1 systemd System and Service Manager
3 CHANGES WITH 230 in spe:
5 * Testing tool /usr/lib/systemd/systemd-activate is renamed to
6 systemd-socket-activate and installed into /usr/bin. It is now fully
11 * The systemd-resolved DNS resolver service has gained a substantial
12 set of new features, most prominently it may now act as a DNSSEC
13 validating stub resolver. DNSSEC mode is currently turned off by
14 default, but is expected to be turned on by default in one of the
15 next releases. For now, we invite everybody to test the DNSSEC logic
16 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
17 service also gained a full set of D-Bus interfaces, including calls
18 to configure DNS and DNSSEC settings per link (for use by external
19 network management software). systemd-resolved and systemd-networkd
20 now distinguish between "search" and "routing" domains. The former
21 are used to qualify single-label names, the latter are used purely
22 for routing lookups within certain domains to specific links.
23 resolved now also synthesizes RRs for all entries from /etc/hosts.
25 * The systemd-resolve tool (which is a client utility for
26 systemd-resolved) has been improved considerably and is now fully
27 supported and documented. Hence it has moved from /usr/lib/systemd to
30 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
33 * The coredump collection logic has been reworked: when a coredump is
34 collected it is now written to disk, compressed and processed
35 (including stacktrace extraction) from a new instantiated service
36 systemd-coredump@.service, instead of directly from the
37 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
38 processing large coredumps can take up a substantial amount of
39 resources and time, and this previously happened entirely outside of
40 systemd's service supervision. With the new logic the core_pattern
41 hook only does minimal metadata collection before passing off control
42 to the new instantiated service, which is configured with a time
43 limit, a nice level and other settings to minimize negative impact on
44 the rest of the system. Also note that the new logic will honour the
45 RLIMIT_CORE setting of the crashed process, which now allows users
46 and processes to turn off coredumping for their processes by setting
49 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
50 and all forked processes by default. Previously, PID 1 would leave
51 the setting at "0" for all processes, as set by the kernel. Note that
52 the resource limit traditionally has no effect on the generated
53 coredumps on the system if the /proc/sys/kernel/core_pattern hook
54 logic is used. Since the limit is now honoured (see above) its
55 default has been changed so that the coredumping logic is enabled by
56 default for all processes, while allowing specific opt-out.
58 * When the stacktrace is extracted from processes of system users, this
59 is now done as "systemd-coredump" user, in order to sandbox this
60 potentially security sensitive parsing operation. (Note that when
61 processing coredumps of normal users this is done under the user ID
62 of process that crashed, as before.) Packagers should take notice
63 that it is now necessary to create the "systemd-coredump" system user
64 and group at package installation time.
66 * The systemd-activate socket activation testing tool gained support
67 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
68 and --seqpacket switches. It also has been extended to support both
69 new-style and inetd-style file descriptor passing. Use the new
70 --inetd switch to request inetd-style file descriptor passing.
72 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
73 variable, which takes a boolean value. If set to false, ANSI color
74 output is disabled in the tools even when run on a terminal that
77 * The VXLAN support in networkd now supports two new settings
78 DestinationPort= and PortRange=.
80 * A new systemd.machine_id= kernel command line switch has been added,
81 that may be used to set the machine ID in /etc/machine-id if it is
82 not initialized yet. This command line option has no effect if the
83 file is already initialized.
85 * systemd-nspawn gained a new --as-pid2 switch that invokes any
86 specified command line as PID 2 rather than PID 1 in the
87 container. In this mode PID 1 is a minimal stub init process that
88 implements the special POSIX and Linux semantics of PID 1 regarding
89 signal and child process management. Note that this stub init process
90 is implemented in nspawn itself and requires no support from the
91 container image. This new logic is useful to support running
92 arbitrary commands in the container, as normal processes are
93 generally not prepared to run as PID 1.
95 * systemd-nspawn gained a new --chdir= switch for setting the current
96 working directory for the process started in the container.
98 * "journalctl /dev/sda" will now output all kernel log messages for
99 specified device from the current boot, in addition to all devices
100 that are parents of it. This should make log output about devices
101 pretty useful, as long as kernel drivers attach enough metadata to
102 the log messages. (The usual SATA drivers do.)
104 * The sd-journal API gained two new calls
105 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
106 that report whether log data from /run or /var has been found.
108 * journalctl gained a new switch "--fields" that prints all journal
109 record field names currently in use in the journal. This is backed
110 by two new sd-journal API calls sd_journal_enumerate_fields() and
111 sd_journal_restart_fields().
113 * Most configurable timeouts in systemd now expect an argument of
114 "infinity" to turn them off, instead of "0" as before. The semantics
115 from now on is that a timeout of "0" means "now", and "infinity"
116 means "never". To maintain backwards compatibility, "0" continues to
117 turn off previously existing timeout settings.
119 * "systemctl reload-or-try-restart" has been renamed to "systemctl
120 try-reload-or-restart" to clarify what it actually does: the "try"
121 logic applies to both reloading and restarting, not just restarting.
122 The old name continues to be accepted for compatibility.
124 * On boot-up, when PID 1 detects that the system clock is behind the
125 release date of the systemd version in use, the clock is now set
126 to the latter. Previously, this was already done in timesyncd, in order
127 to avoid running with clocks set to the various clock epochs such as
128 1902, 1938 or 1970. With this change the logic is now done in PID 1
129 in addition to timesyncd during early boot-up, so that it is enforced
130 before the first process is spawned by systemd. Note that the logic
131 in timesyncd remains, as it is more comprehensive and ensures
132 clock monotonicity by maintaining a persistent timestamp file in
133 /var. Since /var is generally not available in earliest boot or the
134 initrd, this part of the logic remains in timesyncd, and is not done
137 * Support for tweaking details in net_cls.class_id through the
138 NetClass= configuration directive has been removed, as the kernel
139 people have decided to deprecate that controller in cgroup v2.
140 Userspace tools such as nftables are moving over to setting rules
141 that are specific to the full cgroup path of a task, which obsoletes
142 these controllers anyway. The NetClass= directive is kept around for
143 legacy compatibility reasons. For a more in-depth description of the
144 kernel change, please refer to the respective upstream commit:
146 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
148 * A new service setting RuntimeMaxSec= has been added that may be used
149 to specify a maximum runtime for a service. If the timeout is hit, the
150 service is terminated and put into a failure state.
152 * A new service setting AmbientCapabilities= has been added. It allows
153 configuration of additional Linux process capabilities that are
154 passed to the activated processes. This is only available on very
157 * The process resource limit settings in service units may now be used
158 to configure hard and soft limits individually.
160 * The various libsystemd APIs such as sd-bus or sd-event now publicly
161 expose support for gcc's __attribute__((cleanup())) C extension.
162 Specifically, for many object destructor functions alternative
163 versions have been added that have names suffixed with "p" and take a
164 pointer to a pointer to the object to destroy, instead of just a
165 pointer to the object itself. This is useful because these destructor
166 functions may be used directly as parameters to the cleanup
167 construct. Internally, systemd has been a heavy user of this GCC
168 extension for a long time, and with this change similar support is
169 now available to consumers of the library outside of systemd. Note
170 that by using this extension in your sources compatibility with old
171 and strictly ANSI compatible C compilers is lost. However, all gcc or
172 LLVM versions of recent years support this extension.
174 * Timer units gained support for a new setting RandomizedDelaySec= that
175 allows configuring some additional randomized delay to the configured
176 time. This is useful to spread out timer events to avoid load peaks in
177 clusters or larger setups.
179 * Calendar time specifications now support sub-second accuracy.
181 * Socket units now support listening on SCTP and UDP-lite protocol
184 * The sd-event API now comes with a full set of man pages.
186 * Older versions of systemd contained experimental support for
187 compressing journal files and coredumps with the LZ4 compressor that
188 was not compatible with the lz4 binary (due to API limitations of the
189 lz4 library). This support has been removed; only support for files
190 compatible with the lz4 binary remains. This LZ4 logic is now
191 officially supported and no longer considered experimental.
193 * The dkr image import logic has been removed again from importd. dkr's
194 micro-services focus doesn't fit into the machine image focus of
195 importd, and quickly got out of date with the upstream dkr API.
197 * Creation of the /run/lock/lockdev/ directory was dropped from
198 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
199 been available for many years. If you still need this, you need to
200 create your own tmpfiles.d config file with:
202 d /run/lock/lockdev 0775 root lock -
204 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
205 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
206 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
207 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
208 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
209 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
210 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
211 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
212 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
213 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
214 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
215 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
216 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
217 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
218 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
219 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
220 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
221 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
222 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
224 -- Berlin, 2016-02-11
228 * A number of properties previously only settable in unit
229 files are now also available as properties to set when
230 creating transient units programmatically via the bus, as it
231 is exposed with systemd-run's --property=
232 setting. Specifically, these are: SyslogIdentifier=,
233 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
234 EnvironmentFile=, ReadWriteDirectories=,
235 ReadOnlyDirectories=, InaccessibleDirectories=,
236 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
238 * When creating transient services via the bus API it is now
239 possible to pass in a set of file descriptors to use as
240 STDIN/STDOUT/STDERR for the invoked process.
242 * Slice units may now be created transiently via the bus APIs,
243 similar to the way service and scope units may already be
246 * Wherever systemd expects a calendar timestamp specification
247 (like in journalctl's --since= and --until= switches) UTC
248 timestamps are now supported. Timestamps suffixed with "UTC"
249 are now considered to be in Universal Time Coordinated
250 instead of the local timezone. Also, timestamps may now
251 optionally be specified with sub-second accuracy. Both of
252 these additions also apply to recurring calendar event
253 specification, such as OnCalendar= in timer units.
255 * journalctl gained a new "--sync" switch that asks the
256 journal daemon to write all so far unwritten log messages to
257 disk and sync the files, before returning.
259 * systemd-tmpfiles learned two new line types "q" and "Q" that
260 operate like "v", but also set up a basic btrfs quota
261 hierarchy when used on a btrfs file system with quota
264 * tmpfiles' "v", "q" and "Q" will now create a plain directory
265 instead of a subvolume (even on a btrfs file system) if the
266 root directory is a plain directory, and not a
267 subvolume. This should simplify things with certain chroot()
268 environments which are not aware of the concept of btrfs
271 * systemd-detect-virt gained a new --chroot switch to detect
272 whether execution takes place in a chroot() environment.
274 * CPUAffinity= now takes CPU index ranges in addition to
277 * The various memory-related resource limit settings (such as
278 LimitAS=) now understand the usual K, M, G, ... suffixes to
279 the base of 1024 (IEC). Similar, the time-related resource
280 limit settings understand the usual min, h, day, ...
283 * There's a new system.conf setting DefaultTasksMax= to
284 control the default TasksMax= setting for services and
285 scopes running on the system. (TasksMax= is the primary
286 setting that exposes the "pids" cgroup controller on systemd
287 and was introduced in the previous systemd release.) The
288 setting now defaults to 512, which means services that are
289 not explicitly configured otherwise will only be able to
290 create 512 processes or threads at maximum, from this
291 version on. Note that this means that thread- or
292 process-heavy services might need to be reconfigured to set
293 TasksMax= to a higher value. It is sufficient to set
294 TasksMax= in these specific unit files to a higher value, or
295 even "infinity". Similar, there's now a logind.conf setting
296 UserTasksMax= that defaults to 4096 and limits the total
297 number of processes or tasks each user may own
298 concurrently. nspawn containers also have the TasksMax=
299 value set by default now, to 8192. Note that all of this
300 only has an effect if the "pids" cgroup controller is
301 enabled in the kernel. The general benefit of these changes
302 should be a more robust and safer system, that provides a
303 certain amount of per-service fork() bomb protection.
305 * systemd-nspawn gained the new --network-veth-extra= switch
306 to define additional and arbitrarily-named virtual Ethernet
307 links between the host and the container.
309 * A new service execution setting PassEnvironment= has been
310 added that allows importing select environment variables
311 from PID1's environment block into the environment block of
314 * Timer units gained support for a new RemainAfterElapse=
315 setting which takes a boolean argument. It defaults on on,
316 exposing behaviour unchanged to previous releases. If set to
317 off, timer units are unloaded after they elapsed if they
318 cannot elapse again. This is particularly useful for
319 transient timer units, which shall not stay around longer
320 than until they first elapse.
322 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
323 default now (the kernel default is 16). This is beneficial
324 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
325 allows substantially larger numbers of queued
326 datagrams. This should increase the capability of systemd to
327 parallelize boot-up, as logging and sd_notify() are unlikely
328 to stall execution anymore. If you need to change the value
329 from the new defaults, use the usual sysctl.d/ snippets.
331 * The compression framing format used by the journal or
332 coredump processing has changed to be in line with what the
333 official LZ4 tools generate. LZ4 compression support in
334 systemd was considered unsupported previously, as the format
335 was not compatible with the normal tools. With this release
336 this has changed now, and it is hence safe for downstream
337 distributions to turn it on. While not compressing as well
338 as the XZ, LZ4 is substantially faster, which makes
339 it a good default choice for the compression logic in the
340 journal and in coredump handling.
342 * Any reference to /etc/mtab has been dropped from
343 systemd. The file has been obsolete since a while, but
344 systemd refused to work on systems where it was incorrectly
345 set up (it should be a symlink or non-existent). Please make
346 sure to update to util-linux 2.27.1 or newer in conjunction
347 with this systemd release, which also drops any reference to
348 /etc/mtab. If you maintain a distribution make sure that no
349 software you package still references it, as this is a
350 likely source of bugs. There's also a glibc bug pending,
351 asking for removal of any reference to this obsolete file:
353 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
355 * Support for the ".snapshot" unit type has been removed. This
356 feature turned out to be little useful and little used, and
357 has now been removed from the core and from systemctl.
359 * The dependency types RequiresOverridable= and
360 RequisiteOverridable= have been removed from systemd. They
361 have been used only very sparingly to our knowledge and
362 other options that provide a similar effect (such as
363 systemctl --mode=ignore-dependencies) are much more useful
364 and commonly used. Moreover, they were only half-way
365 implemented as the option to control behaviour regarding
366 these dependencies was never added to systemctl. By removing
367 these dependency types the execution engine becomes a bit
368 simpler. Unit files that use these dependencies should be
369 changed to use the non-Overridable dependency types
370 instead. In fact, when parsing unit files with these
371 options, that's what systemd will automatically convert them
372 too, but it will also warn, asking users to fix the unit
373 files accordingly. Removal of these dependency types should
374 only affect a negligible number of unit files in the wild.
376 * Behaviour of networkd's IPForward= option changed
377 (again). It will no longer maintain a per-interface setting,
378 but propagate one way from interfaces where this is enabled
379 to the global kernel setting. The global setting will be
380 enabled when requested by a network that is set up, but
381 never be disabled again. This change was made to make sure
382 IPv4 and IPv6 behaviour regarding packet forwarding is
383 similar (as the Linux IPv6 stack does not support
384 per-interface control of this setting) and to minimize
387 * In unit files the behaviour of %u, %U, %h, %s has
388 changed. These specifiers will now unconditionally resolve
389 to the various user database fields of the user that the
390 systemd instance is running as, instead of the user
391 configured in the specific unit via User=. Note that this
392 effectively doesn't change much, as resolving of these
393 specifiers was already turned off in the --system instance
394 of systemd, as we cannot do NSS lookups from PID 1. In the
395 --user instance of systemd these specifiers where correctly
396 resolved, but hardly made any sense, since the user instance
397 lacks privileges to do user switches anyway, and User= is
398 hence useless. Morever, even in the --user instance of
399 systemd behaviour was awkward as it would only take settings
400 from User= assignment placed before the specifier into
401 account. In order to unify and simplify the logic around
402 this the specifiers will now always resolve to the
403 credentials of the user invoking the manager (which in case
404 of PID 1 is the root user).
406 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
407 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
408 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
409 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
410 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
411 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
412 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
413 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
414 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
415 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
416 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
417 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
418 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
419 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
422 -- Berlin, 2015-11-18
426 * systemd now depends on util-linux v2.27. More specifically,
427 the newly added mount monitor feature in libmount now
428 replaces systemd's former own implementation.
430 * libmount mandates /etc/mtab not to be regular file, and
431 systemd now enforces this condition at early boot.
432 /etc/mtab has been deprecated and warned about for a very
433 long time, so systems running systemd should already have
434 stopped having this file around as anything else than a
435 symlink to /proc/self/mounts.
437 * Support for the "pids" cgroup controller has been added. It
438 allows accounting the number of tasks in a cgroup and
439 enforcing limits on it. This adds two new setting
440 TasksAccounting= and TasksMax= to each unit, as well as a
441 global option DefaultTasksAccounting=.
443 * Support for the "net_cls" cgroup controller has been added.
444 It allows assigning a net class ID to each task in the
445 cgroup, which can then be used in firewall rules and traffic
446 shaping configurations. Note that the kernel netfilter net
447 class code does not currently work reliably for ingress
448 packets on unestablished sockets.
450 This adds a new config directive called NetClass= to CGroup
451 enabled units. Allowed values are positive numbers for fixed
452 assignments and "auto" for picking a free value
455 * 'systemctl is-system-running' now returns 'offline' if the
456 system is not booted with systemd. This command can now be
457 used as a substitute for 'systemd-notify --booted'.
459 * Watchdog timeouts have been increased to 3 minutes for all
460 in-tree service files. Apparently, disk IO issues are more
461 frequent than we hoped, and user reported >1 minute waiting
464 * 'machine-id-commit' functionality has been merged into
465 'machine-id-setup --commit'. The separate binary has been
468 * The WorkingDirectory= directive in unit files may now be set
469 to the special value '~'. In this case, the working
470 directory is set to the home directory of the user
473 * "machinectl shell" will now open the shell in the home
474 directory of the selected user by default.
476 * The CrashChVT= configuration file setting is renamed to
477 CrashChangeVT=, following our usual logic of not
478 abbreviating unnecessarily. The old directive is still
479 supported for compat reasons. Also, this directive now takes
480 an integer value between 1 and 63, or a boolean value. The
481 formerly supported '-1' value for disabling stays around for
484 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
485 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
486 RootDirectory= properties can now be set for transient
489 * The systemd-analyze tool gained a new "set-log-target" verb
490 to change the logging target the system manager logs to
491 dynamically during runtime. This is similar to how
492 "systemd-analyze set-log-level" already changes the log
495 * In nspawn /sys is now mounted as tmpfs, with only a selected
496 set of subdirectories mounted in from the real sysfs. This
497 enhances security slightly, and is useful for ensuring user
498 namespaces work correctly.
500 * Support for USB FunctionFS activation has been added. This
501 allows implementation of USB gadget services that are
502 activated as soon as they are requested, so that they don't
503 have to run continously, similar to classic socket
506 * The "systemctl exit" command now optionally takes an
507 additional parameter that sets the exit code to return from
508 the systemd manager when exiting. This is only relevant when
509 running the systemd user instance, or when running the
510 system instance in a container.
512 * sd-bus gained the new API calls sd_bus_path_encode_many()
513 and sd_bus_path_decode_many() that allow easy encoding and
514 decoding of multiple identifier strings inside a D-Bus
515 object path. Another new call sd_bus_default_flush_close()
516 has been added to flush and close per-thread default
519 * systemd-cgtop gained support for a -M/--machine= switch to
520 show the control groups within a certain container only.
522 * "systemctl kill" gained support for an optional --fail
523 switch. If specified the requested operation will fail of no
524 processes have been killed, because the unit had no
525 processes attached, or similar.
527 * A new systemd.crash_reboot=1 kernel command line option has
528 been added that triggers a reboot after crashing. This can
529 also be set through CrashReboot= in systemd.conf.
531 * The RuntimeDirectory= setting now understands unit
532 specifiers like %i or %f.
534 * A new (still internal) libary API sd-ipv4acd has been added,
535 that implements address conflict detection for IPv4. It's
536 based on code from sd-ipv4ll, and will be useful for
537 detecting DHCP address conflicts.
539 * File descriptors passed during socket activation may now be
540 named. A new API sd_listen_fds_with_names() is added to
541 access the names. The default names may be overridden,
542 either in the .socket file using the FileDescriptorName=
543 parameter, or by passing FDNAME= when storing the file
544 descriptors using sd_notify().
546 * systemd-networkd gained support for:
548 - Setting the IPv6 Router Advertisment settings via
549 IPv6AcceptRouterAdvertisements= in .network files.
551 - Configuring the HelloTimeSec=, MaxAgeSec= and
552 ForwardDelaySec= bridge parameters in .netdev files.
554 - Configuring PreferredSource= for static routes in
557 * The "ask-password" framework used to query for LUKS harddisk
558 passwords or SSL passwords during boot gained support for
559 caching passwords in the kernel keyring, if it is
560 available. This makes sure that the user only has to type in
561 a passphrase once if there are multiple objects to unlock
562 with the same one. Previously, such password caching was
563 available only when Plymouth was used; this moves the
564 caching logic into the systemd codebase itself. The
565 "systemd-ask-password" utility gained a new --keyname=
566 switch to control which kernel keyring key to use for
567 caching a password in. This functionality is also useful for
568 enabling display managers such as gdm to automatically
569 unlock the user's GNOME keyring if its passphrase, the
570 user's password and the harddisk password are the same, if
571 gdm-autologin is used.
573 * When downloading tar or raw images using "machinectl
574 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
575 file is now also downloaded, if it is available and stored
576 next to the image file.
578 * Units of type ".socket" gained a new boolean setting
579 Writable= which is only useful in conjunction with
580 ListenSpecial=. If true, enables opening the specified
581 special file in O_RDWR mode rather than O_RDONLY mode.
583 * systemd-rfkill has been reworked to become a singleton
584 service that is activated through /dev/rfkill on each rfkill
585 state change and saves the settings to disk. This way,
586 systemd-rfkill is now compatible with devices that exist
587 only intermittendly, and even restores state if the previous
588 system shutdown was abrupt rather than clean.
590 * The journal daemon gained support for vacuuming old journal
591 files controlled by the number of files that shall remain,
592 in addition to the already existing control by size and by
593 date. This is useful as journal interleaving performance
594 degrades with too many seperate journal files, and allows
595 putting an effective limit on them. The new setting defaults
596 to 100, but this may be changed by setting SystemMaxFiles=
597 and RuntimeMaxFiles= in journald.conf. Also, the
598 "journalctl" tool gained the new --vacuum-files= switch to
599 manually vacuum journal files to leave only the specified
600 number of files in place.
602 * udev will now create /dev/disk/by-path links for ATA devices
603 on kernels where that is supported.
605 * Galician, Serbian, Turkish and Korean translations were added.
607 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
608 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
609 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
610 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
611 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
612 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
613 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
614 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
615 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
616 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
617 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
618 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
619 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
620 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
621 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
622 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
623 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
624 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
626 -- Berlin, 2015-10-07
630 * The DHCP implementation of systemd-networkd gained a set of
633 - The DHCP server now supports emitting DNS and NTP
634 information. It may be enabled and configured via
635 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
636 and NTP information is enabled, but no servers are
637 configured, the corresponding uplink information (if there
638 is any) is propagated.
640 - Server and client now support transmission and reception
641 of timezone information. It can be configured via the
642 newly introduced network options UseTimezone=,
643 EmitTimezone=, and Timezone=. Transmission of timezone
644 information is enabled between host and containers by
645 default now: the container will change its local timezone
646 to what the host has set.
648 - Lease timeouts can now be configured via
649 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
651 - The DHCP server improved on the stability of
652 leases. Clients are more likely to get the same lease
653 information back, even if the server loses state.
655 - The DHCP server supports two new configuration options to
656 control the lease address pool metrics, PoolOffset= and
659 * The encapsulation limit of tunnels in systemd-networkd may
660 now be configured via 'EncapsulationLimit='. It allows
661 modifying the maximum additional levels of encapsulation
662 that are permitted to be prepended to a packet.
664 * systemd now supports the concept of user buses replacing
665 session buses, if used with dbus-1.10 (and enabled via dbus
666 --enable-user-session). It previously only supported this on
667 kdbus-enabled systems, and this release expands this to
668 'dbus-daemon' systems.
670 * systemd-networkd now supports predictable interface names
673 * systemd now optionally supports the new Linux kernel
674 "unified" control group hierarchy. If enabled via the kernel
675 command-line option 'systemd.unified_cgroup_hierarchy=1',
676 systemd will try to mount the unified cgroup hierarchy
677 directly on /sys/fs/cgroup. If not enabled, or not
678 available, systemd will fall back to the legacy cgroup
679 hierarchy setup, as before. Host system and containers can
680 mix and match legacy and unified hierarchies as they
681 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
682 environment variable to individually select the hierarchy to
683 use for executed containers. By default, nspawn will use the
684 unified hierarchy for the containers if the host uses the
685 unified hierarchy, and the legacy hierarchy otherwise.
686 Please note that at this point the unified hierarchy is an
687 experimental kernel feature and is likely to change in one
688 of the next kernel releases. Therefore, it should not be
689 enabled by default in downstream distributions yet. The
690 minimum required kernel version for the unified hierarchy to
691 work is 4.2. Note that when the unified hierarchy is used
692 for the first time delegated access to controllers is
693 safe. Because of this systemd-nspawn containers will get
694 access to controllers now, as will systemd user
695 sessions. This means containers and user sessions may now
696 manage their own resources, partitioning up what the system
699 * A new special scope unit "init.scope" has been introduced
700 that encapsulates PID 1 of the system. It may be used to
701 determine resource usage and enforce resource limits on PID
702 1 itself. PID 1 hence moved out of the root of the control
705 * The cgtop tool gained support for filtering out kernel
706 threads when counting tasks in a control group. Also, the
707 count of processes is now recursively summed up by
708 default. Two options -k and --recursive= have been added to
709 revert to old behaviour. The tool has also been updated to
710 work correctly in containers now.
712 * systemd-nspawn's --bind= and --bind-ro= options have been
713 extended to allow creation of non-recursive bind mounts.
715 * libsystemd gained two new calls sd_pid_get_cgroup() and
716 sd_peer_get_cgroup() which return the control group path of
717 a process or peer of a connected AF_UNIX socket. This
718 function call is particularly useful when implementing
719 delegated subtrees support in the control group hierarchy.
721 * The "sd-event" event loop API of libsystemd now supports
722 correct dequeuing of real-time signals, without losing
725 * When systemd requests a PolicyKit decision when managing
726 units it will now add additional fields to the request,
727 including unit name and desired operation. This enables more
728 powerful PolicyKit policies, that make decisions depending
731 * nspawn learnt support for .nspawn settings files, that may
732 accompany the image files or directories of containers, and
733 may contain additional settings for the container. This is
734 an alternative to configuring container parameters via the
737 Contributions from: Cristian Rodríguez, Daniel Mack, David
738 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
739 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
740 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
741 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
742 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
743 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
744 Andersen, Tom Gundersen, Torstein Husebø
746 -- Berlin, 2015-09-08
750 * machinectl gained a new verb 'shell' which opens a fresh
751 shell on the target container or the host. It is similar to
752 the existing 'login' command of machinectl, but spawns the
753 shell directly without prompting for username or
754 password. The pseudo machine '.host' now refers to the local
755 host and is used by default. Hence, 'machinectl shell' can
756 be used as replacement for 'su -' which spawns a session as
757 a fresh systemd unit in a way that is fully isolated from
758 the originating session.
760 * systemd-networkd learned to cope with private-zone DHCP
761 options and allows other programs to query the values.
763 * SELinux access control when enabling/disabling units is no
764 longer enforced with this release. The previous
765 implementation was incorrect, and a new corrected
766 implementation is not yet available. As unit file operations
767 are still protected via PolicyKit and D-Bus policy this is
768 not a security problem. Yet, distributions which care about
769 optimal SELinux support should probably not stabilize on
772 * sd-bus gained support for matches of type "arg0has=", that
773 test for membership of strings in string arrays sent in bus
776 * systemd-resolved now dumps the contents of its DNS and LLMNR
777 caches to the logs on reception of the SIGUSR1 signal. This
778 is useful to debug DNS behaviour.
780 * The coredumpctl tool gained a new --directory= option to
781 operate on journal files in a specific directory.
783 * "systemctl reboot" and related commands gained a new
784 "--message=" option which may be used to set a free-text
785 wall message when shutting down or rebooting the
786 system. This message is also logged, which is useful for
787 figuring out the reason for a reboot or shutdown a
790 * The "systemd-resolve-host" tool's -i switch now takes
791 network interface numbers as alternative to interface names.
793 * A new unit file setting for services has been introduced:
794 UtmpMode= allows configuration of how precisely systemd
795 handles utmp and wtmp entries for the service if this is
796 enabled. This allows writing services that appear similar to
797 user sessions in the output of the "w", "who", "last" and
800 * systemd-resolved will now locally synthesize DNS resource
801 records for the "localhost" and "gateway" domains as well as
802 the local hostname. This should ensure that clients querying
803 RRs via resolved will get similar results as those going via
804 NSS, if nss-myhostname is enabled.
806 Contributions from: Alastair Hughes, Alex Crawford, Daniel
807 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
808 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
809 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
810 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
811 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
812 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
813 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
814 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
815 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
816 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
817 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
819 -- Berlin, 2015-08-27
823 * The systemd-efi-boot-generator functionality was merged into
824 systemd-gpt-auto-generator.
826 * systemd-networkd now supports Group Policy for vxlan
827 devices. It can be enabled via the new boolean configuration
828 option called 'GroupPolicyExtension='.
830 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
831 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
832 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
834 -- Berlin, 2015-07-31
838 * The python-systemd code has been removed from the systemd repository.
839 A new repository has been created which accommodates the code from
840 now on, and we kindly ask distributions to create a separate package
841 for this: https://github.com/systemd/python-systemd
843 * The systemd daemon will now reload its main configuration
844 (/etc/systemd/system.conf) on daemon-reload.
846 * sd-dhcp now exposes vendor specific extensions via
847 sd_dhcp_lease_get_vendor_specific().
849 * systemd-networkd gained a number of new configuration options.
851 - A new boolean configuration option for TAP devices called
852 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
853 device, thus allowing to send and receive GSO packets.
855 - A new tunnel configuration option called 'CopyDSCP='.
856 If enabled, the DSCP field of ip6 tunnels is copied into the
859 - A set of boolean bridge configuration options were added.
860 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
861 and 'UnicastFlood=' are now parsed by networkd and applied to the
862 respective bridge link device via the respective IFLA_BRPORT_*
865 - A new string configuration option to override the hostname sent
866 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
867 is true, networkd will use the configured hostname instead of the
868 system hostname when sending DHCP requests.
870 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
871 networkd will configure the IPv6 flow-label of the tunnel device
872 according to RFC2460.
874 - The 'macvtap' virtual network devices are now supported, similar to
875 the already supported 'macvlan' devices.
877 * systemd-resolved now implements RFC5452 to improve resilience against
878 cache poisoning. Additionally, source port randomization is enabled
879 by default to further protect against DNS spoofing attacks.
881 * nss-mymachines now supports translating UIDs and GIDs of running
882 containers with user-namespaces enabled. If a container 'foo'
883 translates a host uid 'UID' to the container uid 'TUID', then
884 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
885 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
886 mapped as 'vg-foo-TGID'.
888 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
889 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
890 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
891 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
892 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
893 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
894 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
895 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
896 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
897 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
899 -- Berlin, 2015-07-29
903 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
904 There are no known issues with current sysfs, and udev does not need
905 or should be used to work around such bugs.
907 * udev does no longer enable USB HID power management. Several reports
908 indicate, that some devices cannot handle that setting.
910 * The udev accelerometer helper was removed. The functionality
911 is now fully included in iio-sensor-proxy. But this means,
912 older iio-sensor-proxy versions will no longer provide
913 accelerometer/orientation data with this systemd version.
914 Please upgrade iio-sensor-proxy to version 1.0.
916 * networkd gained a new configuration option IPv6PrivacyExtensions=
917 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
918 for Stateless Address") on selected networks.
920 * For the sake of fewer build-time dependencies and less code in the
921 main repository, the python bindings are about to be removed in the
922 next release. A new repository has been created which accommodates
923 the code from now on, and we kindly ask distributions to create a
924 separate package for this. The removal will take place in v223.
926 https://github.com/systemd/python-systemd
928 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
929 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
930 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
931 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
932 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
933 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
934 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
935 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
936 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
937 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
939 -- Berlin, 2015-07-07
943 * The sd-bus.h and sd-event.h APIs have now been declared
944 stable and have been added to the official interface of
945 libsystemd.so. sd-bus implements an alternative D-Bus client
946 library, that is relatively easy to use, very efficient and
947 supports both classic D-Bus as well as kdbus as transport
948 backend. sd-event is a generic event loop abstraction that
949 is built around Linux epoll, but adds features such as event
950 prioritization or efficient timer handling. Both APIs are good
951 choices for C programs looking for a bus and/or event loop
952 implementation that is minimal and does not have to be
953 portable to other kernels.
955 * kdbus support is no longer compile-time optional. It is now
956 always built-in. However, it can still be disabled at
957 runtime using the kdbus=0 kernel command line setting, and
958 that setting may be changed to default to off, by specifying
959 --disable-kdbus at build-time. Note though that the kernel
960 command line setting has no effect if the kdbus.ko kernel
961 module is not installed, in which case kdbus is (obviously)
962 also disabled. We encourage all downstream distributions to
963 begin testing kdbus by adding it to the kernel images in the
964 development distributions, and leaving kdbus support in
967 * The minimal required util-linux version has been bumped to
970 * Support for chkconfig (--enable-chkconfig) was removed in
971 favor of calling an abstraction tool
972 /lib/systemd/systemd-sysv-install. This needs to be
973 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
974 in README for details.
976 * If there's a systemd unit and a SysV init script for the
977 same service name, and the user executes "systemctl enable"
978 for it (or a related call), then this will now enable both
979 (or execute the related operation on both), not just the
982 * The libudev API documentation has been converted from gtkdoc
985 * gudev has been removed from the systemd tree, it is now an
988 * The systemd-cgtop tool learnt a new --raw switch to generate
989 "raw" (machine parsable) output.
991 * networkd's IPForwarding= .network file setting learnt the
992 new setting "kernel", which ensures that networkd does not
993 change the IP forwarding sysctl from the default kernel
996 * The systemd-logind bus API now exposes a new boolean
997 property "Docked" that reports whether logind considers the
998 system "docked", i.e. connected to a docking station or not.
1000 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1001 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1002 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1003 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1004 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1005 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1006 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1007 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1008 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1009 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1010 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1011 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1012 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1013 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1014 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1015 Fink, Zbigniew Jędrzejewski-Szmek
1017 -- Berlin, 2015-06-19
1021 * The gudev library has been extracted into a separate repository
1022 available at: https://git.gnome.org/browse/libgudev/
1023 It is now managed as part of the Gnome project. Distributions
1024 are recommended to pass --disable-gudev to systemd and use
1025 gudev from the Gnome project instead. gudev is still included
1026 in systemd, for now. It will be removed soon, though. Please
1027 also see the announcement-thread on systemd-devel:
1028 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1030 * systemd now exposes a CPUUsageNSec= property for each
1031 service unit on the bus, that contains the overall consumed
1032 CPU time of a service (the sum of what each process of the
1033 service consumed). This value is only available if
1034 CPUAccounting= is turned on for a service, and is then shown
1035 in the "systemctl status" output.
1037 * Support for configuring alternative mappings of the old SysV
1038 runlevels to systemd targets has been removed. They are now
1039 hardcoded in a way that runlevels 2, 3, 4 all map to
1040 multi-user.target and 5 to graphical.target (which
1041 previously was already the default behaviour).
1043 * The auto-mounter logic gained support for mount point
1044 expiry, using a new TimeoutIdleSec= setting in .automount
1045 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1047 * The EFI System Partition (ESP) as mounted to /boot by
1048 systemd-efi-boot-generator will now be unmounted
1049 automatically after 2 minutes of not being used. This should
1050 minimize the risk of ESP corruptions.
1052 * New /etc/fstab options x-systemd.requires= and
1053 x-systemd.requires-mounts-for= are now supported to express
1054 additional dependencies for mounts. This is useful for
1055 journalling file systems that support external journal
1056 devices or overlay file systems that require underlying file
1057 systems to be mounted.
1059 * systemd does not support direct live-upgrades (via systemctl
1060 daemon-reexec) from versions older than v44 anymore. As no
1061 distribution we are aware of shipped such old versions in a
1062 stable release this should not be problematic.
1064 * When systemd forks off a new per-connection service instance
1065 it will now set the $REMOTE_ADDR environment variable to the
1066 remote IP address, and $REMOTE_PORT environment variable to
1067 the remote IP port. This behaviour is similar to the
1068 corresponding environment variables defined by CGI.
1070 * systemd-networkd gained support for uplink failure
1071 detection. The BindCarrier= option allows binding interface
1072 configuration dynamically to the link sense of other
1073 interfaces. This is useful to achieve behaviour like in
1076 * systemd-networkd gained support for configuring the DHCP
1077 client identifier to use when requesting leases.
1079 * systemd-networkd now has a per-network UseNTP= option to
1080 configure whether NTP server information acquired via DHCP
1081 is passed on to services like systemd-timesyncd.
1083 * systemd-networkd gained support for vti6 tunnels.
1085 * Note that systemd-networkd manages the sysctl variable
1086 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1087 it is configured for since v219. The variable controls IP
1088 forwarding, and is a per-interface alternative to the global
1089 /proc/sys/net/ipv[46]/ip_forward. This setting is
1090 configurable in the IPForward= option, which defaults to
1091 "no". This means if networkd is used for an interface it is
1092 no longer sufficient to set the global sysctl option to turn
1093 on IP forwarding! Instead, the .network file option
1094 IPForward= needs to be turned on! Note that the
1095 implementation of this behaviour was broken in v219 and has
1098 * Many bonding and vxlan options are now configurable in
1101 * systemd-nspawn gained a new --property= setting to set unit
1102 properties for the container scope. This is useful for
1103 setting resource parameters (e.g "CPUShares=500") on
1104 containers started from the command line.
1106 * systemd-nspawn gained a new --private-users= switch to make
1107 use of user namespacing available on recent Linux kernels.
1109 * systemd-nspawn may now be called as part of a shell pipeline
1110 in which case the pipes used for stdin and stdout are passed
1111 directly to the process invoked in the container, without
1112 indirection via a pseudo tty.
1114 * systemd-nspawn gained a new switch to control the UNIX
1115 signal to use when killing the init process of the container
1118 * systemd-nspawn gained a new --overlay= switch for mounting
1119 overlay file systems into the container using the new kernel
1122 * When a container image is imported via systemd-importd and
1123 the host file system is not btrfs, a loopback block device
1124 file is created in /var/lib/machines.raw with a btrfs file
1125 system inside. It is then mounted to /var/lib/machines to
1126 enable btrfs features for container management. The loopback
1127 file and btrfs file system is grown as needed when container
1128 images are imported via systemd-importd.
1130 * systemd-machined/systemd-importd gained support for btrfs
1131 quota, to enforce container disk space limits on disk. This
1132 is exposed in "machinectl set-limit".
1134 * systemd-importd now can import containers from local .tar,
1135 .raw and .qcow2 images, and export them to .tar and .raw. It
1136 can also import dkr v2 images now from the network (on top
1139 * systemd-importd gained support for verifying downloaded
1140 images with gpg2 (previously only gpg1 was supported).
1142 * systemd-machined, systemd-logind, systemd: most bus calls
1143 are now accessible to unprivileged processes via
1144 PolicyKit. Also, systemd-logind will now allow users to kill
1145 their own sessions without further privileges or
1148 * systemd-shutdownd has been removed. This service was
1149 previously responsible for implementing scheduled shutdowns
1150 as exposed in /usr/bin/shutdown's time parameter. This
1151 functionality has now been moved into systemd-logind and is
1152 accessible via a bus interface.
1154 * "systemctl reboot" gained a new switch --firmware-setup that
1155 can be used to reboot into the EFI firmware setup, if that
1156 is available. systemd-logind now exposes an API on the bus
1157 to trigger such reboots, in case graphical desktop UIs want
1158 to cover this functionality.
1160 * "systemctl enable", "systemctl disable" and "systemctl mask"
1161 now support a new "--now" switch. If specified the units
1162 that are enabled will also be started, and the ones
1163 disabled/masked also stopped.
1165 * The Gummiboot EFI boot loader tool has been merged into
1166 systemd, and renamed to "systemd-boot". The bootctl tool has been
1167 updated to support systemd-boot.
1169 * An EFI kernel stub has been added that may be used to create
1170 kernel EFI binaries that contain not only the actual kernel,
1171 but also an initrd, boot splash, command line and OS release
1172 information. This combined binary can then be signed as a
1173 single image, so that the firmware can verify it all in one
1174 step. systemd-boot has special support for EFI binaries created
1175 like this and can extract OS release information from them
1176 and show them in the boot menu. This functionality is useful
1177 to implement cryptographically verified boot schemes.
1179 * Optional support has been added to systemd-fsck to pass
1180 fsck's progress report to an AF_UNIX socket in the file
1183 * udev will no longer create device symlinks for all block
1184 devices by default. A blacklist for excluding special block
1185 devices from this logic has been turned into a whitelist
1186 that requires picking block devices explicitly that require
1189 * A new (currently still internal) API sd-device.h has been
1190 added to libsystemd. This modernized API is supposed to
1191 replace libudev eventually. In fact, already much of libudev
1192 is now just a wrapper around sd-device.h.
1194 * A new hwdb database for storing metadata about pointing
1195 stick devices has been added.
1197 * systemd-tmpfiles gained support for setting file attributes
1198 similar to the "chattr" tool with new 'h' and 'H' lines.
1200 * systemd-journald will no longer unconditionally set the
1201 btrfs NOCOW flag on new journal files. This is instead done
1202 with tmpfiles snippet using the new 'h' line type. This
1203 allows easy disabling of this logic, by masking the
1204 journal-nocow.conf tmpfiles file.
1206 * systemd-journald will now translate audit message types to
1207 human readable identifiers when writing them to the
1208 journal. This should improve readability of audit messages.
1210 * The LUKS logic gained support for the offset= and skip=
1211 options in /etc/crypttab, as previously implemented by
1214 * /usr/lib/os-release gained a new optional field VARIANT= for
1215 distributions that support multiple variants (such as a
1216 desktop edition, a server edition, ...)
1218 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1219 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1220 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1221 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1222 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1223 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1224 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1225 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1226 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1227 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1228 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1229 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1230 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1231 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1232 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1233 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1234 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1235 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1236 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1237 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1238 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1239 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1240 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1241 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1242 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1243 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1244 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1246 -- Berlin, 2015-05-22
1250 * Introduce a new API "sd-hwdb.h" for querying the hardware
1251 metadata database. With this minimal interface one can query
1252 and enumerate the udev hwdb, decoupled from the old libudev
1253 library. libudev's interface for this is now only a wrapper
1254 around sd-hwdb. A new tool systemd-hwdb has been added to
1255 interface with and update the database.
1257 * When any of systemd's tools copies files (for example due to
1258 tmpfiles' C lines) a btrfs reflink will attempted first,
1259 before bytewise copying is done.
1261 * systemd-nspawn gained a new --ephemeral switch. When
1262 specified a btrfs snapshot is taken of the container's root
1263 directory, and immediately removed when the container
1264 terminates again. Thus, a container can be started whose
1265 changes never alter the container's root directory, and are
1266 lost on container termination. This switch can also be used
1267 for starting a container off the root file system of the
1268 host without affecting the host OS. This switch is only
1269 available on btrfs file systems.
1271 * systemd-nspawn gained a new --template= switch. It takes the
1272 path to a container tree to use as template for the tree
1273 specified via --directory=, should that directory be
1274 missing. This allows instantiating containers dynamically,
1275 on first run. This switch is only available on btrfs file
1278 * When a .mount unit refers to a mount point on which multiple
1279 mounts are stacked, and the .mount unit is stopped all of
1280 the stacked mount points will now be unmounted until no
1281 mount point remains.
1283 * systemd now has an explicit notion of supported and
1284 unsupported unit types. Jobs enqueued for unsupported unit
1285 types will now fail with an "unsupported" error code. More
1286 specifically .swap, .automount and .device units are not
1287 supported in containers, .busname units are not supported on
1288 non-kdbus systems. .swap and .automount are also not
1289 supported if their respective kernel compile time options
1292 * machinectl gained support for two new "copy-from" and
1293 "copy-to" commands for copying files from a running
1294 container to the host or vice versa.
1296 * machinectl gained support for a new "bind" command to bind
1297 mount host directories into local containers. This is
1298 currently only supported for nspawn containers.
1300 * networkd gained support for configuring bridge forwarding
1301 database entries (fdb) from .network files.
1303 * A new tiny daemon "systemd-importd" has been added that can
1304 download container images in tar, raw, qcow2 or dkr formats,
1305 and make them available locally in /var/lib/machines, so
1306 that they can run as nspawn containers. The daemon can GPG
1307 verify the downloads (not supported for dkr, since it has no
1308 provisions for verifying downloads). It will transparently
1309 decompress bz2, xz, gzip compressed downloads if necessary,
1310 and restore sparse files on disk. The daemon uses privilege
1311 separation to ensure the actual download logic runs with
1312 fewer privileges than the daemon itself. machinectl has
1313 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1314 make the functionality of importd available to the
1315 user. With this in place the Fedora and Ubuntu "Cloud"
1316 images can be downloaded and booted as containers unmodified
1317 (the Fedora images lack the appropriate GPG signature files
1318 currently, so they cannot be verified, but this will change
1319 soon, hopefully). Note that downloading images is currently
1320 only fully supported on btrfs.
1322 * machinectl is now able to list container images found in
1323 /var/lib/machines, along with some metadata about sizes of
1324 disk and similar. If the directory is located on btrfs and
1325 quota is enabled, this includes quota display. A new command
1326 "image-status" has been added that shows additional
1327 information about images.
1329 * machinectl is now able to clone container images
1330 efficiently, if the underlying file system (btrfs) supports
1331 it, with the new "machinectl list-images" command. It also
1332 gained commands for renaming and removing images, as well as
1333 marking them read-only or read-write (supported also on
1334 legacy file systems).
1336 * networkd gained support for collecting LLDP network
1337 announcements, from hardware that supports this. This is
1338 shown in networkctl output.
1340 * systemd-run gained support for a new -t (--pty) switch for
1341 invoking a binary on a pty whose input and output is
1342 connected to the invoking terminal. This allows executing
1343 processes as system services while interactively
1344 communicating with them via the terminal. Most interestingly
1345 this is supported across container boundaries. Invoking
1346 "systemd-run -t /bin/bash" is an alternative to running a
1347 full login session, the difference being that the former
1348 will not register a session, nor go through the PAM session
1351 * tmpfiles gained support for a new "v" line type for creating
1352 btrfs subvolumes. If the underlying file system is a legacy
1353 file system, this automatically degrades to creating a
1354 normal directory. Among others /var/lib/machines is now
1355 created like this at boot, should it be missing.
1357 * The directory /var/lib/containers/ has been deprecated and
1358 been replaced by /var/lib/machines. The term "machines" has
1359 been used in the systemd context as generic term for both
1360 VMs and containers, and hence appears more appropriate for
1361 this, as the directory can also contain raw images bootable
1364 * systemd-nspawn when invoked with -M but without --directory=
1365 or --image= is now capable of searching for the container
1366 root directory, subvolume or disk image automatically, in
1367 /var/lib/machines. systemd-nspawn@.service has been updated
1368 to make use of this, thus allowing it to be used for raw
1371 * A new machines.target unit has been introduced that is
1372 supposed to group all containers/VMs invoked as services on
1373 the system. systemd-nspawn@.service has been updated to
1374 integrate with that.
1376 * machinectl gained a new "start" command, for invoking a
1377 container as a service. "machinectl start foo" is mostly
1378 equivalent to "systemctl start systemd-nspawn@foo.service",
1379 but handles escaping in a nicer way.
1381 * systemd-nspawn will now mount most of the cgroupfs tree
1382 read-only into each container, with the exception of the
1383 container's own subtree in the name=systemd hierarchy.
1385 * journald now sets the special FS_NOCOW file flag for its
1386 journal files. This should improve performance on btrfs, by
1387 avoiding heavy fragmentation when journald's write-pattern
1388 is used on COW file systems. It degrades btrfs' data
1389 integrity guarantees for the files to the same levels as for
1390 ext3/ext4 however. This should be OK though as journald does
1391 its own data integrity checks and all its objects are
1392 checksummed on disk. Also, journald should handle btrfs disk
1393 full events a lot more gracefully now, by processing SIGBUS
1394 errors, and not relying on fallocate() anymore.
1396 * When journald detects that journal files it is writing to
1397 have been deleted it will immediately start new journal
1400 * systemd now provides a way to store file descriptors
1401 per-service in PID 1.This is useful for daemons to ensure
1402 that fds they require are not lost during a daemon
1403 restart. The fds are passed to the daemon on the next
1404 invocation in the same way socket activation fds are
1405 passed. This is now used by journald to ensure that the
1406 various sockets connected to all the system's stdout/stderr
1407 are not lost when journald is restarted. File descriptors
1408 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1409 an extension to sd_notify(). Note that a limit is enforced
1410 on the number of fds a service can store in PID 1, and it
1411 defaults to 0, so that no fds may be stored, unless this is
1412 explicitly turned on.
1414 * The default TERM variable to use for units connected to a
1415 terminal, when no other value is explicitly is set is now
1416 vt220 rather than vt102. This should be fairly safe still,
1417 but allows PgUp/PgDn work.
1419 * The /etc/crypttab option header= as known from Debian is now
1422 * "loginctl user-status" and "loginctl session-status" will
1423 now show the last 10 lines of log messages of the
1424 user/session following the status output. Similar,
1425 "machinectl status" will show the last 10 log lines
1426 associated with a virtual machine or container
1427 service. (Note that this is usually not the log messages
1428 done in the VM/container itself, but simply what the
1429 container manager logs. For nspawn this includes all console
1432 * "loginctl session-status" without further argument will now
1433 show the status of the session of the caller. Similar,
1434 "lock-session", "unlock-session", "activate",
1435 "enable-linger", "disable-linger" may now be called without
1436 session/user parameter in which case they apply to the
1437 caller's session/user.
1439 * An X11 session scriptlet is now shipped that uploads
1440 $DISPLAY and $XAUTHORITY into the environment of the systemd
1441 --user daemon if a session begins. This should improve
1442 compatibility with X11 enabled applications run as systemd
1445 * Generators are now subject to masking via /etc and /run, the
1446 same way as unit files.
1448 * networkd .network files gained support for configuring
1449 per-link IPv4/IPv6 packet forwarding as well as IPv4
1450 masquerading. This is by default turned on for veth links to
1451 containers, as registered by systemd-nspawn. This means that
1452 nspawn containers run with --network-veth will now get
1453 automatic routed access to the host's networks without any
1454 further configuration or setup, as long as networkd runs on
1457 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1458 or UDP posts of a container on the host. With this in place
1459 it is possible to run containers with private veth links
1460 (--network-veth), and have their functionality exposed on
1461 the host as if their services were running directly on the
1464 * systemd-nspawn's --network-veth switch now gained a short
1465 version "-n", since with the changes above it is now truly
1466 useful out-of-the-box. The systemd-nspawn@.service has been
1467 updated to make use of it too by default.
1469 * systemd-nspawn will now maintain a per-image R/W lock, to
1470 ensure that the same image is not started more than once
1471 writable. (It's OK to run an image multiple times
1472 simultaneously in read-only mode.)
1474 * systemd-nspawn's --image= option is now capable of
1475 dissecting and booting MBR and GPT disk images that contain
1476 only a single active Linux partition. Previously it
1477 supported only GPT disk images with proper GPT type
1478 IDs. This allows running cloud images from major
1479 distributions directly with systemd-nspawn, without
1482 * In addition to collecting mouse dpi data in the udev
1483 hardware database, there's now support for collecting angle
1484 information for mouse scroll wheels. The database is
1485 supposed to guarantee similar scrolling behavior on mice
1486 that it knows about. There's also support for collecting
1487 information about Touchpad types.
1489 * udev's input_id built-in will now also collect touch screen
1490 dimension data and attach it to probed devices.
1492 * /etc/os-release gained support for a Distribution Privacy
1495 * networkd gained support for creating "ipvlan", "gretap",
1496 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1498 * systemd-tmpfiles gained support for "a" lines for setting
1501 * systemd-nspawn will now mount /tmp in the container to
1502 tmpfs, automatically.
1504 * systemd now exposes the memory.usage_in_bytes cgroup
1505 attribute and shows it for each service in the "systemctl
1506 status" output, if available.
1508 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1509 immediate reboot is triggered. This useful if shutdown is
1510 hung and is unable to complete, to expedite the
1511 operation. Note that this kind of reboot will still unmount
1512 all file systems, and hence should not result in fsck being
1515 * A .device unit for an optical block device will now be
1516 considered active only when a medium is in the drive. Also,
1517 mount units are now bound to their backing devices thus
1518 triggering automatic unmounting when devices become
1519 unavailable. With this in place systemd will now
1520 automatically unmount left-over mounts when a CD-ROM is
1521 ejected or an USB stick is yanked from the system.
1523 * networkd-wait-online now has support for waiting for
1524 specific interfaces only (with globbing), and for giving up
1525 after a configurable timeout.
1527 * networkd now exits when idle. It will be automatically
1528 restarted as soon as interfaces show up, are removed or
1529 change state. networkd will stay around as long as there is
1530 at least one DHCP state machine or similar around, that keep
1533 * networkd may now configure IPv6 link-local addressing in
1534 addition to IPv4 link-local addressing.
1536 * The IPv6 "token" for use in SLAAC may now be configured for
1537 each .network interface in networkd.
1539 * Routes configured with networkd may now be assigned a scope
1542 * networkd's [Match] sections now support globbing and lists
1543 of multiple space-separated matches per item.
1545 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1546 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1547 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1548 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1549 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1550 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1551 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1552 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1553 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1554 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1555 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1556 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1557 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1558 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1559 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1560 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1561 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1562 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1563 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1564 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1565 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1566 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1567 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1568 Hoffmann, Zbigniew Jędrzejewski-Szmek
1570 -- Berlin, 2015-02-16
1574 * When querying unit file enablement status (for example via
1575 "systemctl is-enabled"), a new state "indirect" is now known
1576 which indicates that a unit might not be enabled itself, but
1577 another unit listed in its Also= setting might be.
1579 * Similar to the various existing ConditionXYZ= settings for
1580 units, there are now matching AssertXYZ= settings. While
1581 failing conditions cause a unit to be skipped, but its job
1582 to succeed, failing assertions declared like this will cause
1583 a unit start operation and its job to fail.
1585 * hostnamed now knows a new chassis type "embedded".
1587 * systemctl gained a new "edit" command. When used on a unit
1588 file, this allows extending unit files with .d/ drop-in
1589 configuration snippets or editing the full file (after
1590 copying it from /usr/lib to /etc). This will invoke the
1591 user's editor (as configured with $EDITOR), and reload the
1592 modified configuration after editing.
1594 * "systemctl status" now shows the suggested enablement state
1595 for a unit, as declared in the (usually vendor-supplied)
1596 system preset files.
1598 * nss-myhostname will now resolve the single-label host name
1599 "gateway" to the locally configured default IP routing
1600 gateways, ordered by their metrics. This assigns a stable
1601 name to the used gateways, regardless which ones are
1602 currently configured. Note that the name will only be
1603 resolved after all other name sources (if nss-myhostname is
1604 configured properly) and should hence not negatively impact
1605 systems that use the single-label host name "gateway" in
1608 * systemd-inhibit now allows filtering by mode when listing
1611 * Scope and service units gained a new "Delegate" boolean
1612 property, which, when set, allows processes running inside the
1613 unit to further partition resources. This is primarily
1614 useful for systemd user instances as well as container
1617 * journald will now pick up audit messages directly from
1618 the kernel, and log them like any other log message. The
1619 audit fields are split up and fully indexed. This means that
1620 journalctl in many ways is now a (nicer!) alternative to
1621 ausearch, the traditional audit client. Note that this
1622 implements only a minimal audit client. If you want the
1623 special audit modes like reboot-on-log-overflow, please use
1624 the traditional auditd instead, which can be used in
1625 parallel to journald.
1627 * The ConditionSecurity= unit file option now understands the
1628 special string "audit" to check whether auditing is
1631 * journalctl gained two new commands --vacuum-size= and
1632 --vacuum-time= to delete old journal files until the
1633 remaining ones take up no more than the specified size on disk,
1634 or are not older than the specified time.
1636 * A new, native PPPoE library has been added to sd-network,
1637 systemd's library of light-weight networking protocols. This
1638 library will be used in a future version of networkd to
1639 enable PPPoE communication without an external pppd daemon.
1641 * The busctl tool now understands a new "capture" verb that
1642 works similar to "monitor", but writes a packet capture
1643 trace to STDOUT that can be redirected to a file which is
1644 compatible with libcap's capture file format. This can then
1645 be loaded in Wireshark and similar tools to inspect bus
1648 * The busctl tool now understands a new "tree" verb that shows
1649 the object trees of a specific service on the bus, or of all
1652 * The busctl tool now understands a new "introspect" verb that
1653 shows all interfaces and members of objects on the bus,
1654 including their signature and values. This is particularly
1655 useful to get more information about bus objects shown by
1656 the new "busctl tree" command.
1658 * The busctl tool now understands new verbs "call",
1659 "set-property" and "get-property" for invoking bus method
1660 calls, setting and getting bus object properties in a
1663 * busctl gained a new --augment-creds= argument that controls
1664 whether the tool shall augment credential information it
1665 gets from the bus with data from /proc, in a possibly
1668 * nspawn's --link-journal= switch gained two new values
1669 "try-guest" and "try-host" that work like "guest" and
1670 "host", but do not fail if the host has no persistent
1671 journalling enabled. -j is now equivalent to
1672 --link-journal=try-guest.
1674 * macvlan network devices created by nspawn will now have
1675 stable MAC addresses.
1677 * A new SmackProcessLabel= unit setting has been added, which
1678 controls the SMACK security label processes forked off by
1679 the respective unit shall use.
1681 * If compiled with --enable-xkbcommon, systemd-localed will
1682 verify x11 keymap settings by compiling the given keymap. It
1683 will spew out warnings if the compilation fails. This
1684 requires libxkbcommon to be installed.
1686 * When a coredump is collected, a larger number of metadata
1687 fields is now collected and included in the journal records
1688 created for it. More specifically, control group membership,
1689 environment variables, memory maps, working directory,
1690 chroot directory, /proc/$PID/status, and a list of open file
1691 descriptors is now stored in the log entry.
1693 * The udev hwdb now contains DPI information for mice. For
1696 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1698 * All systemd programs that read standalone configuration
1699 files in /etc now also support a corresponding series of
1700 .conf.d configuration directories in /etc/, /run/,
1701 /usr/local/lib/, /usr/lib/, and (if configured with
1702 --enable-split-usr) /lib/. In particular, the following
1703 configuration files now have corresponding configuration
1704 directories: system.conf user.conf, logind.conf,
1705 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1706 resolved.conf, timesyncd.conf, journal-remote.conf, and
1707 journal-upload.conf. Note that distributions should use the
1708 configuration directories in /usr/lib/; the directories in
1709 /etc/ are reserved for the system administrator.
1711 * systemd-rfkill will no longer take the rfkill device name
1712 into account when storing rfkill state on disk, as the name
1713 might be dynamically assigned and not stable. Instead, the
1714 ID_PATH udev variable combined with the rfkill type (wlan,
1715 bluetooth, ...) is used.
1717 * A new service systemd-machine-id-commit.service has been
1718 added. When used on systems where /etc is read-only during
1719 boot, and /etc/machine-id is not initialized (but an empty
1720 file), this service will copy the temporary machine ID
1721 created as replacement into /etc after the system is fully
1722 booted up. This is useful for systems that are freshly
1723 installed with a non-initialized machine ID, but should get
1724 a fixed machine ID for subsequent boots.
1726 * networkd's .netdev files now provide a large set of
1727 configuration parameters for VXLAN devices. Similarly, the
1728 bridge port cost parameter is now configurable in .network
1729 files. There's also new support for configuring IP source
1730 routing. networkd .link files gained support for a new
1731 OriginalName= match that is useful to match against the
1732 original interface name the kernel assigned. .network files
1733 may include MTU= and MACAddress= fields for altering the MTU
1734 and MAC address while being connected to a specific network
1737 * The LUKS logic gained supported for configuring
1738 UUID-specific key files. There's also new support for naming
1739 LUKS device from the kernel command line, using the new
1740 luks.name= argument.
1742 * Timer units may now be transiently created via the bus API
1743 (this was previously already available for scope and service
1744 units). In addition it is now possible to create multiple
1745 transient units at the same time with a single bus call. The
1746 "systemd-run" tool has been updated to make use of this for
1747 running commands on a specified time, in at(1)-style.
1749 * tmpfiles gained support for "t" lines, for assigning
1750 extended attributes to files. Among other uses this may be
1751 used to assign SMACK labels to files.
1753 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1754 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1755 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1756 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1757 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1758 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1759 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1760 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1761 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1762 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1763 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1764 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1765 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1766 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1767 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1768 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1769 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1770 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1772 -- Berlin, 2014-12-10
1776 * journalctl gained the new options -t/--identifier= to match
1777 on the syslog identifier (aka "tag"), as well as --utc to
1778 show log timestamps in the UTC timezone. journalctl now also
1779 accepts -n/--lines=all to disable line capping in a pager.
1781 * journalctl gained a new switch, --flush, that synchronously
1782 flushes logs from /run/log/journal to /var/log/journal if
1783 persistent storage is enabled. systemd-journal-flush.service
1784 now waits until the operation is complete.
1786 * Services can notify the manager before they start a reload
1787 (by sending RELOADING=1) or shutdown (by sending
1788 STOPPING=1). This allows the manager to track and show the
1789 internal state of daemons and closes a race condition when
1790 the process is still running but has closed its D-Bus
1793 * Services with Type=oneshot do not have to have any ExecStart
1796 * User units are now loaded also from
1797 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1798 /run/systemd/user directory that was already previously
1799 supported, but is under the control of the user.
1801 * Job timeouts (i.e. time-outs on the time a job that is
1802 queued stays in the run queue) can now optionally result in
1803 immediate reboot or power-off actions (JobTimeoutAction= and
1804 JobTimeoutRebootArgument=). This is useful on ".target"
1805 units, to limit the maximum time a target remains
1806 undispatched in the run queue, and to trigger an emergency
1807 operation in such a case. This is now used by default to
1808 turn off the system if boot-up (as defined by everything in
1809 basic.target) hangs and does not complete for at least
1810 15min. Also, if power-off or reboot hang for at least 30min
1811 an immediate power-off/reboot operation is triggered. This
1812 functionality is particularly useful to increase reliability
1813 on embedded devices, but also on laptops which might
1814 accidentally get powered on when carried in a backpack and
1815 whose boot stays stuck in a hard disk encryption passphrase
1818 * systemd-logind can be configured to also handle lid switch
1819 events even when the machine is docked or multiple displays
1820 are attached (HandleLidSwitchDocked= option).
1822 * A helper binary and a service have been added which can be
1823 used to resume from hibernation in the initramfs. A
1824 generator will parse the resume= option on the kernel
1825 command line to trigger resume.
1827 * A user console daemon systemd-consoled has been
1828 added. Currently, it is a preview, and will so far open a
1829 single terminal on each session of the user marked as
1830 Desktop=systemd-console.
1832 * Route metrics can be specified for DHCP routes added by
1835 * The SELinux context of socket-activated services can be set
1836 from the information provided by the networking stack
1837 (SELinuxContextFromNet= option).
1839 * Userspace firmware loading support has been removed and
1840 the minimum supported kernel version is thus bumped to 3.7.
1842 * Timeout for udev workers has been increased from 1 to 3
1843 minutes, but a warning will be printed after 1 minute to
1844 help diagnose kernel modules that take a long time to load.
1846 * Udev rules can now remove tags on devices with TAG-="foobar".
1848 * systemd's readahead implementation has been removed. In many
1849 circumstances it didn't give expected benefits even for
1850 rotational disk drives and was becoming less relevant in the
1851 age of SSDs. As none of the developers has been using
1852 rotating media anymore, and nobody stepped up to actively
1853 maintain this component of systemd it has now been removed.
1855 * Swap units can use Options= to specify discard options.
1856 Discard options specified for swaps in /etc/fstab are now
1859 * Docker containers are now detected as a separate type of
1862 * The Password Agent protocol gained support for queries where
1863 the user input is shown, useful e.g. for user names.
1864 systemd-ask-password gained a new --echo option to turn that
1867 * The default sysctl.d/ snippets will now set:
1869 net.core.default_qdisc = fq_codel
1871 This selects Fair Queuing Controlled Delay as the default
1872 queuing discipline for network interfaces. fq_codel helps
1873 fight the network bufferbloat problem. It is believed to be
1874 a good default with no tuning required for most workloads.
1875 Downstream distributions may override this choice. On 10Gbit
1876 servers that do not do forwarding, "fq" may perform better.
1877 Systems without a good clocksource should use "pfifo_fast".
1879 * If kdbus is enabled during build a new option BusPolicy= is
1880 available for service units, that allows locking all service
1881 processes into a stricter bus policy, in order to limit
1882 access to various bus services, or even hide most of them
1883 from the service's view entirely.
1885 * networkctl will now show the .network and .link file
1886 networkd has applied to a specific interface.
1888 * sd-login gained a new API call sd_session_get_desktop() to
1889 query which desktop environment has been selected for a
1892 * UNIX utmp support is now compile-time optional to support
1893 legacy-free systems.
1895 * systemctl gained two new commands "add-wants" and
1896 "add-requires" for pulling in units from specific targets
1899 * If the word "rescue" is specified on the kernel command line
1900 the system will now boot into rescue mode (aka
1901 rescue.target), which was previously available only by
1902 specifying "1" or "systemd.unit=rescue.target" on the kernel
1903 command line. This new kernel command line option nicely
1904 mirrors the already existing "emergency" kernel command line
1907 * New kernel command line options mount.usr=, mount.usrflags=,
1908 mount.usrfstype= have been added that match root=, rootflags=,
1909 rootfstype= but allow mounting a specific file system to
1912 * The $NOTIFY_SOCKET is now also passed to control processes of
1913 services, not only the main process.
1915 * This version reenables support for fsck's -l switch. This
1916 means at least version v2.25 of util-linux is required for
1917 operation, otherwise dead-locks on device nodes may
1918 occur. Again: you need to update util-linux to at least
1919 v2.25 when updating systemd to v217.
1921 * The "multi-seat-x" tool has been removed from systemd, as
1922 its functionality has been integrated into X servers 1.16,
1923 and the tool is hence redundant. It is recommended to update
1924 display managers invoking this tool to simply invoke X
1925 directly from now on, again.
1927 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1928 message flag has been added for all of systemd's PolicyKit
1929 authenticated method calls has been added. In particular
1930 this now allows optional interactive authorization via
1931 PolicyKit for many of PID1's privileged operations such as
1932 unit file enabling and disabling.
1934 * "udevadm hwdb --update" learnt a new switch "--usr" for
1935 placing the rebuilt hardware database in /usr instead of
1936 /etc. When used only hardware database entries stored in
1937 /usr will be used, and any user database entries in /etc are
1938 ignored. This functionality is useful for vendors to ship a
1939 pre-built database on systems where local configuration is
1940 unnecessary or unlikely.
1942 * Calendar time specifications in .timer units now also
1943 understand the strings "semi-annually", "quarterly" and
1944 "minutely" as shortcuts (in addition to the preexisting
1945 "anually", "hourly", ...).
1947 * systemd-tmpfiles will now correctly create files in /dev
1948 at boot which are marked for creation only at boot. It is
1949 recommended to always create static device nodes with 'c!'
1950 and 'b!', so that they are created only at boot and not
1951 overwritten at runtime.
1953 * When the watchdog logic is used for a service (WatchdogSec=)
1954 and the watchdog timeout is hit the service will now be
1955 terminated with SIGABRT (instead of just SIGTERM), in order
1956 to make sure a proper coredump and backtrace is
1957 generated. This ensures that hanging services will result in
1958 similar coredump/backtrace behaviour as services that hit a
1961 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1962 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1963 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1964 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1965 Herrmann, David Sommerseth, David Strauss, Emil Renner
1966 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1967 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1968 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1969 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1970 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1971 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1972 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1973 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1974 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1975 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1976 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1977 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1978 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1979 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1980 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1981 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1984 -- Berlin, 2014-10-28
1988 * timedated no longer reads NTP implementation unit names from
1989 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1990 implementations should add a
1992 Conflicts=systemd-timesyncd.service
1994 to their unit files to take over and replace systemd's NTP
1995 default functionality.
1997 * systemd-sysusers gained a new line type "r" for configuring
1998 which UID/GID ranges to allocate system users/groups
1999 from. Lines of type "u" may now add an additional column
2000 that specifies the home directory for the system user to be
2001 created. Also, systemd-sysusers may now optionally read user
2002 information from STDIN instead of a file. This is useful for
2003 invoking it from RPM preinst scriptlets that need to create
2004 users before the first RPM file is installed since these
2005 files might need to be owned by them. A new
2006 %sysusers_create_inline RPM macro has been introduced to do
2007 just that. systemd-sysusers now updates the shadow files as
2008 well as the user/group databases, which should enhance
2009 compatibility with certain tools like grpck.
2011 * A number of bus APIs of PID 1 now optionally consult
2012 PolicyKit to permit access for otherwise unprivileged
2013 clients under certain conditions. Note that this currently
2014 doesn't support interactive authentication yet, but this is
2015 expected to be added eventually, too.
2017 * /etc/machine-info now has new fields for configuring the
2018 deployment environment of the machine, as well as the
2019 location of the machine. hostnamectl has been updated with
2020 new command to update these fields.
2022 * systemd-timesyncd has been updated to automatically acquire
2023 NTP server information from systemd-networkd, which might
2024 have been discovered via DHCP.
2026 * systemd-resolved now includes a caching DNS stub resolver
2027 and a complete LLMNR name resolution implementation. A new
2028 NSS module "nss-resolve" has been added which can be used
2029 instead of glibc's own "nss-dns" to resolve hostnames via
2030 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2031 be resolved via systemd-resolved D-Bus APIs. In contrast to
2032 the glibc internal resolver systemd-resolved is aware of
2033 multi-homed system, and keeps DNS server and caches separate
2034 and per-interface. Queries are sent simultaneously on all
2035 interfaces that have DNS servers configured, in order to
2036 properly handle VPNs and local LANs which might resolve
2037 separate sets of domain names. systemd-resolved may acquire
2038 DNS server information from systemd-networkd automatically,
2039 which in turn might have discovered them via DHCP. A tool
2040 "systemd-resolve-host" has been added that may be used to
2041 query the DNS logic in resolved. systemd-resolved implements
2042 IDNA and automatically uses IDNA or UTF-8 encoding depending
2043 on whether classic DNS or LLMNR is used as transport. In the
2044 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2045 implementation to systemd-resolved.
2047 * A new NSS module nss-mymachines has been added, that
2048 automatically resolves the names of all local registered
2049 containers to their respective IP addresses.
2051 * A new client tool "networkctl" for systemd-networkd has been
2052 added. It currently is entirely passive and will query
2053 networking configuration from udev, rtnetlink and networkd,
2054 and present it to the user in a very friendly
2055 way. Eventually, we hope to extend it to become a full
2056 control utility for networkd.
2058 * .socket units gained a new DeferAcceptSec= setting that
2059 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2060 TCP. Similarly, support for controlling TCP keep-alive
2061 settings has been added (KeepAliveTimeSec=,
2062 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2063 turning off Nagle's algorithm on TCP has been added
2066 * logind learned a new session type "web", for use in projects
2067 like Cockpit which register web clients as PAM sessions.
2069 * timer units with at least one OnCalendar= setting will now
2070 be started only after timer-sync.target has been
2071 reached. This way they will not elapse before the system
2072 clock has been corrected by a local NTP client or
2073 similar. This is particular useful on RTC-less embedded
2074 machines, that come up with an invalid system clock.
2076 * systemd-nspawn's --network-veth= switch should now result in
2077 stable MAC addresses for both the outer and the inner side
2080 * systemd-nspawn gained a new --volatile= switch for running
2081 container instances with /etc or /var unpopulated.
2083 * The kdbus client code has been updated to use the new Linux
2084 3.17 memfd subsystem instead of the old kdbus-specific one.
2086 * systemd-networkd's DHCP client and server now support
2087 FORCERENEW. There are also new configuration options to
2088 configure the vendor client identifier and broadcast mode
2091 * systemd will no longer inform the kernel about the current
2092 timezone, as this is necessarily incorrect and racy as the
2093 kernel has no understanding of DST and similar
2094 concepts. This hence means FAT timestamps will be always
2095 considered UTC, similar to what Android is already
2096 doing. Also, when the RTC is configured to the local time
2097 (rather than UTC) systemd will never synchronize back to it,
2098 as this might confuse Windows at a later boot.
2100 * systemd-analyze gained a new command "verify" for offline
2101 validation of unit files.
2103 * systemd-networkd gained support for a couple of additional
2104 settings for bonding networking setups. Also, the metric for
2105 statically configured routes may now be configured. For
2106 network interfaces where this is appropriate the peer IP
2107 address may now be configured.
2109 * systemd-networkd's DHCP client will no longer request
2110 broadcasting by default, as this tripped up some networks.
2111 For hardware where broadcast is required the feature should
2112 be switched back on using RequestBroadcast=yes.
2114 * systemd-networkd will now set up IPv4LL addresses (when
2115 enabled) even if DHCP is configured successfully.
2117 * udev will now default to respect network device names given
2118 by the kernel when the kernel indicates that these are
2119 predictable. This behavior can be tweaked by changing
2120 NamePolicy= in the relevant .link file.
2122 * A new library systemd-terminal has been added that
2123 implements full TTY stream parsing and rendering. This
2124 library is supposed to be used later on for implementing a
2125 full userspace VT subsystem, replacing the current kernel
2128 * A new tool systemd-journal-upload has been added to push
2129 journal data to a remote system running
2130 systemd-journal-remote.
2132 * journald will no longer forward all local data to another
2133 running syslog daemon. This change has been made because
2134 rsyslog (which appears to be the most commonly used syslog
2135 implementation these days) no longer makes use of this, and
2136 instead pulls the data out of the journal on its own. Since
2137 forwarding the messages to a non-existent syslog server is
2138 more expensive than we assumed we have now turned this
2139 off. If you run a syslog server that is not a recent rsyslog
2140 version, you have to turn this option on again
2141 (ForwardToSyslog= in journald.conf).
2143 * journald now optionally supports the LZ4 compressor for
2144 larger journal fields. This compressor should perform much
2145 better than XZ which was the previous default.
2147 * machinectl now shows the IP addresses of local containers,
2148 if it knows them, plus the interface name of the container.
2150 * A new tool "systemd-escape" has been added that makes it
2151 easy to escape strings to build unit names and similar.
2153 * sd_notify() messages may now include a new ERRNO= field
2154 which is parsed and collected by systemd and shown among the
2155 "systemctl status" output for a service.
2157 * A new component "systemd-firstboot" has been added that
2158 queries the most basic systemd information (timezone,
2159 hostname, root password) interactively on first
2160 boot. Alternatively it may also be used to provision these
2161 things offline on OS images installed into directories.
2163 * The default sysctl.d/ snippets will now set
2165 net.ipv4.conf.default.promote_secondaries=1
2167 This has the benefit of no flushing secondary IP addresses
2168 when primary addresses are removed.
2170 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2171 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2172 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2173 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2174 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2175 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2176 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2177 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2178 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2179 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2180 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2181 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2182 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2183 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2184 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2186 -- Berlin, 2014-08-19
2190 * A new tool systemd-sysusers has been added. This tool
2191 creates system users and groups in /etc/passwd and
2192 /etc/group, based on static declarative system user/group
2193 definitions in /usr/lib/sysusers.d/. This is useful to
2194 enable factory resets and volatile systems that boot up with
2195 an empty /etc directory, and thus need system users and
2196 groups created during early boot. systemd now also ships
2197 with two default sysusers.d/ files for the most basic
2198 users and groups systemd and the core operating system
2201 * A new tmpfiles snippet has been added that rebuilds the
2202 essential files in /etc on boot, should they be missing.
2204 * A directive for ensuring automatic clean-up of
2205 /var/cache/man/ has been removed from the default
2206 configuration. This line should now be shipped by the man
2207 implementation. The necessary change has been made to the
2208 man-db implementation. Note that you need to update your man
2209 implementation to one that ships this line, otherwise no
2210 automatic clean-up of /var/cache/man will take place.
2212 * A new condition ConditionNeedsUpdate= has been added that
2213 may conditionalize services to only run when /etc or /var
2214 are "older" than the vendor operating system resources in
2215 /usr. This is useful for reconstructing or updating /etc
2216 after an offline update of /usr or a factory reset, on the
2217 next reboot. Services that want to run once after such an
2218 update or reset should use this condition and order
2219 themselves before the new systemd-update-done.service, which
2220 will mark the two directories as fully updated. A number of
2221 service files have been added making use of this, to rebuild
2222 the udev hardware database, the journald message catalog and
2223 dynamic loader cache (ldconfig). The systemd-sysusers tool
2224 described above also makes use of this now. With this in
2225 place it is now possible to start up a minimal operating
2226 system with /etc empty cleanly. For more information on the
2227 concepts involved see this recent blog story:
2229 http://0pointer.de/blog/projects/stateless.html
2231 * A new system group "input" has been introduced, and all
2232 input device nodes get this group assigned. This is useful
2233 for system-level software to get access to input devices. It
2234 complements what is already done for "audio" and "video".
2236 * systemd-networkd learnt minimal DHCPv4 server support in
2237 addition to the existing DHCPv4 client support. It also
2238 learnt DHCPv6 client and IPv6 Router Solicitation client
2239 support. The DHCPv4 client gained support for static routes
2240 passed in from the server. Note that the [DHCPv4] section
2241 known in older systemd-networkd versions has been renamed to
2242 [DHCP] and is now also used by the DHCPv6 client. Existing
2243 .network files using settings of this section should be
2244 updated, though compatibility is maintained. Optionally, the
2245 client hostname may now be sent to the DHCP server.
2247 * networkd gained support for vxlan virtual networks as well
2248 as tun/tap and dummy devices.
2250 * networkd gained support for automatic allocation of address
2251 ranges for interfaces from a system-wide pool of
2252 addresses. This is useful for dynamically managing a large
2253 number of interfaces with a single network configuration
2254 file. In particular this is useful to easily assign
2255 appropriate IP addresses to the veth links of a large number
2256 of nspawn instances.
2258 * RPM macros for processing sysusers, sysctl and binfmt
2259 drop-in snippets at package installation time have been
2262 * The /etc/os-release file should now be placed in
2263 /usr/lib/os-release. The old location is automatically
2264 created as symlink. /usr/lib is the more appropriate
2265 location of this file, since it shall actually describe the
2266 vendor operating system shipped in /usr, and not the
2267 configuration stored in /etc.
2269 * .mount units gained a new boolean SloppyOptions= setting
2270 that maps to mount(8)'s -s option which enables permissive
2271 parsing of unknown mount options.
2273 * tmpfiles learnt a new "L+" directive which creates a symlink
2274 but (unlike "L") deletes a pre-existing file first, should
2275 it already exist and not already be the correct
2276 symlink. Similarly, "b+", "c+" and "p+" directives have been
2277 added as well, which create block and character devices, as
2278 well as fifos in the filesystem, possibly removing any
2279 pre-existing files of different types.
2281 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2282 'argument' field (which so far specified the source to
2283 symlink/copy the files from) is now optional. If omitted the
2284 same file os copied from /usr/share/factory/ suffixed by the
2285 full destination path. This is useful for populating /etc
2286 with essential files, by copying them from vendor defaults
2287 shipped in /usr/share/factory/etc.
2289 * A new command "systemctl preset-all" has been added that
2290 applies the service preset settings to all installed unit
2291 files. A new switch --preset-mode= has been added that
2292 controls whether only enable or only disable operations
2295 * A new command "systemctl is-system-running" has been added
2296 that allows checking the overall state of the system, for
2297 example whether it is fully up and running.
2299 * When the system boots up with an empty /etc, the equivalent
2300 to "systemctl preset-all" is executed during early boot, to
2301 make sure all default services are enabled after a factory
2304 * systemd now contains a minimal preset file that enables the
2305 most basic services systemd ships by default.
2307 * Unit files' [Install] section gained a new DefaultInstance=
2308 field for defining the default instance to create if a
2309 template unit is enabled with no instance specified.
2311 * A new passive target cryptsetup-pre.target has been added
2312 that may be used by services that need to make they run and
2313 finish before the first LUKS cryptographic device is set up.
2315 * The /dev/loop-control and /dev/btrfs-control device nodes
2316 are now owned by the "disk" group by default, opening up
2317 access to this group.
2319 * systemd-coredump will now automatically generate a
2320 stack trace of all core dumps taking place on the system,
2321 based on elfutils' libdw library. This stack trace is logged
2324 * systemd-coredump may now optionally store coredumps directly
2325 on disk (in /var/lib/systemd/coredump, possibly compressed),
2326 instead of storing them unconditionally in the journal. This
2327 mode is the new default. A new configuration file
2328 /etc/systemd/coredump.conf has been added to configure this
2329 and other parameters of systemd-coredump.
2331 * coredumpctl gained a new "info" verb to show details about a
2332 specific coredump. A new switch "-1" has also been added
2333 that makes sure to only show information about the most
2334 recent entry instead of all entries. Also, as the tool is
2335 generally useful now the "systemd-" prefix of the binary
2336 name has been removed. Distributions that want to maintain
2337 compatibility with the old name should add a symlink from
2338 the old name to the new name.
2340 * journald's SplitMode= now defaults to "uid". This makes sure
2341 that unprivileged users can access their own coredumps with
2342 coredumpctl without restrictions.
2344 * New kernel command line options "systemd.wants=" (for
2345 pulling an additional unit during boot), "systemd.mask="
2346 (for masking a specific unit for the boot), and
2347 "systemd.debug-shell" (for enabling the debug shell on tty9)
2348 have been added. This is implemented in the new generator
2349 "systemd-debug-generator".
2351 * systemd-nspawn will now by default filter a couple of
2352 syscalls for containers, among them those required for
2353 kernel module loading, direct x86 IO port access, swap
2354 management, and kexec. Most importantly though
2355 open_by_handle_at() is now prohibited for containers,
2356 closing a hole similar to a recently discussed vulnerability
2357 in docker regarding access to files on file hierarchies the
2358 container should normally not have access to. Note that, for
2359 nspawn, we generally make no security claims anyway (and
2360 this is explicitly documented in the man page), so this is
2361 just a fix for one of the most obvious problems.
2363 * A new man page file-hierarchy(7) has been added that
2364 contains a minimized, modernized version of the file system
2365 layout systemd expects, similar in style to the FHS
2366 specification or hier(5). A new tool systemd-path(1) has
2367 been added to query many of these paths for the local
2370 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2371 longer done. Since the directory now has a per-user size
2372 limit, and is cleaned on logout this appears unnecessary,
2373 in particular since this now brings the lifecycle of this
2374 directory closer in line with how IPC objects are handled.
2376 * systemd.pc now exports a number of additional directories,
2377 including $libdir (which is useful to identify the library
2378 path for the primary architecture of the system), and a
2379 couple of drop-in directories.
2381 * udev's predictable network interface names now use the dev_port
2382 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2383 distinguish between ports of the same PCI function. dev_id should
2384 only be used for ports using the same HW address, hence the need
2387 * machined has been updated to export the OS version of a
2388 container (read from /etc/os-release and
2389 /usr/lib/os-release) on the bus. This is now shown in
2390 "machinectl status" for a machine.
2392 * A new service setting RestartForceExitStatus= has been
2393 added. If configured to a set of exit signals or process
2394 return values, the service will be restarted when the main
2395 daemon process exits with any of them, regardless of the
2398 * systemctl's -H switch for connecting to remote systemd
2399 machines has been extended so that it may be used to
2400 directly connect to a specific container on the
2401 host. "systemctl -H root@foobar:waldi" will now connect as
2402 user "root" to host "foobar", and then proceed directly to
2403 the container named "waldi". Note that currently you have to
2404 authenticate as user "root" for this to work, as entering
2405 containers is a privileged operation.
2407 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2408 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2409 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2410 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2411 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2412 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2413 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2414 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2415 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2416 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2417 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2418 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2420 -- Berlin, 2014-07-03
2424 * As an experimental feature, udev now tries to lock the
2425 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2426 executes events for the disk or any of its partitions.
2427 Applications like partitioning programs can lock the
2428 disk device node (flock(LOCK_EX)) and claim temporary
2429 device ownership that way; udev will entirely skip all event
2430 handling for this disk and its partitions. If the disk
2431 was opened for writing, the close will trigger a partition
2432 table rescan in udev's "watch" facility, and if needed
2433 synthesize "change" events for the disk and all its partitions.
2434 This is now unconditionally enabled, and if it turns out to
2435 cause major problems, we might turn it on only for specific
2436 devices, or might need to disable it entirely. Device Mapper
2437 devices are excluded from this logic.
2439 * We temporarily dropped the "-l" switch for fsck invocations,
2440 since they collide with the flock() logic above. util-linux
2441 upstream has been changed already to avoid this conflict,
2442 and we will readd "-l" as soon as util-linux with this
2443 change has been released.
2445 * The dependency on libattr has been removed. Since a long
2446 time, the extended attribute calls have moved to glibc, and
2447 libattr is thus unnecessary.
2449 * Virtualization detection works without priviliges now. This
2450 means the systemd-detect-virt binary no longer requires
2451 CAP_SYS_PTRACE file capabilities, and our daemons can run
2452 with fewer privileges.
2454 * systemd-networkd now runs under its own "systemd-network"
2455 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2456 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2457 loses the ability to write to files owned by root this way.
2459 * Similarly, systemd-resolved now runs under its own
2460 "systemd-resolve" user with no capabilities remaining.
2462 * Similarly, systemd-bus-proxyd now runs under its own
2463 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2465 * systemd-networkd gained support for setting up "veth"
2466 virtual Ethernet devices for container connectivity, as well
2467 as GRE and VTI tunnels.
2469 * systemd-networkd will no longer automatically attempt to
2470 manually load kernel modules necessary for certain tunnel
2471 transports. Instead, it is assumed the kernel loads them
2472 automatically when required. This only works correctly on
2473 very new kernels. On older kernels, please consider adding
2474 the kernel modules to /etc/modules-load.d/ as a work-around.
2476 * The resolv.conf file systemd-resolved generates has been
2477 moved to /run/systemd/resolve/. If you have a symlink from
2478 /etc/resolv.conf, it might be necessary to correct it.
2480 * Two new service settings, ProtectHome= and ProtectSystem=,
2481 have been added. When enabled, they will make the user data
2482 (such as /home) inaccessible or read-only and the system
2483 (such as /usr) read-only, for specific services. This allows
2484 very light-weight per-service sandboxing to avoid
2485 modifications of user data or system files from
2486 services. These two new switches have been enabled for all
2487 of systemd's long-running services, where appropriate.
2489 * Socket units gained new SocketUser= and SocketGroup=
2490 settings to set the owner user and group of AF_UNIX sockets
2491 and FIFOs in the file system.
2493 * Socket units gained a new RemoveOnStop= setting. If enabled,
2494 all FIFOS and sockets in the file system will be removed
2495 when the specific socket unit is stopped.
2497 * Socket units gained a new Symlinks= setting. It takes a list
2498 of symlinks to create to file system sockets or FIFOs
2499 created by the specific Unix sockets. This is useful to
2500 manage symlinks to socket nodes with the same life-cycle as
2503 * The /dev/log socket and /dev/initctl FIFO have been moved to
2504 /run, and have been replaced by symlinks. This allows
2505 connecting to these facilities even if PrivateDevices=yes is
2506 used for a service (which makes /dev/log itself unavailable,
2507 but /run is left). This also has the benefit of ensuring
2508 that /dev only contains device nodes, directories and
2509 symlinks, and nothing else.
2511 * sd-daemon gained two new calls sd_pid_notify() and
2512 sd_pid_notifyf(). They are similar to sd_notify() and
2513 sd_notifyf(), but allow overriding of the source PID of
2514 notification messages if permissions permit this. This is
2515 useful to send notify messages on behalf of a different
2516 process (for example, the parent process). The
2517 systemd-notify tool has been updated to make use of this
2518 when sending messages (so that notification messages now
2519 originate from the shell script invoking systemd-notify and
2520 not the systemd-notify process itself. This should minimize
2521 a race where systemd fails to associate notification
2522 messages to services when the originating process already
2525 * A new "on-abnormal" setting for Restart= has been added. If
2526 set, it will result in automatic restarts on all "abnormal"
2527 reasons for a process to exit, which includes unclean
2528 signals, core dumps, timeouts and watchdog timeouts, but
2529 does not include clean and unclean exit codes or clean
2530 signals. Restart=on-abnormal is an alternative for
2531 Restart=on-failure for services that shall be able to
2532 terminate and avoid restarts on certain errors, by
2533 indicating so with an unclean exit code. Restart=on-failure
2534 or Restart=on-abnormal is now the recommended setting for
2535 all long-running services.
2537 * If the InaccessibleDirectories= service setting points to a
2538 mount point (or if there are any submounts contained within
2539 it), it is now attempted to completely unmount it, to make
2540 the file systems truly unavailable for the respective
2543 * The ReadOnlyDirectories= service setting and
2544 systemd-nspawn's --read-only parameter are now recursively
2545 applied to all submounts, too.
2547 * Mount units may now be created transiently via the bus APIs.
2549 * The support for SysV and LSB init scripts has been removed
2550 from the systemd daemon itself. Instead, it is now
2551 implemented as a generator that creates native systemd units
2552 from these scripts when needed. This enables us to remove a
2553 substantial amount of legacy code from PID 1, following the
2554 fact that many distributions only ship a very small number
2555 of LSB/SysV init scripts nowadays.
2557 * Privileged Xen (dom0) domains are not considered
2558 virtualization anymore by the virtualization detection
2559 logic. After all, they generally have unrestricted access to
2560 the hardware and usually are used to manage the unprivileged
2563 * systemd-tmpfiles gained a new "C" line type, for copying
2564 files or entire directories.
2566 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2567 lines. So far, they have been non-globbing versions of the
2568 latter, and have thus been redundant. In future, it is
2569 recommended to only use "z". "m" has hence been removed
2570 from the documentation, even though it stays supported.
2572 * A tmpfiles snippet to recreate the most basic structure in
2573 /var has been added. This is enough to create the /var/run →
2574 /run symlink and create a couple of structural
2575 directories. This allows systems to boot up with an empty or
2576 volatile /var. Of course, while with this change, the core OS
2577 now is capable with dealing with a volatile /var, not all
2578 user services are ready for it. However, we hope that sooner
2579 or later, many service daemons will be changed upstream so
2580 that they are able to automatically create their necessary
2581 directories in /var at boot, should they be missing. This is
2582 the first step to allow state-less systems that only require
2583 the vendor image for /usr to boot.
2585 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2586 empty tmpfs instance to a specific directory. This is
2587 particularly useful for making use of the automatic
2588 reconstruction of /var (see above), by passing --tmpfs=/var.
2590 * Access modes specified in tmpfiles snippets may now be
2591 prefixed with "~", which indicates that they shall be masked
2592 by whether the existing file or directory is currently
2593 writable, readable or executable at all. Also, if specified,
2594 the sgid/suid/sticky bits will be masked for all
2597 * A new passive target unit "network-pre.target" has been
2598 added which is useful for services that shall run before any
2599 network is configured, for example firewall scripts.
2601 * The "floppy" group that previously owned the /dev/fd*
2602 devices is no longer used. The "disk" group is now used
2603 instead. Distributions should probably deprecate usage of
2606 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2607 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2608 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2609 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2610 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2611 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2614 -- Berlin, 2014-06-11
2618 * A new "systemd-timesyncd" daemon has been added for
2619 synchronizing the system clock across the network. It
2620 implements an SNTP client. In contrast to NTP
2621 implementations such as chrony or the NTP reference server,
2622 this only implements a client side, and does not bother with
2623 the full NTP complexity, focusing only on querying time from
2624 one remote server and synchronizing the local clock to
2625 it. Unless you intend to serve NTP to networked clients or
2626 want to connect to local hardware clocks, this simple NTP
2627 client should be more than appropriate for most
2628 installations. The daemon runs with minimal privileges, and
2629 has been hooked up with networkd to only operate when
2630 network connectivity is available. The daemon saves the
2631 current clock to disk every time a new NTP sync has been
2632 acquired, and uses this to possibly correct the system clock
2633 early at bootup, in order to accommodate for systems that
2634 lack an RTC such as the Raspberry Pi and embedded devices,
2635 and to make sure that time monotonically progresses on these
2636 systems, even if it is not always correct. To make use of
2637 this daemon, a new system user and group "systemd-timesync"
2638 needs to be created on installation of systemd.
2640 * The queue "seqnum" interface of libudev has been disabled, as
2641 it was generally incompatible with device namespacing as
2642 sequence numbers of devices go "missing" if the devices are
2643 part of a different namespace.
2645 * "systemctl list-timers" and "systemctl list-sockets" gained
2646 a --recursive switch for showing units of these types also
2647 for all local containers, similar in style to the already
2648 supported --recursive switch for "systemctl list-units".
2650 * A new RebootArgument= setting has been added for service
2651 units, which may be used to specify a kernel reboot argument
2652 to use when triggering reboots with StartLimitAction=.
2654 * A new FailureAction= setting has been added for service
2655 units which may be used to specify an operation to trigger
2656 when a service fails. This works similarly to
2657 StartLimitAction=, but unlike it, controls what is done
2658 immediately rather than only after several attempts to
2659 restart the service in question.
2661 * hostnamed got updated to also expose the kernel name,
2662 release, and version on the bus. This is useful for
2663 executing commands like hostnamectl with the -H switch.
2664 systemd-analyze makes use of this to properly display
2665 details when running non-locally.
2667 * The bootchart tool can now show cgroup information in the
2668 graphs it generates.
2670 * The CFS CPU quota cgroup attribute is now exposed for
2671 services. The new CPUQuota= switch has been added for this
2672 which takes a percentage value. Setting this will have the
2673 result that a service may never get more CPU time than the
2674 specified percentage, even if the machine is otherwise idle.
2676 * systemd-networkd learned IPIP and SIT tunnel support.
2678 * LSB init scripts exposing a dependency on $network will now
2679 get a dependency on network-online.target rather than simply
2680 network.target. This should bring LSB handling closer to
2681 what it was on SysV systems.
2683 * A new fsck.repair= kernel option has been added to control
2684 how fsck shall deal with unclean file systems at boot.
2686 * The (.ini) configuration file parser will now silently
2687 ignore sections whose name begins with "X-". This may be
2688 used to maintain application-specific extension sections in unit
2691 * machined gained a new API to query the IP addresses of
2692 registered containers. "machinectl status" has been updated
2693 to show these addresses in its output.
2695 * A new call sd_uid_get_display() has been added to the
2696 sd-login APIs for querying the "primary" session of a
2697 user. The "primary" session of the user is elected from the
2698 user's sessions and generally a graphical session is
2699 preferred over a text one.
2701 * A minimal systemd-resolved daemon has been added. It
2702 currently simply acts as a companion to systemd-networkd and
2703 manages resolv.conf based on per-interface DNS
2704 configuration, possibly supplied via DHCP. In the long run
2705 we hope to extend this into a local DNSSEC enabled DNS and
2708 * The systemd-networkd-wait-online tool is now enabled by
2709 default. It will delay network-online.target until a network
2710 connection has been configured. The tool primarily integrates
2711 with networkd, but will also make a best effort to make sense
2712 of network configuration performed in some other way.
2714 * Two new service options StartupCPUShares= and
2715 StartupBlockIOWeight= have been added that work similarly to
2716 CPUShares= and BlockIOWeight= however only apply during
2717 system startup. This is useful to prioritize certain services
2718 differently during bootup than during normal runtime.
2720 * hostnamed has been changed to prefer the statically
2721 configured hostname in /etc/hostname (unless set to
2722 'localhost' or empty) over any dynamic one supplied by
2723 dhcp. With this change, the rules for picking the hostname
2724 match more closely the rules of other configuration settings
2725 where the local administrator's configuration in /etc always
2726 overrides any other settings.
2728 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2729 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2730 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2731 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2732 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2733 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2734 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2735 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2736 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2737 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2738 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2739 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2740 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2741 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2742 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2743 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2746 -- Beijing, 2014-05-28
2750 * When restoring the screen brightness at boot, stay away from
2751 the darkest setting or from the lowest 5% of the available
2752 range, depending on which is the larger value of both. This
2753 should effectively protect the user from rebooting into a
2754 black screen, should the brightness have been set to minimum
2757 * sd-login gained a new sd_machine_get_class() call to
2758 determine the class ("vm" or "container") of a machine
2759 registered with machined.
2761 * sd-login gained new calls
2762 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2763 to query the identity of the peer of a local AF_UNIX
2764 connection. They operate similarly to their sd_pid_get_xyz()
2767 * PID 1 will now maintain a system-wide system state engine
2768 with the states "starting", "running", "degraded",
2769 "maintenance", "stopping". These states are bound to system
2770 startup, normal runtime, runtime with at least one failed
2771 service, rescue/emergency mode and system shutdown. This
2772 state is shown in the "systemctl status" output when no unit
2773 name is passed. It is useful to determine system state, in
2774 particularly when doing so for many systems or containers at
2777 * A new command "list-machines" has been added to "systemctl"
2778 that lists all local OS containers and shows their system
2779 state (see above), if systemd runs inside of them.
2781 * systemctl gained a new "-r" switch to recursively enumerate
2782 units on all local containers, when used with the
2783 "list-unit" command (which is the default one that is
2784 executed when no parameters are specified).
2786 * The GPT automatic partition discovery logic will now honour
2787 two GPT partition flags: one may be set on a partition to
2788 cause it to be mounted read-only, and the other may be set
2789 on a partition to ignore it during automatic discovery.
2791 * Two new GPT type UUIDs have been added for automatic root
2792 partition discovery, for 32-bit and 64-bit ARM. This is not
2793 particularly useful for discovering the root directory on
2794 these architectures during bare-metal boots (since UEFI is
2795 not common there), but still very useful to allow booting of
2796 ARM disk images in nspawn with the -i option.
2798 * MAC addresses of interfaces created with nspawn's
2799 --network-interface= switch will now be generated from the
2800 machine name, and thus be stable between multiple invocations
2803 * logind will now automatically remove all IPC objects owned
2804 by a user if she or he fully logs out. This makes sure that
2805 users who are logged out cannot continue to consume IPC
2806 resources. This covers SysV memory, semaphores and message
2807 queues as well as POSIX shared memory and message
2808 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2809 limits. With this functionality, that is corrected. This may
2810 be turned off by using the RemoveIPC= switch of logind.conf.
2812 * The systemd-machine-id-setup and tmpfiles tools gained a
2813 --root= switch to operate on a specific root directory,
2816 * journald can now forward logged messages to the TTYs of all
2817 logged in users ("wall"). This is the default for all
2818 emergency messages now.
2820 * A new tool systemd-journal-remote has been added to stream
2821 journal log messages across the network.
2823 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2824 controller trees are mounted into it. Note that the
2825 directories mounted beneath it are not read-only. This is a
2826 security measure and is particularly useful because glibc
2827 actually includes a search logic to pick any tmpfs it can
2828 find to implement shm_open() if /dev/shm is not available
2829 (which it might very well be in namespaced setups).
2831 * machinectl gained a new "poweroff" command to cleanly power
2832 down a local OS container.
2834 * The PrivateDevices= unit file setting will now also drop the
2835 CAP_MKNOD capability from the capability bound set, and
2836 imply DevicePolicy=closed.
2838 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2839 comprehensively on all long-running systemd services where
2840 this is appropriate.
2842 * systemd-udevd will now run in a disassociated mount
2843 namespace. To mount directories from udev rules, make sure to
2844 pull in mount units via SYSTEMD_WANTS properties.
2846 * The kdbus support gained support for uploading policy into
2847 the kernel. sd-bus gained support for creating "monitoring"
2848 connections that can eavesdrop into all bus communication
2849 for debugging purposes.
2851 * Timestamps may now be specified in seconds since the UNIX
2852 epoch Jan 1st, 1970 by specifying "@" followed by the value
2855 * Native tcpwrap support in systemd has been removed. tcpwrap
2856 is old code, not really maintained anymore and has serious
2857 shortcomings, and better options such as firewalls
2858 exist. For setups that require tcpwrap usage, please
2859 consider invoking your socket-activated service via tcpd,
2860 like on traditional inetd.
2862 * A new system.conf configuration option
2863 DefaultTimerAccuracySec= has been added that controls the
2864 default AccuracySec= setting of .timer units.
2866 * Timer units gained a new WakeSystem= switch. If enabled,
2867 timers configured this way will cause the system to resume
2868 from system suspend (if the system supports that, which most
2871 * Timer units gained a new Persistent= switch. If enabled,
2872 timers configured this way will save to disk when they have
2873 been last triggered. This information is then used on next
2874 reboot to possible execute overdue timer events, that
2875 could not take place because the system was powered off.
2876 This enables simple anacron-like behaviour for timer units.
2878 * systemctl's "list-timers" will now also list the time a
2879 timer unit was last triggered in addition to the next time
2880 it will be triggered.
2882 * systemd-networkd will now assign predictable IPv4LL
2883 addresses to its local interfaces.
2885 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2886 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2887 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2888 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2889 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2890 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2891 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2892 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2895 -- Berlin, 2014-03-25
2899 * A new unit file setting RestrictAddressFamilies= has been
2900 added to restrict which socket address families unit
2901 processes gain access to. This takes address family names
2902 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2903 attack surface of services via exotic protocol stacks. This
2904 is built on seccomp system call filters.
2906 * Two new unit file settings RuntimeDirectory= and
2907 RuntimeDirectoryMode= have been added that may be used to
2908 manage a per-daemon runtime directories below /run. This is
2909 an alternative for setting up directory permissions with
2910 tmpfiles snippets, and has the advantage that the runtime
2911 directory's lifetime is bound to the daemon runtime and that
2912 the daemon starts up with an empty directory each time. This
2913 is particularly useful when writing services that drop
2914 privileges using the User= or Group= setting.
2916 * The DeviceAllow= unit setting now supports globbing for
2917 matching against device group names.
2919 * The systemd configuration file system.conf gained new
2920 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2921 DefaultMemoryAccounting= to globally turn on/off accounting
2922 for specific resources (cgroups) for all units. These
2923 settings may still be overridden individually in each unit
2926 * systemd-gpt-auto-generator is now able to discover /srv and
2927 root partitions in addition to /home and swap partitions. It
2928 also supports LUKS-encrypted partitions now. With this in
2929 place, automatic discovery of partitions to mount following
2930 the Discoverable Partitions Specification
2931 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2932 is now a lot more complete. This allows booting without
2933 /etc/fstab and without root= on the kernel command line on
2934 systems prepared appropriately.
2936 * systemd-nspawn gained a new --image= switch which allows
2937 booting up disk images and Linux installations on any block
2938 device that follow the Discoverable Partitions Specification
2939 (see above). This means that installations made with
2940 appropriately updated installers may now be started and
2941 deployed using container managers, completely
2942 unmodified. (We hope that libvirt-lxc will add support for
2943 this feature soon, too.)
2945 * systemd-nspawn gained a new --network-macvlan= setting to
2946 set up a private macvlan interface for the
2947 container. Similarly, systemd-networkd gained a new
2948 Kind=macvlan setting in .netdev files.
2950 * systemd-networkd now supports configuring local addresses
2953 * A new tool systemd-network-wait-online has been added to
2954 synchronously wait for network connectivity using
2957 * The sd-bus.h bus API gained a new sd_bus_track object for
2958 tracking the life-cycle of bus peers. Note that sd-bus.h is
2959 still not a public API though (unless you specify
2960 --enable-kdbus on the configure command line, which however
2961 voids your warranty and you get no API stability guarantee).
2963 * The $XDG_RUNTIME_DIR runtime directories for each user are
2964 now individual tmpfs instances, which has the benefit of
2965 introducing separate pools for each user, with individual
2966 size limits, and thus making sure that unprivileged clients
2967 can no longer negatively impact the system or other users by
2968 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2969 RuntimeDirectorySize= has been introduced that allows
2970 controlling the default size limit for all users. It
2971 defaults to 10% of the available physical memory. This is no
2972 replacement for quotas on tmpfs though (which the kernel
2973 still does not support), as /dev/shm and /tmp are still
2974 shared resources used by both the system and unprivileged
2977 * logind will now automatically turn off automatic suspending
2978 on laptop lid close when more than one display is
2979 connected. This was previously expected to be implemented
2980 individually in desktop environments (such as GNOME),
2981 however has been added to logind now, in order to fix a
2982 boot-time race where a desktop environment might not have
2983 been started yet and thus not been able to take an inhibitor
2984 lock at the time where logind already suspends the system
2985 due to a closed lid.
2987 * logind will now wait at least 30s after each system
2988 suspend/resume cycle, and 3min after system boot before
2989 suspending the system due to a closed laptop lid. This
2990 should give USB docking stations and similar enough time to
2991 be probed and configured after system resume and boot in
2992 order to then act as suspend blocker.
2994 * systemd-run gained a new --property= setting which allows
2995 initialization of resource control properties (and others)
2996 for the created scope or service unit. Example: "systemd-run
2997 --property=BlockIOWeight=10 updatedb" may be used to run
2998 updatedb at a low block IO scheduling weight.
3000 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3001 now also work in --scope mode.
3003 * When systemd is compiled with kdbus support, basic support
3004 for enforced policies is now in place. (Note that enabling
3005 kdbus still voids your warranty and no API compatibility
3008 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3009 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3010 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3011 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3012 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3013 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3014 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3015 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3016 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3017 Zbigniew Jędrzejewski-Szmek
3019 -- Berlin, 2014-03-12
3023 * systemd will now relabel /dev after loading the SMACK policy
3024 according to SMACK rules.
3026 * A new unit file option AppArmorProfile= has been added to
3027 set the AppArmor profile for the processes of a unit.
3029 * A new condition check ConditionArchitecture= has been added
3030 to conditionalize units based on the system architecture, as
3031 reported by uname()'s "machine" field.
3033 * systemd-networkd now supports matching on the system
3034 virtualization, architecture, kernel command line, host name
3037 * logind is now a lot more aggressive when suspending the
3038 machine due to a closed laptop lid. Instead of acting only
3039 on the lid close action, it will continuously watch the lid
3040 status and act on it. This is useful for laptops where the
3041 power button is on the outside of the chassis so that it can
3042 be reached without opening the lid (such as the Lenovo
3043 Yoga). On those machines, logind will now immediately
3044 re-suspend the machine if the power button has been
3045 accidentally pressed while the laptop was suspended and in a
3046 backpack or similar.
3048 * logind will now watch SW_DOCK switches and inhibit reaction
3049 to the lid switch if it is pressed. This means that logind
3050 will not suspend the machine anymore if the lid is closed
3051 and the system is docked, if the laptop supports SW_DOCK
3052 notifications via the input layer. Note that ACPI docking
3053 stations do not generate this currently. Also note that this
3054 logic is usually not fully sufficient and Desktop
3055 Environments should take a lid switch inhibitor lock when an
3056 external display is connected, as systemd will not watch
3059 * nspawn will now make use of the devices cgroup controller by
3060 default, and only permit creation of and access to the usual
3061 API device nodes like /dev/null or /dev/random, as well as
3062 access to (but not creation of) the pty devices.
3064 * We will now ship a default .network file for
3065 systemd-networkd that automatically configures DHCP for
3066 network interfaces created by nspawn's --network-veth or
3067 --network-bridge= switches.
3069 * systemd will now understand the usual M, K, G, T suffixes
3070 according to SI conventions (i.e. to the base 1000) when
3071 referring to throughput and hardware metrics. It will stay
3072 with IEC conventions (i.e. to the base 1024) for software
3073 metrics, according to what is customary according to
3074 Wikipedia. We explicitly document which base applies for
3075 each configuration option.
3077 * The DeviceAllow= setting in unit files now supports a syntax
3078 to whitelist an entire group of devices node majors at once,
3079 based on the /proc/devices listing. For example, with the
3080 string "char-pts", it is now possible to whitelist all
3081 current and future pseudo-TTYs at once.
3083 * sd-event learned a new "post" event source. Event sources of
3084 this type are triggered by the dispatching of any event
3085 source of a type that is not "post". This is useful for
3086 implementing clean-up and check event sources that are
3087 triggered by other work being done in the program.
3089 * systemd-networkd is no longer statically enabled, but uses
3090 the usual [Install] sections so that it can be
3091 enabled/disabled using systemctl. It still is enabled by
3094 * When creating a veth interface pair with systemd-nspawn, the
3095 host side will now be prefixed with "vb-" if
3096 --network-bridge= is used, and with "ve-" if --network-veth
3097 is used. This way, it is easy to distinguish these cases on
3098 the host, for example to apply different configuration to
3099 them with systemd-networkd.
3101 * The compatibility libraries for libsystemd-journal.so,
3102 libsystem-id128.so, libsystemd-login.so and
3103 libsystemd-daemon.so do not make use of IFUNC
3104 anymore. Instead, we now build libsystemd.so multiple times
3105 under these alternative names. This means that the footprint
3106 is drastically increased, but given that these are
3107 transitional compatibility libraries, this should not matter
3108 much. This change has been made necessary to support the ARM
3109 platform for these compatibility libraries, as the ARM
3110 toolchain is not really at the same level as the toolchain
3111 for other architectures like x86 and does not support
3112 IFUNC. Please make sure to use --enable-compat-libs only
3113 during a transitional period!
3115 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3116 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3117 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3118 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3119 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3120 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3121 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3122 Zbigniew Jędrzejewski-Szmek
3124 -- Berlin, 2014-02-24
3128 * A new component "systemd-networkd" has been added that can
3129 be used to configure local network interfaces statically or
3130 via DHCP. It is capable of bringing up bridges, VLANs, and
3131 bonding. Currently, no hook-ups for interactive network
3132 configuration are provided. Use this for your initrd,
3133 container, embedded, or server setup if you need a simple,
3134 yet powerful, network configuration solution. This
3135 configuration subsystem is quite nifty, as it allows wildcard
3136 hotplug matching in interfaces. For example, with a single
3137 configuration snippet, you can configure that all Ethernet
3138 interfaces showing up are automatically added to a bridge,
3139 or similar. It supports link-sensing and more.
3141 * A new tool "systemd-socket-proxyd" has been added which can
3142 act as a bidirectional proxy for TCP sockets. This is
3143 useful for adding socket activation support to services that
3144 do not actually support socket activation, including virtual
3145 machines and the like.
3147 * Add a new tool to save/restore rfkill state on
3150 * Save/restore state of keyboard backlights in addition to
3151 display backlights on shutdown/boot.
3153 * udev learned a new SECLABEL{} construct to label device
3154 nodes with a specific security label when they appear. For
3155 now, only SECLABEL{selinux} is supported, but the syntax is
3156 prepared for additional security frameworks.
3158 * udev gained a new scheme to configure link-level attributes
3159 from files in /etc/systemd/network/*.link. These files can
3160 match against MAC address, device path, driver name and type,
3161 and will apply attributes like the naming policy, link speed,
3162 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3163 address assignment policy (randomized, ...).
3165 * The configuration of network interface naming rules for
3166 "permanent interface names" has changed: a new NamePolicy=
3167 setting in the [Link] section of .link files determines the
3168 priority of possible naming schemes (onboard, slot, MAC,
3169 path). The default value of this setting is determined by
3170 /usr/lib/net/links/99-default.link. Old
3171 80-net-name-slot.rules udev configuration file has been
3172 removed, so local configuration overriding this file should
3173 be adapated to override 99-default.link instead.
3175 * When the User= switch is used in a unit file, also
3176 initialize $SHELL= based on the user database entry.
3178 * systemd no longer depends on libdbus. All communication is
3179 now done with sd-bus, systemd's low-level bus library
3182 * kdbus support has been added to PID 1 itself. When kdbus is
3183 enabled, this causes PID 1 to set up the system bus and
3184 enable support for a new ".busname" unit type that
3185 encapsulates bus name activation on kdbus. It works a little
3186 bit like ".socket" units, except for bus names. A new
3187 generator has been added that converts classic dbus1 service
3188 activation files automatically into native systemd .busname
3191 * sd-bus: add a light-weight vtable implementation that allows
3192 defining objects on the bus with a simple static const
3193 vtable array of its methods, signals and properties.
3195 * systemd will not generate or install static dbus
3196 introspection data anymore to /usr/share/dbus-1/interfaces,
3197 as the precise format of these files is unclear, and
3198 nothing makes use of it.
3200 * A proxy daemon is now provided to proxy clients connecting
3201 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3202 compatibility with classic D-Bus.
3204 * A bus driver implementation has been added that supports the
3205 classic D-Bus bus driver calls on kdbus, also for
3206 compatibility purposes.
3208 * A new API "sd-event.h" has been added that implements a
3209 minimal event loop API built around epoll. It provides a
3210 couple of features that direct epoll usage is lacking:
3211 prioritization of events, scales to large numbers of timer
3212 events, per-event timer slack (accuracy), system-wide
3213 coalescing of timer events, exit handlers, watchdog
3214 supervision support using systemd's sd_notify() API, child
3217 * A new API "sd-rntl.h" has been added that provides an API
3218 around the route netlink interface of the kernel, similar in
3219 style to "sd-bus.h".
3221 * A new API "sd-dhcp-client.h" has been added that provides a
3222 small DHCPv4 client-side implementation. This is used by
3225 * There is a new kernel command line option
3226 "systemd.restore_state=0|1". When set to "0", none of the
3227 systemd tools will restore saved runtime state to hardware
3228 devices. More specifically, the rfkill and backlight states
3231 * The FsckPassNo= compatibility option in mount/service units
3232 has been removed. The fstab generator will now add the
3233 necessary dependencies automatically, and does not require
3234 PID1's support for that anymore.
3236 * journalctl gained a new switch, --list-boots, that lists
3237 recent boots with their times and boot IDs.
3239 * The various tools like systemctl, loginctl, timedatectl,
3240 busctl, systemd-run, ... have gained a new switch "-M" to
3241 connect to a specific, local OS container (as direct
3242 connection, without requiring SSH). This works on any
3243 container that is registered with machined, such as those
3244 created by libvirt-lxc or nspawn.
3246 * systemd-run and systemd-analyze also gained support for "-H"
3247 to connect to remote hosts via SSH. This is particularly
3248 useful for systemd-run because it enables queuing of jobs
3249 onto remote systems.
3251 * machinectl gained a new command "login" to open a getty
3252 login in any local container. This works with any container
3253 that is registered with machined (such as those created by
3254 libvirt-lxc or nspawn), and which runs systemd inside.
3256 * machinectl gained a new "reboot" command that may be used to
3257 trigger a reboot on a specific container that is registered
3258 with machined. This works on any container that runs an init
3259 system of some kind.
3261 * systemctl gained a new "list-timers" command to print a nice
3262 listing of installed timer units with the times they elapse
3265 * Alternative reboot() parameters may now be specified on the
3266 "systemctl reboot" command line and are passed to the
3267 reboot() system call.
3269 * systemctl gained a new --job-mode= switch to configure the
3270 mode to queue a job with. This is a more generic version of
3271 --fail, --irreversible, and --ignore-dependencies, which are
3272 still available but not advertised anymore.
3274 * /etc/systemd/system.conf gained new settings to configure
3275 various default timeouts of units, as well as the default
3276 start limit interval and burst. These may still be overridden
3279 * PID1 will now export on the bus profile data of the security
3280 policy upload process (such as the SELinux policy upload to
3283 * journald: when forwarding logs to the console, include
3284 timestamps (following the setting in
3285 /sys/module/printk/parameters/time).
3287 * OnCalendar= in timer units now understands the special
3288 strings "yearly" and "annually". (Both are equivalent)
3290 * The accuracy of timer units is now configurable with the new
3291 AccuracySec= setting. It defaults to 1min.
3293 * A new dependency type JoinsNamespaceOf= has been added that
3294 allows running two services within the same /tmp and network
3295 namespace, if PrivateNetwork= or PrivateTmp= are used.
3297 * A new command "cat" has been added to systemctl. It outputs
3298 the original unit file of a unit, and concatenates the
3299 contents of additional "drop-in" unit file snippets, so that
3300 the full configuration is shown.
3302 * systemctl now supports globbing on the various "list-xyz"
3303 commands, like "list-units" or "list-sockets", as well as on
3304 those commands which take multiple unit names.
3306 * journalctl's --unit= switch gained support for globbing.
3308 * All systemd daemons now make use of the watchdog logic so
3309 that systemd automatically notices when they hang.
3311 * If the $container_ttys environment variable is set,
3312 getty-generator will automatically spawn a getty for each
3313 listed tty. This is useful for container managers to request
3314 login gettys to be spawned on as many ttys as needed.
3316 * %h, %s, %U specifier support is not available anymore when
3317 used in unit files for PID 1. This is because NSS calls are
3318 not safe from PID 1. They stay available for --user
3319 instances of systemd, and as special case for the root user.
3321 * loginctl gained a new "--no-legend" switch to turn off output
3324 * The "sd-login.h" API gained three new calls:
3325 sd_session_is_remote(), sd_session_get_remote_user(),
3326 sd_session_get_remote_host() to query information about
3329 * The udev hardware database now also carries vendor/product
3330 information of SDIO devices.
3332 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3333 determine whether watchdog notifications are requested by
3336 * Socket-activated per-connection services now include a
3337 short description of the connection parameters in the
3340 * tmpfiles gained a new "--boot" option. When this is not used,
3341 only lines where the command character is not suffixed with
3342 "!" are executed. When this option is specified, those
3343 options are executed too. This partitions tmpfiles
3344 directives into those that can be safely executed at any
3345 time, and those which should be run only at boot (for
3346 example, a line that creates /run/nologin).
3348 * A new API "sd-resolve.h" has been added which provides a simple
3349 asynchronous wrapper around glibc NSS host name resolution
3350 calls, such as getaddrinfo(). In contrast to glibc's
3351 getaddrinfo_a(), it does not use signals. In contrast to most
3352 other asynchronous name resolution libraries, this one does
3353 not reimplement DNS, but reuses NSS, so that alternate
3354 host name resolution systems continue to work, such as mDNS,
3355 LDAP, etc. This API is based on libasyncns, but it has been
3356 cleaned up for inclusion in systemd.
3358 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3359 "sd-daemon.h" are no longer found in individual libraries
3360 libsystemd-journal.so, libsystemd-login.so,
3361 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3362 merged them into a single library, libsystemd.so, which
3363 provides all symbols. The reason for this is cyclic
3364 dependencies, as these libraries tend to use each other's
3365 symbols. So far, we have managed to workaround that by linking
3366 a copy of a good part of our code into each of these
3367 libraries again and again, which, however, makes certain
3368 things hard to do, like sharing static variables. Also, it
3369 substantially increases footprint. With this change, there
3370 is only one library for the basic APIs systemd
3371 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3372 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3373 library as well, however are subject to the --enable-kdbus
3374 switch (see below). Note that "sd-dhcp-client.h" is not part
3375 of this library (this is because it only consumes, never
3376 provides, services of/to other APIs). To make the transition
3377 easy from the separate libraries to the unified one, we
3378 provide the --enable-compat-libs compile-time switch which
3379 will generate stub libraries that are compatible with the
3380 old ones but redirect all calls to the new one.
3382 * All of the kdbus logic and the new APIs "sd-bus.h",
3383 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3384 and "sd-utf8.h" are compile-time optional via the
3385 "--enable-kdbus" switch, and they are not compiled in by
3386 default. To make use of kdbus, you have to explicitly enable
3387 the switch. Note however, that neither the kernel nor the
3388 userspace API for all of this is considered stable yet. We
3389 want to maintain the freedom to still change the APIs for
3390 now. By specifying this build-time switch, you acknowledge
3391 that you are aware of the instability of the current
3394 * Also, note that while kdbus is pretty much complete,
3395 it lacks one thing: proper policy support. This means you
3396 can build a fully working system with all features; however,
3397 it will be highly insecure. Policy support will be added in
3398 one of the next releases, at the same time that we will
3399 declare the APIs stable.
3401 * When the kernel command line argument "kdbus" is specified,
3402 systemd will automatically load the kdbus.ko kernel module. At
3403 this stage of development, it is only useful for testing kdbus
3404 and should not be used in production. Note: if "--enable-kdbus"
3405 is specified, and the kdbus.ko kernel module is available, and
3406 "kdbus" is added to the kernel command line, the entire system
3407 runs with kdbus instead of dbus-daemon, with the above mentioned
3408 problem of missing the system policy enforcement. Also a future
3409 version of kdbus.ko or a newer systemd will not be compatible with
3410 each other, and will unlikely be able to boot the machine if only
3411 one of them is updated.
3413 * systemctl gained a new "import-environment" command which
3414 uploads the caller's environment (or parts thereof) into the
3415 service manager so that it is inherited by services started
3416 by the manager. This is useful to upload variables like
3417 $DISPLAY into the user service manager.
3419 * A new PrivateDevices= switch has been added to service units
3420 which allows running a service with a namespaced /dev
3421 directory that does not contain any device nodes for
3422 physical devices. More specifically, it only includes devices
3423 such as /dev/null, /dev/urandom, and /dev/zero which are API
3426 * logind has been extended to support behaviour like VT
3427 switching on seats that do not support a VT. This makes
3428 multi-session available on seats that are not the first seat
3429 (seat0), and on systems where kernel support for VTs has
3430 been disabled at compile-time.
3432 * If a process holds a delay lock for system sleep or shutdown
3433 and fails to release it in time, we will now log its
3434 identity. This makes it easier to identify processes that
3435 cause slow suspends or power-offs.
3437 * When parsing /etc/crypttab, support for a new key-slot=
3438 option as supported by Debian is added. It allows indicating
3439 which LUKS slot to use on disk, speeding up key loading.
3441 * The sd_journald_sendv() API call has been checked and
3442 officially declared to be async-signal-safe so that it may
3443 be invoked from signal handlers for logging purposes.
3445 * Boot-time status output is now enabled automatically after a
3446 short timeout if boot does not progress, in order to give
3447 the user an indication what she or he is waiting for.
3449 * The boot-time output has been improved to show how much time
3450 remains until jobs expire.
3452 * The KillMode= switch in service units gained a new possible
3453 value "mixed". If set, and the unit is shut down, then the
3454 initial SIGTERM signal is sent only to the main daemon
3455 process, while the following SIGKILL signal is sent to
3456 all remaining processes of the service.
3458 * When a scope unit is registered, a new property "Controller"
3459 may be set. If set to a valid bus name, systemd will send a
3460 RequestStop() signal to this name when it would like to shut
3461 down the scope. This may be used to hook manager logic into
3462 the shutdown logic of scope units. Also, scope units may now
3463 be put in a special "abandoned" state, in which case the
3464 manager process which created them takes no further
3465 responsibilities for it.
3467 * When reading unit files, systemd will now verify
3468 the access mode of these files, and warn about certain
3469 suspicious combinations. This has been added to make it
3470 easier to track down packaging bugs where unit files are
3471 marked executable or world-writable.
3473 * systemd-nspawn gained a new "--setenv=" switch to set
3474 container-wide environment variables. The similar option in
3475 systemd-activate was renamed from "--environment=" to
3476 "--setenv=" for consistency.
3478 * systemd-nspawn has been updated to create a new kdbus domain
3479 for each container that is invoked, thus allowing each
3480 container to have its own set of system and user buses,
3481 independent of the host.
3483 * systemd-nspawn gained a new --drop-capability= switch to run
3484 the container with less capabilities than the default. Both
3485 --drop-capability= and --capability= now take the special
3486 string "all" for dropping or keeping all capabilities.
3488 * systemd-nspawn gained new switches for executing containers
3489 with specific SELinux labels set.
3491 * systemd-nspawn gained a new --quiet switch to not generate
3492 any additional output but the container's own console
3495 * systemd-nspawn gained a new --share-system switch to run a
3496 container without PID namespacing enabled.
3498 * systemd-nspawn gained a new --register= switch to control
3499 whether the container is registered with systemd-machined or
3500 not. This is useful for containers that do not run full
3501 OS images, but only specific apps.
3503 * systemd-nspawn gained a new --keep-unit which may be used
3504 when invoked as the only program from a service unit, and
3505 results in registration of the unit service itself in
3506 systemd-machined, instead of a newly opened scope unit.
3508 * systemd-nspawn gained a new --network-interface= switch for
3509 moving arbitrary interfaces to the container. The new
3510 --network-veth switch creates a virtual Ethernet connection
3511 between host and container. The new --network-bridge=
3512 switch then allows assigning the host side of this virtual
3513 Ethernet connection to a bridge device.
3515 * systemd-nspawn gained a new --personality= switch for
3516 setting the kernel personality for the container. This is
3517 useful when running a 32-bit container on a 64-bit host. A
3518 similar option Personality= is now also available for service
3521 * logind will now also track a "Desktop" identifier for each
3522 session which encodes the desktop environment of it. This is
3523 useful for desktop environments that want to identify
3524 multiple running sessions of itself easily.
3526 * A new SELinuxContext= setting for service units has been
3527 added that allows setting a specific SELinux execution
3528 context for a service.
3530 * Most systemd client tools will now honour $SYSTEMD_LESS for
3531 settings of the "less" pager. By default, these tools will
3532 override $LESS to allow certain operations to work, such as
3533 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3534 influence this logic.
3536 * systemd's "seccomp" hook-up has been changed to make use of
3537 the libseccomp library instead of using its own
3538 implementation. This has benefits for portability among
3541 * For usage together with SystemCallFilter=, a new
3542 SystemCallErrorNumber= setting has been introduced that
3543 allows configuration of a system error number to be returned
3544 on filtered system calls, instead of immediately killing the
3545 process. Also, SystemCallArchitectures= has been added to
3546 limit access to system calls of a particular architecture
3547 (in order to turn off support for unused secondary
3548 architectures). There is also a global
3549 SystemCallArchitectures= setting in system.conf now to turn
3550 off support for non-native system calls system-wide.
3552 * systemd requires a kernel with a working name_to_handle_at(),
3553 please see the kernel config requirements in the README file.
3555 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3556 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3557 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3558 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3559 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3560 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3561 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3562 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3563 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3564 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3565 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3566 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3567 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3568 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3569 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3570 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3571 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3572 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3573 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3574 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3575 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3576 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3577 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3578 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3580 -- Berlin, 2014-02-20
3584 * logind has gained support for facilitating privileged input
3585 and drm device access for unprivileged clients. This work is
3586 useful to allow Wayland display servers (and similar
3587 programs, such as kmscon) to run under the user's ID and
3588 access input and drm devices which are normally
3589 protected. When this is used (and the kernel is new enough)
3590 logind will "mute" IO on the file descriptors passed to
3591 Wayland as long as it is in the background and "unmute" it
3592 if it returns into the foreground. This allows secure
3593 session switching without allowing background sessions to
3594 eavesdrop on input and display data. This also introduces
3595 session switching support if VT support is turned off in the
3596 kernel, and on seats that are not seat0.
3598 * A new kernel command line option luks.options= is understood
3599 now which allows specifying LUKS options for usage for LUKS
3600 encrypted partitions specified with luks.uuid=.
3602 * tmpfiles.d(5) snippets may now use specifier expansion in
3603 path names. More specifically %m, %b, %H, %v, are now
3604 replaced by the local machine id, boot id, hostname, and
3605 kernel version number.
3607 * A new tmpfiles.d(5) command "m" has been introduced which
3608 may be used to change the owner/group/access mode of a file
3609 or directory if it exists, but do nothing if it does not.
3611 * This release removes high-level support for the
3612 MemorySoftLimit= cgroup setting. The underlying kernel
3613 cgroup attribute memory.soft_limit= is currently badly
3614 designed and likely to be removed from the kernel API in its
3615 current form, hence we should not expose it for now.
3617 * The memory.use_hierarchy cgroup attribute is now enabled for
3618 all cgroups systemd creates in the memory cgroup
3619 hierarchy. This option is likely to be come the built-in
3620 default in the kernel anyway, and the non-hierarchical mode
3621 never made much sense in the intrinsically hierarchical
3624 * A new field _SYSTEMD_SLICE= is logged along with all journal
3625 messages containing the slice a message was generated
3626 from. This is useful to allow easy per-customer filtering of
3627 logs among other things.
3629 * systemd-journald will no longer adjust the group of journal
3630 files it creates to the "systemd-journal" group. Instead we
3631 rely on the journal directory to be owned by the
3632 "systemd-journal" group, and its setgid bit set, so that the
3633 kernel file system layer will automatically enforce that
3634 journal files inherit this group assignment. The reason for
3635 this change is that we cannot allow NSS look-ups from
3636 journald which would be necessary to resolve
3637 "systemd-journal" to a numeric GID, because this might
3638 create deadlocks if NSS involves synchronous queries to
3639 other daemons (such as nscd, or sssd) which in turn are
3640 logging clients of journald and might block on it, which
3641 would then dead lock. A tmpfiles.d(5) snippet included in
3642 systemd will make sure the setgid bit and group are
3643 properly set on the journal directory if it exists on every
3644 boot. However, we recommend adjusting it manually after
3645 upgrades too (or from RPM scriptlets), so that the change is
3646 not delayed until next reboot.
3648 * Backlight and random seed files in /var/lib/ have moved into
3649 the /var/lib/systemd/ directory, in order to centralize all
3650 systemd generated files in one directory.
3652 * Boot time performance measurements (as displayed by
3653 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3654 performance information if that's available to determine how
3655 much time BIOS and boot loader initialization required. With
3656 a sufficiently new BIOS you hence no longer need to boot
3657 with Gummiboot to get access to such information.
3659 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3660 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3661 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3662 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3663 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3664 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3665 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3667 -- Berlin, 2013-10-02
3671 * The Restart= option for services now understands a new
3672 on-watchdog setting, which will restart the service
3673 automatically if the service stops sending out watchdog keep
3674 alive messages (as configured with WatchdogSec=).
3676 * The getty generator (which is responsible for bringing up a
3677 getty on configured serial consoles) will no longer only
3678 start a getty on the primary kernel console but on all
3679 others, too. This makes the order in which console= is
3680 specified on the kernel command line less important.
3682 * libsystemd-logind gained a new sd_session_get_vt() call to
3683 retrieve the VT number of a session.
3685 * If the option "tries=0" is set for an entry of /etc/crypttab
3686 its passphrase is queried indefinitely instead of any
3687 maximum number of tries.
3689 * If a service with a configure PID file terminates its PID
3690 file will now be removed automatically if it still exists
3691 afterwards. This should put an end to stale PID files.
3693 * systemd-run will now also take relative binary path names
3694 for execution and no longer insists on absolute paths.
3696 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3697 paths that are optionally prefixed with "-" to indicate that
3698 it should not be considered a failure if they do not exist.
3700 * journalctl -o (and similar commands) now understands a new
3701 output mode "short-precise", it is similar to "short" but
3702 shows timestamps with usec accuracy.
3704 * The option "discard" (as known from Debian) is now
3705 synonymous to "allow-discards" in /etc/crypttab. In fact,
3706 "discard" is preferred now (since it is easier to remember
3709 * Some licensing clean-ups were made, so that more code is now
3710 LGPL-2.1 licensed than before.
3712 * A minimal tool to save/restore the display backlight
3713 brightness across reboots has been added. It will store the
3714 backlight setting as late as possible at shutdown, and
3715 restore it as early as possible during reboot.
3717 * A logic to automatically discover and enable home and swap
3718 partitions on GPT disks has been added. With this in place
3719 /etc/fstab becomes optional for many setups as systemd can
3720 discover certain partitions located on the root disk
3721 automatically. Home partitions are recognized under their
3722 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3723 partitions are recognized under their GPT type ID
3724 0657fd6da4ab43c484e50933c84b4f4f.
3726 * systemd will no longer pass any environment from the kernel
3727 or initrd to system services. If you want to set an
3728 environment for all services, do so via the kernel command
3729 line systemd.setenv= assignment.
3731 * The systemd-sysctl tool no longer natively reads the file
3732 /etc/sysctl.conf. If desired, the file should be symlinked
3733 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3734 legacy support by a symlink rather than built-in code, it
3735 also makes the otherwise hidden order of application of the
3736 different files visible. (Note that this partly reverts to a
3737 pre-198 application order of sysctl knobs!)
3739 * The "systemctl set-log-level" and "systemctl dump" commands
3740 have been moved to systemd-analyze.
3742 * systemd-run learned the new --remain-after-exit switch,
3743 which causes the scope unit not to be cleaned up
3744 automatically after the process terminated.
3746 * tmpfiles learned a new --exclude-prefix= switch to exclude
3747 certain paths from operation.
3749 * journald will now automatically flush all messages to disk
3750 as soon as a message at the log level CRIT, ALERT or EMERG
3753 Contributions from: Andrew Cook, Brandon Philips, Christian
3754 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3755 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3756 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3757 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3758 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3759 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3760 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3761 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3762 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3763 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3764 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3765 William Giokas, Zbigniew Jędrzejewski-Szmek
3767 -- Berlin, 2013-09-13
3771 * The documentation has been updated to cover the various new
3772 concepts introduced with 205.
3774 * Unit files now understand the new %v specifier which
3775 resolves to the kernel version string as returned by "uname
3778 * systemctl now supports filtering the unit list output by
3779 load state, active state and sub state, using the new
3782 * "systemctl status" will now show the results of the
3783 condition checks (like ConditionPathExists= and similar) of
3784 the last start attempts of the unit. They are also logged to
3787 * "journalctl -b" may now be used to look for boot output of a
3788 specific boot. Try "journalctl -b -1" for the previous boot,
3789 but the syntax is substantially more powerful.
3791 * "journalctl --show-cursor" has been added which prints the
3792 cursor string the last shown log line. This may then be used
3793 with the new "journalctl --after-cursor=" switch to continue
3794 browsing logs from that point on.
3796 * "journalctl --force" may now be used to force regeneration
3799 * Creation of "dead" device nodes has been moved from udev
3800 into kmod and tmpfiles. Previously, udev would read the kmod
3801 databases to pre-generate dead device nodes based on meta
3802 information contained in kernel modules, so that these would
3803 be auto-loaded on access rather then at boot. As this
3804 does not really have much to do with the exposing actual
3805 kernel devices to userspace this has always been slightly
3806 alien in the udev codebase. Following the new scheme kmod
3807 will now generate a runtime snippet for tmpfiles from the
3808 module meta information and it now is tmpfiles' job to the
3809 create the nodes. This also allows overriding access and
3810 other parameters for the nodes using the usual tmpfiles
3811 facilities. As side effect this allows us to remove the
3812 CAP_SYS_MKNOD capability bit from udevd entirely.
3814 * logind's device ACLs may now be applied to these "dead"
3815 devices nodes too, thus finally allowing managed access to
3816 devices such as /dev/snd/sequencer whithout loading the
3817 backing module right-away.
3819 * A new RPM macro has been added that may be used to apply
3820 tmpfiles configuration during package installation.
3822 * systemd-detect-virt and ConditionVirtualization= now can
3823 detect User-Mode-Linux machines (UML).
3825 * journald will now implicitly log the effective capabilities
3826 set of processes in the message metadata.
3828 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3830 * The initrd interface has been simplified (more specifically,
3831 support for passing performance data via environment
3832 variables and fsck results via files in /run has been
3833 removed). These features were non-essential, and are
3834 nowadays available in a much nicer way by having systemd in
3835 the initrd serialize its state and have the hosts systemd
3836 deserialize it again.
3838 * The udev "keymap" data files and tools to apply keyboard
3839 specific mappings of scan to key codes, and force-release
3840 scan code lists have been entirely replaced by a udev
3841 "keyboard" builtin and a hwdb data file.
3843 * systemd will now honour the kernel's "quiet" command line
3844 argument also during late shutdown, resulting in a
3845 completely silent shutdown when used.
3847 * There's now an option to control the SO_REUSEPORT socket
3848 option in .socket units.
3850 * Instance units will now automatically get a per-template
3851 subslice of system.slice unless something else is explicitly
3852 configured. For example, instances of sshd@.service will now
3853 implicitly be placed in system-sshd.slice rather than
3854 system.slice as before.
3856 * Test coverage support may now be enabled at build time.
3858 Contributions from: Dave Reisner, Frederic Crozat, Harald
3859 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3860 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3861 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3862 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3863 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3864 Giokas, Zbigniew Jędrzejewski-Szmek
3866 -- Berlin, 2013-07-23
3870 * Two new unit types have been introduced:
3872 Scope units are very similar to service units, however, are
3873 created out of pre-existing processes -- instead of PID 1
3874 forking off the processes. By using scope units it is
3875 possible for system services and applications to group their
3876 own child processes (worker processes) in a powerful way
3877 which then maybe used to organize them, or kill them
3878 together, or apply resource limits on them.
3880 Slice units may be used to partition system resources in an
3881 hierarchical fashion and then assign other units to them. By
3882 default there are now three slices: system.slice (for all
3883 system services), user.slice (for all user sessions),
3884 machine.slice (for VMs and containers).
3886 Slices and scopes have been introduced primarily in
3887 context of the work to move cgroup handling to a
3888 single-writer scheme, where only PID 1
3889 creates/removes/manages cgroups.
3891 * There's a new concept of "transient" units. In contrast to
3892 normal units these units are created via an API at runtime,
3893 not from configuration from disk. More specifically this
3894 means it is now possible to run arbitrary programs as
3895 independent services, with all execution parameters passed
3896 in via bus APIs rather than read from disk. Transient units
3897 make systemd substantially more dynamic then it ever was,
3898 and useful as a general batch manager.
3900 * logind has been updated to make use of scope and slice units
3901 for managing user sessions. As a user logs in he will get
3902 his own private slice unit, to which all sessions are added
3903 as scope units. We also added support for automatically
3904 adding an instance of user@.service for the user into the
3905 slice. Effectively logind will no longer create cgroup
3906 hierarchies on its own now, it will defer entirely to PID 1
3907 for this by means of scope, service and slice units. Since
3908 user sessions this way become entities managed by PID 1
3909 the output of "systemctl" is now a lot more comprehensive.
3911 * A new mini-daemon "systemd-machined" has been added which
3912 may be used by virtualization managers to register local
3913 VMs/containers. nspawn has been updated accordingly, and
3914 libvirt will be updated shortly. machined will collect a bit
3915 of meta information about the VMs/containers, and assign
3916 them their own scope unit (see above). The collected
3917 meta-data is then made available via the "machinectl" tool,
3918 and exposed in "ps" and similar tools. machined/machinectl
3919 is compile-time optional.
3921 * As discussed earlier, the low-level cgroup configuration
3922 options ControlGroup=, ControlGroupModify=,
3923 ControlGroupPersistent=, ControlGroupAttribute= have been
3924 removed. Please use high-level attribute settings instead as
3925 well as slice units.
3927 * A new bus call SetUnitProperties() has been added to alter
3928 various runtime parameters of a unit. This is primarily
3929 useful to alter cgroup parameters dynamically in a nice way,
3930 but will be extended later on to make more properties
3931 modifiable at runtime. systemctl gained a new set-properties
3932 command that wraps this call.
3934 * A new tool "systemd-run" has been added which can be used to
3935 run arbitrary command lines as transient services or scopes,
3936 while configuring a number of settings via the command
3937 line. This tool is currently very basic, however already
3938 very useful. We plan to extend this tool to even allow
3939 queuing of execution jobs with time triggers from the
3940 command line, similar in fashion to "at".
3942 * nspawn will now inform the user explicitly that kernels with
3943 audit enabled break containers, and suggest the user to turn
3946 * Support for detecting the IMA and AppArmor security
3947 frameworks with ConditionSecurity= has been added.
3949 * journalctl gained a new "-k" switch for showing only kernel
3950 messages, mimicking dmesg output; in addition to "--user"
3951 and "--system" switches for showing only user's own logs
3954 * systemd-delta can now show information about drop-in
3955 snippets extending unit files.
3957 * libsystemd-bus has been substantially updated but is still
3958 not available as public API.
3960 * systemd will now look for the "debug" argument on the kernel
3961 command line and enable debug logging, similar to what
3962 "systemd.log_level=debug" already did before.
3964 * "systemctl set-default", "systemctl get-default" has been
3965 added to configure the default.target symlink, which
3966 controls what to boot into by default.
3968 * "systemctl set-log-level" has been added as a convenient
3969 way to raise and lower systemd logging threshold.
3971 * "systemd-analyze plot" will now show the time the various
3972 generators needed for execution, as well as information
3973 about the unit file loading.
3975 * libsystemd-journal gained a new sd_journal_open_files() call
3976 for opening specific journal files. journactl also gained a
3977 new switch to expose this new functionality. Previously we
3978 only supported opening all files from a directory, or all
3979 files from the system, as opening individual files only is
3980 racy due to journal file rotation.
3982 * systemd gained the new DefaultEnvironment= setting in
3983 /etc/systemd/system.conf to set environment variables for
3986 * If a privileged process logs a journal message with the
3987 OBJECT_PID= field set, then journald will automatically
3988 augment this with additional OBJECT_UID=, OBJECT_GID=,
3989 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3990 system services want to log events about specific client
3991 processes. journactl/systemctl has been updated to make use
3992 of this information if all log messages regarding a specific
3995 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3996 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3997 Reisner, David Coppa, David King, David Strauss, Eelco
3998 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3999 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4000 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4001 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4002 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4003 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4004 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4005 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4006 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4007 Łukasz Stelmach, 장동준
4011 * The Python bindings gained some minimal support for the APIs
4012 exposed by libsystemd-logind.
4014 * ConditionSecurity= gained support for detecting SMACK. Since
4015 this condition already supports SELinux and AppArmor we only
4016 miss IMA for this. Patches welcome!
4018 Contributions from: Karol Lewandowski, Lennart Poettering,
4019 Zbigniew Jędrzejewski-Szmek
4023 * systemd-nspawn will now create /etc/resolv.conf if
4024 necessary, before bind-mounting the host's file onto it.
4026 * systemd-nspawn will now store meta information about a
4027 container on the container's cgroup as extended attribute
4028 fields, including the root directory.
4030 * The cgroup hierarchy has been reworked in many ways. All
4031 objects any of the components systemd creates in the cgroup
4032 tree are now suffixed. More specifically, user sessions are
4033 now placed in cgroups suffixed with ".session", users in
4034 cgroups suffixed with ".user", and nspawn containers in
4035 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4036 names are now escaped in a simple scheme to avoid collision
4037 of userspace object names with kernel filenames. This work
4038 is preparation for making these objects relocatable in the
4039 cgroup tree, in order to allow easy resource partitioning of
4040 these objects without causing naming conflicts.
4042 * systemctl list-dependencies gained the new switches
4043 --plain, --reverse, --after and --before.
4045 * systemd-inhibit now shows the process name of processes that
4046 have taken an inhibitor lock.
4048 * nss-myhostname will now also resolve "localhost"
4049 implicitly. This makes /etc/hosts an optional file and
4050 nicely handles that on IPv6 ::1 maps to both "localhost" and
4053 * libsystemd-logind.so gained a new call
4054 sd_get_machine_names() to enumerate running containers and
4055 VMs (currently only supported by very new libvirt and
4056 nspawn). sd_login_monitor can now be used to watch
4057 VMs/containers coming and going.
4059 * .include is not allowed recursively anymore, and only in
4060 unit files. Usually it is better to use drop-in snippets in
4061 .d/*.conf anyway, as introduced with systemd 198.
4063 * systemd-analyze gained a new "critical-chain" command that
4064 determines the slowest chain of units run during system
4065 boot-up. It is very useful for tracking down where
4066 optimizing boot time is the most beneficial.
4068 * systemd will no longer allow manipulating service paths in
4069 the name=systemd:/system cgroup tree using ControlGroup= in
4070 units. (But is still fine with it in all other dirs.)
4072 * There's a new systemd-nspawn@.service service file that may
4073 be used to easily run nspawn containers as system
4074 services. With the container's root directory in
4075 /var/lib/container/foobar it is now sufficient to run
4076 "systemctl start systemd-nspawn@foobar.service" to boot it.
4078 * systemd-cgls gained a new parameter "--machine" to list only
4079 the processes within a certain container.
4081 * ConditionSecurity= now can check for "apparmor". We still
4082 are lacking checks for SMACK and IMA for this condition
4083 check though. Patches welcome!
4085 * A new configuration file /etc/systemd/sleep.conf has been
4086 added that may be used to configure which kernel operation
4087 systemd is supposed to execute when "suspend", "hibernate"
4088 or "hybrid-sleep" is requested. This makes the new kernel
4089 "freeze" state accessible to the user.
4091 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4092 the passed argument if applicable.
4094 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4095 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4096 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4097 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4098 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4099 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4100 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4105 * The output of 'systemctl list-jobs' got some polishing. The
4106 '--type=' argument may now be passed more than once. A new
4107 command 'systemctl list-sockets' has been added which shows
4108 a list of kernel sockets systemd is listening on with the
4109 socket units they belong to, plus the units these socket
4112 * The experimental libsystemd-bus library got substantial
4113 updates to work in conjunction with the (also experimental)
4114 kdbus kernel project. It works well enough to exchange
4115 messages with some sophistication. Note that kdbus is not
4116 ready yet, and the library is mostly an elaborate test case
4117 for now, and not installable.
4119 * systemd gained a new unit 'systemd-static-nodes.service'
4120 that generates static device nodes earlier during boot, and
4121 can run in conjunction with udev.
4123 * libsystemd-login gained a new call sd_pid_get_user_unit()
4124 to retrieve the user systemd unit a process is running
4125 in. This is useful for systems where systemd is used as
4128 * systemd-nspawn now places all containers in the new /machine
4129 top-level cgroup directory in the name=systemd
4130 hierarchy. libvirt will soon do the same, so that we get a
4131 uniform separation of /system, /user and /machine for system
4132 services, user processes and containers/virtual
4133 machines. This new cgroup hierarchy is also useful to stick
4134 stable names to specific container instances, which can be
4135 recognized later this way (this name may be controlled
4136 via systemd-nspawn's new -M switch). libsystemd-login also
4137 gained a new call sd_pid_get_machine_name() to retrieve the
4138 name of the container/VM a specific process belongs to.
4140 * bootchart can now store its data in the journal.
4142 * libsystemd-journal gained a new call
4143 sd_journal_add_conjunction() for AND expressions to the
4144 matching logic. This can be used to express more complex
4145 logical expressions.
4147 * journactl can now take multiple --unit= and --user-unit=
4150 * The cryptsetup logic now understands the "luks.key=" kernel
4151 command line switch for specifying a file to read the
4152 decryption key from. Also, if a configured key file is not
4153 found the tool will now automatically fall back to prompting
4156 * Python systemd.journal module was updated to wrap recently
4157 added functions from libsystemd-journal. The interface was
4158 changed to bring the low level interface in s.j._Reader
4159 closer to the C API, and the high level interface in
4160 s.j.Reader was updated to wrap and convert all data about
4163 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4164 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4165 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4166 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4167 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4168 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4172 * journalctl --update-catalog now understands a new --root=
4173 option to operate on catalogs found in a different root
4176 * During shutdown after systemd has terminated all running
4177 services a final killing loop kills all remaining left-over
4178 processes. We will now print the name of these processes
4179 when we send SIGKILL to them, since this usually indicates a
4182 * If /etc/crypttab refers to password files stored on
4183 configured mount points automatic dependencies will now be
4184 generated to ensure the specific mount is established first
4185 before the key file is attempted to be read.
4187 * 'systemctl status' will now show information about the
4188 network sockets a socket unit is listening on.
4190 * 'systemctl status' will also shown information about any
4191 drop-in configuration file for units. (Drop-In configuration
4192 files in this context are files such as
4193 /etc/systemd/systemd/foobar.service.d/*.conf)
4195 * systemd-cgtop now optionally shows summed up CPU times of
4196 cgroups. Press '%' while running cgtop to switch between
4197 percentage and absolute mode. This is useful to determine
4198 which cgroups use up the most CPU time over the entire
4199 runtime of the system. systemd-cgtop has also been updated
4200 to be 'pipeable' for processing with further shell tools.
4202 * 'hostnamectl set-hostname' will now allow setting of FQDN
4205 * The formatting and parsing of time span values has been
4206 changed. The parser now understands fractional expressions
4207 such as "5.5h". The formatter will now output fractional
4208 expressions for all time spans under 1min, i.e. "5.123456s"
4209 rather than "5s 123ms 456us". For time spans under 1s
4210 millisecond values are shown, for those under 1ms
4211 microsecond values are shown. This should greatly improve
4212 all time-related output of systemd.
4214 * libsystemd-login and libsystemd-journal gained new
4215 functions for querying the poll() events mask and poll()
4216 timeout value for integration into arbitrary event
4219 * localectl gained the ability to list available X11 keymaps
4220 (models, layouts, variants, options).
4222 * 'systemd-analyze dot' gained the ability to filter for
4223 specific units via shell-style globs, to create smaller,
4224 more useful graphs. I.e. it is now possible to create simple
4225 graphs of all the dependencies between only target units, or
4226 of all units that Avahi has dependencies with.
4228 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4229 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4230 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4231 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4232 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4233 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4234 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4238 * The boot-time readahead implementation for rotating media
4239 will now read the read-ahead data in multiple passes which
4240 consist of all read requests made in equidistant time
4241 intervals. This means instead of strictly reading read-ahead
4242 data in its physical order on disk we now try to find a
4243 middle ground between physical and access time order.
4245 * /etc/os-release files gained a new BUILD_ID= field for usage
4246 on operating systems that provide continuous builds of OS
4249 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4250 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4251 William Douglas, Zbigniew Jędrzejewski-Szmek
4255 * systemd-python gained an API exposing libsystemd-daemon.
4257 * The SMACK setup logic gained support for uploading CIPSO
4260 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4261 ReadOnlyDirectories= and InaccessibleDirectories= has
4262 changed. The private /tmp and /var/tmp directories are now
4263 shared by all processes of a service (which means
4264 ExecStartPre= may now leave data in /tmp that ExecStart= of
4265 the same service can still access). When a service is
4266 stopped its temporary directories are immediately deleted
4267 (normal clean-up with tmpfiles is still done in addition to
4270 * By default, systemd will now set a couple of sysctl
4271 variables in the kernel: the safe sysrq options are turned
4272 on, IP route verification is turned on, and source routing
4273 disabled. The recently added hardlink and softlink
4274 protection of the kernel is turned on. These settings should
4275 be reasonably safe, and good defaults for all new systems.
4277 * The predictable network naming logic may now be turned off
4278 with a new kernel command line switch: net.ifnames=0.
4280 * A new libsystemd-bus module has been added that implements a
4281 pretty complete D-Bus client library. For details see:
4283 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4285 * journald will now explicitly flush the journal files to disk
4286 at the latest 5min after each write. The file will then also
4287 be marked offline until the next write. This should increase
4288 reliability in case of a crash. The synchronization delay
4289 can be configured via SyncIntervalSec= in journald.conf.
4291 * There's a new remote-fs-setup.target unit that can be used
4292 to pull in specific services when at least one remote file
4293 system is to be mounted.
4295 * There are new targets timers.target and paths.target as
4296 canonical targets to pull user timer and path units in
4297 from. This complements sockets.target with a similar
4298 purpose for socket units.
4300 * libudev gained a new call udev_device_set_attribute_value()
4301 to set sysfs attributes of a device.
4303 * The udev daemon now sets the default number of worker
4304 processes executed in parallel based on the number of available
4305 CPUs instead of the amount of available RAM. This is supposed
4306 to provide a more reliable default and limit a too aggressive
4307 paralellism for setups with 1000s of devices connected.
4309 Contributions from: Auke Kok, Colin Walters, Cristian
4310 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4311 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4312 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4313 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4314 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4315 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4316 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4317 Zbigniew Jędrzejewski-Szmek
4321 * Configuration of unit files may now be extended via drop-in
4322 files without having to edit/override the unit files
4323 themselves. More specifically, if the administrator wants to
4324 change one value for a service file foobar.service he can
4325 now do so by dropping in a configuration snippet into
4326 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4327 will load all these snippets and apply them on top of the
4328 main unit configuration file, possibly extending or
4329 overriding its settings. Using these drop-in snippets is
4330 generally nicer than the two earlier options for changing
4331 unit files locally: copying the files from
4332 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4333 them there; or creating a new file in /etc/systemd/system/
4334 that incorporates the original one via ".include". Drop-in
4335 snippets into these .d/ directories can be placed in any
4336 directory systemd looks for units in, and the usual
4337 overriding semantics between /usr/lib, /etc and /run apply
4340 * Most unit file settings which take lists of items can now be
4341 reset by assigning the empty string to them. For example,
4342 normally, settings such as Environment=FOO=BAR append a new
4343 environment variable assignment to the environment block,
4344 each time they are used. By assigning Environment= the empty
4345 string the environment block can be reset to empty. This is
4346 particularly useful with the .d/*.conf drop-in snippets
4347 mentioned above, since this adds the ability to reset list
4348 settings from vendor unit files via these drop-ins.
4350 * systemctl gained a new "list-dependencies" command for
4351 listing the dependencies of a unit recursively.
4353 * Inhibitors are now honored and listed by "systemctl
4354 suspend", "systemctl poweroff" (and similar) too, not only
4355 GNOME. These commands will also list active sessions by
4358 * Resource limits (as exposed by the various control group
4359 controllers) can now be controlled dynamically at runtime
4360 for all units. More specifically, you can now use a command
4361 like "systemctl set-cgroup-attr foobar.service cpu.shares
4362 2000" to alter the CPU shares a specific service gets. These
4363 settings are stored persistently on disk, and thus allow the
4364 administrator to easily adjust the resource usage of
4365 services with a few simple commands. This dynamic resource
4366 management logic is also available to other programs via the
4367 bus. Almost any kernel cgroup attribute and controller is
4370 * systemd-vconsole-setup will now copy all font settings to
4371 all allocated VTs, where it previously applied them only to
4374 * libsystemd-login gained the new sd_session_get_tty() API
4377 * This release drops support for a few legacy or
4378 distribution-specific LSB facility names when parsing init
4379 scripts: $x-display-manager, $mail-transfer-agent,
4380 $mail-transport-agent, $mail-transfer-agent, $smtp,
4381 $null. Also, the mail-transfer-agent.target unit backing
4382 this has been removed. Distributions which want to retain
4383 compatibility with this should carry the burden for
4384 supporting this themselves and patch support for these back
4385 in, if they really need to. Also, the facilities $syslog and
4386 $local_fs are now ignored, since systemd does not support
4387 early-boot LSB init scripts anymore, and these facilities
4388 are implied anyway for normal services. syslog.target has
4391 * There are new bus calls on PID1's Manager object for
4392 cancelling jobs, and removing snapshot units. Previously,
4393 both calls were only available on the Job and Snapshot
4396 * systemd-journal-gatewayd gained SSL support.
4398 * The various "environment" files, such as /etc/locale.conf
4399 now support continuation lines with a backslash ("\") as
4400 last character in the line, similarly in style (but different)
4401 to how this is supported in shells.
4403 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4404 now implicitly appended to every log entry logged. systemctl
4405 has been updated to filter by this field when operating on a
4406 user systemd instance.
4408 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4409 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4410 the container. This makes it easier to boot unmodified
4411 Fedora systems in a container, which however still requires
4412 audit=0 to be passed on the kernel command line. Auditing in
4413 kernel and userspace is unfortunately still too broken in
4414 context of containers, hence we recommend compiling it out
4415 of the kernel or using audit=0. Hopefully this will be fixed
4416 one day for good in the kernel.
4418 * nspawn gained the new --bind= and --bind-ro= parameters to
4419 bind mount specific directories from the host into the
4422 * nspawn will now mount its own devpts file system instance
4423 into the container, in order not to leak pty devices from
4424 the host into the container.
4426 * systemd will now read the firmware boot time performance
4427 information from the EFI variables, if the used boot loader
4428 supports this, and takes it into account for boot performance
4429 analysis via "systemd-analyze". This is currently supported
4430 only in conjunction with Gummiboot, but could be supported
4431 by other boot loaders too. For details see:
4433 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4435 * A new generator has been added that automatically mounts the
4436 EFI System Partition (ESP) to /boot, if that directory
4437 exists, is empty, and no other file system has been
4438 configured to be mounted there.
4440 * logind will now send out PrepareForSleep(false) out
4441 unconditionally, after coming back from suspend. This may be
4442 used by applications as asynchronous notification for
4443 system resume events.
4445 * "systemctl unlock-sessions" has been added, that allows
4446 unlocking the screens of all user sessions at once, similar
4447 to how "systemctl lock-sessions" already locked all users
4448 sessions. This is backed by a new D-Bus call UnlockSessions().
4450 * "loginctl seat-status" will now show the master device of a
4451 seat. (i.e. the device of a seat that needs to be around for
4452 the seat to be considered available, usually the graphics
4455 * tmpfiles gained a new "X" line type, that allows
4456 configuration of files and directories (with wildcards) that
4457 shall be excluded from automatic cleanup ("aging").
4459 * udev default rules set the device node permissions now only
4460 at "add" events, and do not change them any longer with a
4461 later "change" event.
4463 * The log messages for lid events and power/sleep keypresses
4464 now carry a message ID.
4466 * We now have a substantially larger unit test suite, but this
4467 continues to be work in progress.
4469 * udevadm hwdb gained a new --root= parameter to change the
4470 root directory to operate relative to.
4472 * logind will now issue a background sync() request to the kernel
4473 early at shutdown, so that dirty buffers are flushed to disk early
4474 instead of at the last moment, in order to optimize shutdown
4477 * A new bootctl tool has been added that is an interface for
4478 certain boot loader operations. This is currently a preview
4479 and is likely to be extended into a small mechanism daemon
4480 like timedated, localed, hostnamed, and can be used by
4481 graphical UIs to enumerate available boot options, and
4482 request boot into firmware operations.
4484 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4485 the rest of the package. It also has been updated to work
4486 correctly in initrds.
4488 * Policykit previously has been runtime optional, and is now
4489 also compile time optional via a configure switch.
4491 * systemd-analyze has been reimplemented in C. Also "systemctl
4492 dot" has moved into systemd-analyze.
4494 * "systemctl status" with no further parameters will now print
4495 the status of all active or failed units.
4497 * Operations such as "systemctl start" can now be executed
4498 with a new mode "--irreversible" which may be used to queue
4499 operations that cannot accidentally be reversed by a later
4500 job queuing. This is by default used to make shutdown
4501 requests more robust.
4503 * The Python API of systemd now gained a new module for
4504 reading journal files.
4506 * A new tool kernel-install has been added that can install
4507 kernel images according to the Boot Loader Specification:
4509 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4511 * Boot time console output has been improved to provide
4512 animated boot time output for hanging jobs.
4514 * A new tool systemd-activate has been added which can be used
4515 to test socket activation with, directly from the command
4516 line. This should make it much easier to test and debug
4517 socket activation in daemons.
4519 * journalctl gained a new "--reverse" (or -r) option to show
4520 journal output in reverse order (i.e. newest line first).
4522 * journalctl gained a new "--pager-end" (or -e) option to jump
4523 to immediately jump to the end of the journal in the
4524 pager. This is only supported in conjunction with "less".
4526 * journalctl gained a new "--user-unit=" option, that works
4527 similarly to "--unit=" but filters for user units rather than
4530 * A number of unit files to ease adoption of systemd in
4531 initrds has been added. This moves some minimal logic from
4532 the various initrd implementations into systemd proper.
4534 * The journal files are now owned by a new group
4535 "systemd-journal", which exists specifically to allow access
4536 to the journal, and nothing else. Previously, we used the
4537 "adm" group for that, which however possibly covers more
4538 than just journal/log file access. This new group is now
4539 already used by systemd-journal-gatewayd to ensure this
4540 daemon gets access to the journal files and as little else
4541 as possible. Note that "make install" will also set FS ACLs
4542 up for /var/log/journal to give "adm" and "wheel" read
4543 access to it, in addition to "systemd-journal" which owns
4544 the journal files. We recommend that packaging scripts also
4545 add read access to "adm" + "wheel" to /var/log/journal, and
4546 all existing/future journal files. To normal users and
4547 administrators little changes, however packagers need to
4548 ensure to create the "systemd-journal" system group at
4549 package installation time.
4551 * The systemd-journal-gatewayd now runs as unprivileged user
4552 systemd-journal-gateway:systemd-journal-gateway. Packaging
4553 scripts need to create these system user/group at
4556 * timedated now exposes a new boolean property CanNTP that
4557 indicates whether a local NTP service is available or not.
4559 * systemd-detect-virt will now also detect xen PVs
4561 * The pstore file system is now mounted by default, if it is
4564 * In addition to the SELinux and IMA policies we will now also
4565 load SMACK policies at early boot.
4567 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4568 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4569 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4570 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4571 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4572 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4573 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4574 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4575 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4576 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4577 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4578 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4579 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4580 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4584 * Timer units now support calendar time events in addition to
4585 monotonic time events. That means you can now trigger a unit
4586 based on a calendar time specification such as "Thu,Fri
4587 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4588 or fifth day of any month of the year 2013, given that it is
4589 a thursday or friday. This brings timer event support
4590 considerably closer to cron's capabilities. For details on
4591 the supported calendar time specification language see
4594 * udev now supports a number of different naming policies for
4595 network interfaces for predictable names, and a combination
4596 of these policies is now the default. Please see this wiki
4597 document for details:
4599 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4601 * Auke Kok's bootchart implementation has been added to the
4602 systemd tree. It is an optional component that can graph the
4603 boot in quite some detail. It is one of the best bootchart
4604 implementations around and minimal in its code and
4607 * nss-myhostname has been integrated into the systemd source
4608 tree. nss-myhostname guarantees that the local hostname
4609 always stays resolvable via NSS. It has been a weak
4610 requirement of systemd-hostnamed since a long time, and
4611 since its code is actually trivial we decided to just
4612 include it in systemd's source tree. It can be turned off
4613 with a configure switch.
4615 * The read-ahead logic is now capable of properly detecting
4616 whether a btrfs file system is on SSD or rotating media, in
4617 order to optimize the read-ahead scheme. Previously, it was
4618 only capable of detecting this on traditional file systems
4621 * In udev, additional device properties are now read from the
4622 IAB in addition to the OUI database. Also, Bluetooth company
4623 identities are attached to the devices as well.
4625 * In service files %U may be used as specifier that is
4626 replaced by the configured user name of the service.
4628 * nspawn may now be invoked without a controlling TTY. This
4629 makes it suitable for invocation as its own service. This
4630 may be used to set up a simple containerized server system
4631 using only core OS tools.
4633 * systemd and nspawn can now accept socket file descriptors
4634 when they are started for socket activation. This enables
4635 implementation of socket activated nspawn
4636 containers. i.e. think about autospawning an entire OS image
4637 when the first SSH or HTTP connection is received. We expect
4638 that similar functionality will also be added to libvirt-lxc
4641 * journalctl will now suppress ANSI color codes when
4642 presenting log data.
4644 * systemctl will no longer show control group information for
4645 a unit if a the control group is empty anyway.
4647 * logind can now automatically suspend/hibernate/shutdown the
4650 * /etc/machine-info and hostnamed now also expose the chassis
4651 type of the system. This can be used to determine whether
4652 the local system is a laptop, desktop, handset or
4653 tablet. This information may either be configured by the
4654 user/vendor or is automatically determined from ACPI and DMI
4655 information if possible.
4657 * A number of PolicyKit actions are now bound together with
4658 "imply" rules. This should simplify creating UIs because
4659 many actions will now authenticate similar ones as well.
4661 * Unit files learnt a new condition ConditionACPower= which
4662 may be used to conditionalize a unit depending on whether an
4663 AC power source is connected or not, of whether the system
4664 is running on battery power.
4666 * systemctl gained a new "is-failed" verb that may be used in
4667 shell scripts and suchlike to check whether a specific unit
4668 is in the "failed" state.
4670 * The EnvironmentFile= setting in unit files now supports file
4671 globbing, and can hence be used to easily read a number of
4672 environment files at once.
4674 * systemd will no longer detect and recognize specific
4675 distributions. All distribution-specific #ifdeffery has been
4676 removed, systemd is now fully generic and
4677 distribution-agnostic. Effectively, not too much is lost as
4678 a lot of the code is still accessible via explicit configure
4679 switches. However, support for some distribution specific
4680 legacy configuration file formats has been dropped. We
4681 recommend distributions to simply adopt the configuration
4682 files everybody else uses now and convert the old
4683 configuration from packaging scripts. Most distributions
4684 already did that. If that's not possible or desirable,
4685 distributions are welcome to forward port the specific
4686 pieces of code locally from the git history.
4688 * When logging a message about a unit systemd will now always
4689 log the unit name in the message meta data.
4691 * localectl will now also discover system locale data that is
4692 not stored in locale archives, but directly unpacked.
4694 * logind will no longer unconditionally use framebuffer
4695 devices as seat masters, i.e. as devices that are required
4696 to be existing before a seat is considered preset. Instead,
4697 it will now look for all devices that are tagged as
4698 "seat-master" in udev. By default, framebuffer devices will
4699 be marked as such, but depending on local systems, other
4700 devices might be marked as well. This may be used to
4701 integrate graphics cards using closed source drivers (such
4702 as NVidia ones) more nicely into logind. Note however, that
4703 we recommend using the open source NVidia drivers instead,
4704 and no udev rules for the closed-source drivers will be
4705 shipped from us upstream.
4707 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4708 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4709 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4710 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4711 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4712 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4713 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4714 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4715 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4716 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4717 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4718 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4723 * udev gained support for loading additional device properties
4724 from an indexed database that is keyed by vendor/product IDs
4725 and similar device identifiers. For the beginning this
4726 "hwdb" is populated with data from the well-known PCI and
4727 USB database, but also includes PNP, ACPI and OID data. In
4728 the longer run this indexed database shall grow into
4729 becoming the one central database for non-essential
4730 userspace device metadata. Previously, data from the PCI/USB
4731 database was only attached to select devices, since the
4732 lookup was a relatively expensive operation due to O(n) time
4733 complexity (with n being the number of entries in the
4734 database). Since this is now O(1), we decided to add in this
4735 data for all devices where this is available, by
4736 default. Note that the indexed database needs to be rebuilt
4737 when new data files are installed. To achieve this you need
4738 to update your packaging scripts to invoke "udevadm hwdb
4739 --update" after installation of hwdb data files. For
4740 RPM-based distributions we introduced the new
4741 %udev_hwdb_update macro for this purpose.
4743 * The Journal gained support for the "Message Catalog", an
4744 indexed database to link up additional information with
4745 journal entries. For further details please check:
4747 http://www.freedesktop.org/wiki/Software/systemd/catalog
4749 The indexed message catalog database also needs to be
4750 rebuilt after installation of message catalog files. Use
4751 "journalctl --update-catalog" for this. For RPM-based
4752 distributions we introduced the %journal_catalog_update
4753 macro for this purpose.
4755 * The Python Journal bindings gained support for the standard
4756 Python logging framework.
4758 * The Journal API gained new functions for checking whether
4759 the underlying file system of a journal file is capable of
4760 properly reporting file change notifications, or whether
4761 applications that want to reflect journal changes "live"
4762 need to recheck journal files continuously in appropriate
4765 * It is now possible to set the "age" field for tmpfiles
4766 entries to 0, indicating that files matching this entry
4767 shall always be removed when the directories are cleaned up.
4769 * coredumpctl gained a new "gdb" verb which invokes gdb
4770 right-away on the selected coredump.
4772 * There's now support for "hybrid sleep" on kernels that
4773 support this, in addition to "suspend" and "hibernate". Use
4774 "systemctl hybrid-sleep" to make use of this.
4776 * logind's HandleSuspendKey= setting (and related settings)
4777 now gained support for a new "lock" setting to simply
4778 request the screen lock on all local sessions, instead of
4779 actually executing a suspend or hibernation.
4781 * systemd will now mount the EFI variables file system by
4784 * Socket units now gained support for configuration of the
4785 SMACK security label.
4787 * timedatectl will now output the time of the last and next
4788 daylight saving change.
4790 * We dropped support for various legacy and distro-specific
4791 concepts, such as insserv, early-boot SysV services
4792 (i.e. those for non-standard runlevels such as 'b' or 'S')
4793 or ArchLinux /etc/rc.conf support. We recommend the
4794 distributions who still need support this to either continue
4795 to maintain the necessary patches downstream, or find a
4796 different solution. (Talk to us if you have questions!)
4798 * Various systemd components will now bypass PolicyKit checks
4799 for root and otherwise handle properly if PolicyKit is not
4800 found to be around. This should fix most issues for
4801 PolicyKit-less systems. Quite frankly this should have been
4802 this way since day one. It is absolutely our intention to
4803 make systemd work fine on PolicyKit-less systems, and we
4804 consider it a bug if something does not work as it should if
4805 PolicyKit is not around.
4807 * For embedded systems it is now possible to build udev and
4808 systemd without blkid and/or kmod support.
4810 * "systemctl switch-root" is now capable of switching root
4811 more than once. I.e. in addition to transitions from the
4812 initrd to the host OS it is now possible to transition to
4813 further OS images from the host. This is useful to implement
4814 offline updating tools.
4816 * Various other additions have been made to the RPM macros
4817 shipped with systemd. Use %udev_rules_update() after
4818 installing new udev rules files. %_udevhwdbdir,
4819 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4820 %_sysctldir are now available which resolve to the right
4821 directories for packages to place various data files in.
4823 * journalctl gained the new --full switch (in addition to
4824 --all, to disable ellipsation for long messages.
4826 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4827 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4828 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4829 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4830 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4831 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4832 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4833 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4834 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4838 * journalctl gained new --since= and --until= switches to
4839 filter by time. It also now supports nice filtering for
4840 units via --unit=/-u.
4842 * Type=oneshot services may use ExecReload= and do the
4845 * The journal daemon now supports time-based rotation and
4846 vacuuming, in addition to the usual disk-space based
4849 * The journal will now index the available field values for
4850 each field name. This enables clients to show pretty drop
4851 downs of available match values when filtering. The bash
4852 completion of journalctl has been updated
4853 accordingly. journalctl gained a new switch -F to list all
4854 values a certain field takes in the journal database.
4856 * More service events are now written as structured messages
4857 to the journal, and made recognizable via message IDs.
4859 * The timedated, localed and hostnamed mini-services which
4860 previously only provided support for changing time, locale
4861 and hostname settings from graphical DEs such as GNOME now
4862 also have a minimal (but very useful) text-based client
4863 utility each. This is probably the nicest way to changing
4864 these settings from the command line now, especially since
4865 it lists available options and is fully integrated with bash
4868 * There's now a new tool "systemd-coredumpctl" to list and
4869 extract coredumps from the journal.
4871 * We now install a README each in /var/log/ and
4872 /etc/rc.d/init.d explaining where the system logs and init
4873 scripts went. This hopefully should help folks who go to
4874 that dirs and look into the otherwise now empty void and
4875 scratch their heads.
4877 * When user-services are invoked (by systemd --user) the
4878 $MANAGERPID env var is set to the PID of systemd.
4880 * SIGRTMIN+24 when sent to a --user instance will now result
4881 in immediate termination of systemd.
4883 * gatewayd received numerous feature additions such as a
4884 "follow" mode, for live syncing and filtering.
4886 * browse.html now allows filtering and showing detailed
4887 information on specific entries. Keyboard navigation and
4888 mouse screen support has been added.
4890 * gatewayd/journalctl now supports HTML5/JSON
4891 Server-Sent-Events as output.
4893 * The SysV init script compatibility logic will now
4894 heuristically determine whether a script supports the
4895 "reload" verb, and only then make this available as
4898 * "systemctl status --follow" has been removed, use "journalctl
4901 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4902 have been removed since they are hardly useful to be
4905 * And I'd like to take the opportunity to specifically mention
4906 Zbigniew for his great contributions. Zbigniew, you rock!
4908 Contributions from: Andrew Eikum, Christian Hesse, Colin
4909 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4910 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4911 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4912 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4913 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4914 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4918 * If /etc/vconsole.conf is non-existent or empty we will no
4919 longer load any console font or key map at boot by
4920 default. Instead the kernel defaults will be left
4921 intact. This is definitely the right thing to do, as no
4922 configuration should mean no configuration, and hard-coding
4923 font names that are different on all archs is probably a bad
4924 idea. Also, the kernel default key map and font should be
4925 good enough for most cases anyway, and mostly identical to
4926 the userspace fonts/key maps we previously overloaded them
4927 with. If distributions want to continue to default to a
4928 non-kernel font or key map they should ship a default
4929 /etc/vconsole.conf with the appropriate contents.
4931 Contributions from: Colin Walters, Daniel J Walsh, Dave
4932 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4933 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4937 * journalctl gained a new --cursor= switch to show entries
4938 starting from the specified location in the journal.
4940 * We now enforce a size limit on journal entry fields exported
4941 with "-o json" in journalctl. Fields larger than 4K will be
4942 assigned null. This can be turned off with --all.
4944 * An (optional) journal gateway daemon is now available as
4945 "systemd-journal-gatewayd.service". This service provides
4946 access to the journal via HTTP and JSON. This functionality
4947 will be used to implement live log synchronization in both
4948 pull and push modes, but has various other users too, such
4949 as easy log access for debugging of embedded devices. Right
4950 now it is already useful to retrieve the journal via HTTP:
4952 # systemctl start systemd-journal-gatewayd.service
4953 # wget http://localhost:19531/entries
4955 This will download the journal contents in a
4956 /var/log/messages compatible format. The same as JSON:
4958 # curl -H"Accept: application/json" http://localhost:19531/entries
4960 This service is also accessible via a web browser where a
4961 single static HTML5 app is served that uses the JSON logic
4962 to enable the user to do some basic browsing of the
4963 journal. This will be extended later on. Here's an example
4964 screenshot of this app in its current state:
4966 http://0pointer.de/public/journal-gatewayd
4968 Contributions from: Kay Sievers, Lennart Poettering, Robert
4969 Milasan, Tom Gundersen
4973 * The bash completion logic is now available for journalctl
4976 * We do not mount the "cpuset" controller anymore together with
4977 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4978 started if no parameters are assigned to it. "cpuset" hence
4979 broke code that assumed it it could create "cpu" groups and
4982 * journalctl -f will now subscribe to terminal size changes,
4983 and line break accordingly.
4985 Contributions from: Dave Reisner, Kay Sievers, Lennart
4986 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4990 * nspawn will now create a symlink /etc/localtime in the
4991 container environment, copying the host's timezone
4992 setting. Previously this has been done via a bind mount, but
4993 since symlinks cannot be bind mounted this has now been
4994 changed to create/update the appropriate symlink.
4996 * journalctl -n's line number argument is now optional, and
4997 will default to 10 if omitted.
4999 * journald will now log the maximum size the journal files may
5000 take up on disk. This is particularly useful if the default
5001 built-in logic of determining this parameter from the file
5002 system size is used. Use "systemctl status
5003 systemd-journald.service" to see this information.
5005 * The multi-seat X wrapper tool has been stripped down. As X
5006 is now capable of enumerating graphics devices via udev in a
5007 seat-aware way the wrapper is not strictly necessary
5008 anymore. A stripped down temporary stop-gap is still shipped
5009 until the upstream display managers have been updated to
5010 fully support the new X logic. Expect this wrapper to be
5011 removed entirely in one of the next releases.
5013 * HandleSleepKey= in logind.conf has been split up into
5014 HandleSuspendKey= and HandleHibernateKey=. The old setting
5015 is not available anymore. X11 and the kernel are
5016 distinguishing between these keys and we should too. This
5017 also means the inhibition lock for these keys has been split
5020 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5021 Poettering, Lukas Nykryn, Václav Pavlín
5025 * Whenever a unit changes state we will now log this to the
5026 journal and show along the unit's own log output in
5029 * ConditionPathIsMountPoint= can now properly detect bind
5030 mount points too. (Previously, a bind mount of one file
5031 system to another place in the same file system could not be
5032 detected as mount, since they shared struct stat's st_dev
5035 * We will now mount the cgroup controllers cpu, cpuacct,
5036 cpuset and the controllers net_cls, net_prio together by
5039 * nspawn containers will now have a virtualized boot
5040 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5041 over with a randomized ID at container initialization). This
5042 has the effect of making "journalctl -b" do the right thing
5045 * The JSON output journal serialization has been updated not
5046 to generate "endless" list objects anymore, but rather one
5047 JSON object per line. This is more in line how most JSON
5048 parsers expect JSON objects. The new output mode
5049 "json-pretty" has been added to provide similar output, but
5050 neatly aligned for readability by humans.
5052 * We dropped all explicit sync() invocations in the shutdown
5053 code. The kernel does this implicitly anyway in the kernel
5054 reboot() syscall. halt(8)'s -n option is now a compatibility
5057 * We now support virtualized reboot() in containers, as
5058 supported by newer kernels. We will fall back to exit() if
5059 CAP_SYS_REBOOT is not available to the container. Also,
5060 nspawn makes use of this now and will actually reboot the
5061 container if the containerized OS asks for that.
5063 * journalctl will only show local log output by default
5064 now. Use --merge (-m) to show remote log output, too.
5066 * libsystemd-journal gained the new sd_journal_get_usage()
5067 call to determine the current disk usage of all journal
5068 files. This is exposed in the new "journalctl --disk-usage"
5071 * journald gained a new configuration setting SplitMode= in
5072 journald.conf which may be used to control how user journals
5073 are split off. See journald.conf(5) for details.
5075 * A new condition type ConditionFileNotEmpty= has been added.
5077 * tmpfiles' "w" lines now support file globbing, to write
5078 multiple files at once.
5080 * We added Python bindings for the journal submission
5081 APIs. More Python APIs for a number of selected APIs will
5082 likely follow. Note that we intend to add native bindings
5083 only for the Python language, as we consider it common
5084 enough to deserve bindings shipped within systemd. There are
5085 various projects outside of systemd that provide bindings
5086 for languages such as PHP or Lua.
5088 * Many conditions will now resolve specifiers such as %i. In
5089 addition, PathChanged= and related directives of .path units
5090 now support specifiers as well.
5092 * There's now a new RPM macro definition for the system preset
5095 * journald will now warn if it ca not forward a message to the
5096 syslog daemon because its socket is full.
5098 * timedated will no longer write or process /etc/timezone,
5099 except on Debian. As we do not support late mounted /usr
5100 anymore /etc/localtime always being a symlink is now safe,
5101 and hence the information in /etc/timezone is not necessary
5104 * logind will now always reserve one VT for a text getty (VT6
5105 by default). Previously if more than 6 X sessions where
5106 started they took up all the VTs with auto-spawned gettys,
5107 so that no text gettys were available anymore.
5109 * udev will now automatically inform the btrfs kernel logic
5110 about btrfs RAID components showing up. This should make
5111 simple hotplug based btrfs RAID assembly work.
5113 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5114 (but not for its children which will stay at the kernel
5115 default). This should allow setups with a lot more listening
5118 * systemd will now always pass the configured timezone to the
5119 kernel at boot. timedated will do the same when the timezone
5122 * logind's inhibition logic has been updated. By default,
5123 logind will now handle the lid switch, the power and sleep
5124 keys all the time, even in graphical sessions. If DEs want
5125 to handle these events on their own they should take the new
5126 handle-power-key, handle-sleep-key and handle-lid-switch
5127 inhibitors during their runtime. A simple way to achieve
5128 that is to invoke the DE wrapped in an invocation of:
5130 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5132 * Access to unit operations is now checked via SELinux taking
5133 the unit file label and client process label into account.
5135 * systemd will now notify the administrator in the journal
5136 when he over-mounts a non-empty directory.
5138 * There are new specifiers that are resolved in unit files,
5139 for the host name (%H), the machine ID (%m) and the boot ID
5142 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5143 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5144 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5145 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5146 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5147 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5148 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5152 * Support for reading structured kernel messages from
5153 /dev/kmsg has now been added and is enabled by default.
5155 * Support for reading kernel messages from /proc/kmsg has now
5156 been removed. If you want kernel messages in the journal
5157 make sure to run a recent kernel (>= 3.5) that supports
5158 reading structured messages from /dev/kmsg (see
5159 above). /proc/kmsg is now exclusive property of classic
5160 syslog daemons again.
5162 * The libudev API gained the new
5163 udev_device_new_from_device_id() call.
5165 * The logic for file system namespace (ReadOnlyDirectory=,
5166 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5167 require pivot_root() anymore. This means fewer temporary
5168 directories are created below /tmp for this feature.
5170 * nspawn containers will now see and receive all submounts
5171 made on the host OS below the root file system of the
5174 * Forward Secure Sealing is now supported for Journal files,
5175 which provide cryptographical sealing of journal files so
5176 that attackers cannot alter log history anymore without this
5177 being detectable. Lennart will soon post a blog story about
5178 this explaining it in more detail.
5180 * There are two new service settings RestartPreventExitStatus=
5181 and SuccessExitStatus= which allow configuration of exit
5182 status (exit code or signal) which will be excepted from the
5183 restart logic, resp. consider successful.
5185 * journalctl gained the new --verify switch that can be used
5186 to check the integrity of the structure of journal files and
5187 (if Forward Secure Sealing is enabled) the contents of
5190 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5191 and similar symlinks pre-created. This makes running shells
5192 as container init process a lot more fun.
5194 * The fstab support can now handle PARTUUID= and PARTLABEL=
5197 * A new ConditionHost= condition has been added to match
5198 against the hostname (with globs) and machine ID. This is
5199 useful for clusters where a single OS image is used to
5200 provision a large number of hosts which shall run slightly
5201 different sets of services.
5203 * Services which hit the restart limit will now be placed in a
5206 Contributions from: Bertram Poettering, Dave Reisner, Huang
5207 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5208 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5212 * When running in --user mode systemd will now become a
5213 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5214 tree a lot more organized.
5216 * A new PartOf= unit dependency type has been introduced that
5217 may be used to group services in a natural way.
5219 * "systemctl enable" may now be used to enable instances of
5222 * journalctl now prints error log levels in red, and
5223 warning/notice log levels in bright white. It also supports
5224 filtering by log level now.
5226 * cgtop gained a new -n switch (similar to top), to configure
5227 the maximum number of iterations to run for. It also gained
5228 -b, to run in batch mode (accepting no input).
5230 * The suffix ".service" may now be omitted on most systemctl
5231 command lines involving service unit names.
5233 * There's a new bus call in logind to lock all sessions, as
5234 well as a loginctl verb for it "lock-sessions".
5236 * libsystemd-logind.so gained a new call sd_journal_perror()
5237 that works similar to libc perror() but logs to the journal
5238 and encodes structured information about the error number.
5240 * /etc/crypttab entries now understand the new keyfile-size=
5243 * shutdown(8) now can send a (configurable) wall message when
5244 a shutdown is cancelled.
5246 * The mount propagation mode for the root file system will now
5247 default to "shared", which is useful to make containers work
5248 nicely out-of-the-box so that they receive new mounts from
5249 the host. This can be undone locally by running "mount
5250 --make-rprivate /" if needed.
5252 * The prefdm.service file has been removed. Distributions
5253 should maintain this unit downstream if they intend to keep
5254 it around. However, we recommend writing normal unit files
5255 for display managers instead.
5257 * Since systemd is a crucial part of the OS we will now
5258 default to a number of compiler switches that improve
5259 security (hardening) such as read-only relocations, stack
5260 protection, and suchlike.
5262 * The TimeoutSec= setting for services is now split into
5263 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5264 of individual time outs for the start and the stop phase of
5267 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5268 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5269 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5270 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5271 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5272 Gundersen, Zbigniew Jędrzejewski-Szmek
5276 * The journal and id128 C APIs are now fully documented as man
5279 * Extra safety checks have been added when transitioning from
5280 the initial RAM disk to the main system to avoid accidental
5283 * /etc/crypttab entries now understand the new keyfile-offset=
5286 * systemctl -t can now be used to filter by unit load state.
5288 * The journal C API gained the new sd_journal_wait() call to
5289 make writing synchronous journal clients easier.
5291 * journalctl gained the new -D switch to show journals from a
5294 * journalctl now displays a special marker between log
5295 messages of two different boots.
5297 * The journal is now explicitly flushed to /var via a service
5298 systemd-journal-flush.service, rather than implicitly simply
5299 by seeing /var/log/journal to be writable.
5301 * journalctl (and the journal C APIs) can now match for much
5302 more complex expressions, with alternatives and
5305 * When transitioning from the initial RAM disk to the main
5306 system we will now kill all processes in a killing spree to
5307 ensure no processes stay around by accident.
5309 * Three new specifiers may be used in unit files: %u, %h, %s
5310 resolve to the user name, user home directory resp. user
5311 shell. This is useful for running systemd user instances.
5313 * We now automatically rotate journal files if their data
5314 object hash table gets a fill level > 75%. We also size the
5315 hash table based on the configured maximum file size. This
5316 together should lower hash collisions drastically and thus
5317 speed things up a bit.
5319 * journalctl gained the new "--header" switch to introspect
5320 header data of journal files.
5322 * A new setting SystemCallFilters= has been added to services
5323 which may be used to apply blacklists or whitelists to
5324 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5326 * nspawn gained a new --link-journal= switch (and quicker: -j)
5327 to link the container journal with the host. This makes it
5328 very easy to centralize log viewing on the host for all
5329 guests while still keeping the journal files separated.
5331 * Many bugfixes and optimizations
5333 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5334 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5335 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5340 * Several tools now understand kernel command line arguments,
5341 which are only read when run in an initial RAM disk. They
5342 usually follow closely their normal counterparts, but are
5345 * There's a new tool to analyze the readahead files that are
5346 automatically generated at boot. Use:
5348 /usr/lib/systemd/systemd-readahead analyze /.readahead
5350 * We now provide an early debug shell on tty9 if this enabled. Use:
5352 systemctl enable debug-shell.service
5354 * All plymouth related units have been moved into the Plymouth
5355 package. Please make sure to upgrade your Plymouth version
5358 * systemd-tmpfiles now supports getting passed the basename of
5359 a configuration file only, in which case it will look for it
5360 in all appropriate directories automatically.
5362 * udevadm info now takes a /dev or /sys path as argument, and
5363 does the right thing. Example:
5365 udevadm info /dev/sda
5366 udevadm info /sys/class/block/sda
5368 * systemctl now prints a warning if a unit is stopped but a
5369 unit that might trigger it continues to run. Example: a
5370 service is stopped but the socket that activates it is left
5373 * "systemctl status" will now mention if the log output was
5374 shortened due to rotation since a service has been started.
5376 * The journal API now exposes functions to determine the
5377 "cutoff" times due to rotation.
5379 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5380 immediately flushing of runtime logs to /var if possible,
5381 resp. for triggering immediate rotation of the journal
5384 * It is now considered an error if a service is attempted to
5385 be stopped that is not loaded.
5387 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5389 * systemd-analyze now supports Python 3
5391 * tmpfiles now supports cleaning up directories via aging
5392 where the first level dirs are always kept around but
5393 directories beneath it automatically aged. This is enabled
5394 by prefixing the age field with '~'.
5396 * Seat objects now expose CanGraphical, CanTTY properties
5397 which is required to deal with very fast bootups where the
5398 display manager might be running before the graphics drivers
5399 completed initialization.
5401 * Seat objects now expose a State property.
5403 * We now include RPM macros for service enabling/disabling
5404 based on the preset logic. We recommend RPM based
5405 distributions to make use of these macros if possible. This
5406 makes it simpler to reuse RPM spec files across
5409 * We now make sure that the collected systemd unit name is
5410 always valid when services log to the journal via
5413 * There's a new man page kernel-command-line(7) detailing all
5414 command line options we understand.
5416 * The fstab generator may now be disabled at boot by passing
5417 fstab=0 on the kernel command line.
5419 * A new kernel command line option modules-load= is now understood
5420 to load a specific kernel module statically, early at boot.
5422 * Unit names specified on the systemctl command line are now
5423 automatically escaped as needed. Also, if file system or
5424 device paths are specified they are automatically turned
5425 into the appropriate mount or device unit names. Example:
5427 systemctl status /home
5428 systemctl status /dev/sda
5430 * The SysVConsole= configuration option has been removed from
5431 system.conf parsing.
5433 * The SysV search path is no longer exported on the D-Bus
5436 * The Names= option is been removed from unit file parsing.
5438 * There's a new man page bootup(7) detailing the boot process.
5440 * Every unit and every generator we ship with systemd now
5441 comes with full documentation. The self-explanatory boot is
5444 * A couple of services gained "systemd-" prefixes in their
5445 name if they wrap systemd code, rather than only external
5446 code. Among them fsck@.service which is now
5447 systemd-fsck@.service.
5449 * The HaveWatchdog property has been removed from the D-Bus
5452 * systemd.confirm_spawn= on the kernel command line should now
5455 * There's a new man page crypttab(5) which details all options
5456 we actually understand.
5458 * systemd-nspawn gained a new --capability= switch to pass
5459 additional capabilities to the container.
5461 * timedated will now read known NTP implementation unit names
5462 from /usr/lib/systemd/ntp-units.d/*.list,
5463 systemd-timedated-ntp.target has been removed.
5465 * journalctl gained a new switch "-b" that lists log data of
5466 the current boot only.
5468 * The notify socket is in the abstract namespace again, in
5469 order to support daemons which chroot() at start-up.
5471 * There is a new Storage= configuration option for journald
5472 which allows configuration of where log data should go. This
5473 also provides a way to disable journal logging entirely, so
5474 that data collected is only forwarded to the console, the
5475 kernel log buffer or another syslog implementation.
5477 * Many bugfixes and optimizations
5479 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5480 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5481 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5482 Shawn Landden, Tom Gundersen
5486 * "systemctl help <unit>" now shows the man page if one is
5489 * Several new man pages have been added.
5491 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5492 MaxLevelConsole= can now be specified in
5493 journald.conf. These options allow reducing the amount of
5494 data stored on disk or forwarded by the log level.
5496 * TimerSlackNSec= can now be specified in system.conf for
5497 PID1. This allows system-wide power savings.
5499 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5500 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5505 * logind is now capable of (optionally) handling power and
5506 sleep keys as well as the lid switch.
5508 * journalctl now understands the syntax "journalctl
5509 /usr/bin/avahi-daemon" to get all log output of a specific
5512 * CapabilityBoundingSet= in system.conf now also influences
5513 the capability bound set of usermode helpers of the kernel.
5515 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5516 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5517 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5518 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5522 * Note that we skipped 139 releases here in order to set the
5523 new version to something that is greater than both udev's
5524 and systemd's most recent version number.
5526 * udev: all udev sources are merged into the systemd source tree now.
5527 All future udev development will happen in the systemd tree. It
5528 is still fully supported to use the udev daemon and tools without
5529 systemd running, like in initramfs or other init systems. Building
5530 udev though, will require the *build* of the systemd tree, but
5531 udev can be properly *run* without systemd.
5533 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5534 should be used to create dead device nodes as workarounds for broken
5537 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5538 no longer supported. udev_monitor_new_from_netlink() needs to be
5539 used to subscribe to events.
5541 * udev: when udevd is started by systemd, processes which are left
5542 behind by forking them off of udev rules, are unconditionally cleaned
5543 up and killed now after the event handling has finished. Services or
5544 daemons must be started as systemd services. Services can be
5545 pulled-in by udev to get started, but they can no longer be directly
5546 forked by udev rules.
5548 * udev: the daemon binary is called systemd-udevd now and installed
5549 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5550 to adapt to that, create symlink, or rename the binary after building
5553 * libudev no longer provides these symbols:
5554 udev_monitor_from_socket()
5555 udev_queue_get_failed_list_entry()
5556 udev_get_{dev,sys,run}_path()
5557 The versions number was bumped and symbol versioning introduced.
5559 * systemd-loginctl and systemd-journalctl have been renamed
5560 to loginctl and journalctl to match systemctl.
5562 * The config files: /etc/systemd/systemd-logind.conf and
5563 /etc/systemd/systemd-journald.conf have been renamed to
5564 logind.conf and journald.conf. Package updates should rename
5565 the files to the new names on upgrade.
5567 * For almost all files the license is now LGPL2.1+, changed
5568 from the previous GPL2.0+. Exceptions are some minor stuff
5569 of udev (which will be changed to LGPL2.1 eventually, too),
5570 and the MIT licensed sd-daemon.[ch] library that is suitable
5571 to be used as drop-in files.
5573 * systemd and logind now handle system sleep states, in
5574 particular suspending and hibernating.
5576 * logind now implements a sleep/shutdown/idle inhibiting logic
5577 suitable for a variety of uses. Soonishly Lennart will blog
5578 about this in more detail.
5580 * var-run.mount and var-lock.mount are no longer provided
5581 (which prevously bind mounted these directories to their new
5582 places). Distributions which have not converted these
5583 directories to symlinks should consider stealing these files
5584 from git history and add them downstream.
5586 * We introduced the Documentation= field for units and added
5587 this to all our shipped units. This is useful to make it
5588 easier to explore the boot and the purpose of the various
5591 * All smaller setup units (such as
5592 systemd-vconsole-setup.service) now detect properly if they
5593 are run in a container and are skipped when
5594 appropriate. This guarantees an entirely noise-free boot in
5595 Linux container environments such as systemd-nspawn.
5597 * A framework for implementing offline system updates is now
5598 integrated, for details see:
5599 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5601 * A new service type Type=idle is available now which helps us
5602 avoiding ugly interleaving of getty output and boot status
5605 * There's now a system-wide CapabilityBoundingSet= option to
5606 globally reduce the set of capabilities for the
5607 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5608 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5609 even CAP_NET_ADMIN system-wide for secure systems.
5611 * There are now system-wide DefaultLimitXXX= options to
5612 globally change the defaults of the various resource limits
5613 for all units started by PID 1.
5615 * Harald Hoyer's systemd test suite has been integrated into
5616 systemd which allows easy testing of systemd builds in qemu
5617 and nspawn. (This is really awesome! Ask us for details!)
5619 * The fstab parser is now implemented as generator, not inside
5622 * systemctl will now warn you if .mount units generated from
5623 /etc/fstab are out of date due to changes in fstab that
5624 have not been read by systemd yet.
5626 * systemd is now suitable for usage in initrds. Dracut has
5627 already been updated to make use of this. With this in place
5628 initrds get a slight bit faster but primarily are much
5629 easier to introspect and debug since "systemctl status" in
5630 the host system can be used to introspect initrd services,
5631 and the journal from the initrd is kept around too.
5633 * systemd-delta has been added, a tool to explore differences
5634 between user/admin configuration and vendor defaults.
5636 * PrivateTmp= now affects both /tmp and /var/tmp.
5638 * Boot time status messages are now much prettier and feature
5639 proper english language. Booting up systemd has never been
5642 * Read-ahead pack files now include the inode number of all
5643 files to pre-cache. When the inode changes the pre-caching
5644 is not attempted. This should be nicer to deal with updated
5645 packages which might result in changes of read-ahead
5648 * We now temporaritly lower the kernel's read_ahead_kb variable
5649 when collecting read-ahead data to ensure the kernel's
5650 built-in read-ahead does not add noise to our measurements
5651 of necessary blocks to pre-cache.
5653 * There's now RequiresMountsFor= to add automatic dependencies
5654 for all mounts necessary for a specific file system path.
5656 * MountAuto= and SwapAuto= have been removed from
5657 system.conf. Mounting file systems at boot has to take place
5660 * nspawn now learned a new switch --uuid= to set the machine
5661 ID on the command line.
5663 * nspawn now learned the -b switch to automatically search
5666 * vt102 is now the default TERM for serial TTYs, upgraded from
5669 * systemd-logind now works on VT-less systems.
5671 * The build tree has been reorganized. The individual
5672 components now have directories of their own.
5674 * A new condition type ConditionPathIsReadWrite= is now available.
5676 * nspawn learned the new -C switch to create cgroups for the
5677 container in other hierarchies.
5679 * We now have support for hardware watchdogs, configurable in
5682 * The scheduled shutdown logic now has a public API.
5684 * We now mount /tmp as tmpfs by default, but this can be
5685 masked and /etc/fstab can override it.
5687 * Since udisks does not make use of /media anymore we are not
5688 mounting a tmpfs on it anymore.
5690 * journalctl gained a new --local switch to only interleave
5691 locally generated journal files.
5693 * We can now load the IMA policy at boot automatically.
5695 * The GTK tools have been split off into a systemd-ui.
5697 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5698 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5699 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5700 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5701 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5702 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5703 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5704 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5705 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5710 * This is mostly a bugfix release
5712 * Support optional initialization of the machine ID from the
5713 KVM or container configured UUID.
5715 * Support immediate reboots with "systemctl reboot -ff"
5717 * Show /etc/os-release data in systemd-analyze output
5719 * Many bugfixes for the journal, including endianness fixes and
5720 ensuring that disk space enforcement works
5722 * sd-login.h is C++ comptaible again
5724 * Extend the /etc/os-release format on request of the Debian
5727 * We now refuse non-UTF8 strings used in various configuration
5728 and unit files. This is done to ensure we do not pass invalid
5729 data over D-Bus or expose it elsewhere.
5731 * Register Mimo USB Screens as suitable for automatic seat
5734 * Read SELinux client context from journal clients in a race
5737 * Reorder configuration file lookup order. /etc now always
5738 overrides /run in order to allow the administrator to always
5739 and unconditionally override vendor-supplied or
5740 automatically generated data.
5742 * The various user visible bits of the journal now have man
5743 pages. We still lack man pages for the journal API calls
5746 * We now ship all man pages in HTML format again in the
5749 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5750 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5751 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5756 * This is mostly a bugfix release
5758 * systems lacking /etc/os-release are no longer supported.
5760 * Various functionality updates to libsystemd-login.so
5762 * Track class of PAM logins to distinguish greeters from
5765 Contributions from: Kay Sievers, Lennart Poettering, Michael
5770 * This is an important bugfix release for v41.
5772 * Building man pages is now optional which should be useful
5773 for those building systemd from git but unwilling to install
5776 * Watchdog support for supervising services is now usable. In
5777 a future release support for hardware watchdogs
5778 (i.e. /dev/watchdog) will be added building on this.
5780 * Service start rate limiting is now configurable and can be
5781 turned off per service. When a start rate limit is hit a
5782 reboot can automatically be triggered.
5784 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5786 Contributions from: Benjamin Franzke, Bill Nottingham,
5787 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5788 Schmidt, Michał Górny, Piotr Drąg
5792 * The systemd binary is installed /usr/lib/systemd/systemd now;
5793 An existing /sbin/init symlink needs to be adapted with the
5796 * The code that loads kernel modules has been ported to invoke
5797 libkmod directly, instead of modprobe. This means we do not
5798 support systems with module-init-tools anymore.
5800 * Watchdog support is now already useful, but still not
5803 * A new kernel command line option systemd.setenv= is
5804 understood to set system wide environment variables
5805 dynamically at boot.
5807 * We now limit the set of capabilities of systemd-journald.
5809 * We now set SIGPIPE to ignore by default, since it only is
5810 useful in shell pipelines, and has little use in general
5811 code. This can be disabled with IgnoreSIPIPE=no in unit
5814 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5815 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5820 * This is mostly a bugfix release
5822 * We now expose the reason why a service failed in the
5823 "Result" D-Bus property.
5825 * Rudimentary service watchdog support (will be completed over
5826 the next few releases.)
5828 * When systemd forks off in order execute some service we will
5829 now immediately changes its argv[0] to reflect which process
5830 it will execute. This is useful to minimize the time window
5831 with a generic argv[0], which makes bootcharts more useful
5833 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5834 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5835 Mike Kazantsev, Ray Strode
5839 * This is mostly a test release, but incorporates many
5842 * New systemd-cgtop tool to show control groups by their
5845 * Linking against libacl for ACLs is optional again. If
5846 disabled, support tracking device access for active logins
5847 goes becomes unavailable, and so does access to the user
5848 journals by the respective users.
5850 * If a group "adm" exists, journal files are automatically
5851 owned by them, thus allow members of this group full access
5852 to the system journal as well as all user journals.
5854 * The journal now stores the SELinux context of the logging
5855 client for all entries.
5857 * Add C++ inclusion guards to all public headers
5859 * New output mode "cat" in the journal to print only text
5860 messages, without any meta data like date or time.
5862 * Include tiny X server wrapper as a temporary stop-gap to
5863 teach XOrg udev display enumeration. This is used by display
5864 managers such as gdm, and will go away as soon as XOrg
5865 learned native udev hotplugging for display devices.
5867 * Add new systemd-cat tool for executing arbitrary programs
5868 with STDERR/STDOUT connected to the journal. Can also act as
5869 BSD logger replacement, and does so by default.
5871 * Optionally store all locally generated coredumps in the
5872 journal along with meta data.
5874 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5875 writing short strings to files (for usage for /sys), and for
5876 creating symlinks, character and block device nodes.
5878 * New unit file option ControlGroupPersistent= to make cgroups
5879 persistent, following the mechanisms outlined in
5880 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5882 * Support multiple local RTCs in a sane way
5884 * No longer monopolize IO when replaying readahead data on
5885 rotating disks, since we might starve non-file-system IO to
5886 death, since fanotify() will not see accesses done by blkid,
5889 * Do not show kernel threads in systemd-cgls anymore, unless
5890 requested with new -k switch.
5892 Contributions from: Dan Horák, Kay Sievers, Lennart
5893 Poettering, Michal Schmidt
5897 * This is mostly a test release, but incorporates many
5900 * The git repository moved to:
5901 git://anongit.freedesktop.org/systemd/systemd
5902 ssh://git.freedesktop.org/git/systemd/systemd
5904 * First release with the journal
5905 http://0pointer.de/blog/projects/the-journal.html
5907 * The journal replaces both systemd-kmsg-syslogd and
5908 systemd-stdout-bridge.
5910 * New sd_pid_get_unit() API call in libsystemd-logind
5912 * Many systemadm clean-ups
5914 * Introduce remote-fs-pre.target which is ordered before all
5915 remote mounts and may be used to start services before all
5918 * Added Mageia support
5920 * Add bash completion for systemd-loginctl
5922 * Actively monitor PID file creation for daemons which exit in
5923 the parent process before having finished writing the PID
5924 file in the daemon process. Daemons which do this need to be
5925 fixed (i.e. PID file creation must have finished before the
5926 parent exits), but we now react a bit more gracefully to them.
5928 * Add colourful boot output, mimicking the well-known output
5929 of existing distributions.
5931 * New option PassCredentials= for socket units, for
5932 compatibility with a recent kernel ABI breakage.
5934 * /etc/rc.local is now hooked in via a generator binary, and
5935 thus will no longer act as synchronization point during
5938 * systemctl list-unit-files now supports --root=.
5940 * systemd-tmpfiles now understands two new commands: z, Z for
5941 relabelling files according to the SELinux database. This is
5942 useful to apply SELinux labels to specific files in /sys,
5945 * Output of SysV services is now forwarded to both the console
5946 and the journal by default, not only just the console.
5948 * New man pages for all APIs from libsystemd-login.
5950 * The build tree got reorganized and a the build system is a
5951 lot more modular allowing embedded setups to specifically
5952 select the components of systemd they are interested in.
5954 * Support for Linux systems lacking the kernel VT subsystem is
5957 * configure's --with-rootdir= got renamed to
5958 --with-rootprefix= to follow the naming used by udev and
5961 * Unless specified otherwise we will now install to /usr instead
5962 of /usr/local by default.
5964 * Processes with '@' in argv[0][0] are now excluded from the
5965 final shut-down killing spree, following the logic explained
5967 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5969 * All processes remaining in a service cgroup when we enter
5970 the START or START_PRE states are now killed with
5971 SIGKILL. That means it is no longer possible to spawn
5972 background processes from ExecStart= lines (which was never
5973 supported anyway, and bad style).
5975 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5976 reloading of units together.
5978 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5979 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5980 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5981 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5982 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek