1 systemd System and Service Manager
5 * The systemd-resolved DNS resolver service has gained a substantial
6 set of new features, most prominently it may now act as a DNSSEC
7 validating stub resolver. DNSSEC mode is currently turned off by
8 default, but is expected to be turned on by default in one of the
9 next releases. For now, we invite everybody to test the DNSSEC logic
10 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
11 service also gained a full set of D-Bus interfaces, including calls
12 to configure DNS and DNSSEC settings per link (for use by external
13 network management software). systemd-resolved and systemd-networkd
14 now distinguish between "search" and "routing" domains. The former
15 are used to qualify single-label names, the latter are used purely
16 for routing lookups within certain domains to specific links.
17 resolved now also synthesizes RRs for all entries from /etc/hosts.
19 * The systemd-resolve tool (which is a client utility for
20 systemd-resolved) has been improved considerably and is now fully
21 supported and documented. Hence it has moved from /usr/lib/systemd to
24 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
27 * The coredump collection logic has been reworked: when a coredump is
28 collected it is now written to disk, compressed and processed
29 (including stacktrace extraction) from a new instantiated service
30 systemd-coredump@.service, instead of directly from the
31 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
32 processing large coredumps can take up a substantial amount of
33 resources and time, and this previously happened entirely outside of
34 systemd's service supervision. With the new logic the core_pattern
35 hook only does minimal metadata collection before passing off control
36 to the new instantiated service, which is configured with a time
37 limit, a nice level and other settings to minimize negative impact on
38 the rest of the system. Also note that the new logic will honour the
39 RLIMIT_CORE setting of the crashed process, which now allows users
40 and processes to turn off coredumping for their processes by setting
43 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
44 and all forked processes by default. Previously, PID 1 would leave
45 the setting at "0" for all processes, as set by the kernel. Note that
46 the resource limit traditionally has no effect on the generated
47 coredumps on the system if the /proc/sys/kernel/core_pattern hook
48 logic is used. Since the limit is now honoured (see above) its
49 default has been changed so that the coredumping logic is enabled by
50 default for all processes, while allowing specific opt-out.
52 * When the stacktrace is extracted from processes of system users, this
53 is now done as "systemd-coredump" user, in order to sandbox this
54 potentially security sensitive parsing operation. (Note that when
55 processing coredumps of normal users this is done under the user ID
56 of process that crashed, as before.) Packagers should take notice
57 that it is now necessary to create the "systemd-coredump" system user
58 and group at package installation time.
60 * The systemd-activate socket activation testing tool gained support
61 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
62 and --seqpacket switches. It also has been extended to support both
63 new-style and inetd-style file descriptor passing. Use the new
64 --inetd switch to request inetd-style file descriptor passing.
66 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
67 variable, which takes a boolean value. If set to false, ANSI color
68 output is disabled in the tools even when run on a terminal that
71 * The VXLAN support in networkd now supports two new settings
72 DestinationPort= and PortRange=.
74 * A new systemd.machine_id= kernel command line switch has been added,
75 that may be used to set the machine ID in /etc/machine-id if it is
76 not initialized yet. This command line option has no effect if the
77 file is already initialized.
79 * systemd-nspawn gained a new --as-pid2 switch that invokes any
80 specified command line as PID 2 rather than PID 1 in the
81 container. In this mode PID 1 is a minimal stub init process that
82 implements the special POSIX and Linux semantics of PID 1 regarding
83 signal and child process management. Note that this stub init process
84 is implemented in nspawn itself and requires no support from the
85 container image. This new logic is useful to support running
86 arbitrary commands in the container, as normal processes are
87 generally not prepared to run as PID 1.
89 * systemd-nspawn gained a new --chdir= switch for setting the current
90 working directory for the process started in the container.
92 * "journalctl /dev/sda" will now output all kernel log messages for
93 specified device from the current boot, in addition to all devices
94 that are parents of it. This should make log output about devices
95 pretty useful, as long as kernel drivers attach enough metadata to
96 the log messages. (The usual SATA drivers do.)
98 * The sd-journal API gained two new calls
99 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
100 that report whether log data from /run or /var has been found.
102 * journalctl gained a new switch "--fields" that prints all journal
103 record field names currently in use in the journal. This is backed
104 by two new sd-journal API calls sd_journal_enumerate_fields() and
105 sd_journal_restart_fields().
107 * Most configurable timeouts in systemd now expect an argument of
108 "infinity" to turn them off, instead of "0" as before. The semantics
109 from now on is that a timeout of "0" means "now", and "infinity"
110 means "never". To maintain backwards compatibility, "0" continues to
111 turn off previously existing timeout settings.
113 * "systemctl reload-or-try-restart" has been renamed to "systemctl
114 try-reload-or-restart" to clarify what it actually does: the "try"
115 logic applies to both reloading and restarting, not just restarting.
116 The old name continues to be accepted for compatibility.
118 * On boot-up, when PID 1 detects that the system clock is behind the
119 release date of the systemd version in use, the clock is now set
120 to the latter. Previously, this was already done in timesyncd, in order
121 to avoid running with clocks set to the various clock epochs such as
122 1902, 1938 or 1970. With this change the logic is now done in PID 1
123 in addition to timesyncd during early boot-up, so that it is enforced
124 before the first process is spawned by systemd. Note that the logic
125 in timesyncd remains, as it is more comprehensive and ensures
126 clock monotonicity by maintaining a persistent timestamp file in
127 /var. Since /var is generally not available in earliest boot or the
128 initrd, this part of the logic remains in timesyncd, and is not done
131 * Support for tweaking details in net_cls.class_id through the
132 NetClass= configuration directive has been removed, as the kernel
133 people have decided to deprecate that controller in cgroup v2.
134 Userspace tools such as nftables are moving over to setting rules
135 that are specific to the full cgroup path of a task, which obsoletes
136 these controllers anyway. The NetClass= directive is kept around for
137 legacy compatibility reasons. For a more in-depth description of the
138 kernel change, please refer to the respective upstream commit:
140 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
142 * A new service setting RuntimeMaxSec= has been added that may be used
143 to specify a maximum runtime for a service. If the timeout is hit, the
144 service is terminated and put into a failure state.
146 * A new service setting AmbientCapabilities= has been added. It allows
147 configuration of additional Linux process capabilities that are
148 passed to the activated processes. This is only available on very
151 * The process resource limit settings in service units may now be used
152 to configure hard and soft limits individually.
154 * The various libsystemd APIs such as sd-bus or sd-event now publicly
155 expose support for gcc's __attribute__((cleanup())) C extension.
156 Specifically, for many object destructor functions alternative
157 versions have been added that have names suffixed with "p" and take a
158 pointer to a pointer to the object to destroy, instead of just a
159 pointer to the object itself. This is useful because these destructor
160 functions may be used directly as parameters to the cleanup
161 construct. Internally, systemd has been a heavy user of this GCC
162 extension for a long time, and with this change similar support is
163 now available to consumers of the library outside of systemd. Note
164 that by using this extension in your sources compatibility with old
165 and strictly ANSI compatible C compilers is lost. However, all gcc or
166 LLVM versions of recent years support this extension.
168 * Timer units gained support for a new setting RandomizedDelaySec= that
169 allows configuring some additional randomized delay to the configured
170 time. This is useful to spread out timer events to avoid load peaks in
171 clusters or larger setups.
173 * Calendar time specifications now support sub-second accuracy.
175 * Socket units now support listening on SCTP and UDP-lite protocol
178 * The sd-event API now comes with a full set of man pages.
180 * Older versions of systemd contained experimental support for
181 compressing journal files and coredumps with the LZ4 compressor that
182 was not compatible with the lz4 binary (due to API limitations of the
183 lz4 library). This support has been removed; only support for files
184 compatible with the lz4 binary remains. This LZ4 logic is now
185 officially supported and no longer considered experimental.
187 * The dkr image import logic has been removed again from importd. dkr's
188 micro-services focus doesn't fit into the machine image focus of
189 importd, and quickly got out of date with the upstream dkr API.
191 * Creation of the /run/lock/lockdev/ directory was dropped from
192 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
193 been available for many years. If you still need this, you need to
194 create your own tmpfiles.d config file with:
196 d /run/lock/lockdev 0775 root lock -
198 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
199 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
200 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
201 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
202 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
203 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
204 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
205 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
206 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
207 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
208 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
209 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
210 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
211 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
212 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
213 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
214 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
215 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
216 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
218 -- Berlin, 2016-02-11
222 * A number of properties previously only settable in unit
223 files are now also available as properties to set when
224 creating transient units programmatically via the bus, as it
225 is exposed with systemd-run's --property=
226 setting. Specifically, these are: SyslogIdentifier=,
227 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
228 EnvironmentFile=, ReadWriteDirectories=,
229 ReadOnlyDirectories=, InaccessibleDirectories=,
230 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
232 * When creating transient services via the bus API it is now
233 possible to pass in a set of file descriptors to use as
234 STDIN/STDOUT/STDERR for the invoked process.
236 * Slice units may now be created transiently via the bus APIs,
237 similar to the way service and scope units may already be
240 * Wherever systemd expects a calendar timestamp specification
241 (like in journalctl's --since= and --until= switches) UTC
242 timestamps are now supported. Timestamps suffixed with "UTC"
243 are now considered to be in Universal Time Coordinated
244 instead of the local timezone. Also, timestamps may now
245 optionally be specified with sub-second accuracy. Both of
246 these additions also apply to recurring calendar event
247 specification, such as OnCalendar= in timer units.
249 * journalctl gained a new "--sync" switch that asks the
250 journal daemon to write all so far unwritten log messages to
251 disk and sync the files, before returning.
253 * systemd-tmpfiles learned two new line types "q" and "Q" that
254 operate like "v", but also set up a basic btrfs quota
255 hierarchy when used on a btrfs file system with quota
258 * tmpfiles' "v", "q" and "Q" will now create a plain directory
259 instead of a subvolume (even on a btrfs file system) if the
260 root directory is a plain directory, and not a
261 subvolume. This should simplify things with certain chroot()
262 environments which are not aware of the concept of btrfs
265 * systemd-detect-virt gained a new --chroot switch to detect
266 whether execution takes place in a chroot() environment.
268 * CPUAffinity= now takes CPU index ranges in addition to
271 * The various memory-related resource limit settings (such as
272 LimitAS=) now understand the usual K, M, G, ... suffixes to
273 the base of 1024 (IEC). Similar, the time-related resource
274 limit settings understand the usual min, h, day, ...
277 * There's a new system.conf setting DefaultTasksMax= to
278 control the default TasksMax= setting for services and
279 scopes running on the system. (TasksMax= is the primary
280 setting that exposes the "pids" cgroup controller on systemd
281 and was introduced in the previous systemd release.) The
282 setting now defaults to 512, which means services that are
283 not explicitly configured otherwise will only be able to
284 create 512 processes or threads at maximum, from this
285 version on. Note that this means that thread- or
286 process-heavy services might need to be reconfigured to set
287 TasksMax= to a higher value. It is sufficient to set
288 TasksMax= in these specific unit files to a higher value, or
289 even "infinity". Similar, there's now a logind.conf setting
290 UserTasksMax= that defaults to 4096 and limits the total
291 number of processes or tasks each user may own
292 concurrently. nspawn containers also have the TasksMax=
293 value set by default now, to 8192. Note that all of this
294 only has an effect if the "pids" cgroup controller is
295 enabled in the kernel. The general benefit of these changes
296 should be a more robust and safer system, that provides a
297 certain amount of per-service fork() bomb protection.
299 * systemd-nspawn gained the new --network-veth-extra= switch
300 to define additional and arbitrarily-named virtual Ethernet
301 links between the host and the container.
303 * A new service execution setting PassEnvironment= has been
304 added that allows importing select environment variables
305 from PID1's environment block into the environment block of
308 * Timer units gained support for a new RemainAfterElapse=
309 setting which takes a boolean argument. It defaults on on,
310 exposing behaviour unchanged to previous releases. If set to
311 off, timer units are unloaded after they elapsed if they
312 cannot elapse again. This is particularly useful for
313 transient timer units, which shall not stay around longer
314 than until they first elapse.
316 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
317 default now (the kernel default is 16). This is beneficial
318 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
319 allows substantially larger numbers of queued
320 datagrams. This should increase the capability of systemd to
321 parallelize boot-up, as logging and sd_notify() are unlikely
322 to stall execution anymore. If you need to change the value
323 from the new defaults, use the usual sysctl.d/ snippets.
325 * The compression framing format used by the journal or
326 coredump processing has changed to be in line with what the
327 official LZ4 tools generate. LZ4 compression support in
328 systemd was considered unsupported previously, as the format
329 was not compatible with the normal tools. With this release
330 this has changed now, and it is hence safe for downstream
331 distributions to turn it on. While not compressing as well
332 as the XZ, LZ4 is substantially faster, which makes
333 it a good default choice for the compression logic in the
334 journal and in coredump handling.
336 * Any reference to /etc/mtab has been dropped from
337 systemd. The file has been obsolete since a while, but
338 systemd refused to work on systems where it was incorrectly
339 set up (it should be a symlink or non-existent). Please make
340 sure to update to util-linux 2.27.1 or newer in conjunction
341 with this systemd release, which also drops any reference to
342 /etc/mtab. If you maintain a distribution make sure that no
343 software you package still references it, as this is a
344 likely source of bugs. There's also a glibc bug pending,
345 asking for removal of any reference to this obsolete file:
347 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
349 * Support for the ".snapshot" unit type has been removed. This
350 feature turned out to be little useful and little used, and
351 has now been removed from the core and from systemctl.
353 * The dependency types RequiresOverridable= and
354 RequisiteOverridable= have been removed from systemd. They
355 have been used only very sparingly to our knowledge and
356 other options that provide a similar effect (such as
357 systemctl --mode=ignore-dependencies) are much more useful
358 and commonly used. Moreover, they were only half-way
359 implemented as the option to control behaviour regarding
360 these dependencies was never added to systemctl. By removing
361 these dependency types the execution engine becomes a bit
362 simpler. Unit files that use these dependencies should be
363 changed to use the non-Overridable dependency types
364 instead. In fact, when parsing unit files with these
365 options, that's what systemd will automatically convert them
366 too, but it will also warn, asking users to fix the unit
367 files accordingly. Removal of these dependency types should
368 only affect a negligible number of unit files in the wild.
370 * Behaviour of networkd's IPForward= option changed
371 (again). It will no longer maintain a per-interface setting,
372 but propagate one way from interfaces where this is enabled
373 to the global kernel setting. The global setting will be
374 enabled when requested by a network that is set up, but
375 never be disabled again. This change was made to make sure
376 IPv4 and IPv6 behaviour regarding packet forwarding is
377 similar (as the Linux IPv6 stack does not support
378 per-interface control of this setting) and to minimize
381 * In unit files the behaviour of %u, %U, %h, %s has
382 changed. These specifiers will now unconditionally resolve
383 to the various user database fields of the user that the
384 systemd instance is running as, instead of the user
385 configured in the specific unit via User=. Note that this
386 effectively doesn't change much, as resolving of these
387 specifiers was already turned off in the --system instance
388 of systemd, as we cannot do NSS lookups from PID 1. In the
389 --user instance of systemd these specifiers where correctly
390 resolved, but hardly made any sense, since the user instance
391 lacks privileges to do user switches anyway, and User= is
392 hence useless. Morever, even in the --user instance of
393 systemd behaviour was awkward as it would only take settings
394 from User= assignment placed before the specifier into
395 account. In order to unify and simplify the logic around
396 this the specifiers will now always resolve to the
397 credentials of the user invoking the manager (which in case
398 of PID 1 is the root user).
400 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
401 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
402 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
403 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
404 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
405 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
406 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
407 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
408 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
409 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
410 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
411 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
412 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
413 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
416 -- Berlin, 2015-11-18
420 * systemd now depends on util-linux v2.27. More specifically,
421 the newly added mount monitor feature in libmount now
422 replaces systemd's former own implementation.
424 * libmount mandates /etc/mtab not to be regular file, and
425 systemd now enforces this condition at early boot.
426 /etc/mtab has been deprecated and warned about for a very
427 long time, so systems running systemd should already have
428 stopped having this file around as anything else than a
429 symlink to /proc/self/mounts.
431 * Support for the "pids" cgroup controller has been added. It
432 allows accounting the number of tasks in a cgroup and
433 enforcing limits on it. This adds two new setting
434 TasksAccounting= and TasksMax= to each unit, as well as a
435 global option DefaultTasksAccounting=.
437 * Support for the "net_cls" cgroup controller has been added.
438 It allows assigning a net class ID to each task in the
439 cgroup, which can then be used in firewall rules and traffic
440 shaping configurations. Note that the kernel netfilter net
441 class code does not currently work reliably for ingress
442 packets on unestablished sockets.
444 This adds a new config directive called NetClass= to CGroup
445 enabled units. Allowed values are positive numbers for fixed
446 assignments and "auto" for picking a free value
449 * 'systemctl is-system-running' now returns 'offline' if the
450 system is not booted with systemd. This command can now be
451 used as a substitute for 'systemd-notify --booted'.
453 * Watchdog timeouts have been increased to 3 minutes for all
454 in-tree service files. Apparently, disk IO issues are more
455 frequent than we hoped, and user reported >1 minute waiting
458 * 'machine-id-commit' functionality has been merged into
459 'machine-id-setup --commit'. The separate binary has been
462 * The WorkingDirectory= directive in unit files may now be set
463 to the special value '~'. In this case, the working
464 directory is set to the home directory of the user
467 * "machinectl shell" will now open the shell in the home
468 directory of the selected user by default.
470 * The CrashChVT= configuration file setting is renamed to
471 CrashChangeVT=, following our usual logic of not
472 abbreviating unnecessarily. The old directive is still
473 supported for compat reasons. Also, this directive now takes
474 an integer value between 1 and 63, or a boolean value. The
475 formerly supported '-1' value for disabling stays around for
478 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
479 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
480 RootDirectory= properties can now be set for transient
483 * The systemd-analyze tool gained a new "set-log-target" verb
484 to change the logging target the system manager logs to
485 dynamically during runtime. This is similar to how
486 "systemd-analyze set-log-level" already changes the log
489 * In nspawn /sys is now mounted as tmpfs, with only a selected
490 set of subdirectories mounted in from the real sysfs. This
491 enhances security slightly, and is useful for ensuring user
492 namespaces work correctly.
494 * Support for USB FunctionFS activation has been added. This
495 allows implementation of USB gadget services that are
496 activated as soon as they are requested, so that they don't
497 have to run continously, similar to classic socket
500 * The "systemctl exit" command now optionally takes an
501 additional parameter that sets the exit code to return from
502 the systemd manager when exiting. This is only relevant when
503 running the systemd user instance, or when running the
504 system instance in a container.
506 * sd-bus gained the new API calls sd_bus_path_encode_many()
507 and sd_bus_path_decode_many() that allow easy encoding and
508 decoding of multiple identifier strings inside a D-Bus
509 object path. Another new call sd_bus_default_flush_close()
510 has been added to flush and close per-thread default
513 * systemd-cgtop gained support for a -M/--machine= switch to
514 show the control groups within a certain container only.
516 * "systemctl kill" gained support for an optional --fail
517 switch. If specified the requested operation will fail of no
518 processes have been killed, because the unit had no
519 processes attached, or similar.
521 * A new systemd.crash_reboot=1 kernel command line option has
522 been added that triggers a reboot after crashing. This can
523 also be set through CrashReboot= in systemd.conf.
525 * The RuntimeDirectory= setting now understands unit
526 specifiers like %i or %f.
528 * A new (still internal) libary API sd-ipv4acd has been added,
529 that implements address conflict detection for IPv4. It's
530 based on code from sd-ipv4ll, and will be useful for
531 detecting DHCP address conflicts.
533 * File descriptors passed during socket activation may now be
534 named. A new API sd_listen_fds_with_names() is added to
535 access the names. The default names may be overridden,
536 either in the .socket file using the FileDescriptorName=
537 parameter, or by passing FDNAME= when storing the file
538 descriptors using sd_notify().
540 * systemd-networkd gained support for:
542 - Setting the IPv6 Router Advertisment settings via
543 IPv6AcceptRouterAdvertisements= in .network files.
545 - Configuring the HelloTimeSec=, MaxAgeSec= and
546 ForwardDelaySec= bridge parameters in .netdev files.
548 - Configuring PreferredSource= for static routes in
551 * The "ask-password" framework used to query for LUKS harddisk
552 passwords or SSL passwords during boot gained support for
553 caching passwords in the kernel keyring, if it is
554 available. This makes sure that the user only has to type in
555 a passphrase once if there are multiple objects to unlock
556 with the same one. Previously, such password caching was
557 available only when Plymouth was used; this moves the
558 caching logic into the systemd codebase itself. The
559 "systemd-ask-password" utility gained a new --keyname=
560 switch to control which kernel keyring key to use for
561 caching a password in. This functionality is also useful for
562 enabling display managers such as gdm to automatically
563 unlock the user's GNOME keyring if its passphrase, the
564 user's password and the harddisk password are the same, if
565 gdm-autologin is used.
567 * When downloading tar or raw images using "machinectl
568 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
569 file is now also downloaded, if it is available and stored
570 next to the image file.
572 * Units of type ".socket" gained a new boolean setting
573 Writable= which is only useful in conjunction with
574 ListenSpecial=. If true, enables opening the specified
575 special file in O_RDWR mode rather than O_RDONLY mode.
577 * systemd-rfkill has been reworked to become a singleton
578 service that is activated through /dev/rfkill on each rfkill
579 state change and saves the settings to disk. This way,
580 systemd-rfkill is now compatible with devices that exist
581 only intermittendly, and even restores state if the previous
582 system shutdown was abrupt rather than clean.
584 * The journal daemon gained support for vacuuming old journal
585 files controlled by the number of files that shall remain,
586 in addition to the already existing control by size and by
587 date. This is useful as journal interleaving performance
588 degrades with too many seperate journal files, and allows
589 putting an effective limit on them. The new setting defaults
590 to 100, but this may be changed by setting SystemMaxFiles=
591 and RuntimeMaxFiles= in journald.conf. Also, the
592 "journalctl" tool gained the new --vacuum-files= switch to
593 manually vacuum journal files to leave only the specified
594 number of files in place.
596 * udev will now create /dev/disk/by-path links for ATA devices
597 on kernels where that is supported.
599 * Galician, Serbian, Turkish and Korean translations were added.
601 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
602 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
603 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
604 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
605 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
606 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
607 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
608 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
609 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
610 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
611 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
612 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
613 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
614 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
615 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
616 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
617 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
618 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
620 -- Berlin, 2015-10-07
624 * The DHCP implementation of systemd-networkd gained a set of
627 - The DHCP server now supports emitting DNS and NTP
628 information. It may be enabled and configured via
629 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
630 and NTP information is enabled, but no servers are
631 configured, the corresponding uplink information (if there
632 is any) is propagated.
634 - Server and client now support transmission and reception
635 of timezone information. It can be configured via the
636 newly introduced network options UseTimezone=,
637 EmitTimezone=, and Timezone=. Transmission of timezone
638 information is enabled between host and containers by
639 default now: the container will change its local timezone
640 to what the host has set.
642 - Lease timeouts can now be configured via
643 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
645 - The DHCP server improved on the stability of
646 leases. Clients are more likely to get the same lease
647 information back, even if the server loses state.
649 - The DHCP server supports two new configuration options to
650 control the lease address pool metrics, PoolOffset= and
653 * The encapsulation limit of tunnels in systemd-networkd may
654 now be configured via 'EncapsulationLimit='. It allows
655 modifying the maximum additional levels of encapsulation
656 that are permitted to be prepended to a packet.
658 * systemd now supports the concept of user buses replacing
659 session buses, if used with dbus-1.10 (and enabled via dbus
660 --enable-user-session). It previously only supported this on
661 kdbus-enabled systems, and this release expands this to
662 'dbus-daemon' systems.
664 * systemd-networkd now supports predictable interface names
667 * systemd now optionally supports the new Linux kernel
668 "unified" control group hierarchy. If enabled via the kernel
669 command-line option 'systemd.unified_cgroup_hierarchy=1',
670 systemd will try to mount the unified cgroup hierarchy
671 directly on /sys/fs/cgroup. If not enabled, or not
672 available, systemd will fall back to the legacy cgroup
673 hierarchy setup, as before. Host system and containers can
674 mix and match legacy and unified hierarchies as they
675 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
676 environment variable to individually select the hierarchy to
677 use for executed containers. By default, nspawn will use the
678 unified hierarchy for the containers if the host uses the
679 unified hierarchy, and the legacy hierarchy otherwise.
680 Please note that at this point the unified hierarchy is an
681 experimental kernel feature and is likely to change in one
682 of the next kernel releases. Therefore, it should not be
683 enabled by default in downstream distributions yet. The
684 minimum required kernel version for the unified hierarchy to
685 work is 4.2. Note that when the unified hierarchy is used
686 for the first time delegated access to controllers is
687 safe. Because of this systemd-nspawn containers will get
688 access to controllers now, as will systemd user
689 sessions. This means containers and user sessions may now
690 manage their own resources, partitioning up what the system
693 * A new special scope unit "init.scope" has been introduced
694 that encapsulates PID 1 of the system. It may be used to
695 determine resource usage and enforce resource limits on PID
696 1 itself. PID 1 hence moved out of the root of the control
699 * The cgtop tool gained support for filtering out kernel
700 threads when counting tasks in a control group. Also, the
701 count of processes is now recursively summed up by
702 default. Two options -k and --recursive= have been added to
703 revert to old behaviour. The tool has also been updated to
704 work correctly in containers now.
706 * systemd-nspawn's --bind= and --bind-ro= options have been
707 extended to allow creation of non-recursive bind mounts.
709 * libsystemd gained two new calls sd_pid_get_cgroup() and
710 sd_peer_get_cgroup() which return the control group path of
711 a process or peer of a connected AF_UNIX socket. This
712 function call is particularly useful when implementing
713 delegated subtrees support in the control group hierarchy.
715 * The "sd-event" event loop API of libsystemd now supports
716 correct dequeuing of real-time signals, without losing
719 * When systemd requests a PolicyKit decision when managing
720 units it will now add additional fields to the request,
721 including unit name and desired operation. This enables more
722 powerful PolicyKit policies, that make decisions depending
725 * nspawn learnt support for .nspawn settings files, that may
726 accompany the image files or directories of containers, and
727 may contain additional settings for the container. This is
728 an alternative to configuring container parameters via the
731 Contributions from: Cristian Rodríguez, Daniel Mack, David
732 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
733 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
734 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
735 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
736 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
737 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
738 Andersen, Tom Gundersen, Torstein Husebø
740 -- Berlin, 2015-09-08
744 * machinectl gained a new verb 'shell' which opens a fresh
745 shell on the target container or the host. It is similar to
746 the existing 'login' command of machinectl, but spawns the
747 shell directly without prompting for username or
748 password. The pseudo machine '.host' now refers to the local
749 host and is used by default. Hence, 'machinectl shell' can
750 be used as replacement for 'su -' which spawns a session as
751 a fresh systemd unit in a way that is fully isolated from
752 the originating session.
754 * systemd-networkd learned to cope with private-zone DHCP
755 options and allows other programs to query the values.
757 * SELinux access control when enabling/disabling units is no
758 longer enforced with this release. The previous
759 implementation was incorrect, and a new corrected
760 implementation is not yet available. As unit file operations
761 are still protected via PolicyKit and D-Bus policy this is
762 not a security problem. Yet, distributions which care about
763 optimal SELinux support should probably not stabilize on
766 * sd-bus gained support for matches of type "arg0has=", that
767 test for membership of strings in string arrays sent in bus
770 * systemd-resolved now dumps the contents of its DNS and LLMNR
771 caches to the logs on reception of the SIGUSR1 signal. This
772 is useful to debug DNS behaviour.
774 * The coredumpctl tool gained a new --directory= option to
775 operate on journal files in a specific directory.
777 * "systemctl reboot" and related commands gained a new
778 "--message=" option which may be used to set a free-text
779 wall message when shutting down or rebooting the
780 system. This message is also logged, which is useful for
781 figuring out the reason for a reboot or shutdown a
784 * The "systemd-resolve-host" tool's -i switch now takes
785 network interface numbers as alternative to interface names.
787 * A new unit file setting for services has been introduced:
788 UtmpMode= allows configuration of how precisely systemd
789 handles utmp and wtmp entries for the service if this is
790 enabled. This allows writing services that appear similar to
791 user sessions in the output of the "w", "who", "last" and
794 * systemd-resolved will now locally synthesize DNS resource
795 records for the "localhost" and "gateway" domains as well as
796 the local hostname. This should ensure that clients querying
797 RRs via resolved will get similar results as those going via
798 NSS, if nss-myhostname is enabled.
800 Contributions from: Alastair Hughes, Alex Crawford, Daniel
801 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
802 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
803 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
804 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
805 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
806 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
807 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
808 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
809 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
810 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
811 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
813 -- Berlin, 2015-08-27
817 * The systemd-efi-boot-generator functionality was merged into
818 systemd-gpt-auto-generator.
820 * systemd-networkd now supports Group Policy for vxlan
821 devices. It can be enabled via the new boolean configuration
822 option called 'GroupPolicyExtension='.
824 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
825 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
826 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
828 -- Berlin, 2015-07-31
832 * The python-systemd code has been removed from the systemd repository.
833 A new repository has been created which accommodates the code from
834 now on, and we kindly ask distributions to create a separate package
835 for this: https://github.com/systemd/python-systemd
837 * The systemd daemon will now reload its main configuration
838 (/etc/systemd/system.conf) on daemon-reload.
840 * sd-dhcp now exposes vendor specific extensions via
841 sd_dhcp_lease_get_vendor_specific().
843 * systemd-networkd gained a number of new configuration options.
845 - A new boolean configuration option for TAP devices called
846 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
847 device, thus allowing to send and receive GSO packets.
849 - A new tunnel configuration option called 'CopyDSCP='.
850 If enabled, the DSCP field of ip6 tunnels is copied into the
853 - A set of boolean bridge configuration options were added.
854 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
855 and 'UnicastFlood=' are now parsed by networkd and applied to the
856 respective bridge link device via the respective IFLA_BRPORT_*
859 - A new string configuration option to override the hostname sent
860 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
861 is true, networkd will use the configured hostname instead of the
862 system hostname when sending DHCP requests.
864 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
865 networkd will configure the IPv6 flow-label of the tunnel device
866 according to RFC2460.
868 - The 'macvtap' virtual network devices are now supported, similar to
869 the already supported 'macvlan' devices.
871 * systemd-resolved now implements RFC5452 to improve resilience against
872 cache poisoning. Additionally, source port randomization is enabled
873 by default to further protect against DNS spoofing attacks.
875 * nss-mymachines now supports translating UIDs and GIDs of running
876 containers with user-namespaces enabled. If a container 'foo'
877 translates a host uid 'UID' to the container uid 'TUID', then
878 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
879 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
880 mapped as 'vg-foo-TGID'.
882 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
883 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
884 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
885 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
886 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
887 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
888 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
889 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
890 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
891 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
893 -- Berlin, 2015-07-29
897 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
898 There are no known issues with current sysfs, and udev does not need
899 or should be used to work around such bugs.
901 * udev does no longer enable USB HID power management. Several reports
902 indicate, that some devices cannot handle that setting.
904 * The udev accelerometer helper was removed. The functionality
905 is now fully included in iio-sensor-proxy. But this means,
906 older iio-sensor-proxy versions will no longer provide
907 accelerometer/orientation data with this systemd version.
908 Please upgrade iio-sensor-proxy to version 1.0.
910 * networkd gained a new configuration option IPv6PrivacyExtensions=
911 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
912 for Stateless Address") on selected networks.
914 * For the sake of fewer build-time dependencies and less code in the
915 main repository, the python bindings are about to be removed in the
916 next release. A new repository has been created which accommodates
917 the code from now on, and we kindly ask distributions to create a
918 separate package for this. The removal will take place in v223.
920 https://github.com/systemd/python-systemd
922 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
923 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
924 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
925 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
926 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
927 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
928 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
929 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
930 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
931 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
933 -- Berlin, 2015-07-07
937 * The sd-bus.h and sd-event.h APIs have now been declared
938 stable and have been added to the official interface of
939 libsystemd.so. sd-bus implements an alternative D-Bus client
940 library, that is relatively easy to use, very efficient and
941 supports both classic D-Bus as well as kdbus as transport
942 backend. sd-event is a generic event loop abstraction that
943 is built around Linux epoll, but adds features such as event
944 prioritization or efficient timer handling. Both APIs are good
945 choices for C programs looking for a bus and/or event loop
946 implementation that is minimal and does not have to be
947 portable to other kernels.
949 * kdbus support is no longer compile-time optional. It is now
950 always built-in. However, it can still be disabled at
951 runtime using the kdbus=0 kernel command line setting, and
952 that setting may be changed to default to off, by specifying
953 --disable-kdbus at build-time. Note though that the kernel
954 command line setting has no effect if the kdbus.ko kernel
955 module is not installed, in which case kdbus is (obviously)
956 also disabled. We encourage all downstream distributions to
957 begin testing kdbus by adding it to the kernel images in the
958 development distributions, and leaving kdbus support in
961 * The minimal required util-linux version has been bumped to
964 * Support for chkconfig (--enable-chkconfig) was removed in
965 favor of calling an abstraction tool
966 /lib/systemd/systemd-sysv-install. This needs to be
967 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
968 in README for details.
970 * If there's a systemd unit and a SysV init script for the
971 same service name, and the user executes "systemctl enable"
972 for it (or a related call), then this will now enable both
973 (or execute the related operation on both), not just the
976 * The libudev API documentation has been converted from gtkdoc
979 * gudev has been removed from the systemd tree, it is now an
982 * The systemd-cgtop tool learnt a new --raw switch to generate
983 "raw" (machine parsable) output.
985 * networkd's IPForwarding= .network file setting learnt the
986 new setting "kernel", which ensures that networkd does not
987 change the IP forwarding sysctl from the default kernel
990 * The systemd-logind bus API now exposes a new boolean
991 property "Docked" that reports whether logind considers the
992 system "docked", i.e. connected to a docking station or not.
994 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
995 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
996 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
997 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
998 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
999 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1000 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1001 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1002 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1003 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1004 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1005 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1006 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1007 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1008 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1009 Fink, Zbigniew Jędrzejewski-Szmek
1011 -- Berlin, 2015-06-19
1015 * The gudev library has been extracted into a separate repository
1016 available at: https://git.gnome.org/browse/libgudev/
1017 It is now managed as part of the Gnome project. Distributions
1018 are recommended to pass --disable-gudev to systemd and use
1019 gudev from the Gnome project instead. gudev is still included
1020 in systemd, for now. It will be removed soon, though. Please
1021 also see the announcement-thread on systemd-devel:
1022 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1024 * systemd now exposes a CPUUsageNSec= property for each
1025 service unit on the bus, that contains the overall consumed
1026 CPU time of a service (the sum of what each process of the
1027 service consumed). This value is only available if
1028 CPUAccounting= is turned on for a service, and is then shown
1029 in the "systemctl status" output.
1031 * Support for configuring alternative mappings of the old SysV
1032 runlevels to systemd targets has been removed. They are now
1033 hardcoded in a way that runlevels 2, 3, 4 all map to
1034 multi-user.target and 5 to graphical.target (which
1035 previously was already the default behaviour).
1037 * The auto-mounter logic gained support for mount point
1038 expiry, using a new TimeoutIdleSec= setting in .automount
1039 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1041 * The EFI System Partition (ESP) as mounted to /boot by
1042 systemd-efi-boot-generator will now be unmounted
1043 automatically after 2 minutes of not being used. This should
1044 minimize the risk of ESP corruptions.
1046 * New /etc/fstab options x-systemd.requires= and
1047 x-systemd.requires-mounts-for= are now supported to express
1048 additional dependencies for mounts. This is useful for
1049 journalling file systems that support external journal
1050 devices or overlay file systems that require underlying file
1051 systems to be mounted.
1053 * systemd does not support direct live-upgrades (via systemctl
1054 daemon-reexec) from versions older than v44 anymore. As no
1055 distribution we are aware of shipped such old versions in a
1056 stable release this should not be problematic.
1058 * When systemd forks off a new per-connection service instance
1059 it will now set the $REMOTE_ADDR environment variable to the
1060 remote IP address, and $REMOTE_PORT environment variable to
1061 the remote IP port. This behaviour is similar to the
1062 corresponding environment variables defined by CGI.
1064 * systemd-networkd gained support for uplink failure
1065 detection. The BindCarrier= option allows binding interface
1066 configuration dynamically to the link sense of other
1067 interfaces. This is useful to achieve behaviour like in
1070 * systemd-networkd gained support for configuring the DHCP
1071 client identifier to use when requesting leases.
1073 * systemd-networkd now has a per-network UseNTP= option to
1074 configure whether NTP server information acquired via DHCP
1075 is passed on to services like systemd-timesyncd.
1077 * systemd-networkd gained support for vti6 tunnels.
1079 * Note that systemd-networkd manages the sysctl variable
1080 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1081 it is configured for since v219. The variable controls IP
1082 forwarding, and is a per-interface alternative to the global
1083 /proc/sys/net/ipv[46]/ip_forward. This setting is
1084 configurable in the IPForward= option, which defaults to
1085 "no". This means if networkd is used for an interface it is
1086 no longer sufficient to set the global sysctl option to turn
1087 on IP forwarding! Instead, the .network file option
1088 IPForward= needs to be turned on! Note that the
1089 implementation of this behaviour was broken in v219 and has
1092 * Many bonding and vxlan options are now configurable in
1095 * systemd-nspawn gained a new --property= setting to set unit
1096 properties for the container scope. This is useful for
1097 setting resource parameters (e.g "CPUShares=500") on
1098 containers started from the command line.
1100 * systemd-nspawn gained a new --private-users= switch to make
1101 use of user namespacing available on recent Linux kernels.
1103 * systemd-nspawn may now be called as part of a shell pipeline
1104 in which case the pipes used for stdin and stdout are passed
1105 directly to the process invoked in the container, without
1106 indirection via a pseudo tty.
1108 * systemd-nspawn gained a new switch to control the UNIX
1109 signal to use when killing the init process of the container
1112 * systemd-nspawn gained a new --overlay= switch for mounting
1113 overlay file systems into the container using the new kernel
1116 * When a container image is imported via systemd-importd and
1117 the host file system is not btrfs, a loopback block device
1118 file is created in /var/lib/machines.raw with a btrfs file
1119 system inside. It is then mounted to /var/lib/machines to
1120 enable btrfs features for container management. The loopback
1121 file and btrfs file system is grown as needed when container
1122 images are imported via systemd-importd.
1124 * systemd-machined/systemd-importd gained support for btrfs
1125 quota, to enforce container disk space limits on disk. This
1126 is exposed in "machinectl set-limit".
1128 * systemd-importd now can import containers from local .tar,
1129 .raw and .qcow2 images, and export them to .tar and .raw. It
1130 can also import dkr v2 images now from the network (on top
1133 * systemd-importd gained support for verifying downloaded
1134 images with gpg2 (previously only gpg1 was supported).
1136 * systemd-machined, systemd-logind, systemd: most bus calls
1137 are now accessible to unprivileged processes via
1138 PolicyKit. Also, systemd-logind will now allow users to kill
1139 their own sessions without further privileges or
1142 * systemd-shutdownd has been removed. This service was
1143 previously responsible for implementing scheduled shutdowns
1144 as exposed in /usr/bin/shutdown's time parameter. This
1145 functionality has now been moved into systemd-logind and is
1146 accessible via a bus interface.
1148 * "systemctl reboot" gained a new switch --firmware-setup that
1149 can be used to reboot into the EFI firmware setup, if that
1150 is available. systemd-logind now exposes an API on the bus
1151 to trigger such reboots, in case graphical desktop UIs want
1152 to cover this functionality.
1154 * "systemctl enable", "systemctl disable" and "systemctl mask"
1155 now support a new "--now" switch. If specified the units
1156 that are enabled will also be started, and the ones
1157 disabled/masked also stopped.
1159 * The Gummiboot EFI boot loader tool has been merged into
1160 systemd, and renamed to "systemd-boot". The bootctl tool has been
1161 updated to support systemd-boot.
1163 * An EFI kernel stub has been added that may be used to create
1164 kernel EFI binaries that contain not only the actual kernel,
1165 but also an initrd, boot splash, command line and OS release
1166 information. This combined binary can then be signed as a
1167 single image, so that the firmware can verify it all in one
1168 step. systemd-boot has special support for EFI binaries created
1169 like this and can extract OS release information from them
1170 and show them in the boot menu. This functionality is useful
1171 to implement cryptographically verified boot schemes.
1173 * Optional support has been added to systemd-fsck to pass
1174 fsck's progress report to an AF_UNIX socket in the file
1177 * udev will no longer create device symlinks for all block
1178 devices by default. A blacklist for excluding special block
1179 devices from this logic has been turned into a whitelist
1180 that requires picking block devices explicitly that require
1183 * A new (currently still internal) API sd-device.h has been
1184 added to libsystemd. This modernized API is supposed to
1185 replace libudev eventually. In fact, already much of libudev
1186 is now just a wrapper around sd-device.h.
1188 * A new hwdb database for storing metadata about pointing
1189 stick devices has been added.
1191 * systemd-tmpfiles gained support for setting file attributes
1192 similar to the "chattr" tool with new 'h' and 'H' lines.
1194 * systemd-journald will no longer unconditionally set the
1195 btrfs NOCOW flag on new journal files. This is instead done
1196 with tmpfiles snippet using the new 'h' line type. This
1197 allows easy disabling of this logic, by masking the
1198 journal-nocow.conf tmpfiles file.
1200 * systemd-journald will now translate audit message types to
1201 human readable identifiers when writing them to the
1202 journal. This should improve readability of audit messages.
1204 * The LUKS logic gained support for the offset= and skip=
1205 options in /etc/crypttab, as previously implemented by
1208 * /usr/lib/os-release gained a new optional field VARIANT= for
1209 distributions that support multiple variants (such as a
1210 desktop edition, a server edition, ...)
1212 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1213 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1214 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1215 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1216 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1217 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1218 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1219 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1220 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1221 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1222 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1223 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1224 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1225 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1226 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1227 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1228 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1229 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1230 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1231 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1232 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1233 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1234 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1235 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1236 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1237 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1238 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1240 -- Berlin, 2015-05-22
1244 * Introduce a new API "sd-hwdb.h" for querying the hardware
1245 metadata database. With this minimal interface one can query
1246 and enumerate the udev hwdb, decoupled from the old libudev
1247 library. libudev's interface for this is now only a wrapper
1248 around sd-hwdb. A new tool systemd-hwdb has been added to
1249 interface with and update the database.
1251 * When any of systemd's tools copies files (for example due to
1252 tmpfiles' C lines) a btrfs reflink will attempted first,
1253 before bytewise copying is done.
1255 * systemd-nspawn gained a new --ephemeral switch. When
1256 specified a btrfs snapshot is taken of the container's root
1257 directory, and immediately removed when the container
1258 terminates again. Thus, a container can be started whose
1259 changes never alter the container's root directory, and are
1260 lost on container termination. This switch can also be used
1261 for starting a container off the root file system of the
1262 host without affecting the host OS. This switch is only
1263 available on btrfs file systems.
1265 * systemd-nspawn gained a new --template= switch. It takes the
1266 path to a container tree to use as template for the tree
1267 specified via --directory=, should that directory be
1268 missing. This allows instantiating containers dynamically,
1269 on first run. This switch is only available on btrfs file
1272 * When a .mount unit refers to a mount point on which multiple
1273 mounts are stacked, and the .mount unit is stopped all of
1274 the stacked mount points will now be unmounted until no
1275 mount point remains.
1277 * systemd now has an explicit notion of supported and
1278 unsupported unit types. Jobs enqueued for unsupported unit
1279 types will now fail with an "unsupported" error code. More
1280 specifically .swap, .automount and .device units are not
1281 supported in containers, .busname units are not supported on
1282 non-kdbus systems. .swap and .automount are also not
1283 supported if their respective kernel compile time options
1286 * machinectl gained support for two new "copy-from" and
1287 "copy-to" commands for copying files from a running
1288 container to the host or vice versa.
1290 * machinectl gained support for a new "bind" command to bind
1291 mount host directories into local containers. This is
1292 currently only supported for nspawn containers.
1294 * networkd gained support for configuring bridge forwarding
1295 database entries (fdb) from .network files.
1297 * A new tiny daemon "systemd-importd" has been added that can
1298 download container images in tar, raw, qcow2 or dkr formats,
1299 and make them available locally in /var/lib/machines, so
1300 that they can run as nspawn containers. The daemon can GPG
1301 verify the downloads (not supported for dkr, since it has no
1302 provisions for verifying downloads). It will transparently
1303 decompress bz2, xz, gzip compressed downloads if necessary,
1304 and restore sparse files on disk. The daemon uses privilege
1305 separation to ensure the actual download logic runs with
1306 fewer privileges than the daemon itself. machinectl has
1307 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1308 make the functionality of importd available to the
1309 user. With this in place the Fedora and Ubuntu "Cloud"
1310 images can be downloaded and booted as containers unmodified
1311 (the Fedora images lack the appropriate GPG signature files
1312 currently, so they cannot be verified, but this will change
1313 soon, hopefully). Note that downloading images is currently
1314 only fully supported on btrfs.
1316 * machinectl is now able to list container images found in
1317 /var/lib/machines, along with some metadata about sizes of
1318 disk and similar. If the directory is located on btrfs and
1319 quota is enabled, this includes quota display. A new command
1320 "image-status" has been added that shows additional
1321 information about images.
1323 * machinectl is now able to clone container images
1324 efficiently, if the underlying file system (btrfs) supports
1325 it, with the new "machinectl list-images" command. It also
1326 gained commands for renaming and removing images, as well as
1327 marking them read-only or read-write (supported also on
1328 legacy file systems).
1330 * networkd gained support for collecting LLDP network
1331 announcements, from hardware that supports this. This is
1332 shown in networkctl output.
1334 * systemd-run gained support for a new -t (--pty) switch for
1335 invoking a binary on a pty whose input and output is
1336 connected to the invoking terminal. This allows executing
1337 processes as system services while interactively
1338 communicating with them via the terminal. Most interestingly
1339 this is supported across container boundaries. Invoking
1340 "systemd-run -t /bin/bash" is an alternative to running a
1341 full login session, the difference being that the former
1342 will not register a session, nor go through the PAM session
1345 * tmpfiles gained support for a new "v" line type for creating
1346 btrfs subvolumes. If the underlying file system is a legacy
1347 file system, this automatically degrades to creating a
1348 normal directory. Among others /var/lib/machines is now
1349 created like this at boot, should it be missing.
1351 * The directory /var/lib/containers/ has been deprecated and
1352 been replaced by /var/lib/machines. The term "machines" has
1353 been used in the systemd context as generic term for both
1354 VMs and containers, and hence appears more appropriate for
1355 this, as the directory can also contain raw images bootable
1358 * systemd-nspawn when invoked with -M but without --directory=
1359 or --image= is now capable of searching for the container
1360 root directory, subvolume or disk image automatically, in
1361 /var/lib/machines. systemd-nspawn@.service has been updated
1362 to make use of this, thus allowing it to be used for raw
1365 * A new machines.target unit has been introduced that is
1366 supposed to group all containers/VMs invoked as services on
1367 the system. systemd-nspawn@.service has been updated to
1368 integrate with that.
1370 * machinectl gained a new "start" command, for invoking a
1371 container as a service. "machinectl start foo" is mostly
1372 equivalent to "systemctl start systemd-nspawn@foo.service",
1373 but handles escaping in a nicer way.
1375 * systemd-nspawn will now mount most of the cgroupfs tree
1376 read-only into each container, with the exception of the
1377 container's own subtree in the name=systemd hierarchy.
1379 * journald now sets the special FS_NOCOW file flag for its
1380 journal files. This should improve performance on btrfs, by
1381 avoiding heavy fragmentation when journald's write-pattern
1382 is used on COW file systems. It degrades btrfs' data
1383 integrity guarantees for the files to the same levels as for
1384 ext3/ext4 however. This should be OK though as journald does
1385 its own data integrity checks and all its objects are
1386 checksummed on disk. Also, journald should handle btrfs disk
1387 full events a lot more gracefully now, by processing SIGBUS
1388 errors, and not relying on fallocate() anymore.
1390 * When journald detects that journal files it is writing to
1391 have been deleted it will immediately start new journal
1394 * systemd now provides a way to store file descriptors
1395 per-service in PID 1.This is useful for daemons to ensure
1396 that fds they require are not lost during a daemon
1397 restart. The fds are passed to the daemon on the next
1398 invocation in the same way socket activation fds are
1399 passed. This is now used by journald to ensure that the
1400 various sockets connected to all the system's stdout/stderr
1401 are not lost when journald is restarted. File descriptors
1402 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1403 an extension to sd_notify(). Note that a limit is enforced
1404 on the number of fds a service can store in PID 1, and it
1405 defaults to 0, so that no fds may be stored, unless this is
1406 explicitly turned on.
1408 * The default TERM variable to use for units connected to a
1409 terminal, when no other value is explicitly is set is now
1410 vt220 rather than vt102. This should be fairly safe still,
1411 but allows PgUp/PgDn work.
1413 * The /etc/crypttab option header= as known from Debian is now
1416 * "loginctl user-status" and "loginctl session-status" will
1417 now show the last 10 lines of log messages of the
1418 user/session following the status output. Similar,
1419 "machinectl status" will show the last 10 log lines
1420 associated with a virtual machine or container
1421 service. (Note that this is usually not the log messages
1422 done in the VM/container itself, but simply what the
1423 container manager logs. For nspawn this includes all console
1426 * "loginctl session-status" without further argument will now
1427 show the status of the session of the caller. Similar,
1428 "lock-session", "unlock-session", "activate",
1429 "enable-linger", "disable-linger" may now be called without
1430 session/user parameter in which case they apply to the
1431 caller's session/user.
1433 * An X11 session scriptlet is now shipped that uploads
1434 $DISPLAY and $XAUTHORITY into the environment of the systemd
1435 --user daemon if a session begins. This should improve
1436 compatibility with X11 enabled applications run as systemd
1439 * Generators are now subject to masking via /etc and /run, the
1440 same way as unit files.
1442 * networkd .network files gained support for configuring
1443 per-link IPv4/IPv6 packet forwarding as well as IPv4
1444 masquerading. This is by default turned on for veth links to
1445 containers, as registered by systemd-nspawn. This means that
1446 nspawn containers run with --network-veth will now get
1447 automatic routed access to the host's networks without any
1448 further configuration or setup, as long as networkd runs on
1451 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1452 or UDP posts of a container on the host. With this in place
1453 it is possible to run containers with private veth links
1454 (--network-veth), and have their functionality exposed on
1455 the host as if their services were running directly on the
1458 * systemd-nspawn's --network-veth switch now gained a short
1459 version "-n", since with the changes above it is now truly
1460 useful out-of-the-box. The systemd-nspawn@.service has been
1461 updated to make use of it too by default.
1463 * systemd-nspawn will now maintain a per-image R/W lock, to
1464 ensure that the same image is not started more than once
1465 writable. (It's OK to run an image multiple times
1466 simultaneously in read-only mode.)
1468 * systemd-nspawn's --image= option is now capable of
1469 dissecting and booting MBR and GPT disk images that contain
1470 only a single active Linux partition. Previously it
1471 supported only GPT disk images with proper GPT type
1472 IDs. This allows running cloud images from major
1473 distributions directly with systemd-nspawn, without
1476 * In addition to collecting mouse dpi data in the udev
1477 hardware database, there's now support for collecting angle
1478 information for mouse scroll wheels. The database is
1479 supposed to guarantee similar scrolling behavior on mice
1480 that it knows about. There's also support for collecting
1481 information about Touchpad types.
1483 * udev's input_id built-in will now also collect touch screen
1484 dimension data and attach it to probed devices.
1486 * /etc/os-release gained support for a Distribution Privacy
1489 * networkd gained support for creating "ipvlan", "gretap",
1490 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1492 * systemd-tmpfiles gained support for "a" lines for setting
1495 * systemd-nspawn will now mount /tmp in the container to
1496 tmpfs, automatically.
1498 * systemd now exposes the memory.usage_in_bytes cgroup
1499 attribute and shows it for each service in the "systemctl
1500 status" output, if available.
1502 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1503 immediate reboot is triggered. This useful if shutdown is
1504 hung and is unable to complete, to expedite the
1505 operation. Note that this kind of reboot will still unmount
1506 all file systems, and hence should not result in fsck being
1509 * A .device unit for an optical block device will now be
1510 considered active only when a medium is in the drive. Also,
1511 mount units are now bound to their backing devices thus
1512 triggering automatic unmounting when devices become
1513 unavailable. With this in place systemd will now
1514 automatically unmount left-over mounts when a CD-ROM is
1515 ejected or an USB stick is yanked from the system.
1517 * networkd-wait-online now has support for waiting for
1518 specific interfaces only (with globbing), and for giving up
1519 after a configurable timeout.
1521 * networkd now exits when idle. It will be automatically
1522 restarted as soon as interfaces show up, are removed or
1523 change state. networkd will stay around as long as there is
1524 at least one DHCP state machine or similar around, that keep
1527 * networkd may now configure IPv6 link-local addressing in
1528 addition to IPv4 link-local addressing.
1530 * The IPv6 "token" for use in SLAAC may now be configured for
1531 each .network interface in networkd.
1533 * Routes configured with networkd may now be assigned a scope
1536 * networkd's [Match] sections now support globbing and lists
1537 of multiple space-separated matches per item.
1539 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1540 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1541 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1542 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1543 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1544 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1545 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1546 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1547 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1548 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1549 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1550 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1551 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1552 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1553 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1554 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1555 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1556 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1557 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1558 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1559 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1560 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1561 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1562 Hoffmann, Zbigniew Jędrzejewski-Szmek
1564 -- Berlin, 2015-02-16
1568 * When querying unit file enablement status (for example via
1569 "systemctl is-enabled"), a new state "indirect" is now known
1570 which indicates that a unit might not be enabled itself, but
1571 another unit listed in its Also= setting might be.
1573 * Similar to the various existing ConditionXYZ= settings for
1574 units, there are now matching AssertXYZ= settings. While
1575 failing conditions cause a unit to be skipped, but its job
1576 to succeed, failing assertions declared like this will cause
1577 a unit start operation and its job to fail.
1579 * hostnamed now knows a new chassis type "embedded".
1581 * systemctl gained a new "edit" command. When used on a unit
1582 file, this allows extending unit files with .d/ drop-in
1583 configuration snippets or editing the full file (after
1584 copying it from /usr/lib to /etc). This will invoke the
1585 user's editor (as configured with $EDITOR), and reload the
1586 modified configuration after editing.
1588 * "systemctl status" now shows the suggested enablement state
1589 for a unit, as declared in the (usually vendor-supplied)
1590 system preset files.
1592 * nss-myhostname will now resolve the single-label host name
1593 "gateway" to the locally configured default IP routing
1594 gateways, ordered by their metrics. This assigns a stable
1595 name to the used gateways, regardless which ones are
1596 currently configured. Note that the name will only be
1597 resolved after all other name sources (if nss-myhostname is
1598 configured properly) and should hence not negatively impact
1599 systems that use the single-label host name "gateway" in
1602 * systemd-inhibit now allows filtering by mode when listing
1605 * Scope and service units gained a new "Delegate" boolean
1606 property, which, when set, allows processes running inside the
1607 unit to further partition resources. This is primarily
1608 useful for systemd user instances as well as container
1611 * journald will now pick up audit messages directly from
1612 the kernel, and log them like any other log message. The
1613 audit fields are split up and fully indexed. This means that
1614 journalctl in many ways is now a (nicer!) alternative to
1615 ausearch, the traditional audit client. Note that this
1616 implements only a minimal audit client. If you want the
1617 special audit modes like reboot-on-log-overflow, please use
1618 the traditional auditd instead, which can be used in
1619 parallel to journald.
1621 * The ConditionSecurity= unit file option now understands the
1622 special string "audit" to check whether auditing is
1625 * journalctl gained two new commands --vacuum-size= and
1626 --vacuum-time= to delete old journal files until the
1627 remaining ones take up no more than the specified size on disk,
1628 or are not older than the specified time.
1630 * A new, native PPPoE library has been added to sd-network,
1631 systemd's library of light-weight networking protocols. This
1632 library will be used in a future version of networkd to
1633 enable PPPoE communication without an external pppd daemon.
1635 * The busctl tool now understands a new "capture" verb that
1636 works similar to "monitor", but writes a packet capture
1637 trace to STDOUT that can be redirected to a file which is
1638 compatible with libcap's capture file format. This can then
1639 be loaded in Wireshark and similar tools to inspect bus
1642 * The busctl tool now understands a new "tree" verb that shows
1643 the object trees of a specific service on the bus, or of all
1646 * The busctl tool now understands a new "introspect" verb that
1647 shows all interfaces and members of objects on the bus,
1648 including their signature and values. This is particularly
1649 useful to get more information about bus objects shown by
1650 the new "busctl tree" command.
1652 * The busctl tool now understands new verbs "call",
1653 "set-property" and "get-property" for invoking bus method
1654 calls, setting and getting bus object properties in a
1657 * busctl gained a new --augment-creds= argument that controls
1658 whether the tool shall augment credential information it
1659 gets from the bus with data from /proc, in a possibly
1662 * nspawn's --link-journal= switch gained two new values
1663 "try-guest" and "try-host" that work like "guest" and
1664 "host", but do not fail if the host has no persistent
1665 journalling enabled. -j is now equivalent to
1666 --link-journal=try-guest.
1668 * macvlan network devices created by nspawn will now have
1669 stable MAC addresses.
1671 * A new SmackProcessLabel= unit setting has been added, which
1672 controls the SMACK security label processes forked off by
1673 the respective unit shall use.
1675 * If compiled with --enable-xkbcommon, systemd-localed will
1676 verify x11 keymap settings by compiling the given keymap. It
1677 will spew out warnings if the compilation fails. This
1678 requires libxkbcommon to be installed.
1680 * When a coredump is collected, a larger number of metadata
1681 fields is now collected and included in the journal records
1682 created for it. More specifically, control group membership,
1683 environment variables, memory maps, working directory,
1684 chroot directory, /proc/$PID/status, and a list of open file
1685 descriptors is now stored in the log entry.
1687 * The udev hwdb now contains DPI information for mice. For
1690 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1692 * All systemd programs that read standalone configuration
1693 files in /etc now also support a corresponding series of
1694 .conf.d configuration directories in /etc/, /run/,
1695 /usr/local/lib/, /usr/lib/, and (if configured with
1696 --enable-split-usr) /lib/. In particular, the following
1697 configuration files now have corresponding configuration
1698 directories: system.conf user.conf, logind.conf,
1699 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1700 resolved.conf, timesyncd.conf, journal-remote.conf, and
1701 journal-upload.conf. Note that distributions should use the
1702 configuration directories in /usr/lib/; the directories in
1703 /etc/ are reserved for the system administrator.
1705 * systemd-rfkill will no longer take the rfkill device name
1706 into account when storing rfkill state on disk, as the name
1707 might be dynamically assigned and not stable. Instead, the
1708 ID_PATH udev variable combined with the rfkill type (wlan,
1709 bluetooth, ...) is used.
1711 * A new service systemd-machine-id-commit.service has been
1712 added. When used on systems where /etc is read-only during
1713 boot, and /etc/machine-id is not initialized (but an empty
1714 file), this service will copy the temporary machine ID
1715 created as replacement into /etc after the system is fully
1716 booted up. This is useful for systems that are freshly
1717 installed with a non-initialized machine ID, but should get
1718 a fixed machine ID for subsequent boots.
1720 * networkd's .netdev files now provide a large set of
1721 configuration parameters for VXLAN devices. Similarly, the
1722 bridge port cost parameter is now configurable in .network
1723 files. There's also new support for configuring IP source
1724 routing. networkd .link files gained support for a new
1725 OriginalName= match that is useful to match against the
1726 original interface name the kernel assigned. .network files
1727 may include MTU= and MACAddress= fields for altering the MTU
1728 and MAC address while being connected to a specific network
1731 * The LUKS logic gained supported for configuring
1732 UUID-specific key files. There's also new support for naming
1733 LUKS device from the kernel command line, using the new
1734 luks.name= argument.
1736 * Timer units may now be transiently created via the bus API
1737 (this was previously already available for scope and service
1738 units). In addition it is now possible to create multiple
1739 transient units at the same time with a single bus call. The
1740 "systemd-run" tool has been updated to make use of this for
1741 running commands on a specified time, in at(1)-style.
1743 * tmpfiles gained support for "t" lines, for assigning
1744 extended attributes to files. Among other uses this may be
1745 used to assign SMACK labels to files.
1747 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1748 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1749 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1750 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1751 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1752 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1753 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1754 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1755 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1756 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1757 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1758 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1759 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1760 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1761 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1762 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1763 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1764 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1766 -- Berlin, 2014-12-10
1770 * journalctl gained the new options -t/--identifier= to match
1771 on the syslog identifier (aka "tag"), as well as --utc to
1772 show log timestamps in the UTC timezone. journalctl now also
1773 accepts -n/--lines=all to disable line capping in a pager.
1775 * journalctl gained a new switch, --flush, that synchronously
1776 flushes logs from /run/log/journal to /var/log/journal if
1777 persistent storage is enabled. systemd-journal-flush.service
1778 now waits until the operation is complete.
1780 * Services can notify the manager before they start a reload
1781 (by sending RELOADING=1) or shutdown (by sending
1782 STOPPING=1). This allows the manager to track and show the
1783 internal state of daemons and closes a race condition when
1784 the process is still running but has closed its D-Bus
1787 * Services with Type=oneshot do not have to have any ExecStart
1790 * User units are now loaded also from
1791 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1792 /run/systemd/user directory that was already previously
1793 supported, but is under the control of the user.
1795 * Job timeouts (i.e. time-outs on the time a job that is
1796 queued stays in the run queue) can now optionally result in
1797 immediate reboot or power-off actions (JobTimeoutAction= and
1798 JobTimeoutRebootArgument=). This is useful on ".target"
1799 units, to limit the maximum time a target remains
1800 undispatched in the run queue, and to trigger an emergency
1801 operation in such a case. This is now used by default to
1802 turn off the system if boot-up (as defined by everything in
1803 basic.target) hangs and does not complete for at least
1804 15min. Also, if power-off or reboot hang for at least 30min
1805 an immediate power-off/reboot operation is triggered. This
1806 functionality is particularly useful to increase reliability
1807 on embedded devices, but also on laptops which might
1808 accidentally get powered on when carried in a backpack and
1809 whose boot stays stuck in a hard disk encryption passphrase
1812 * systemd-logind can be configured to also handle lid switch
1813 events even when the machine is docked or multiple displays
1814 are attached (HandleLidSwitchDocked= option).
1816 * A helper binary and a service have been added which can be
1817 used to resume from hibernation in the initramfs. A
1818 generator will parse the resume= option on the kernel
1819 command line to trigger resume.
1821 * A user console daemon systemd-consoled has been
1822 added. Currently, it is a preview, and will so far open a
1823 single terminal on each session of the user marked as
1824 Desktop=systemd-console.
1826 * Route metrics can be specified for DHCP routes added by
1829 * The SELinux context of socket-activated services can be set
1830 from the information provided by the networking stack
1831 (SELinuxContextFromNet= option).
1833 * Userspace firmware loading support has been removed and
1834 the minimum supported kernel version is thus bumped to 3.7.
1836 * Timeout for udev workers has been increased from 1 to 3
1837 minutes, but a warning will be printed after 1 minute to
1838 help diagnose kernel modules that take a long time to load.
1840 * Udev rules can now remove tags on devices with TAG-="foobar".
1842 * systemd's readahead implementation has been removed. In many
1843 circumstances it didn't give expected benefits even for
1844 rotational disk drives and was becoming less relevant in the
1845 age of SSDs. As none of the developers has been using
1846 rotating media anymore, and nobody stepped up to actively
1847 maintain this component of systemd it has now been removed.
1849 * Swap units can use Options= to specify discard options.
1850 Discard options specified for swaps in /etc/fstab are now
1853 * Docker containers are now detected as a separate type of
1856 * The Password Agent protocol gained support for queries where
1857 the user input is shown, useful e.g. for user names.
1858 systemd-ask-password gained a new --echo option to turn that
1861 * The default sysctl.d/ snippets will now set:
1863 net.core.default_qdisc = fq_codel
1865 This selects Fair Queuing Controlled Delay as the default
1866 queuing discipline for network interfaces. fq_codel helps
1867 fight the network bufferbloat problem. It is believed to be
1868 a good default with no tuning required for most workloads.
1869 Downstream distributions may override this choice. On 10Gbit
1870 servers that do not do forwarding, "fq" may perform better.
1871 Systems without a good clocksource should use "pfifo_fast".
1873 * If kdbus is enabled during build a new option BusPolicy= is
1874 available for service units, that allows locking all service
1875 processes into a stricter bus policy, in order to limit
1876 access to various bus services, or even hide most of them
1877 from the service's view entirely.
1879 * networkctl will now show the .network and .link file
1880 networkd has applied to a specific interface.
1882 * sd-login gained a new API call sd_session_get_desktop() to
1883 query which desktop environment has been selected for a
1886 * UNIX utmp support is now compile-time optional to support
1887 legacy-free systems.
1889 * systemctl gained two new commands "add-wants" and
1890 "add-requires" for pulling in units from specific targets
1893 * If the word "rescue" is specified on the kernel command line
1894 the system will now boot into rescue mode (aka
1895 rescue.target), which was previously available only by
1896 specifying "1" or "systemd.unit=rescue.target" on the kernel
1897 command line. This new kernel command line option nicely
1898 mirrors the already existing "emergency" kernel command line
1901 * New kernel command line options mount.usr=, mount.usrflags=,
1902 mount.usrfstype= have been added that match root=, rootflags=,
1903 rootfstype= but allow mounting a specific file system to
1906 * The $NOTIFY_SOCKET is now also passed to control processes of
1907 services, not only the main process.
1909 * This version reenables support for fsck's -l switch. This
1910 means at least version v2.25 of util-linux is required for
1911 operation, otherwise dead-locks on device nodes may
1912 occur. Again: you need to update util-linux to at least
1913 v2.25 when updating systemd to v217.
1915 * The "multi-seat-x" tool has been removed from systemd, as
1916 its functionality has been integrated into X servers 1.16,
1917 and the tool is hence redundant. It is recommended to update
1918 display managers invoking this tool to simply invoke X
1919 directly from now on, again.
1921 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1922 message flag has been added for all of systemd's PolicyKit
1923 authenticated method calls has been added. In particular
1924 this now allows optional interactive authorization via
1925 PolicyKit for many of PID1's privileged operations such as
1926 unit file enabling and disabling.
1928 * "udevadm hwdb --update" learnt a new switch "--usr" for
1929 placing the rebuilt hardware database in /usr instead of
1930 /etc. When used only hardware database entries stored in
1931 /usr will be used, and any user database entries in /etc are
1932 ignored. This functionality is useful for vendors to ship a
1933 pre-built database on systems where local configuration is
1934 unnecessary or unlikely.
1936 * Calendar time specifications in .timer units now also
1937 understand the strings "semi-annually", "quarterly" and
1938 "minutely" as shortcuts (in addition to the preexisting
1939 "anually", "hourly", ...).
1941 * systemd-tmpfiles will now correctly create files in /dev
1942 at boot which are marked for creation only at boot. It is
1943 recommended to always create static device nodes with 'c!'
1944 and 'b!', so that they are created only at boot and not
1945 overwritten at runtime.
1947 * When the watchdog logic is used for a service (WatchdogSec=)
1948 and the watchdog timeout is hit the service will now be
1949 terminated with SIGABRT (instead of just SIGTERM), in order
1950 to make sure a proper coredump and backtrace is
1951 generated. This ensures that hanging services will result in
1952 similar coredump/backtrace behaviour as services that hit a
1955 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1956 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1957 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1958 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1959 Herrmann, David Sommerseth, David Strauss, Emil Renner
1960 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1961 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1962 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1963 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1964 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1965 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1966 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1967 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1968 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1969 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1970 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1971 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1972 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1973 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1974 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1975 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1978 -- Berlin, 2014-10-28
1982 * timedated no longer reads NTP implementation unit names from
1983 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1984 implementations should add a
1986 Conflicts=systemd-timesyncd.service
1988 to their unit files to take over and replace systemd's NTP
1989 default functionality.
1991 * systemd-sysusers gained a new line type "r" for configuring
1992 which UID/GID ranges to allocate system users/groups
1993 from. Lines of type "u" may now add an additional column
1994 that specifies the home directory for the system user to be
1995 created. Also, systemd-sysusers may now optionally read user
1996 information from STDIN instead of a file. This is useful for
1997 invoking it from RPM preinst scriptlets that need to create
1998 users before the first RPM file is installed since these
1999 files might need to be owned by them. A new
2000 %sysusers_create_inline RPM macro has been introduced to do
2001 just that. systemd-sysusers now updates the shadow files as
2002 well as the user/group databases, which should enhance
2003 compatibility with certain tools like grpck.
2005 * A number of bus APIs of PID 1 now optionally consult
2006 PolicyKit to permit access for otherwise unprivileged
2007 clients under certain conditions. Note that this currently
2008 doesn't support interactive authentication yet, but this is
2009 expected to be added eventually, too.
2011 * /etc/machine-info now has new fields for configuring the
2012 deployment environment of the machine, as well as the
2013 location of the machine. hostnamectl has been updated with
2014 new command to update these fields.
2016 * systemd-timesyncd has been updated to automatically acquire
2017 NTP server information from systemd-networkd, which might
2018 have been discovered via DHCP.
2020 * systemd-resolved now includes a caching DNS stub resolver
2021 and a complete LLMNR name resolution implementation. A new
2022 NSS module "nss-resolve" has been added which can be used
2023 instead of glibc's own "nss-dns" to resolve hostnames via
2024 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2025 be resolved via systemd-resolved D-Bus APIs. In contrast to
2026 the glibc internal resolver systemd-resolved is aware of
2027 multi-homed system, and keeps DNS server and caches separate
2028 and per-interface. Queries are sent simultaneously on all
2029 interfaces that have DNS servers configured, in order to
2030 properly handle VPNs and local LANs which might resolve
2031 separate sets of domain names. systemd-resolved may acquire
2032 DNS server information from systemd-networkd automatically,
2033 which in turn might have discovered them via DHCP. A tool
2034 "systemd-resolve-host" has been added that may be used to
2035 query the DNS logic in resolved. systemd-resolved implements
2036 IDNA and automatically uses IDNA or UTF-8 encoding depending
2037 on whether classic DNS or LLMNR is used as transport. In the
2038 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2039 implementation to systemd-resolved.
2041 * A new NSS module nss-mymachines has been added, that
2042 automatically resolves the names of all local registered
2043 containers to their respective IP addresses.
2045 * A new client tool "networkctl" for systemd-networkd has been
2046 added. It currently is entirely passive and will query
2047 networking configuration from udev, rtnetlink and networkd,
2048 and present it to the user in a very friendly
2049 way. Eventually, we hope to extend it to become a full
2050 control utility for networkd.
2052 * .socket units gained a new DeferAcceptSec= setting that
2053 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2054 TCP. Similarly, support for controlling TCP keep-alive
2055 settings has been added (KeepAliveTimeSec=,
2056 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2057 turning off Nagle's algorithm on TCP has been added
2060 * logind learned a new session type "web", for use in projects
2061 like Cockpit which register web clients as PAM sessions.
2063 * timer units with at least one OnCalendar= setting will now
2064 be started only after timer-sync.target has been
2065 reached. This way they will not elapse before the system
2066 clock has been corrected by a local NTP client or
2067 similar. This is particular useful on RTC-less embedded
2068 machines, that come up with an invalid system clock.
2070 * systemd-nspawn's --network-veth= switch should now result in
2071 stable MAC addresses for both the outer and the inner side
2074 * systemd-nspawn gained a new --volatile= switch for running
2075 container instances with /etc or /var unpopulated.
2077 * The kdbus client code has been updated to use the new Linux
2078 3.17 memfd subsystem instead of the old kdbus-specific one.
2080 * systemd-networkd's DHCP client and server now support
2081 FORCERENEW. There are also new configuration options to
2082 configure the vendor client identifier and broadcast mode
2085 * systemd will no longer inform the kernel about the current
2086 timezone, as this is necessarily incorrect and racy as the
2087 kernel has no understanding of DST and similar
2088 concepts. This hence means FAT timestamps will be always
2089 considered UTC, similar to what Android is already
2090 doing. Also, when the RTC is configured to the local time
2091 (rather than UTC) systemd will never synchronize back to it,
2092 as this might confuse Windows at a later boot.
2094 * systemd-analyze gained a new command "verify" for offline
2095 validation of unit files.
2097 * systemd-networkd gained support for a couple of additional
2098 settings for bonding networking setups. Also, the metric for
2099 statically configured routes may now be configured. For
2100 network interfaces where this is appropriate the peer IP
2101 address may now be configured.
2103 * systemd-networkd's DHCP client will no longer request
2104 broadcasting by default, as this tripped up some networks.
2105 For hardware where broadcast is required the feature should
2106 be switched back on using RequestBroadcast=yes.
2108 * systemd-networkd will now set up IPv4LL addresses (when
2109 enabled) even if DHCP is configured successfully.
2111 * udev will now default to respect network device names given
2112 by the kernel when the kernel indicates that these are
2113 predictable. This behavior can be tweaked by changing
2114 NamePolicy= in the relevant .link file.
2116 * A new library systemd-terminal has been added that
2117 implements full TTY stream parsing and rendering. This
2118 library is supposed to be used later on for implementing a
2119 full userspace VT subsystem, replacing the current kernel
2122 * A new tool systemd-journal-upload has been added to push
2123 journal data to a remote system running
2124 systemd-journal-remote.
2126 * journald will no longer forward all local data to another
2127 running syslog daemon. This change has been made because
2128 rsyslog (which appears to be the most commonly used syslog
2129 implementation these days) no longer makes use of this, and
2130 instead pulls the data out of the journal on its own. Since
2131 forwarding the messages to a non-existent syslog server is
2132 more expensive than we assumed we have now turned this
2133 off. If you run a syslog server that is not a recent rsyslog
2134 version, you have to turn this option on again
2135 (ForwardToSyslog= in journald.conf).
2137 * journald now optionally supports the LZ4 compressor for
2138 larger journal fields. This compressor should perform much
2139 better than XZ which was the previous default.
2141 * machinectl now shows the IP addresses of local containers,
2142 if it knows them, plus the interface name of the container.
2144 * A new tool "systemd-escape" has been added that makes it
2145 easy to escape strings to build unit names and similar.
2147 * sd_notify() messages may now include a new ERRNO= field
2148 which is parsed and collected by systemd and shown among the
2149 "systemctl status" output for a service.
2151 * A new component "systemd-firstboot" has been added that
2152 queries the most basic systemd information (timezone,
2153 hostname, root password) interactively on first
2154 boot. Alternatively it may also be used to provision these
2155 things offline on OS images installed into directories.
2157 * The default sysctl.d/ snippets will now set
2159 net.ipv4.conf.default.promote_secondaries=1
2161 This has the benefit of no flushing secondary IP addresses
2162 when primary addresses are removed.
2164 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2165 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2166 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2167 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2168 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2169 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2170 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2171 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2172 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2173 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2174 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2175 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2176 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2177 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2178 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2180 -- Berlin, 2014-08-19
2184 * A new tool systemd-sysusers has been added. This tool
2185 creates system users and groups in /etc/passwd and
2186 /etc/group, based on static declarative system user/group
2187 definitions in /usr/lib/sysusers.d/. This is useful to
2188 enable factory resets and volatile systems that boot up with
2189 an empty /etc directory, and thus need system users and
2190 groups created during early boot. systemd now also ships
2191 with two default sysusers.d/ files for the most basic
2192 users and groups systemd and the core operating system
2195 * A new tmpfiles snippet has been added that rebuilds the
2196 essential files in /etc on boot, should they be missing.
2198 * A directive for ensuring automatic clean-up of
2199 /var/cache/man/ has been removed from the default
2200 configuration. This line should now be shipped by the man
2201 implementation. The necessary change has been made to the
2202 man-db implementation. Note that you need to update your man
2203 implementation to one that ships this line, otherwise no
2204 automatic clean-up of /var/cache/man will take place.
2206 * A new condition ConditionNeedsUpdate= has been added that
2207 may conditionalize services to only run when /etc or /var
2208 are "older" than the vendor operating system resources in
2209 /usr. This is useful for reconstructing or updating /etc
2210 after an offline update of /usr or a factory reset, on the
2211 next reboot. Services that want to run once after such an
2212 update or reset should use this condition and order
2213 themselves before the new systemd-update-done.service, which
2214 will mark the two directories as fully updated. A number of
2215 service files have been added making use of this, to rebuild
2216 the udev hardware database, the journald message catalog and
2217 dynamic loader cache (ldconfig). The systemd-sysusers tool
2218 described above also makes use of this now. With this in
2219 place it is now possible to start up a minimal operating
2220 system with /etc empty cleanly. For more information on the
2221 concepts involved see this recent blog story:
2223 http://0pointer.de/blog/projects/stateless.html
2225 * A new system group "input" has been introduced, and all
2226 input device nodes get this group assigned. This is useful
2227 for system-level software to get access to input devices. It
2228 complements what is already done for "audio" and "video".
2230 * systemd-networkd learnt minimal DHCPv4 server support in
2231 addition to the existing DHCPv4 client support. It also
2232 learnt DHCPv6 client and IPv6 Router Solicitation client
2233 support. The DHCPv4 client gained support for static routes
2234 passed in from the server. Note that the [DHCPv4] section
2235 known in older systemd-networkd versions has been renamed to
2236 [DHCP] and is now also used by the DHCPv6 client. Existing
2237 .network files using settings of this section should be
2238 updated, though compatibility is maintained. Optionally, the
2239 client hostname may now be sent to the DHCP server.
2241 * networkd gained support for vxlan virtual networks as well
2242 as tun/tap and dummy devices.
2244 * networkd gained support for automatic allocation of address
2245 ranges for interfaces from a system-wide pool of
2246 addresses. This is useful for dynamically managing a large
2247 number of interfaces with a single network configuration
2248 file. In particular this is useful to easily assign
2249 appropriate IP addresses to the veth links of a large number
2250 of nspawn instances.
2252 * RPM macros for processing sysusers, sysctl and binfmt
2253 drop-in snippets at package installation time have been
2256 * The /etc/os-release file should now be placed in
2257 /usr/lib/os-release. The old location is automatically
2258 created as symlink. /usr/lib is the more appropriate
2259 location of this file, since it shall actually describe the
2260 vendor operating system shipped in /usr, and not the
2261 configuration stored in /etc.
2263 * .mount units gained a new boolean SloppyOptions= setting
2264 that maps to mount(8)'s -s option which enables permissive
2265 parsing of unknown mount options.
2267 * tmpfiles learnt a new "L+" directive which creates a symlink
2268 but (unlike "L") deletes a pre-existing file first, should
2269 it already exist and not already be the correct
2270 symlink. Similarly, "b+", "c+" and "p+" directives have been
2271 added as well, which create block and character devices, as
2272 well as fifos in the filesystem, possibly removing any
2273 pre-existing files of different types.
2275 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2276 'argument' field (which so far specified the source to
2277 symlink/copy the files from) is now optional. If omitted the
2278 same file os copied from /usr/share/factory/ suffixed by the
2279 full destination path. This is useful for populating /etc
2280 with essential files, by copying them from vendor defaults
2281 shipped in /usr/share/factory/etc.
2283 * A new command "systemctl preset-all" has been added that
2284 applies the service preset settings to all installed unit
2285 files. A new switch --preset-mode= has been added that
2286 controls whether only enable or only disable operations
2289 * A new command "systemctl is-system-running" has been added
2290 that allows checking the overall state of the system, for
2291 example whether it is fully up and running.
2293 * When the system boots up with an empty /etc, the equivalent
2294 to "systemctl preset-all" is executed during early boot, to
2295 make sure all default services are enabled after a factory
2298 * systemd now contains a minimal preset file that enables the
2299 most basic services systemd ships by default.
2301 * Unit files' [Install] section gained a new DefaultInstance=
2302 field for defining the default instance to create if a
2303 template unit is enabled with no instance specified.
2305 * A new passive target cryptsetup-pre.target has been added
2306 that may be used by services that need to make they run and
2307 finish before the first LUKS cryptographic device is set up.
2309 * The /dev/loop-control and /dev/btrfs-control device nodes
2310 are now owned by the "disk" group by default, opening up
2311 access to this group.
2313 * systemd-coredump will now automatically generate a
2314 stack trace of all core dumps taking place on the system,
2315 based on elfutils' libdw library. This stack trace is logged
2318 * systemd-coredump may now optionally store coredumps directly
2319 on disk (in /var/lib/systemd/coredump, possibly compressed),
2320 instead of storing them unconditionally in the journal. This
2321 mode is the new default. A new configuration file
2322 /etc/systemd/coredump.conf has been added to configure this
2323 and other parameters of systemd-coredump.
2325 * coredumpctl gained a new "info" verb to show details about a
2326 specific coredump. A new switch "-1" has also been added
2327 that makes sure to only show information about the most
2328 recent entry instead of all entries. Also, as the tool is
2329 generally useful now the "systemd-" prefix of the binary
2330 name has been removed. Distributions that want to maintain
2331 compatibility with the old name should add a symlink from
2332 the old name to the new name.
2334 * journald's SplitMode= now defaults to "uid". This makes sure
2335 that unprivileged users can access their own coredumps with
2336 coredumpctl without restrictions.
2338 * New kernel command line options "systemd.wants=" (for
2339 pulling an additional unit during boot), "systemd.mask="
2340 (for masking a specific unit for the boot), and
2341 "systemd.debug-shell" (for enabling the debug shell on tty9)
2342 have been added. This is implemented in the new generator
2343 "systemd-debug-generator".
2345 * systemd-nspawn will now by default filter a couple of
2346 syscalls for containers, among them those required for
2347 kernel module loading, direct x86 IO port access, swap
2348 management, and kexec. Most importantly though
2349 open_by_handle_at() is now prohibited for containers,
2350 closing a hole similar to a recently discussed vulnerability
2351 in docker regarding access to files on file hierarchies the
2352 container should normally not have access to. Note that, for
2353 nspawn, we generally make no security claims anyway (and
2354 this is explicitly documented in the man page), so this is
2355 just a fix for one of the most obvious problems.
2357 * A new man page file-hierarchy(7) has been added that
2358 contains a minimized, modernized version of the file system
2359 layout systemd expects, similar in style to the FHS
2360 specification or hier(5). A new tool systemd-path(1) has
2361 been added to query many of these paths for the local
2364 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2365 longer done. Since the directory now has a per-user size
2366 limit, and is cleaned on logout this appears unnecessary,
2367 in particular since this now brings the lifecycle of this
2368 directory closer in line with how IPC objects are handled.
2370 * systemd.pc now exports a number of additional directories,
2371 including $libdir (which is useful to identify the library
2372 path for the primary architecture of the system), and a
2373 couple of drop-in directories.
2375 * udev's predictable network interface names now use the dev_port
2376 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2377 distinguish between ports of the same PCI function. dev_id should
2378 only be used for ports using the same HW address, hence the need
2381 * machined has been updated to export the OS version of a
2382 container (read from /etc/os-release and
2383 /usr/lib/os-release) on the bus. This is now shown in
2384 "machinectl status" for a machine.
2386 * A new service setting RestartForceExitStatus= has been
2387 added. If configured to a set of exit signals or process
2388 return values, the service will be restarted when the main
2389 daemon process exits with any of them, regardless of the
2392 * systemctl's -H switch for connecting to remote systemd
2393 machines has been extended so that it may be used to
2394 directly connect to a specific container on the
2395 host. "systemctl -H root@foobar:waldi" will now connect as
2396 user "root" to host "foobar", and then proceed directly to
2397 the container named "waldi". Note that currently you have to
2398 authenticate as user "root" for this to work, as entering
2399 containers is a privileged operation.
2401 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2402 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2403 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2404 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2405 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2406 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2407 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2408 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2409 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2410 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2411 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2412 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2414 -- Berlin, 2014-07-03
2418 * As an experimental feature, udev now tries to lock the
2419 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2420 executes events for the disk or any of its partitions.
2421 Applications like partitioning programs can lock the
2422 disk device node (flock(LOCK_EX)) and claim temporary
2423 device ownership that way; udev will entirely skip all event
2424 handling for this disk and its partitions. If the disk
2425 was opened for writing, the close will trigger a partition
2426 table rescan in udev's "watch" facility, and if needed
2427 synthesize "change" events for the disk and all its partitions.
2428 This is now unconditionally enabled, and if it turns out to
2429 cause major problems, we might turn it on only for specific
2430 devices, or might need to disable it entirely. Device Mapper
2431 devices are excluded from this logic.
2433 * We temporarily dropped the "-l" switch for fsck invocations,
2434 since they collide with the flock() logic above. util-linux
2435 upstream has been changed already to avoid this conflict,
2436 and we will readd "-l" as soon as util-linux with this
2437 change has been released.
2439 * The dependency on libattr has been removed. Since a long
2440 time, the extended attribute calls have moved to glibc, and
2441 libattr is thus unnecessary.
2443 * Virtualization detection works without priviliges now. This
2444 means the systemd-detect-virt binary no longer requires
2445 CAP_SYS_PTRACE file capabilities, and our daemons can run
2446 with fewer privileges.
2448 * systemd-networkd now runs under its own "systemd-network"
2449 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2450 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2451 loses the ability to write to files owned by root this way.
2453 * Similarly, systemd-resolved now runs under its own
2454 "systemd-resolve" user with no capabilities remaining.
2456 * Similarly, systemd-bus-proxyd now runs under its own
2457 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2459 * systemd-networkd gained support for setting up "veth"
2460 virtual Ethernet devices for container connectivity, as well
2461 as GRE and VTI tunnels.
2463 * systemd-networkd will no longer automatically attempt to
2464 manually load kernel modules necessary for certain tunnel
2465 transports. Instead, it is assumed the kernel loads them
2466 automatically when required. This only works correctly on
2467 very new kernels. On older kernels, please consider adding
2468 the kernel modules to /etc/modules-load.d/ as a work-around.
2470 * The resolv.conf file systemd-resolved generates has been
2471 moved to /run/systemd/resolve/. If you have a symlink from
2472 /etc/resolv.conf, it might be necessary to correct it.
2474 * Two new service settings, ProtectHome= and ProtectSystem=,
2475 have been added. When enabled, they will make the user data
2476 (such as /home) inaccessible or read-only and the system
2477 (such as /usr) read-only, for specific services. This allows
2478 very light-weight per-service sandboxing to avoid
2479 modifications of user data or system files from
2480 services. These two new switches have been enabled for all
2481 of systemd's long-running services, where appropriate.
2483 * Socket units gained new SocketUser= and SocketGroup=
2484 settings to set the owner user and group of AF_UNIX sockets
2485 and FIFOs in the file system.
2487 * Socket units gained a new RemoveOnStop= setting. If enabled,
2488 all FIFOS and sockets in the file system will be removed
2489 when the specific socket unit is stopped.
2491 * Socket units gained a new Symlinks= setting. It takes a list
2492 of symlinks to create to file system sockets or FIFOs
2493 created by the specific Unix sockets. This is useful to
2494 manage symlinks to socket nodes with the same life-cycle as
2497 * The /dev/log socket and /dev/initctl FIFO have been moved to
2498 /run, and have been replaced by symlinks. This allows
2499 connecting to these facilities even if PrivateDevices=yes is
2500 used for a service (which makes /dev/log itself unavailable,
2501 but /run is left). This also has the benefit of ensuring
2502 that /dev only contains device nodes, directories and
2503 symlinks, and nothing else.
2505 * sd-daemon gained two new calls sd_pid_notify() and
2506 sd_pid_notifyf(). They are similar to sd_notify() and
2507 sd_notifyf(), but allow overriding of the source PID of
2508 notification messages if permissions permit this. This is
2509 useful to send notify messages on behalf of a different
2510 process (for example, the parent process). The
2511 systemd-notify tool has been updated to make use of this
2512 when sending messages (so that notification messages now
2513 originate from the shell script invoking systemd-notify and
2514 not the systemd-notify process itself. This should minimize
2515 a race where systemd fails to associate notification
2516 messages to services when the originating process already
2519 * A new "on-abnormal" setting for Restart= has been added. If
2520 set, it will result in automatic restarts on all "abnormal"
2521 reasons for a process to exit, which includes unclean
2522 signals, core dumps, timeouts and watchdog timeouts, but
2523 does not include clean and unclean exit codes or clean
2524 signals. Restart=on-abnormal is an alternative for
2525 Restart=on-failure for services that shall be able to
2526 terminate and avoid restarts on certain errors, by
2527 indicating so with an unclean exit code. Restart=on-failure
2528 or Restart=on-abnormal is now the recommended setting for
2529 all long-running services.
2531 * If the InaccessibleDirectories= service setting points to a
2532 mount point (or if there are any submounts contained within
2533 it), it is now attempted to completely unmount it, to make
2534 the file systems truly unavailable for the respective
2537 * The ReadOnlyDirectories= service setting and
2538 systemd-nspawn's --read-only parameter are now recursively
2539 applied to all submounts, too.
2541 * Mount units may now be created transiently via the bus APIs.
2543 * The support for SysV and LSB init scripts has been removed
2544 from the systemd daemon itself. Instead, it is now
2545 implemented as a generator that creates native systemd units
2546 from these scripts when needed. This enables us to remove a
2547 substantial amount of legacy code from PID 1, following the
2548 fact that many distributions only ship a very small number
2549 of LSB/SysV init scripts nowadays.
2551 * Privileged Xen (dom0) domains are not considered
2552 virtualization anymore by the virtualization detection
2553 logic. After all, they generally have unrestricted access to
2554 the hardware and usually are used to manage the unprivileged
2557 * systemd-tmpfiles gained a new "C" line type, for copying
2558 files or entire directories.
2560 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2561 lines. So far, they have been non-globbing versions of the
2562 latter, and have thus been redundant. In future, it is
2563 recommended to only use "z". "m" has hence been removed
2564 from the documentation, even though it stays supported.
2566 * A tmpfiles snippet to recreate the most basic structure in
2567 /var has been added. This is enough to create the /var/run →
2568 /run symlink and create a couple of structural
2569 directories. This allows systems to boot up with an empty or
2570 volatile /var. Of course, while with this change, the core OS
2571 now is capable with dealing with a volatile /var, not all
2572 user services are ready for it. However, we hope that sooner
2573 or later, many service daemons will be changed upstream so
2574 that they are able to automatically create their necessary
2575 directories in /var at boot, should they be missing. This is
2576 the first step to allow state-less systems that only require
2577 the vendor image for /usr to boot.
2579 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2580 empty tmpfs instance to a specific directory. This is
2581 particularly useful for making use of the automatic
2582 reconstruction of /var (see above), by passing --tmpfs=/var.
2584 * Access modes specified in tmpfiles snippets may now be
2585 prefixed with "~", which indicates that they shall be masked
2586 by whether the existing file or directory is currently
2587 writable, readable or executable at all. Also, if specified,
2588 the sgid/suid/sticky bits will be masked for all
2591 * A new passive target unit "network-pre.target" has been
2592 added which is useful for services that shall run before any
2593 network is configured, for example firewall scripts.
2595 * The "floppy" group that previously owned the /dev/fd*
2596 devices is no longer used. The "disk" group is now used
2597 instead. Distributions should probably deprecate usage of
2600 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2601 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2602 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2603 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2604 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2605 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2608 -- Berlin, 2014-06-11
2612 * A new "systemd-timesyncd" daemon has been added for
2613 synchronizing the system clock across the network. It
2614 implements an SNTP client. In contrast to NTP
2615 implementations such as chrony or the NTP reference server,
2616 this only implements a client side, and does not bother with
2617 the full NTP complexity, focusing only on querying time from
2618 one remote server and synchronizing the local clock to
2619 it. Unless you intend to serve NTP to networked clients or
2620 want to connect to local hardware clocks, this simple NTP
2621 client should be more than appropriate for most
2622 installations. The daemon runs with minimal privileges, and
2623 has been hooked up with networkd to only operate when
2624 network connectivity is available. The daemon saves the
2625 current clock to disk every time a new NTP sync has been
2626 acquired, and uses this to possibly correct the system clock
2627 early at bootup, in order to accommodate for systems that
2628 lack an RTC such as the Raspberry Pi and embedded devices,
2629 and to make sure that time monotonically progresses on these
2630 systems, even if it is not always correct. To make use of
2631 this daemon, a new system user and group "systemd-timesync"
2632 needs to be created on installation of systemd.
2634 * The queue "seqnum" interface of libudev has been disabled, as
2635 it was generally incompatible with device namespacing as
2636 sequence numbers of devices go "missing" if the devices are
2637 part of a different namespace.
2639 * "systemctl list-timers" and "systemctl list-sockets" gained
2640 a --recursive switch for showing units of these types also
2641 for all local containers, similar in style to the already
2642 supported --recursive switch for "systemctl list-units".
2644 * A new RebootArgument= setting has been added for service
2645 units, which may be used to specify a kernel reboot argument
2646 to use when triggering reboots with StartLimitAction=.
2648 * A new FailureAction= setting has been added for service
2649 units which may be used to specify an operation to trigger
2650 when a service fails. This works similarly to
2651 StartLimitAction=, but unlike it, controls what is done
2652 immediately rather than only after several attempts to
2653 restart the service in question.
2655 * hostnamed got updated to also expose the kernel name,
2656 release, and version on the bus. This is useful for
2657 executing commands like hostnamectl with the -H switch.
2658 systemd-analyze makes use of this to properly display
2659 details when running non-locally.
2661 * The bootchart tool can now show cgroup information in the
2662 graphs it generates.
2664 * The CFS CPU quota cgroup attribute is now exposed for
2665 services. The new CPUQuota= switch has been added for this
2666 which takes a percentage value. Setting this will have the
2667 result that a service may never get more CPU time than the
2668 specified percentage, even if the machine is otherwise idle.
2670 * systemd-networkd learned IPIP and SIT tunnel support.
2672 * LSB init scripts exposing a dependency on $network will now
2673 get a dependency on network-online.target rather than simply
2674 network.target. This should bring LSB handling closer to
2675 what it was on SysV systems.
2677 * A new fsck.repair= kernel option has been added to control
2678 how fsck shall deal with unclean file systems at boot.
2680 * The (.ini) configuration file parser will now silently
2681 ignore sections whose name begins with "X-". This may be
2682 used to maintain application-specific extension sections in unit
2685 * machined gained a new API to query the IP addresses of
2686 registered containers. "machinectl status" has been updated
2687 to show these addresses in its output.
2689 * A new call sd_uid_get_display() has been added to the
2690 sd-login APIs for querying the "primary" session of a
2691 user. The "primary" session of the user is elected from the
2692 user's sessions and generally a graphical session is
2693 preferred over a text one.
2695 * A minimal systemd-resolved daemon has been added. It
2696 currently simply acts as a companion to systemd-networkd and
2697 manages resolv.conf based on per-interface DNS
2698 configuration, possibly supplied via DHCP. In the long run
2699 we hope to extend this into a local DNSSEC enabled DNS and
2702 * The systemd-networkd-wait-online tool is now enabled by
2703 default. It will delay network-online.target until a network
2704 connection has been configured. The tool primarily integrates
2705 with networkd, but will also make a best effort to make sense
2706 of network configuration performed in some other way.
2708 * Two new service options StartupCPUShares= and
2709 StartupBlockIOWeight= have been added that work similarly to
2710 CPUShares= and BlockIOWeight= however only apply during
2711 system startup. This is useful to prioritize certain services
2712 differently during bootup than during normal runtime.
2714 * hostnamed has been changed to prefer the statically
2715 configured hostname in /etc/hostname (unless set to
2716 'localhost' or empty) over any dynamic one supplied by
2717 dhcp. With this change, the rules for picking the hostname
2718 match more closely the rules of other configuration settings
2719 where the local administrator's configuration in /etc always
2720 overrides any other settings.
2722 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2723 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2724 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2725 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2726 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2727 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2728 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2729 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2730 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2731 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2732 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2733 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2734 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2735 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2736 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2737 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2740 -- Beijing, 2014-05-28
2744 * When restoring the screen brightness at boot, stay away from
2745 the darkest setting or from the lowest 5% of the available
2746 range, depending on which is the larger value of both. This
2747 should effectively protect the user from rebooting into a
2748 black screen, should the brightness have been set to minimum
2751 * sd-login gained a new sd_machine_get_class() call to
2752 determine the class ("vm" or "container") of a machine
2753 registered with machined.
2755 * sd-login gained new calls
2756 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2757 to query the identity of the peer of a local AF_UNIX
2758 connection. They operate similarly to their sd_pid_get_xyz()
2761 * PID 1 will now maintain a system-wide system state engine
2762 with the states "starting", "running", "degraded",
2763 "maintenance", "stopping". These states are bound to system
2764 startup, normal runtime, runtime with at least one failed
2765 service, rescue/emergency mode and system shutdown. This
2766 state is shown in the "systemctl status" output when no unit
2767 name is passed. It is useful to determine system state, in
2768 particularly when doing so for many systems or containers at
2771 * A new command "list-machines" has been added to "systemctl"
2772 that lists all local OS containers and shows their system
2773 state (see above), if systemd runs inside of them.
2775 * systemctl gained a new "-r" switch to recursively enumerate
2776 units on all local containers, when used with the
2777 "list-unit" command (which is the default one that is
2778 executed when no parameters are specified).
2780 * The GPT automatic partition discovery logic will now honour
2781 two GPT partition flags: one may be set on a partition to
2782 cause it to be mounted read-only, and the other may be set
2783 on a partition to ignore it during automatic discovery.
2785 * Two new GPT type UUIDs have been added for automatic root
2786 partition discovery, for 32-bit and 64-bit ARM. This is not
2787 particularly useful for discovering the root directory on
2788 these architectures during bare-metal boots (since UEFI is
2789 not common there), but still very useful to allow booting of
2790 ARM disk images in nspawn with the -i option.
2792 * MAC addresses of interfaces created with nspawn's
2793 --network-interface= switch will now be generated from the
2794 machine name, and thus be stable between multiple invocations
2797 * logind will now automatically remove all IPC objects owned
2798 by a user if she or he fully logs out. This makes sure that
2799 users who are logged out cannot continue to consume IPC
2800 resources. This covers SysV memory, semaphores and message
2801 queues as well as POSIX shared memory and message
2802 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2803 limits. With this functionality, that is corrected. This may
2804 be turned off by using the RemoveIPC= switch of logind.conf.
2806 * The systemd-machine-id-setup and tmpfiles tools gained a
2807 --root= switch to operate on a specific root directory,
2810 * journald can now forward logged messages to the TTYs of all
2811 logged in users ("wall"). This is the default for all
2812 emergency messages now.
2814 * A new tool systemd-journal-remote has been added to stream
2815 journal log messages across the network.
2817 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2818 controller trees are mounted into it. Note that the
2819 directories mounted beneath it are not read-only. This is a
2820 security measure and is particularly useful because glibc
2821 actually includes a search logic to pick any tmpfs it can
2822 find to implement shm_open() if /dev/shm is not available
2823 (which it might very well be in namespaced setups).
2825 * machinectl gained a new "poweroff" command to cleanly power
2826 down a local OS container.
2828 * The PrivateDevices= unit file setting will now also drop the
2829 CAP_MKNOD capability from the capability bound set, and
2830 imply DevicePolicy=closed.
2832 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2833 comprehensively on all long-running systemd services where
2834 this is appropriate.
2836 * systemd-udevd will now run in a disassociated mount
2837 namespace. To mount directories from udev rules, make sure to
2838 pull in mount units via SYSTEMD_WANTS properties.
2840 * The kdbus support gained support for uploading policy into
2841 the kernel. sd-bus gained support for creating "monitoring"
2842 connections that can eavesdrop into all bus communication
2843 for debugging purposes.
2845 * Timestamps may now be specified in seconds since the UNIX
2846 epoch Jan 1st, 1970 by specifying "@" followed by the value
2849 * Native tcpwrap support in systemd has been removed. tcpwrap
2850 is old code, not really maintained anymore and has serious
2851 shortcomings, and better options such as firewalls
2852 exist. For setups that require tcpwrap usage, please
2853 consider invoking your socket-activated service via tcpd,
2854 like on traditional inetd.
2856 * A new system.conf configuration option
2857 DefaultTimerAccuracySec= has been added that controls the
2858 default AccuracySec= setting of .timer units.
2860 * Timer units gained a new WakeSystem= switch. If enabled,
2861 timers configured this way will cause the system to resume
2862 from system suspend (if the system supports that, which most
2865 * Timer units gained a new Persistent= switch. If enabled,
2866 timers configured this way will save to disk when they have
2867 been last triggered. This information is then used on next
2868 reboot to possible execute overdue timer events, that
2869 could not take place because the system was powered off.
2870 This enables simple anacron-like behaviour for timer units.
2872 * systemctl's "list-timers" will now also list the time a
2873 timer unit was last triggered in addition to the next time
2874 it will be triggered.
2876 * systemd-networkd will now assign predictable IPv4LL
2877 addresses to its local interfaces.
2879 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2880 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2881 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2882 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2883 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2884 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2885 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2886 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2889 -- Berlin, 2014-03-25
2893 * A new unit file setting RestrictAddressFamilies= has been
2894 added to restrict which socket address families unit
2895 processes gain access to. This takes address family names
2896 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2897 attack surface of services via exotic protocol stacks. This
2898 is built on seccomp system call filters.
2900 * Two new unit file settings RuntimeDirectory= and
2901 RuntimeDirectoryMode= have been added that may be used to
2902 manage a per-daemon runtime directories below /run. This is
2903 an alternative for setting up directory permissions with
2904 tmpfiles snippets, and has the advantage that the runtime
2905 directory's lifetime is bound to the daemon runtime and that
2906 the daemon starts up with an empty directory each time. This
2907 is particularly useful when writing services that drop
2908 privileges using the User= or Group= setting.
2910 * The DeviceAllow= unit setting now supports globbing for
2911 matching against device group names.
2913 * The systemd configuration file system.conf gained new
2914 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2915 DefaultMemoryAccounting= to globally turn on/off accounting
2916 for specific resources (cgroups) for all units. These
2917 settings may still be overridden individually in each unit
2920 * systemd-gpt-auto-generator is now able to discover /srv and
2921 root partitions in addition to /home and swap partitions. It
2922 also supports LUKS-encrypted partitions now. With this in
2923 place, automatic discovery of partitions to mount following
2924 the Discoverable Partitions Specification
2925 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2926 is now a lot more complete. This allows booting without
2927 /etc/fstab and without root= on the kernel command line on
2928 systems prepared appropriately.
2930 * systemd-nspawn gained a new --image= switch which allows
2931 booting up disk images and Linux installations on any block
2932 device that follow the Discoverable Partitions Specification
2933 (see above). This means that installations made with
2934 appropriately updated installers may now be started and
2935 deployed using container managers, completely
2936 unmodified. (We hope that libvirt-lxc will add support for
2937 this feature soon, too.)
2939 * systemd-nspawn gained a new --network-macvlan= setting to
2940 set up a private macvlan interface for the
2941 container. Similarly, systemd-networkd gained a new
2942 Kind=macvlan setting in .netdev files.
2944 * systemd-networkd now supports configuring local addresses
2947 * A new tool systemd-network-wait-online has been added to
2948 synchronously wait for network connectivity using
2951 * The sd-bus.h bus API gained a new sd_bus_track object for
2952 tracking the life-cycle of bus peers. Note that sd-bus.h is
2953 still not a public API though (unless you specify
2954 --enable-kdbus on the configure command line, which however
2955 voids your warranty and you get no API stability guarantee).
2957 * The $XDG_RUNTIME_DIR runtime directories for each user are
2958 now individual tmpfs instances, which has the benefit of
2959 introducing separate pools for each user, with individual
2960 size limits, and thus making sure that unprivileged clients
2961 can no longer negatively impact the system or other users by
2962 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2963 RuntimeDirectorySize= has been introduced that allows
2964 controlling the default size limit for all users. It
2965 defaults to 10% of the available physical memory. This is no
2966 replacement for quotas on tmpfs though (which the kernel
2967 still does not support), as /dev/shm and /tmp are still
2968 shared resources used by both the system and unprivileged
2971 * logind will now automatically turn off automatic suspending
2972 on laptop lid close when more than one display is
2973 connected. This was previously expected to be implemented
2974 individually in desktop environments (such as GNOME),
2975 however has been added to logind now, in order to fix a
2976 boot-time race where a desktop environment might not have
2977 been started yet and thus not been able to take an inhibitor
2978 lock at the time where logind already suspends the system
2979 due to a closed lid.
2981 * logind will now wait at least 30s after each system
2982 suspend/resume cycle, and 3min after system boot before
2983 suspending the system due to a closed laptop lid. This
2984 should give USB docking stations and similar enough time to
2985 be probed and configured after system resume and boot in
2986 order to then act as suspend blocker.
2988 * systemd-run gained a new --property= setting which allows
2989 initialization of resource control properties (and others)
2990 for the created scope or service unit. Example: "systemd-run
2991 --property=BlockIOWeight=10 updatedb" may be used to run
2992 updatedb at a low block IO scheduling weight.
2994 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2995 now also work in --scope mode.
2997 * When systemd is compiled with kdbus support, basic support
2998 for enforced policies is now in place. (Note that enabling
2999 kdbus still voids your warranty and no API compatibility
3002 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3003 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3004 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3005 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3006 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3007 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3008 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3009 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3010 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3011 Zbigniew Jędrzejewski-Szmek
3013 -- Berlin, 2014-03-12
3017 * systemd will now relabel /dev after loading the SMACK policy
3018 according to SMACK rules.
3020 * A new unit file option AppArmorProfile= has been added to
3021 set the AppArmor profile for the processes of a unit.
3023 * A new condition check ConditionArchitecture= has been added
3024 to conditionalize units based on the system architecture, as
3025 reported by uname()'s "machine" field.
3027 * systemd-networkd now supports matching on the system
3028 virtualization, architecture, kernel command line, host name
3031 * logind is now a lot more aggressive when suspending the
3032 machine due to a closed laptop lid. Instead of acting only
3033 on the lid close action, it will continuously watch the lid
3034 status and act on it. This is useful for laptops where the
3035 power button is on the outside of the chassis so that it can
3036 be reached without opening the lid (such as the Lenovo
3037 Yoga). On those machines, logind will now immediately
3038 re-suspend the machine if the power button has been
3039 accidentally pressed while the laptop was suspended and in a
3040 backpack or similar.
3042 * logind will now watch SW_DOCK switches and inhibit reaction
3043 to the lid switch if it is pressed. This means that logind
3044 will not suspend the machine anymore if the lid is closed
3045 and the system is docked, if the laptop supports SW_DOCK
3046 notifications via the input layer. Note that ACPI docking
3047 stations do not generate this currently. Also note that this
3048 logic is usually not fully sufficient and Desktop
3049 Environments should take a lid switch inhibitor lock when an
3050 external display is connected, as systemd will not watch
3053 * nspawn will now make use of the devices cgroup controller by
3054 default, and only permit creation of and access to the usual
3055 API device nodes like /dev/null or /dev/random, as well as
3056 access to (but not creation of) the pty devices.
3058 * We will now ship a default .network file for
3059 systemd-networkd that automatically configures DHCP for
3060 network interfaces created by nspawn's --network-veth or
3061 --network-bridge= switches.
3063 * systemd will now understand the usual M, K, G, T suffixes
3064 according to SI conventions (i.e. to the base 1000) when
3065 referring to throughput and hardware metrics. It will stay
3066 with IEC conventions (i.e. to the base 1024) for software
3067 metrics, according to what is customary according to
3068 Wikipedia. We explicitly document which base applies for
3069 each configuration option.
3071 * The DeviceAllow= setting in unit files now supports a syntax
3072 to whitelist an entire group of devices node majors at once,
3073 based on the /proc/devices listing. For example, with the
3074 string "char-pts", it is now possible to whitelist all
3075 current and future pseudo-TTYs at once.
3077 * sd-event learned a new "post" event source. Event sources of
3078 this type are triggered by the dispatching of any event
3079 source of a type that is not "post". This is useful for
3080 implementing clean-up and check event sources that are
3081 triggered by other work being done in the program.
3083 * systemd-networkd is no longer statically enabled, but uses
3084 the usual [Install] sections so that it can be
3085 enabled/disabled using systemctl. It still is enabled by
3088 * When creating a veth interface pair with systemd-nspawn, the
3089 host side will now be prefixed with "vb-" if
3090 --network-bridge= is used, and with "ve-" if --network-veth
3091 is used. This way, it is easy to distinguish these cases on
3092 the host, for example to apply different configuration to
3093 them with systemd-networkd.
3095 * The compatibility libraries for libsystemd-journal.so,
3096 libsystem-id128.so, libsystemd-login.so and
3097 libsystemd-daemon.so do not make use of IFUNC
3098 anymore. Instead, we now build libsystemd.so multiple times
3099 under these alternative names. This means that the footprint
3100 is drastically increased, but given that these are
3101 transitional compatibility libraries, this should not matter
3102 much. This change has been made necessary to support the ARM
3103 platform for these compatibility libraries, as the ARM
3104 toolchain is not really at the same level as the toolchain
3105 for other architectures like x86 and does not support
3106 IFUNC. Please make sure to use --enable-compat-libs only
3107 during a transitional period!
3109 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3110 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3111 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3112 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3113 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3114 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3115 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3116 Zbigniew Jędrzejewski-Szmek
3118 -- Berlin, 2014-02-24
3122 * A new component "systemd-networkd" has been added that can
3123 be used to configure local network interfaces statically or
3124 via DHCP. It is capable of bringing up bridges, VLANs, and
3125 bonding. Currently, no hook-ups for interactive network
3126 configuration are provided. Use this for your initrd,
3127 container, embedded, or server setup if you need a simple,
3128 yet powerful, network configuration solution. This
3129 configuration subsystem is quite nifty, as it allows wildcard
3130 hotplug matching in interfaces. For example, with a single
3131 configuration snippet, you can configure that all Ethernet
3132 interfaces showing up are automatically added to a bridge,
3133 or similar. It supports link-sensing and more.
3135 * A new tool "systemd-socket-proxyd" has been added which can
3136 act as a bidirectional proxy for TCP sockets. This is
3137 useful for adding socket activation support to services that
3138 do not actually support socket activation, including virtual
3139 machines and the like.
3141 * Add a new tool to save/restore rfkill state on
3144 * Save/restore state of keyboard backlights in addition to
3145 display backlights on shutdown/boot.
3147 * udev learned a new SECLABEL{} construct to label device
3148 nodes with a specific security label when they appear. For
3149 now, only SECLABEL{selinux} is supported, but the syntax is
3150 prepared for additional security frameworks.
3152 * udev gained a new scheme to configure link-level attributes
3153 from files in /etc/systemd/network/*.link. These files can
3154 match against MAC address, device path, driver name and type,
3155 and will apply attributes like the naming policy, link speed,
3156 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3157 address assignment policy (randomized, ...).
3159 * The configuration of network interface naming rules for
3160 "permanent interface names" has changed: a new NamePolicy=
3161 setting in the [Link] section of .link files determines the
3162 priority of possible naming schemes (onboard, slot, MAC,
3163 path). The default value of this setting is determined by
3164 /usr/lib/net/links/99-default.link. Old
3165 80-net-name-slot.rules udev configuration file has been
3166 removed, so local configuration overriding this file should
3167 be adapated to override 99-default.link instead.
3169 * When the User= switch is used in a unit file, also
3170 initialize $SHELL= based on the user database entry.
3172 * systemd no longer depends on libdbus. All communication is
3173 now done with sd-bus, systemd's low-level bus library
3176 * kdbus support has been added to PID 1 itself. When kdbus is
3177 enabled, this causes PID 1 to set up the system bus and
3178 enable support for a new ".busname" unit type that
3179 encapsulates bus name activation on kdbus. It works a little
3180 bit like ".socket" units, except for bus names. A new
3181 generator has been added that converts classic dbus1 service
3182 activation files automatically into native systemd .busname
3185 * sd-bus: add a light-weight vtable implementation that allows
3186 defining objects on the bus with a simple static const
3187 vtable array of its methods, signals and properties.
3189 * systemd will not generate or install static dbus
3190 introspection data anymore to /usr/share/dbus-1/interfaces,
3191 as the precise format of these files is unclear, and
3192 nothing makes use of it.
3194 * A proxy daemon is now provided to proxy clients connecting
3195 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3196 compatibility with classic D-Bus.
3198 * A bus driver implementation has been added that supports the
3199 classic D-Bus bus driver calls on kdbus, also for
3200 compatibility purposes.
3202 * A new API "sd-event.h" has been added that implements a
3203 minimal event loop API built around epoll. It provides a
3204 couple of features that direct epoll usage is lacking:
3205 prioritization of events, scales to large numbers of timer
3206 events, per-event timer slack (accuracy), system-wide
3207 coalescing of timer events, exit handlers, watchdog
3208 supervision support using systemd's sd_notify() API, child
3211 * A new API "sd-rntl.h" has been added that provides an API
3212 around the route netlink interface of the kernel, similar in
3213 style to "sd-bus.h".
3215 * A new API "sd-dhcp-client.h" has been added that provides a
3216 small DHCPv4 client-side implementation. This is used by
3219 * There is a new kernel command line option
3220 "systemd.restore_state=0|1". When set to "0", none of the
3221 systemd tools will restore saved runtime state to hardware
3222 devices. More specifically, the rfkill and backlight states
3225 * The FsckPassNo= compatibility option in mount/service units
3226 has been removed. The fstab generator will now add the
3227 necessary dependencies automatically, and does not require
3228 PID1's support for that anymore.
3230 * journalctl gained a new switch, --list-boots, that lists
3231 recent boots with their times and boot IDs.
3233 * The various tools like systemctl, loginctl, timedatectl,
3234 busctl, systemd-run, ... have gained a new switch "-M" to
3235 connect to a specific, local OS container (as direct
3236 connection, without requiring SSH). This works on any
3237 container that is registered with machined, such as those
3238 created by libvirt-lxc or nspawn.
3240 * systemd-run and systemd-analyze also gained support for "-H"
3241 to connect to remote hosts via SSH. This is particularly
3242 useful for systemd-run because it enables queuing of jobs
3243 onto remote systems.
3245 * machinectl gained a new command "login" to open a getty
3246 login in any local container. This works with any container
3247 that is registered with machined (such as those created by
3248 libvirt-lxc or nspawn), and which runs systemd inside.
3250 * machinectl gained a new "reboot" command that may be used to
3251 trigger a reboot on a specific container that is registered
3252 with machined. This works on any container that runs an init
3253 system of some kind.
3255 * systemctl gained a new "list-timers" command to print a nice
3256 listing of installed timer units with the times they elapse
3259 * Alternative reboot() parameters may now be specified on the
3260 "systemctl reboot" command line and are passed to the
3261 reboot() system call.
3263 * systemctl gained a new --job-mode= switch to configure the
3264 mode to queue a job with. This is a more generic version of
3265 --fail, --irreversible, and --ignore-dependencies, which are
3266 still available but not advertised anymore.
3268 * /etc/systemd/system.conf gained new settings to configure
3269 various default timeouts of units, as well as the default
3270 start limit interval and burst. These may still be overridden
3273 * PID1 will now export on the bus profile data of the security
3274 policy upload process (such as the SELinux policy upload to
3277 * journald: when forwarding logs to the console, include
3278 timestamps (following the setting in
3279 /sys/module/printk/parameters/time).
3281 * OnCalendar= in timer units now understands the special
3282 strings "yearly" and "annually". (Both are equivalent)
3284 * The accuracy of timer units is now configurable with the new
3285 AccuracySec= setting. It defaults to 1min.
3287 * A new dependency type JoinsNamespaceOf= has been added that
3288 allows running two services within the same /tmp and network
3289 namespace, if PrivateNetwork= or PrivateTmp= are used.
3291 * A new command "cat" has been added to systemctl. It outputs
3292 the original unit file of a unit, and concatenates the
3293 contents of additional "drop-in" unit file snippets, so that
3294 the full configuration is shown.
3296 * systemctl now supports globbing on the various "list-xyz"
3297 commands, like "list-units" or "list-sockets", as well as on
3298 those commands which take multiple unit names.
3300 * journalctl's --unit= switch gained support for globbing.
3302 * All systemd daemons now make use of the watchdog logic so
3303 that systemd automatically notices when they hang.
3305 * If the $container_ttys environment variable is set,
3306 getty-generator will automatically spawn a getty for each
3307 listed tty. This is useful for container managers to request
3308 login gettys to be spawned on as many ttys as needed.
3310 * %h, %s, %U specifier support is not available anymore when
3311 used in unit files for PID 1. This is because NSS calls are
3312 not safe from PID 1. They stay available for --user
3313 instances of systemd, and as special case for the root user.
3315 * loginctl gained a new "--no-legend" switch to turn off output
3318 * The "sd-login.h" API gained three new calls:
3319 sd_session_is_remote(), sd_session_get_remote_user(),
3320 sd_session_get_remote_host() to query information about
3323 * The udev hardware database now also carries vendor/product
3324 information of SDIO devices.
3326 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3327 determine whether watchdog notifications are requested by
3330 * Socket-activated per-connection services now include a
3331 short description of the connection parameters in the
3334 * tmpfiles gained a new "--boot" option. When this is not used,
3335 only lines where the command character is not suffixed with
3336 "!" are executed. When this option is specified, those
3337 options are executed too. This partitions tmpfiles
3338 directives into those that can be safely executed at any
3339 time, and those which should be run only at boot (for
3340 example, a line that creates /run/nologin).
3342 * A new API "sd-resolve.h" has been added which provides a simple
3343 asynchronous wrapper around glibc NSS host name resolution
3344 calls, such as getaddrinfo(). In contrast to glibc's
3345 getaddrinfo_a(), it does not use signals. In contrast to most
3346 other asynchronous name resolution libraries, this one does
3347 not reimplement DNS, but reuses NSS, so that alternate
3348 host name resolution systems continue to work, such as mDNS,
3349 LDAP, etc. This API is based on libasyncns, but it has been
3350 cleaned up for inclusion in systemd.
3352 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3353 "sd-daemon.h" are no longer found in individual libraries
3354 libsystemd-journal.so, libsystemd-login.so,
3355 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3356 merged them into a single library, libsystemd.so, which
3357 provides all symbols. The reason for this is cyclic
3358 dependencies, as these libraries tend to use each other's
3359 symbols. So far, we have managed to workaround that by linking
3360 a copy of a good part of our code into each of these
3361 libraries again and again, which, however, makes certain
3362 things hard to do, like sharing static variables. Also, it
3363 substantially increases footprint. With this change, there
3364 is only one library for the basic APIs systemd
3365 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3366 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3367 library as well, however are subject to the --enable-kdbus
3368 switch (see below). Note that "sd-dhcp-client.h" is not part
3369 of this library (this is because it only consumes, never
3370 provides, services of/to other APIs). To make the transition
3371 easy from the separate libraries to the unified one, we
3372 provide the --enable-compat-libs compile-time switch which
3373 will generate stub libraries that are compatible with the
3374 old ones but redirect all calls to the new one.
3376 * All of the kdbus logic and the new APIs "sd-bus.h",
3377 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3378 and "sd-utf8.h" are compile-time optional via the
3379 "--enable-kdbus" switch, and they are not compiled in by
3380 default. To make use of kdbus, you have to explicitly enable
3381 the switch. Note however, that neither the kernel nor the
3382 userspace API for all of this is considered stable yet. We
3383 want to maintain the freedom to still change the APIs for
3384 now. By specifying this build-time switch, you acknowledge
3385 that you are aware of the instability of the current
3388 * Also, note that while kdbus is pretty much complete,
3389 it lacks one thing: proper policy support. This means you
3390 can build a fully working system with all features; however,
3391 it will be highly insecure. Policy support will be added in
3392 one of the next releases, at the same time that we will
3393 declare the APIs stable.
3395 * When the kernel command line argument "kdbus" is specified,
3396 systemd will automatically load the kdbus.ko kernel module. At
3397 this stage of development, it is only useful for testing kdbus
3398 and should not be used in production. Note: if "--enable-kdbus"
3399 is specified, and the kdbus.ko kernel module is available, and
3400 "kdbus" is added to the kernel command line, the entire system
3401 runs with kdbus instead of dbus-daemon, with the above mentioned
3402 problem of missing the system policy enforcement. Also a future
3403 version of kdbus.ko or a newer systemd will not be compatible with
3404 each other, and will unlikely be able to boot the machine if only
3405 one of them is updated.
3407 * systemctl gained a new "import-environment" command which
3408 uploads the caller's environment (or parts thereof) into the
3409 service manager so that it is inherited by services started
3410 by the manager. This is useful to upload variables like
3411 $DISPLAY into the user service manager.
3413 * A new PrivateDevices= switch has been added to service units
3414 which allows running a service with a namespaced /dev
3415 directory that does not contain any device nodes for
3416 physical devices. More specifically, it only includes devices
3417 such as /dev/null, /dev/urandom, and /dev/zero which are API
3420 * logind has been extended to support behaviour like VT
3421 switching on seats that do not support a VT. This makes
3422 multi-session available on seats that are not the first seat
3423 (seat0), and on systems where kernel support for VTs has
3424 been disabled at compile-time.
3426 * If a process holds a delay lock for system sleep or shutdown
3427 and fails to release it in time, we will now log its
3428 identity. This makes it easier to identify processes that
3429 cause slow suspends or power-offs.
3431 * When parsing /etc/crypttab, support for a new key-slot=
3432 option as supported by Debian is added. It allows indicating
3433 which LUKS slot to use on disk, speeding up key loading.
3435 * The sd_journald_sendv() API call has been checked and
3436 officially declared to be async-signal-safe so that it may
3437 be invoked from signal handlers for logging purposes.
3439 * Boot-time status output is now enabled automatically after a
3440 short timeout if boot does not progress, in order to give
3441 the user an indication what she or he is waiting for.
3443 * The boot-time output has been improved to show how much time
3444 remains until jobs expire.
3446 * The KillMode= switch in service units gained a new possible
3447 value "mixed". If set, and the unit is shut down, then the
3448 initial SIGTERM signal is sent only to the main daemon
3449 process, while the following SIGKILL signal is sent to
3450 all remaining processes of the service.
3452 * When a scope unit is registered, a new property "Controller"
3453 may be set. If set to a valid bus name, systemd will send a
3454 RequestStop() signal to this name when it would like to shut
3455 down the scope. This may be used to hook manager logic into
3456 the shutdown logic of scope units. Also, scope units may now
3457 be put in a special "abandoned" state, in which case the
3458 manager process which created them takes no further
3459 responsibilities for it.
3461 * When reading unit files, systemd will now verify
3462 the access mode of these files, and warn about certain
3463 suspicious combinations. This has been added to make it
3464 easier to track down packaging bugs where unit files are
3465 marked executable or world-writable.
3467 * systemd-nspawn gained a new "--setenv=" switch to set
3468 container-wide environment variables. The similar option in
3469 systemd-activate was renamed from "--environment=" to
3470 "--setenv=" for consistency.
3472 * systemd-nspawn has been updated to create a new kdbus domain
3473 for each container that is invoked, thus allowing each
3474 container to have its own set of system and user buses,
3475 independent of the host.
3477 * systemd-nspawn gained a new --drop-capability= switch to run
3478 the container with less capabilities than the default. Both
3479 --drop-capability= and --capability= now take the special
3480 string "all" for dropping or keeping all capabilities.
3482 * systemd-nspawn gained new switches for executing containers
3483 with specific SELinux labels set.
3485 * systemd-nspawn gained a new --quiet switch to not generate
3486 any additional output but the container's own console
3489 * systemd-nspawn gained a new --share-system switch to run a
3490 container without PID namespacing enabled.
3492 * systemd-nspawn gained a new --register= switch to control
3493 whether the container is registered with systemd-machined or
3494 not. This is useful for containers that do not run full
3495 OS images, but only specific apps.
3497 * systemd-nspawn gained a new --keep-unit which may be used
3498 when invoked as the only program from a service unit, and
3499 results in registration of the unit service itself in
3500 systemd-machined, instead of a newly opened scope unit.
3502 * systemd-nspawn gained a new --network-interface= switch for
3503 moving arbitrary interfaces to the container. The new
3504 --network-veth switch creates a virtual Ethernet connection
3505 between host and container. The new --network-bridge=
3506 switch then allows assigning the host side of this virtual
3507 Ethernet connection to a bridge device.
3509 * systemd-nspawn gained a new --personality= switch for
3510 setting the kernel personality for the container. This is
3511 useful when running a 32-bit container on a 64-bit host. A
3512 similar option Personality= is now also available for service
3515 * logind will now also track a "Desktop" identifier for each
3516 session which encodes the desktop environment of it. This is
3517 useful for desktop environments that want to identify
3518 multiple running sessions of itself easily.
3520 * A new SELinuxContext= setting for service units has been
3521 added that allows setting a specific SELinux execution
3522 context for a service.
3524 * Most systemd client tools will now honour $SYSTEMD_LESS for
3525 settings of the "less" pager. By default, these tools will
3526 override $LESS to allow certain operations to work, such as
3527 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3528 influence this logic.
3530 * systemd's "seccomp" hook-up has been changed to make use of
3531 the libseccomp library instead of using its own
3532 implementation. This has benefits for portability among
3535 * For usage together with SystemCallFilter=, a new
3536 SystemCallErrorNumber= setting has been introduced that
3537 allows configuration of a system error number to be returned
3538 on filtered system calls, instead of immediately killing the
3539 process. Also, SystemCallArchitectures= has been added to
3540 limit access to system calls of a particular architecture
3541 (in order to turn off support for unused secondary
3542 architectures). There is also a global
3543 SystemCallArchitectures= setting in system.conf now to turn
3544 off support for non-native system calls system-wide.
3546 * systemd requires a kernel with a working name_to_handle_at(),
3547 please see the kernel config requirements in the README file.
3549 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3550 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3551 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3552 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3553 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3554 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3555 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3556 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3557 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3558 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3559 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3560 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3561 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3562 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3563 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3564 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3565 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3566 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3567 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3568 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3569 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3570 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3571 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3572 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3574 -- Berlin, 2014-02-20
3578 * logind has gained support for facilitating privileged input
3579 and drm device access for unprivileged clients. This work is
3580 useful to allow Wayland display servers (and similar
3581 programs, such as kmscon) to run under the user's ID and
3582 access input and drm devices which are normally
3583 protected. When this is used (and the kernel is new enough)
3584 logind will "mute" IO on the file descriptors passed to
3585 Wayland as long as it is in the background and "unmute" it
3586 if it returns into the foreground. This allows secure
3587 session switching without allowing background sessions to
3588 eavesdrop on input and display data. This also introduces
3589 session switching support if VT support is turned off in the
3590 kernel, and on seats that are not seat0.
3592 * A new kernel command line option luks.options= is understood
3593 now which allows specifying LUKS options for usage for LUKS
3594 encrypted partitions specified with luks.uuid=.
3596 * tmpfiles.d(5) snippets may now use specifier expansion in
3597 path names. More specifically %m, %b, %H, %v, are now
3598 replaced by the local machine id, boot id, hostname, and
3599 kernel version number.
3601 * A new tmpfiles.d(5) command "m" has been introduced which
3602 may be used to change the owner/group/access mode of a file
3603 or directory if it exists, but do nothing if it does not.
3605 * This release removes high-level support for the
3606 MemorySoftLimit= cgroup setting. The underlying kernel
3607 cgroup attribute memory.soft_limit= is currently badly
3608 designed and likely to be removed from the kernel API in its
3609 current form, hence we should not expose it for now.
3611 * The memory.use_hierarchy cgroup attribute is now enabled for
3612 all cgroups systemd creates in the memory cgroup
3613 hierarchy. This option is likely to be come the built-in
3614 default in the kernel anyway, and the non-hierarchical mode
3615 never made much sense in the intrinsically hierarchical
3618 * A new field _SYSTEMD_SLICE= is logged along with all journal
3619 messages containing the slice a message was generated
3620 from. This is useful to allow easy per-customer filtering of
3621 logs among other things.
3623 * systemd-journald will no longer adjust the group of journal
3624 files it creates to the "systemd-journal" group. Instead we
3625 rely on the journal directory to be owned by the
3626 "systemd-journal" group, and its setgid bit set, so that the
3627 kernel file system layer will automatically enforce that
3628 journal files inherit this group assignment. The reason for
3629 this change is that we cannot allow NSS look-ups from
3630 journald which would be necessary to resolve
3631 "systemd-journal" to a numeric GID, because this might
3632 create deadlocks if NSS involves synchronous queries to
3633 other daemons (such as nscd, or sssd) which in turn are
3634 logging clients of journald and might block on it, which
3635 would then dead lock. A tmpfiles.d(5) snippet included in
3636 systemd will make sure the setgid bit and group are
3637 properly set on the journal directory if it exists on every
3638 boot. However, we recommend adjusting it manually after
3639 upgrades too (or from RPM scriptlets), so that the change is
3640 not delayed until next reboot.
3642 * Backlight and random seed files in /var/lib/ have moved into
3643 the /var/lib/systemd/ directory, in order to centralize all
3644 systemd generated files in one directory.
3646 * Boot time performance measurements (as displayed by
3647 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3648 performance information if that's available to determine how
3649 much time BIOS and boot loader initialization required. With
3650 a sufficiently new BIOS you hence no longer need to boot
3651 with Gummiboot to get access to such information.
3653 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3654 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3655 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3656 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3657 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3658 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3659 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3661 -- Berlin, 2013-10-02
3665 * The Restart= option for services now understands a new
3666 on-watchdog setting, which will restart the service
3667 automatically if the service stops sending out watchdog keep
3668 alive messages (as configured with WatchdogSec=).
3670 * The getty generator (which is responsible for bringing up a
3671 getty on configured serial consoles) will no longer only
3672 start a getty on the primary kernel console but on all
3673 others, too. This makes the order in which console= is
3674 specified on the kernel command line less important.
3676 * libsystemd-logind gained a new sd_session_get_vt() call to
3677 retrieve the VT number of a session.
3679 * If the option "tries=0" is set for an entry of /etc/crypttab
3680 its passphrase is queried indefinitely instead of any
3681 maximum number of tries.
3683 * If a service with a configure PID file terminates its PID
3684 file will now be removed automatically if it still exists
3685 afterwards. This should put an end to stale PID files.
3687 * systemd-run will now also take relative binary path names
3688 for execution and no longer insists on absolute paths.
3690 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3691 paths that are optionally prefixed with "-" to indicate that
3692 it should not be considered a failure if they do not exist.
3694 * journalctl -o (and similar commands) now understands a new
3695 output mode "short-precise", it is similar to "short" but
3696 shows timestamps with usec accuracy.
3698 * The option "discard" (as known from Debian) is now
3699 synonymous to "allow-discards" in /etc/crypttab. In fact,
3700 "discard" is preferred now (since it is easier to remember
3703 * Some licensing clean-ups were made, so that more code is now
3704 LGPL-2.1 licensed than before.
3706 * A minimal tool to save/restore the display backlight
3707 brightness across reboots has been added. It will store the
3708 backlight setting as late as possible at shutdown, and
3709 restore it as early as possible during reboot.
3711 * A logic to automatically discover and enable home and swap
3712 partitions on GPT disks has been added. With this in place
3713 /etc/fstab becomes optional for many setups as systemd can
3714 discover certain partitions located on the root disk
3715 automatically. Home partitions are recognized under their
3716 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3717 partitions are recognized under their GPT type ID
3718 0657fd6da4ab43c484e50933c84b4f4f.
3720 * systemd will no longer pass any environment from the kernel
3721 or initrd to system services. If you want to set an
3722 environment for all services, do so via the kernel command
3723 line systemd.setenv= assignment.
3725 * The systemd-sysctl tool no longer natively reads the file
3726 /etc/sysctl.conf. If desired, the file should be symlinked
3727 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3728 legacy support by a symlink rather than built-in code, it
3729 also makes the otherwise hidden order of application of the
3730 different files visible. (Note that this partly reverts to a
3731 pre-198 application order of sysctl knobs!)
3733 * The "systemctl set-log-level" and "systemctl dump" commands
3734 have been moved to systemd-analyze.
3736 * systemd-run learned the new --remain-after-exit switch,
3737 which causes the scope unit not to be cleaned up
3738 automatically after the process terminated.
3740 * tmpfiles learned a new --exclude-prefix= switch to exclude
3741 certain paths from operation.
3743 * journald will now automatically flush all messages to disk
3744 as soon as a message at the log level CRIT, ALERT or EMERG
3747 Contributions from: Andrew Cook, Brandon Philips, Christian
3748 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3749 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3750 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3751 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3752 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3753 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3754 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3755 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3756 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3757 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3758 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3759 William Giokas, Zbigniew Jędrzejewski-Szmek
3761 -- Berlin, 2013-09-13
3765 * The documentation has been updated to cover the various new
3766 concepts introduced with 205.
3768 * Unit files now understand the new %v specifier which
3769 resolves to the kernel version string as returned by "uname
3772 * systemctl now supports filtering the unit list output by
3773 load state, active state and sub state, using the new
3776 * "systemctl status" will now show the results of the
3777 condition checks (like ConditionPathExists= and similar) of
3778 the last start attempts of the unit. They are also logged to
3781 * "journalctl -b" may now be used to look for boot output of a
3782 specific boot. Try "journalctl -b -1" for the previous boot,
3783 but the syntax is substantially more powerful.
3785 * "journalctl --show-cursor" has been added which prints the
3786 cursor string the last shown log line. This may then be used
3787 with the new "journalctl --after-cursor=" switch to continue
3788 browsing logs from that point on.
3790 * "journalctl --force" may now be used to force regeneration
3793 * Creation of "dead" device nodes has been moved from udev
3794 into kmod and tmpfiles. Previously, udev would read the kmod
3795 databases to pre-generate dead device nodes based on meta
3796 information contained in kernel modules, so that these would
3797 be auto-loaded on access rather then at boot. As this
3798 does not really have much to do with the exposing actual
3799 kernel devices to userspace this has always been slightly
3800 alien in the udev codebase. Following the new scheme kmod
3801 will now generate a runtime snippet for tmpfiles from the
3802 module meta information and it now is tmpfiles' job to the
3803 create the nodes. This also allows overriding access and
3804 other parameters for the nodes using the usual tmpfiles
3805 facilities. As side effect this allows us to remove the
3806 CAP_SYS_MKNOD capability bit from udevd entirely.
3808 * logind's device ACLs may now be applied to these "dead"
3809 devices nodes too, thus finally allowing managed access to
3810 devices such as /dev/snd/sequencer whithout loading the
3811 backing module right-away.
3813 * A new RPM macro has been added that may be used to apply
3814 tmpfiles configuration during package installation.
3816 * systemd-detect-virt and ConditionVirtualization= now can
3817 detect User-Mode-Linux machines (UML).
3819 * journald will now implicitly log the effective capabilities
3820 set of processes in the message metadata.
3822 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3824 * The initrd interface has been simplified (more specifically,
3825 support for passing performance data via environment
3826 variables and fsck results via files in /run has been
3827 removed). These features were non-essential, and are
3828 nowadays available in a much nicer way by having systemd in
3829 the initrd serialize its state and have the hosts systemd
3830 deserialize it again.
3832 * The udev "keymap" data files and tools to apply keyboard
3833 specific mappings of scan to key codes, and force-release
3834 scan code lists have been entirely replaced by a udev
3835 "keyboard" builtin and a hwdb data file.
3837 * systemd will now honour the kernel's "quiet" command line
3838 argument also during late shutdown, resulting in a
3839 completely silent shutdown when used.
3841 * There's now an option to control the SO_REUSEPORT socket
3842 option in .socket units.
3844 * Instance units will now automatically get a per-template
3845 subslice of system.slice unless something else is explicitly
3846 configured. For example, instances of sshd@.service will now
3847 implicitly be placed in system-sshd.slice rather than
3848 system.slice as before.
3850 * Test coverage support may now be enabled at build time.
3852 Contributions from: Dave Reisner, Frederic Crozat, Harald
3853 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3854 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3855 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3856 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3857 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3858 Giokas, Zbigniew Jędrzejewski-Szmek
3860 -- Berlin, 2013-07-23
3864 * Two new unit types have been introduced:
3866 Scope units are very similar to service units, however, are
3867 created out of pre-existing processes -- instead of PID 1
3868 forking off the processes. By using scope units it is
3869 possible for system services and applications to group their
3870 own child processes (worker processes) in a powerful way
3871 which then maybe used to organize them, or kill them
3872 together, or apply resource limits on them.
3874 Slice units may be used to partition system resources in an
3875 hierarchical fashion and then assign other units to them. By
3876 default there are now three slices: system.slice (for all
3877 system services), user.slice (for all user sessions),
3878 machine.slice (for VMs and containers).
3880 Slices and scopes have been introduced primarily in
3881 context of the work to move cgroup handling to a
3882 single-writer scheme, where only PID 1
3883 creates/removes/manages cgroups.
3885 * There's a new concept of "transient" units. In contrast to
3886 normal units these units are created via an API at runtime,
3887 not from configuration from disk. More specifically this
3888 means it is now possible to run arbitrary programs as
3889 independent services, with all execution parameters passed
3890 in via bus APIs rather than read from disk. Transient units
3891 make systemd substantially more dynamic then it ever was,
3892 and useful as a general batch manager.
3894 * logind has been updated to make use of scope and slice units
3895 for managing user sessions. As a user logs in he will get
3896 his own private slice unit, to which all sessions are added
3897 as scope units. We also added support for automatically
3898 adding an instance of user@.service for the user into the
3899 slice. Effectively logind will no longer create cgroup
3900 hierarchies on its own now, it will defer entirely to PID 1
3901 for this by means of scope, service and slice units. Since
3902 user sessions this way become entities managed by PID 1
3903 the output of "systemctl" is now a lot more comprehensive.
3905 * A new mini-daemon "systemd-machined" has been added which
3906 may be used by virtualization managers to register local
3907 VMs/containers. nspawn has been updated accordingly, and
3908 libvirt will be updated shortly. machined will collect a bit
3909 of meta information about the VMs/containers, and assign
3910 them their own scope unit (see above). The collected
3911 meta-data is then made available via the "machinectl" tool,
3912 and exposed in "ps" and similar tools. machined/machinectl
3913 is compile-time optional.
3915 * As discussed earlier, the low-level cgroup configuration
3916 options ControlGroup=, ControlGroupModify=,
3917 ControlGroupPersistent=, ControlGroupAttribute= have been
3918 removed. Please use high-level attribute settings instead as
3919 well as slice units.
3921 * A new bus call SetUnitProperties() has been added to alter
3922 various runtime parameters of a unit. This is primarily
3923 useful to alter cgroup parameters dynamically in a nice way,
3924 but will be extended later on to make more properties
3925 modifiable at runtime. systemctl gained a new set-properties
3926 command that wraps this call.
3928 * A new tool "systemd-run" has been added which can be used to
3929 run arbitrary command lines as transient services or scopes,
3930 while configuring a number of settings via the command
3931 line. This tool is currently very basic, however already
3932 very useful. We plan to extend this tool to even allow
3933 queuing of execution jobs with time triggers from the
3934 command line, similar in fashion to "at".
3936 * nspawn will now inform the user explicitly that kernels with
3937 audit enabled break containers, and suggest the user to turn
3940 * Support for detecting the IMA and AppArmor security
3941 frameworks with ConditionSecurity= has been added.
3943 * journalctl gained a new "-k" switch for showing only kernel
3944 messages, mimicking dmesg output; in addition to "--user"
3945 and "--system" switches for showing only user's own logs
3948 * systemd-delta can now show information about drop-in
3949 snippets extending unit files.
3951 * libsystemd-bus has been substantially updated but is still
3952 not available as public API.
3954 * systemd will now look for the "debug" argument on the kernel
3955 command line and enable debug logging, similar to what
3956 "systemd.log_level=debug" already did before.
3958 * "systemctl set-default", "systemctl get-default" has been
3959 added to configure the default.target symlink, which
3960 controls what to boot into by default.
3962 * "systemctl set-log-level" has been added as a convenient
3963 way to raise and lower systemd logging threshold.
3965 * "systemd-analyze plot" will now show the time the various
3966 generators needed for execution, as well as information
3967 about the unit file loading.
3969 * libsystemd-journal gained a new sd_journal_open_files() call
3970 for opening specific journal files. journactl also gained a
3971 new switch to expose this new functionality. Previously we
3972 only supported opening all files from a directory, or all
3973 files from the system, as opening individual files only is
3974 racy due to journal file rotation.
3976 * systemd gained the new DefaultEnvironment= setting in
3977 /etc/systemd/system.conf to set environment variables for
3980 * If a privileged process logs a journal message with the
3981 OBJECT_PID= field set, then journald will automatically
3982 augment this with additional OBJECT_UID=, OBJECT_GID=,
3983 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3984 system services want to log events about specific client
3985 processes. journactl/systemctl has been updated to make use
3986 of this information if all log messages regarding a specific
3989 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3990 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3991 Reisner, David Coppa, David King, David Strauss, Eelco
3992 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3993 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3994 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3995 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3996 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3997 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3998 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3999 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4000 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4001 Łukasz Stelmach, 장동준
4005 * The Python bindings gained some minimal support for the APIs
4006 exposed by libsystemd-logind.
4008 * ConditionSecurity= gained support for detecting SMACK. Since
4009 this condition already supports SELinux and AppArmor we only
4010 miss IMA for this. Patches welcome!
4012 Contributions from: Karol Lewandowski, Lennart Poettering,
4013 Zbigniew Jędrzejewski-Szmek
4017 * systemd-nspawn will now create /etc/resolv.conf if
4018 necessary, before bind-mounting the host's file onto it.
4020 * systemd-nspawn will now store meta information about a
4021 container on the container's cgroup as extended attribute
4022 fields, including the root directory.
4024 * The cgroup hierarchy has been reworked in many ways. All
4025 objects any of the components systemd creates in the cgroup
4026 tree are now suffixed. More specifically, user sessions are
4027 now placed in cgroups suffixed with ".session", users in
4028 cgroups suffixed with ".user", and nspawn containers in
4029 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4030 names are now escaped in a simple scheme to avoid collision
4031 of userspace object names with kernel filenames. This work
4032 is preparation for making these objects relocatable in the
4033 cgroup tree, in order to allow easy resource partitioning of
4034 these objects without causing naming conflicts.
4036 * systemctl list-dependencies gained the new switches
4037 --plain, --reverse, --after and --before.
4039 * systemd-inhibit now shows the process name of processes that
4040 have taken an inhibitor lock.
4042 * nss-myhostname will now also resolve "localhost"
4043 implicitly. This makes /etc/hosts an optional file and
4044 nicely handles that on IPv6 ::1 maps to both "localhost" and
4047 * libsystemd-logind.so gained a new call
4048 sd_get_machine_names() to enumerate running containers and
4049 VMs (currently only supported by very new libvirt and
4050 nspawn). sd_login_monitor can now be used to watch
4051 VMs/containers coming and going.
4053 * .include is not allowed recursively anymore, and only in
4054 unit files. Usually it is better to use drop-in snippets in
4055 .d/*.conf anyway, as introduced with systemd 198.
4057 * systemd-analyze gained a new "critical-chain" command that
4058 determines the slowest chain of units run during system
4059 boot-up. It is very useful for tracking down where
4060 optimizing boot time is the most beneficial.
4062 * systemd will no longer allow manipulating service paths in
4063 the name=systemd:/system cgroup tree using ControlGroup= in
4064 units. (But is still fine with it in all other dirs.)
4066 * There's a new systemd-nspawn@.service service file that may
4067 be used to easily run nspawn containers as system
4068 services. With the container's root directory in
4069 /var/lib/container/foobar it is now sufficient to run
4070 "systemctl start systemd-nspawn@foobar.service" to boot it.
4072 * systemd-cgls gained a new parameter "--machine" to list only
4073 the processes within a certain container.
4075 * ConditionSecurity= now can check for "apparmor". We still
4076 are lacking checks for SMACK and IMA for this condition
4077 check though. Patches welcome!
4079 * A new configuration file /etc/systemd/sleep.conf has been
4080 added that may be used to configure which kernel operation
4081 systemd is supposed to execute when "suspend", "hibernate"
4082 or "hybrid-sleep" is requested. This makes the new kernel
4083 "freeze" state accessible to the user.
4085 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4086 the passed argument if applicable.
4088 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4089 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4090 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4091 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4092 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4093 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4094 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4099 * The output of 'systemctl list-jobs' got some polishing. The
4100 '--type=' argument may now be passed more than once. A new
4101 command 'systemctl list-sockets' has been added which shows
4102 a list of kernel sockets systemd is listening on with the
4103 socket units they belong to, plus the units these socket
4106 * The experimental libsystemd-bus library got substantial
4107 updates to work in conjunction with the (also experimental)
4108 kdbus kernel project. It works well enough to exchange
4109 messages with some sophistication. Note that kdbus is not
4110 ready yet, and the library is mostly an elaborate test case
4111 for now, and not installable.
4113 * systemd gained a new unit 'systemd-static-nodes.service'
4114 that generates static device nodes earlier during boot, and
4115 can run in conjunction with udev.
4117 * libsystemd-login gained a new call sd_pid_get_user_unit()
4118 to retrieve the user systemd unit a process is running
4119 in. This is useful for systems where systemd is used as
4122 * systemd-nspawn now places all containers in the new /machine
4123 top-level cgroup directory in the name=systemd
4124 hierarchy. libvirt will soon do the same, so that we get a
4125 uniform separation of /system, /user and /machine for system
4126 services, user processes and containers/virtual
4127 machines. This new cgroup hierarchy is also useful to stick
4128 stable names to specific container instances, which can be
4129 recognized later this way (this name may be controlled
4130 via systemd-nspawn's new -M switch). libsystemd-login also
4131 gained a new call sd_pid_get_machine_name() to retrieve the
4132 name of the container/VM a specific process belongs to.
4134 * bootchart can now store its data in the journal.
4136 * libsystemd-journal gained a new call
4137 sd_journal_add_conjunction() for AND expressions to the
4138 matching logic. This can be used to express more complex
4139 logical expressions.
4141 * journactl can now take multiple --unit= and --user-unit=
4144 * The cryptsetup logic now understands the "luks.key=" kernel
4145 command line switch for specifying a file to read the
4146 decryption key from. Also, if a configured key file is not
4147 found the tool will now automatically fall back to prompting
4150 * Python systemd.journal module was updated to wrap recently
4151 added functions from libsystemd-journal. The interface was
4152 changed to bring the low level interface in s.j._Reader
4153 closer to the C API, and the high level interface in
4154 s.j.Reader was updated to wrap and convert all data about
4157 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4158 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4159 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4160 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4161 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4162 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4166 * journalctl --update-catalog now understands a new --root=
4167 option to operate on catalogs found in a different root
4170 * During shutdown after systemd has terminated all running
4171 services a final killing loop kills all remaining left-over
4172 processes. We will now print the name of these processes
4173 when we send SIGKILL to them, since this usually indicates a
4176 * If /etc/crypttab refers to password files stored on
4177 configured mount points automatic dependencies will now be
4178 generated to ensure the specific mount is established first
4179 before the key file is attempted to be read.
4181 * 'systemctl status' will now show information about the
4182 network sockets a socket unit is listening on.
4184 * 'systemctl status' will also shown information about any
4185 drop-in configuration file for units. (Drop-In configuration
4186 files in this context are files such as
4187 /etc/systemd/systemd/foobar.service.d/*.conf)
4189 * systemd-cgtop now optionally shows summed up CPU times of
4190 cgroups. Press '%' while running cgtop to switch between
4191 percentage and absolute mode. This is useful to determine
4192 which cgroups use up the most CPU time over the entire
4193 runtime of the system. systemd-cgtop has also been updated
4194 to be 'pipeable' for processing with further shell tools.
4196 * 'hostnamectl set-hostname' will now allow setting of FQDN
4199 * The formatting and parsing of time span values has been
4200 changed. The parser now understands fractional expressions
4201 such as "5.5h". The formatter will now output fractional
4202 expressions for all time spans under 1min, i.e. "5.123456s"
4203 rather than "5s 123ms 456us". For time spans under 1s
4204 millisecond values are shown, for those under 1ms
4205 microsecond values are shown. This should greatly improve
4206 all time-related output of systemd.
4208 * libsystemd-login and libsystemd-journal gained new
4209 functions for querying the poll() events mask and poll()
4210 timeout value for integration into arbitrary event
4213 * localectl gained the ability to list available X11 keymaps
4214 (models, layouts, variants, options).
4216 * 'systemd-analyze dot' gained the ability to filter for
4217 specific units via shell-style globs, to create smaller,
4218 more useful graphs. I.e. it is now possible to create simple
4219 graphs of all the dependencies between only target units, or
4220 of all units that Avahi has dependencies with.
4222 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4223 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4224 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4225 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4226 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4227 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4228 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4232 * The boot-time readahead implementation for rotating media
4233 will now read the read-ahead data in multiple passes which
4234 consist of all read requests made in equidistant time
4235 intervals. This means instead of strictly reading read-ahead
4236 data in its physical order on disk we now try to find a
4237 middle ground between physical and access time order.
4239 * /etc/os-release files gained a new BUILD_ID= field for usage
4240 on operating systems that provide continuous builds of OS
4243 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4244 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4245 William Douglas, Zbigniew Jędrzejewski-Szmek
4249 * systemd-python gained an API exposing libsystemd-daemon.
4251 * The SMACK setup logic gained support for uploading CIPSO
4254 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4255 ReadOnlyDirectories= and InaccessibleDirectories= has
4256 changed. The private /tmp and /var/tmp directories are now
4257 shared by all processes of a service (which means
4258 ExecStartPre= may now leave data in /tmp that ExecStart= of
4259 the same service can still access). When a service is
4260 stopped its temporary directories are immediately deleted
4261 (normal clean-up with tmpfiles is still done in addition to
4264 * By default, systemd will now set a couple of sysctl
4265 variables in the kernel: the safe sysrq options are turned
4266 on, IP route verification is turned on, and source routing
4267 disabled. The recently added hardlink and softlink
4268 protection of the kernel is turned on. These settings should
4269 be reasonably safe, and good defaults for all new systems.
4271 * The predictable network naming logic may now be turned off
4272 with a new kernel command line switch: net.ifnames=0.
4274 * A new libsystemd-bus module has been added that implements a
4275 pretty complete D-Bus client library. For details see:
4277 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4279 * journald will now explicitly flush the journal files to disk
4280 at the latest 5min after each write. The file will then also
4281 be marked offline until the next write. This should increase
4282 reliability in case of a crash. The synchronization delay
4283 can be configured via SyncIntervalSec= in journald.conf.
4285 * There's a new remote-fs-setup.target unit that can be used
4286 to pull in specific services when at least one remote file
4287 system is to be mounted.
4289 * There are new targets timers.target and paths.target as
4290 canonical targets to pull user timer and path units in
4291 from. This complements sockets.target with a similar
4292 purpose for socket units.
4294 * libudev gained a new call udev_device_set_attribute_value()
4295 to set sysfs attributes of a device.
4297 * The udev daemon now sets the default number of worker
4298 processes executed in parallel based on the number of available
4299 CPUs instead of the amount of available RAM. This is supposed
4300 to provide a more reliable default and limit a too aggressive
4301 paralellism for setups with 1000s of devices connected.
4303 Contributions from: Auke Kok, Colin Walters, Cristian
4304 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4305 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4306 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4307 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4308 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4309 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4310 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4311 Zbigniew Jędrzejewski-Szmek
4315 * Configuration of unit files may now be extended via drop-in
4316 files without having to edit/override the unit files
4317 themselves. More specifically, if the administrator wants to
4318 change one value for a service file foobar.service he can
4319 now do so by dropping in a configuration snippet into
4320 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4321 will load all these snippets and apply them on top of the
4322 main unit configuration file, possibly extending or
4323 overriding its settings. Using these drop-in snippets is
4324 generally nicer than the two earlier options for changing
4325 unit files locally: copying the files from
4326 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4327 them there; or creating a new file in /etc/systemd/system/
4328 that incorporates the original one via ".include". Drop-in
4329 snippets into these .d/ directories can be placed in any
4330 directory systemd looks for units in, and the usual
4331 overriding semantics between /usr/lib, /etc and /run apply
4334 * Most unit file settings which take lists of items can now be
4335 reset by assigning the empty string to them. For example,
4336 normally, settings such as Environment=FOO=BAR append a new
4337 environment variable assignment to the environment block,
4338 each time they are used. By assigning Environment= the empty
4339 string the environment block can be reset to empty. This is
4340 particularly useful with the .d/*.conf drop-in snippets
4341 mentioned above, since this adds the ability to reset list
4342 settings from vendor unit files via these drop-ins.
4344 * systemctl gained a new "list-dependencies" command for
4345 listing the dependencies of a unit recursively.
4347 * Inhibitors are now honored and listed by "systemctl
4348 suspend", "systemctl poweroff" (and similar) too, not only
4349 GNOME. These commands will also list active sessions by
4352 * Resource limits (as exposed by the various control group
4353 controllers) can now be controlled dynamically at runtime
4354 for all units. More specifically, you can now use a command
4355 like "systemctl set-cgroup-attr foobar.service cpu.shares
4356 2000" to alter the CPU shares a specific service gets. These
4357 settings are stored persistently on disk, and thus allow the
4358 administrator to easily adjust the resource usage of
4359 services with a few simple commands. This dynamic resource
4360 management logic is also available to other programs via the
4361 bus. Almost any kernel cgroup attribute and controller is
4364 * systemd-vconsole-setup will now copy all font settings to
4365 all allocated VTs, where it previously applied them only to
4368 * libsystemd-login gained the new sd_session_get_tty() API
4371 * This release drops support for a few legacy or
4372 distribution-specific LSB facility names when parsing init
4373 scripts: $x-display-manager, $mail-transfer-agent,
4374 $mail-transport-agent, $mail-transfer-agent, $smtp,
4375 $null. Also, the mail-transfer-agent.target unit backing
4376 this has been removed. Distributions which want to retain
4377 compatibility with this should carry the burden for
4378 supporting this themselves and patch support for these back
4379 in, if they really need to. Also, the facilities $syslog and
4380 $local_fs are now ignored, since systemd does not support
4381 early-boot LSB init scripts anymore, and these facilities
4382 are implied anyway for normal services. syslog.target has
4385 * There are new bus calls on PID1's Manager object for
4386 cancelling jobs, and removing snapshot units. Previously,
4387 both calls were only available on the Job and Snapshot
4390 * systemd-journal-gatewayd gained SSL support.
4392 * The various "environment" files, such as /etc/locale.conf
4393 now support continuation lines with a backslash ("\") as
4394 last character in the line, similarly in style (but different)
4395 to how this is supported in shells.
4397 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4398 now implicitly appended to every log entry logged. systemctl
4399 has been updated to filter by this field when operating on a
4400 user systemd instance.
4402 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4403 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4404 the container. This makes it easier to boot unmodified
4405 Fedora systems in a container, which however still requires
4406 audit=0 to be passed on the kernel command line. Auditing in
4407 kernel and userspace is unfortunately still too broken in
4408 context of containers, hence we recommend compiling it out
4409 of the kernel or using audit=0. Hopefully this will be fixed
4410 one day for good in the kernel.
4412 * nspawn gained the new --bind= and --bind-ro= parameters to
4413 bind mount specific directories from the host into the
4416 * nspawn will now mount its own devpts file system instance
4417 into the container, in order not to leak pty devices from
4418 the host into the container.
4420 * systemd will now read the firmware boot time performance
4421 information from the EFI variables, if the used boot loader
4422 supports this, and takes it into account for boot performance
4423 analysis via "systemd-analyze". This is currently supported
4424 only in conjunction with Gummiboot, but could be supported
4425 by other boot loaders too. For details see:
4427 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4429 * A new generator has been added that automatically mounts the
4430 EFI System Partition (ESP) to /boot, if that directory
4431 exists, is empty, and no other file system has been
4432 configured to be mounted there.
4434 * logind will now send out PrepareForSleep(false) out
4435 unconditionally, after coming back from suspend. This may be
4436 used by applications as asynchronous notification for
4437 system resume events.
4439 * "systemctl unlock-sessions" has been added, that allows
4440 unlocking the screens of all user sessions at once, similar
4441 to how "systemctl lock-sessions" already locked all users
4442 sessions. This is backed by a new D-Bus call UnlockSessions().
4444 * "loginctl seat-status" will now show the master device of a
4445 seat. (i.e. the device of a seat that needs to be around for
4446 the seat to be considered available, usually the graphics
4449 * tmpfiles gained a new "X" line type, that allows
4450 configuration of files and directories (with wildcards) that
4451 shall be excluded from automatic cleanup ("aging").
4453 * udev default rules set the device node permissions now only
4454 at "add" events, and do not change them any longer with a
4455 later "change" event.
4457 * The log messages for lid events and power/sleep keypresses
4458 now carry a message ID.
4460 * We now have a substantially larger unit test suite, but this
4461 continues to be work in progress.
4463 * udevadm hwdb gained a new --root= parameter to change the
4464 root directory to operate relative to.
4466 * logind will now issue a background sync() request to the kernel
4467 early at shutdown, so that dirty buffers are flushed to disk early
4468 instead of at the last moment, in order to optimize shutdown
4471 * A new bootctl tool has been added that is an interface for
4472 certain boot loader operations. This is currently a preview
4473 and is likely to be extended into a small mechanism daemon
4474 like timedated, localed, hostnamed, and can be used by
4475 graphical UIs to enumerate available boot options, and
4476 request boot into firmware operations.
4478 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4479 the rest of the package. It also has been updated to work
4480 correctly in initrds.
4482 * Policykit previously has been runtime optional, and is now
4483 also compile time optional via a configure switch.
4485 * systemd-analyze has been reimplemented in C. Also "systemctl
4486 dot" has moved into systemd-analyze.
4488 * "systemctl status" with no further parameters will now print
4489 the status of all active or failed units.
4491 * Operations such as "systemctl start" can now be executed
4492 with a new mode "--irreversible" which may be used to queue
4493 operations that cannot accidentally be reversed by a later
4494 job queuing. This is by default used to make shutdown
4495 requests more robust.
4497 * The Python API of systemd now gained a new module for
4498 reading journal files.
4500 * A new tool kernel-install has been added that can install
4501 kernel images according to the Boot Loader Specification:
4503 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4505 * Boot time console output has been improved to provide
4506 animated boot time output for hanging jobs.
4508 * A new tool systemd-activate has been added which can be used
4509 to test socket activation with, directly from the command
4510 line. This should make it much easier to test and debug
4511 socket activation in daemons.
4513 * journalctl gained a new "--reverse" (or -r) option to show
4514 journal output in reverse order (i.e. newest line first).
4516 * journalctl gained a new "--pager-end" (or -e) option to jump
4517 to immediately jump to the end of the journal in the
4518 pager. This is only supported in conjunction with "less".
4520 * journalctl gained a new "--user-unit=" option, that works
4521 similarly to "--unit=" but filters for user units rather than
4524 * A number of unit files to ease adoption of systemd in
4525 initrds has been added. This moves some minimal logic from
4526 the various initrd implementations into systemd proper.
4528 * The journal files are now owned by a new group
4529 "systemd-journal", which exists specifically to allow access
4530 to the journal, and nothing else. Previously, we used the
4531 "adm" group for that, which however possibly covers more
4532 than just journal/log file access. This new group is now
4533 already used by systemd-journal-gatewayd to ensure this
4534 daemon gets access to the journal files and as little else
4535 as possible. Note that "make install" will also set FS ACLs
4536 up for /var/log/journal to give "adm" and "wheel" read
4537 access to it, in addition to "systemd-journal" which owns
4538 the journal files. We recommend that packaging scripts also
4539 add read access to "adm" + "wheel" to /var/log/journal, and
4540 all existing/future journal files. To normal users and
4541 administrators little changes, however packagers need to
4542 ensure to create the "systemd-journal" system group at
4543 package installation time.
4545 * The systemd-journal-gatewayd now runs as unprivileged user
4546 systemd-journal-gateway:systemd-journal-gateway. Packaging
4547 scripts need to create these system user/group at
4550 * timedated now exposes a new boolean property CanNTP that
4551 indicates whether a local NTP service is available or not.
4553 * systemd-detect-virt will now also detect xen PVs
4555 * The pstore file system is now mounted by default, if it is
4558 * In addition to the SELinux and IMA policies we will now also
4559 load SMACK policies at early boot.
4561 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4562 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4563 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4564 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4565 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4566 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4567 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4568 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4569 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4570 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4571 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4572 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4573 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4574 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4578 * Timer units now support calendar time events in addition to
4579 monotonic time events. That means you can now trigger a unit
4580 based on a calendar time specification such as "Thu,Fri
4581 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4582 or fifth day of any month of the year 2013, given that it is
4583 a thursday or friday. This brings timer event support
4584 considerably closer to cron's capabilities. For details on
4585 the supported calendar time specification language see
4588 * udev now supports a number of different naming policies for
4589 network interfaces for predictable names, and a combination
4590 of these policies is now the default. Please see this wiki
4591 document for details:
4593 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4595 * Auke Kok's bootchart implementation has been added to the
4596 systemd tree. It is an optional component that can graph the
4597 boot in quite some detail. It is one of the best bootchart
4598 implementations around and minimal in its code and
4601 * nss-myhostname has been integrated into the systemd source
4602 tree. nss-myhostname guarantees that the local hostname
4603 always stays resolvable via NSS. It has been a weak
4604 requirement of systemd-hostnamed since a long time, and
4605 since its code is actually trivial we decided to just
4606 include it in systemd's source tree. It can be turned off
4607 with a configure switch.
4609 * The read-ahead logic is now capable of properly detecting
4610 whether a btrfs file system is on SSD or rotating media, in
4611 order to optimize the read-ahead scheme. Previously, it was
4612 only capable of detecting this on traditional file systems
4615 * In udev, additional device properties are now read from the
4616 IAB in addition to the OUI database. Also, Bluetooth company
4617 identities are attached to the devices as well.
4619 * In service files %U may be used as specifier that is
4620 replaced by the configured user name of the service.
4622 * nspawn may now be invoked without a controlling TTY. This
4623 makes it suitable for invocation as its own service. This
4624 may be used to set up a simple containerized server system
4625 using only core OS tools.
4627 * systemd and nspawn can now accept socket file descriptors
4628 when they are started for socket activation. This enables
4629 implementation of socket activated nspawn
4630 containers. i.e. think about autospawning an entire OS image
4631 when the first SSH or HTTP connection is received. We expect
4632 that similar functionality will also be added to libvirt-lxc
4635 * journalctl will now suppress ANSI color codes when
4636 presenting log data.
4638 * systemctl will no longer show control group information for
4639 a unit if a the control group is empty anyway.
4641 * logind can now automatically suspend/hibernate/shutdown the
4644 * /etc/machine-info and hostnamed now also expose the chassis
4645 type of the system. This can be used to determine whether
4646 the local system is a laptop, desktop, handset or
4647 tablet. This information may either be configured by the
4648 user/vendor or is automatically determined from ACPI and DMI
4649 information if possible.
4651 * A number of PolicyKit actions are now bound together with
4652 "imply" rules. This should simplify creating UIs because
4653 many actions will now authenticate similar ones as well.
4655 * Unit files learnt a new condition ConditionACPower= which
4656 may be used to conditionalize a unit depending on whether an
4657 AC power source is connected or not, of whether the system
4658 is running on battery power.
4660 * systemctl gained a new "is-failed" verb that may be used in
4661 shell scripts and suchlike to check whether a specific unit
4662 is in the "failed" state.
4664 * The EnvironmentFile= setting in unit files now supports file
4665 globbing, and can hence be used to easily read a number of
4666 environment files at once.
4668 * systemd will no longer detect and recognize specific
4669 distributions. All distribution-specific #ifdeffery has been
4670 removed, systemd is now fully generic and
4671 distribution-agnostic. Effectively, not too much is lost as
4672 a lot of the code is still accessible via explicit configure
4673 switches. However, support for some distribution specific
4674 legacy configuration file formats has been dropped. We
4675 recommend distributions to simply adopt the configuration
4676 files everybody else uses now and convert the old
4677 configuration from packaging scripts. Most distributions
4678 already did that. If that's not possible or desirable,
4679 distributions are welcome to forward port the specific
4680 pieces of code locally from the git history.
4682 * When logging a message about a unit systemd will now always
4683 log the unit name in the message meta data.
4685 * localectl will now also discover system locale data that is
4686 not stored in locale archives, but directly unpacked.
4688 * logind will no longer unconditionally use framebuffer
4689 devices as seat masters, i.e. as devices that are required
4690 to be existing before a seat is considered preset. Instead,
4691 it will now look for all devices that are tagged as
4692 "seat-master" in udev. By default, framebuffer devices will
4693 be marked as such, but depending on local systems, other
4694 devices might be marked as well. This may be used to
4695 integrate graphics cards using closed source drivers (such
4696 as NVidia ones) more nicely into logind. Note however, that
4697 we recommend using the open source NVidia drivers instead,
4698 and no udev rules for the closed-source drivers will be
4699 shipped from us upstream.
4701 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4702 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4703 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4704 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4705 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4706 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4707 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4708 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4709 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4710 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4711 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4712 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4717 * udev gained support for loading additional device properties
4718 from an indexed database that is keyed by vendor/product IDs
4719 and similar device identifiers. For the beginning this
4720 "hwdb" is populated with data from the well-known PCI and
4721 USB database, but also includes PNP, ACPI and OID data. In
4722 the longer run this indexed database shall grow into
4723 becoming the one central database for non-essential
4724 userspace device metadata. Previously, data from the PCI/USB
4725 database was only attached to select devices, since the
4726 lookup was a relatively expensive operation due to O(n) time
4727 complexity (with n being the number of entries in the
4728 database). Since this is now O(1), we decided to add in this
4729 data for all devices where this is available, by
4730 default. Note that the indexed database needs to be rebuilt
4731 when new data files are installed. To achieve this you need
4732 to update your packaging scripts to invoke "udevadm hwdb
4733 --update" after installation of hwdb data files. For
4734 RPM-based distributions we introduced the new
4735 %udev_hwdb_update macro for this purpose.
4737 * The Journal gained support for the "Message Catalog", an
4738 indexed database to link up additional information with
4739 journal entries. For further details please check:
4741 http://www.freedesktop.org/wiki/Software/systemd/catalog
4743 The indexed message catalog database also needs to be
4744 rebuilt after installation of message catalog files. Use
4745 "journalctl --update-catalog" for this. For RPM-based
4746 distributions we introduced the %journal_catalog_update
4747 macro for this purpose.
4749 * The Python Journal bindings gained support for the standard
4750 Python logging framework.
4752 * The Journal API gained new functions for checking whether
4753 the underlying file system of a journal file is capable of
4754 properly reporting file change notifications, or whether
4755 applications that want to reflect journal changes "live"
4756 need to recheck journal files continuously in appropriate
4759 * It is now possible to set the "age" field for tmpfiles
4760 entries to 0, indicating that files matching this entry
4761 shall always be removed when the directories are cleaned up.
4763 * coredumpctl gained a new "gdb" verb which invokes gdb
4764 right-away on the selected coredump.
4766 * There's now support for "hybrid sleep" on kernels that
4767 support this, in addition to "suspend" and "hibernate". Use
4768 "systemctl hybrid-sleep" to make use of this.
4770 * logind's HandleSuspendKey= setting (and related settings)
4771 now gained support for a new "lock" setting to simply
4772 request the screen lock on all local sessions, instead of
4773 actually executing a suspend or hibernation.
4775 * systemd will now mount the EFI variables file system by
4778 * Socket units now gained support for configuration of the
4779 SMACK security label.
4781 * timedatectl will now output the time of the last and next
4782 daylight saving change.
4784 * We dropped support for various legacy and distro-specific
4785 concepts, such as insserv, early-boot SysV services
4786 (i.e. those for non-standard runlevels such as 'b' or 'S')
4787 or ArchLinux /etc/rc.conf support. We recommend the
4788 distributions who still need support this to either continue
4789 to maintain the necessary patches downstream, or find a
4790 different solution. (Talk to us if you have questions!)
4792 * Various systemd components will now bypass PolicyKit checks
4793 for root and otherwise handle properly if PolicyKit is not
4794 found to be around. This should fix most issues for
4795 PolicyKit-less systems. Quite frankly this should have been
4796 this way since day one. It is absolutely our intention to
4797 make systemd work fine on PolicyKit-less systems, and we
4798 consider it a bug if something does not work as it should if
4799 PolicyKit is not around.
4801 * For embedded systems it is now possible to build udev and
4802 systemd without blkid and/or kmod support.
4804 * "systemctl switch-root" is now capable of switching root
4805 more than once. I.e. in addition to transitions from the
4806 initrd to the host OS it is now possible to transition to
4807 further OS images from the host. This is useful to implement
4808 offline updating tools.
4810 * Various other additions have been made to the RPM macros
4811 shipped with systemd. Use %udev_rules_update() after
4812 installing new udev rules files. %_udevhwdbdir,
4813 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4814 %_sysctldir are now available which resolve to the right
4815 directories for packages to place various data files in.
4817 * journalctl gained the new --full switch (in addition to
4818 --all, to disable ellipsation for long messages.
4820 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4821 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4822 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4823 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4824 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4825 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4826 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4827 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4828 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4832 * journalctl gained new --since= and --until= switches to
4833 filter by time. It also now supports nice filtering for
4834 units via --unit=/-u.
4836 * Type=oneshot services may use ExecReload= and do the
4839 * The journal daemon now supports time-based rotation and
4840 vacuuming, in addition to the usual disk-space based
4843 * The journal will now index the available field values for
4844 each field name. This enables clients to show pretty drop
4845 downs of available match values when filtering. The bash
4846 completion of journalctl has been updated
4847 accordingly. journalctl gained a new switch -F to list all
4848 values a certain field takes in the journal database.
4850 * More service events are now written as structured messages
4851 to the journal, and made recognizable via message IDs.
4853 * The timedated, localed and hostnamed mini-services which
4854 previously only provided support for changing time, locale
4855 and hostname settings from graphical DEs such as GNOME now
4856 also have a minimal (but very useful) text-based client
4857 utility each. This is probably the nicest way to changing
4858 these settings from the command line now, especially since
4859 it lists available options and is fully integrated with bash
4862 * There's now a new tool "systemd-coredumpctl" to list and
4863 extract coredumps from the journal.
4865 * We now install a README each in /var/log/ and
4866 /etc/rc.d/init.d explaining where the system logs and init
4867 scripts went. This hopefully should help folks who go to
4868 that dirs and look into the otherwise now empty void and
4869 scratch their heads.
4871 * When user-services are invoked (by systemd --user) the
4872 $MANAGERPID env var is set to the PID of systemd.
4874 * SIGRTMIN+24 when sent to a --user instance will now result
4875 in immediate termination of systemd.
4877 * gatewayd received numerous feature additions such as a
4878 "follow" mode, for live syncing and filtering.
4880 * browse.html now allows filtering and showing detailed
4881 information on specific entries. Keyboard navigation and
4882 mouse screen support has been added.
4884 * gatewayd/journalctl now supports HTML5/JSON
4885 Server-Sent-Events as output.
4887 * The SysV init script compatibility logic will now
4888 heuristically determine whether a script supports the
4889 "reload" verb, and only then make this available as
4892 * "systemctl status --follow" has been removed, use "journalctl
4895 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4896 have been removed since they are hardly useful to be
4899 * And I'd like to take the opportunity to specifically mention
4900 Zbigniew for his great contributions. Zbigniew, you rock!
4902 Contributions from: Andrew Eikum, Christian Hesse, Colin
4903 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4904 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4905 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4906 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4907 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4908 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4912 * If /etc/vconsole.conf is non-existent or empty we will no
4913 longer load any console font or key map at boot by
4914 default. Instead the kernel defaults will be left
4915 intact. This is definitely the right thing to do, as no
4916 configuration should mean no configuration, and hard-coding
4917 font names that are different on all archs is probably a bad
4918 idea. Also, the kernel default key map and font should be
4919 good enough for most cases anyway, and mostly identical to
4920 the userspace fonts/key maps we previously overloaded them
4921 with. If distributions want to continue to default to a
4922 non-kernel font or key map they should ship a default
4923 /etc/vconsole.conf with the appropriate contents.
4925 Contributions from: Colin Walters, Daniel J Walsh, Dave
4926 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4927 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4931 * journalctl gained a new --cursor= switch to show entries
4932 starting from the specified location in the journal.
4934 * We now enforce a size limit on journal entry fields exported
4935 with "-o json" in journalctl. Fields larger than 4K will be
4936 assigned null. This can be turned off with --all.
4938 * An (optional) journal gateway daemon is now available as
4939 "systemd-journal-gatewayd.service". This service provides
4940 access to the journal via HTTP and JSON. This functionality
4941 will be used to implement live log synchronization in both
4942 pull and push modes, but has various other users too, such
4943 as easy log access for debugging of embedded devices. Right
4944 now it is already useful to retrieve the journal via HTTP:
4946 # systemctl start systemd-journal-gatewayd.service
4947 # wget http://localhost:19531/entries
4949 This will download the journal contents in a
4950 /var/log/messages compatible format. The same as JSON:
4952 # curl -H"Accept: application/json" http://localhost:19531/entries
4954 This service is also accessible via a web browser where a
4955 single static HTML5 app is served that uses the JSON logic
4956 to enable the user to do some basic browsing of the
4957 journal. This will be extended later on. Here's an example
4958 screenshot of this app in its current state:
4960 http://0pointer.de/public/journal-gatewayd
4962 Contributions from: Kay Sievers, Lennart Poettering, Robert
4963 Milasan, Tom Gundersen
4967 * The bash completion logic is now available for journalctl
4970 * We do not mount the "cpuset" controller anymore together with
4971 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4972 started if no parameters are assigned to it. "cpuset" hence
4973 broke code that assumed it it could create "cpu" groups and
4976 * journalctl -f will now subscribe to terminal size changes,
4977 and line break accordingly.
4979 Contributions from: Dave Reisner, Kay Sievers, Lennart
4980 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4984 * nspawn will now create a symlink /etc/localtime in the
4985 container environment, copying the host's timezone
4986 setting. Previously this has been done via a bind mount, but
4987 since symlinks cannot be bind mounted this has now been
4988 changed to create/update the appropriate symlink.
4990 * journalctl -n's line number argument is now optional, and
4991 will default to 10 if omitted.
4993 * journald will now log the maximum size the journal files may
4994 take up on disk. This is particularly useful if the default
4995 built-in logic of determining this parameter from the file
4996 system size is used. Use "systemctl status
4997 systemd-journald.service" to see this information.
4999 * The multi-seat X wrapper tool has been stripped down. As X
5000 is now capable of enumerating graphics devices via udev in a
5001 seat-aware way the wrapper is not strictly necessary
5002 anymore. A stripped down temporary stop-gap is still shipped
5003 until the upstream display managers have been updated to
5004 fully support the new X logic. Expect this wrapper to be
5005 removed entirely in one of the next releases.
5007 * HandleSleepKey= in logind.conf has been split up into
5008 HandleSuspendKey= and HandleHibernateKey=. The old setting
5009 is not available anymore. X11 and the kernel are
5010 distinguishing between these keys and we should too. This
5011 also means the inhibition lock for these keys has been split
5014 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5015 Poettering, Lukas Nykryn, Václav Pavlín
5019 * Whenever a unit changes state we will now log this to the
5020 journal and show along the unit's own log output in
5023 * ConditionPathIsMountPoint= can now properly detect bind
5024 mount points too. (Previously, a bind mount of one file
5025 system to another place in the same file system could not be
5026 detected as mount, since they shared struct stat's st_dev
5029 * We will now mount the cgroup controllers cpu, cpuacct,
5030 cpuset and the controllers net_cls, net_prio together by
5033 * nspawn containers will now have a virtualized boot
5034 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5035 over with a randomized ID at container initialization). This
5036 has the effect of making "journalctl -b" do the right thing
5039 * The JSON output journal serialization has been updated not
5040 to generate "endless" list objects anymore, but rather one
5041 JSON object per line. This is more in line how most JSON
5042 parsers expect JSON objects. The new output mode
5043 "json-pretty" has been added to provide similar output, but
5044 neatly aligned for readability by humans.
5046 * We dropped all explicit sync() invocations in the shutdown
5047 code. The kernel does this implicitly anyway in the kernel
5048 reboot() syscall. halt(8)'s -n option is now a compatibility
5051 * We now support virtualized reboot() in containers, as
5052 supported by newer kernels. We will fall back to exit() if
5053 CAP_SYS_REBOOT is not available to the container. Also,
5054 nspawn makes use of this now and will actually reboot the
5055 container if the containerized OS asks for that.
5057 * journalctl will only show local log output by default
5058 now. Use --merge (-m) to show remote log output, too.
5060 * libsystemd-journal gained the new sd_journal_get_usage()
5061 call to determine the current disk usage of all journal
5062 files. This is exposed in the new "journalctl --disk-usage"
5065 * journald gained a new configuration setting SplitMode= in
5066 journald.conf which may be used to control how user journals
5067 are split off. See journald.conf(5) for details.
5069 * A new condition type ConditionFileNotEmpty= has been added.
5071 * tmpfiles' "w" lines now support file globbing, to write
5072 multiple files at once.
5074 * We added Python bindings for the journal submission
5075 APIs. More Python APIs for a number of selected APIs will
5076 likely follow. Note that we intend to add native bindings
5077 only for the Python language, as we consider it common
5078 enough to deserve bindings shipped within systemd. There are
5079 various projects outside of systemd that provide bindings
5080 for languages such as PHP or Lua.
5082 * Many conditions will now resolve specifiers such as %i. In
5083 addition, PathChanged= and related directives of .path units
5084 now support specifiers as well.
5086 * There's now a new RPM macro definition for the system preset
5089 * journald will now warn if it ca not forward a message to the
5090 syslog daemon because its socket is full.
5092 * timedated will no longer write or process /etc/timezone,
5093 except on Debian. As we do not support late mounted /usr
5094 anymore /etc/localtime always being a symlink is now safe,
5095 and hence the information in /etc/timezone is not necessary
5098 * logind will now always reserve one VT for a text getty (VT6
5099 by default). Previously if more than 6 X sessions where
5100 started they took up all the VTs with auto-spawned gettys,
5101 so that no text gettys were available anymore.
5103 * udev will now automatically inform the btrfs kernel logic
5104 about btrfs RAID components showing up. This should make
5105 simple hotplug based btrfs RAID assembly work.
5107 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5108 (but not for its children which will stay at the kernel
5109 default). This should allow setups with a lot more listening
5112 * systemd will now always pass the configured timezone to the
5113 kernel at boot. timedated will do the same when the timezone
5116 * logind's inhibition logic has been updated. By default,
5117 logind will now handle the lid switch, the power and sleep
5118 keys all the time, even in graphical sessions. If DEs want
5119 to handle these events on their own they should take the new
5120 handle-power-key, handle-sleep-key and handle-lid-switch
5121 inhibitors during their runtime. A simple way to achieve
5122 that is to invoke the DE wrapped in an invocation of:
5124 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5126 * Access to unit operations is now checked via SELinux taking
5127 the unit file label and client process label into account.
5129 * systemd will now notify the administrator in the journal
5130 when he over-mounts a non-empty directory.
5132 * There are new specifiers that are resolved in unit files,
5133 for the host name (%H), the machine ID (%m) and the boot ID
5136 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5137 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5138 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5139 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5140 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5141 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5142 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5146 * Support for reading structured kernel messages from
5147 /dev/kmsg has now been added and is enabled by default.
5149 * Support for reading kernel messages from /proc/kmsg has now
5150 been removed. If you want kernel messages in the journal
5151 make sure to run a recent kernel (>= 3.5) that supports
5152 reading structured messages from /dev/kmsg (see
5153 above). /proc/kmsg is now exclusive property of classic
5154 syslog daemons again.
5156 * The libudev API gained the new
5157 udev_device_new_from_device_id() call.
5159 * The logic for file system namespace (ReadOnlyDirectory=,
5160 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5161 require pivot_root() anymore. This means fewer temporary
5162 directories are created below /tmp for this feature.
5164 * nspawn containers will now see and receive all submounts
5165 made on the host OS below the root file system of the
5168 * Forward Secure Sealing is now supported for Journal files,
5169 which provide cryptographical sealing of journal files so
5170 that attackers cannot alter log history anymore without this
5171 being detectable. Lennart will soon post a blog story about
5172 this explaining it in more detail.
5174 * There are two new service settings RestartPreventExitStatus=
5175 and SuccessExitStatus= which allow configuration of exit
5176 status (exit code or signal) which will be excepted from the
5177 restart logic, resp. consider successful.
5179 * journalctl gained the new --verify switch that can be used
5180 to check the integrity of the structure of journal files and
5181 (if Forward Secure Sealing is enabled) the contents of
5184 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5185 and similar symlinks pre-created. This makes running shells
5186 as container init process a lot more fun.
5188 * The fstab support can now handle PARTUUID= and PARTLABEL=
5191 * A new ConditionHost= condition has been added to match
5192 against the hostname (with globs) and machine ID. This is
5193 useful for clusters where a single OS image is used to
5194 provision a large number of hosts which shall run slightly
5195 different sets of services.
5197 * Services which hit the restart limit will now be placed in a
5200 Contributions from: Bertram Poettering, Dave Reisner, Huang
5201 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5202 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5206 * When running in --user mode systemd will now become a
5207 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5208 tree a lot more organized.
5210 * A new PartOf= unit dependency type has been introduced that
5211 may be used to group services in a natural way.
5213 * "systemctl enable" may now be used to enable instances of
5216 * journalctl now prints error log levels in red, and
5217 warning/notice log levels in bright white. It also supports
5218 filtering by log level now.
5220 * cgtop gained a new -n switch (similar to top), to configure
5221 the maximum number of iterations to run for. It also gained
5222 -b, to run in batch mode (accepting no input).
5224 * The suffix ".service" may now be omitted on most systemctl
5225 command lines involving service unit names.
5227 * There's a new bus call in logind to lock all sessions, as
5228 well as a loginctl verb for it "lock-sessions".
5230 * libsystemd-logind.so gained a new call sd_journal_perror()
5231 that works similar to libc perror() but logs to the journal
5232 and encodes structured information about the error number.
5234 * /etc/crypttab entries now understand the new keyfile-size=
5237 * shutdown(8) now can send a (configurable) wall message when
5238 a shutdown is cancelled.
5240 * The mount propagation mode for the root file system will now
5241 default to "shared", which is useful to make containers work
5242 nicely out-of-the-box so that they receive new mounts from
5243 the host. This can be undone locally by running "mount
5244 --make-rprivate /" if needed.
5246 * The prefdm.service file has been removed. Distributions
5247 should maintain this unit downstream if they intend to keep
5248 it around. However, we recommend writing normal unit files
5249 for display managers instead.
5251 * Since systemd is a crucial part of the OS we will now
5252 default to a number of compiler switches that improve
5253 security (hardening) such as read-only relocations, stack
5254 protection, and suchlike.
5256 * The TimeoutSec= setting for services is now split into
5257 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5258 of individual time outs for the start and the stop phase of
5261 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5262 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5263 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5264 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5265 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5266 Gundersen, Zbigniew Jędrzejewski-Szmek
5270 * The journal and id128 C APIs are now fully documented as man
5273 * Extra safety checks have been added when transitioning from
5274 the initial RAM disk to the main system to avoid accidental
5277 * /etc/crypttab entries now understand the new keyfile-offset=
5280 * systemctl -t can now be used to filter by unit load state.
5282 * The journal C API gained the new sd_journal_wait() call to
5283 make writing synchronous journal clients easier.
5285 * journalctl gained the new -D switch to show journals from a
5288 * journalctl now displays a special marker between log
5289 messages of two different boots.
5291 * The journal is now explicitly flushed to /var via a service
5292 systemd-journal-flush.service, rather than implicitly simply
5293 by seeing /var/log/journal to be writable.
5295 * journalctl (and the journal C APIs) can now match for much
5296 more complex expressions, with alternatives and
5299 * When transitioning from the initial RAM disk to the main
5300 system we will now kill all processes in a killing spree to
5301 ensure no processes stay around by accident.
5303 * Three new specifiers may be used in unit files: %u, %h, %s
5304 resolve to the user name, user home directory resp. user
5305 shell. This is useful for running systemd user instances.
5307 * We now automatically rotate journal files if their data
5308 object hash table gets a fill level > 75%. We also size the
5309 hash table based on the configured maximum file size. This
5310 together should lower hash collisions drastically and thus
5311 speed things up a bit.
5313 * journalctl gained the new "--header" switch to introspect
5314 header data of journal files.
5316 * A new setting SystemCallFilters= has been added to services
5317 which may be used to apply blacklists or whitelists to
5318 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5320 * nspawn gained a new --link-journal= switch (and quicker: -j)
5321 to link the container journal with the host. This makes it
5322 very easy to centralize log viewing on the host for all
5323 guests while still keeping the journal files separated.
5325 * Many bugfixes and optimizations
5327 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5328 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5329 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5334 * Several tools now understand kernel command line arguments,
5335 which are only read when run in an initial RAM disk. They
5336 usually follow closely their normal counterparts, but are
5339 * There's a new tool to analyze the readahead files that are
5340 automatically generated at boot. Use:
5342 /usr/lib/systemd/systemd-readahead analyze /.readahead
5344 * We now provide an early debug shell on tty9 if this enabled. Use:
5346 systemctl enable debug-shell.service
5348 * All plymouth related units have been moved into the Plymouth
5349 package. Please make sure to upgrade your Plymouth version
5352 * systemd-tmpfiles now supports getting passed the basename of
5353 a configuration file only, in which case it will look for it
5354 in all appropriate directories automatically.
5356 * udevadm info now takes a /dev or /sys path as argument, and
5357 does the right thing. Example:
5359 udevadm info /dev/sda
5360 udevadm info /sys/class/block/sda
5362 * systemctl now prints a warning if a unit is stopped but a
5363 unit that might trigger it continues to run. Example: a
5364 service is stopped but the socket that activates it is left
5367 * "systemctl status" will now mention if the log output was
5368 shortened due to rotation since a service has been started.
5370 * The journal API now exposes functions to determine the
5371 "cutoff" times due to rotation.
5373 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5374 immediately flushing of runtime logs to /var if possible,
5375 resp. for triggering immediate rotation of the journal
5378 * It is now considered an error if a service is attempted to
5379 be stopped that is not loaded.
5381 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5383 * systemd-analyze now supports Python 3
5385 * tmpfiles now supports cleaning up directories via aging
5386 where the first level dirs are always kept around but
5387 directories beneath it automatically aged. This is enabled
5388 by prefixing the age field with '~'.
5390 * Seat objects now expose CanGraphical, CanTTY properties
5391 which is required to deal with very fast bootups where the
5392 display manager might be running before the graphics drivers
5393 completed initialization.
5395 * Seat objects now expose a State property.
5397 * We now include RPM macros for service enabling/disabling
5398 based on the preset logic. We recommend RPM based
5399 distributions to make use of these macros if possible. This
5400 makes it simpler to reuse RPM spec files across
5403 * We now make sure that the collected systemd unit name is
5404 always valid when services log to the journal via
5407 * There's a new man page kernel-command-line(7) detailing all
5408 command line options we understand.
5410 * The fstab generator may now be disabled at boot by passing
5411 fstab=0 on the kernel command line.
5413 * A new kernel command line option modules-load= is now understood
5414 to load a specific kernel module statically, early at boot.
5416 * Unit names specified on the systemctl command line are now
5417 automatically escaped as needed. Also, if file system or
5418 device paths are specified they are automatically turned
5419 into the appropriate mount or device unit names. Example:
5421 systemctl status /home
5422 systemctl status /dev/sda
5424 * The SysVConsole= configuration option has been removed from
5425 system.conf parsing.
5427 * The SysV search path is no longer exported on the D-Bus
5430 * The Names= option is been removed from unit file parsing.
5432 * There's a new man page bootup(7) detailing the boot process.
5434 * Every unit and every generator we ship with systemd now
5435 comes with full documentation. The self-explanatory boot is
5438 * A couple of services gained "systemd-" prefixes in their
5439 name if they wrap systemd code, rather than only external
5440 code. Among them fsck@.service which is now
5441 systemd-fsck@.service.
5443 * The HaveWatchdog property has been removed from the D-Bus
5446 * systemd.confirm_spawn= on the kernel command line should now
5449 * There's a new man page crypttab(5) which details all options
5450 we actually understand.
5452 * systemd-nspawn gained a new --capability= switch to pass
5453 additional capabilities to the container.
5455 * timedated will now read known NTP implementation unit names
5456 from /usr/lib/systemd/ntp-units.d/*.list,
5457 systemd-timedated-ntp.target has been removed.
5459 * journalctl gained a new switch "-b" that lists log data of
5460 the current boot only.
5462 * The notify socket is in the abstract namespace again, in
5463 order to support daemons which chroot() at start-up.
5465 * There is a new Storage= configuration option for journald
5466 which allows configuration of where log data should go. This
5467 also provides a way to disable journal logging entirely, so
5468 that data collected is only forwarded to the console, the
5469 kernel log buffer or another syslog implementation.
5471 * Many bugfixes and optimizations
5473 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5474 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5475 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5476 Shawn Landden, Tom Gundersen
5480 * "systemctl help <unit>" now shows the man page if one is
5483 * Several new man pages have been added.
5485 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5486 MaxLevelConsole= can now be specified in
5487 journald.conf. These options allow reducing the amount of
5488 data stored on disk or forwarded by the log level.
5490 * TimerSlackNSec= can now be specified in system.conf for
5491 PID1. This allows system-wide power savings.
5493 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5494 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5499 * logind is now capable of (optionally) handling power and
5500 sleep keys as well as the lid switch.
5502 * journalctl now understands the syntax "journalctl
5503 /usr/bin/avahi-daemon" to get all log output of a specific
5506 * CapabilityBoundingSet= in system.conf now also influences
5507 the capability bound set of usermode helpers of the kernel.
5509 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5510 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5511 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5512 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5516 * Note that we skipped 139 releases here in order to set the
5517 new version to something that is greater than both udev's
5518 and systemd's most recent version number.
5520 * udev: all udev sources are merged into the systemd source tree now.
5521 All future udev development will happen in the systemd tree. It
5522 is still fully supported to use the udev daemon and tools without
5523 systemd running, like in initramfs or other init systems. Building
5524 udev though, will require the *build* of the systemd tree, but
5525 udev can be properly *run* without systemd.
5527 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5528 should be used to create dead device nodes as workarounds for broken
5531 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5532 no longer supported. udev_monitor_new_from_netlink() needs to be
5533 used to subscribe to events.
5535 * udev: when udevd is started by systemd, processes which are left
5536 behind by forking them off of udev rules, are unconditionally cleaned
5537 up and killed now after the event handling has finished. Services or
5538 daemons must be started as systemd services. Services can be
5539 pulled-in by udev to get started, but they can no longer be directly
5540 forked by udev rules.
5542 * udev: the daemon binary is called systemd-udevd now and installed
5543 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5544 to adapt to that, create symlink, or rename the binary after building
5547 * libudev no longer provides these symbols:
5548 udev_monitor_from_socket()
5549 udev_queue_get_failed_list_entry()
5550 udev_get_{dev,sys,run}_path()
5551 The versions number was bumped and symbol versioning introduced.
5553 * systemd-loginctl and systemd-journalctl have been renamed
5554 to loginctl and journalctl to match systemctl.
5556 * The config files: /etc/systemd/systemd-logind.conf and
5557 /etc/systemd/systemd-journald.conf have been renamed to
5558 logind.conf and journald.conf. Package updates should rename
5559 the files to the new names on upgrade.
5561 * For almost all files the license is now LGPL2.1+, changed
5562 from the previous GPL2.0+. Exceptions are some minor stuff
5563 of udev (which will be changed to LGPL2.1 eventually, too),
5564 and the MIT licensed sd-daemon.[ch] library that is suitable
5565 to be used as drop-in files.
5567 * systemd and logind now handle system sleep states, in
5568 particular suspending and hibernating.
5570 * logind now implements a sleep/shutdown/idle inhibiting logic
5571 suitable for a variety of uses. Soonishly Lennart will blog
5572 about this in more detail.
5574 * var-run.mount and var-lock.mount are no longer provided
5575 (which prevously bind mounted these directories to their new
5576 places). Distributions which have not converted these
5577 directories to symlinks should consider stealing these files
5578 from git history and add them downstream.
5580 * We introduced the Documentation= field for units and added
5581 this to all our shipped units. This is useful to make it
5582 easier to explore the boot and the purpose of the various
5585 * All smaller setup units (such as
5586 systemd-vconsole-setup.service) now detect properly if they
5587 are run in a container and are skipped when
5588 appropriate. This guarantees an entirely noise-free boot in
5589 Linux container environments such as systemd-nspawn.
5591 * A framework for implementing offline system updates is now
5592 integrated, for details see:
5593 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5595 * A new service type Type=idle is available now which helps us
5596 avoiding ugly interleaving of getty output and boot status
5599 * There's now a system-wide CapabilityBoundingSet= option to
5600 globally reduce the set of capabilities for the
5601 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5602 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5603 even CAP_NET_ADMIN system-wide for secure systems.
5605 * There are now system-wide DefaultLimitXXX= options to
5606 globally change the defaults of the various resource limits
5607 for all units started by PID 1.
5609 * Harald Hoyer's systemd test suite has been integrated into
5610 systemd which allows easy testing of systemd builds in qemu
5611 and nspawn. (This is really awesome! Ask us for details!)
5613 * The fstab parser is now implemented as generator, not inside
5616 * systemctl will now warn you if .mount units generated from
5617 /etc/fstab are out of date due to changes in fstab that
5618 have not been read by systemd yet.
5620 * systemd is now suitable for usage in initrds. Dracut has
5621 already been updated to make use of this. With this in place
5622 initrds get a slight bit faster but primarily are much
5623 easier to introspect and debug since "systemctl status" in
5624 the host system can be used to introspect initrd services,
5625 and the journal from the initrd is kept around too.
5627 * systemd-delta has been added, a tool to explore differences
5628 between user/admin configuration and vendor defaults.
5630 * PrivateTmp= now affects both /tmp and /var/tmp.
5632 * Boot time status messages are now much prettier and feature
5633 proper english language. Booting up systemd has never been
5636 * Read-ahead pack files now include the inode number of all
5637 files to pre-cache. When the inode changes the pre-caching
5638 is not attempted. This should be nicer to deal with updated
5639 packages which might result in changes of read-ahead
5642 * We now temporaritly lower the kernel's read_ahead_kb variable
5643 when collecting read-ahead data to ensure the kernel's
5644 built-in read-ahead does not add noise to our measurements
5645 of necessary blocks to pre-cache.
5647 * There's now RequiresMountsFor= to add automatic dependencies
5648 for all mounts necessary for a specific file system path.
5650 * MountAuto= and SwapAuto= have been removed from
5651 system.conf. Mounting file systems at boot has to take place
5654 * nspawn now learned a new switch --uuid= to set the machine
5655 ID on the command line.
5657 * nspawn now learned the -b switch to automatically search
5660 * vt102 is now the default TERM for serial TTYs, upgraded from
5663 * systemd-logind now works on VT-less systems.
5665 * The build tree has been reorganized. The individual
5666 components now have directories of their own.
5668 * A new condition type ConditionPathIsReadWrite= is now available.
5670 * nspawn learned the new -C switch to create cgroups for the
5671 container in other hierarchies.
5673 * We now have support for hardware watchdogs, configurable in
5676 * The scheduled shutdown logic now has a public API.
5678 * We now mount /tmp as tmpfs by default, but this can be
5679 masked and /etc/fstab can override it.
5681 * Since udisks does not make use of /media anymore we are not
5682 mounting a tmpfs on it anymore.
5684 * journalctl gained a new --local switch to only interleave
5685 locally generated journal files.
5687 * We can now load the IMA policy at boot automatically.
5689 * The GTK tools have been split off into a systemd-ui.
5691 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5692 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5693 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5694 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5695 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5696 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5697 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5698 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5699 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5704 * This is mostly a bugfix release
5706 * Support optional initialization of the machine ID from the
5707 KVM or container configured UUID.
5709 * Support immediate reboots with "systemctl reboot -ff"
5711 * Show /etc/os-release data in systemd-analyze output
5713 * Many bugfixes for the journal, including endianness fixes and
5714 ensuring that disk space enforcement works
5716 * sd-login.h is C++ comptaible again
5718 * Extend the /etc/os-release format on request of the Debian
5721 * We now refuse non-UTF8 strings used in various configuration
5722 and unit files. This is done to ensure we do not pass invalid
5723 data over D-Bus or expose it elsewhere.
5725 * Register Mimo USB Screens as suitable for automatic seat
5728 * Read SELinux client context from journal clients in a race
5731 * Reorder configuration file lookup order. /etc now always
5732 overrides /run in order to allow the administrator to always
5733 and unconditionally override vendor-supplied or
5734 automatically generated data.
5736 * The various user visible bits of the journal now have man
5737 pages. We still lack man pages for the journal API calls
5740 * We now ship all man pages in HTML format again in the
5743 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5744 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5745 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5750 * This is mostly a bugfix release
5752 * systems lacking /etc/os-release are no longer supported.
5754 * Various functionality updates to libsystemd-login.so
5756 * Track class of PAM logins to distinguish greeters from
5759 Contributions from: Kay Sievers, Lennart Poettering, Michael
5764 * This is an important bugfix release for v41.
5766 * Building man pages is now optional which should be useful
5767 for those building systemd from git but unwilling to install
5770 * Watchdog support for supervising services is now usable. In
5771 a future release support for hardware watchdogs
5772 (i.e. /dev/watchdog) will be added building on this.
5774 * Service start rate limiting is now configurable and can be
5775 turned off per service. When a start rate limit is hit a
5776 reboot can automatically be triggered.
5778 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5780 Contributions from: Benjamin Franzke, Bill Nottingham,
5781 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5782 Schmidt, Michał Górny, Piotr Drąg
5786 * The systemd binary is installed /usr/lib/systemd/systemd now;
5787 An existing /sbin/init symlink needs to be adapted with the
5790 * The code that loads kernel modules has been ported to invoke
5791 libkmod directly, instead of modprobe. This means we do not
5792 support systems with module-init-tools anymore.
5794 * Watchdog support is now already useful, but still not
5797 * A new kernel command line option systemd.setenv= is
5798 understood to set system wide environment variables
5799 dynamically at boot.
5801 * We now limit the set of capabilities of systemd-journald.
5803 * We now set SIGPIPE to ignore by default, since it only is
5804 useful in shell pipelines, and has little use in general
5805 code. This can be disabled with IgnoreSIPIPE=no in unit
5808 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5809 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5814 * This is mostly a bugfix release
5816 * We now expose the reason why a service failed in the
5817 "Result" D-Bus property.
5819 * Rudimentary service watchdog support (will be completed over
5820 the next few releases.)
5822 * When systemd forks off in order execute some service we will
5823 now immediately changes its argv[0] to reflect which process
5824 it will execute. This is useful to minimize the time window
5825 with a generic argv[0], which makes bootcharts more useful
5827 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5828 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5829 Mike Kazantsev, Ray Strode
5833 * This is mostly a test release, but incorporates many
5836 * New systemd-cgtop tool to show control groups by their
5839 * Linking against libacl for ACLs is optional again. If
5840 disabled, support tracking device access for active logins
5841 goes becomes unavailable, and so does access to the user
5842 journals by the respective users.
5844 * If a group "adm" exists, journal files are automatically
5845 owned by them, thus allow members of this group full access
5846 to the system journal as well as all user journals.
5848 * The journal now stores the SELinux context of the logging
5849 client for all entries.
5851 * Add C++ inclusion guards to all public headers
5853 * New output mode "cat" in the journal to print only text
5854 messages, without any meta data like date or time.
5856 * Include tiny X server wrapper as a temporary stop-gap to
5857 teach XOrg udev display enumeration. This is used by display
5858 managers such as gdm, and will go away as soon as XOrg
5859 learned native udev hotplugging for display devices.
5861 * Add new systemd-cat tool for executing arbitrary programs
5862 with STDERR/STDOUT connected to the journal. Can also act as
5863 BSD logger replacement, and does so by default.
5865 * Optionally store all locally generated coredumps in the
5866 journal along with meta data.
5868 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5869 writing short strings to files (for usage for /sys), and for
5870 creating symlinks, character and block device nodes.
5872 * New unit file option ControlGroupPersistent= to make cgroups
5873 persistent, following the mechanisms outlined in
5874 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5876 * Support multiple local RTCs in a sane way
5878 * No longer monopolize IO when replaying readahead data on
5879 rotating disks, since we might starve non-file-system IO to
5880 death, since fanotify() will not see accesses done by blkid,
5883 * Do not show kernel threads in systemd-cgls anymore, unless
5884 requested with new -k switch.
5886 Contributions from: Dan Horák, Kay Sievers, Lennart
5887 Poettering, Michal Schmidt
5891 * This is mostly a test release, but incorporates many
5894 * The git repository moved to:
5895 git://anongit.freedesktop.org/systemd/systemd
5896 ssh://git.freedesktop.org/git/systemd/systemd
5898 * First release with the journal
5899 http://0pointer.de/blog/projects/the-journal.html
5901 * The journal replaces both systemd-kmsg-syslogd and
5902 systemd-stdout-bridge.
5904 * New sd_pid_get_unit() API call in libsystemd-logind
5906 * Many systemadm clean-ups
5908 * Introduce remote-fs-pre.target which is ordered before all
5909 remote mounts and may be used to start services before all
5912 * Added Mageia support
5914 * Add bash completion for systemd-loginctl
5916 * Actively monitor PID file creation for daemons which exit in
5917 the parent process before having finished writing the PID
5918 file in the daemon process. Daemons which do this need to be
5919 fixed (i.e. PID file creation must have finished before the
5920 parent exits), but we now react a bit more gracefully to them.
5922 * Add colourful boot output, mimicking the well-known output
5923 of existing distributions.
5925 * New option PassCredentials= for socket units, for
5926 compatibility with a recent kernel ABI breakage.
5928 * /etc/rc.local is now hooked in via a generator binary, and
5929 thus will no longer act as synchronization point during
5932 * systemctl list-unit-files now supports --root=.
5934 * systemd-tmpfiles now understands two new commands: z, Z for
5935 relabelling files according to the SELinux database. This is
5936 useful to apply SELinux labels to specific files in /sys,
5939 * Output of SysV services is now forwarded to both the console
5940 and the journal by default, not only just the console.
5942 * New man pages for all APIs from libsystemd-login.
5944 * The build tree got reorganized and a the build system is a
5945 lot more modular allowing embedded setups to specifically
5946 select the components of systemd they are interested in.
5948 * Support for Linux systems lacking the kernel VT subsystem is
5951 * configure's --with-rootdir= got renamed to
5952 --with-rootprefix= to follow the naming used by udev and
5955 * Unless specified otherwise we will now install to /usr instead
5956 of /usr/local by default.
5958 * Processes with '@' in argv[0][0] are now excluded from the
5959 final shut-down killing spree, following the logic explained
5961 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5963 * All processes remaining in a service cgroup when we enter
5964 the START or START_PRE states are now killed with
5965 SIGKILL. That means it is no longer possible to spawn
5966 background processes from ExecStart= lines (which was never
5967 supported anyway, and bad style).
5969 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5970 reloading of units together.
5972 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5973 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5974 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5975 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5976 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob