 * systemd-nspawn gained a new --register= switch to control
3495 whether the container is registered with systemd-machined or
3496 not. This is useful for containers that do not run full
3497 OS images, but only specific apps.
3499 * systemd-nspawn gained a new --keep-unit which may be used
3500 when invoked as the only program from a service unit, and
3501 results in registration of the unit service itself in
3502 systemd-machined, instead of a newly opened scope unit.
3504 * systemd-nspawn gained a new --network-interface= switch for
3505 moving arbitrary interfaces to the container. The new
3506 --network-veth switch creates a virtual Ethernet connection
3507 between host and container. The new --network-bridge=
3508 switch then allows assigning the host side of this virtual
3509 Ethernet connection to a bridge device.
3511 * systemd-nspawn gained a new --personality= switch for
3512 setting the kernel personality for the container. This is
3513 useful when running a 32-bit container on a 64-bit host. A
3514 similar option Personality= is now also available for service
3517 * logind will now also track a "Desktop" identifier for each
3518 session which encodes the desktop environment of it. This is
3519 useful for desktop environments that want to identify
3520 multiple running sessions of itself easily.
3522 * A new SELinuxContext= setting for service units has been
3523 added that allows setting a specific SELinux execution
3524 context for a service.
3526 * Most systemd client tools will now honour $SYSTEMD_LESS for
3527 settings of the "less" pager. By default, these tools will
3528 override $LESS to allow certain operations to work, such as
3529 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3530 influence this logic.
3532 * systemd's "seccomp" hook-up has been changed to make use of
3533 the libseccomp library instead of using its own
3534 implementation. This has benefits for portability among
3537 * For usage together with SystemCallFilter=, a new
3538 SystemCallErrorNumber= setting has been introduced that
3539 allows configuration of a system error number to be returned
3540 on filtered system calls, instead of immediately killing the
3541 process. Also, SystemCallArchitectures= has been added to
3542 limit access to system calls of a particular architecture
3543 (in order to turn off support for unused secondary
3544 architectures). There is also a global
3545 SystemCallArchitectures= setting in system.conf now to turn
3546 off support for non-native system calls system-wide.
3548 * systemd requires a kernel with a working name_to_handle_at(),
3549 please see the kernel config requirements in the README file.
3551 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3552 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3553 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3554 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3555 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3556 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3557 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3558 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3559 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3560 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3561 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3562 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3563 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3564 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3565 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3566 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3567 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3568 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3569 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3570 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3571 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3572 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3573 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3574 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3576 -- Berlin, 2014-02-20
3580 * logind has gained support for facilitating privileged input
3581 and drm device access for unprivileged clients. This work is
3582 useful to allow Wayland display servers (and similar
3583 programs, such as kmscon) to run under the user's ID and
3584 access input and drm devices which are normally
3585 protected. When this is used (and the kernel is new enough)
3586 logind will "mute" IO on the file descriptors passed to
3587 Wayland as long as it is in the background and "unmute" it
3588 if it returns into the foreground. This allows secure
3589 session switching without allowing background sessions to
3590 eavesdrop on input and display data. This also introduces
3591 session switching support if VT support is turned off in the
3592 kernel, and on seats that are not seat0.
3594 * A new kernel command line option luks.options= is understood
3595 now which allows specifying LUKS options for usage for LUKS
3596 encrypted partitions specified with luks.uuid=.
3598 * tmpfiles.d(5) snippets may now use specifier expansion in
3599 path names. More specifically %m, %b, %H, %v, are now
3600 replaced by the local machine id, boot id, hostname, and
3601 kernel version number.
3603 * A new tmpfiles.d(5) command "m" has been introduced which
3604 may be used to change the owner/group/access mode of a file
3605 or directory if it exists, but do nothing if it does not.
3607 * This release removes high-level support for the
3608 MemorySoftLimit= cgroup setting. The underlying kernel
3609 cgroup attribute memory.soft_limit= is currently badly
3610 designed and likely to be removed from the kernel API in its
3611 current form, hence we should not expose it for now.
3613 * The memory.use_hierarchy cgroup attribute is now enabled for
3614 all cgroups systemd creates in the memory cgroup
3615 hierarchy. This option is likely to be come the built-in
3616 default in the kernel anyway, and the non-hierarchical mode
3617 never made much sense in the intrinsically hierarchical
3620 * A new field _SYSTEMD_SLICE= is logged along with all journal
3621 messages containing the slice a message was generated
3622 from. This is useful to allow easy per-customer filtering of
3623 logs among other things.
3625 * systemd-journald will no longer adjust the group of journal
3626 files it creates to the "systemd-journal" group. Instead we
3627 rely on the journal directory to be owned by the
3628 "systemd-journal" group, and its setgid bit set, so that the
3629 kernel file system layer will automatically enforce that
3630 journal files inherit this group assignment. The reason for
3631 this change is that we cannot allow NSS look-ups from
3632 journald which would be necessary to resolve
3633 "systemd-journal" to a numeric GID, because this might
3634 create deadlocks if NSS involves synchronous queries to
3635 other daemons (such as nscd, or sssd) which in turn are
3636 logging clients of journald and might block on it, which
3637 would then dead lock. A tmpfiles.d(5) snippet included in
3638 systemd will make sure the setgid bit and group are
3639 properly set on the journal directory if it exists on every
3640 boot. However, we recommend adjusting it manually after
3641 upgrades too (or from RPM scriptlets), so that the change is
3642 not delayed until next reboot.
3644 * Backlight and random seed files in /var/lib/ have moved into
3645 the /var/lib/systemd/ directory, in order to centralize all
3646 systemd generated files in one directory.
3648 * Boot time performance measurements (as displayed by
3649 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3650 performance information if that's available to determine how
3651 much time BIOS and boot loader initialization required. With
3652 a sufficiently new BIOS you hence no longer need to boot
3653 with Gummiboot to get access to such information.
3655 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3656 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3657 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3658 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3659 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3660 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3661 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3663 -- Berlin, 2013-10-02
3667 * The Restart= option for services now understands a new
3668 on-watchdog setting, which will restart the service
3669 automatically if the service stops sending out watchdog keep
3670 alive messages (as configured with WatchdogSec=).
3672 * The getty generator (which is responsible for bringing up a
3673 getty on configured serial consoles) will no longer only
3674 start a getty on the primary kernel console but on all
3675 others, too. This makes the order in which console= is
3676 specified on the kernel command line less important.
3678 * libsystemd-logind gained a new sd_session_get_vt() call to
3679 retrieve the VT number of a session.
3681 * If the option "tries=0" is set for an entry of /etc/crypttab
3682 its passphrase is queried indefinitely instead of any
3683 maximum number of tries.
3685 * If a service with a configure PID file terminates its PID
3686 file will now be removed automatically if it still exists
3687 afterwards. This should put an end to stale PID files.
3689 * systemd-run will now also take relative binary path names
3690 for execution and no longer insists on absolute paths.
3692 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3693 paths that are optionally prefixed with "-" to indicate that
3694 it should not be considered a failure if they do not exist.
3696 * journalctl -o (and similar commands) now understands a new
3697 output mode "short-precise", it is similar to "short" but
3698 shows timestamps with usec accuracy.
3700 * The option "discard" (as known from Debian) is now
3701 synonymous to "allow-discards" in /etc/crypttab. In fact,
3702 "discard" is preferred now (since it is easier to remember
3705 * Some licensing clean-ups were made, so that more code is now
3706 LGPL-2.1 licensed than before.
3708 * A minimal tool to save/restore the display backlight
3709 brightness across reboots has been added. It will store the
3710 backlight setting as late as possible at shutdown, and
3711 restore it as early as possible during reboot.
3713 * A logic to automatically discover and enable home and swap
3714 partitions on GPT disks has been added. With this in place
3715 /etc/fstab becomes optional for many setups as systemd can
3716 discover certain partitions located on the root disk
3717 automatically. Home partitions are recognized under their
3718 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3719 partitions are recognized under their GPT type ID
3720 0657fd6da4ab43c484e50933c84b4f4f.
3722 * systemd will no longer pass any environment from the kernel
3723 or initrd to system services. If you want to set an
3724 environment for all services, do so via the kernel command
3725 line systemd.setenv= assignment.
3727 * The systemd-sysctl tool no longer natively reads the file
3728 /etc/sysctl.conf. If desired, the file should be symlinked
3729 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3730 legacy support by a symlink rather than built-in code, it
3731 also makes the otherwise hidden order of application of the
3732 different files visible. (Note that this partly reverts to a
3733 pre-198 application order of sysctl knobs!)
3735 * The "systemctl set-log-level" and "systemctl dump" commands
3736 have been moved to systemd-analyze.
3738 * systemd-run learned the new --remain-after-exit switch,
3739 which causes the scope unit not to be cleaned up
3740 automatically after the process terminated.
3742 * tmpfiles learned a new --exclude-prefix= switch to exclude
3743 certain paths from operation.
3745 * journald will now automatically flush all messages to disk
3746 as soon as a message at the log level CRIT, ALERT or EMERG
3749 Contributions from: Andrew Cook, Brandon Philips, Christian
3750 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3751 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3752 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3753 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3754 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3755 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3756 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3757 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3758 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3759 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3760 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3761 William Giokas, Zbigniew Jędrzejewski-Szmek
3763 -- Berlin, 2013-09-13
3767 * The documentation has been updated to cover the various new
3768 concepts introduced with 205.
3770 * Unit files now understand the new %v specifier which
3771 resolves to the kernel version string as returned by "uname
3774 * systemctl now supports filtering the unit list output by
3775 load state, active state and sub state, using the new
3778 * "systemctl status" will now show the results of the
3779 condition checks (like ConditionPathExists= and similar) of
3780 the last start attempts of the unit. They are also logged to
3783 * "journalctl -b" may now be used to look for boot output of a
3784 specific boot. Try "journalctl -b -1" for the previous boot,
3785 but the syntax is substantially more powerful.
3787 * "journalctl --show-cursor" has been added which prints the
3788 cursor string the last shown log line. This may then be used
3789 with the new "journalctl --after-cursor=" switch to continue
3790 browsing logs from that point on.
3792 * "journalctl --force" may now be used to force regeneration
3795 * Creation of "dead" device nodes has been moved from udev
3796 into kmod and tmpfiles. Previously, udev would read the kmod
3797 databases to pre-generate dead device nodes based on meta
3798 information contained in kernel modules, so that these would
3799 be auto-loaded on access rather then at boot. As this
3800 does not really have much to do with the exposing actual
3801 kernel devices to userspace this has always been slightly
3802 alien in the udev codebase. Following the new scheme kmod
3803 will now generate a runtime snippet for tmpfiles from the
3804 module meta information and it now is tmpfiles' job to the
3805 create the nodes. This also allows overriding access and
3806 other parameters for the nodes using the usual tmpfiles
3807 facilities. As side effect this allows us to remove the
3808 CAP_SYS_MKNOD capability bit from udevd entirely.
3810 * logind's device ACLs may now be applied to these "dead"
3811 devices nodes too, thus finally allowing managed access to
3812 devices such as /dev/snd/sequencer whithout loading the
3813 backing module right-away.
3815 * A new RPM macro has been added that may be used to apply
3816 tmpfiles configuration during package installation.
3818 * systemd-detect-virt and ConditionVirtualization= now can
3819 detect User-Mode-Linux machines (UML).
3821 * journald will now implicitly log the effective capabilities
3822 set of processes in the message metadata.
3824 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3826 * The initrd interface has been simplified (more specifically,
3827 support for passing performance data via environment
3828 variables and fsck results via files in /run has been
3829 removed). These features were non-essential, and are
3830 nowadays available in a much nicer way by having systemd in
3831 the initrd serialize its state and have the hosts systemd
3832 deserialize it again.
3834 * The udev "keymap" data files and tools to apply keyboard
3835 specific mappings of scan to key codes, and force-release
3836 scan code lists have been entirely replaced by a udev
3837 "keyboard" builtin and a hwdb data file.
3839 * systemd will now honour the kernel's "quiet" command line
3840 argument also during late shutdown, resulting in a
3841 completely silent shutdown when used.
3843 * There's now an option to control the SO_REUSEPORT socket
3844 option in .socket units.
3846 * Instance units will now automatically get a per-template
3847 subslice of system.slice unless something else is explicitly
3848 configured. For example, instances of sshd@.service will now
3849 implicitly be placed in system-sshd.slice rather than
3850 system.slice as before.
3852 * Test coverage support may now be enabled at build time.
3854 Contributions from: Dave Reisner, Frederic Crozat, Harald
3855 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3856 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3857 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3858 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3859 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3860 Giokas, Zbigniew Jędrzejewski-Szmek
3862 -- Berlin, 2013-07-23
3866 * Two new unit types have been introduced:
3868 Scope units are very similar to service units, however, are
3869 created out of pre-existing processes -- instead of PID 1
3870 forking off the processes. By using scope units it is
3871 possible for system services and applications to group their
3872 own child processes (worker processes) in a powerful way
3873 which then maybe used to organize them, or kill them
3874 together, or apply resource limits on them.
3876 Slice units may be used to partition system resources in an
3877 hierarchical fashion and then assign other units to them. By
3878 default there are now three slices: system.slice (for all
3879 system services), user.slice (for all user sessions),
3880 machine.slice (for VMs and containers).
3882 Slices and scopes have been introduced primarily in
3883 context of the work to move cgroup handling to a
3884 single-writer scheme, where only PID 1
3885 creates/removes/manages cgroups.
3887 * There's a new concept of "transient" units. In contrast to
3888 normal units these units are created via an API at runtime,
3889 not from configuration from disk. More specifically this
3890 means it is now possible to run arbitrary programs as
3891 independent services, with all execution parameters passed
3892 in via bus APIs rather than read from disk. Transient units
3893 make systemd substantially more dynamic then it ever was,
3894 and useful as a general batch manager.
3896 * logind has been updated to make use of scope and slice units
3897 for managing user sessions. As a user logs in he will get
3898 his own private slice unit, to which all sessions are added
3899 as scope units. We also added support for automatically
3900 adding an instance of user@.service for the user into the
3901 slice. Effectively logind will no longer create cgroup
3902 hierarchies on its own now, it will defer entirely to PID 1
3903 for this by means of scope, service and slice units. Since
3904 user sessions this way become entities managed by PID 1
3905 the output of "systemctl" is now a lot more comprehensive.
3907 * A new mini-daemon "systemd-machined" has been added which
3908 may be used by virtualization managers to register local
3909 VMs/containers. nspawn has been updated accordingly, and
3910 libvirt will be updated shortly. machined will collect a bit
3911 of meta information about the VMs/containers, and assign
3912 them their own scope unit (see above). The collected
3913 meta-data is then made available via the "machinectl" tool,
3914 and exposed in "ps" and similar tools. machined/machinectl
3915 is compile-time optional.
3917 * As discussed earlier, the low-level cgroup configuration
3918 options ControlGroup=, ControlGroupModify=,
3919 ControlGroupPersistent=, ControlGroupAttribute= have been
3920 removed. Please use high-level attribute settings instead as
3921 well as slice units.
3923 * A new bus call SetUnitProperties() has been added to alter
3924 various runtime parameters of a unit. This is primarily
3925 useful to alter cgroup parameters dynamically in a nice way,
3926 but will be extended later on to make more properties
3927 modifiable at runtime. systemctl gained a new set-properties
3928 command that wraps this call.
3930 * A new tool "systemd-run" has been added which can be used to
3931 run arbitrary command lines as transient services or scopes,
3932 while configuring a number of settings via the command
3933 line. This tool is currently very basic, however already
3934 very useful. We plan to extend this tool to even allow
3935 queuing of execution jobs with time triggers from the
3936 command line, similar in fashion to "at".
3938 * nspawn will now inform the user explicitly that kernels with
3939 audit enabled break containers, and suggest the user to turn
3942 * Support for detecting the IMA and AppArmor security
3943 frameworks with ConditionSecurity= has been added.
3945 * journalctl gained a new "-k" switch for showing only kernel
3946 messages, mimicking dmesg output; in addition to "--user"
3947 and "--system" switches for showing only user's own logs
3950 * systemd-delta can now show information about drop-in
3951 snippets extending unit files.
3953 * libsystemd-bus has been substantially updated but is still
3954 not available as public API.
3956 * systemd will now look for the "debug" argument on the kernel
3957 command line and enable debug logging, similar to what
3958 "systemd.log_level=debug" already did before.
3960 * "systemctl set-default", "systemctl get-default" has been
3961 added to configure the default.target symlink, which
3962 controls what to boot into by default.
3964 * "systemctl set-log-level" has been added as a convenient
3965 way to raise and lower systemd logging threshold.
3967 * "systemd-analyze plot" will now show the time the various
3968 generators needed for execution, as well as information
3969 about the unit file loading.
3971 * libsystemd-journal gained a new sd_journal_open_files() call
3972 for opening specific journal files. journactl also gained a
3973 new switch to expose this new functionality. Previously we
3974 only supported opening all files from a directory, or all
3975 files from the system, as opening individual files only is
3976 racy due to journal file rotation.
3978 * systemd gained the new DefaultEnvironment= setting in
3979 /etc/systemd/system.conf to set environment variables for
3982 * If a privileged process logs a journal message with the
3983 OBJECT_PID= field set, then journald will automatically
3984 augment this with additional OBJECT_UID=, OBJECT_GID=,
3985 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3986 system services want to log events about specific client
3987 processes. journactl/systemctl has been updated to make use
3988 of this information if all log messages regarding a specific
3991 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3992 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3993 Reisner, David Coppa, David King, David Strauss, Eelco
3994 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3995 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3996 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3997 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3998 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3999 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4000 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4001 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4002 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4003 Łukasz Stelmach, 장동준
4007 * The Python bindings gained some minimal support for the APIs
4008 exposed by libsystemd-logind.
4010 * ConditionSecurity= gained support for detecting SMACK. Since
4011 this condition already supports SELinux and AppArmor we only
4012 miss IMA for this. Patches welcome!
4014 Contributions from: Karol Lewandowski, Lennart Poettering,
4015 Zbigniew Jędrzejewski-Szmek
4019 * systemd-nspawn will now create /etc/resolv.conf if
4020 necessary, before bind-mounting the host's file onto it.
4022 * systemd-nspawn will now store meta information about a
4023 container on the container's cgroup as extended attribute
4024 fields, including the root directory.
4026 * The cgroup hierarchy has been reworked in many ways. All
4027 objects any of the components systemd creates in the cgroup
4028 tree are now suffixed. More specifically, user sessions are
4029 now placed in cgroups suffixed with ".session", users in
4030 cgroups suffixed with ".user", and nspawn containers in
4031 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4032 names are now escaped in a simple scheme to avoid collision
4033 of userspace object names with kernel filenames. This work
4034 is preparation for making these objects relocatable in the
4035 cgroup tree, in order to allow easy resource partitioning of
4036 these objects without causing naming conflicts.
4038 * systemctl list-dependencies gained the new switches
4039 --plain, --reverse, --after and --before.
4041 * systemd-inhibit now shows the process name of processes that
4042 have taken an inhibitor lock.
4044 * nss-myhostname will now also resolve "localhost"
4045 implicitly. This makes /etc/hosts an optional file and
4046 nicely handles that on IPv6 ::1 maps to both "localhost" and
4049 * libsystemd-logind.so gained a new call
4050 sd_get_machine_names() to enumerate running containers and
4051 VMs (currently only supported by very new libvirt and
4052 nspawn). sd_login_monitor can now be used to watch
4053 VMs/containers coming and going.
4055 * .include is not allowed recursively anymore, and only in
4056 unit files. Usually it is better to use drop-in snippets in
4057 .d/*.conf anyway, as introduced with systemd 198.
4059 * systemd-analyze gained a new "critical-chain" command that
4060 determines the slowest chain of units run during system
4061 boot-up. It is very useful for tracking down where
4062 optimizing boot time is the most beneficial.
4064 * systemd will no longer allow manipulating service paths in
4065 the name=systemd:/system cgroup tree using ControlGroup= in
4066 units. (But is still fine with it in all other dirs.)
4068 * There's a new systemd-nspawn@.service service file that may
4069 be used to easily run nspawn containers as system
4070 services. With the container's root directory in
4071 /var/lib/container/foobar it is now sufficient to run
4072 "systemctl start systemd-nspawn@foobar.service" to boot it.
4074 * systemd-cgls gained a new parameter "--machine" to list only
4075 the processes within a certain container.
4077 * ConditionSecurity= now can check for "apparmor". We still
4078 are lacking checks for SMACK and IMA for this condition
4079 check though. Patches welcome!
4081 * A new configuration file /etc/systemd/sleep.conf has been
4082 added that may be used to configure which kernel operation
4083 systemd is supposed to execute when "suspend", "hibernate"
4084 or "hybrid-sleep" is requested. This makes the new kernel
4085 "freeze" state accessible to the user.
4087 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4088 the passed argument if applicable.
4090 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4091 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4092 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4093 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4094 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4095 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4096 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4101 * The output of 'systemctl list-jobs' got some polishing. The
4102 '--type=' argument may now be passed more than once. A new
4103 command 'systemctl list-sockets' has been added which shows
4104 a list of kernel sockets systemd is listening on with the
4105 socket units they belong to, plus the units these socket
4108 * The experimental libsystemd-bus library got substantial
4109 updates to work in conjunction with the (also experimental)
4110 kdbus kernel project. It works well enough to exchange
4111 messages with some sophistication. Note that kdbus is not
4112 ready yet, and the library is mostly an elaborate test case
4113 for now, and not installable.
4115 * systemd gained a new unit 'systemd-static-nodes.service'
4116 that generates static device nodes earlier during boot, and
4117 can run in conjunction with udev.
4119 * libsystemd-login gained a new call sd_pid_get_user_unit()
4120 to retrieve the user systemd unit a process is running
4121 in. This is useful for systems where systemd is used as
4124 * systemd-nspawn now places all containers in the new /machine
4125 top-level cgroup directory in the name=systemd
4126 hierarchy. libvirt will soon do the same, so that we get a
4127 uniform separation of /system, /user and /machine for system
4128 services, user processes and containers/virtual
4129 machines. This new cgroup hierarchy is also useful to stick
4130 stable names to specific container instances, which can be
4131 recognized later this way (this name may be controlled
4132 via systemd-nspawn's new -M switch). libsystemd-login also
4133 gained a new call sd_pid_get_machine_name() to retrieve the
4134 name of the container/VM a specific process belongs to.
4136 * bootchart can now store its data in the journal.
4138 * libsystemd-journal gained a new call
4139 sd_journal_add_conjunction() for AND expressions to the
4140 matching logic. This can be used to express more complex
4141 logical expressions.
4143 * journactl can now take multiple --unit= and --user-unit=
4146 * The cryptsetup logic now understands the "luks.key=" kernel
4147 command line switch for specifying a file to read the
4148 decryption key from. Also, if a configured key file is not
4149 found the tool will now automatically fall back to prompting
4152 * Python systemd.journal module was updated to wrap recently
4153 added functions from libsystemd-journal. The interface was
4154 changed to bring the low level interface in s.j._Reader
4155 closer to the C API, and the high level interface in
4156 s.j.Reader was updated to wrap and convert all data about
4159 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4160 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4161 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4162 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4163 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4164 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4168 * journalctl --update-catalog now understands a new --root=
4169 option to operate on catalogs found in a different root
4172 * During shutdown after systemd has terminated all running
4173 services a final killing loop kills all remaining left-over
4174 processes. We will now print the name of these processes
4175 when we send SIGKILL to them, since this usually indicates a
4178 * If /etc/crypttab refers to password files stored on
4179 configured mount points automatic dependencies will now be
4180 generated to ensure the specific mount is established first
4181 before the key file is attempted to be read.
4183 * 'systemctl status' will now show information about the
4184 network sockets a socket unit is listening on.
4186 * 'systemctl status' will also shown information about any
4187 drop-in configuration file for units. (Drop-In configuration
4188 files in this context are files such as
4189 /etc/systemd/systemd/foobar.service.d/*.conf)
4191 * systemd-cgtop now optionally shows summed up CPU times of
4192 cgroups. Press '%' while running cgtop to switch between
4193 percentage and absolute mode. This is useful to determine
4194 which cgroups use up the most CPU time over the entire
4195 runtime of the system. systemd-cgtop has also been updated
4196 to be 'pipeable' for processing with further shell tools.
4198 * 'hostnamectl set-hostname' will now allow setting of FQDN
4201 * The formatting and parsing of time span values has been
4202 changed. The parser now understands fractional expressions
4203 such as "5.5h". The formatter will now output fractional
4204 expressions for all time spans under 1min, i.e. "5.123456s"
4205 rather than "5s 123ms 456us". For time spans under 1s
4206 millisecond values are shown, for those under 1ms
4207 microsecond values are shown. This should greatly improve
4208 all time-related output of systemd.
4210 * libsystemd-login and libsystemd-journal gained new
4211 functions for querying the poll() events mask and poll()
4212 timeout value for integration into arbitrary event
4215 * localectl gained the ability to list available X11 keymaps
4216 (models, layouts, variants, options).
4218 * 'systemd-analyze dot' gained the ability to filter for
4219 specific units via shell-style globs, to create smaller,
4220 more useful graphs. I.e. it is now possible to create simple
4221 graphs of all the dependencies between only target units, or
4222 of all units that Avahi has dependencies with.
4224 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4225 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4226 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4227 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4228 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4229 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4230 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4234 * The boot-time readahead implementation for rotating media
4235 will now read the read-ahead data in multiple passes which
4236 consist of all read requests made in equidistant time
4237 intervals. This means instead of strictly reading read-ahead
4238 data in its physical order on disk we now try to find a
4239 middle ground between physical and access time order.
4241 * /etc/os-release files gained a new BUILD_ID= field for usage
4242 on operating systems that provide continuous builds of OS
4245 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4246 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4247 William Douglas, Zbigniew Jędrzejewski-Szmek
4251 * systemd-python gained an API exposing libsystemd-daemon.
4253 * The SMACK setup logic gained support for uploading CIPSO
4256 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4257 ReadOnlyDirectories= and InaccessibleDirectories= has
4258 changed. The private /tmp and /var/tmp directories are now
4259 shared by all processes of a service (which means
4260 ExecStartPre= may now leave data in /tmp that ExecStart= of
4261 the same service can still access). When a service is
4262 stopped its temporary directories are immediately deleted
4263 (normal clean-up with tmpfiles is still done in addition to
4266 * By default, systemd will now set a couple of sysctl
4267 variables in the kernel: the safe sysrq options are turned
4268 on, IP route verification is turned on, and source routing
4269 disabled. The recently added hardlink and softlink
4270 protection of the kernel is turned on. These settings should
4271 be reasonably safe, and good defaults for all new systems.
4273 * The predictable network naming logic may now be turned off
4274 with a new kernel command line switch: net.ifnames=0.
4276 * A new libsystemd-bus module has been added that implements a
4277 pretty complete D-Bus client library. For details see:
4279 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4281 * journald will now explicitly flush the journal files to disk
4282 at the latest 5min after each write. The file will then also
4283 be marked offline until the next write. This should increase
4284 reliability in case of a crash. The synchronization delay
4285 can be configured via SyncIntervalSec= in journald.conf.
4287 * There's a new remote-fs-setup.target unit that can be used
4288 to pull in specific services when at least one remote file
4289 system is to be mounted.
4291 * There are new targets timers.target and paths.target as
4292 canonical targets to pull user timer and path units in
4293 from. This complements sockets.target with a similar
4294 purpose for socket units.
4296 * libudev gained a new call udev_device_set_attribute_value()
4297 to set sysfs attributes of a device.
4299 * The udev daemon now sets the default number of worker
4300 processes executed in parallel based on the number of available
4301 CPUs instead of the amount of available RAM. This is supposed
4302 to provide a more reliable default and limit a too aggressive
4303 paralellism for setups with 1000s of devices connected.
4305 Contributions from: Auke Kok, Colin Walters, Cristian
4306 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4307 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4308 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4309 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4310 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4311 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4312 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4313 Zbigniew Jędrzejewski-Szmek
4317 * Configuration of unit files may now be extended via drop-in
4318 files without having to edit/override the unit files
4319 themselves. More specifically, if the administrator wants to
4320 change one value for a service file foobar.service he can
4321 now do so by dropping in a configuration snippet into
4322 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4323 will load all these snippets and apply them on top of the
4324 main unit configuration file, possibly extending or
4325 overriding its settings. Using these drop-in snippets is
4326 generally nicer than the two earlier options for changing
4327 unit files locally: copying the files from
4328 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4329 them there; or creating a new file in /etc/systemd/system/
4330 that incorporates the original one via ".include". Drop-in
4331 snippets into these .d/ directories can be placed in any
4332 directory systemd looks for units in, and the usual
4333 overriding semantics between /usr/lib, /etc and /run apply
4336 * Most unit file settings which take lists of items can now be
4337 reset by assigning the empty string to them. For example,
4338 normally, settings such as Environment=FOO=BAR append a new
4339 environment variable assignment to the environment block,
4340 each time they are used. By assigning Environment= the empty
4341 string the environment block can be reset to empty. This is
4342 particularly useful with the .d/*.conf drop-in snippets
4343 mentioned above, since this adds the ability to reset list
4344 settings from vendor unit files via these drop-ins.
4346 * systemctl gained a new "list-dependencies" command for
4347 listing the dependencies of a unit recursively.
4349 * Inhibitors are now honored and listed by "systemctl
4350 suspend", "systemctl poweroff" (and similar) too, not only
4351 GNOME. These commands will also list active sessions by
4354 * Resource limits (as exposed by the various control group
4355 controllers) can now be controlled dynamically at runtime
4356 for all units. More specifically, you can now use a command
4357 like "systemctl set-cgroup-attr foobar.service cpu.shares
4358 2000" to alter the CPU shares a specific service gets. These
4359 settings are stored persistently on disk, and thus allow the
4360 administrator to easily adjust the resource usage of
4361 services with a few simple commands. This dynamic resource
4362 management logic is also available to other programs via the
4363 bus. Almost any kernel cgroup attribute and controller is
4366 * systemd-vconsole-setup will now copy all font settings to
4367 all allocated VTs, where it previously applied them only to
4370 * libsystemd-login gained the new sd_session_get_tty() API
4373 * This release drops support for a few legacy or
4374 distribution-specific LSB facility names when parsing init
4375 scripts: $x-display-manager, $mail-transfer-agent,
4376 $mail-transport-agent, $mail-transfer-agent, $smtp,
4377 $null. Also, the mail-transfer-agent.target unit backing
4378 this has been removed. Distributions which want to retain
4379 compatibility with this should carry the burden for
4380 supporting this themselves and patch support for these back
4381 in, if they really need to. Also, the facilities $syslog and
4382 $local_fs are now ignored, since systemd does not support
4383 early-boot LSB init scripts anymore, and these facilities
4384 are implied anyway for normal services. syslog.target has
4387 * There are new bus calls on PID1's Manager object for
4388 cancelling jobs, and removing snapshot units. Previously,
4389 both calls were only available on the Job and Snapshot
4392 * systemd-journal-gatewayd gained SSL support.
4394 * The various "environment" files, such as /etc/locale.conf
4395 now support continuation lines with a backslash ("\") as
4396 last character in the line, similarly in style (but different)
4397 to how this is supported in shells.
4399 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4400 now implicitly appended to every log entry logged. systemctl
4401 has been updated to filter by this field when operating on a
4402 user systemd instance.
4404 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4405 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4406 the container. This makes it easier to boot unmodified
4407 Fedora systems in a container, which however still requires
4408 audit=0 to be passed on the kernel command line. Auditing in
4409 kernel and userspace is unfortunately still too broken in
4410 context of containers, hence we recommend compiling it out
4411 of the kernel or using audit=0. Hopefully this will be fixed
4412 one day for good in the kernel.
4414 * nspawn gained the new --bind= and --bind-ro= parameters to
4415 bind mount specific directories from the host into the
4418 * nspawn will now mount its own devpts file system instance
4419 into the container, in order not to leak pty devices from
4420 the host into the container.
4422 * systemd will now read the firmware boot time performance
4423 information from the EFI variables, if the used boot loader
4424 supports this, and takes it into account for boot performance
4425 analysis via "systemd-analyze". This is currently supported
4426 only in conjunction with Gummiboot, but could be supported
4427 by other boot loaders too. For details see:
4429 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4431 * A new generator has been added that automatically mounts the
4432 EFI System Partition (ESP) to /boot, if that directory
4433 exists, is empty, and no other file system has been
4434 configured to be mounted there.
4436 * logind will now send out PrepareForSleep(false) out
4437 unconditionally, after coming back from suspend. This may be
4438 used by applications as asynchronous notification for
4439 system resume events.
4441 * "systemctl unlock-sessions" has been added, that allows
4442 unlocking the screens of all user sessions at once, similar
4443 to how "systemctl lock-sessions" already locked all users
4444 sessions. This is backed by a new D-Bus call UnlockSessions().
4446 * "loginctl seat-status" will now show the master device of a
4447 seat. (i.e. the device of a seat that needs to be around for
4448 the seat to be considered available, usually the graphics
4451 * tmpfiles gained a new "X" line type, that allows
4452 configuration of files and directories (with wildcards) that
4453 shall be excluded from automatic cleanup ("aging").
4455 * udev default rules set the device node permissions now only
4456 at "add" events, and do not change them any longer with a
4457 later "change" event.
4459 * The log messages for lid events and power/sleep keypresses
4460 now carry a message ID.
4462 * We now have a substantially larger unit test suite, but this
4463 continues to be work in progress.
4465 * udevadm hwdb gained a new --root= parameter to change the
4466 root directory to operate relative to.
4468 * logind will now issue a background sync() request to the kernel
4469 early at shutdown, so that dirty buffers are flushed to disk early
4470 instead of at the last moment, in order to optimize shutdown
4473 * A new bootctl tool has been added that is an interface for
4474 certain boot loader operations. This is currently a preview
4475 and is likely to be extended into a small mechanism daemon
4476 like timedated, localed, hostnamed, and can be used by
4477 graphical UIs to enumerate available boot options, and
4478 request boot into firmware operations.
4480 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4481 the rest of the package. It also has been updated to work
4482 correctly in initrds.
4484 * Policykit previously has been runtime optional, and is now
4485 also compile time optional via a configure switch.
4487 * systemd-analyze has been reimplemented in C. Also "systemctl
4488 dot" has moved into systemd-analyze.
4490 * "systemctl status" with no further parameters will now print
4491 the status of all active or failed units.
4493 * Operations such as "systemctl start" can now be executed
4494 with a new mode "--irreversible" which may be used to queue
4495 operations that cannot accidentally be reversed by a later
4496 job queuing. This is by default used to make shutdown
4497 requests more robust.
4499 * The Python API of systemd now gained a new module for
4500 reading journal files.
4502 * A new tool kernel-install has been added that can install
4503 kernel images according to the Boot Loader Specification:
4505 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4507 * Boot time console output has been improved to provide
4508 animated boot time output for hanging jobs.
4510 * A new tool systemd-activate has been added which can be used
4511 to test socket activation with, directly from the command
4512 line. This should make it much easier to test and debug
4513 socket activation in daemons.
4515 * journalctl gained a new "--reverse" (or -r) option to show
4516 journal output in reverse order (i.e. newest line first).
4518 * journalctl gained a new "--pager-end" (or -e) option to jump
4519 to immediately jump to the end of the journal in the
4520 pager. This is only supported in conjunction with "less".
4522 * journalctl gained a new "--user-unit=" option, that works
4523 similarly to "--unit=" but filters for user units rather than
4526 * A number of unit files to ease adoption of systemd in
4527 initrds has been added. This moves some minimal logic from
4528 the various initrd implementations into systemd proper.
4530 * The journal files are now owned by a new group
4531 "systemd-journal", which exists specifically to allow access
4532 to the journal, and nothing else. Previously, we used the
4533 "adm" group for that, which however possibly covers more
4534 than just journal/log file access. This new group is now
4535 already used by systemd-journal-gatewayd to ensure this
4536 daemon gets access to the journal files and as little else
4537 as possible. Note that "make install" will also set FS ACLs
4538 up for /var/log/journal to give "adm" and "wheel" read
4539 access to it, in addition to "systemd-journal" which owns
4540 the journal files. We recommend that packaging scripts also
4541 add read access to "adm" + "wheel" to /var/log/journal, and
4542 all existing/future journal files. To normal users and
4543 administrators little changes, however packagers need to
4544 ensure to create the "systemd-journal" system group at
4545 package installation time.
4547 * The systemd-journal-gatewayd now runs as unprivileged user
4548 systemd-journal-gateway:systemd-journal-gateway. Packaging
4549 scripts need to create these system user/group at
4552 * timedated now exposes a new boolean property CanNTP that
4553 indicates whether a local NTP service is available or not.
4555 * systemd-detect-virt will now also detect xen PVs
4557 * The pstore file system is now mounted by default, if it is
4560 * In addition to the SELinux and IMA policies we will now also
4561 load SMACK policies at early boot.
4563 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4564 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4565 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4566 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4567 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4568 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4569 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4570 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4571 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4572 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4573 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4574 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4575 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4576 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4580 * Timer units now support calendar time events in addition to
4581 monotonic time events. That means you can now trigger a unit
4582 based on a calendar time specification such as "Thu,Fri
4583 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4584 or fifth day of any month of the year 2013, given that it is
4585 a thursday or friday. This brings timer event support
4586 considerably closer to cron's capabilities. For details on
4587 the supported calendar time specification language see
4590 * udev now supports a number of different naming policies for
4591 network interfaces for predictable names, and a combination
4592 of these policies is now the default. Please see this wiki
4593 document for details:
4595 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4597 * Auke Kok's bootchart implementation has been added to the
4598 systemd tree. It is an optional component that can graph the
4599 boot in quite some detail. It is one of the best bootchart
4600 implementations around and minimal in its code and
4603 * nss-myhostname has been integrated into the systemd source
4604 tree. nss-myhostname guarantees that the local hostname
4605 always stays resolvable via NSS. It has been a weak
4606 requirement of systemd-hostnamed since a long time, and
4607 since its code is actually trivial we decided to just
4608 include it in systemd's source tree. It can be turned off
4609 with a configure switch.
4611 * The read-ahead logic is now capable of properly detecting
4612 whether a btrfs file system is on SSD or rotating media, in
4613 order to optimize the read-ahead scheme. Previously, it was
4614 only capable of detecting this on traditional file systems
4617 * In udev, additional device properties are now read from the
4618 IAB in addition to the OUI database. Also, Bluetooth company
4619 identities are attached to the devices as well.
4621 * In service files %U may be used as specifier that is
4622 replaced by the configured user name of the service.
4624 * nspawn may now be invoked without a controlling TTY. This
4625 makes it suitable for invocation as its own service. This
4626 may be used to set up a simple containerized server system
4627 using only core OS tools.
4629 * systemd and nspawn can now accept socket file descriptors
4630 when they are started for socket activation. This enables
4631 implementation of socket activated nspawn
4632 containers. i.e. think about autospawning an entire OS image
4633 when the first SSH or HTTP connection is received. We expect
4634 that similar functionality will also be added to libvirt-lxc
4637 * journalctl will now suppress ANSI color codes when
4638 presenting log data.
4640 * systemctl will no longer show control group information for
4641 a unit if a the control group is empty anyway.
4643 * logind can now automatically suspend/hibernate/shutdown the
4646 * /etc/machine-info and hostnamed now also expose the chassis
4647 type of the system. This can be used to determine whether
4648 the local system is a laptop, desktop, handset or
4649 tablet. This information may either be configured by the
4650 user/vendor or is automatically determined from ACPI and DMI
4651 information if possible.
4653 * A number of PolicyKit actions are now bound together with
4654 "imply" rules. This should simplify creating UIs because
4655 many actions will now authenticate similar ones as well.
4657 * Unit files learnt a new condition ConditionACPower= which
4658 may be used to conditionalize a unit depending on whether an
4659 AC power source is connected or not, of whether the system
4660 is running on battery power.
4662 * systemctl gained a new "is-failed" verb that may be used in
4663 shell scripts and suchlike to check whether a specific unit
4664 is in the "failed" state.
4666 * The EnvironmentFile= setting in unit files now supports file
4667 globbing, and can hence be used to easily read a number of
4668 environment files at once.
4670 * systemd will no longer detect and recognize specific
4671 distributions. All distribution-specific #ifdeffery has been
4672 removed, systemd is now fully generic and
4673 distribution-agnostic. Effectively, not too much is lost as
4674 a lot of the code is still accessible via explicit configure
4675 switches. However, support for some distribution specific
4676 legacy configuration file formats has been dropped. We
4677 recommend distributions to simply adopt the configuration
4678 files everybody else uses now and convert the old
4679 configuration from packaging scripts. Most distributions
4680 already did that. If that's not possible or desirable,
4681 distributions are welcome to forward port the specific
4682 pieces of code locally from the git history.
4684 * When logging a message about a unit systemd will now always
4685 log the unit name in the message meta data.
4687 * localectl will now also discover system locale data that is
4688 not stored in locale archives, but directly unpacked.
4690 * logind will no longer unconditionally use framebuffer
4691 devices as seat masters, i.e. as devices that are required
4692 to be existing before a seat is considered preset. Instead,
4693 it will now look for all devices that are tagged as
4694 "seat-master" in udev. By default, framebuffer devices will
4695 be marked as such, but depending on local systems, other
4696 devices might be marked as well. This may be used to
4697 integrate graphics cards using closed source drivers (such
4698 as NVidia ones) more nicely into logind. Note however, that
4699 we recommend using the open source NVidia drivers instead,
4700 and no udev rules for the closed-source drivers will be
4701 shipped from us upstream.
4703 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4704 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4705 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4706 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4707 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4708 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4709 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4710 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4711 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4712 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4713 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4714 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4719 * udev gained support for loading additional device properties
4720 from an indexed database that is keyed by vendor/product IDs
4721 and similar device identifiers. For the beginning this
4722 "hwdb" is populated with data from the well-known PCI and
4723 USB database, but also includes PNP, ACPI and OID data. In
4724 the longer run this indexed database shall grow into
4725 becoming the one central database for non-essential
4726 userspace device metadata. Previously, data from the PCI/USB
4727 database was only attached to select devices, since the
4728 lookup was a relatively expensive operation due to O(n) time
4729 complexity (with n being the number of entries in the
4730 database). Since this is now O(1), we decided to add in this
4731 data for all devices where this is available, by
4732 default. Note that the indexed database needs to be rebuilt
4733 when new data files are installed. To achieve this you need
4734 to update your packaging scripts to invoke "udevadm hwdb
4735 --update" after installation of hwdb data files. For
4736 RPM-based distributions we introduced the new
4737 %udev_hwdb_update macro for this purpose.
4739 * The Journal gained support for the "Message Catalog", an
4740 indexed database to link up additional information with
4741 journal entries. For further details please check:
4743 http://www.freedesktop.org/wiki/Software/systemd/catalog
4745 The indexed message catalog database also needs to be
4746 rebuilt after installation of message catalog files. Use
4747 "journalctl --update-catalog" for this. For RPM-based
4748 distributions we introduced the %journal_catalog_update
4749 macro for this purpose.
4751 * The Python Journal bindings gained support for the standard
4752 Python logging framework.
4754 * The Journal API gained new functions for checking whether
4755 the underlying file system of a journal file is capable of
4756 properly reporting file change notifications, or whether
4757 applications that want to reflect journal changes "live"
4758 need to recheck journal files continuously in appropriate
4761 * It is now possible to set the "age" field for tmpfiles
4762 entries to 0, indicating that files matching this entry
4763 shall always be removed when the directories are cleaned up.
4765 * coredumpctl gained a new "gdb" verb which invokes gdb
4766 right-away on the selected coredump.
4768 * There's now support for "hybrid sleep" on kernels that
4769 support this, in addition to "suspend" and "hibernate". Use
4770 "systemctl hybrid-sleep" to make use of this.
4772 * logind's HandleSuspendKey= setting (and related settings)
4773 now gained support for a new "lock" setting to simply
4774 request the screen lock on all local sessions, instead of
4775 actually executing a suspend or hibernation.
4777 * systemd will now mount the EFI variables file system by
4780 * Socket units now gained support for configuration of the
4781 SMACK security label.
4783 * timedatectl will now output the time of the last and next
4784 daylight saving change.
4786 * We dropped support for various legacy and distro-specific
4787 concepts, such as insserv, early-boot SysV services
4788 (i.e. those for non-standard runlevels such as 'b' or 'S')
4789 or ArchLinux /etc/rc.conf support. We recommend the
4790 distributions who still need support this to either continue
4791 to maintain the necessary patches downstream, or find a
4792 different solution. (Talk to us if you have questions!)
4794 * Various systemd components will now bypass PolicyKit checks
4795 for root and otherwise handle properly if PolicyKit is not
4796 found to be around. This should fix most issues for
4797 PolicyKit-less systems. Quite frankly this should have been
4798 this way since day one. It is absolutely our intention to
4799 make systemd work fine on PolicyKit-less systems, and we
4800 consider it a bug if something does not work as it should if
4801 PolicyKit is not around.
4803 * For embedded systems it is now possible to build udev and
4804 systemd without blkid and/or kmod support.
4806 * "systemctl switch-root" is now capable of switching root
4807 more than once. I.e. in addition to transitions from the
4808 initrd to the host OS it is now possible to transition to
4809 further OS images from the host. This is useful to implement
4810 offline updating tools.
4812 * Various other additions have been made to the RPM macros
4813 shipped with systemd. Use %udev_rules_update() after
4814 installing new udev rules files. %_udevhwdbdir,
4815 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4816 %_sysctldir are now available which resolve to the right
4817 directories for packages to place various data files in.
4819 * journalctl gained the new --full switch (in addition to
4820 --all, to disable ellipsation for long messages.
4822 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4823 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4824 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4825 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4826 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4827 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4828 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4829 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4830 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4834 * journalctl gained new --since= and --until= switches to
4835 filter by time. It also now supports nice filtering for
4836 units via --unit=/-u.
4838 * Type=oneshot services may use ExecReload= and do the
4841 * The journal daemon now supports time-based rotation and
4842 vacuuming, in addition to the usual disk-space based
4845 * The journal will now index the available field values for
4846 each field name. This enables clients to show pretty drop
4847 downs of available match values when filtering. The bash
4848 completion of journalctl has been updated
4849 accordingly. journalctl gained a new switch -F to list all
4850 values a certain field takes in the journal database.
4852 * More service events are now written as structured messages
4853 to the journal, and made recognizable via message IDs.
4855 * The timedated, localed and hostnamed mini-services which
4856 previously only provided support for changing time, locale
4857 and hostname settings from graphical DEs such as GNOME now
4858 also have a minimal (but very useful) text-based client
4859 utility each. This is probably the nicest way to changing
4860 these settings from the command line now, especially since
4861 it lists available options and is fully integrated with bash
4864 * There's now a new tool "systemd-coredumpctl" to list and
4865 extract coredumps from the journal.
4867 * We now install a README each in /var/log/ and
4868 /etc/rc.d/init.d explaining where the system logs and init
4869 scripts went. This hopefully should help folks who go to
4870 that dirs and look into the otherwise now empty void and
4871 scratch their heads.
4873 * When user-services are invoked (by systemd --user) the
4874 $MANAGERPID env var is set to the PID of systemd.
4876 * SIGRTMIN+24 when sent to a --user instance will now result
4877 in immediate termination of systemd.
4879 * gatewayd received numerous feature additions such as a
4880 "follow" mode, for live syncing and filtering.
4882 * browse.html now allows filtering and showing detailed
4883 information on specific entries. Keyboard navigation and
4884 mouse screen support has been added.
4886 * gatewayd/journalctl now supports HTML5/JSON
4887 Server-Sent-Events as output.
4889 * The SysV init script compatibility logic will now
4890 heuristically determine whether a script supports the
4891 "reload" verb, and only then make this available as
4894 * "systemctl status --follow" has been removed, use "journalctl
4897 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4898 have been removed since they are hardly useful to be
4901 * And I'd like to take the opportunity to specifically mention
4902 Zbigniew for his great contributions. Zbigniew, you rock!
4904 Contributions from: Andrew Eikum, Christian Hesse, Colin
4905 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4906 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4907 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4908 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4909 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4910 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4914 * If /etc/vconsole.conf is non-existent or empty we will no
4915 longer load any console font or key map at boot by
4916 default. Instead the kernel defaults will be left
4917 intact. This is definitely the right thing to do, as no
4918 configuration should mean no configuration, and hard-coding
4919 font names that are different on all archs is probably a bad
4920 idea. Also, the kernel default key map and font should be
4921 good enough for most cases anyway, and mostly identical to
4922 the userspace fonts/key maps we previously overloaded them
4923 with. If distributions want to continue to default to a
4924 non-kernel font or key map they should ship a default
4925 /etc/vconsole.conf with the appropriate contents.
4927 Contributions from: Colin Walters, Daniel J Walsh, Dave
4928 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4929 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4933 * journalctl gained a new --cursor= switch to show entries
4934 starting from the specified location in the journal.
4936 * We now enforce a size limit on journal entry fields exported
4937 with "-o json" in journalctl. Fields larger than 4K will be
4938 assigned null. This can be turned off with --all.
4940 * An (optional) journal gateway daemon is now available as
4941 "systemd-journal-gatewayd.service". This service provides
4942 access to the journal via HTTP and JSON. This functionality
4943 will be used to implement live log synchronization in both
4944 pull and push modes, but has various other users too, such
4945 as easy log access for debugging of embedded devices. Right
4946 now it is already useful to retrieve the journal via HTTP:
4948 # systemctl start systemd-journal-gatewayd.service
4949 # wget http://localhost:19531/entries
4951 This will download the journal contents in a
4952 /var/log/messages compatible format. The same as JSON:
4954 # curl -H"Accept: application/json" http://localhost:19531/entries
4956 This service is also accessible via a web browser where a
4957 single static HTML5 app is served that uses the JSON logic
4958 to enable the user to do some basic browsing of the
4959 journal. This will be extended later on. Here's an example
4960 screenshot of this app in its current state:
4962 http://0pointer.de/public/journal-gatewayd
4964 Contributions from: Kay Sievers, Lennart Poettering, Robert
4965 Milasan, Tom Gundersen
4969 * The bash completion logic is now available for journalctl
4972 * We do not mount the "cpuset" controller anymore together with
4973 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4974 started if no parameters are assigned to it. "cpuset" hence
4975 broke code that assumed it it could create "cpu" groups and
4978 * journalctl -f will now subscribe to terminal size changes,
4979 and line break accordingly.
4981 Contributions from: Dave Reisner, Kay Sievers, Lennart
4982 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4986 * nspawn will now create a symlink /etc/localtime in the
4987 container environment, copying the host's timezone
4988 setting. Previously this has been done via a bind mount, but
4989 since symlinks cannot be bind mounted this has now been
4990 changed to create/update the appropriate symlink.
4992 * journalctl -n's line number argument is now optional, and
4993 will default to 10 if omitted.
4995 * journald will now log the maximum size the journal files may
4996 take up on disk. This is particularly useful if the default
4997 built-in logic of determining this parameter from the file
4998 system size is used. Use "systemctl status
4999 systemd-journald.service" to see this information.
5001 * The multi-seat X wrapper tool has been stripped down. As X
5002 is now capable of enumerating graphics devices via udev in a
5003 seat-aware way the wrapper is not strictly necessary
5004 anymore. A stripped down temporary stop-gap is still shipped
5005 until the upstream display managers have been updated to
5006 fully support the new X logic. Expect this wrapper to be
5007 removed entirely in one of the next releases.
5009 * HandleSleepKey= in logind.conf has been split up into
5010 HandleSuspendKey= and HandleHibernateKey=. The old setting
5011 is not available anymore. X11 and the kernel are
5012 distinguishing between these keys and we should too. This
5013 also means the inhibition lock for these keys has been split
5016 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5017 Poettering, Lukas Nykryn, Václav Pavlín
5021 * Whenever a unit changes state we will now log this to the
5022 journal and show along the unit's own log output in
5025 * ConditionPathIsMountPoint= can now properly detect bind
5026 mount points too. (Previously, a bind mount of one file
5027 system to another place in the same file system could not be
5028 detected as mount, since they shared struct stat's st_dev
5031 * We will now mount the cgroup controllers cpu, cpuacct,
5032 cpuset and the controllers net_cls, net_prio together by
5035 * nspawn containers will now have a virtualized boot
5036 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5037 over with a randomized ID at container initialization). This
5038 has the effect of making "journalctl -b" do the right thing
5041 * The JSON output journal serialization has been updated not
5042 to generate "endless" list objects anymore, but rather one
5043 JSON object per line. This is more in line how most JSON
5044 parsers expect JSON objects. The new output mode
5045 "json-pretty" has been added to provide similar output, but
5046 neatly aligned for readability by humans.
5048 * We dropped all explicit sync() invocations in the shutdown
5049 code. The kernel does this implicitly anyway in the kernel
5050 reboot() syscall. halt(8)'s -n option is now a compatibility
5053 * We now support virtualized reboot() in containers, as
5054 supported by newer kernels. We will fall back to exit() if
5055 CAP_SYS_REBOOT is not available to the container. Also,
5056 nspawn makes use of this now and will actually reboot the
5057 container if the containerized OS asks for that.
5059 * journalctl will only show local log output by default
5060 now. Use --merge (-m) to show remote log output, too.
5062 * libsystemd-journal gained the new sd_journal_get_usage()
5063 call to determine the current disk usage of all journal
5064 files. This is exposed in the new "journalctl --disk-usage"
5067 * journald gained a new configuration setting SplitMode= in
5068 journald.conf which may be used to control how user journals
5069 are split off. See journald.conf(5) for details.
5071 * A new condition type ConditionFileNotEmpty= has been added.
5073 * tmpfiles' "w" lines now support file globbing, to write
5074 multiple files at once.
5076 * We added Python bindings for the journal submission
5077 APIs. More Python APIs for a number of selected APIs will
5078 likely follow. Note that we intend to add native bindings
5079 only for the Python language, as we consider it common
5080 enough to deserve bindings shipped within systemd. There are
5081 various projects outside of systemd that provide bindings
5082 for languages such as PHP or Lua.
5084 * Many conditions will now resolve specifiers such as %i. In
5085 addition, PathChanged= and related directives of .path units
5086 now support specifiers as well.
5088 * There's now a new RPM macro definition for the system preset
5091 * journald will now warn if it ca not forward a message to the
5092 syslog daemon because its socket is full.
5094 * timedated will no longer write or process /etc/timezone,
5095 except on Debian. As we do not support late mounted /usr
5096 anymore /etc/localtime always being a symlink is now safe,
5097 and hence the information in /etc/timezone is not necessary
5100 * logind will now always reserve one VT for a text getty (VT6
5101 by default). Previously if more than 6 X sessions where
5102 started they took up all the VTs with auto-spawned gettys,
5103 so that no text gettys were available anymore.
5105 * udev will now automatically inform the btrfs kernel logic
5106 about btrfs RAID components showing up. This should make
5107 simple hotplug based btrfs RAID assembly work.
5109 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5110 (but not for its children which will stay at the kernel
5111 default). This should allow setups with a lot more listening
5114 * systemd will now always pass the configured timezone to the
5115 kernel at boot. timedated will do the same when the timezone
5118 * logind's inhibition logic has been updated. By default,
5119 logind will now handle the lid switch, the power and sleep
5120 keys all the time, even in graphical sessions. If DEs want
5121 to handle these events on their own they should take the new
5122 handle-power-key, handle-sleep-key and handle-lid-switch
5123 inhibitors during their runtime. A simple way to achieve
5124 that is to invoke the DE wrapped in an invocation of:
5126 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5128 * Access to unit operations is now checked via SELinux taking
5129 the unit file label and client process label into account.
5131 * systemd will now notify the administrator in the journal
5132 when he over-mounts a non-empty directory.
5134 * There are new specifiers that are resolved in unit files,
5135 for the host name (%H), the machine ID (%m) and the boot ID
5138 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5139 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5140 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5141 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5142 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5143 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5144 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5148 * Support for reading structured kernel messages from
5149 /dev/kmsg has now been added and is enabled by default.
5151 * Support for reading kernel messages from /proc/kmsg has now
5152 been removed. If you want kernel messages in the journal
5153 make sure to run a recent kernel (>= 3.5) that supports
5154 reading structured messages from /dev/kmsg (see
5155 above). /proc/kmsg is now exclusive property of classic
5156 syslog daemons again.
5158 * The libudev API gained the new
5159 udev_device_new_from_device_id() call.
5161 * The logic for file system namespace (ReadOnlyDirectory=,
5162 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5163 require pivot_root() anymore. This means fewer temporary
5164 directories are created below /tmp for this feature.
5166 * nspawn containers will now see and receive all submounts
5167 made on the host OS below the root file system of the
5170 * Forward Secure Sealing is now supported for Journal files,
5171 which provide cryptographical sealing of journal files so
5172 that attackers cannot alter log history anymore without this
5173 being detectable. Lennart will soon post a blog story about
5174 this explaining it in more detail.
5176 * There are two new service settings RestartPreventExitStatus=
5177 and SuccessExitStatus= which allow configuration of exit
5178 status (exit code or signal) which will be excepted from the
5179 restart logic, resp. consider successful.
5181 * journalctl gained the new --verify switch that can be used
5182 to check the integrity of the structure of journal files and
5183 (if Forward Secure Sealing is enabled) the contents of
5186 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5187 and similar symlinks pre-created. This makes running shells
5188 as container init process a lot more fun.
5190 * The fstab support can now handle PARTUUID= and PARTLABEL=
5193 * A new ConditionHost= condition has been added to match
5194 against the hostname (with globs) and machine ID. This is
5195 useful for clusters where a single OS image is used to
5196 provision a large number of hosts which shall run slightly
5197 different sets of services.
5199 * Services which hit the restart limit will now be placed in a
5202 Contributions from: Bertram Poettering, Dave Reisner, Huang
5203 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5204 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5208 * When running in --user mode systemd will now become a
5209 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5210 tree a lot more organized.
5212 * A new PartOf= unit dependency type has been introduced that
5213 may be used to group services in a natural way.
5215 * "systemctl enable" may now be used to enable instances of
5218 * journalctl now prints error log levels in red, and
5219 warning/notice log levels in bright white. It also supports
5220 filtering by log level now.
5222 * cgtop gained a new -n switch (similar to top), to configure
5223 the maximum number of iterations to run for. It also gained
5224 -b, to run in batch mode (accepting no input).
5226 * The suffix ".service" may now be omitted on most systemctl
5227 command lines involving service unit names.
5229 * There's a new bus call in logind to lock all sessions, as
5230 well as a loginctl verb for it "lock-sessions".
5232 * libsystemd-logind.so gained a new call sd_journal_perror()
5233 that works similar to libc perror() but logs to the journal
5234 and encodes structured information about the error number.
5236 * /etc/crypttab entries now understand the new keyfile-size=
5239 * shutdown(8) now can send a (configurable) wall message when
5240 a shutdown is cancelled.
5242 * The mount propagation mode for the root file system will now
5243 default to "shared", which is useful to make containers work
5244 nicely out-of-the-box so that they receive new mounts from
5245 the host. This can be undone locally by running "mount
5246 --make-rprivate /" if needed.
5248 * The prefdm.service file has been removed. Distributions
5249 should maintain this unit downstream if they intend to keep
5250 it around. However, we recommend writing normal unit files
5251 for display managers instead.
5253 * Since systemd is a crucial part of the OS we will now
5254 default to a number of compiler switches that improve
5255 security (hardening) such as read-only relocations, stack
5256 protection, and suchlike.
5258 * The TimeoutSec= setting for services is now split into
5259 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5260 of individual time outs for the start and the stop phase of
5263 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5264 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5265 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5266 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5267 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5268 Gundersen, Zbigniew Jędrzejewski-Szmek
5272 * The journal and id128 C APIs are now fully documented as man
5275 * Extra safety checks have been added when transitioning from
5276 the initial RAM disk to the main system to avoid accidental
5279 * /etc/crypttab entries now understand the new keyfile-offset=
5282 * systemctl -t can now be used to filter by unit load state.
5284 * The journal C API gained the new sd_journal_wait() call to
5285 make writing synchronous journal clients easier.
5287 * journalctl gained the new -D switch to show journals from a
5290 * journalctl now displays a special marker between log
5291 messages of two different boots.
5293 * The journal is now explicitly flushed to /var via a service
5294 systemd-journal-flush.service, rather than implicitly simply
5295 by seeing /var/log/journal to be writable.
5297 * journalctl (and the journal C APIs) can now match for much
5298 more complex expressions, with alternatives and
5301 * When transitioning from the initial RAM disk to the main
5302 system we will now kill all processes in a killing spree to
5303 ensure no processes stay around by accident.
5305 * Three new specifiers may be used in unit files: %u, %h, %s
5306 resolve to the user name, user home directory resp. user
5307 shell. This is useful for running systemd user instances.
5309 * We now automatically rotate journal files if their data
5310 object hash table gets a fill level > 75%. We also size the
5311 hash table based on the configured maximum file size. This
5312 together should lower hash collisions drastically and thus
5313 speed things up a bit.
5315 * journalctl gained the new "--header" switch to introspect
5316 header data of journal files.
5318 * A new setting SystemCallFilters= has been added to services
5319 which may be used to apply blacklists or whitelists to
5320 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5322 * nspawn gained a new --link-journal= switch (and quicker: -j)
5323 to link the container journal with the host. This makes it
5324 very easy to centralize log viewing on the host for all
5325 guests while still keeping the journal files separated.
5327 * Many bugfixes and optimizations
5329 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5330 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5331 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5336 * Several tools now understand kernel command line arguments,
5337 which are only read when run in an initial RAM disk. They
5338 usually follow closely their normal counterparts, but are
5341 * There's a new tool to analyze the readahead files that are
5342 automatically generated at boot. Use:
5344 /usr/lib/systemd/systemd-readahead analyze /.readahead
5346 * We now provide an early debug shell on tty9 if this enabled. Use:
5348 systemctl enable debug-shell.service
5350 * All plymouth related units have been moved into the Plymouth
5351 package. Please make sure to upgrade your Plymouth version
5354 * systemd-tmpfiles now supports getting passed the basename of
5355 a configuration file only, in which case it will look for it
5356 in all appropriate directories automatically.
5358 * udevadm info now takes a /dev or /sys path as argument, and
5359 does the right thing. Example:
5361 udevadm info /dev/sda
5362 udevadm info /sys/class/block/sda
5364 * systemctl now prints a warning if a unit is stopped but a
5365 unit that might trigger it continues to run. Example: a
5366 service is stopped but the socket that activates it is left
5369 * "systemctl status" will now mention if the log output was
5370 shortened due to rotation since a service has been started.
5372 * The journal API now exposes functions to determine the
5373 "cutoff" times due to rotation.
5375 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5376 immediately flushing of runtime logs to /var if possible,
5377 resp. for triggering immediate rotation of the journal
5380 * It is now considered an error if a service is attempted to
5381 be stopped that is not loaded.
5383 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5385 * systemd-analyze now supports Python 3
5387 * tmpfiles now supports cleaning up directories via aging
5388 where the first level dirs are always kept around but
5389 directories beneath it automatically aged. This is enabled
5390 by prefixing the age field with '~'.
5392 * Seat objects now expose CanGraphical, CanTTY properties
5393 which is required to deal with very fast bootups where the
5394 display manager might be running before the graphics drivers
5395 completed initialization.
5397 * Seat objects now expose a State property.
5399 * We now include RPM macros for service enabling/disabling
5400 based on the preset logic. We recommend RPM based
5401 distributions to make use of these macros if possible. This
5402 makes it simpler to reuse RPM spec files across
5405 * We now make sure that the collected systemd unit name is
5406 always valid when services log to the journal via
5409 * There's a new man page kernel-command-line(7) detailing all
5410 command line options we understand.
5412 * The fstab generator may now be disabled at boot by passing
5413 fstab=0 on the kernel command line.
5415 * A new kernel command line option modules-load= is now understood
5416 to load a specific kernel module statically, early at boot.
5418 * Unit names specified on the systemctl command line are now
5419 automatically escaped as needed. Also, if file system or
5420 device paths are specified they are automatically turned
5421 into the appropriate mount or device unit names. Example:
5423 systemctl status /home
5424 systemctl status /dev/sda
5426 * The SysVConsole= configuration option has been removed from
5427 system.conf parsing.
5429 * The SysV search path is no longer exported on the D-Bus
5432 * The Names= option is been removed from unit file parsing.
5434 * There's a new man page bootup(7) detailing the boot process.
5436 * Every unit and every generator we ship with systemd now
5437 comes with full documentation. The self-explanatory boot is
5440 * A couple of services gained "systemd-" prefixes in their
5441 name if they wrap systemd code, rather than only external
5442 code. Among them fsck@.service which is now
5443 systemd-fsck@.service.
5445 * The HaveWatchdog property has been removed from the D-Bus
5448 * systemd.confirm_spawn= on the kernel command line should now
5451 * There's a new man page crypttab(5) which details all options
5452 we actually understand.
5454 * systemd-nspawn gained a new --capability= switch to pass
5455 additional capabilities to the container.
5457 * timedated will now read known NTP implementation unit names
5458 from /usr/lib/systemd/ntp-units.d/*.list,
5459 systemd-timedated-ntp.target has been removed.
5461 * journalctl gained a new switch "-b" that lists log data of
5462 the current boot only.
5464 * The notify socket is in the abstract namespace again, in
5465 order to support daemons which chroot() at start-up.
5467 * There is a new Storage= configuration option for journald
5468 which allows configuration of where log data should go. This
5469 also provides a way to disable journal logging entirely, so
5470 that data collected is only forwarded to the console, the
5471 kernel log buffer or another syslog implementation.
5473 * Many bugfixes and optimizations
5475 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5476 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5477 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5478 Shawn Landden, Tom Gundersen
5482 * "systemctl help <unit>" now shows the man page if one is
5485 * Several new man pages have been added.
5487 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5488 MaxLevelConsole= can now be specified in
5489 journald.conf. These options allow reducing the amount of
5490 data stored on disk or forwarded by the log level.
5492 * TimerSlackNSec= can now be specified in system.conf for
5493 PID1. This allows system-wide power savings.
5495 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5496 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5501 * logind is now capable of (optionally) handling power and
5502 sleep keys as well as the lid switch.
5504 * journalctl now understands the syntax "journalctl
5505 /usr/bin/avahi-daemon" to get all log output of a specific
5508 * CapabilityBoundingSet= in system.conf now also influences
5509 the capability bound set of usermode helpers of the kernel.
5511 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5512 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5513 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5514 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5518 * Note that we skipped 139 releases here in order to set the
5519 new version to something that is greater than both udev's
5520 and systemd's most recent version number.
5522 * udev: all udev sources are merged into the systemd source tree now.
5523 All future udev development will happen in the systemd tree. It
5524 is still fully supported to use the udev daemon and tools without
5525 systemd running, like in initramfs or other init systems. Building
5526 udev though, will require the *build* of the systemd tree, but
5527 udev can be properly *run* without systemd.
5529 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5530 should be used to create dead device nodes as workarounds for broken
5533 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5534 no longer supported. udev_monitor_new_from_netlink() needs to be
5535 used to subscribe to events.
5537 * udev: when udevd is started by systemd, processes which are left
5538 behind by forking them off of udev rules, are unconditionally cleaned
5539 up and killed now after the event handling has finished. Services or
5540 daemons must be started as systemd services. Services can be
5541 pulled-in by udev to get started, but they can no longer be directly
5542 forked by udev rules.
5544 * udev: the daemon binary is called systemd-udevd now and installed
5545 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5546 to adapt to that, create symlink, or rename the binary after building
5549 * libudev no longer provides these symbols:
5550 udev_monitor_from_socket()
5551 udev_queue_get_failed_list_entry()
5552 udev_get_{dev,sys,run}_path()
5553 The versions number was bumped and symbol versioning introduced.
5555 * systemd-loginctl and systemd-journalctl have been renamed
5556 to loginctl and journalctl to match systemctl.
5558 * The config files: /etc/systemd/systemd-logind.conf and
5559 /etc/systemd/systemd-journald.conf have been renamed to
5560 logind.conf and journald.conf. Package updates should rename
5561 the files to the new names on upgrade.
5563 * For almost all files the license is now LGPL2.1+, changed
5564 from the previous GPL2.0+. Exceptions are some minor stuff
5565 of udev (which will be changed to LGPL2.1 eventually, too),
5566 and the MIT licensed sd-daemon.[ch] library that is suitable
5567 to be used as drop-in files.
5569 * systemd and logind now handle system sleep states, in
5570 particular suspending and hibernating.
5572 * logind now implements a sleep/shutdown/idle inhibiting logic
5573 suitable for a variety of uses. Soonishly Lennart will blog
5574 about this in more detail.
5576 * var-run.mount and var-lock.mount are no longer provided
5577 (which prevously bind mounted these directories to their new
5578 places). Distributions which have not converted these
5579 directories to symlinks should consider stealing these files
5580 from git history and add them downstream.
5582 * We introduced the Documentation= field for units and added
5583 this to all our shipped units. This is useful to make it
5584 easier to explore the boot and the purpose of the various
5587 * All smaller setup units (such as
5588 systemd-vconsole-setup.service) now detect properly if they
5589 are run in a container and are skipped when
5590 appropriate. This guarantees an entirely noise-free boot in
5591 Linux container environments such as systemd-nspawn.
5593 * A framework for implementing offline system updates is now
5594 integrated, for details see:
5595 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5597 * A new service type Type=idle is available now which helps us
5598 avoiding ugly interleaving of getty output and boot status
5601 * There's now a system-wide CapabilityBoundingSet= option to
5602 globally reduce the set of capabilities for the
5603 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5604 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5605 even CAP_NET_ADMIN system-wide for secure systems.
5607 * There are now system-wide DefaultLimitXXX= options to
5608 globally change the defaults of the various resource limits
5609 for all units started by PID 1.
5611 * Harald Hoyer's systemd test suite has been integrated into
5612 systemd which allows easy testing of systemd builds in qemu
5613 and nspawn. (This is really awesome! Ask us for details!)
5615 * The fstab parser is now implemented as generator, not inside
5618 * systemctl will now warn you if .mount units generated from
5619 /etc/fstab are out of date due to changes in fstab that
5620 have not been read by systemd yet.
5622 * systemd is now suitable for usage in initrds. Dracut has
5623 already been updated to make use of this. With this in place
5624 initrds get a slight bit faster but primarily are much
5625 easier to introspect and debug since "systemctl status" in
5626 the host system can be used to introspect initrd services,
5627 and the journal from the initrd is kept around too.
5629 * systemd-delta has been added, a tool to explore differences
5630 between user/admin configuration and vendor defaults.
5632 * PrivateTmp= now affects both /tmp and /var/tmp.
5634 * Boot time status messages are now much prettier and feature
5635 proper english language. Booting up systemd has never been
5638 * Read-ahead pack files now include the inode number of all
5639 files to pre-cache. When the inode changes the pre-caching
5640 is not attempted. This should be nicer to deal with updated
5641 packages which might result in changes of read-ahead
5644 * We now temporaritly lower the kernel's read_ahead_kb variable
5645 when collecting read-ahead data to ensure the kernel's
5646 built-in read-ahead does not add noise to our measurements
5647 of necessary blocks to pre-cache.
5649 * There's now RequiresMountsFor= to add automatic dependencies
5650 for all mounts necessary for a specific file system path.
5652 * MountAuto= and SwapAuto= have been removed from
5653 system.conf. Mounting file systems at boot has to take place
5656 * nspawn now learned a new switch --uuid= to set the machine
5657 ID on the command line.
5659 * nspawn now learned the -b switch to automatically search
5662 * vt102 is now the default TERM for serial TTYs, upgraded from
5665 * systemd-logind now works on VT-less systems.
5667 * The build tree has been reorganized. The individual
5668 components now have directories of their own.
5670 * A new condition type ConditionPathIsReadWrite= is now available.
5672 * nspawn learned the new -C switch to create cgroups for the
5673 container in other hierarchies.
5675 * We now have support for hardware watchdogs, configurable in
5678 * The scheduled shutdown logic now has a public API.
5680 * We now mount /tmp as tmpfs by default, but this can be
5681 masked and /etc/fstab can override it.
5683 * Since udisks does not make use of /media anymore we are not
5684 mounting a tmpfs on it anymore.
5686 * journalctl gained a new --local switch to only interleave
5687 locally generated journal files.
5689 * We can now load the IMA policy at boot automatically.
5691 * The GTK tools have been split off into a systemd-ui.
5693 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5694 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5695 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5696 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5697 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5698 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5699 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5700 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5701 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5706 * This is mostly a bugfix release
5708 * Support optional initialization of the machine ID from the
5709 KVM or container configured UUID.
5711 * Support immediate reboots with "systemctl reboot -ff"
5713 * Show /etc/os-release data in systemd-analyze output
5715 * Many bugfixes for the journal, including endianness fixes and
5716 ensuring that disk space enforcement works
5718 * sd-login.h is C++ comptaible again
5720 * Extend the /etc/os-release format on request of the Debian
5723 * We now refuse non-UTF8 strings used in various configuration
5724 and unit files. This is done to ensure we do not pass invalid
5725 data over D-Bus or expose it elsewhere.
5727 * Register Mimo USB Screens as suitable for automatic seat
5730 * Read SELinux client context from journal clients in a race
5733 * Reorder configuration file lookup order. /etc now always
5734 overrides /run in order to allow the administrator to always
5735 and unconditionally override vendor-supplied or
5736 automatically generated data.
5738 * The various user visible bits of the journal now have man
5739 pages. We still lack man pages for the journal API calls
5742 * We now ship all man pages in HTML format again in the
5745 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5746 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5747 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5752 * This is mostly a bugfix release
5754 * systems lacking /etc/os-release are no longer supported.
5756 * Various functionality updates to libsystemd-login.so
5758 * Track class of PAM logins to distinguish greeters from
5761 Contributions from: Kay Sievers, Lennart Poettering, Michael
5766 * This is an important bugfix release for v41.
5768 * Building man pages is now optional which should be useful
5769 for those building systemd from git but unwilling to install
5772 * Watchdog support for supervising services is now usable. In
5773 a future release support for hardware watchdogs
5774 (i.e. /dev/watchdog) will be added building on this.
5776 * Service start rate limiting is now configurable and can be
5777 turned off per service. When a start rate limit is hit a
5778 reboot can automatically be triggered.
5780 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5782 Contributions from: Benjamin Franzke, Bill Nottingham,
5783 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5784 Schmidt, Michał Górny, Piotr Drąg
5788 * The systemd binary is installed /usr/lib/systemd/systemd now;
5789 An existing /sbin/init symlink needs to be adapted with the
5792 * The code that loads kernel modules has been ported to invoke
5793 libkmod directly, instead of modprobe. This means we do not
5794 support systems with module-init-tools anymore.
5796 * Watchdog support is now already useful, but still not
5799 * A new kernel command line option systemd.setenv= is
5800 understood to set system wide environment variables
5801 dynamically at boot.
5803 * We now limit the set of capabilities of systemd-journald.
5805 * We now set SIGPIPE to ignore by default, since it only is
5806 useful in shell pipelines, and has little use in general
5807 code. This can be disabled with IgnoreSIPIPE=no in unit
5810 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5811 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5816 * This is mostly a bugfix release
5818 * We now expose the reason why a service failed in the
5819 "Result" D-Bus property.
5821 * Rudimentary service watchdog support (will be completed over
5822 the next few releases.)
5824 * When systemd forks off in order execute some service we will
5825 now immediately changes its argv[0] to reflect which process
5826 it will execute. This is useful to minimize the time window
5827 with a generic argv[0], which makes bootcharts more useful
5829 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5830 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5831 Mike Kazantsev, Ray Strode
5835 * This is mostly a test release, but incorporates many
5838 * New systemd-cgtop tool to show control groups by their
5841 * Linking against libacl for ACLs is optional again. If
5842 disabled, support tracking device access for active logins
5843 goes becomes unavailable, and so does access to the user
5844 journals by the respective users.
5846 * If a group "adm" exists, journal files are automatically
5847 owned by them, thus allow members of this group full access
5848 to the system journal as well as all user journals.
5850 * The journal now stores the SELinux context of the logging
5851 client for all entries.
5853 * Add C++ inclusion guards to all public headers
5855 * New output mode "cat" in the journal to print only text
5856 messages, without any meta data like date or time.
5858 * Include tiny X server wrapper as a temporary stop-gap to
5859 teach XOrg udev display enumeration. This is used by display
5860 managers such as gdm, and will go away as soon as XOrg
5861 learned native udev hotplugging for display devices.
5863 * Add new systemd-cat tool for executing arbitrary programs
5864 with STDERR/STDOUT connected to the journal. Can also act as
5865 BSD logger replacement, and does so by default.
5867 * Optionally store all locally generated coredumps in the
5868 journal along with meta data.
5870 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5871 writing short strings to files (for usage for /sys), and for
5872 creating symlinks, character and block device nodes.
5874 * New unit file option ControlGroupPersistent= to make cgroups
5875 persistent, following the mechanisms outlined in
5876 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5878 * Support multiple local RTCs in a sane way
5880 * No longer monopolize IO when replaying readahead data on
5881 rotating disks, since we might starve non-file-system IO to
5882 death, since fanotify() will not see accesses done by blkid,
5885 * Do not show kernel threads in systemd-cgls anymore, unless
5886 requested with new -k switch.
5888 Contributions from: Dan Horák, Kay Sievers, Lennart
5889 Poettering, Michal Schmidt
5893 * This is mostly a test release, but incorporates many
5896 * The git repository moved to:
5897 git://anongit.freedesktop.org/systemd/systemd
5898 ssh://git.freedesktop.org/git/systemd/systemd
5900 * First release with the journal
5901 http://0pointer.de/blog/projects/the-journal.html
5903 * The journal replaces both systemd-kmsg-syslogd and
5904 systemd-stdout-bridge.
5906 * New sd_pid_get_unit() API call in libsystemd-logind
5908 * Many systemadm clean-ups
5910 * Introduce remote-fs-pre.target which is ordered before all
5911 remote mounts and may be used to start services before all
5914 * Added Mageia support
5916 * Add bash completion for systemd-loginctl
5918 * Actively monitor PID file creation for daemons which exit in
5919 the parent process before having finished writing the PID
5920 file in the daemon process. Daemons which do this need to be
5921 fixed (i.e. PID file creation must have finished before the
5922 parent exits), but we now react a bit more gracefully to them.
5924 * Add colourful boot output, mimicking the well-known output
5925 of existing distributions.
5927 * New option PassCredentials= for socket units, for
5928 compatibility with a recent kernel ABI breakage.
5930 * /etc/rc.local is now hooked in via a generator binary, and
5931 thus will no longer act as synchronization point during
5934 * systemctl list-unit-files now supports --root=.
5936 * systemd-tmpfiles now understands two new commands: z, Z for
5937 relabelling files according to the SELinux database. This is
5938 useful to apply SELinux labels to specific files in /sys,
5941 * Output of SysV services is now forwarded to both the console
5942 and the journal by default, not only just the console.
5944 * New man pages for all APIs from libsystemd-login.
5946 * The build tree got reorganized and a the build system is a
5947 lot more modular allowing embedded setups to specifically
5948 select the components of systemd they are interested in.
5950 * Support for Linux systems lacking the kernel VT subsystem is
5953 * configure's --with-rootdir= got renamed to
5954 --with-rootprefix= to follow the naming used by udev and
5957 * Unless specified otherwise we will now install to /usr instead
5958 of /usr/local by default.
5960 * Processes with '@' in argv[0][0] are now excluded from the
5961 final shut-down killing spree, following the logic explained
5963 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5965 * All processes remaining in a service cgroup when we enter
5966 the START or START_PRE states are now killed with
5967 SIGKILL. That means it is no longer possible to spawn
5968 background processes from ExecStart= lines (which was never
5969 supported anyway, and bad style).
5971 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5972 reloading of units together.
5974 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5975 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5976 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5977 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5978 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob