1 systemd System and Service Manager
3 CHANGES WITH 230 in spe:
5 * DNSSEC is now turned on by default in systemd-resolved (in
6 "allow-downgrade" mode), but may be turned off during compile time by
7 passing "--with-default-dnssec=no" to "configure" (and of course,
8 during runtime with DNSSEC= in resolved.conf). We recommend
9 downstreams to leave this on at least during development cycles and
10 report any issues with the DNSSEC logic upstream. We are very
11 interested in collecting feedback about the DNSSEC validator and its
12 limitations in the wild. Note however, that DNSSEC support is
13 probably nothing downstreams should turn on in stable distros just
14 yet, as it might create incompatibilities with a few DNS servers and
15 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
16 automatically whenever we detect such incompatible setups, but there
17 might be systems we do not cover yet. Hence: please help us testing
18 the DNSSEC code, leave this on where you can, report back, but then
19 again don't consider turning this on in your stable, LTS or
20 production release just yet.
22 * Testing tool /usr/lib/systemd/systemd-activate is renamed to
23 systemd-socket-activate and installed into /usr/bin. It is now fully
28 * The systemd-resolved DNS resolver service has gained a substantial
29 set of new features, most prominently it may now act as a DNSSEC
30 validating stub resolver. DNSSEC mode is currently turned off by
31 default, but is expected to be turned on by default in one of the
32 next releases. For now, we invite everybody to test the DNSSEC logic
33 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
34 service also gained a full set of D-Bus interfaces, including calls
35 to configure DNS and DNSSEC settings per link (for use by external
36 network management software). systemd-resolved and systemd-networkd
37 now distinguish between "search" and "routing" domains. The former
38 are used to qualify single-label names, the latter are used purely
39 for routing lookups within certain domains to specific links.
40 resolved now also synthesizes RRs for all entries from /etc/hosts.
42 * The systemd-resolve tool (which is a client utility for
43 systemd-resolved) has been improved considerably and is now fully
44 supported and documented. Hence it has moved from /usr/lib/systemd to
47 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
50 * The coredump collection logic has been reworked: when a coredump is
51 collected it is now written to disk, compressed and processed
52 (including stacktrace extraction) from a new instantiated service
53 systemd-coredump@.service, instead of directly from the
54 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
55 processing large coredumps can take up a substantial amount of
56 resources and time, and this previously happened entirely outside of
57 systemd's service supervision. With the new logic the core_pattern
58 hook only does minimal metadata collection before passing off control
59 to the new instantiated service, which is configured with a time
60 limit, a nice level and other settings to minimize negative impact on
61 the rest of the system. Also note that the new logic will honour the
62 RLIMIT_CORE setting of the crashed process, which now allows users
63 and processes to turn off coredumping for their processes by setting
66 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
67 and all forked processes by default. Previously, PID 1 would leave
68 the setting at "0" for all processes, as set by the kernel. Note that
69 the resource limit traditionally has no effect on the generated
70 coredumps on the system if the /proc/sys/kernel/core_pattern hook
71 logic is used. Since the limit is now honoured (see above) its
72 default has been changed so that the coredumping logic is enabled by
73 default for all processes, while allowing specific opt-out.
75 * When the stacktrace is extracted from processes of system users, this
76 is now done as "systemd-coredump" user, in order to sandbox this
77 potentially security sensitive parsing operation. (Note that when
78 processing coredumps of normal users this is done under the user ID
79 of process that crashed, as before.) Packagers should take notice
80 that it is now necessary to create the "systemd-coredump" system user
81 and group at package installation time.
83 * The systemd-activate socket activation testing tool gained support
84 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
85 and --seqpacket switches. It also has been extended to support both
86 new-style and inetd-style file descriptor passing. Use the new
87 --inetd switch to request inetd-style file descriptor passing.
89 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
90 variable, which takes a boolean value. If set to false, ANSI color
91 output is disabled in the tools even when run on a terminal that
94 * The VXLAN support in networkd now supports two new settings
95 DestinationPort= and PortRange=.
97 * A new systemd.machine_id= kernel command line switch has been added,
98 that may be used to set the machine ID in /etc/machine-id if it is
99 not initialized yet. This command line option has no effect if the
100 file is already initialized.
102 * systemd-nspawn gained a new --as-pid2 switch that invokes any
103 specified command line as PID 2 rather than PID 1 in the
104 container. In this mode PID 1 is a minimal stub init process that
105 implements the special POSIX and Linux semantics of PID 1 regarding
106 signal and child process management. Note that this stub init process
107 is implemented in nspawn itself and requires no support from the
108 container image. This new logic is useful to support running
109 arbitrary commands in the container, as normal processes are
110 generally not prepared to run as PID 1.
112 * systemd-nspawn gained a new --chdir= switch for setting the current
113 working directory for the process started in the container.
115 * "journalctl /dev/sda" will now output all kernel log messages for
116 specified device from the current boot, in addition to all devices
117 that are parents of it. This should make log output about devices
118 pretty useful, as long as kernel drivers attach enough metadata to
119 the log messages. (The usual SATA drivers do.)
121 * The sd-journal API gained two new calls
122 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
123 that report whether log data from /run or /var has been found.
125 * journalctl gained a new switch "--fields" that prints all journal
126 record field names currently in use in the journal. This is backed
127 by two new sd-journal API calls sd_journal_enumerate_fields() and
128 sd_journal_restart_fields().
130 * Most configurable timeouts in systemd now expect an argument of
131 "infinity" to turn them off, instead of "0" as before. The semantics
132 from now on is that a timeout of "0" means "now", and "infinity"
133 means "never". To maintain backwards compatibility, "0" continues to
134 turn off previously existing timeout settings.
136 * "systemctl reload-or-try-restart" has been renamed to "systemctl
137 try-reload-or-restart" to clarify what it actually does: the "try"
138 logic applies to both reloading and restarting, not just restarting.
139 The old name continues to be accepted for compatibility.
141 * On boot-up, when PID 1 detects that the system clock is behind the
142 release date of the systemd version in use, the clock is now set
143 to the latter. Previously, this was already done in timesyncd, in order
144 to avoid running with clocks set to the various clock epochs such as
145 1902, 1938 or 1970. With this change the logic is now done in PID 1
146 in addition to timesyncd during early boot-up, so that it is enforced
147 before the first process is spawned by systemd. Note that the logic
148 in timesyncd remains, as it is more comprehensive and ensures
149 clock monotonicity by maintaining a persistent timestamp file in
150 /var. Since /var is generally not available in earliest boot or the
151 initrd, this part of the logic remains in timesyncd, and is not done
154 * Support for tweaking details in net_cls.class_id through the
155 NetClass= configuration directive has been removed, as the kernel
156 people have decided to deprecate that controller in cgroup v2.
157 Userspace tools such as nftables are moving over to setting rules
158 that are specific to the full cgroup path of a task, which obsoletes
159 these controllers anyway. The NetClass= directive is kept around for
160 legacy compatibility reasons. For a more in-depth description of the
161 kernel change, please refer to the respective upstream commit:
163 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
165 * A new service setting RuntimeMaxSec= has been added that may be used
166 to specify a maximum runtime for a service. If the timeout is hit, the
167 service is terminated and put into a failure state.
169 * A new service setting AmbientCapabilities= has been added. It allows
170 configuration of additional Linux process capabilities that are
171 passed to the activated processes. This is only available on very
174 * The process resource limit settings in service units may now be used
175 to configure hard and soft limits individually.
177 * The various libsystemd APIs such as sd-bus or sd-event now publicly
178 expose support for gcc's __attribute__((cleanup())) C extension.
179 Specifically, for many object destructor functions alternative
180 versions have been added that have names suffixed with "p" and take a
181 pointer to a pointer to the object to destroy, instead of just a
182 pointer to the object itself. This is useful because these destructor
183 functions may be used directly as parameters to the cleanup
184 construct. Internally, systemd has been a heavy user of this GCC
185 extension for a long time, and with this change similar support is
186 now available to consumers of the library outside of systemd. Note
187 that by using this extension in your sources compatibility with old
188 and strictly ANSI compatible C compilers is lost. However, all gcc or
189 LLVM versions of recent years support this extension.
191 * Timer units gained support for a new setting RandomizedDelaySec= that
192 allows configuring some additional randomized delay to the configured
193 time. This is useful to spread out timer events to avoid load peaks in
194 clusters or larger setups.
196 * Calendar time specifications now support sub-second accuracy.
198 * Socket units now support listening on SCTP and UDP-lite protocol
201 * The sd-event API now comes with a full set of man pages.
203 * Older versions of systemd contained experimental support for
204 compressing journal files and coredumps with the LZ4 compressor that
205 was not compatible with the lz4 binary (due to API limitations of the
206 lz4 library). This support has been removed; only support for files
207 compatible with the lz4 binary remains. This LZ4 logic is now
208 officially supported and no longer considered experimental.
210 * The dkr image import logic has been removed again from importd. dkr's
211 micro-services focus doesn't fit into the machine image focus of
212 importd, and quickly got out of date with the upstream dkr API.
214 * Creation of the /run/lock/lockdev/ directory was dropped from
215 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
216 been available for many years. If you still need this, you need to
217 create your own tmpfiles.d config file with:
219 d /run/lock/lockdev 0775 root lock -
221 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
222 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
223 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
224 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
225 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
226 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
227 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
228 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
229 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
230 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
231 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
232 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
233 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
234 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
235 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
236 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
237 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
238 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
239 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
241 -- Berlin, 2016-02-11
245 * A number of properties previously only settable in unit
246 files are now also available as properties to set when
247 creating transient units programmatically via the bus, as it
248 is exposed with systemd-run's --property=
249 setting. Specifically, these are: SyslogIdentifier=,
250 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
251 EnvironmentFile=, ReadWriteDirectories=,
252 ReadOnlyDirectories=, InaccessibleDirectories=,
253 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
255 * When creating transient services via the bus API it is now
256 possible to pass in a set of file descriptors to use as
257 STDIN/STDOUT/STDERR for the invoked process.
259 * Slice units may now be created transiently via the bus APIs,
260 similar to the way service and scope units may already be
263 * Wherever systemd expects a calendar timestamp specification
264 (like in journalctl's --since= and --until= switches) UTC
265 timestamps are now supported. Timestamps suffixed with "UTC"
266 are now considered to be in Universal Time Coordinated
267 instead of the local timezone. Also, timestamps may now
268 optionally be specified with sub-second accuracy. Both of
269 these additions also apply to recurring calendar event
270 specification, such as OnCalendar= in timer units.
272 * journalctl gained a new "--sync" switch that asks the
273 journal daemon to write all so far unwritten log messages to
274 disk and sync the files, before returning.
276 * systemd-tmpfiles learned two new line types "q" and "Q" that
277 operate like "v", but also set up a basic btrfs quota
278 hierarchy when used on a btrfs file system with quota
281 * tmpfiles' "v", "q" and "Q" will now create a plain directory
282 instead of a subvolume (even on a btrfs file system) if the
283 root directory is a plain directory, and not a
284 subvolume. This should simplify things with certain chroot()
285 environments which are not aware of the concept of btrfs
288 * systemd-detect-virt gained a new --chroot switch to detect
289 whether execution takes place in a chroot() environment.
291 * CPUAffinity= now takes CPU index ranges in addition to
294 * The various memory-related resource limit settings (such as
295 LimitAS=) now understand the usual K, M, G, ... suffixes to
296 the base of 1024 (IEC). Similar, the time-related resource
297 limit settings understand the usual min, h, day, ...
300 * There's a new system.conf setting DefaultTasksMax= to
301 control the default TasksMax= setting for services and
302 scopes running on the system. (TasksMax= is the primary
303 setting that exposes the "pids" cgroup controller on systemd
304 and was introduced in the previous systemd release.) The
305 setting now defaults to 512, which means services that are
306 not explicitly configured otherwise will only be able to
307 create 512 processes or threads at maximum, from this
308 version on. Note that this means that thread- or
309 process-heavy services might need to be reconfigured to set
310 TasksMax= to a higher value. It is sufficient to set
311 TasksMax= in these specific unit files to a higher value, or
312 even "infinity". Similar, there's now a logind.conf setting
313 UserTasksMax= that defaults to 4096 and limits the total
314 number of processes or tasks each user may own
315 concurrently. nspawn containers also have the TasksMax=
316 value set by default now, to 8192. Note that all of this
317 only has an effect if the "pids" cgroup controller is
318 enabled in the kernel. The general benefit of these changes
319 should be a more robust and safer system, that provides a
320 certain amount of per-service fork() bomb protection.
322 * systemd-nspawn gained the new --network-veth-extra= switch
323 to define additional and arbitrarily-named virtual Ethernet
324 links between the host and the container.
326 * A new service execution setting PassEnvironment= has been
327 added that allows importing select environment variables
328 from PID1's environment block into the environment block of
331 * Timer units gained support for a new RemainAfterElapse=
332 setting which takes a boolean argument. It defaults on on,
333 exposing behaviour unchanged to previous releases. If set to
334 off, timer units are unloaded after they elapsed if they
335 cannot elapse again. This is particularly useful for
336 transient timer units, which shall not stay around longer
337 than until they first elapse.
339 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
340 default now (the kernel default is 16). This is beneficial
341 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
342 allows substantially larger numbers of queued
343 datagrams. This should increase the capability of systemd to
344 parallelize boot-up, as logging and sd_notify() are unlikely
345 to stall execution anymore. If you need to change the value
346 from the new defaults, use the usual sysctl.d/ snippets.
348 * The compression framing format used by the journal or
349 coredump processing has changed to be in line with what the
350 official LZ4 tools generate. LZ4 compression support in
351 systemd was considered unsupported previously, as the format
352 was not compatible with the normal tools. With this release
353 this has changed now, and it is hence safe for downstream
354 distributions to turn it on. While not compressing as well
355 as the XZ, LZ4 is substantially faster, which makes
356 it a good default choice for the compression logic in the
357 journal and in coredump handling.
359 * Any reference to /etc/mtab has been dropped from
360 systemd. The file has been obsolete since a while, but
361 systemd refused to work on systems where it was incorrectly
362 set up (it should be a symlink or non-existent). Please make
363 sure to update to util-linux 2.27.1 or newer in conjunction
364 with this systemd release, which also drops any reference to
365 /etc/mtab. If you maintain a distribution make sure that no
366 software you package still references it, as this is a
367 likely source of bugs. There's also a glibc bug pending,
368 asking for removal of any reference to this obsolete file:
370 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
372 * Support for the ".snapshot" unit type has been removed. This
373 feature turned out to be little useful and little used, and
374 has now been removed from the core and from systemctl.
376 * The dependency types RequiresOverridable= and
377 RequisiteOverridable= have been removed from systemd. They
378 have been used only very sparingly to our knowledge and
379 other options that provide a similar effect (such as
380 systemctl --mode=ignore-dependencies) are much more useful
381 and commonly used. Moreover, they were only half-way
382 implemented as the option to control behaviour regarding
383 these dependencies was never added to systemctl. By removing
384 these dependency types the execution engine becomes a bit
385 simpler. Unit files that use these dependencies should be
386 changed to use the non-Overridable dependency types
387 instead. In fact, when parsing unit files with these
388 options, that's what systemd will automatically convert them
389 too, but it will also warn, asking users to fix the unit
390 files accordingly. Removal of these dependency types should
391 only affect a negligible number of unit files in the wild.
393 * Behaviour of networkd's IPForward= option changed
394 (again). It will no longer maintain a per-interface setting,
395 but propagate one way from interfaces where this is enabled
396 to the global kernel setting. The global setting will be
397 enabled when requested by a network that is set up, but
398 never be disabled again. This change was made to make sure
399 IPv4 and IPv6 behaviour regarding packet forwarding is
400 similar (as the Linux IPv6 stack does not support
401 per-interface control of this setting) and to minimize
404 * In unit files the behaviour of %u, %U, %h, %s has
405 changed. These specifiers will now unconditionally resolve
406 to the various user database fields of the user that the
407 systemd instance is running as, instead of the user
408 configured in the specific unit via User=. Note that this
409 effectively doesn't change much, as resolving of these
410 specifiers was already turned off in the --system instance
411 of systemd, as we cannot do NSS lookups from PID 1. In the
412 --user instance of systemd these specifiers where correctly
413 resolved, but hardly made any sense, since the user instance
414 lacks privileges to do user switches anyway, and User= is
415 hence useless. Morever, even in the --user instance of
416 systemd behaviour was awkward as it would only take settings
417 from User= assignment placed before the specifier into
418 account. In order to unify and simplify the logic around
419 this the specifiers will now always resolve to the
420 credentials of the user invoking the manager (which in case
421 of PID 1 is the root user).
423 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
424 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
425 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
426 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
427 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
428 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
429 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
430 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
431 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
432 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
433 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
434 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
435 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
436 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
439 -- Berlin, 2015-11-18
443 * systemd now depends on util-linux v2.27. More specifically,
444 the newly added mount monitor feature in libmount now
445 replaces systemd's former own implementation.
447 * libmount mandates /etc/mtab not to be regular file, and
448 systemd now enforces this condition at early boot.
449 /etc/mtab has been deprecated and warned about for a very
450 long time, so systems running systemd should already have
451 stopped having this file around as anything else than a
452 symlink to /proc/self/mounts.
454 * Support for the "pids" cgroup controller has been added. It
455 allows accounting the number of tasks in a cgroup and
456 enforcing limits on it. This adds two new setting
457 TasksAccounting= and TasksMax= to each unit, as well as a
458 global option DefaultTasksAccounting=.
460 * Support for the "net_cls" cgroup controller has been added.
461 It allows assigning a net class ID to each task in the
462 cgroup, which can then be used in firewall rules and traffic
463 shaping configurations. Note that the kernel netfilter net
464 class code does not currently work reliably for ingress
465 packets on unestablished sockets.
467 This adds a new config directive called NetClass= to CGroup
468 enabled units. Allowed values are positive numbers for fixed
469 assignments and "auto" for picking a free value
472 * 'systemctl is-system-running' now returns 'offline' if the
473 system is not booted with systemd. This command can now be
474 used as a substitute for 'systemd-notify --booted'.
476 * Watchdog timeouts have been increased to 3 minutes for all
477 in-tree service files. Apparently, disk IO issues are more
478 frequent than we hoped, and user reported >1 minute waiting
481 * 'machine-id-commit' functionality has been merged into
482 'machine-id-setup --commit'. The separate binary has been
485 * The WorkingDirectory= directive in unit files may now be set
486 to the special value '~'. In this case, the working
487 directory is set to the home directory of the user
490 * "machinectl shell" will now open the shell in the home
491 directory of the selected user by default.
493 * The CrashChVT= configuration file setting is renamed to
494 CrashChangeVT=, following our usual logic of not
495 abbreviating unnecessarily. The old directive is still
496 supported for compat reasons. Also, this directive now takes
497 an integer value between 1 and 63, or a boolean value. The
498 formerly supported '-1' value for disabling stays around for
501 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
502 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
503 RootDirectory= properties can now be set for transient
506 * The systemd-analyze tool gained a new "set-log-target" verb
507 to change the logging target the system manager logs to
508 dynamically during runtime. This is similar to how
509 "systemd-analyze set-log-level" already changes the log
512 * In nspawn /sys is now mounted as tmpfs, with only a selected
513 set of subdirectories mounted in from the real sysfs. This
514 enhances security slightly, and is useful for ensuring user
515 namespaces work correctly.
517 * Support for USB FunctionFS activation has been added. This
518 allows implementation of USB gadget services that are
519 activated as soon as they are requested, so that they don't
520 have to run continously, similar to classic socket
523 * The "systemctl exit" command now optionally takes an
524 additional parameter that sets the exit code to return from
525 the systemd manager when exiting. This is only relevant when
526 running the systemd user instance, or when running the
527 system instance in a container.
529 * sd-bus gained the new API calls sd_bus_path_encode_many()
530 and sd_bus_path_decode_many() that allow easy encoding and
531 decoding of multiple identifier strings inside a D-Bus
532 object path. Another new call sd_bus_default_flush_close()
533 has been added to flush and close per-thread default
536 * systemd-cgtop gained support for a -M/--machine= switch to
537 show the control groups within a certain container only.
539 * "systemctl kill" gained support for an optional --fail
540 switch. If specified the requested operation will fail of no
541 processes have been killed, because the unit had no
542 processes attached, or similar.
544 * A new systemd.crash_reboot=1 kernel command line option has
545 been added that triggers a reboot after crashing. This can
546 also be set through CrashReboot= in systemd.conf.
548 * The RuntimeDirectory= setting now understands unit
549 specifiers like %i or %f.
551 * A new (still internal) libary API sd-ipv4acd has been added,
552 that implements address conflict detection for IPv4. It's
553 based on code from sd-ipv4ll, and will be useful for
554 detecting DHCP address conflicts.
556 * File descriptors passed during socket activation may now be
557 named. A new API sd_listen_fds_with_names() is added to
558 access the names. The default names may be overridden,
559 either in the .socket file using the FileDescriptorName=
560 parameter, or by passing FDNAME= when storing the file
561 descriptors using sd_notify().
563 * systemd-networkd gained support for:
565 - Setting the IPv6 Router Advertisment settings via
566 IPv6AcceptRouterAdvertisements= in .network files.
568 - Configuring the HelloTimeSec=, MaxAgeSec= and
569 ForwardDelaySec= bridge parameters in .netdev files.
571 - Configuring PreferredSource= for static routes in
574 * The "ask-password" framework used to query for LUKS harddisk
575 passwords or SSL passwords during boot gained support for
576 caching passwords in the kernel keyring, if it is
577 available. This makes sure that the user only has to type in
578 a passphrase once if there are multiple objects to unlock
579 with the same one. Previously, such password caching was
580 available only when Plymouth was used; this moves the
581 caching logic into the systemd codebase itself. The
582 "systemd-ask-password" utility gained a new --keyname=
583 switch to control which kernel keyring key to use for
584 caching a password in. This functionality is also useful for
585 enabling display managers such as gdm to automatically
586 unlock the user's GNOME keyring if its passphrase, the
587 user's password and the harddisk password are the same, if
588 gdm-autologin is used.
590 * When downloading tar or raw images using "machinectl
591 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
592 file is now also downloaded, if it is available and stored
593 next to the image file.
595 * Units of type ".socket" gained a new boolean setting
596 Writable= which is only useful in conjunction with
597 ListenSpecial=. If true, enables opening the specified
598 special file in O_RDWR mode rather than O_RDONLY mode.
600 * systemd-rfkill has been reworked to become a singleton
601 service that is activated through /dev/rfkill on each rfkill
602 state change and saves the settings to disk. This way,
603 systemd-rfkill is now compatible with devices that exist
604 only intermittendly, and even restores state if the previous
605 system shutdown was abrupt rather than clean.
607 * The journal daemon gained support for vacuuming old journal
608 files controlled by the number of files that shall remain,
609 in addition to the already existing control by size and by
610 date. This is useful as journal interleaving performance
611 degrades with too many seperate journal files, and allows
612 putting an effective limit on them. The new setting defaults
613 to 100, but this may be changed by setting SystemMaxFiles=
614 and RuntimeMaxFiles= in journald.conf. Also, the
615 "journalctl" tool gained the new --vacuum-files= switch to
616 manually vacuum journal files to leave only the specified
617 number of files in place.
619 * udev will now create /dev/disk/by-path links for ATA devices
620 on kernels where that is supported.
622 * Galician, Serbian, Turkish and Korean translations were added.
624 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
625 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
626 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
627 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
628 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
629 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
630 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
631 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
632 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
633 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
634 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
635 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
636 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
637 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
638 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
639 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
640 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
641 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
643 -- Berlin, 2015-10-07
647 * The DHCP implementation of systemd-networkd gained a set of
650 - The DHCP server now supports emitting DNS and NTP
651 information. It may be enabled and configured via
652 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
653 and NTP information is enabled, but no servers are
654 configured, the corresponding uplink information (if there
655 is any) is propagated.
657 - Server and client now support transmission and reception
658 of timezone information. It can be configured via the
659 newly introduced network options UseTimezone=,
660 EmitTimezone=, and Timezone=. Transmission of timezone
661 information is enabled between host and containers by
662 default now: the container will change its local timezone
663 to what the host has set.
665 - Lease timeouts can now be configured via
666 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
668 - The DHCP server improved on the stability of
669 leases. Clients are more likely to get the same lease
670 information back, even if the server loses state.
672 - The DHCP server supports two new configuration options to
673 control the lease address pool metrics, PoolOffset= and
676 * The encapsulation limit of tunnels in systemd-networkd may
677 now be configured via 'EncapsulationLimit='. It allows
678 modifying the maximum additional levels of encapsulation
679 that are permitted to be prepended to a packet.
681 * systemd now supports the concept of user buses replacing
682 session buses, if used with dbus-1.10 (and enabled via dbus
683 --enable-user-session). It previously only supported this on
684 kdbus-enabled systems, and this release expands this to
685 'dbus-daemon' systems.
687 * systemd-networkd now supports predictable interface names
690 * systemd now optionally supports the new Linux kernel
691 "unified" control group hierarchy. If enabled via the kernel
692 command-line option 'systemd.unified_cgroup_hierarchy=1',
693 systemd will try to mount the unified cgroup hierarchy
694 directly on /sys/fs/cgroup. If not enabled, or not
695 available, systemd will fall back to the legacy cgroup
696 hierarchy setup, as before. Host system and containers can
697 mix and match legacy and unified hierarchies as they
698 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
699 environment variable to individually select the hierarchy to
700 use for executed containers. By default, nspawn will use the
701 unified hierarchy for the containers if the host uses the
702 unified hierarchy, and the legacy hierarchy otherwise.
703 Please note that at this point the unified hierarchy is an
704 experimental kernel feature and is likely to change in one
705 of the next kernel releases. Therefore, it should not be
706 enabled by default in downstream distributions yet. The
707 minimum required kernel version for the unified hierarchy to
708 work is 4.2. Note that when the unified hierarchy is used
709 for the first time delegated access to controllers is
710 safe. Because of this systemd-nspawn containers will get
711 access to controllers now, as will systemd user
712 sessions. This means containers and user sessions may now
713 manage their own resources, partitioning up what the system
716 * A new special scope unit "init.scope" has been introduced
717 that encapsulates PID 1 of the system. It may be used to
718 determine resource usage and enforce resource limits on PID
719 1 itself. PID 1 hence moved out of the root of the control
722 * The cgtop tool gained support for filtering out kernel
723 threads when counting tasks in a control group. Also, the
724 count of processes is now recursively summed up by
725 default. Two options -k and --recursive= have been added to
726 revert to old behaviour. The tool has also been updated to
727 work correctly in containers now.
729 * systemd-nspawn's --bind= and --bind-ro= options have been
730 extended to allow creation of non-recursive bind mounts.
732 * libsystemd gained two new calls sd_pid_get_cgroup() and
733 sd_peer_get_cgroup() which return the control group path of
734 a process or peer of a connected AF_UNIX socket. This
735 function call is particularly useful when implementing
736 delegated subtrees support in the control group hierarchy.
738 * The "sd-event" event loop API of libsystemd now supports
739 correct dequeuing of real-time signals, without losing
742 * When systemd requests a PolicyKit decision when managing
743 units it will now add additional fields to the request,
744 including unit name and desired operation. This enables more
745 powerful PolicyKit policies, that make decisions depending
748 * nspawn learnt support for .nspawn settings files, that may
749 accompany the image files or directories of containers, and
750 may contain additional settings for the container. This is
751 an alternative to configuring container parameters via the
754 Contributions from: Cristian Rodríguez, Daniel Mack, David
755 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
756 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
757 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
758 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
759 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
760 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
761 Andersen, Tom Gundersen, Torstein Husebø
763 -- Berlin, 2015-09-08
767 * machinectl gained a new verb 'shell' which opens a fresh
768 shell on the target container or the host. It is similar to
769 the existing 'login' command of machinectl, but spawns the
770 shell directly without prompting for username or
771 password. The pseudo machine '.host' now refers to the local
772 host and is used by default. Hence, 'machinectl shell' can
773 be used as replacement for 'su -' which spawns a session as
774 a fresh systemd unit in a way that is fully isolated from
775 the originating session.
777 * systemd-networkd learned to cope with private-zone DHCP
778 options and allows other programs to query the values.
780 * SELinux access control when enabling/disabling units is no
781 longer enforced with this release. The previous
782 implementation was incorrect, and a new corrected
783 implementation is not yet available. As unit file operations
784 are still protected via PolicyKit and D-Bus policy this is
785 not a security problem. Yet, distributions which care about
786 optimal SELinux support should probably not stabilize on
789 * sd-bus gained support for matches of type "arg0has=", that
790 test for membership of strings in string arrays sent in bus
793 * systemd-resolved now dumps the contents of its DNS and LLMNR
794 caches to the logs on reception of the SIGUSR1 signal. This
795 is useful to debug DNS behaviour.
797 * The coredumpctl tool gained a new --directory= option to
798 operate on journal files in a specific directory.
800 * "systemctl reboot" and related commands gained a new
801 "--message=" option which may be used to set a free-text
802 wall message when shutting down or rebooting the
803 system. This message is also logged, which is useful for
804 figuring out the reason for a reboot or shutdown a
807 * The "systemd-resolve-host" tool's -i switch now takes
808 network interface numbers as alternative to interface names.
810 * A new unit file setting for services has been introduced:
811 UtmpMode= allows configuration of how precisely systemd
812 handles utmp and wtmp entries for the service if this is
813 enabled. This allows writing services that appear similar to
814 user sessions in the output of the "w", "who", "last" and
817 * systemd-resolved will now locally synthesize DNS resource
818 records for the "localhost" and "gateway" domains as well as
819 the local hostname. This should ensure that clients querying
820 RRs via resolved will get similar results as those going via
821 NSS, if nss-myhostname is enabled.
823 Contributions from: Alastair Hughes, Alex Crawford, Daniel
824 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
825 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
826 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
827 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
828 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
829 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
830 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
831 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
832 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
833 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
834 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
836 -- Berlin, 2015-08-27
840 * The systemd-efi-boot-generator functionality was merged into
841 systemd-gpt-auto-generator.
843 * systemd-networkd now supports Group Policy for vxlan
844 devices. It can be enabled via the new boolean configuration
845 option called 'GroupPolicyExtension='.
847 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
848 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
849 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
851 -- Berlin, 2015-07-31
855 * The python-systemd code has been removed from the systemd repository.
856 A new repository has been created which accommodates the code from
857 now on, and we kindly ask distributions to create a separate package
858 for this: https://github.com/systemd/python-systemd
860 * The systemd daemon will now reload its main configuration
861 (/etc/systemd/system.conf) on daemon-reload.
863 * sd-dhcp now exposes vendor specific extensions via
864 sd_dhcp_lease_get_vendor_specific().
866 * systemd-networkd gained a number of new configuration options.
868 - A new boolean configuration option for TAP devices called
869 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
870 device, thus allowing to send and receive GSO packets.
872 - A new tunnel configuration option called 'CopyDSCP='.
873 If enabled, the DSCP field of ip6 tunnels is copied into the
876 - A set of boolean bridge configuration options were added.
877 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
878 and 'UnicastFlood=' are now parsed by networkd and applied to the
879 respective bridge link device via the respective IFLA_BRPORT_*
882 - A new string configuration option to override the hostname sent
883 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
884 is true, networkd will use the configured hostname instead of the
885 system hostname when sending DHCP requests.
887 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
888 networkd will configure the IPv6 flow-label of the tunnel device
889 according to RFC2460.
891 - The 'macvtap' virtual network devices are now supported, similar to
892 the already supported 'macvlan' devices.
894 * systemd-resolved now implements RFC5452 to improve resilience against
895 cache poisoning. Additionally, source port randomization is enabled
896 by default to further protect against DNS spoofing attacks.
898 * nss-mymachines now supports translating UIDs and GIDs of running
899 containers with user-namespaces enabled. If a container 'foo'
900 translates a host uid 'UID' to the container uid 'TUID', then
901 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
902 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
903 mapped as 'vg-foo-TGID'.
905 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
906 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
907 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
908 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
909 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
910 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
911 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
912 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
913 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
914 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
916 -- Berlin, 2015-07-29
920 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
921 There are no known issues with current sysfs, and udev does not need
922 or should be used to work around such bugs.
924 * udev does no longer enable USB HID power management. Several reports
925 indicate, that some devices cannot handle that setting.
927 * The udev accelerometer helper was removed. The functionality
928 is now fully included in iio-sensor-proxy. But this means,
929 older iio-sensor-proxy versions will no longer provide
930 accelerometer/orientation data with this systemd version.
931 Please upgrade iio-sensor-proxy to version 1.0.
933 * networkd gained a new configuration option IPv6PrivacyExtensions=
934 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
935 for Stateless Address") on selected networks.
937 * For the sake of fewer build-time dependencies and less code in the
938 main repository, the python bindings are about to be removed in the
939 next release. A new repository has been created which accommodates
940 the code from now on, and we kindly ask distributions to create a
941 separate package for this. The removal will take place in v223.
943 https://github.com/systemd/python-systemd
945 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
946 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
947 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
948 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
949 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
950 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
951 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
952 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
953 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
954 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
956 -- Berlin, 2015-07-07
960 * The sd-bus.h and sd-event.h APIs have now been declared
961 stable and have been added to the official interface of
962 libsystemd.so. sd-bus implements an alternative D-Bus client
963 library, that is relatively easy to use, very efficient and
964 supports both classic D-Bus as well as kdbus as transport
965 backend. sd-event is a generic event loop abstraction that
966 is built around Linux epoll, but adds features such as event
967 prioritization or efficient timer handling. Both APIs are good
968 choices for C programs looking for a bus and/or event loop
969 implementation that is minimal and does not have to be
970 portable to other kernels.
972 * kdbus support is no longer compile-time optional. It is now
973 always built-in. However, it can still be disabled at
974 runtime using the kdbus=0 kernel command line setting, and
975 that setting may be changed to default to off, by specifying
976 --disable-kdbus at build-time. Note though that the kernel
977 command line setting has no effect if the kdbus.ko kernel
978 module is not installed, in which case kdbus is (obviously)
979 also disabled. We encourage all downstream distributions to
980 begin testing kdbus by adding it to the kernel images in the
981 development distributions, and leaving kdbus support in
984 * The minimal required util-linux version has been bumped to
987 * Support for chkconfig (--enable-chkconfig) was removed in
988 favor of calling an abstraction tool
989 /lib/systemd/systemd-sysv-install. This needs to be
990 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
991 in README for details.
993 * If there's a systemd unit and a SysV init script for the
994 same service name, and the user executes "systemctl enable"
995 for it (or a related call), then this will now enable both
996 (or execute the related operation on both), not just the
999 * The libudev API documentation has been converted from gtkdoc
1002 * gudev has been removed from the systemd tree, it is now an
1005 * The systemd-cgtop tool learnt a new --raw switch to generate
1006 "raw" (machine parsable) output.
1008 * networkd's IPForwarding= .network file setting learnt the
1009 new setting "kernel", which ensures that networkd does not
1010 change the IP forwarding sysctl from the default kernel
1013 * The systemd-logind bus API now exposes a new boolean
1014 property "Docked" that reports whether logind considers the
1015 system "docked", i.e. connected to a docking station or not.
1017 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1018 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1019 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1020 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1021 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1022 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1023 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1024 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1025 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1026 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1027 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1028 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1029 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1030 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1031 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1032 Fink, Zbigniew Jędrzejewski-Szmek
1034 -- Berlin, 2015-06-19
1038 * The gudev library has been extracted into a separate repository
1039 available at: https://git.gnome.org/browse/libgudev/
1040 It is now managed as part of the Gnome project. Distributions
1041 are recommended to pass --disable-gudev to systemd and use
1042 gudev from the Gnome project instead. gudev is still included
1043 in systemd, for now. It will be removed soon, though. Please
1044 also see the announcement-thread on systemd-devel:
1045 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1047 * systemd now exposes a CPUUsageNSec= property for each
1048 service unit on the bus, that contains the overall consumed
1049 CPU time of a service (the sum of what each process of the
1050 service consumed). This value is only available if
1051 CPUAccounting= is turned on for a service, and is then shown
1052 in the "systemctl status" output.
1054 * Support for configuring alternative mappings of the old SysV
1055 runlevels to systemd targets has been removed. They are now
1056 hardcoded in a way that runlevels 2, 3, 4 all map to
1057 multi-user.target and 5 to graphical.target (which
1058 previously was already the default behaviour).
1060 * The auto-mounter logic gained support for mount point
1061 expiry, using a new TimeoutIdleSec= setting in .automount
1062 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1064 * The EFI System Partition (ESP) as mounted to /boot by
1065 systemd-efi-boot-generator will now be unmounted
1066 automatically after 2 minutes of not being used. This should
1067 minimize the risk of ESP corruptions.
1069 * New /etc/fstab options x-systemd.requires= and
1070 x-systemd.requires-mounts-for= are now supported to express
1071 additional dependencies for mounts. This is useful for
1072 journalling file systems that support external journal
1073 devices or overlay file systems that require underlying file
1074 systems to be mounted.
1076 * systemd does not support direct live-upgrades (via systemctl
1077 daemon-reexec) from versions older than v44 anymore. As no
1078 distribution we are aware of shipped such old versions in a
1079 stable release this should not be problematic.
1081 * When systemd forks off a new per-connection service instance
1082 it will now set the $REMOTE_ADDR environment variable to the
1083 remote IP address, and $REMOTE_PORT environment variable to
1084 the remote IP port. This behaviour is similar to the
1085 corresponding environment variables defined by CGI.
1087 * systemd-networkd gained support for uplink failure
1088 detection. The BindCarrier= option allows binding interface
1089 configuration dynamically to the link sense of other
1090 interfaces. This is useful to achieve behaviour like in
1093 * systemd-networkd gained support for configuring the DHCP
1094 client identifier to use when requesting leases.
1096 * systemd-networkd now has a per-network UseNTP= option to
1097 configure whether NTP server information acquired via DHCP
1098 is passed on to services like systemd-timesyncd.
1100 * systemd-networkd gained support for vti6 tunnels.
1102 * Note that systemd-networkd manages the sysctl variable
1103 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1104 it is configured for since v219. The variable controls IP
1105 forwarding, and is a per-interface alternative to the global
1106 /proc/sys/net/ipv[46]/ip_forward. This setting is
1107 configurable in the IPForward= option, which defaults to
1108 "no". This means if networkd is used for an interface it is
1109 no longer sufficient to set the global sysctl option to turn
1110 on IP forwarding! Instead, the .network file option
1111 IPForward= needs to be turned on! Note that the
1112 implementation of this behaviour was broken in v219 and has
1115 * Many bonding and vxlan options are now configurable in
1118 * systemd-nspawn gained a new --property= setting to set unit
1119 properties for the container scope. This is useful for
1120 setting resource parameters (e.g "CPUShares=500") on
1121 containers started from the command line.
1123 * systemd-nspawn gained a new --private-users= switch to make
1124 use of user namespacing available on recent Linux kernels.
1126 * systemd-nspawn may now be called as part of a shell pipeline
1127 in which case the pipes used for stdin and stdout are passed
1128 directly to the process invoked in the container, without
1129 indirection via a pseudo tty.
1131 * systemd-nspawn gained a new switch to control the UNIX
1132 signal to use when killing the init process of the container
1135 * systemd-nspawn gained a new --overlay= switch for mounting
1136 overlay file systems into the container using the new kernel
1139 * When a container image is imported via systemd-importd and
1140 the host file system is not btrfs, a loopback block device
1141 file is created in /var/lib/machines.raw with a btrfs file
1142 system inside. It is then mounted to /var/lib/machines to
1143 enable btrfs features for container management. The loopback
1144 file and btrfs file system is grown as needed when container
1145 images are imported via systemd-importd.
1147 * systemd-machined/systemd-importd gained support for btrfs
1148 quota, to enforce container disk space limits on disk. This
1149 is exposed in "machinectl set-limit".
1151 * systemd-importd now can import containers from local .tar,
1152 .raw and .qcow2 images, and export them to .tar and .raw. It
1153 can also import dkr v2 images now from the network (on top
1156 * systemd-importd gained support for verifying downloaded
1157 images with gpg2 (previously only gpg1 was supported).
1159 * systemd-machined, systemd-logind, systemd: most bus calls
1160 are now accessible to unprivileged processes via
1161 PolicyKit. Also, systemd-logind will now allow users to kill
1162 their own sessions without further privileges or
1165 * systemd-shutdownd has been removed. This service was
1166 previously responsible for implementing scheduled shutdowns
1167 as exposed in /usr/bin/shutdown's time parameter. This
1168 functionality has now been moved into systemd-logind and is
1169 accessible via a bus interface.
1171 * "systemctl reboot" gained a new switch --firmware-setup that
1172 can be used to reboot into the EFI firmware setup, if that
1173 is available. systemd-logind now exposes an API on the bus
1174 to trigger such reboots, in case graphical desktop UIs want
1175 to cover this functionality.
1177 * "systemctl enable", "systemctl disable" and "systemctl mask"
1178 now support a new "--now" switch. If specified the units
1179 that are enabled will also be started, and the ones
1180 disabled/masked also stopped.
1182 * The Gummiboot EFI boot loader tool has been merged into
1183 systemd, and renamed to "systemd-boot". The bootctl tool has been
1184 updated to support systemd-boot.
1186 * An EFI kernel stub has been added that may be used to create
1187 kernel EFI binaries that contain not only the actual kernel,
1188 but also an initrd, boot splash, command line and OS release
1189 information. This combined binary can then be signed as a
1190 single image, so that the firmware can verify it all in one
1191 step. systemd-boot has special support for EFI binaries created
1192 like this and can extract OS release information from them
1193 and show them in the boot menu. This functionality is useful
1194 to implement cryptographically verified boot schemes.
1196 * Optional support has been added to systemd-fsck to pass
1197 fsck's progress report to an AF_UNIX socket in the file
1200 * udev will no longer create device symlinks for all block
1201 devices by default. A blacklist for excluding special block
1202 devices from this logic has been turned into a whitelist
1203 that requires picking block devices explicitly that require
1206 * A new (currently still internal) API sd-device.h has been
1207 added to libsystemd. This modernized API is supposed to
1208 replace libudev eventually. In fact, already much of libudev
1209 is now just a wrapper around sd-device.h.
1211 * A new hwdb database for storing metadata about pointing
1212 stick devices has been added.
1214 * systemd-tmpfiles gained support for setting file attributes
1215 similar to the "chattr" tool with new 'h' and 'H' lines.
1217 * systemd-journald will no longer unconditionally set the
1218 btrfs NOCOW flag on new journal files. This is instead done
1219 with tmpfiles snippet using the new 'h' line type. This
1220 allows easy disabling of this logic, by masking the
1221 journal-nocow.conf tmpfiles file.
1223 * systemd-journald will now translate audit message types to
1224 human readable identifiers when writing them to the
1225 journal. This should improve readability of audit messages.
1227 * The LUKS logic gained support for the offset= and skip=
1228 options in /etc/crypttab, as previously implemented by
1231 * /usr/lib/os-release gained a new optional field VARIANT= for
1232 distributions that support multiple variants (such as a
1233 desktop edition, a server edition, ...)
1235 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1236 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1237 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1238 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1239 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1240 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1241 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1242 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1243 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1244 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1245 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1246 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1247 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1248 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1249 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1250 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1251 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1252 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1253 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1254 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1255 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1256 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1257 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1258 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1259 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1260 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1261 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1263 -- Berlin, 2015-05-22
1267 * Introduce a new API "sd-hwdb.h" for querying the hardware
1268 metadata database. With this minimal interface one can query
1269 and enumerate the udev hwdb, decoupled from the old libudev
1270 library. libudev's interface for this is now only a wrapper
1271 around sd-hwdb. A new tool systemd-hwdb has been added to
1272 interface with and update the database.
1274 * When any of systemd's tools copies files (for example due to
1275 tmpfiles' C lines) a btrfs reflink will attempted first,
1276 before bytewise copying is done.
1278 * systemd-nspawn gained a new --ephemeral switch. When
1279 specified a btrfs snapshot is taken of the container's root
1280 directory, and immediately removed when the container
1281 terminates again. Thus, a container can be started whose
1282 changes never alter the container's root directory, and are
1283 lost on container termination. This switch can also be used
1284 for starting a container off the root file system of the
1285 host without affecting the host OS. This switch is only
1286 available on btrfs file systems.
1288 * systemd-nspawn gained a new --template= switch. It takes the
1289 path to a container tree to use as template for the tree
1290 specified via --directory=, should that directory be
1291 missing. This allows instantiating containers dynamically,
1292 on first run. This switch is only available on btrfs file
1295 * When a .mount unit refers to a mount point on which multiple
1296 mounts are stacked, and the .mount unit is stopped all of
1297 the stacked mount points will now be unmounted until no
1298 mount point remains.
1300 * systemd now has an explicit notion of supported and
1301 unsupported unit types. Jobs enqueued for unsupported unit
1302 types will now fail with an "unsupported" error code. More
1303 specifically .swap, .automount and .device units are not
1304 supported in containers, .busname units are not supported on
1305 non-kdbus systems. .swap and .automount are also not
1306 supported if their respective kernel compile time options
1309 * machinectl gained support for two new "copy-from" and
1310 "copy-to" commands for copying files from a running
1311 container to the host or vice versa.
1313 * machinectl gained support for a new "bind" command to bind
1314 mount host directories into local containers. This is
1315 currently only supported for nspawn containers.
1317 * networkd gained support for configuring bridge forwarding
1318 database entries (fdb) from .network files.
1320 * A new tiny daemon "systemd-importd" has been added that can
1321 download container images in tar, raw, qcow2 or dkr formats,
1322 and make them available locally in /var/lib/machines, so
1323 that they can run as nspawn containers. The daemon can GPG
1324 verify the downloads (not supported for dkr, since it has no
1325 provisions for verifying downloads). It will transparently
1326 decompress bz2, xz, gzip compressed downloads if necessary,
1327 and restore sparse files on disk. The daemon uses privilege
1328 separation to ensure the actual download logic runs with
1329 fewer privileges than the daemon itself. machinectl has
1330 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1331 make the functionality of importd available to the
1332 user. With this in place the Fedora and Ubuntu "Cloud"
1333 images can be downloaded and booted as containers unmodified
1334 (the Fedora images lack the appropriate GPG signature files
1335 currently, so they cannot be verified, but this will change
1336 soon, hopefully). Note that downloading images is currently
1337 only fully supported on btrfs.
1339 * machinectl is now able to list container images found in
1340 /var/lib/machines, along with some metadata about sizes of
1341 disk and similar. If the directory is located on btrfs and
1342 quota is enabled, this includes quota display. A new command
1343 "image-status" has been added that shows additional
1344 information about images.
1346 * machinectl is now able to clone container images
1347 efficiently, if the underlying file system (btrfs) supports
1348 it, with the new "machinectl list-images" command. It also
1349 gained commands for renaming and removing images, as well as
1350 marking them read-only or read-write (supported also on
1351 legacy file systems).
1353 * networkd gained support for collecting LLDP network
1354 announcements, from hardware that supports this. This is
1355 shown in networkctl output.
1357 * systemd-run gained support for a new -t (--pty) switch for
1358 invoking a binary on a pty whose input and output is
1359 connected to the invoking terminal. This allows executing
1360 processes as system services while interactively
1361 communicating with them via the terminal. Most interestingly
1362 this is supported across container boundaries. Invoking
1363 "systemd-run -t /bin/bash" is an alternative to running a
1364 full login session, the difference being that the former
1365 will not register a session, nor go through the PAM session
1368 * tmpfiles gained support for a new "v" line type for creating
1369 btrfs subvolumes. If the underlying file system is a legacy
1370 file system, this automatically degrades to creating a
1371 normal directory. Among others /var/lib/machines is now
1372 created like this at boot, should it be missing.
1374 * The directory /var/lib/containers/ has been deprecated and
1375 been replaced by /var/lib/machines. The term "machines" has
1376 been used in the systemd context as generic term for both
1377 VMs and containers, and hence appears more appropriate for
1378 this, as the directory can also contain raw images bootable
1381 * systemd-nspawn when invoked with -M but without --directory=
1382 or --image= is now capable of searching for the container
1383 root directory, subvolume or disk image automatically, in
1384 /var/lib/machines. systemd-nspawn@.service has been updated
1385 to make use of this, thus allowing it to be used for raw
1388 * A new machines.target unit has been introduced that is
1389 supposed to group all containers/VMs invoked as services on
1390 the system. systemd-nspawn@.service has been updated to
1391 integrate with that.
1393 * machinectl gained a new "start" command, for invoking a
1394 container as a service. "machinectl start foo" is mostly
1395 equivalent to "systemctl start systemd-nspawn@foo.service",
1396 but handles escaping in a nicer way.
1398 * systemd-nspawn will now mount most of the cgroupfs tree
1399 read-only into each container, with the exception of the
1400 container's own subtree in the name=systemd hierarchy.
1402 * journald now sets the special FS_NOCOW file flag for its
1403 journal files. This should improve performance on btrfs, by
1404 avoiding heavy fragmentation when journald's write-pattern
1405 is used on COW file systems. It degrades btrfs' data
1406 integrity guarantees for the files to the same levels as for
1407 ext3/ext4 however. This should be OK though as journald does
1408 its own data integrity checks and all its objects are
1409 checksummed on disk. Also, journald should handle btrfs disk
1410 full events a lot more gracefully now, by processing SIGBUS
1411 errors, and not relying on fallocate() anymore.
1413 * When journald detects that journal files it is writing to
1414 have been deleted it will immediately start new journal
1417 * systemd now provides a way to store file descriptors
1418 per-service in PID 1.This is useful for daemons to ensure
1419 that fds they require are not lost during a daemon
1420 restart. The fds are passed to the daemon on the next
1421 invocation in the same way socket activation fds are
1422 passed. This is now used by journald to ensure that the
1423 various sockets connected to all the system's stdout/stderr
1424 are not lost when journald is restarted. File descriptors
1425 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1426 an extension to sd_notify(). Note that a limit is enforced
1427 on the number of fds a service can store in PID 1, and it
1428 defaults to 0, so that no fds may be stored, unless this is
1429 explicitly turned on.
1431 * The default TERM variable to use for units connected to a
1432 terminal, when no other value is explicitly is set is now
1433 vt220 rather than vt102. This should be fairly safe still,
1434 but allows PgUp/PgDn work.
1436 * The /etc/crypttab option header= as known from Debian is now
1439 * "loginctl user-status" and "loginctl session-status" will
1440 now show the last 10 lines of log messages of the
1441 user/session following the status output. Similar,
1442 "machinectl status" will show the last 10 log lines
1443 associated with a virtual machine or container
1444 service. (Note that this is usually not the log messages
1445 done in the VM/container itself, but simply what the
1446 container manager logs. For nspawn this includes all console
1449 * "loginctl session-status" without further argument will now
1450 show the status of the session of the caller. Similar,
1451 "lock-session", "unlock-session", "activate",
1452 "enable-linger", "disable-linger" may now be called without
1453 session/user parameter in which case they apply to the
1454 caller's session/user.
1456 * An X11 session scriptlet is now shipped that uploads
1457 $DISPLAY and $XAUTHORITY into the environment of the systemd
1458 --user daemon if a session begins. This should improve
1459 compatibility with X11 enabled applications run as systemd
1462 * Generators are now subject to masking via /etc and /run, the
1463 same way as unit files.
1465 * networkd .network files gained support for configuring
1466 per-link IPv4/IPv6 packet forwarding as well as IPv4
1467 masquerading. This is by default turned on for veth links to
1468 containers, as registered by systemd-nspawn. This means that
1469 nspawn containers run with --network-veth will now get
1470 automatic routed access to the host's networks without any
1471 further configuration or setup, as long as networkd runs on
1474 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1475 or UDP posts of a container on the host. With this in place
1476 it is possible to run containers with private veth links
1477 (--network-veth), and have their functionality exposed on
1478 the host as if their services were running directly on the
1481 * systemd-nspawn's --network-veth switch now gained a short
1482 version "-n", since with the changes above it is now truly
1483 useful out-of-the-box. The systemd-nspawn@.service has been
1484 updated to make use of it too by default.
1486 * systemd-nspawn will now maintain a per-image R/W lock, to
1487 ensure that the same image is not started more than once
1488 writable. (It's OK to run an image multiple times
1489 simultaneously in read-only mode.)
1491 * systemd-nspawn's --image= option is now capable of
1492 dissecting and booting MBR and GPT disk images that contain
1493 only a single active Linux partition. Previously it
1494 supported only GPT disk images with proper GPT type
1495 IDs. This allows running cloud images from major
1496 distributions directly with systemd-nspawn, without
1499 * In addition to collecting mouse dpi data in the udev
1500 hardware database, there's now support for collecting angle
1501 information for mouse scroll wheels. The database is
1502 supposed to guarantee similar scrolling behavior on mice
1503 that it knows about. There's also support for collecting
1504 information about Touchpad types.
1506 * udev's input_id built-in will now also collect touch screen
1507 dimension data and attach it to probed devices.
1509 * /etc/os-release gained support for a Distribution Privacy
1512 * networkd gained support for creating "ipvlan", "gretap",
1513 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1515 * systemd-tmpfiles gained support for "a" lines for setting
1518 * systemd-nspawn will now mount /tmp in the container to
1519 tmpfs, automatically.
1521 * systemd now exposes the memory.usage_in_bytes cgroup
1522 attribute and shows it for each service in the "systemctl
1523 status" output, if available.
1525 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1526 immediate reboot is triggered. This useful if shutdown is
1527 hung and is unable to complete, to expedite the
1528 operation. Note that this kind of reboot will still unmount
1529 all file systems, and hence should not result in fsck being
1532 * A .device unit for an optical block device will now be
1533 considered active only when a medium is in the drive. Also,
1534 mount units are now bound to their backing devices thus
1535 triggering automatic unmounting when devices become
1536 unavailable. With this in place systemd will now
1537 automatically unmount left-over mounts when a CD-ROM is
1538 ejected or an USB stick is yanked from the system.
1540 * networkd-wait-online now has support for waiting for
1541 specific interfaces only (with globbing), and for giving up
1542 after a configurable timeout.
1544 * networkd now exits when idle. It will be automatically
1545 restarted as soon as interfaces show up, are removed or
1546 change state. networkd will stay around as long as there is
1547 at least one DHCP state machine or similar around, that keep
1550 * networkd may now configure IPv6 link-local addressing in
1551 addition to IPv4 link-local addressing.
1553 * The IPv6 "token" for use in SLAAC may now be configured for
1554 each .network interface in networkd.
1556 * Routes configured with networkd may now be assigned a scope
1559 * networkd's [Match] sections now support globbing and lists
1560 of multiple space-separated matches per item.
1562 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1563 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1564 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1565 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1566 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1567 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1568 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1569 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1570 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1571 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1572 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1573 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1574 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1575 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1576 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1577 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1578 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1579 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1580 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1581 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1582 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1583 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1584 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1585 Hoffmann, Zbigniew Jędrzejewski-Szmek
1587 -- Berlin, 2015-02-16
1591 * When querying unit file enablement status (for example via
1592 "systemctl is-enabled"), a new state "indirect" is now known
1593 which indicates that a unit might not be enabled itself, but
1594 another unit listed in its Also= setting might be.
1596 * Similar to the various existing ConditionXYZ= settings for
1597 units, there are now matching AssertXYZ= settings. While
1598 failing conditions cause a unit to be skipped, but its job
1599 to succeed, failing assertions declared like this will cause
1600 a unit start operation and its job to fail.
1602 * hostnamed now knows a new chassis type "embedded".
1604 * systemctl gained a new "edit" command. When used on a unit
1605 file, this allows extending unit files with .d/ drop-in
1606 configuration snippets or editing the full file (after
1607 copying it from /usr/lib to /etc). This will invoke the
1608 user's editor (as configured with $EDITOR), and reload the
1609 modified configuration after editing.
1611 * "systemctl status" now shows the suggested enablement state
1612 for a unit, as declared in the (usually vendor-supplied)
1613 system preset files.
1615 * nss-myhostname will now resolve the single-label host name
1616 "gateway" to the locally configured default IP routing
1617 gateways, ordered by their metrics. This assigns a stable
1618 name to the used gateways, regardless which ones are
1619 currently configured. Note that the name will only be
1620 resolved after all other name sources (if nss-myhostname is
1621 configured properly) and should hence not negatively impact
1622 systems that use the single-label host name "gateway" in
1625 * systemd-inhibit now allows filtering by mode when listing
1628 * Scope and service units gained a new "Delegate" boolean
1629 property, which, when set, allows processes running inside the
1630 unit to further partition resources. This is primarily
1631 useful for systemd user instances as well as container
1634 * journald will now pick up audit messages directly from
1635 the kernel, and log them like any other log message. The
1636 audit fields are split up and fully indexed. This means that
1637 journalctl in many ways is now a (nicer!) alternative to
1638 ausearch, the traditional audit client. Note that this
1639 implements only a minimal audit client. If you want the
1640 special audit modes like reboot-on-log-overflow, please use
1641 the traditional auditd instead, which can be used in
1642 parallel to journald.
1644 * The ConditionSecurity= unit file option now understands the
1645 special string "audit" to check whether auditing is
1648 * journalctl gained two new commands --vacuum-size= and
1649 --vacuum-time= to delete old journal files until the
1650 remaining ones take up no more than the specified size on disk,
1651 or are not older than the specified time.
1653 * A new, native PPPoE library has been added to sd-network,
1654 systemd's library of light-weight networking protocols. This
1655 library will be used in a future version of networkd to
1656 enable PPPoE communication without an external pppd daemon.
1658 * The busctl tool now understands a new "capture" verb that
1659 works similar to "monitor", but writes a packet capture
1660 trace to STDOUT that can be redirected to a file which is
1661 compatible with libcap's capture file format. This can then
1662 be loaded in Wireshark and similar tools to inspect bus
1665 * The busctl tool now understands a new "tree" verb that shows
1666 the object trees of a specific service on the bus, or of all
1669 * The busctl tool now understands a new "introspect" verb that
1670 shows all interfaces and members of objects on the bus,
1671 including their signature and values. This is particularly
1672 useful to get more information about bus objects shown by
1673 the new "busctl tree" command.
1675 * The busctl tool now understands new verbs "call",
1676 "set-property" and "get-property" for invoking bus method
1677 calls, setting and getting bus object properties in a
1680 * busctl gained a new --augment-creds= argument that controls
1681 whether the tool shall augment credential information it
1682 gets from the bus with data from /proc, in a possibly
1685 * nspawn's --link-journal= switch gained two new values
1686 "try-guest" and "try-host" that work like "guest" and
1687 "host", but do not fail if the host has no persistent
1688 journalling enabled. -j is now equivalent to
1689 --link-journal=try-guest.
1691 * macvlan network devices created by nspawn will now have
1692 stable MAC addresses.
1694 * A new SmackProcessLabel= unit setting has been added, which
1695 controls the SMACK security label processes forked off by
1696 the respective unit shall use.
1698 * If compiled with --enable-xkbcommon, systemd-localed will
1699 verify x11 keymap settings by compiling the given keymap. It
1700 will spew out warnings if the compilation fails. This
1701 requires libxkbcommon to be installed.
1703 * When a coredump is collected, a larger number of metadata
1704 fields is now collected and included in the journal records
1705 created for it. More specifically, control group membership,
1706 environment variables, memory maps, working directory,
1707 chroot directory, /proc/$PID/status, and a list of open file
1708 descriptors is now stored in the log entry.
1710 * The udev hwdb now contains DPI information for mice. For
1713 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1715 * All systemd programs that read standalone configuration
1716 files in /etc now also support a corresponding series of
1717 .conf.d configuration directories in /etc/, /run/,
1718 /usr/local/lib/, /usr/lib/, and (if configured with
1719 --enable-split-usr) /lib/. In particular, the following
1720 configuration files now have corresponding configuration
1721 directories: system.conf user.conf, logind.conf,
1722 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1723 resolved.conf, timesyncd.conf, journal-remote.conf, and
1724 journal-upload.conf. Note that distributions should use the
1725 configuration directories in /usr/lib/; the directories in
1726 /etc/ are reserved for the system administrator.
1728 * systemd-rfkill will no longer take the rfkill device name
1729 into account when storing rfkill state on disk, as the name
1730 might be dynamically assigned and not stable. Instead, the
1731 ID_PATH udev variable combined with the rfkill type (wlan,
1732 bluetooth, ...) is used.
1734 * A new service systemd-machine-id-commit.service has been
1735 added. When used on systems where /etc is read-only during
1736 boot, and /etc/machine-id is not initialized (but an empty
1737 file), this service will copy the temporary machine ID
1738 created as replacement into /etc after the system is fully
1739 booted up. This is useful for systems that are freshly
1740 installed with a non-initialized machine ID, but should get
1741 a fixed machine ID for subsequent boots.
1743 * networkd's .netdev files now provide a large set of
1744 configuration parameters for VXLAN devices. Similarly, the
1745 bridge port cost parameter is now configurable in .network
1746 files. There's also new support for configuring IP source
1747 routing. networkd .link files gained support for a new
1748 OriginalName= match that is useful to match against the
1749 original interface name the kernel assigned. .network files
1750 may include MTU= and MACAddress= fields for altering the MTU
1751 and MAC address while being connected to a specific network
1754 * The LUKS logic gained supported for configuring
1755 UUID-specific key files. There's also new support for naming
1756 LUKS device from the kernel command line, using the new
1757 luks.name= argument.
1759 * Timer units may now be transiently created via the bus API
1760 (this was previously already available for scope and service
1761 units). In addition it is now possible to create multiple
1762 transient units at the same time with a single bus call. The
1763 "systemd-run" tool has been updated to make use of this for
1764 running commands on a specified time, in at(1)-style.
1766 * tmpfiles gained support for "t" lines, for assigning
1767 extended attributes to files. Among other uses this may be
1768 used to assign SMACK labels to files.
1770 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1771 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1772 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1773 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1774 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1775 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1776 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1777 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1778 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1779 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1780 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1781 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1782 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1783 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1784 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1785 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1786 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1787 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1789 -- Berlin, 2014-12-10
1793 * journalctl gained the new options -t/--identifier= to match
1794 on the syslog identifier (aka "tag"), as well as --utc to
1795 show log timestamps in the UTC timezone. journalctl now also
1796 accepts -n/--lines=all to disable line capping in a pager.
1798 * journalctl gained a new switch, --flush, that synchronously
1799 flushes logs from /run/log/journal to /var/log/journal if
1800 persistent storage is enabled. systemd-journal-flush.service
1801 now waits until the operation is complete.
1803 * Services can notify the manager before they start a reload
1804 (by sending RELOADING=1) or shutdown (by sending
1805 STOPPING=1). This allows the manager to track and show the
1806 internal state of daemons and closes a race condition when
1807 the process is still running but has closed its D-Bus
1810 * Services with Type=oneshot do not have to have any ExecStart
1813 * User units are now loaded also from
1814 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1815 /run/systemd/user directory that was already previously
1816 supported, but is under the control of the user.
1818 * Job timeouts (i.e. time-outs on the time a job that is
1819 queued stays in the run queue) can now optionally result in
1820 immediate reboot or power-off actions (JobTimeoutAction= and
1821 JobTimeoutRebootArgument=). This is useful on ".target"
1822 units, to limit the maximum time a target remains
1823 undispatched in the run queue, and to trigger an emergency
1824 operation in such a case. This is now used by default to
1825 turn off the system if boot-up (as defined by everything in
1826 basic.target) hangs and does not complete for at least
1827 15min. Also, if power-off or reboot hang for at least 30min
1828 an immediate power-off/reboot operation is triggered. This
1829 functionality is particularly useful to increase reliability
1830 on embedded devices, but also on laptops which might
1831 accidentally get powered on when carried in a backpack and
1832 whose boot stays stuck in a hard disk encryption passphrase
1835 * systemd-logind can be configured to also handle lid switch
1836 events even when the machine is docked or multiple displays
1837 are attached (HandleLidSwitchDocked= option).
1839 * A helper binary and a service have been added which can be
1840 used to resume from hibernation in the initramfs. A
1841 generator will parse the resume= option on the kernel
1842 command line to trigger resume.
1844 * A user console daemon systemd-consoled has been
1845 added. Currently, it is a preview, and will so far open a
1846 single terminal on each session of the user marked as
1847 Desktop=systemd-console.
1849 * Route metrics can be specified for DHCP routes added by
1852 * The SELinux context of socket-activated services can be set
1853 from the information provided by the networking stack
1854 (SELinuxContextFromNet= option).
1856 * Userspace firmware loading support has been removed and
1857 the minimum supported kernel version is thus bumped to 3.7.
1859 * Timeout for udev workers has been increased from 1 to 3
1860 minutes, but a warning will be printed after 1 minute to
1861 help diagnose kernel modules that take a long time to load.
1863 * Udev rules can now remove tags on devices with TAG-="foobar".
1865 * systemd's readahead implementation has been removed. In many
1866 circumstances it didn't give expected benefits even for
1867 rotational disk drives and was becoming less relevant in the
1868 age of SSDs. As none of the developers has been using
1869 rotating media anymore, and nobody stepped up to actively
1870 maintain this component of systemd it has now been removed.
1872 * Swap units can use Options= to specify discard options.
1873 Discard options specified for swaps in /etc/fstab are now
1876 * Docker containers are now detected as a separate type of
1879 * The Password Agent protocol gained support for queries where
1880 the user input is shown, useful e.g. for user names.
1881 systemd-ask-password gained a new --echo option to turn that
1884 * The default sysctl.d/ snippets will now set:
1886 net.core.default_qdisc = fq_codel
1888 This selects Fair Queuing Controlled Delay as the default
1889 queuing discipline for network interfaces. fq_codel helps
1890 fight the network bufferbloat problem. It is believed to be
1891 a good default with no tuning required for most workloads.
1892 Downstream distributions may override this choice. On 10Gbit
1893 servers that do not do forwarding, "fq" may perform better.
1894 Systems without a good clocksource should use "pfifo_fast".
1896 * If kdbus is enabled during build a new option BusPolicy= is
1897 available for service units, that allows locking all service
1898 processes into a stricter bus policy, in order to limit
1899 access to various bus services, or even hide most of them
1900 from the service's view entirely.
1902 * networkctl will now show the .network and .link file
1903 networkd has applied to a specific interface.
1905 * sd-login gained a new API call sd_session_get_desktop() to
1906 query which desktop environment has been selected for a
1909 * UNIX utmp support is now compile-time optional to support
1910 legacy-free systems.
1912 * systemctl gained two new commands "add-wants" and
1913 "add-requires" for pulling in units from specific targets
1916 * If the word "rescue" is specified on the kernel command line
1917 the system will now boot into rescue mode (aka
1918 rescue.target), which was previously available only by
1919 specifying "1" or "systemd.unit=rescue.target" on the kernel
1920 command line. This new kernel command line option nicely
1921 mirrors the already existing "emergency" kernel command line
1924 * New kernel command line options mount.usr=, mount.usrflags=,
1925 mount.usrfstype= have been added that match root=, rootflags=,
1926 rootfstype= but allow mounting a specific file system to
1929 * The $NOTIFY_SOCKET is now also passed to control processes of
1930 services, not only the main process.
1932 * This version reenables support for fsck's -l switch. This
1933 means at least version v2.25 of util-linux is required for
1934 operation, otherwise dead-locks on device nodes may
1935 occur. Again: you need to update util-linux to at least
1936 v2.25 when updating systemd to v217.
1938 * The "multi-seat-x" tool has been removed from systemd, as
1939 its functionality has been integrated into X servers 1.16,
1940 and the tool is hence redundant. It is recommended to update
1941 display managers invoking this tool to simply invoke X
1942 directly from now on, again.
1944 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1945 message flag has been added for all of systemd's PolicyKit
1946 authenticated method calls has been added. In particular
1947 this now allows optional interactive authorization via
1948 PolicyKit for many of PID1's privileged operations such as
1949 unit file enabling and disabling.
1951 * "udevadm hwdb --update" learnt a new switch "--usr" for
1952 placing the rebuilt hardware database in /usr instead of
1953 /etc. When used only hardware database entries stored in
1954 /usr will be used, and any user database entries in /etc are
1955 ignored. This functionality is useful for vendors to ship a
1956 pre-built database on systems where local configuration is
1957 unnecessary or unlikely.
1959 * Calendar time specifications in .timer units now also
1960 understand the strings "semi-annually", "quarterly" and
1961 "minutely" as shortcuts (in addition to the preexisting
1962 "anually", "hourly", ...).
1964 * systemd-tmpfiles will now correctly create files in /dev
1965 at boot which are marked for creation only at boot. It is
1966 recommended to always create static device nodes with 'c!'
1967 and 'b!', so that they are created only at boot and not
1968 overwritten at runtime.
1970 * When the watchdog logic is used for a service (WatchdogSec=)
1971 and the watchdog timeout is hit the service will now be
1972 terminated with SIGABRT (instead of just SIGTERM), in order
1973 to make sure a proper coredump and backtrace is
1974 generated. This ensures that hanging services will result in
1975 similar coredump/backtrace behaviour as services that hit a
1978 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1979 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1980 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1981 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1982 Herrmann, David Sommerseth, David Strauss, Emil Renner
1983 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1984 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1985 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1986 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1987 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1988 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1989 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1990 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1991 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1992 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1993 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1994 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1995 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1996 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1997 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1998 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2001 -- Berlin, 2014-10-28
2005 * timedated no longer reads NTP implementation unit names from
2006 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2007 implementations should add a
2009 Conflicts=systemd-timesyncd.service
2011 to their unit files to take over and replace systemd's NTP
2012 default functionality.
2014 * systemd-sysusers gained a new line type "r" for configuring
2015 which UID/GID ranges to allocate system users/groups
2016 from. Lines of type "u" may now add an additional column
2017 that specifies the home directory for the system user to be
2018 created. Also, systemd-sysusers may now optionally read user
2019 information from STDIN instead of a file. This is useful for
2020 invoking it from RPM preinst scriptlets that need to create
2021 users before the first RPM file is installed since these
2022 files might need to be owned by them. A new
2023 %sysusers_create_inline RPM macro has been introduced to do
2024 just that. systemd-sysusers now updates the shadow files as
2025 well as the user/group databases, which should enhance
2026 compatibility with certain tools like grpck.
2028 * A number of bus APIs of PID 1 now optionally consult
2029 PolicyKit to permit access for otherwise unprivileged
2030 clients under certain conditions. Note that this currently
2031 doesn't support interactive authentication yet, but this is
2032 expected to be added eventually, too.
2034 * /etc/machine-info now has new fields for configuring the
2035 deployment environment of the machine, as well as the
2036 location of the machine. hostnamectl has been updated with
2037 new command to update these fields.
2039 * systemd-timesyncd has been updated to automatically acquire
2040 NTP server information from systemd-networkd, which might
2041 have been discovered via DHCP.
2043 * systemd-resolved now includes a caching DNS stub resolver
2044 and a complete LLMNR name resolution implementation. A new
2045 NSS module "nss-resolve" has been added which can be used
2046 instead of glibc's own "nss-dns" to resolve hostnames via
2047 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2048 be resolved via systemd-resolved D-Bus APIs. In contrast to
2049 the glibc internal resolver systemd-resolved is aware of
2050 multi-homed system, and keeps DNS server and caches separate
2051 and per-interface. Queries are sent simultaneously on all
2052 interfaces that have DNS servers configured, in order to
2053 properly handle VPNs and local LANs which might resolve
2054 separate sets of domain names. systemd-resolved may acquire
2055 DNS server information from systemd-networkd automatically,
2056 which in turn might have discovered them via DHCP. A tool
2057 "systemd-resolve-host" has been added that may be used to
2058 query the DNS logic in resolved. systemd-resolved implements
2059 IDNA and automatically uses IDNA or UTF-8 encoding depending
2060 on whether classic DNS or LLMNR is used as transport. In the
2061 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2062 implementation to systemd-resolved.
2064 * A new NSS module nss-mymachines has been added, that
2065 automatically resolves the names of all local registered
2066 containers to their respective IP addresses.
2068 * A new client tool "networkctl" for systemd-networkd has been
2069 added. It currently is entirely passive and will query
2070 networking configuration from udev, rtnetlink and networkd,
2071 and present it to the user in a very friendly
2072 way. Eventually, we hope to extend it to become a full
2073 control utility for networkd.
2075 * .socket units gained a new DeferAcceptSec= setting that
2076 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2077 TCP. Similarly, support for controlling TCP keep-alive
2078 settings has been added (KeepAliveTimeSec=,
2079 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2080 turning off Nagle's algorithm on TCP has been added
2083 * logind learned a new session type "web", for use in projects
2084 like Cockpit which register web clients as PAM sessions.
2086 * timer units with at least one OnCalendar= setting will now
2087 be started only after timer-sync.target has been
2088 reached. This way they will not elapse before the system
2089 clock has been corrected by a local NTP client or
2090 similar. This is particular useful on RTC-less embedded
2091 machines, that come up with an invalid system clock.
2093 * systemd-nspawn's --network-veth= switch should now result in
2094 stable MAC addresses for both the outer and the inner side
2097 * systemd-nspawn gained a new --volatile= switch for running
2098 container instances with /etc or /var unpopulated.
2100 * The kdbus client code has been updated to use the new Linux
2101 3.17 memfd subsystem instead of the old kdbus-specific one.
2103 * systemd-networkd's DHCP client and server now support
2104 FORCERENEW. There are also new configuration options to
2105 configure the vendor client identifier and broadcast mode
2108 * systemd will no longer inform the kernel about the current
2109 timezone, as this is necessarily incorrect and racy as the
2110 kernel has no understanding of DST and similar
2111 concepts. This hence means FAT timestamps will be always
2112 considered UTC, similar to what Android is already
2113 doing. Also, when the RTC is configured to the local time
2114 (rather than UTC) systemd will never synchronize back to it,
2115 as this might confuse Windows at a later boot.
2117 * systemd-analyze gained a new command "verify" for offline
2118 validation of unit files.
2120 * systemd-networkd gained support for a couple of additional
2121 settings for bonding networking setups. Also, the metric for
2122 statically configured routes may now be configured. For
2123 network interfaces where this is appropriate the peer IP
2124 address may now be configured.
2126 * systemd-networkd's DHCP client will no longer request
2127 broadcasting by default, as this tripped up some networks.
2128 For hardware where broadcast is required the feature should
2129 be switched back on using RequestBroadcast=yes.
2131 * systemd-networkd will now set up IPv4LL addresses (when
2132 enabled) even if DHCP is configured successfully.
2134 * udev will now default to respect network device names given
2135 by the kernel when the kernel indicates that these are
2136 predictable. This behavior can be tweaked by changing
2137 NamePolicy= in the relevant .link file.
2139 * A new library systemd-terminal has been added that
2140 implements full TTY stream parsing and rendering. This
2141 library is supposed to be used later on for implementing a
2142 full userspace VT subsystem, replacing the current kernel
2145 * A new tool systemd-journal-upload has been added to push
2146 journal data to a remote system running
2147 systemd-journal-remote.
2149 * journald will no longer forward all local data to another
2150 running syslog daemon. This change has been made because
2151 rsyslog (which appears to be the most commonly used syslog
2152 implementation these days) no longer makes use of this, and
2153 instead pulls the data out of the journal on its own. Since
2154 forwarding the messages to a non-existent syslog server is
2155 more expensive than we assumed we have now turned this
2156 off. If you run a syslog server that is not a recent rsyslog
2157 version, you have to turn this option on again
2158 (ForwardToSyslog= in journald.conf).
2160 * journald now optionally supports the LZ4 compressor for
2161 larger journal fields. This compressor should perform much
2162 better than XZ which was the previous default.
2164 * machinectl now shows the IP addresses of local containers,
2165 if it knows them, plus the interface name of the container.
2167 * A new tool "systemd-escape" has been added that makes it
2168 easy to escape strings to build unit names and similar.
2170 * sd_notify() messages may now include a new ERRNO= field
2171 which is parsed and collected by systemd and shown among the
2172 "systemctl status" output for a service.
2174 * A new component "systemd-firstboot" has been added that
2175 queries the most basic systemd information (timezone,
2176 hostname, root password) interactively on first
2177 boot. Alternatively it may also be used to provision these
2178 things offline on OS images installed into directories.
2180 * The default sysctl.d/ snippets will now set
2182 net.ipv4.conf.default.promote_secondaries=1
2184 This has the benefit of no flushing secondary IP addresses
2185 when primary addresses are removed.
2187 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2188 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2189 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2190 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2191 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2192 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2193 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2194 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2195 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2196 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2197 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2198 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2199 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2200 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2201 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2203 -- Berlin, 2014-08-19
2207 * A new tool systemd-sysusers has been added. This tool
2208 creates system users and groups in /etc/passwd and
2209 /etc/group, based on static declarative system user/group
2210 definitions in /usr/lib/sysusers.d/. This is useful to
2211 enable factory resets and volatile systems that boot up with
2212 an empty /etc directory, and thus need system users and
2213 groups created during early boot. systemd now also ships
2214 with two default sysusers.d/ files for the most basic
2215 users and groups systemd and the core operating system
2218 * A new tmpfiles snippet has been added that rebuilds the
2219 essential files in /etc on boot, should they be missing.
2221 * A directive for ensuring automatic clean-up of
2222 /var/cache/man/ has been removed from the default
2223 configuration. This line should now be shipped by the man
2224 implementation. The necessary change has been made to the
2225 man-db implementation. Note that you need to update your man
2226 implementation to one that ships this line, otherwise no
2227 automatic clean-up of /var/cache/man will take place.
2229 * A new condition ConditionNeedsUpdate= has been added that
2230 may conditionalize services to only run when /etc or /var
2231 are "older" than the vendor operating system resources in
2232 /usr. This is useful for reconstructing or updating /etc
2233 after an offline update of /usr or a factory reset, on the
2234 next reboot. Services that want to run once after such an
2235 update or reset should use this condition and order
2236 themselves before the new systemd-update-done.service, which
2237 will mark the two directories as fully updated. A number of
2238 service files have been added making use of this, to rebuild
2239 the udev hardware database, the journald message catalog and
2240 dynamic loader cache (ldconfig). The systemd-sysusers tool
2241 described above also makes use of this now. With this in
2242 place it is now possible to start up a minimal operating
2243 system with /etc empty cleanly. For more information on the
2244 concepts involved see this recent blog story:
2246 http://0pointer.de/blog/projects/stateless.html
2248 * A new system group "input" has been introduced, and all
2249 input device nodes get this group assigned. This is useful
2250 for system-level software to get access to input devices. It
2251 complements what is already done for "audio" and "video".
2253 * systemd-networkd learnt minimal DHCPv4 server support in
2254 addition to the existing DHCPv4 client support. It also
2255 learnt DHCPv6 client and IPv6 Router Solicitation client
2256 support. The DHCPv4 client gained support for static routes
2257 passed in from the server. Note that the [DHCPv4] section
2258 known in older systemd-networkd versions has been renamed to
2259 [DHCP] and is now also used by the DHCPv6 client. Existing
2260 .network files using settings of this section should be
2261 updated, though compatibility is maintained. Optionally, the
2262 client hostname may now be sent to the DHCP server.
2264 * networkd gained support for vxlan virtual networks as well
2265 as tun/tap and dummy devices.
2267 * networkd gained support for automatic allocation of address
2268 ranges for interfaces from a system-wide pool of
2269 addresses. This is useful for dynamically managing a large
2270 number of interfaces with a single network configuration
2271 file. In particular this is useful to easily assign
2272 appropriate IP addresses to the veth links of a large number
2273 of nspawn instances.
2275 * RPM macros for processing sysusers, sysctl and binfmt
2276 drop-in snippets at package installation time have been
2279 * The /etc/os-release file should now be placed in
2280 /usr/lib/os-release. The old location is automatically
2281 created as symlink. /usr/lib is the more appropriate
2282 location of this file, since it shall actually describe the
2283 vendor operating system shipped in /usr, and not the
2284 configuration stored in /etc.
2286 * .mount units gained a new boolean SloppyOptions= setting
2287 that maps to mount(8)'s -s option which enables permissive
2288 parsing of unknown mount options.
2290 * tmpfiles learnt a new "L+" directive which creates a symlink
2291 but (unlike "L") deletes a pre-existing file first, should
2292 it already exist and not already be the correct
2293 symlink. Similarly, "b+", "c+" and "p+" directives have been
2294 added as well, which create block and character devices, as
2295 well as fifos in the filesystem, possibly removing any
2296 pre-existing files of different types.
2298 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2299 'argument' field (which so far specified the source to
2300 symlink/copy the files from) is now optional. If omitted the
2301 same file os copied from /usr/share/factory/ suffixed by the
2302 full destination path. This is useful for populating /etc
2303 with essential files, by copying them from vendor defaults
2304 shipped in /usr/share/factory/etc.
2306 * A new command "systemctl preset-all" has been added that
2307 applies the service preset settings to all installed unit
2308 files. A new switch --preset-mode= has been added that
2309 controls whether only enable or only disable operations
2312 * A new command "systemctl is-system-running" has been added
2313 that allows checking the overall state of the system, for
2314 example whether it is fully up and running.
2316 * When the system boots up with an empty /etc, the equivalent
2317 to "systemctl preset-all" is executed during early boot, to
2318 make sure all default services are enabled after a factory
2321 * systemd now contains a minimal preset file that enables the
2322 most basic services systemd ships by default.
2324 * Unit files' [Install] section gained a new DefaultInstance=
2325 field for defining the default instance to create if a
2326 template unit is enabled with no instance specified.
2328 * A new passive target cryptsetup-pre.target has been added
2329 that may be used by services that need to make they run and
2330 finish before the first LUKS cryptographic device is set up.
2332 * The /dev/loop-control and /dev/btrfs-control device nodes
2333 are now owned by the "disk" group by default, opening up
2334 access to this group.
2336 * systemd-coredump will now automatically generate a
2337 stack trace of all core dumps taking place on the system,
2338 based on elfutils' libdw library. This stack trace is logged
2341 * systemd-coredump may now optionally store coredumps directly
2342 on disk (in /var/lib/systemd/coredump, possibly compressed),
2343 instead of storing them unconditionally in the journal. This
2344 mode is the new default. A new configuration file
2345 /etc/systemd/coredump.conf has been added to configure this
2346 and other parameters of systemd-coredump.
2348 * coredumpctl gained a new "info" verb to show details about a
2349 specific coredump. A new switch "-1" has also been added
2350 that makes sure to only show information about the most
2351 recent entry instead of all entries. Also, as the tool is
2352 generally useful now the "systemd-" prefix of the binary
2353 name has been removed. Distributions that want to maintain
2354 compatibility with the old name should add a symlink from
2355 the old name to the new name.
2357 * journald's SplitMode= now defaults to "uid". This makes sure
2358 that unprivileged users can access their own coredumps with
2359 coredumpctl without restrictions.
2361 * New kernel command line options "systemd.wants=" (for
2362 pulling an additional unit during boot), "systemd.mask="
2363 (for masking a specific unit for the boot), and
2364 "systemd.debug-shell" (for enabling the debug shell on tty9)
2365 have been added. This is implemented in the new generator
2366 "systemd-debug-generator".
2368 * systemd-nspawn will now by default filter a couple of
2369 syscalls for containers, among them those required for
2370 kernel module loading, direct x86 IO port access, swap
2371 management, and kexec. Most importantly though
2372 open_by_handle_at() is now prohibited for containers,
2373 closing a hole similar to a recently discussed vulnerability
2374 in docker regarding access to files on file hierarchies the
2375 container should normally not have access to. Note that, for
2376 nspawn, we generally make no security claims anyway (and
2377 this is explicitly documented in the man page), so this is
2378 just a fix for one of the most obvious problems.
2380 * A new man page file-hierarchy(7) has been added that
2381 contains a minimized, modernized version of the file system
2382 layout systemd expects, similar in style to the FHS
2383 specification or hier(5). A new tool systemd-path(1) has
2384 been added to query many of these paths for the local
2387 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2388 longer done. Since the directory now has a per-user size
2389 limit, and is cleaned on logout this appears unnecessary,
2390 in particular since this now brings the lifecycle of this
2391 directory closer in line with how IPC objects are handled.
2393 * systemd.pc now exports a number of additional directories,
2394 including $libdir (which is useful to identify the library
2395 path for the primary architecture of the system), and a
2396 couple of drop-in directories.
2398 * udev's predictable network interface names now use the dev_port
2399 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2400 distinguish between ports of the same PCI function. dev_id should
2401 only be used for ports using the same HW address, hence the need
2404 * machined has been updated to export the OS version of a
2405 container (read from /etc/os-release and
2406 /usr/lib/os-release) on the bus. This is now shown in
2407 "machinectl status" for a machine.
2409 * A new service setting RestartForceExitStatus= has been
2410 added. If configured to a set of exit signals or process
2411 return values, the service will be restarted when the main
2412 daemon process exits with any of them, regardless of the
2415 * systemctl's -H switch for connecting to remote systemd
2416 machines has been extended so that it may be used to
2417 directly connect to a specific container on the
2418 host. "systemctl -H root@foobar:waldi" will now connect as
2419 user "root" to host "foobar", and then proceed directly to
2420 the container named "waldi". Note that currently you have to
2421 authenticate as user "root" for this to work, as entering
2422 containers is a privileged operation.
2424 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2425 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2426 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2427 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2428 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2429 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2430 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2431 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2432 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2433 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2434 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2435 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2437 -- Berlin, 2014-07-03
2441 * As an experimental feature, udev now tries to lock the
2442 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2443 executes events for the disk or any of its partitions.
2444 Applications like partitioning programs can lock the
2445 disk device node (flock(LOCK_EX)) and claim temporary
2446 device ownership that way; udev will entirely skip all event
2447 handling for this disk and its partitions. If the disk
2448 was opened for writing, the close will trigger a partition
2449 table rescan in udev's "watch" facility, and if needed
2450 synthesize "change" events for the disk and all its partitions.
2451 This is now unconditionally enabled, and if it turns out to
2452 cause major problems, we might turn it on only for specific
2453 devices, or might need to disable it entirely. Device Mapper
2454 devices are excluded from this logic.
2456 * We temporarily dropped the "-l" switch for fsck invocations,
2457 since they collide with the flock() logic above. util-linux
2458 upstream has been changed already to avoid this conflict,
2459 and we will readd "-l" as soon as util-linux with this
2460 change has been released.
2462 * The dependency on libattr has been removed. Since a long
2463 time, the extended attribute calls have moved to glibc, and
2464 libattr is thus unnecessary.
2466 * Virtualization detection works without priviliges now. This
2467 means the systemd-detect-virt binary no longer requires
2468 CAP_SYS_PTRACE file capabilities, and our daemons can run
2469 with fewer privileges.
2471 * systemd-networkd now runs under its own "systemd-network"
2472 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2473 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2474 loses the ability to write to files owned by root this way.
2476 * Similarly, systemd-resolved now runs under its own
2477 "systemd-resolve" user with no capabilities remaining.
2479 * Similarly, systemd-bus-proxyd now runs under its own
2480 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2482 * systemd-networkd gained support for setting up "veth"
2483 virtual Ethernet devices for container connectivity, as well
2484 as GRE and VTI tunnels.
2486 * systemd-networkd will no longer automatically attempt to
2487 manually load kernel modules necessary for certain tunnel
2488 transports. Instead, it is assumed the kernel loads them
2489 automatically when required. This only works correctly on
2490 very new kernels. On older kernels, please consider adding
2491 the kernel modules to /etc/modules-load.d/ as a work-around.
2493 * The resolv.conf file systemd-resolved generates has been
2494 moved to /run/systemd/resolve/. If you have a symlink from
2495 /etc/resolv.conf, it might be necessary to correct it.
2497 * Two new service settings, ProtectHome= and ProtectSystem=,
2498 have been added. When enabled, they will make the user data
2499 (such as /home) inaccessible or read-only and the system
2500 (such as /usr) read-only, for specific services. This allows
2501 very light-weight per-service sandboxing to avoid
2502 modifications of user data or system files from
2503 services. These two new switches have been enabled for all
2504 of systemd's long-running services, where appropriate.
2506 * Socket units gained new SocketUser= and SocketGroup=
2507 settings to set the owner user and group of AF_UNIX sockets
2508 and FIFOs in the file system.
2510 * Socket units gained a new RemoveOnStop= setting. If enabled,
2511 all FIFOS and sockets in the file system will be removed
2512 when the specific socket unit is stopped.
2514 * Socket units gained a new Symlinks= setting. It takes a list
2515 of symlinks to create to file system sockets or FIFOs
2516 created by the specific Unix sockets. This is useful to
2517 manage symlinks to socket nodes with the same life-cycle as
2520 * The /dev/log socket and /dev/initctl FIFO have been moved to
2521 /run, and have been replaced by symlinks. This allows
2522 connecting to these facilities even if PrivateDevices=yes is
2523 used for a service (which makes /dev/log itself unavailable,
2524 but /run is left). This also has the benefit of ensuring
2525 that /dev only contains device nodes, directories and
2526 symlinks, and nothing else.
2528 * sd-daemon gained two new calls sd_pid_notify() and
2529 sd_pid_notifyf(). They are similar to sd_notify() and
2530 sd_notifyf(), but allow overriding of the source PID of
2531 notification messages if permissions permit this. This is
2532 useful to send notify messages on behalf of a different
2533 process (for example, the parent process). The
2534 systemd-notify tool has been updated to make use of this
2535 when sending messages (so that notification messages now
2536 originate from the shell script invoking systemd-notify and
2537 not the systemd-notify process itself. This should minimize
2538 a race where systemd fails to associate notification
2539 messages to services when the originating process already
2542 * A new "on-abnormal" setting for Restart= has been added. If
2543 set, it will result in automatic restarts on all "abnormal"
2544 reasons for a process to exit, which includes unclean
2545 signals, core dumps, timeouts and watchdog timeouts, but
2546 does not include clean and unclean exit codes or clean
2547 signals. Restart=on-abnormal is an alternative for
2548 Restart=on-failure for services that shall be able to
2549 terminate and avoid restarts on certain errors, by
2550 indicating so with an unclean exit code. Restart=on-failure
2551 or Restart=on-abnormal is now the recommended setting for
2552 all long-running services.
2554 * If the InaccessibleDirectories= service setting points to a
2555 mount point (or if there are any submounts contained within
2556 it), it is now attempted to completely unmount it, to make
2557 the file systems truly unavailable for the respective
2560 * The ReadOnlyDirectories= service setting and
2561 systemd-nspawn's --read-only parameter are now recursively
2562 applied to all submounts, too.
2564 * Mount units may now be created transiently via the bus APIs.
2566 * The support for SysV and LSB init scripts has been removed
2567 from the systemd daemon itself. Instead, it is now
2568 implemented as a generator that creates native systemd units
2569 from these scripts when needed. This enables us to remove a
2570 substantial amount of legacy code from PID 1, following the
2571 fact that many distributions only ship a very small number
2572 of LSB/SysV init scripts nowadays.
2574 * Privileged Xen (dom0) domains are not considered
2575 virtualization anymore by the virtualization detection
2576 logic. After all, they generally have unrestricted access to
2577 the hardware and usually are used to manage the unprivileged
2580 * systemd-tmpfiles gained a new "C" line type, for copying
2581 files or entire directories.
2583 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2584 lines. So far, they have been non-globbing versions of the
2585 latter, and have thus been redundant. In future, it is
2586 recommended to only use "z". "m" has hence been removed
2587 from the documentation, even though it stays supported.
2589 * A tmpfiles snippet to recreate the most basic structure in
2590 /var has been added. This is enough to create the /var/run →
2591 /run symlink and create a couple of structural
2592 directories. This allows systems to boot up with an empty or
2593 volatile /var. Of course, while with this change, the core OS
2594 now is capable with dealing with a volatile /var, not all
2595 user services are ready for it. However, we hope that sooner
2596 or later, many service daemons will be changed upstream so
2597 that they are able to automatically create their necessary
2598 directories in /var at boot, should they be missing. This is
2599 the first step to allow state-less systems that only require
2600 the vendor image for /usr to boot.
2602 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2603 empty tmpfs instance to a specific directory. This is
2604 particularly useful for making use of the automatic
2605 reconstruction of /var (see above), by passing --tmpfs=/var.
2607 * Access modes specified in tmpfiles snippets may now be
2608 prefixed with "~", which indicates that they shall be masked
2609 by whether the existing file or directory is currently
2610 writable, readable or executable at all. Also, if specified,
2611 the sgid/suid/sticky bits will be masked for all
2614 * A new passive target unit "network-pre.target" has been
2615 added which is useful for services that shall run before any
2616 network is configured, for example firewall scripts.
2618 * The "floppy" group that previously owned the /dev/fd*
2619 devices is no longer used. The "disk" group is now used
2620 instead. Distributions should probably deprecate usage of
2623 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2624 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2625 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2626 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2627 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2628 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2631 -- Berlin, 2014-06-11
2635 * A new "systemd-timesyncd" daemon has been added for
2636 synchronizing the system clock across the network. It
2637 implements an SNTP client. In contrast to NTP
2638 implementations such as chrony or the NTP reference server,
2639 this only implements a client side, and does not bother with
2640 the full NTP complexity, focusing only on querying time from
2641 one remote server and synchronizing the local clock to
2642 it. Unless you intend to serve NTP to networked clients or
2643 want to connect to local hardware clocks, this simple NTP
2644 client should be more than appropriate for most
2645 installations. The daemon runs with minimal privileges, and
2646 has been hooked up with networkd to only operate when
2647 network connectivity is available. The daemon saves the
2648 current clock to disk every time a new NTP sync has been
2649 acquired, and uses this to possibly correct the system clock
2650 early at bootup, in order to accommodate for systems that
2651 lack an RTC such as the Raspberry Pi and embedded devices,
2652 and to make sure that time monotonically progresses on these
2653 systems, even if it is not always correct. To make use of
2654 this daemon, a new system user and group "systemd-timesync"
2655 needs to be created on installation of systemd.
2657 * The queue "seqnum" interface of libudev has been disabled, as
2658 it was generally incompatible with device namespacing as
2659 sequence numbers of devices go "missing" if the devices are
2660 part of a different namespace.
2662 * "systemctl list-timers" and "systemctl list-sockets" gained
2663 a --recursive switch for showing units of these types also
2664 for all local containers, similar in style to the already
2665 supported --recursive switch for "systemctl list-units".
2667 * A new RebootArgument= setting has been added for service
2668 units, which may be used to specify a kernel reboot argument
2669 to use when triggering reboots with StartLimitAction=.
2671 * A new FailureAction= setting has been added for service
2672 units which may be used to specify an operation to trigger
2673 when a service fails. This works similarly to
2674 StartLimitAction=, but unlike it, controls what is done
2675 immediately rather than only after several attempts to
2676 restart the service in question.
2678 * hostnamed got updated to also expose the kernel name,
2679 release, and version on the bus. This is useful for
2680 executing commands like hostnamectl with the -H switch.
2681 systemd-analyze makes use of this to properly display
2682 details when running non-locally.
2684 * The bootchart tool can now show cgroup information in the
2685 graphs it generates.
2687 * The CFS CPU quota cgroup attribute is now exposed for
2688 services. The new CPUQuota= switch has been added for this
2689 which takes a percentage value. Setting this will have the
2690 result that a service may never get more CPU time than the
2691 specified percentage, even if the machine is otherwise idle.
2693 * systemd-networkd learned IPIP and SIT tunnel support.
2695 * LSB init scripts exposing a dependency on $network will now
2696 get a dependency on network-online.target rather than simply
2697 network.target. This should bring LSB handling closer to
2698 what it was on SysV systems.
2700 * A new fsck.repair= kernel option has been added to control
2701 how fsck shall deal with unclean file systems at boot.
2703 * The (.ini) configuration file parser will now silently
2704 ignore sections whose name begins with "X-". This may be
2705 used to maintain application-specific extension sections in unit
2708 * machined gained a new API to query the IP addresses of
2709 registered containers. "machinectl status" has been updated
2710 to show these addresses in its output.
2712 * A new call sd_uid_get_display() has been added to the
2713 sd-login APIs for querying the "primary" session of a
2714 user. The "primary" session of the user is elected from the
2715 user's sessions and generally a graphical session is
2716 preferred over a text one.
2718 * A minimal systemd-resolved daemon has been added. It
2719 currently simply acts as a companion to systemd-networkd and
2720 manages resolv.conf based on per-interface DNS
2721 configuration, possibly supplied via DHCP. In the long run
2722 we hope to extend this into a local DNSSEC enabled DNS and
2725 * The systemd-networkd-wait-online tool is now enabled by
2726 default. It will delay network-online.target until a network
2727 connection has been configured. The tool primarily integrates
2728 with networkd, but will also make a best effort to make sense
2729 of network configuration performed in some other way.
2731 * Two new service options StartupCPUShares= and
2732 StartupBlockIOWeight= have been added that work similarly to
2733 CPUShares= and BlockIOWeight= however only apply during
2734 system startup. This is useful to prioritize certain services
2735 differently during bootup than during normal runtime.
2737 * hostnamed has been changed to prefer the statically
2738 configured hostname in /etc/hostname (unless set to
2739 'localhost' or empty) over any dynamic one supplied by
2740 dhcp. With this change, the rules for picking the hostname
2741 match more closely the rules of other configuration settings
2742 where the local administrator's configuration in /etc always
2743 overrides any other settings.
2745 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2746 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2747 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2748 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2749 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2750 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2751 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2752 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2753 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2754 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2755 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2756 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2757 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2758 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2759 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2760 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2763 -- Beijing, 2014-05-28
2767 * When restoring the screen brightness at boot, stay away from
2768 the darkest setting or from the lowest 5% of the available
2769 range, depending on which is the larger value of both. This
2770 should effectively protect the user from rebooting into a
2771 black screen, should the brightness have been set to minimum
2774 * sd-login gained a new sd_machine_get_class() call to
2775 determine the class ("vm" or "container") of a machine
2776 registered with machined.
2778 * sd-login gained new calls
2779 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2780 to query the identity of the peer of a local AF_UNIX
2781 connection. They operate similarly to their sd_pid_get_xyz()
2784 * PID 1 will now maintain a system-wide system state engine
2785 with the states "starting", "running", "degraded",
2786 "maintenance", "stopping". These states are bound to system
2787 startup, normal runtime, runtime with at least one failed
2788 service, rescue/emergency mode and system shutdown. This
2789 state is shown in the "systemctl status" output when no unit
2790 name is passed. It is useful to determine system state, in
2791 particularly when doing so for many systems or containers at
2794 * A new command "list-machines" has been added to "systemctl"
2795 that lists all local OS containers and shows their system
2796 state (see above), if systemd runs inside of them.
2798 * systemctl gained a new "-r" switch to recursively enumerate
2799 units on all local containers, when used with the
2800 "list-unit" command (which is the default one that is
2801 executed when no parameters are specified).
2803 * The GPT automatic partition discovery logic will now honour
2804 two GPT partition flags: one may be set on a partition to
2805 cause it to be mounted read-only, and the other may be set
2806 on a partition to ignore it during automatic discovery.
2808 * Two new GPT type UUIDs have been added for automatic root
2809 partition discovery, for 32-bit and 64-bit ARM. This is not
2810 particularly useful for discovering the root directory on
2811 these architectures during bare-metal boots (since UEFI is
2812 not common there), but still very useful to allow booting of
2813 ARM disk images in nspawn with the -i option.
2815 * MAC addresses of interfaces created with nspawn's
2816 --network-interface= switch will now be generated from the
2817 machine name, and thus be stable between multiple invocations
2820 * logind will now automatically remove all IPC objects owned
2821 by a user if she or he fully logs out. This makes sure that
2822 users who are logged out cannot continue to consume IPC
2823 resources. This covers SysV memory, semaphores and message
2824 queues as well as POSIX shared memory and message
2825 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2826 limits. With this functionality, that is corrected. This may
2827 be turned off by using the RemoveIPC= switch of logind.conf.
2829 * The systemd-machine-id-setup and tmpfiles tools gained a
2830 --root= switch to operate on a specific root directory,
2833 * journald can now forward logged messages to the TTYs of all
2834 logged in users ("wall"). This is the default for all
2835 emergency messages now.
2837 * A new tool systemd-journal-remote has been added to stream
2838 journal log messages across the network.
2840 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2841 controller trees are mounted into it. Note that the
2842 directories mounted beneath it are not read-only. This is a
2843 security measure and is particularly useful because glibc
2844 actually includes a search logic to pick any tmpfs it can
2845 find to implement shm_open() if /dev/shm is not available
2846 (which it might very well be in namespaced setups).
2848 * machinectl gained a new "poweroff" command to cleanly power
2849 down a local OS container.
2851 * The PrivateDevices= unit file setting will now also drop the
2852 CAP_MKNOD capability from the capability bound set, and
2853 imply DevicePolicy=closed.
2855 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2856 comprehensively on all long-running systemd services where
2857 this is appropriate.
2859 * systemd-udevd will now run in a disassociated mount
2860 namespace. To mount directories from udev rules, make sure to
2861 pull in mount units via SYSTEMD_WANTS properties.
2863 * The kdbus support gained support for uploading policy into
2864 the kernel. sd-bus gained support for creating "monitoring"
2865 connections that can eavesdrop into all bus communication
2866 for debugging purposes.
2868 * Timestamps may now be specified in seconds since the UNIX
2869 epoch Jan 1st, 1970 by specifying "@" followed by the value
2872 * Native tcpwrap support in systemd has been removed. tcpwrap
2873 is old code, not really maintained anymore and has serious
2874 shortcomings, and better options such as firewalls
2875 exist. For setups that require tcpwrap usage, please
2876 consider invoking your socket-activated service via tcpd,
2877 like on traditional inetd.
2879 * A new system.conf configuration option
2880 DefaultTimerAccuracySec= has been added that controls the
2881 default AccuracySec= setting of .timer units.
2883 * Timer units gained a new WakeSystem= switch. If enabled,
2884 timers configured this way will cause the system to resume
2885 from system suspend (if the system supports that, which most
2888 * Timer units gained a new Persistent= switch. If enabled,
2889 timers configured this way will save to disk when they have
2890 been last triggered. This information is then used on next
2891 reboot to possible execute overdue timer events, that
2892 could not take place because the system was powered off.
2893 This enables simple anacron-like behaviour for timer units.
2895 * systemctl's "list-timers" will now also list the time a
2896 timer unit was last triggered in addition to the next time
2897 it will be triggered.
2899 * systemd-networkd will now assign predictable IPv4LL
2900 addresses to its local interfaces.
2902 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2903 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2904 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2905 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2906 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2907 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2908 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2909 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2912 -- Berlin, 2014-03-25
2916 * A new unit file setting RestrictAddressFamilies= has been
2917 added to restrict which socket address families unit
2918 processes gain access to. This takes address family names
2919 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2920 attack surface of services via exotic protocol stacks. This
2921 is built on seccomp system call filters.
2923 * Two new unit file settings RuntimeDirectory= and
2924 RuntimeDirectoryMode= have been added that may be used to
2925 manage a per-daemon runtime directories below /run. This is
2926 an alternative for setting up directory permissions with
2927 tmpfiles snippets, and has the advantage that the runtime
2928 directory's lifetime is bound to the daemon runtime and that
2929 the daemon starts up with an empty directory each time. This
2930 is particularly useful when writing services that drop
2931 privileges using the User= or Group= setting.
2933 * The DeviceAllow= unit setting now supports globbing for
2934 matching against device group names.
2936 * The systemd configuration file system.conf gained new
2937 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2938 DefaultMemoryAccounting= to globally turn on/off accounting
2939 for specific resources (cgroups) for all units. These
2940 settings may still be overridden individually in each unit
2943 * systemd-gpt-auto-generator is now able to discover /srv and
2944 root partitions in addition to /home and swap partitions. It
2945 also supports LUKS-encrypted partitions now. With this in
2946 place, automatic discovery of partitions to mount following
2947 the Discoverable Partitions Specification
2948 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2949 is now a lot more complete. This allows booting without
2950 /etc/fstab and without root= on the kernel command line on
2951 systems prepared appropriately.
2953 * systemd-nspawn gained a new --image= switch which allows
2954 booting up disk images and Linux installations on any block
2955 device that follow the Discoverable Partitions Specification
2956 (see above). This means that installations made with
2957 appropriately updated installers may now be started and
2958 deployed using container managers, completely
2959 unmodified. (We hope that libvirt-lxc will add support for
2960 this feature soon, too.)
2962 * systemd-nspawn gained a new --network-macvlan= setting to
2963 set up a private macvlan interface for the
2964 container. Similarly, systemd-networkd gained a new
2965 Kind=macvlan setting in .netdev files.
2967 * systemd-networkd now supports configuring local addresses
2970 * A new tool systemd-network-wait-online has been added to
2971 synchronously wait for network connectivity using
2974 * The sd-bus.h bus API gained a new sd_bus_track object for
2975 tracking the life-cycle of bus peers. Note that sd-bus.h is
2976 still not a public API though (unless you specify
2977 --enable-kdbus on the configure command line, which however
2978 voids your warranty and you get no API stability guarantee).
2980 * The $XDG_RUNTIME_DIR runtime directories for each user are
2981 now individual tmpfs instances, which has the benefit of
2982 introducing separate pools for each user, with individual
2983 size limits, and thus making sure that unprivileged clients
2984 can no longer negatively impact the system or other users by
2985 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2986 RuntimeDirectorySize= has been introduced that allows
2987 controlling the default size limit for all users. It
2988 defaults to 10% of the available physical memory. This is no
2989 replacement for quotas on tmpfs though (which the kernel
2990 still does not support), as /dev/shm and /tmp are still
2991 shared resources used by both the system and unprivileged
2994 * logind will now automatically turn off automatic suspending
2995 on laptop lid close when more than one display is
2996 connected. This was previously expected to be implemented
2997 individually in desktop environments (such as GNOME),
2998 however has been added to logind now, in order to fix a
2999 boot-time race where a desktop environment might not have
3000 been started yet and thus not been able to take an inhibitor
3001 lock at the time where logind already suspends the system
3002 due to a closed lid.
3004 * logind will now wait at least 30s after each system
3005 suspend/resume cycle, and 3min after system boot before
3006 suspending the system due to a closed laptop lid. This
3007 should give USB docking stations and similar enough time to
3008 be probed and configured after system resume and boot in
3009 order to then act as suspend blocker.
3011 * systemd-run gained a new --property= setting which allows
3012 initialization of resource control properties (and others)
3013 for the created scope or service unit. Example: "systemd-run
3014 --property=BlockIOWeight=10 updatedb" may be used to run
3015 updatedb at a low block IO scheduling weight.
3017 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3018 now also work in --scope mode.
3020 * When systemd is compiled with kdbus support, basic support
3021 for enforced policies is now in place. (Note that enabling
3022 kdbus still voids your warranty and no API compatibility
3025 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3026 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3027 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3028 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3029 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3030 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3031 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3032 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3033 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3034 Zbigniew Jędrzejewski-Szmek
3036 -- Berlin, 2014-03-12
3040 * systemd will now relabel /dev after loading the SMACK policy
3041 according to SMACK rules.
3043 * A new unit file option AppArmorProfile= has been added to
3044 set the AppArmor profile for the processes of a unit.
3046 * A new condition check ConditionArchitecture= has been added
3047 to conditionalize units based on the system architecture, as
3048 reported by uname()'s "machine" field.
3050 * systemd-networkd now supports matching on the system
3051 virtualization, architecture, kernel command line, host name
3054 * logind is now a lot more aggressive when suspending the
3055 machine due to a closed laptop lid. Instead of acting only
3056 on the lid close action, it will continuously watch the lid
3057 status and act on it. This is useful for laptops where the
3058 power button is on the outside of the chassis so that it can
3059 be reached without opening the lid (such as the Lenovo
3060 Yoga). On those machines, logind will now immediately
3061 re-suspend the machine if the power button has been
3062 accidentally pressed while the laptop was suspended and in a
3063 backpack or similar.
3065 * logind will now watch SW_DOCK switches and inhibit reaction
3066 to the lid switch if it is pressed. This means that logind
3067 will not suspend the machine anymore if the lid is closed
3068 and the system is docked, if the laptop supports SW_DOCK
3069 notifications via the input layer. Note that ACPI docking
3070 stations do not generate this currently. Also note that this
3071 logic is usually not fully sufficient and Desktop
3072 Environments should take a lid switch inhibitor lock when an
3073 external display is connected, as systemd will not watch
3076 * nspawn will now make use of the devices cgroup controller by
3077 default, and only permit creation of and access to the usual
3078 API device nodes like /dev/null or /dev/random, as well as
3079 access to (but not creation of) the pty devices.
3081 * We will now ship a default .network file for
3082 systemd-networkd that automatically configures DHCP for
3083 network interfaces created by nspawn's --network-veth or
3084 --network-bridge= switches.
3086 * systemd will now understand the usual M, K, G, T suffixes
3087 according to SI conventions (i.e. to the base 1000) when
3088 referring to throughput and hardware metrics. It will stay
3089 with IEC conventions (i.e. to the base 1024) for software
3090 metrics, according to what is customary according to
3091 Wikipedia. We explicitly document which base applies for
3092 each configuration option.
3094 * The DeviceAllow= setting in unit files now supports a syntax
3095 to whitelist an entire group of devices node majors at once,
3096 based on the /proc/devices listing. For example, with the
3097 string "char-pts", it is now possible to whitelist all
3098 current and future pseudo-TTYs at once.
3100 * sd-event learned a new "post" event source. Event sources of
3101 this type are triggered by the dispatching of any event
3102 source of a type that is not "post". This is useful for
3103 implementing clean-up and check event sources that are
3104 triggered by other work being done in the program.
3106 * systemd-networkd is no longer statically enabled, but uses
3107 the usual [Install] sections so that it can be
3108 enabled/disabled using systemctl. It still is enabled by
3111 * When creating a veth interface pair with systemd-nspawn, the
3112 host side will now be prefixed with "vb-" if
3113 --network-bridge= is used, and with "ve-" if --network-veth
3114 is used. This way, it is easy to distinguish these cases on
3115 the host, for example to apply different configuration to
3116 them with systemd-networkd.
3118 * The compatibility libraries for libsystemd-journal.so,
3119 libsystem-id128.so, libsystemd-login.so and
3120 libsystemd-daemon.so do not make use of IFUNC
3121 anymore. Instead, we now build libsystemd.so multiple times
3122 under these alternative names. This means that the footprint
3123 is drastically increased, but given that these are
3124 transitional compatibility libraries, this should not matter
3125 much. This change has been made necessary to support the ARM
3126 platform for these compatibility libraries, as the ARM
3127 toolchain is not really at the same level as the toolchain
3128 for other architectures like x86 and does not support
3129 IFUNC. Please make sure to use --enable-compat-libs only
3130 during a transitional period!
3132 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3133 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3134 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3135 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3136 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3137 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3138 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3139 Zbigniew Jędrzejewski-Szmek
3141 -- Berlin, 2014-02-24
3145 * A new component "systemd-networkd" has been added that can
3146 be used to configure local network interfaces statically or
3147 via DHCP. It is capable of bringing up bridges, VLANs, and
3148 bonding. Currently, no hook-ups for interactive network
3149 configuration are provided. Use this for your initrd,
3150 container, embedded, or server setup if you need a simple,
3151 yet powerful, network configuration solution. This
3152 configuration subsystem is quite nifty, as it allows wildcard
3153 hotplug matching in interfaces. For example, with a single
3154 configuration snippet, you can configure that all Ethernet
3155 interfaces showing up are automatically added to a bridge,
3156 or similar. It supports link-sensing and more.
3158 * A new tool "systemd-socket-proxyd" has been added which can
3159 act as a bidirectional proxy for TCP sockets. This is
3160 useful for adding socket activation support to services that
3161 do not actually support socket activation, including virtual
3162 machines and the like.
3164 * Add a new tool to save/restore rfkill state on
3167 * Save/restore state of keyboard backlights in addition to
3168 display backlights on shutdown/boot.
3170 * udev learned a new SECLABEL{} construct to label device
3171 nodes with a specific security label when they appear. For
3172 now, only SECLABEL{selinux} is supported, but the syntax is
3173 prepared for additional security frameworks.
3175 * udev gained a new scheme to configure link-level attributes
3176 from files in /etc/systemd/network/*.link. These files can
3177 match against MAC address, device path, driver name and type,
3178 and will apply attributes like the naming policy, link speed,
3179 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3180 address assignment policy (randomized, ...).
3182 * The configuration of network interface naming rules for
3183 "permanent interface names" has changed: a new NamePolicy=
3184 setting in the [Link] section of .link files determines the
3185 priority of possible naming schemes (onboard, slot, MAC,
3186 path). The default value of this setting is determined by
3187 /usr/lib/net/links/99-default.link. Old
3188 80-net-name-slot.rules udev configuration file has been
3189 removed, so local configuration overriding this file should
3190 be adapated to override 99-default.link instead.
3192 * When the User= switch is used in a unit file, also
3193 initialize $SHELL= based on the user database entry.
3195 * systemd no longer depends on libdbus. All communication is
3196 now done with sd-bus, systemd's low-level bus library
3199 * kdbus support has been added to PID 1 itself. When kdbus is
3200 enabled, this causes PID 1 to set up the system bus and
3201 enable support for a new ".busname" unit type that
3202 encapsulates bus name activation on kdbus. It works a little
3203 bit like ".socket" units, except for bus names. A new
3204 generator has been added that converts classic dbus1 service
3205 activation files automatically into native systemd .busname
3208 * sd-bus: add a light-weight vtable implementation that allows
3209 defining objects on the bus with a simple static const
3210 vtable array of its methods, signals and properties.
3212 * systemd will not generate or install static dbus
3213 introspection data anymore to /usr/share/dbus-1/interfaces,
3214 as the precise format of these files is unclear, and
3215 nothing makes use of it.
3217 * A proxy daemon is now provided to proxy clients connecting
3218 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3219 compatibility with classic D-Bus.
3221 * A bus driver implementation has been added that supports the
3222 classic D-Bus bus driver calls on kdbus, also for
3223 compatibility purposes.
3225 * A new API "sd-event.h" has been added that implements a
3226 minimal event loop API built around epoll. It provides a
3227 couple of features that direct epoll usage is lacking:
3228 prioritization of events, scales to large numbers of timer
3229 events, per-event timer slack (accuracy), system-wide
3230 coalescing of timer events, exit handlers, watchdog
3231 supervision support using systemd's sd_notify() API, child
3234 * A new API "sd-rntl.h" has been added that provides an API
3235 around the route netlink interface of the kernel, similar in
3236 style to "sd-bus.h".
3238 * A new API "sd-dhcp-client.h" has been added that provides a
3239 small DHCPv4 client-side implementation. This is used by
3242 * There is a new kernel command line option
3243 "systemd.restore_state=0|1". When set to "0", none of the
3244 systemd tools will restore saved runtime state to hardware
3245 devices. More specifically, the rfkill and backlight states
3248 * The FsckPassNo= compatibility option in mount/service units
3249 has been removed. The fstab generator will now add the
3250 necessary dependencies automatically, and does not require
3251 PID1's support for that anymore.
3253 * journalctl gained a new switch, --list-boots, that lists
3254 recent boots with their times and boot IDs.
3256 * The various tools like systemctl, loginctl, timedatectl,
3257 busctl, systemd-run, ... have gained a new switch "-M" to
3258 connect to a specific, local OS container (as direct
3259 connection, without requiring SSH). This works on any
3260 container that is registered with machined, such as those
3261 created by libvirt-lxc or nspawn.
3263 * systemd-run and systemd-analyze also gained support for "-H"
3264 to connect to remote hosts via SSH. This is particularly
3265 useful for systemd-run because it enables queuing of jobs
3266 onto remote systems.
3268 * machinectl gained a new command "login" to open a getty
3269 login in any local container. This works with any container
3270 that is registered with machined (such as those created by
3271 libvirt-lxc or nspawn), and which runs systemd inside.
3273 * machinectl gained a new "reboot" command that may be used to
3274 trigger a reboot on a specific container that is registered
3275 with machined. This works on any container that runs an init
3276 system of some kind.
3278 * systemctl gained a new "list-timers" command to print a nice
3279 listing of installed timer units with the times they elapse
3282 * Alternative reboot() parameters may now be specified on the
3283 "systemctl reboot" command line and are passed to the
3284 reboot() system call.
3286 * systemctl gained a new --job-mode= switch to configure the
3287 mode to queue a job with. This is a more generic version of
3288 --fail, --irreversible, and --ignore-dependencies, which are
3289 still available but not advertised anymore.
3291 * /etc/systemd/system.conf gained new settings to configure
3292 various default timeouts of units, as well as the default
3293 start limit interval and burst. These may still be overridden
3296 * PID1 will now export on the bus profile data of the security
3297 policy upload process (such as the SELinux policy upload to
3300 * journald: when forwarding logs to the console, include
3301 timestamps (following the setting in
3302 /sys/module/printk/parameters/time).
3304 * OnCalendar= in timer units now understands the special
3305 strings "yearly" and "annually". (Both are equivalent)
3307 * The accuracy of timer units is now configurable with the new
3308 AccuracySec= setting. It defaults to 1min.
3310 * A new dependency type JoinsNamespaceOf= has been added that
3311 allows running two services within the same /tmp and network
3312 namespace, if PrivateNetwork= or PrivateTmp= are used.
3314 * A new command "cat" has been added to systemctl. It outputs
3315 the original unit file of a unit, and concatenates the
3316 contents of additional "drop-in" unit file snippets, so that
3317 the full configuration is shown.
3319 * systemctl now supports globbing on the various "list-xyz"
3320 commands, like "list-units" or "list-sockets", as well as on
3321 those commands which take multiple unit names.
3323 * journalctl's --unit= switch gained support for globbing.
3325 * All systemd daemons now make use of the watchdog logic so
3326 that systemd automatically notices when they hang.
3328 * If the $container_ttys environment variable is set,
3329 getty-generator will automatically spawn a getty for each
3330 listed tty. This is useful for container managers to request
3331 login gettys to be spawned on as many ttys as needed.
3333 * %h, %s, %U specifier support is not available anymore when
3334 used in unit files for PID 1. This is because NSS calls are
3335 not safe from PID 1. They stay available for --user
3336 instances of systemd, and as special case for the root user.
3338 * loginctl gained a new "--no-legend" switch to turn off output
3341 * The "sd-login.h" API gained three new calls:
3342 sd_session_is_remote(), sd_session_get_remote_user(),
3343 sd_session_get_remote_host() to query information about
3346 * The udev hardware database now also carries vendor/product
3347 information of SDIO devices.
3349 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3350 determine whether watchdog notifications are requested by
3353 * Socket-activated per-connection services now include a
3354 short description of the connection parameters in the
3357 * tmpfiles gained a new "--boot" option. When this is not used,
3358 only lines where the command character is not suffixed with
3359 "!" are executed. When this option is specified, those
3360 options are executed too. This partitions tmpfiles
3361 directives into those that can be safely executed at any
3362 time, and those which should be run only at boot (for
3363 example, a line that creates /run/nologin).
3365 * A new API "sd-resolve.h" has been added which provides a simple
3366 asynchronous wrapper around glibc NSS host name resolution
3367 calls, such as getaddrinfo(). In contrast to glibc's
3368 getaddrinfo_a(), it does not use signals. In contrast to most
3369 other asynchronous name resolution libraries, this one does
3370 not reimplement DNS, but reuses NSS, so that alternate
3371 host name resolution systems continue to work, such as mDNS,
3372 LDAP, etc. This API is based on libasyncns, but it has been
3373 cleaned up for inclusion in systemd.
3375 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3376 "sd-daemon.h" are no longer found in individual libraries
3377 libsystemd-journal.so, libsystemd-login.so,
3378 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3379 merged them into a single library, libsystemd.so, which
3380 provides all symbols. The reason for this is cyclic
3381 dependencies, as these libraries tend to use each other's
3382 symbols. So far, we have managed to workaround that by linking
3383 a copy of a good part of our code into each of these
3384 libraries again and again, which, however, makes certain
3385 things hard to do, like sharing static variables. Also, it
3386 substantially increases footprint. With this change, there
3387 is only one library for the basic APIs systemd
3388 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3389 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3390 library as well, however are subject to the --enable-kdbus
3391 switch (see below). Note that "sd-dhcp-client.h" is not part
3392 of this library (this is because it only consumes, never
3393 provides, services of/to other APIs). To make the transition
3394 easy from the separate libraries to the unified one, we
3395 provide the --enable-compat-libs compile-time switch which
3396 will generate stub libraries that are compatible with the
3397 old ones but redirect all calls to the new one.
3399 * All of the kdbus logic and the new APIs "sd-bus.h",
3400 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3401 and "sd-utf8.h" are compile-time optional via the
3402 "--enable-kdbus" switch, and they are not compiled in by
3403 default. To make use of kdbus, you have to explicitly enable
3404 the switch. Note however, that neither the kernel nor the
3405 userspace API for all of this is considered stable yet. We
3406 want to maintain the freedom to still change the APIs for
3407 now. By specifying this build-time switch, you acknowledge
3408 that you are aware of the instability of the current
3411 * Also, note that while kdbus is pretty much complete,
3412 it lacks one thing: proper policy support. This means you
3413 can build a fully working system with all features; however,
3414 it will be highly insecure. Policy support will be added in
3415 one of the next releases, at the same time that we will
3416 declare the APIs stable.
3418 * When the kernel command line argument "kdbus" is specified,
3419 systemd will automatically load the kdbus.ko kernel module. At
3420 this stage of development, it is only useful for testing kdbus
3421 and should not be used in production. Note: if "--enable-kdbus"
3422 is specified, and the kdbus.ko kernel module is available, and
3423 "kdbus" is added to the kernel command line, the entire system
3424 runs with kdbus instead of dbus-daemon, with the above mentioned
3425 problem of missing the system policy enforcement. Also a future
3426 version of kdbus.ko or a newer systemd will not be compatible with
3427 each other, and will unlikely be able to boot the machine if only
3428 one of them is updated.
3430 * systemctl gained a new "import-environment" command which
3431 uploads the caller's environment (or parts thereof) into the
3432 service manager so that it is inherited by services started
3433 by the manager. This is useful to upload variables like
3434 $DISPLAY into the user service manager.
3436 * A new PrivateDevices= switch has been added to service units
3437 which allows running a service with a namespaced /dev
3438 directory that does not contain any device nodes for
3439 physical devices. More specifically, it only includes devices
3440 such as /dev/null, /dev/urandom, and /dev/zero which are API
3443 * logind has been extended to support behaviour like VT
3444 switching on seats that do not support a VT. This makes
3445 multi-session available on seats that are not the first seat
3446 (seat0), and on systems where kernel support for VTs has
3447 been disabled at compile-time.
3449 * If a process holds a delay lock for system sleep or shutdown
3450 and fails to release it in time, we will now log its
3451 identity. This makes it easier to identify processes that
3452 cause slow suspends or power-offs.
3454 * When parsing /etc/crypttab, support for a new key-slot=
3455 option as supported by Debian is added. It allows indicating
3456 which LUKS slot to use on disk, speeding up key loading.
3458 * The sd_journald_sendv() API call has been checked and
3459 officially declared to be async-signal-safe so that it may
3460 be invoked from signal handlers for logging purposes.
3462 * Boot-time status output is now enabled automatically after a
3463 short timeout if boot does not progress, in order to give
3464 the user an indication what she or he is waiting for.
3466 * The boot-time output has been improved to show how much time
3467 remains until jobs expire.
3469 * The KillMode= switch in service units gained a new possible
3470 value "mixed". If set, and the unit is shut down, then the
3471 initial SIGTERM signal is sent only to the main daemon
3472 process, while the following SIGKILL signal is sent to
3473 all remaining processes of the service.
3475 * When a scope unit is registered, a new property "Controller"
3476 may be set. If set to a valid bus name, systemd will send a
3477 RequestStop() signal to this name when it would like to shut
3478 down the scope. This may be used to hook manager logic into
3479 the shutdown logic of scope units. Also, scope units may now
3480 be put in a special "abandoned" state, in which case the
3481 manager process which created them takes no further
3482 responsibilities for it.
3484 * When reading unit files, systemd will now verify
3485 the access mode of these files, and warn about certain
3486 suspicious combinations. This has been added to make it
3487 easier to track down packaging bugs where unit files are
3488 marked executable or world-writable.
3490 * systemd-nspawn gained a new "--setenv=" switch to set
3491 container-wide environment variables. The similar option in
3492 systemd-activate was renamed from "--environment=" to
3493 "--setenv=" for consistency.
3495 * systemd-nspawn has been updated to create a new kdbus domain
3496 for each container that is invoked, thus allowing each
3497 container to have its own set of system and user buses,
3498 independent of the host.
3500 * systemd-nspawn gained a new --drop-capability= switch to run
3501 the container with less capabilities than the default. Both
3502 --drop-capability= and --capability= now take the special
3503 string "all" for dropping or keeping all capabilities.
3505 * systemd-nspawn gained new switches for executing containers
3506 with specific SELinux labels set.
3508 * systemd-nspawn gained a new --quiet switch to not generate
3509 any additional output but the container's own console
3512 * systemd-nspawn gained a new --share-system switch to run a
3513 container without PID namespacing enabled.
3515 * systemd-nspawn gained a new --register= switch to control
3516 whether the container is registered with systemd-machined or
3517 not. This is useful for containers that do not run full
3518 OS images, but only specific apps.
3520 * systemd-nspawn gained a new --keep-unit which may be used
3521 when invoked as the only program from a service unit, and
3522 results in registration of the unit service itself in
3523 systemd-machined, instead of a newly opened scope unit.
3525 * systemd-nspawn gained a new --network-interface= switch for
3526 moving arbitrary interfaces to the container. The new
3527 --network-veth switch creates a virtual Ethernet connection
3528 between host and container. The new --network-bridge=
3529 switch then allows assigning the host side of this virtual
3530 Ethernet connection to a bridge device.
3532 * systemd-nspawn gained a new --personality= switch for
3533 setting the kernel personality for the container. This is
3534 useful when running a 32-bit container on a 64-bit host. A
3535 similar option Personality= is now also available for service
3538 * logind will now also track a "Desktop" identifier for each
3539 session which encodes the desktop environment of it. This is
3540 useful for desktop environments that want to identify
3541 multiple running sessions of itself easily.
3543 * A new SELinuxContext= setting for service units has been
3544 added that allows setting a specific SELinux execution
3545 context for a service.
3547 * Most systemd client tools will now honour $SYSTEMD_LESS for
3548 settings of the "less" pager. By default, these tools will
3549 override $LESS to allow certain operations to work, such as
3550 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3551 influence this logic.
3553 * systemd's "seccomp" hook-up has been changed to make use of
3554 the libseccomp library instead of using its own
3555 implementation. This has benefits for portability among
3558 * For usage together with SystemCallFilter=, a new
3559 SystemCallErrorNumber= setting has been introduced that
3560 allows configuration of a system error number to be returned
3561 on filtered system calls, instead of immediately killing the
3562 process. Also, SystemCallArchitectures= has been added to
3563 limit access to system calls of a particular architecture
3564 (in order to turn off support for unused secondary
3565 architectures). There is also a global
3566 SystemCallArchitectures= setting in system.conf now to turn
3567 off support for non-native system calls system-wide.
3569 * systemd requires a kernel with a working name_to_handle_at(),
3570 please see the kernel config requirements in the README file.
3572 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3573 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3574 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3575 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3576 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3577 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3578 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3579 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3580 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3581 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3582 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3583 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3584 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3585 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3586 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3587 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3588 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3589 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3590 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3591 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3592 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3593 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3594 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3595 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3597 -- Berlin, 2014-02-20
3601 * logind has gained support for facilitating privileged input
3602 and drm device access for unprivileged clients. This work is
3603 useful to allow Wayland display servers (and similar
3604 programs, such as kmscon) to run under the user's ID and
3605 access input and drm devices which are normally
3606 protected. When this is used (and the kernel is new enough)
3607 logind will "mute" IO on the file descriptors passed to
3608 Wayland as long as it is in the background and "unmute" it
3609 if it returns into the foreground. This allows secure
3610 session switching without allowing background sessions to
3611 eavesdrop on input and display data. This also introduces
3612 session switching support if VT support is turned off in the
3613 kernel, and on seats that are not seat0.
3615 * A new kernel command line option luks.options= is understood
3616 now which allows specifying LUKS options for usage for LUKS
3617 encrypted partitions specified with luks.uuid=.
3619 * tmpfiles.d(5) snippets may now use specifier expansion in
3620 path names. More specifically %m, %b, %H, %v, are now
3621 replaced by the local machine id, boot id, hostname, and
3622 kernel version number.
3624 * A new tmpfiles.d(5) command "m" has been introduced which
3625 may be used to change the owner/group/access mode of a file
3626 or directory if it exists, but do nothing if it does not.
3628 * This release removes high-level support for the
3629 MemorySoftLimit= cgroup setting. The underlying kernel
3630 cgroup attribute memory.soft_limit= is currently badly
3631 designed and likely to be removed from the kernel API in its
3632 current form, hence we should not expose it for now.
3634 * The memory.use_hierarchy cgroup attribute is now enabled for
3635 all cgroups systemd creates in the memory cgroup
3636 hierarchy. This option is likely to be come the built-in
3637 default in the kernel anyway, and the non-hierarchical mode
3638 never made much sense in the intrinsically hierarchical
3641 * A new field _SYSTEMD_SLICE= is logged along with all journal
3642 messages containing the slice a message was generated
3643 from. This is useful to allow easy per-customer filtering of
3644 logs among other things.
3646 * systemd-journald will no longer adjust the group of journal
3647 files it creates to the "systemd-journal" group. Instead we
3648 rely on the journal directory to be owned by the
3649 "systemd-journal" group, and its setgid bit set, so that the
3650 kernel file system layer will automatically enforce that
3651 journal files inherit this group assignment. The reason for
3652 this change is that we cannot allow NSS look-ups from
3653 journald which would be necessary to resolve
3654 "systemd-journal" to a numeric GID, because this might
3655 create deadlocks if NSS involves synchronous queries to
3656 other daemons (such as nscd, or sssd) which in turn are
3657 logging clients of journald and might block on it, which
3658 would then dead lock. A tmpfiles.d(5) snippet included in
3659 systemd will make sure the setgid bit and group are
3660 properly set on the journal directory if it exists on every
3661 boot. However, we recommend adjusting it manually after
3662 upgrades too (or from RPM scriptlets), so that the change is
3663 not delayed until next reboot.
3665 * Backlight and random seed files in /var/lib/ have moved into
3666 the /var/lib/systemd/ directory, in order to centralize all
3667 systemd generated files in one directory.
3669 * Boot time performance measurements (as displayed by
3670 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3671 performance information if that's available to determine how
3672 much time BIOS and boot loader initialization required. With
3673 a sufficiently new BIOS you hence no longer need to boot
3674 with Gummiboot to get access to such information.
3676 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3677 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3678 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3679 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3680 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3681 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3682 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3684 -- Berlin, 2013-10-02
3688 * The Restart= option for services now understands a new
3689 on-watchdog setting, which will restart the service
3690 automatically if the service stops sending out watchdog keep
3691 alive messages (as configured with WatchdogSec=).
3693 * The getty generator (which is responsible for bringing up a
3694 getty on configured serial consoles) will no longer only
3695 start a getty on the primary kernel console but on all
3696 others, too. This makes the order in which console= is
3697 specified on the kernel command line less important.
3699 * libsystemd-logind gained a new sd_session_get_vt() call to
3700 retrieve the VT number of a session.
3702 * If the option "tries=0" is set for an entry of /etc/crypttab
3703 its passphrase is queried indefinitely instead of any
3704 maximum number of tries.
3706 * If a service with a configure PID file terminates its PID
3707 file will now be removed automatically if it still exists
3708 afterwards. This should put an end to stale PID files.
3710 * systemd-run will now also take relative binary path names
3711 for execution and no longer insists on absolute paths.
3713 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3714 paths that are optionally prefixed with "-" to indicate that
3715 it should not be considered a failure if they do not exist.
3717 * journalctl -o (and similar commands) now understands a new
3718 output mode "short-precise", it is similar to "short" but
3719 shows timestamps with usec accuracy.
3721 * The option "discard" (as known from Debian) is now
3722 synonymous to "allow-discards" in /etc/crypttab. In fact,
3723 "discard" is preferred now (since it is easier to remember
3726 * Some licensing clean-ups were made, so that more code is now
3727 LGPL-2.1 licensed than before.
3729 * A minimal tool to save/restore the display backlight
3730 brightness across reboots has been added. It will store the
3731 backlight setting as late as possible at shutdown, and
3732 restore it as early as possible during reboot.
3734 * A logic to automatically discover and enable home and swap
3735 partitions on GPT disks has been added. With this in place
3736 /etc/fstab becomes optional for many setups as systemd can
3737 discover certain partitions located on the root disk
3738 automatically. Home partitions are recognized under their
3739 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3740 partitions are recognized under their GPT type ID
3741 0657fd6da4ab43c484e50933c84b4f4f.
3743 * systemd will no longer pass any environment from the kernel
3744 or initrd to system services. If you want to set an
3745 environment for all services, do so via the kernel command
3746 line systemd.setenv= assignment.
3748 * The systemd-sysctl tool no longer natively reads the file
3749 /etc/sysctl.conf. If desired, the file should be symlinked
3750 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3751 legacy support by a symlink rather than built-in code, it
3752 also makes the otherwise hidden order of application of the
3753 different files visible. (Note that this partly reverts to a
3754 pre-198 application order of sysctl knobs!)
3756 * The "systemctl set-log-level" and "systemctl dump" commands
3757 have been moved to systemd-analyze.
3759 * systemd-run learned the new --remain-after-exit switch,
3760 which causes the scope unit not to be cleaned up
3761 automatically after the process terminated.
3763 * tmpfiles learned a new --exclude-prefix= switch to exclude
3764 certain paths from operation.
3766 * journald will now automatically flush all messages to disk
3767 as soon as a message at the log level CRIT, ALERT or EMERG
3770 Contributions from: Andrew Cook, Brandon Philips, Christian
3771 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3772 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3773 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3774 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3775 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3776 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3777 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3778 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3779 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3780 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3781 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3782 William Giokas, Zbigniew Jędrzejewski-Szmek
3784 -- Berlin, 2013-09-13
3788 * The documentation has been updated to cover the various new
3789 concepts introduced with 205.
3791 * Unit files now understand the new %v specifier which
3792 resolves to the kernel version string as returned by "uname
3795 * systemctl now supports filtering the unit list output by
3796 load state, active state and sub state, using the new
3799 * "systemctl status" will now show the results of the
3800 condition checks (like ConditionPathExists= and similar) of
3801 the last start attempts of the unit. They are also logged to
3804 * "journalctl -b" may now be used to look for boot output of a
3805 specific boot. Try "journalctl -b -1" for the previous boot,
3806 but the syntax is substantially more powerful.
3808 * "journalctl --show-cursor" has been added which prints the
3809 cursor string the last shown log line. This may then be used
3810 with the new "journalctl --after-cursor=" switch to continue
3811 browsing logs from that point on.
3813 * "journalctl --force" may now be used to force regeneration
3816 * Creation of "dead" device nodes has been moved from udev
3817 into kmod and tmpfiles. Previously, udev would read the kmod
3818 databases to pre-generate dead device nodes based on meta
3819 information contained in kernel modules, so that these would
3820 be auto-loaded on access rather then at boot. As this
3821 does not really have much to do with the exposing actual
3822 kernel devices to userspace this has always been slightly
3823 alien in the udev codebase. Following the new scheme kmod
3824 will now generate a runtime snippet for tmpfiles from the
3825 module meta information and it now is tmpfiles' job to the
3826 create the nodes. This also allows overriding access and
3827 other parameters for the nodes using the usual tmpfiles
3828 facilities. As side effect this allows us to remove the
3829 CAP_SYS_MKNOD capability bit from udevd entirely.
3831 * logind's device ACLs may now be applied to these "dead"
3832 devices nodes too, thus finally allowing managed access to
3833 devices such as /dev/snd/sequencer whithout loading the
3834 backing module right-away.
3836 * A new RPM macro has been added that may be used to apply
3837 tmpfiles configuration during package installation.
3839 * systemd-detect-virt and ConditionVirtualization= now can
3840 detect User-Mode-Linux machines (UML).
3842 * journald will now implicitly log the effective capabilities
3843 set of processes in the message metadata.
3845 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3847 * The initrd interface has been simplified (more specifically,
3848 support for passing performance data via environment
3849 variables and fsck results via files in /run has been
3850 removed). These features were non-essential, and are
3851 nowadays available in a much nicer way by having systemd in
3852 the initrd serialize its state and have the hosts systemd
3853 deserialize it again.
3855 * The udev "keymap" data files and tools to apply keyboard
3856 specific mappings of scan to key codes, and force-release
3857 scan code lists have been entirely replaced by a udev
3858 "keyboard" builtin and a hwdb data file.
3860 * systemd will now honour the kernel's "quiet" command line
3861 argument also during late shutdown, resulting in a
3862 completely silent shutdown when used.
3864 * There's now an option to control the SO_REUSEPORT socket
3865 option in .socket units.
3867 * Instance units will now automatically get a per-template
3868 subslice of system.slice unless something else is explicitly
3869 configured. For example, instances of sshd@.service will now
3870 implicitly be placed in system-sshd.slice rather than
3871 system.slice as before.
3873 * Test coverage support may now be enabled at build time.
3875 Contributions from: Dave Reisner, Frederic Crozat, Harald
3876 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3877 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3878 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3879 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3880 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3881 Giokas, Zbigniew Jędrzejewski-Szmek
3883 -- Berlin, 2013-07-23
3887 * Two new unit types have been introduced:
3889 Scope units are very similar to service units, however, are
3890 created out of pre-existing processes -- instead of PID 1
3891 forking off the processes. By using scope units it is
3892 possible for system services and applications to group their
3893 own child processes (worker processes) in a powerful way
3894 which then maybe used to organize them, or kill them
3895 together, or apply resource limits on them.
3897 Slice units may be used to partition system resources in an
3898 hierarchical fashion and then assign other units to them. By
3899 default there are now three slices: system.slice (for all
3900 system services), user.slice (for all user sessions),
3901 machine.slice (for VMs and containers).
3903 Slices and scopes have been introduced primarily in
3904 context of the work to move cgroup handling to a
3905 single-writer scheme, where only PID 1
3906 creates/removes/manages cgroups.
3908 * There's a new concept of "transient" units. In contrast to
3909 normal units these units are created via an API at runtime,
3910 not from configuration from disk. More specifically this
3911 means it is now possible to run arbitrary programs as
3912 independent services, with all execution parameters passed
3913 in via bus APIs rather than read from disk. Transient units
3914 make systemd substantially more dynamic then it ever was,
3915 and useful as a general batch manager.
3917 * logind has been updated to make use of scope and slice units
3918 for managing user sessions. As a user logs in he will get
3919 his own private slice unit, to which all sessions are added
3920 as scope units. We also added support for automatically
3921 adding an instance of user@.service for the user into the
3922 slice. Effectively logind will no longer create cgroup
3923 hierarchies on its own now, it will defer entirely to PID 1
3924 for this by means of scope, service and slice units. Since
3925 user sessions this way become entities managed by PID 1
3926 the output of "systemctl" is now a lot more comprehensive.
3928 * A new mini-daemon "systemd-machined" has been added which
3929 may be used by virtualization managers to register local
3930 VMs/containers. nspawn has been updated accordingly, and
3931 libvirt will be updated shortly. machined will collect a bit
3932 of meta information about the VMs/containers, and assign
3933 them their own scope unit (see above). The collected
3934 meta-data is then made available via the "machinectl" tool,
3935 and exposed in "ps" and similar tools. machined/machinectl
3936 is compile-time optional.
3938 * As discussed earlier, the low-level cgroup configuration
3939 options ControlGroup=, ControlGroupModify=,
3940 ControlGroupPersistent=, ControlGroupAttribute= have been
3941 removed. Please use high-level attribute settings instead as
3942 well as slice units.
3944 * A new bus call SetUnitProperties() has been added to alter
3945 various runtime parameters of a unit. This is primarily
3946 useful to alter cgroup parameters dynamically in a nice way,
3947 but will be extended later on to make more properties
3948 modifiable at runtime. systemctl gained a new set-properties
3949 command that wraps this call.
3951 * A new tool "systemd-run" has been added which can be used to
3952 run arbitrary command lines as transient services or scopes,
3953 while configuring a number of settings via the command
3954 line. This tool is currently very basic, however already
3955 very useful. We plan to extend this tool to even allow
3956 queuing of execution jobs with time triggers from the
3957 command line, similar in fashion to "at".
3959 * nspawn will now inform the user explicitly that kernels with
3960 audit enabled break containers, and suggest the user to turn
3963 * Support for detecting the IMA and AppArmor security
3964 frameworks with ConditionSecurity= has been added.
3966 * journalctl gained a new "-k" switch for showing only kernel
3967 messages, mimicking dmesg output; in addition to "--user"
3968 and "--system" switches for showing only user's own logs
3971 * systemd-delta can now show information about drop-in
3972 snippets extending unit files.
3974 * libsystemd-bus has been substantially updated but is still
3975 not available as public API.
3977 * systemd will now look for the "debug" argument on the kernel
3978 command line and enable debug logging, similar to what
3979 "systemd.log_level=debug" already did before.
3981 * "systemctl set-default", "systemctl get-default" has been
3982 added to configure the default.target symlink, which
3983 controls what to boot into by default.
3985 * "systemctl set-log-level" has been added as a convenient
3986 way to raise and lower systemd logging threshold.
3988 * "systemd-analyze plot" will now show the time the various
3989 generators needed for execution, as well as information
3990 about the unit file loading.
3992 * libsystemd-journal gained a new sd_journal_open_files() call
3993 for opening specific journal files. journactl also gained a
3994 new switch to expose this new functionality. Previously we
3995 only supported opening all files from a directory, or all
3996 files from the system, as opening individual files only is
3997 racy due to journal file rotation.
3999 * systemd gained the new DefaultEnvironment= setting in
4000 /etc/systemd/system.conf to set environment variables for
4003 * If a privileged process logs a journal message with the
4004 OBJECT_PID= field set, then journald will automatically
4005 augment this with additional OBJECT_UID=, OBJECT_GID=,
4006 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4007 system services want to log events about specific client
4008 processes. journactl/systemctl has been updated to make use
4009 of this information if all log messages regarding a specific
4012 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4013 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4014 Reisner, David Coppa, David King, David Strauss, Eelco
4015 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4016 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4017 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4018 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4019 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4020 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4021 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4022 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4023 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4024 Łukasz Stelmach, 장동준
4028 * The Python bindings gained some minimal support for the APIs
4029 exposed by libsystemd-logind.
4031 * ConditionSecurity= gained support for detecting SMACK. Since
4032 this condition already supports SELinux and AppArmor we only
4033 miss IMA for this. Patches welcome!
4035 Contributions from: Karol Lewandowski, Lennart Poettering,
4036 Zbigniew Jędrzejewski-Szmek
4040 * systemd-nspawn will now create /etc/resolv.conf if
4041 necessary, before bind-mounting the host's file onto it.
4043 * systemd-nspawn will now store meta information about a
4044 container on the container's cgroup as extended attribute
4045 fields, including the root directory.
4047 * The cgroup hierarchy has been reworked in many ways. All
4048 objects any of the components systemd creates in the cgroup
4049 tree are now suffixed. More specifically, user sessions are
4050 now placed in cgroups suffixed with ".session", users in
4051 cgroups suffixed with ".user", and nspawn containers in
4052 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4053 names are now escaped in a simple scheme to avoid collision
4054 of userspace object names with kernel filenames. This work
4055 is preparation for making these objects relocatable in the
4056 cgroup tree, in order to allow easy resource partitioning of
4057 these objects without causing naming conflicts.
4059 * systemctl list-dependencies gained the new switches
4060 --plain, --reverse, --after and --before.
4062 * systemd-inhibit now shows the process name of processes that
4063 have taken an inhibitor lock.
4065 * nss-myhostname will now also resolve "localhost"
4066 implicitly. This makes /etc/hosts an optional file and
4067 nicely handles that on IPv6 ::1 maps to both "localhost" and
4070 * libsystemd-logind.so gained a new call
4071 sd_get_machine_names() to enumerate running containers and
4072 VMs (currently only supported by very new libvirt and
4073 nspawn). sd_login_monitor can now be used to watch
4074 VMs/containers coming and going.
4076 * .include is not allowed recursively anymore, and only in
4077 unit files. Usually it is better to use drop-in snippets in
4078 .d/*.conf anyway, as introduced with systemd 198.
4080 * systemd-analyze gained a new "critical-chain" command that
4081 determines the slowest chain of units run during system
4082 boot-up. It is very useful for tracking down where
4083 optimizing boot time is the most beneficial.
4085 * systemd will no longer allow manipulating service paths in
4086 the name=systemd:/system cgroup tree using ControlGroup= in
4087 units. (But is still fine with it in all other dirs.)
4089 * There's a new systemd-nspawn@.service service file that may
4090 be used to easily run nspawn containers as system
4091 services. With the container's root directory in
4092 /var/lib/container/foobar it is now sufficient to run
4093 "systemctl start systemd-nspawn@foobar.service" to boot it.
4095 * systemd-cgls gained a new parameter "--machine" to list only
4096 the processes within a certain container.
4098 * ConditionSecurity= now can check for "apparmor". We still
4099 are lacking checks for SMACK and IMA for this condition
4100 check though. Patches welcome!
4102 * A new configuration file /etc/systemd/sleep.conf has been
4103 added that may be used to configure which kernel operation
4104 systemd is supposed to execute when "suspend", "hibernate"
4105 or "hybrid-sleep" is requested. This makes the new kernel
4106 "freeze" state accessible to the user.
4108 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4109 the passed argument if applicable.
4111 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4112 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4113 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4114 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4115 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4116 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4117 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4122 * The output of 'systemctl list-jobs' got some polishing. The
4123 '--type=' argument may now be passed more than once. A new
4124 command 'systemctl list-sockets' has been added which shows
4125 a list of kernel sockets systemd is listening on with the
4126 socket units they belong to, plus the units these socket
4129 * The experimental libsystemd-bus library got substantial
4130 updates to work in conjunction with the (also experimental)
4131 kdbus kernel project. It works well enough to exchange
4132 messages with some sophistication. Note that kdbus is not
4133 ready yet, and the library is mostly an elaborate test case
4134 for now, and not installable.
4136 * systemd gained a new unit 'systemd-static-nodes.service'
4137 that generates static device nodes earlier during boot, and
4138 can run in conjunction with udev.
4140 * libsystemd-login gained a new call sd_pid_get_user_unit()
4141 to retrieve the user systemd unit a process is running
4142 in. This is useful for systems where systemd is used as
4145 * systemd-nspawn now places all containers in the new /machine
4146 top-level cgroup directory in the name=systemd
4147 hierarchy. libvirt will soon do the same, so that we get a
4148 uniform separation of /system, /user and /machine for system
4149 services, user processes and containers/virtual
4150 machines. This new cgroup hierarchy is also useful to stick
4151 stable names to specific container instances, which can be
4152 recognized later this way (this name may be controlled
4153 via systemd-nspawn's new -M switch). libsystemd-login also
4154 gained a new call sd_pid_get_machine_name() to retrieve the
4155 name of the container/VM a specific process belongs to.
4157 * bootchart can now store its data in the journal.
4159 * libsystemd-journal gained a new call
4160 sd_journal_add_conjunction() for AND expressions to the
4161 matching logic. This can be used to express more complex
4162 logical expressions.
4164 * journactl can now take multiple --unit= and --user-unit=
4167 * The cryptsetup logic now understands the "luks.key=" kernel
4168 command line switch for specifying a file to read the
4169 decryption key from. Also, if a configured key file is not
4170 found the tool will now automatically fall back to prompting
4173 * Python systemd.journal module was updated to wrap recently
4174 added functions from libsystemd-journal. The interface was
4175 changed to bring the low level interface in s.j._Reader
4176 closer to the C API, and the high level interface in
4177 s.j.Reader was updated to wrap and convert all data about
4180 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4181 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4182 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4183 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4184 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4185 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4189 * journalctl --update-catalog now understands a new --root=
4190 option to operate on catalogs found in a different root
4193 * During shutdown after systemd has terminated all running
4194 services a final killing loop kills all remaining left-over
4195 processes. We will now print the name of these processes
4196 when we send SIGKILL to them, since this usually indicates a
4199 * If /etc/crypttab refers to password files stored on
4200 configured mount points automatic dependencies will now be
4201 generated to ensure the specific mount is established first
4202 before the key file is attempted to be read.
4204 * 'systemctl status' will now show information about the
4205 network sockets a socket unit is listening on.
4207 * 'systemctl status' will also shown information about any
4208 drop-in configuration file for units. (Drop-In configuration
4209 files in this context are files such as
4210 /etc/systemd/systemd/foobar.service.d/*.conf)
4212 * systemd-cgtop now optionally shows summed up CPU times of
4213 cgroups. Press '%' while running cgtop to switch between
4214 percentage and absolute mode. This is useful to determine
4215 which cgroups use up the most CPU time over the entire
4216 runtime of the system. systemd-cgtop has also been updated
4217 to be 'pipeable' for processing with further shell tools.
4219 * 'hostnamectl set-hostname' will now allow setting of FQDN
4222 * The formatting and parsing of time span values has been
4223 changed. The parser now understands fractional expressions
4224 such as "5.5h". The formatter will now output fractional
4225 expressions for all time spans under 1min, i.e. "5.123456s"
4226 rather than "5s 123ms 456us". For time spans under 1s
4227 millisecond values are shown, for those under 1ms
4228 microsecond values are shown. This should greatly improve
4229 all time-related output of systemd.
4231 * libsystemd-login and libsystemd-journal gained new
4232 functions for querying the poll() events mask and poll()
4233 timeout value for integration into arbitrary event
4236 * localectl gained the ability to list available X11 keymaps
4237 (models, layouts, variants, options).
4239 * 'systemd-analyze dot' gained the ability to filter for
4240 specific units via shell-style globs, to create smaller,
4241 more useful graphs. I.e. it is now possible to create simple
4242 graphs of all the dependencies between only target units, or
4243 of all units that Avahi has dependencies with.
4245 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4246 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4247 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4248 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4249 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4250 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4251 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4255 * The boot-time readahead implementation for rotating media
4256 will now read the read-ahead data in multiple passes which
4257 consist of all read requests made in equidistant time
4258 intervals. This means instead of strictly reading read-ahead
4259 data in its physical order on disk we now try to find a
4260 middle ground between physical and access time order.
4262 * /etc/os-release files gained a new BUILD_ID= field for usage
4263 on operating systems that provide continuous builds of OS
4266 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4267 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4268 William Douglas, Zbigniew Jędrzejewski-Szmek
4272 * systemd-python gained an API exposing libsystemd-daemon.
4274 * The SMACK setup logic gained support for uploading CIPSO
4277 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4278 ReadOnlyDirectories= and InaccessibleDirectories= has
4279 changed. The private /tmp and /var/tmp directories are now
4280 shared by all processes of a service (which means
4281 ExecStartPre= may now leave data in /tmp that ExecStart= of
4282 the same service can still access). When a service is
4283 stopped its temporary directories are immediately deleted
4284 (normal clean-up with tmpfiles is still done in addition to
4287 * By default, systemd will now set a couple of sysctl
4288 variables in the kernel: the safe sysrq options are turned
4289 on, IP route verification is turned on, and source routing
4290 disabled. The recently added hardlink and softlink
4291 protection of the kernel is turned on. These settings should
4292 be reasonably safe, and good defaults for all new systems.
4294 * The predictable network naming logic may now be turned off
4295 with a new kernel command line switch: net.ifnames=0.
4297 * A new libsystemd-bus module has been added that implements a
4298 pretty complete D-Bus client library. For details see:
4300 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4302 * journald will now explicitly flush the journal files to disk
4303 at the latest 5min after each write. The file will then also
4304 be marked offline until the next write. This should increase
4305 reliability in case of a crash. The synchronization delay
4306 can be configured via SyncIntervalSec= in journald.conf.
4308 * There's a new remote-fs-setup.target unit that can be used
4309 to pull in specific services when at least one remote file
4310 system is to be mounted.
4312 * There are new targets timers.target and paths.target as
4313 canonical targets to pull user timer and path units in
4314 from. This complements sockets.target with a similar
4315 purpose for socket units.
4317 * libudev gained a new call udev_device_set_attribute_value()
4318 to set sysfs attributes of a device.
4320 * The udev daemon now sets the default number of worker
4321 processes executed in parallel based on the number of available
4322 CPUs instead of the amount of available RAM. This is supposed
4323 to provide a more reliable default and limit a too aggressive
4324 paralellism for setups with 1000s of devices connected.
4326 Contributions from: Auke Kok, Colin Walters, Cristian
4327 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4328 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4329 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4330 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4331 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4332 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4333 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4334 Zbigniew Jędrzejewski-Szmek
4338 * Configuration of unit files may now be extended via drop-in
4339 files without having to edit/override the unit files
4340 themselves. More specifically, if the administrator wants to
4341 change one value for a service file foobar.service he can
4342 now do so by dropping in a configuration snippet into
4343 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4344 will load all these snippets and apply them on top of the
4345 main unit configuration file, possibly extending or
4346 overriding its settings. Using these drop-in snippets is
4347 generally nicer than the two earlier options for changing
4348 unit files locally: copying the files from
4349 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4350 them there; or creating a new file in /etc/systemd/system/
4351 that incorporates the original one via ".include". Drop-in
4352 snippets into these .d/ directories can be placed in any
4353 directory systemd looks for units in, and the usual
4354 overriding semantics between /usr/lib, /etc and /run apply
4357 * Most unit file settings which take lists of items can now be
4358 reset by assigning the empty string to them. For example,
4359 normally, settings such as Environment=FOO=BAR append a new
4360 environment variable assignment to the environment block,
4361 each time they are used. By assigning Environment= the empty
4362 string the environment block can be reset to empty. This is
4363 particularly useful with the .d/*.conf drop-in snippets
4364 mentioned above, since this adds the ability to reset list
4365 settings from vendor unit files via these drop-ins.
4367 * systemctl gained a new "list-dependencies" command for
4368 listing the dependencies of a unit recursively.
4370 * Inhibitors are now honored and listed by "systemctl
4371 suspend", "systemctl poweroff" (and similar) too, not only
4372 GNOME. These commands will also list active sessions by
4375 * Resource limits (as exposed by the various control group
4376 controllers) can now be controlled dynamically at runtime
4377 for all units. More specifically, you can now use a command
4378 like "systemctl set-cgroup-attr foobar.service cpu.shares
4379 2000" to alter the CPU shares a specific service gets. These
4380 settings are stored persistently on disk, and thus allow the
4381 administrator to easily adjust the resource usage of
4382 services with a few simple commands. This dynamic resource
4383 management logic is also available to other programs via the
4384 bus. Almost any kernel cgroup attribute and controller is
4387 * systemd-vconsole-setup will now copy all font settings to
4388 all allocated VTs, where it previously applied them only to
4391 * libsystemd-login gained the new sd_session_get_tty() API
4394 * This release drops support for a few legacy or
4395 distribution-specific LSB facility names when parsing init
4396 scripts: $x-display-manager, $mail-transfer-agent,
4397 $mail-transport-agent, $mail-transfer-agent, $smtp,
4398 $null. Also, the mail-transfer-agent.target unit backing
4399 this has been removed. Distributions which want to retain
4400 compatibility with this should carry the burden for
4401 supporting this themselves and patch support for these back
4402 in, if they really need to. Also, the facilities $syslog and
4403 $local_fs are now ignored, since systemd does not support
4404 early-boot LSB init scripts anymore, and these facilities
4405 are implied anyway for normal services. syslog.target has
4408 * There are new bus calls on PID1's Manager object for
4409 cancelling jobs, and removing snapshot units. Previously,
4410 both calls were only available on the Job and Snapshot
4413 * systemd-journal-gatewayd gained SSL support.
4415 * The various "environment" files, such as /etc/locale.conf
4416 now support continuation lines with a backslash ("\") as
4417 last character in the line, similarly in style (but different)
4418 to how this is supported in shells.
4420 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4421 now implicitly appended to every log entry logged. systemctl
4422 has been updated to filter by this field when operating on a
4423 user systemd instance.
4425 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4426 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4427 the container. This makes it easier to boot unmodified
4428 Fedora systems in a container, which however still requires
4429 audit=0 to be passed on the kernel command line. Auditing in
4430 kernel and userspace is unfortunately still too broken in
4431 context of containers, hence we recommend compiling it out
4432 of the kernel or using audit=0. Hopefully this will be fixed
4433 one day for good in the kernel.
4435 * nspawn gained the new --bind= and --bind-ro= parameters to
4436 bind mount specific directories from the host into the
4439 * nspawn will now mount its own devpts file system instance
4440 into the container, in order not to leak pty devices from
4441 the host into the container.
4443 * systemd will now read the firmware boot time performance
4444 information from the EFI variables, if the used boot loader
4445 supports this, and takes it into account for boot performance
4446 analysis via "systemd-analyze". This is currently supported
4447 only in conjunction with Gummiboot, but could be supported
4448 by other boot loaders too. For details see:
4450 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4452 * A new generator has been added that automatically mounts the
4453 EFI System Partition (ESP) to /boot, if that directory
4454 exists, is empty, and no other file system has been
4455 configured to be mounted there.
4457 * logind will now send out PrepareForSleep(false) out
4458 unconditionally, after coming back from suspend. This may be
4459 used by applications as asynchronous notification for
4460 system resume events.
4462 * "systemctl unlock-sessions" has been added, that allows
4463 unlocking the screens of all user sessions at once, similar
4464 to how "systemctl lock-sessions" already locked all users
4465 sessions. This is backed by a new D-Bus call UnlockSessions().
4467 * "loginctl seat-status" will now show the master device of a
4468 seat. (i.e. the device of a seat that needs to be around for
4469 the seat to be considered available, usually the graphics
4472 * tmpfiles gained a new "X" line type, that allows
4473 configuration of files and directories (with wildcards) that
4474 shall be excluded from automatic cleanup ("aging").
4476 * udev default rules set the device node permissions now only
4477 at "add" events, and do not change them any longer with a
4478 later "change" event.
4480 * The log messages for lid events and power/sleep keypresses
4481 now carry a message ID.
4483 * We now have a substantially larger unit test suite, but this
4484 continues to be work in progress.
4486 * udevadm hwdb gained a new --root= parameter to change the
4487 root directory to operate relative to.
4489 * logind will now issue a background sync() request to the kernel
4490 early at shutdown, so that dirty buffers are flushed to disk early
4491 instead of at the last moment, in order to optimize shutdown
4494 * A new bootctl tool has been added that is an interface for
4495 certain boot loader operations. This is currently a preview
4496 and is likely to be extended into a small mechanism daemon
4497 like timedated, localed, hostnamed, and can be used by
4498 graphical UIs to enumerate available boot options, and
4499 request boot into firmware operations.
4501 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4502 the rest of the package. It also has been updated to work
4503 correctly in initrds.
4505 * Policykit previously has been runtime optional, and is now
4506 also compile time optional via a configure switch.
4508 * systemd-analyze has been reimplemented in C. Also "systemctl
4509 dot" has moved into systemd-analyze.
4511 * "systemctl status" with no further parameters will now print
4512 the status of all active or failed units.
4514 * Operations such as "systemctl start" can now be executed
4515 with a new mode "--irreversible" which may be used to queue
4516 operations that cannot accidentally be reversed by a later
4517 job queuing. This is by default used to make shutdown
4518 requests more robust.
4520 * The Python API of systemd now gained a new module for
4521 reading journal files.
4523 * A new tool kernel-install has been added that can install
4524 kernel images according to the Boot Loader Specification:
4526 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4528 * Boot time console output has been improved to provide
4529 animated boot time output for hanging jobs.
4531 * A new tool systemd-activate has been added which can be used
4532 to test socket activation with, directly from the command
4533 line. This should make it much easier to test and debug
4534 socket activation in daemons.
4536 * journalctl gained a new "--reverse" (or -r) option to show
4537 journal output in reverse order (i.e. newest line first).
4539 * journalctl gained a new "--pager-end" (or -e) option to jump
4540 to immediately jump to the end of the journal in the
4541 pager. This is only supported in conjunction with "less".
4543 * journalctl gained a new "--user-unit=" option, that works
4544 similarly to "--unit=" but filters for user units rather than
4547 * A number of unit files to ease adoption of systemd in
4548 initrds has been added. This moves some minimal logic from
4549 the various initrd implementations into systemd proper.
4551 * The journal files are now owned by a new group
4552 "systemd-journal", which exists specifically to allow access
4553 to the journal, and nothing else. Previously, we used the
4554 "adm" group for that, which however possibly covers more
4555 than just journal/log file access. This new group is now
4556 already used by systemd-journal-gatewayd to ensure this
4557 daemon gets access to the journal files and as little else
4558 as possible. Note that "make install" will also set FS ACLs
4559 up for /var/log/journal to give "adm" and "wheel" read
4560 access to it, in addition to "systemd-journal" which owns
4561 the journal files. We recommend that packaging scripts also
4562 add read access to "adm" + "wheel" to /var/log/journal, and
4563 all existing/future journal files. To normal users and
4564 administrators little changes, however packagers need to
4565 ensure to create the "systemd-journal" system group at
4566 package installation time.
4568 * The systemd-journal-gatewayd now runs as unprivileged user
4569 systemd-journal-gateway:systemd-journal-gateway. Packaging
4570 scripts need to create these system user/group at
4573 * timedated now exposes a new boolean property CanNTP that
4574 indicates whether a local NTP service is available or not.
4576 * systemd-detect-virt will now also detect xen PVs
4578 * The pstore file system is now mounted by default, if it is
4581 * In addition to the SELinux and IMA policies we will now also
4582 load SMACK policies at early boot.
4584 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4585 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4586 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4587 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4588 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4589 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4590 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4591 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4592 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4593 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4594 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4595 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4596 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4597 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4601 * Timer units now support calendar time events in addition to
4602 monotonic time events. That means you can now trigger a unit
4603 based on a calendar time specification such as "Thu,Fri
4604 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4605 or fifth day of any month of the year 2013, given that it is
4606 a thursday or friday. This brings timer event support
4607 considerably closer to cron's capabilities. For details on
4608 the supported calendar time specification language see
4611 * udev now supports a number of different naming policies for
4612 network interfaces for predictable names, and a combination
4613 of these policies is now the default. Please see this wiki
4614 document for details:
4616 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4618 * Auke Kok's bootchart implementation has been added to the
4619 systemd tree. It is an optional component that can graph the
4620 boot in quite some detail. It is one of the best bootchart
4621 implementations around and minimal in its code and
4624 * nss-myhostname has been integrated into the systemd source
4625 tree. nss-myhostname guarantees that the local hostname
4626 always stays resolvable via NSS. It has been a weak
4627 requirement of systemd-hostnamed since a long time, and
4628 since its code is actually trivial we decided to just
4629 include it in systemd's source tree. It can be turned off
4630 with a configure switch.
4632 * The read-ahead logic is now capable of properly detecting
4633 whether a btrfs file system is on SSD or rotating media, in
4634 order to optimize the read-ahead scheme. Previously, it was
4635 only capable of detecting this on traditional file systems
4638 * In udev, additional device properties are now read from the
4639 IAB in addition to the OUI database. Also, Bluetooth company
4640 identities are attached to the devices as well.
4642 * In service files %U may be used as specifier that is
4643 replaced by the configured user name of the service.
4645 * nspawn may now be invoked without a controlling TTY. This
4646 makes it suitable for invocation as its own service. This
4647 may be used to set up a simple containerized server system
4648 using only core OS tools.
4650 * systemd and nspawn can now accept socket file descriptors
4651 when they are started for socket activation. This enables
4652 implementation of socket activated nspawn
4653 containers. i.e. think about autospawning an entire OS image
4654 when the first SSH or HTTP connection is received. We expect
4655 that similar functionality will also be added to libvirt-lxc
4658 * journalctl will now suppress ANSI color codes when
4659 presenting log data.
4661 * systemctl will no longer show control group information for
4662 a unit if a the control group is empty anyway.
4664 * logind can now automatically suspend/hibernate/shutdown the
4667 * /etc/machine-info and hostnamed now also expose the chassis
4668 type of the system. This can be used to determine whether
4669 the local system is a laptop, desktop, handset or
4670 tablet. This information may either be configured by the
4671 user/vendor or is automatically determined from ACPI and DMI
4672 information if possible.
4674 * A number of PolicyKit actions are now bound together with
4675 "imply" rules. This should simplify creating UIs because
4676 many actions will now authenticate similar ones as well.
4678 * Unit files learnt a new condition ConditionACPower= which
4679 may be used to conditionalize a unit depending on whether an
4680 AC power source is connected or not, of whether the system
4681 is running on battery power.
4683 * systemctl gained a new "is-failed" verb that may be used in
4684 shell scripts and suchlike to check whether a specific unit
4685 is in the "failed" state.
4687 * The EnvironmentFile= setting in unit files now supports file
4688 globbing, and can hence be used to easily read a number of
4689 environment files at once.
4691 * systemd will no longer detect and recognize specific
4692 distributions. All distribution-specific #ifdeffery has been
4693 removed, systemd is now fully generic and
4694 distribution-agnostic. Effectively, not too much is lost as
4695 a lot of the code is still accessible via explicit configure
4696 switches. However, support for some distribution specific
4697 legacy configuration file formats has been dropped. We
4698 recommend distributions to simply adopt the configuration
4699 files everybody else uses now and convert the old
4700 configuration from packaging scripts. Most distributions
4701 already did that. If that's not possible or desirable,
4702 distributions are welcome to forward port the specific
4703 pieces of code locally from the git history.
4705 * When logging a message about a unit systemd will now always
4706 log the unit name in the message meta data.
4708 * localectl will now also discover system locale data that is
4709 not stored in locale archives, but directly unpacked.
4711 * logind will no longer unconditionally use framebuffer
4712 devices as seat masters, i.e. as devices that are required
4713 to be existing before a seat is considered preset. Instead,
4714 it will now look for all devices that are tagged as
4715 "seat-master" in udev. By default, framebuffer devices will
4716 be marked as such, but depending on local systems, other
4717 devices might be marked as well. This may be used to
4718 integrate graphics cards using closed source drivers (such
4719 as NVidia ones) more nicely into logind. Note however, that
4720 we recommend using the open source NVidia drivers instead,
4721 and no udev rules for the closed-source drivers will be
4722 shipped from us upstream.
4724 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4725 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4726 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4727 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4728 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4729 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4730 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4731 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4732 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4733 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4734 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4735 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4740 * udev gained support for loading additional device properties
4741 from an indexed database that is keyed by vendor/product IDs
4742 and similar device identifiers. For the beginning this
4743 "hwdb" is populated with data from the well-known PCI and
4744 USB database, but also includes PNP, ACPI and OID data. In
4745 the longer run this indexed database shall grow into
4746 becoming the one central database for non-essential
4747 userspace device metadata. Previously, data from the PCI/USB
4748 database was only attached to select devices, since the
4749 lookup was a relatively expensive operation due to O(n) time
4750 complexity (with n being the number of entries in the
4751 database). Since this is now O(1), we decided to add in this
4752 data for all devices where this is available, by
4753 default. Note that the indexed database needs to be rebuilt
4754 when new data files are installed. To achieve this you need
4755 to update your packaging scripts to invoke "udevadm hwdb
4756 --update" after installation of hwdb data files. For
4757 RPM-based distributions we introduced the new
4758 %udev_hwdb_update macro for this purpose.
4760 * The Journal gained support for the "Message Catalog", an
4761 indexed database to link up additional information with
4762 journal entries. For further details please check:
4764 http://www.freedesktop.org/wiki/Software/systemd/catalog
4766 The indexed message catalog database also needs to be
4767 rebuilt after installation of message catalog files. Use
4768 "journalctl --update-catalog" for this. For RPM-based
4769 distributions we introduced the %journal_catalog_update
4770 macro for this purpose.
4772 * The Python Journal bindings gained support for the standard
4773 Python logging framework.
4775 * The Journal API gained new functions for checking whether
4776 the underlying file system of a journal file is capable of
4777 properly reporting file change notifications, or whether
4778 applications that want to reflect journal changes "live"
4779 need to recheck journal files continuously in appropriate
4782 * It is now possible to set the "age" field for tmpfiles
4783 entries to 0, indicating that files matching this entry
4784 shall always be removed when the directories are cleaned up.
4786 * coredumpctl gained a new "gdb" verb which invokes gdb
4787 right-away on the selected coredump.
4789 * There's now support for "hybrid sleep" on kernels that
4790 support this, in addition to "suspend" and "hibernate". Use
4791 "systemctl hybrid-sleep" to make use of this.
4793 * logind's HandleSuspendKey= setting (and related settings)
4794 now gained support for a new "lock" setting to simply
4795 request the screen lock on all local sessions, instead of
4796 actually executing a suspend or hibernation.
4798 * systemd will now mount the EFI variables file system by
4801 * Socket units now gained support for configuration of the
4802 SMACK security label.
4804 * timedatectl will now output the time of the last and next
4805 daylight saving change.
4807 * We dropped support for various legacy and distro-specific
4808 concepts, such as insserv, early-boot SysV services
4809 (i.e. those for non-standard runlevels such as 'b' or 'S')
4810 or ArchLinux /etc/rc.conf support. We recommend the
4811 distributions who still need support this to either continue
4812 to maintain the necessary patches downstream, or find a
4813 different solution. (Talk to us if you have questions!)
4815 * Various systemd components will now bypass PolicyKit checks
4816 for root and otherwise handle properly if PolicyKit is not
4817 found to be around. This should fix most issues for
4818 PolicyKit-less systems. Quite frankly this should have been
4819 this way since day one. It is absolutely our intention to
4820 make systemd work fine on PolicyKit-less systems, and we
4821 consider it a bug if something does not work as it should if
4822 PolicyKit is not around.
4824 * For embedded systems it is now possible to build udev and
4825 systemd without blkid and/or kmod support.
4827 * "systemctl switch-root" is now capable of switching root
4828 more than once. I.e. in addition to transitions from the
4829 initrd to the host OS it is now possible to transition to
4830 further OS images from the host. This is useful to implement
4831 offline updating tools.
4833 * Various other additions have been made to the RPM macros
4834 shipped with systemd. Use %udev_rules_update() after
4835 installing new udev rules files. %_udevhwdbdir,
4836 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4837 %_sysctldir are now available which resolve to the right
4838 directories for packages to place various data files in.
4840 * journalctl gained the new --full switch (in addition to
4841 --all, to disable ellipsation for long messages.
4843 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4844 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4845 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4846 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4847 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4848 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4849 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4850 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4851 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4855 * journalctl gained new --since= and --until= switches to
4856 filter by time. It also now supports nice filtering for
4857 units via --unit=/-u.
4859 * Type=oneshot services may use ExecReload= and do the
4862 * The journal daemon now supports time-based rotation and
4863 vacuuming, in addition to the usual disk-space based
4866 * The journal will now index the available field values for
4867 each field name. This enables clients to show pretty drop
4868 downs of available match values when filtering. The bash
4869 completion of journalctl has been updated
4870 accordingly. journalctl gained a new switch -F to list all
4871 values a certain field takes in the journal database.
4873 * More service events are now written as structured messages
4874 to the journal, and made recognizable via message IDs.
4876 * The timedated, localed and hostnamed mini-services which
4877 previously only provided support for changing time, locale
4878 and hostname settings from graphical DEs such as GNOME now
4879 also have a minimal (but very useful) text-based client
4880 utility each. This is probably the nicest way to changing
4881 these settings from the command line now, especially since
4882 it lists available options and is fully integrated with bash
4885 * There's now a new tool "systemd-coredumpctl" to list and
4886 extract coredumps from the journal.
4888 * We now install a README each in /var/log/ and
4889 /etc/rc.d/init.d explaining where the system logs and init
4890 scripts went. This hopefully should help folks who go to
4891 that dirs and look into the otherwise now empty void and
4892 scratch their heads.
4894 * When user-services are invoked (by systemd --user) the
4895 $MANAGERPID env var is set to the PID of systemd.
4897 * SIGRTMIN+24 when sent to a --user instance will now result
4898 in immediate termination of systemd.
4900 * gatewayd received numerous feature additions such as a
4901 "follow" mode, for live syncing and filtering.
4903 * browse.html now allows filtering and showing detailed
4904 information on specific entries. Keyboard navigation and
4905 mouse screen support has been added.
4907 * gatewayd/journalctl now supports HTML5/JSON
4908 Server-Sent-Events as output.
4910 * The SysV init script compatibility logic will now
4911 heuristically determine whether a script supports the
4912 "reload" verb, and only then make this available as
4915 * "systemctl status --follow" has been removed, use "journalctl
4918 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4919 have been removed since they are hardly useful to be
4922 * And I'd like to take the opportunity to specifically mention
4923 Zbigniew for his great contributions. Zbigniew, you rock!
4925 Contributions from: Andrew Eikum, Christian Hesse, Colin
4926 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4927 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4928 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4929 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4930 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4931 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4935 * If /etc/vconsole.conf is non-existent or empty we will no
4936 longer load any console font or key map at boot by
4937 default. Instead the kernel defaults will be left
4938 intact. This is definitely the right thing to do, as no
4939 configuration should mean no configuration, and hard-coding
4940 font names that are different on all archs is probably a bad
4941 idea. Also, the kernel default key map and font should be
4942 good enough for most cases anyway, and mostly identical to
4943 the userspace fonts/key maps we previously overloaded them
4944 with. If distributions want to continue to default to a
4945 non-kernel font or key map they should ship a default
4946 /etc/vconsole.conf with the appropriate contents.
4948 Contributions from: Colin Walters, Daniel J Walsh, Dave
4949 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4950 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4954 * journalctl gained a new --cursor= switch to show entries
4955 starting from the specified location in the journal.
4957 * We now enforce a size limit on journal entry fields exported
4958 with "-o json" in journalctl. Fields larger than 4K will be
4959 assigned null. This can be turned off with --all.
4961 * An (optional) journal gateway daemon is now available as
4962 "systemd-journal-gatewayd.service". This service provides
4963 access to the journal via HTTP and JSON. This functionality
4964 will be used to implement live log synchronization in both
4965 pull and push modes, but has various other users too, such
4966 as easy log access for debugging of embedded devices. Right
4967 now it is already useful to retrieve the journal via HTTP:
4969 # systemctl start systemd-journal-gatewayd.service
4970 # wget http://localhost:19531/entries
4972 This will download the journal contents in a
4973 /var/log/messages compatible format. The same as JSON:
4975 # curl -H"Accept: application/json" http://localhost:19531/entries
4977 This service is also accessible via a web browser where a
4978 single static HTML5 app is served that uses the JSON logic
4979 to enable the user to do some basic browsing of the
4980 journal. This will be extended later on. Here's an example
4981 screenshot of this app in its current state:
4983 http://0pointer.de/public/journal-gatewayd
4985 Contributions from: Kay Sievers, Lennart Poettering, Robert
4986 Milasan, Tom Gundersen
4990 * The bash completion logic is now available for journalctl
4993 * We do not mount the "cpuset" controller anymore together with
4994 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4995 started if no parameters are assigned to it. "cpuset" hence
4996 broke code that assumed it it could create "cpu" groups and
4999 * journalctl -f will now subscribe to terminal size changes,
5000 and line break accordingly.
5002 Contributions from: Dave Reisner, Kay Sievers, Lennart
5003 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5007 * nspawn will now create a symlink /etc/localtime in the
5008 container environment, copying the host's timezone
5009 setting. Previously this has been done via a bind mount, but
5010 since symlinks cannot be bind mounted this has now been
5011 changed to create/update the appropriate symlink.
5013 * journalctl -n's line number argument is now optional, and
5014 will default to 10 if omitted.
5016 * journald will now log the maximum size the journal files may
5017 take up on disk. This is particularly useful if the default
5018 built-in logic of determining this parameter from the file
5019 system size is used. Use "systemctl status
5020 systemd-journald.service" to see this information.
5022 * The multi-seat X wrapper tool has been stripped down. As X
5023 is now capable of enumerating graphics devices via udev in a
5024 seat-aware way the wrapper is not strictly necessary
5025 anymore. A stripped down temporary stop-gap is still shipped
5026 until the upstream display managers have been updated to
5027 fully support the new X logic. Expect this wrapper to be
5028 removed entirely in one of the next releases.
5030 * HandleSleepKey= in logind.conf has been split up into
5031 HandleSuspendKey= and HandleHibernateKey=. The old setting
5032 is not available anymore. X11 and the kernel are
5033 distinguishing between these keys and we should too. This
5034 also means the inhibition lock for these keys has been split
5037 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5038 Poettering, Lukas Nykryn, Václav Pavlín
5042 * Whenever a unit changes state we will now log this to the
5043 journal and show along the unit's own log output in
5046 * ConditionPathIsMountPoint= can now properly detect bind
5047 mount points too. (Previously, a bind mount of one file
5048 system to another place in the same file system could not be
5049 detected as mount, since they shared struct stat's st_dev
5052 * We will now mount the cgroup controllers cpu, cpuacct,
5053 cpuset and the controllers net_cls, net_prio together by
5056 * nspawn containers will now have a virtualized boot
5057 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5058 over with a randomized ID at container initialization). This
5059 has the effect of making "journalctl -b" do the right thing
5062 * The JSON output journal serialization has been updated not
5063 to generate "endless" list objects anymore, but rather one
5064 JSON object per line. This is more in line how most JSON
5065 parsers expect JSON objects. The new output mode
5066 "json-pretty" has been added to provide similar output, but
5067 neatly aligned for readability by humans.
5069 * We dropped all explicit sync() invocations in the shutdown
5070 code. The kernel does this implicitly anyway in the kernel
5071 reboot() syscall. halt(8)'s -n option is now a compatibility
5074 * We now support virtualized reboot() in containers, as
5075 supported by newer kernels. We will fall back to exit() if
5076 CAP_SYS_REBOOT is not available to the container. Also,
5077 nspawn makes use of this now and will actually reboot the
5078 container if the containerized OS asks for that.
5080 * journalctl will only show local log output by default
5081 now. Use --merge (-m) to show remote log output, too.
5083 * libsystemd-journal gained the new sd_journal_get_usage()
5084 call to determine the current disk usage of all journal
5085 files. This is exposed in the new "journalctl --disk-usage"
5088 * journald gained a new configuration setting SplitMode= in
5089 journald.conf which may be used to control how user journals
5090 are split off. See journald.conf(5) for details.
5092 * A new condition type ConditionFileNotEmpty= has been added.
5094 * tmpfiles' "w" lines now support file globbing, to write
5095 multiple files at once.
5097 * We added Python bindings for the journal submission
5098 APIs. More Python APIs for a number of selected APIs will
5099 likely follow. Note that we intend to add native bindings
5100 only for the Python language, as we consider it common
5101 enough to deserve bindings shipped within systemd. There are
5102 various projects outside of systemd that provide bindings
5103 for languages such as PHP or Lua.
5105 * Many conditions will now resolve specifiers such as %i. In
5106 addition, PathChanged= and related directives of .path units
5107 now support specifiers as well.
5109 * There's now a new RPM macro definition for the system preset
5112 * journald will now warn if it ca not forward a message to the
5113 syslog daemon because its socket is full.
5115 * timedated will no longer write or process /etc/timezone,
5116 except on Debian. As we do not support late mounted /usr
5117 anymore /etc/localtime always being a symlink is now safe,
5118 and hence the information in /etc/timezone is not necessary
5121 * logind will now always reserve one VT for a text getty (VT6
5122 by default). Previously if more than 6 X sessions where
5123 started they took up all the VTs with auto-spawned gettys,
5124 so that no text gettys were available anymore.
5126 * udev will now automatically inform the btrfs kernel logic
5127 about btrfs RAID components showing up. This should make
5128 simple hotplug based btrfs RAID assembly work.
5130 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5131 (but not for its children which will stay at the kernel
5132 default). This should allow setups with a lot more listening
5135 * systemd will now always pass the configured timezone to the
5136 kernel at boot. timedated will do the same when the timezone
5139 * logind's inhibition logic has been updated. By default,
5140 logind will now handle the lid switch, the power and sleep
5141 keys all the time, even in graphical sessions. If DEs want
5142 to handle these events on their own they should take the new
5143 handle-power-key, handle-sleep-key and handle-lid-switch
5144 inhibitors during their runtime. A simple way to achieve
5145 that is to invoke the DE wrapped in an invocation of:
5147 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5149 * Access to unit operations is now checked via SELinux taking
5150 the unit file label and client process label into account.
5152 * systemd will now notify the administrator in the journal
5153 when he over-mounts a non-empty directory.
5155 * There are new specifiers that are resolved in unit files,
5156 for the host name (%H), the machine ID (%m) and the boot ID
5159 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5160 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5161 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5162 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5163 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5164 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5165 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5169 * Support for reading structured kernel messages from
5170 /dev/kmsg has now been added and is enabled by default.
5172 * Support for reading kernel messages from /proc/kmsg has now
5173 been removed. If you want kernel messages in the journal
5174 make sure to run a recent kernel (>= 3.5) that supports
5175 reading structured messages from /dev/kmsg (see
5176 above). /proc/kmsg is now exclusive property of classic
5177 syslog daemons again.
5179 * The libudev API gained the new
5180 udev_device_new_from_device_id() call.
5182 * The logic for file system namespace (ReadOnlyDirectory=,
5183 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5184 require pivot_root() anymore. This means fewer temporary
5185 directories are created below /tmp for this feature.
5187 * nspawn containers will now see and receive all submounts
5188 made on the host OS below the root file system of the
5191 * Forward Secure Sealing is now supported for Journal files,
5192 which provide cryptographical sealing of journal files so
5193 that attackers cannot alter log history anymore without this
5194 being detectable. Lennart will soon post a blog story about
5195 this explaining it in more detail.
5197 * There are two new service settings RestartPreventExitStatus=
5198 and SuccessExitStatus= which allow configuration of exit
5199 status (exit code or signal) which will be excepted from the
5200 restart logic, resp. consider successful.
5202 * journalctl gained the new --verify switch that can be used
5203 to check the integrity of the structure of journal files and
5204 (if Forward Secure Sealing is enabled) the contents of
5207 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5208 and similar symlinks pre-created. This makes running shells
5209 as container init process a lot more fun.
5211 * The fstab support can now handle PARTUUID= and PARTLABEL=
5214 * A new ConditionHost= condition has been added to match
5215 against the hostname (with globs) and machine ID. This is
5216 useful for clusters where a single OS image is used to
5217 provision a large number of hosts which shall run slightly
5218 different sets of services.
5220 * Services which hit the restart limit will now be placed in a
5223 Contributions from: Bertram Poettering, Dave Reisner, Huang
5224 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5225 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5229 * When running in --user mode systemd will now become a
5230 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5231 tree a lot more organized.
5233 * A new PartOf= unit dependency type has been introduced that
5234 may be used to group services in a natural way.
5236 * "systemctl enable" may now be used to enable instances of
5239 * journalctl now prints error log levels in red, and
5240 warning/notice log levels in bright white. It also supports
5241 filtering by log level now.
5243 * cgtop gained a new -n switch (similar to top), to configure
5244 the maximum number of iterations to run for. It also gained
5245 -b, to run in batch mode (accepting no input).
5247 * The suffix ".service" may now be omitted on most systemctl
5248 command lines involving service unit names.
5250 * There's a new bus call in logind to lock all sessions, as
5251 well as a loginctl verb for it "lock-sessions".
5253 * libsystemd-logind.so gained a new call sd_journal_perror()
5254 that works similar to libc perror() but logs to the journal
5255 and encodes structured information about the error number.
5257 * /etc/crypttab entries now understand the new keyfile-size=
5260 * shutdown(8) now can send a (configurable) wall message when
5261 a shutdown is cancelled.
5263 * The mount propagation mode for the root file system will now
5264 default to "shared", which is useful to make containers work
5265 nicely out-of-the-box so that they receive new mounts from
5266 the host. This can be undone locally by running "mount
5267 --make-rprivate /" if needed.
5269 * The prefdm.service file has been removed. Distributions
5270 should maintain this unit downstream if they intend to keep
5271 it around. However, we recommend writing normal unit files
5272 for display managers instead.
5274 * Since systemd is a crucial part of the OS we will now
5275 default to a number of compiler switches that improve
5276 security (hardening) such as read-only relocations, stack
5277 protection, and suchlike.
5279 * The TimeoutSec= setting for services is now split into
5280 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5281 of individual time outs for the start and the stop phase of
5284 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5285 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5286 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5287 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5288 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5289 Gundersen, Zbigniew Jędrzejewski-Szmek
5293 * The journal and id128 C APIs are now fully documented as man
5296 * Extra safety checks have been added when transitioning from
5297 the initial RAM disk to the main system to avoid accidental
5300 * /etc/crypttab entries now understand the new keyfile-offset=
5303 * systemctl -t can now be used to filter by unit load state.
5305 * The journal C API gained the new sd_journal_wait() call to
5306 make writing synchronous journal clients easier.
5308 * journalctl gained the new -D switch to show journals from a
5311 * journalctl now displays a special marker between log
5312 messages of two different boots.
5314 * The journal is now explicitly flushed to /var via a service
5315 systemd-journal-flush.service, rather than implicitly simply
5316 by seeing /var/log/journal to be writable.
5318 * journalctl (and the journal C APIs) can now match for much
5319 more complex expressions, with alternatives and
5322 * When transitioning from the initial RAM disk to the main
5323 system we will now kill all processes in a killing spree to
5324 ensure no processes stay around by accident.
5326 * Three new specifiers may be used in unit files: %u, %h, %s
5327 resolve to the user name, user home directory resp. user
5328 shell. This is useful for running systemd user instances.
5330 * We now automatically rotate journal files if their data
5331 object hash table gets a fill level > 75%. We also size the
5332 hash table based on the configured maximum file size. This
5333 together should lower hash collisions drastically and thus
5334 speed things up a bit.
5336 * journalctl gained the new "--header" switch to introspect
5337 header data of journal files.
5339 * A new setting SystemCallFilters= has been added to services
5340 which may be used to apply blacklists or whitelists to
5341 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5343 * nspawn gained a new --link-journal= switch (and quicker: -j)
5344 to link the container journal with the host. This makes it
5345 very easy to centralize log viewing on the host for all
5346 guests while still keeping the journal files separated.
5348 * Many bugfixes and optimizations
5350 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5351 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5352 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5357 * Several tools now understand kernel command line arguments,
5358 which are only read when run in an initial RAM disk. They
5359 usually follow closely their normal counterparts, but are
5362 * There's a new tool to analyze the readahead files that are
5363 automatically generated at boot. Use:
5365 /usr/lib/systemd/systemd-readahead analyze /.readahead
5367 * We now provide an early debug shell on tty9 if this enabled. Use:
5369 systemctl enable debug-shell.service
5371 * All plymouth related units have been moved into the Plymouth
5372 package. Please make sure to upgrade your Plymouth version
5375 * systemd-tmpfiles now supports getting passed the basename of
5376 a configuration file only, in which case it will look for it
5377 in all appropriate directories automatically.
5379 * udevadm info now takes a /dev or /sys path as argument, and
5380 does the right thing. Example:
5382 udevadm info /dev/sda
5383 udevadm info /sys/class/block/sda
5385 * systemctl now prints a warning if a unit is stopped but a
5386 unit that might trigger it continues to run. Example: a
5387 service is stopped but the socket that activates it is left
5390 * "systemctl status" will now mention if the log output was
5391 shortened due to rotation since a service has been started.
5393 * The journal API now exposes functions to determine the
5394 "cutoff" times due to rotation.
5396 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5397 immediately flushing of runtime logs to /var if possible,
5398 resp. for triggering immediate rotation of the journal
5401 * It is now considered an error if a service is attempted to
5402 be stopped that is not loaded.
5404 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5406 * systemd-analyze now supports Python 3
5408 * tmpfiles now supports cleaning up directories via aging
5409 where the first level dirs are always kept around but
5410 directories beneath it automatically aged. This is enabled
5411 by prefixing the age field with '~'.
5413 * Seat objects now expose CanGraphical, CanTTY properties
5414 which is required to deal with very fast bootups where the
5415 display manager might be running before the graphics drivers
5416 completed initialization.
5418 * Seat objects now expose a State property.
5420 * We now include RPM macros for service enabling/disabling
5421 based on the preset logic. We recommend RPM based
5422 distributions to make use of these macros if possible. This
5423 makes it simpler to reuse RPM spec files across
5426 * We now make sure that the collected systemd unit name is
5427 always valid when services log to the journal via
5430 * There's a new man page kernel-command-line(7) detailing all
5431 command line options we understand.
5433 * The fstab generator may now be disabled at boot by passing
5434 fstab=0 on the kernel command line.
5436 * A new kernel command line option modules-load= is now understood
5437 to load a specific kernel module statically, early at boot.
5439 * Unit names specified on the systemctl command line are now
5440 automatically escaped as needed. Also, if file system or
5441 device paths are specified they are automatically turned
5442 into the appropriate mount or device unit names. Example:
5444 systemctl status /home
5445 systemctl status /dev/sda
5447 * The SysVConsole= configuration option has been removed from
5448 system.conf parsing.
5450 * The SysV search path is no longer exported on the D-Bus
5453 * The Names= option is been removed from unit file parsing.
5455 * There's a new man page bootup(7) detailing the boot process.
5457 * Every unit and every generator we ship with systemd now
5458 comes with full documentation. The self-explanatory boot is
5461 * A couple of services gained "systemd-" prefixes in their
5462 name if they wrap systemd code, rather than only external
5463 code. Among them fsck@.service which is now
5464 systemd-fsck@.service.
5466 * The HaveWatchdog property has been removed from the D-Bus
5469 * systemd.confirm_spawn= on the kernel command line should now
5472 * There's a new man page crypttab(5) which details all options
5473 we actually understand.
5475 * systemd-nspawn gained a new --capability= switch to pass
5476 additional capabilities to the container.
5478 * timedated will now read known NTP implementation unit names
5479 from /usr/lib/systemd/ntp-units.d/*.list,
5480 systemd-timedated-ntp.target has been removed.
5482 * journalctl gained a new switch "-b" that lists log data of
5483 the current boot only.
5485 * The notify socket is in the abstract namespace again, in
5486 order to support daemons which chroot() at start-up.
5488 * There is a new Storage= configuration option for journald
5489 which allows configuration of where log data should go. This
5490 also provides a way to disable journal logging entirely, so
5491 that data collected is only forwarded to the console, the
5492 kernel log buffer or another syslog implementation.
5494 * Many bugfixes and optimizations
5496 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5497 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5498 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5499 Shawn Landden, Tom Gundersen
5503 * "systemctl help <unit>" now shows the man page if one is
5506 * Several new man pages have been added.
5508 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5509 MaxLevelConsole= can now be specified in
5510 journald.conf. These options allow reducing the amount of
5511 data stored on disk or forwarded by the log level.
5513 * TimerSlackNSec= can now be specified in system.conf for
5514 PID1. This allows system-wide power savings.
5516 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5517 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5522 * logind is now capable of (optionally) handling power and
5523 sleep keys as well as the lid switch.
5525 * journalctl now understands the syntax "journalctl
5526 /usr/bin/avahi-daemon" to get all log output of a specific
5529 * CapabilityBoundingSet= in system.conf now also influences
5530 the capability bound set of usermode helpers of the kernel.
5532 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5533 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5534 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5535 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5539 * Note that we skipped 139 releases here in order to set the
5540 new version to something that is greater than both udev's
5541 and systemd's most recent version number.
5543 * udev: all udev sources are merged into the systemd source tree now.
5544 All future udev development will happen in the systemd tree. It
5545 is still fully supported to use the udev daemon and tools without
5546 systemd running, like in initramfs or other init systems. Building
5547 udev though, will require the *build* of the systemd tree, but
5548 udev can be properly *run* without systemd.
5550 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5551 should be used to create dead device nodes as workarounds for broken
5554 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5555 no longer supported. udev_monitor_new_from_netlink() needs to be
5556 used to subscribe to events.
5558 * udev: when udevd is started by systemd, processes which are left
5559 behind by forking them off of udev rules, are unconditionally cleaned
5560 up and killed now after the event handling has finished. Services or
5561 daemons must be started as systemd services. Services can be
5562 pulled-in by udev to get started, but they can no longer be directly
5563 forked by udev rules.
5565 * udev: the daemon binary is called systemd-udevd now and installed
5566 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5567 to adapt to that, create symlink, or rename the binary after building
5570 * libudev no longer provides these symbols:
5571 udev_monitor_from_socket()
5572 udev_queue_get_failed_list_entry()
5573 udev_get_{dev,sys,run}_path()
5574 The versions number was bumped and symbol versioning introduced.
5576 * systemd-loginctl and systemd-journalctl have been renamed
5577 to loginctl and journalctl to match systemctl.
5579 * The config files: /etc/systemd/systemd-logind.conf and
5580 /etc/systemd/systemd-journald.conf have been renamed to
5581 logind.conf and journald.conf. Package updates should rename
5582 the files to the new names on upgrade.
5584 * For almost all files the license is now LGPL2.1+, changed
5585 from the previous GPL2.0+. Exceptions are some minor stuff
5586 of udev (which will be changed to LGPL2.1 eventually, too),
5587 and the MIT licensed sd-daemon.[ch] library that is suitable
5588 to be used as drop-in files.
5590 * systemd and logind now handle system sleep states, in
5591 particular suspending and hibernating.
5593 * logind now implements a sleep/shutdown/idle inhibiting logic
5594 suitable for a variety of uses. Soonishly Lennart will blog
5595 about this in more detail.
5597 * var-run.mount and var-lock.mount are no longer provided
5598 (which prevously bind mounted these directories to their new
5599 places). Distributions which have not converted these
5600 directories to symlinks should consider stealing these files
5601 from git history and add them downstream.
5603 * We introduced the Documentation= field for units and added
5604 this to all our shipped units. This is useful to make it
5605 easier to explore the boot and the purpose of the various
5608 * All smaller setup units (such as
5609 systemd-vconsole-setup.service) now detect properly if they
5610 are run in a container and are skipped when
5611 appropriate. This guarantees an entirely noise-free boot in
5612 Linux container environments such as systemd-nspawn.
5614 * A framework for implementing offline system updates is now
5615 integrated, for details see:
5616 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5618 * A new service type Type=idle is available now which helps us
5619 avoiding ugly interleaving of getty output and boot status
5622 * There's now a system-wide CapabilityBoundingSet= option to
5623 globally reduce the set of capabilities for the
5624 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5625 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5626 even CAP_NET_ADMIN system-wide for secure systems.
5628 * There are now system-wide DefaultLimitXXX= options to
5629 globally change the defaults of the various resource limits
5630 for all units started by PID 1.
5632 * Harald Hoyer's systemd test suite has been integrated into
5633 systemd which allows easy testing of systemd builds in qemu
5634 and nspawn. (This is really awesome! Ask us for details!)
5636 * The fstab parser is now implemented as generator, not inside
5639 * systemctl will now warn you if .mount units generated from
5640 /etc/fstab are out of date due to changes in fstab that
5641 have not been read by systemd yet.
5643 * systemd is now suitable for usage in initrds. Dracut has
5644 already been updated to make use of this. With this in place
5645 initrds get a slight bit faster but primarily are much
5646 easier to introspect and debug since "systemctl status" in
5647 the host system can be used to introspect initrd services,
5648 and the journal from the initrd is kept around too.
5650 * systemd-delta has been added, a tool to explore differences
5651 between user/admin configuration and vendor defaults.
5653 * PrivateTmp= now affects both /tmp and /var/tmp.
5655 * Boot time status messages are now much prettier and feature
5656 proper english language. Booting up systemd has never been
5659 * Read-ahead pack files now include the inode number of all
5660 files to pre-cache. When the inode changes the pre-caching
5661 is not attempted. This should be nicer to deal with updated
5662 packages which might result in changes of read-ahead
5665 * We now temporaritly lower the kernel's read_ahead_kb variable
5666 when collecting read-ahead data to ensure the kernel's
5667 built-in read-ahead does not add noise to our measurements
5668 of necessary blocks to pre-cache.
5670 * There's now RequiresMountsFor= to add automatic dependencies
5671 for all mounts necessary for a specific file system path.
5673 * MountAuto= and SwapAuto= have been removed from
5674 system.conf. Mounting file systems at boot has to take place
5677 * nspawn now learned a new switch --uuid= to set the machine
5678 ID on the command line.
5680 * nspawn now learned the -b switch to automatically search
5683 * vt102 is now the default TERM for serial TTYs, upgraded from
5686 * systemd-logind now works on VT-less systems.
5688 * The build tree has been reorganized. The individual
5689 components now have directories of their own.
5691 * A new condition type ConditionPathIsReadWrite= is now available.
5693 * nspawn learned the new -C switch to create cgroups for the
5694 container in other hierarchies.
5696 * We now have support for hardware watchdogs, configurable in
5699 * The scheduled shutdown logic now has a public API.
5701 * We now mount /tmp as tmpfs by default, but this can be
5702 masked and /etc/fstab can override it.
5704 * Since udisks does not make use of /media anymore we are not
5705 mounting a tmpfs on it anymore.
5707 * journalctl gained a new --local switch to only interleave
5708 locally generated journal files.
5710 * We can now load the IMA policy at boot automatically.
5712 * The GTK tools have been split off into a systemd-ui.
5714 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5715 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5716 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5717 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5718 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5719 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5720 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5721 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5722 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5727 * This is mostly a bugfix release
5729 * Support optional initialization of the machine ID from the
5730 KVM or container configured UUID.
5732 * Support immediate reboots with "systemctl reboot -ff"
5734 * Show /etc/os-release data in systemd-analyze output
5736 * Many bugfixes for the journal, including endianness fixes and
5737 ensuring that disk space enforcement works
5739 * sd-login.h is C++ comptaible again
5741 * Extend the /etc/os-release format on request of the Debian
5744 * We now refuse non-UTF8 strings used in various configuration
5745 and unit files. This is done to ensure we do not pass invalid
5746 data over D-Bus or expose it elsewhere.
5748 * Register Mimo USB Screens as suitable for automatic seat
5751 * Read SELinux client context from journal clients in a race
5754 * Reorder configuration file lookup order. /etc now always
5755 overrides /run in order to allow the administrator to always
5756 and unconditionally override vendor-supplied or
5757 automatically generated data.
5759 * The various user visible bits of the journal now have man
5760 pages. We still lack man pages for the journal API calls
5763 * We now ship all man pages in HTML format again in the
5766 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5767 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5768 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5773 * This is mostly a bugfix release
5775 * systems lacking /etc/os-release are no longer supported.
5777 * Various functionality updates to libsystemd-login.so
5779 * Track class of PAM logins to distinguish greeters from
5782 Contributions from: Kay Sievers, Lennart Poettering, Michael
5787 * This is an important bugfix release for v41.
5789 * Building man pages is now optional which should be useful
5790 for those building systemd from git but unwilling to install
5793 * Watchdog support for supervising services is now usable. In
5794 a future release support for hardware watchdogs
5795 (i.e. /dev/watchdog) will be added building on this.
5797 * Service start rate limiting is now configurable and can be
5798 turned off per service. When a start rate limit is hit a
5799 reboot can automatically be triggered.
5801 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5803 Contributions from: Benjamin Franzke, Bill Nottingham,
5804 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5805 Schmidt, Michał Górny, Piotr Drąg
5809 * The systemd binary is installed /usr/lib/systemd/systemd now;
5810 An existing /sbin/init symlink needs to be adapted with the
5813 * The code that loads kernel modules has been ported to invoke
5814 libkmod directly, instead of modprobe. This means we do not
5815 support systems with module-init-tools anymore.
5817 * Watchdog support is now already useful, but still not
5820 * A new kernel command line option systemd.setenv= is
5821 understood to set system wide environment variables
5822 dynamically at boot.
5824 * We now limit the set of capabilities of systemd-journald.
5826 * We now set SIGPIPE to ignore by default, since it only is
5827 useful in shell pipelines, and has little use in general
5828 code. This can be disabled with IgnoreSIPIPE=no in unit
5831 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5832 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5837 * This is mostly a bugfix release
5839 * We now expose the reason why a service failed in the
5840 "Result" D-Bus property.
5842 * Rudimentary service watchdog support (will be completed over
5843 the next few releases.)
5845 * When systemd forks off in order execute some service we will
5846 now immediately changes its argv[0] to reflect which process
5847 it will execute. This is useful to minimize the time window
5848 with a generic argv[0], which makes bootcharts more useful
5850 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5851 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5852 Mike Kazantsev, Ray Strode
5856 * This is mostly a test release, but incorporates many
5859 * New systemd-cgtop tool to show control groups by their
5862 * Linking against libacl for ACLs is optional again. If
5863 disabled, support tracking device access for active logins
5864 goes becomes unavailable, and so does access to the user
5865 journals by the respective users.
5867 * If a group "adm" exists, journal files are automatically
5868 owned by them, thus allow members of this group full access
5869 to the system journal as well as all user journals.
5871 * The journal now stores the SELinux context of the logging
5872 client for all entries.
5874 * Add C++ inclusion guards to all public headers
5876 * New output mode "cat" in the journal to print only text
5877 messages, without any meta data like date or time.
5879 * Include tiny X server wrapper as a temporary stop-gap to
5880 teach XOrg udev display enumeration. This is used by display
5881 managers such as gdm, and will go away as soon as XOrg
5882 learned native udev hotplugging for display devices.
5884 * Add new systemd-cat tool for executing arbitrary programs
5885 with STDERR/STDOUT connected to the journal. Can also act as
5886 BSD logger replacement, and does so by default.
5888 * Optionally store all locally generated coredumps in the
5889 journal along with meta data.
5891 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5892 writing short strings to files (for usage for /sys), and for
5893 creating symlinks, character and block device nodes.
5895 * New unit file option ControlGroupPersistent= to make cgroups
5896 persistent, following the mechanisms outlined in
5897 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5899 * Support multiple local RTCs in a sane way
5901 * No longer monopolize IO when replaying readahead data on
5902 rotating disks, since we might starve non-file-system IO to
5903 death, since fanotify() will not see accesses done by blkid,
5906 * Do not show kernel threads in systemd-cgls anymore, unless
5907 requested with new -k switch.
5909 Contributions from: Dan Horák, Kay Sievers, Lennart
5910 Poettering, Michal Schmidt
5914 * This is mostly a test release, but incorporates many
5917 * The git repository moved to:
5918 git://anongit.freedesktop.org/systemd/systemd
5919 ssh://git.freedesktop.org/git/systemd/systemd
5921 * First release with the journal
5922 http://0pointer.de/blog/projects/the-journal.html
5924 * The journal replaces both systemd-kmsg-syslogd and
5925 systemd-stdout-bridge.
5927 * New sd_pid_get_unit() API call in libsystemd-logind
5929 * Many systemadm clean-ups
5931 * Introduce remote-fs-pre.target which is ordered before all
5932 remote mounts and may be used to start services before all
5935 * Added Mageia support
5937 * Add bash completion for systemd-loginctl
5939 * Actively monitor PID file creation for daemons which exit in
5940 the parent process before having finished writing the PID
5941 file in the daemon process. Daemons which do this need to be
5942 fixed (i.e. PID file creation must have finished before the
5943 parent exits), but we now react a bit more gracefully to them.
5945 * Add colourful boot output, mimicking the well-known output
5946 of existing distributions.
5948 * New option PassCredentials= for socket units, for
5949 compatibility with a recent kernel ABI breakage.
5951 * /etc/rc.local is now hooked in via a generator binary, and
5952 thus will no longer act as synchronization point during
5955 * systemctl list-unit-files now supports --root=.
5957 * systemd-tmpfiles now understands two new commands: z, Z for
5958 relabelling files according to the SELinux database. This is
5959 useful to apply SELinux labels to specific files in /sys,
5962 * Output of SysV services is now forwarded to both the console
5963 and the journal by default, not only just the console.
5965 * New man pages for all APIs from libsystemd-login.
5967 * The build tree got reorganized and a the build system is a
5968 lot more modular allowing embedded setups to specifically
5969 select the components of systemd they are interested in.
5971 * Support for Linux systems lacking the kernel VT subsystem is
5974 * configure's --with-rootdir= got renamed to
5975 --with-rootprefix= to follow the naming used by udev and
5978 * Unless specified otherwise we will now install to /usr instead
5979 of /usr/local by default.
5981 * Processes with '@' in argv[0][0] are now excluded from the
5982 final shut-down killing spree, following the logic explained
5984 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5986 * All processes remaining in a service cgroup when we enter
5987 the START or START_PRE states are now killed with
5988 SIGKILL. That means it is no longer possible to spawn
5989 background processes from ExecStart= lines (which was never
5990 supported anyway, and bad style).
5992 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5993 reloading of units together.
5995 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5996 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5997 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5998 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5999 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob