1 systemd System and Service Manager
3 CHANGES WITH 230 in spe:
5 * DNSSEC is now turned on by default in systemd-resolved (in
6 "allow-downgrade" mode), but may be turned off during compile time by
7 passing "--with-default-dnssec=no" to "configure" (and of course,
8 during runtime with DNSSEC= in resolved.conf). We recommend
9 downstreams to leave this on at least during development cycles and
10 report any issues with the DNSSEC logic upstream. We are very
11 interested in collecting feedback about the DNSSEC validator and its
12 limitations in the wild. Note however, that DNSSEC support is
13 probably nothing downstreams should turn on in stable distros just
14 yet, as it might create incompatibilities with a few DNS servers and
15 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
16 automatically whenever we detect such incompatible setups, but there
17 might be systems we do not cover yet. Hence: please help us testing
18 the DNSSEC code, leave this on where you can, report back, but then
19 again don't consider turning this on in your stable, LTS or
20 production release just yet.
22 * systemd-resolve conveniently resolves DANE records with the --tlsa
23 option and OPENPGPKEY records with the --openpgp option.
25 * Testing tool /usr/lib/systemd/systemd-activate is renamed to
26 systemd-socket-activate and installed into /usr/bin. It is now fully
29 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
30 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line
32 WARNING: it is not possible to use previous systemd versions with
33 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
34 is necessary to also update systemd in the initramfs if using the
35 unified hierarchy. Updated selinux policy is also required.
37 * LLDP support has been extended, and both passive (receive-only)
38 and active (sender) modes are supported. Passive mode
39 ("routers-only") is enabled by default in systemd-networkd.
40 Active LLDP mode is enabled by default for containers on the
42 "networkctl lldp" can be used to list information gathered.
44 * Headers for LLDP support (sd-lldp.h) are now public.
46 * The Unique Identifier sent in DHCP requests can be configured.
48 * systemd-journald now uses separate threads to flush changes to
49 disk when closing journal files.
51 * systemd-ask-password skips printing of the password to stdout
52 with --no-output which can be useful in scripts.
54 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
55 (devices tagged with ID_MAKER_TOOL) are now tagged with
56 "uaccess" and are available to logged in users.
58 * systemd-bootchart has been split out to a separate repository:
59 https://github.com/systemd/systemd-bootchart
61 * Compatibility libraries libsystemd-daemon.so, libsystemd-journal.so,
62 libsystemd-id128.so, and libsystemd-login.so which have been
63 deprecated since systemd-209 have been removed along along with the
64 corresponding pkg-config files. All symbols provided by the those
65 libraries are provided by libsystemd.so.
67 * Capabilities= setting has been removed (it is ignored for backwards
68 compatibility). AmbientCapabilities= and CapabilityBoundingSet=
69 should be used instead.
71 * systemd-bus-proxyd has been removed, as kdbus will not be merged
76 * The systemd-resolved DNS resolver service has gained a substantial
77 set of new features, most prominently it may now act as a DNSSEC
78 validating stub resolver. DNSSEC mode is currently turned off by
79 default, but is expected to be turned on by default in one of the
80 next releases. For now, we invite everybody to test the DNSSEC logic
81 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
82 service also gained a full set of D-Bus interfaces, including calls
83 to configure DNS and DNSSEC settings per link (for use by external
84 network management software). systemd-resolved and systemd-networkd
85 now distinguish between "search" and "routing" domains. The former
86 are used to qualify single-label names, the latter are used purely
87 for routing lookups within certain domains to specific links.
88 resolved now also synthesizes RRs for all entries from /etc/hosts.
90 * The systemd-resolve tool (which is a client utility for
91 systemd-resolved) has been improved considerably and is now fully
92 supported and documented. Hence it has moved from /usr/lib/systemd to
95 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
98 * The coredump collection logic has been reworked: when a coredump is
99 collected it is now written to disk, compressed and processed
100 (including stacktrace extraction) from a new instantiated service
101 systemd-coredump@.service, instead of directly from the
102 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
103 processing large coredumps can take up a substantial amount of
104 resources and time, and this previously happened entirely outside of
105 systemd's service supervision. With the new logic the core_pattern
106 hook only does minimal metadata collection before passing off control
107 to the new instantiated service, which is configured with a time
108 limit, a nice level and other settings to minimize negative impact on
109 the rest of the system. Also note that the new logic will honour the
110 RLIMIT_CORE setting of the crashed process, which now allows users
111 and processes to turn off coredumping for their processes by setting
114 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
115 and all forked processes by default. Previously, PID 1 would leave
116 the setting at "0" for all processes, as set by the kernel. Note that
117 the resource limit traditionally has no effect on the generated
118 coredumps on the system if the /proc/sys/kernel/core_pattern hook
119 logic is used. Since the limit is now honoured (see above) its
120 default has been changed so that the coredumping logic is enabled by
121 default for all processes, while allowing specific opt-out.
123 * When the stacktrace is extracted from processes of system users, this
124 is now done as "systemd-coredump" user, in order to sandbox this
125 potentially security sensitive parsing operation. (Note that when
126 processing coredumps of normal users this is done under the user ID
127 of process that crashed, as before.) Packagers should take notice
128 that it is now necessary to create the "systemd-coredump" system user
129 and group at package installation time.
131 * The systemd-activate socket activation testing tool gained support
132 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
133 and --seqpacket switches. It also has been extended to support both
134 new-style and inetd-style file descriptor passing. Use the new
135 --inetd switch to request inetd-style file descriptor passing.
137 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
138 variable, which takes a boolean value. If set to false, ANSI color
139 output is disabled in the tools even when run on a terminal that
142 * The VXLAN support in networkd now supports two new settings
143 DestinationPort= and PortRange=.
145 * A new systemd.machine_id= kernel command line switch has been added,
146 that may be used to set the machine ID in /etc/machine-id if it is
147 not initialized yet. This command line option has no effect if the
148 file is already initialized.
150 * systemd-nspawn gained a new --as-pid2 switch that invokes any
151 specified command line as PID 2 rather than PID 1 in the
152 container. In this mode PID 1 is a minimal stub init process that
153 implements the special POSIX and Linux semantics of PID 1 regarding
154 signal and child process management. Note that this stub init process
155 is implemented in nspawn itself and requires no support from the
156 container image. This new logic is useful to support running
157 arbitrary commands in the container, as normal processes are
158 generally not prepared to run as PID 1.
160 * systemd-nspawn gained a new --chdir= switch for setting the current
161 working directory for the process started in the container.
163 * "journalctl /dev/sda" will now output all kernel log messages for
164 specified device from the current boot, in addition to all devices
165 that are parents of it. This should make log output about devices
166 pretty useful, as long as kernel drivers attach enough metadata to
167 the log messages. (The usual SATA drivers do.)
169 * The sd-journal API gained two new calls
170 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
171 that report whether log data from /run or /var has been found.
173 * journalctl gained a new switch "--fields" that prints all journal
174 record field names currently in use in the journal. This is backed
175 by two new sd-journal API calls sd_journal_enumerate_fields() and
176 sd_journal_restart_fields().
178 * Most configurable timeouts in systemd now expect an argument of
179 "infinity" to turn them off, instead of "0" as before. The semantics
180 from now on is that a timeout of "0" means "now", and "infinity"
181 means "never". To maintain backwards compatibility, "0" continues to
182 turn off previously existing timeout settings.
184 * "systemctl reload-or-try-restart" has been renamed to "systemctl
185 try-reload-or-restart" to clarify what it actually does: the "try"
186 logic applies to both reloading and restarting, not just restarting.
187 The old name continues to be accepted for compatibility.
189 * On boot-up, when PID 1 detects that the system clock is behind the
190 release date of the systemd version in use, the clock is now set
191 to the latter. Previously, this was already done in timesyncd, in order
192 to avoid running with clocks set to the various clock epochs such as
193 1902, 1938 or 1970. With this change the logic is now done in PID 1
194 in addition to timesyncd during early boot-up, so that it is enforced
195 before the first process is spawned by systemd. Note that the logic
196 in timesyncd remains, as it is more comprehensive and ensures
197 clock monotonicity by maintaining a persistent timestamp file in
198 /var. Since /var is generally not available in earliest boot or the
199 initrd, this part of the logic remains in timesyncd, and is not done
202 * Support for tweaking details in net_cls.class_id through the
203 NetClass= configuration directive has been removed, as the kernel
204 people have decided to deprecate that controller in cgroup v2.
205 Userspace tools such as nftables are moving over to setting rules
206 that are specific to the full cgroup path of a task, which obsoletes
207 these controllers anyway. The NetClass= directive is kept around for
208 legacy compatibility reasons. For a more in-depth description of the
209 kernel change, please refer to the respective upstream commit:
211 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
213 * A new service setting RuntimeMaxSec= has been added that may be used
214 to specify a maximum runtime for a service. If the timeout is hit, the
215 service is terminated and put into a failure state.
217 * A new service setting AmbientCapabilities= has been added. It allows
218 configuration of additional Linux process capabilities that are
219 passed to the activated processes. This is only available on very
222 * The process resource limit settings in service units may now be used
223 to configure hard and soft limits individually.
225 * The various libsystemd APIs such as sd-bus or sd-event now publicly
226 expose support for gcc's __attribute__((cleanup())) C extension.
227 Specifically, for many object destructor functions alternative
228 versions have been added that have names suffixed with "p" and take a
229 pointer to a pointer to the object to destroy, instead of just a
230 pointer to the object itself. This is useful because these destructor
231 functions may be used directly as parameters to the cleanup
232 construct. Internally, systemd has been a heavy user of this GCC
233 extension for a long time, and with this change similar support is
234 now available to consumers of the library outside of systemd. Note
235 that by using this extension in your sources compatibility with old
236 and strictly ANSI compatible C compilers is lost. However, all gcc or
237 LLVM versions of recent years support this extension.
239 * Timer units gained support for a new setting RandomizedDelaySec= that
240 allows configuring some additional randomized delay to the configured
241 time. This is useful to spread out timer events to avoid load peaks in
242 clusters or larger setups.
244 * Calendar time specifications now support sub-second accuracy.
246 * Socket units now support listening on SCTP and UDP-lite protocol
249 * The sd-event API now comes with a full set of man pages.
251 * Older versions of systemd contained experimental support for
252 compressing journal files and coredumps with the LZ4 compressor that
253 was not compatible with the lz4 binary (due to API limitations of the
254 lz4 library). This support has been removed; only support for files
255 compatible with the lz4 binary remains. This LZ4 logic is now
256 officially supported and no longer considered experimental.
258 * The dkr image import logic has been removed again from importd. dkr's
259 micro-services focus doesn't fit into the machine image focus of
260 importd, and quickly got out of date with the upstream dkr API.
262 * Creation of the /run/lock/lockdev/ directory was dropped from
263 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
264 been available for many years. If you still need this, you need to
265 create your own tmpfiles.d config file with:
267 d /run/lock/lockdev 0775 root lock -
269 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
270 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
271 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
272 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
273 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
274 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
275 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
276 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
277 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
278 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
279 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
280 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
281 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
282 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
283 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
284 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
285 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
286 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
287 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
289 -- Berlin, 2016-02-11
293 * A number of properties previously only settable in unit
294 files are now also available as properties to set when
295 creating transient units programmatically via the bus, as it
296 is exposed with systemd-run's --property=
297 setting. Specifically, these are: SyslogIdentifier=,
298 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
299 EnvironmentFile=, ReadWriteDirectories=,
300 ReadOnlyDirectories=, InaccessibleDirectories=,
301 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
303 * When creating transient services via the bus API it is now
304 possible to pass in a set of file descriptors to use as
305 STDIN/STDOUT/STDERR for the invoked process.
307 * Slice units may now be created transiently via the bus APIs,
308 similar to the way service and scope units may already be
311 * Wherever systemd expects a calendar timestamp specification
312 (like in journalctl's --since= and --until= switches) UTC
313 timestamps are now supported. Timestamps suffixed with "UTC"
314 are now considered to be in Universal Time Coordinated
315 instead of the local timezone. Also, timestamps may now
316 optionally be specified with sub-second accuracy. Both of
317 these additions also apply to recurring calendar event
318 specification, such as OnCalendar= in timer units.
320 * journalctl gained a new "--sync" switch that asks the
321 journal daemon to write all so far unwritten log messages to
322 disk and sync the files, before returning.
324 * systemd-tmpfiles learned two new line types "q" and "Q" that
325 operate like "v", but also set up a basic btrfs quota
326 hierarchy when used on a btrfs file system with quota
329 * tmpfiles' "v", "q" and "Q" will now create a plain directory
330 instead of a subvolume (even on a btrfs file system) if the
331 root directory is a plain directory, and not a
332 subvolume. This should simplify things with certain chroot()
333 environments which are not aware of the concept of btrfs
336 * systemd-detect-virt gained a new --chroot switch to detect
337 whether execution takes place in a chroot() environment.
339 * CPUAffinity= now takes CPU index ranges in addition to
342 * The various memory-related resource limit settings (such as
343 LimitAS=) now understand the usual K, M, G, ... suffixes to
344 the base of 1024 (IEC). Similar, the time-related resource
345 limit settings understand the usual min, h, day, ...
348 * There's a new system.conf setting DefaultTasksMax= to
349 control the default TasksMax= setting for services and
350 scopes running on the system. (TasksMax= is the primary
351 setting that exposes the "pids" cgroup controller on systemd
352 and was introduced in the previous systemd release.) The
353 setting now defaults to 512, which means services that are
354 not explicitly configured otherwise will only be able to
355 create 512 processes or threads at maximum, from this
356 version on. Note that this means that thread- or
357 process-heavy services might need to be reconfigured to set
358 TasksMax= to a higher value. It is sufficient to set
359 TasksMax= in these specific unit files to a higher value, or
360 even "infinity". Similar, there's now a logind.conf setting
361 UserTasksMax= that defaults to 4096 and limits the total
362 number of processes or tasks each user may own
363 concurrently. nspawn containers also have the TasksMax=
364 value set by default now, to 8192. Note that all of this
365 only has an effect if the "pids" cgroup controller is
366 enabled in the kernel. The general benefit of these changes
367 should be a more robust and safer system, that provides a
368 certain amount of per-service fork() bomb protection.
370 * systemd-nspawn gained the new --network-veth-extra= switch
371 to define additional and arbitrarily-named virtual Ethernet
372 links between the host and the container.
374 * A new service execution setting PassEnvironment= has been
375 added that allows importing select environment variables
376 from PID1's environment block into the environment block of
379 * Timer units gained support for a new RemainAfterElapse=
380 setting which takes a boolean argument. It defaults on on,
381 exposing behaviour unchanged to previous releases. If set to
382 off, timer units are unloaded after they elapsed if they
383 cannot elapse again. This is particularly useful for
384 transient timer units, which shall not stay around longer
385 than until they first elapse.
387 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
388 default now (the kernel default is 16). This is beneficial
389 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
390 allows substantially larger numbers of queued
391 datagrams. This should increase the capability of systemd to
392 parallelize boot-up, as logging and sd_notify() are unlikely
393 to stall execution anymore. If you need to change the value
394 from the new defaults, use the usual sysctl.d/ snippets.
396 * The compression framing format used by the journal or
397 coredump processing has changed to be in line with what the
398 official LZ4 tools generate. LZ4 compression support in
399 systemd was considered unsupported previously, as the format
400 was not compatible with the normal tools. With this release
401 this has changed now, and it is hence safe for downstream
402 distributions to turn it on. While not compressing as well
403 as the XZ, LZ4 is substantially faster, which makes
404 it a good default choice for the compression logic in the
405 journal and in coredump handling.
407 * Any reference to /etc/mtab has been dropped from
408 systemd. The file has been obsolete since a while, but
409 systemd refused to work on systems where it was incorrectly
410 set up (it should be a symlink or non-existent). Please make
411 sure to update to util-linux 2.27.1 or newer in conjunction
412 with this systemd release, which also drops any reference to
413 /etc/mtab. If you maintain a distribution make sure that no
414 software you package still references it, as this is a
415 likely source of bugs. There's also a glibc bug pending,
416 asking for removal of any reference to this obsolete file:
418 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
420 * Support for the ".snapshot" unit type has been removed. This
421 feature turned out to be little useful and little used, and
422 has now been removed from the core and from systemctl.
424 * The dependency types RequiresOverridable= and
425 RequisiteOverridable= have been removed from systemd. They
426 have been used only very sparingly to our knowledge and
427 other options that provide a similar effect (such as
428 systemctl --mode=ignore-dependencies) are much more useful
429 and commonly used. Moreover, they were only half-way
430 implemented as the option to control behaviour regarding
431 these dependencies was never added to systemctl. By removing
432 these dependency types the execution engine becomes a bit
433 simpler. Unit files that use these dependencies should be
434 changed to use the non-Overridable dependency types
435 instead. In fact, when parsing unit files with these
436 options, that's what systemd will automatically convert them
437 too, but it will also warn, asking users to fix the unit
438 files accordingly. Removal of these dependency types should
439 only affect a negligible number of unit files in the wild.
441 * Behaviour of networkd's IPForward= option changed
442 (again). It will no longer maintain a per-interface setting,
443 but propagate one way from interfaces where this is enabled
444 to the global kernel setting. The global setting will be
445 enabled when requested by a network that is set up, but
446 never be disabled again. This change was made to make sure
447 IPv4 and IPv6 behaviour regarding packet forwarding is
448 similar (as the Linux IPv6 stack does not support
449 per-interface control of this setting) and to minimize
452 * In unit files the behaviour of %u, %U, %h, %s has
453 changed. These specifiers will now unconditionally resolve
454 to the various user database fields of the user that the
455 systemd instance is running as, instead of the user
456 configured in the specific unit via User=. Note that this
457 effectively doesn't change much, as resolving of these
458 specifiers was already turned off in the --system instance
459 of systemd, as we cannot do NSS lookups from PID 1. In the
460 --user instance of systemd these specifiers where correctly
461 resolved, but hardly made any sense, since the user instance
462 lacks privileges to do user switches anyway, and User= is
463 hence useless. Morever, even in the --user instance of
464 systemd behaviour was awkward as it would only take settings
465 from User= assignment placed before the specifier into
466 account. In order to unify and simplify the logic around
467 this the specifiers will now always resolve to the
468 credentials of the user invoking the manager (which in case
469 of PID 1 is the root user).
471 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
472 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
473 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
474 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
475 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
476 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
477 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
478 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
479 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
480 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
481 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
482 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
483 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
484 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
487 -- Berlin, 2015-11-18
491 * systemd now depends on util-linux v2.27. More specifically,
492 the newly added mount monitor feature in libmount now
493 replaces systemd's former own implementation.
495 * libmount mandates /etc/mtab not to be regular file, and
496 systemd now enforces this condition at early boot.
497 /etc/mtab has been deprecated and warned about for a very
498 long time, so systems running systemd should already have
499 stopped having this file around as anything else than a
500 symlink to /proc/self/mounts.
502 * Support for the "pids" cgroup controller has been added. It
503 allows accounting the number of tasks in a cgroup and
504 enforcing limits on it. This adds two new setting
505 TasksAccounting= and TasksMax= to each unit, as well as a
506 global option DefaultTasksAccounting=.
508 * Support for the "net_cls" cgroup controller has been added.
509 It allows assigning a net class ID to each task in the
510 cgroup, which can then be used in firewall rules and traffic
511 shaping configurations. Note that the kernel netfilter net
512 class code does not currently work reliably for ingress
513 packets on unestablished sockets.
515 This adds a new config directive called NetClass= to CGroup
516 enabled units. Allowed values are positive numbers for fixed
517 assignments and "auto" for picking a free value
520 * 'systemctl is-system-running' now returns 'offline' if the
521 system is not booted with systemd. This command can now be
522 used as a substitute for 'systemd-notify --booted'.
524 * Watchdog timeouts have been increased to 3 minutes for all
525 in-tree service files. Apparently, disk IO issues are more
526 frequent than we hoped, and user reported >1 minute waiting
529 * 'machine-id-commit' functionality has been merged into
530 'machine-id-setup --commit'. The separate binary has been
533 * The WorkingDirectory= directive in unit files may now be set
534 to the special value '~'. In this case, the working
535 directory is set to the home directory of the user
538 * "machinectl shell" will now open the shell in the home
539 directory of the selected user by default.
541 * The CrashChVT= configuration file setting is renamed to
542 CrashChangeVT=, following our usual logic of not
543 abbreviating unnecessarily. The old directive is still
544 supported for compat reasons. Also, this directive now takes
545 an integer value between 1 and 63, or a boolean value. The
546 formerly supported '-1' value for disabling stays around for
549 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
550 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
551 RootDirectory= properties can now be set for transient
554 * The systemd-analyze tool gained a new "set-log-target" verb
555 to change the logging target the system manager logs to
556 dynamically during runtime. This is similar to how
557 "systemd-analyze set-log-level" already changes the log
560 * In nspawn /sys is now mounted as tmpfs, with only a selected
561 set of subdirectories mounted in from the real sysfs. This
562 enhances security slightly, and is useful for ensuring user
563 namespaces work correctly.
565 * Support for USB FunctionFS activation has been added. This
566 allows implementation of USB gadget services that are
567 activated as soon as they are requested, so that they don't
568 have to run continously, similar to classic socket
571 * The "systemctl exit" command now optionally takes an
572 additional parameter that sets the exit code to return from
573 the systemd manager when exiting. This is only relevant when
574 running the systemd user instance, or when running the
575 system instance in a container.
577 * sd-bus gained the new API calls sd_bus_path_encode_many()
578 and sd_bus_path_decode_many() that allow easy encoding and
579 decoding of multiple identifier strings inside a D-Bus
580 object path. Another new call sd_bus_default_flush_close()
581 has been added to flush and close per-thread default
584 * systemd-cgtop gained support for a -M/--machine= switch to
585 show the control groups within a certain container only.
587 * "systemctl kill" gained support for an optional --fail
588 switch. If specified the requested operation will fail of no
589 processes have been killed, because the unit had no
590 processes attached, or similar.
592 * A new systemd.crash_reboot=1 kernel command line option has
593 been added that triggers a reboot after crashing. This can
594 also be set through CrashReboot= in systemd.conf.
596 * The RuntimeDirectory= setting now understands unit
597 specifiers like %i or %f.
599 * A new (still internal) libary API sd-ipv4acd has been added,
600 that implements address conflict detection for IPv4. It's
601 based on code from sd-ipv4ll, and will be useful for
602 detecting DHCP address conflicts.
604 * File descriptors passed during socket activation may now be
605 named. A new API sd_listen_fds_with_names() is added to
606 access the names. The default names may be overridden,
607 either in the .socket file using the FileDescriptorName=
608 parameter, or by passing FDNAME= when storing the file
609 descriptors using sd_notify().
611 * systemd-networkd gained support for:
613 - Setting the IPv6 Router Advertisment settings via
614 IPv6AcceptRouterAdvertisements= in .network files.
616 - Configuring the HelloTimeSec=, MaxAgeSec= and
617 ForwardDelaySec= bridge parameters in .netdev files.
619 - Configuring PreferredSource= for static routes in
622 * The "ask-password" framework used to query for LUKS harddisk
623 passwords or SSL passwords during boot gained support for
624 caching passwords in the kernel keyring, if it is
625 available. This makes sure that the user only has to type in
626 a passphrase once if there are multiple objects to unlock
627 with the same one. Previously, such password caching was
628 available only when Plymouth was used; this moves the
629 caching logic into the systemd codebase itself. The
630 "systemd-ask-password" utility gained a new --keyname=
631 switch to control which kernel keyring key to use for
632 caching a password in. This functionality is also useful for
633 enabling display managers such as gdm to automatically
634 unlock the user's GNOME keyring if its passphrase, the
635 user's password and the harddisk password are the same, if
636 gdm-autologin is used.
638 * When downloading tar or raw images using "machinectl
639 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
640 file is now also downloaded, if it is available and stored
641 next to the image file.
643 * Units of type ".socket" gained a new boolean setting
644 Writable= which is only useful in conjunction with
645 ListenSpecial=. If true, enables opening the specified
646 special file in O_RDWR mode rather than O_RDONLY mode.
648 * systemd-rfkill has been reworked to become a singleton
649 service that is activated through /dev/rfkill on each rfkill
650 state change and saves the settings to disk. This way,
651 systemd-rfkill is now compatible with devices that exist
652 only intermittendly, and even restores state if the previous
653 system shutdown was abrupt rather than clean.
655 * The journal daemon gained support for vacuuming old journal
656 files controlled by the number of files that shall remain,
657 in addition to the already existing control by size and by
658 date. This is useful as journal interleaving performance
659 degrades with too many seperate journal files, and allows
660 putting an effective limit on them. The new setting defaults
661 to 100, but this may be changed by setting SystemMaxFiles=
662 and RuntimeMaxFiles= in journald.conf. Also, the
663 "journalctl" tool gained the new --vacuum-files= switch to
664 manually vacuum journal files to leave only the specified
665 number of files in place.
667 * udev will now create /dev/disk/by-path links for ATA devices
668 on kernels where that is supported.
670 * Galician, Serbian, Turkish and Korean translations were added.
672 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
673 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
674 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
675 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
676 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
677 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
678 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
679 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
680 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
681 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
682 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
683 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
684 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
685 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
686 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
687 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
688 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
689 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
691 -- Berlin, 2015-10-07
695 * The DHCP implementation of systemd-networkd gained a set of
698 - The DHCP server now supports emitting DNS and NTP
699 information. It may be enabled and configured via
700 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
701 and NTP information is enabled, but no servers are
702 configured, the corresponding uplink information (if there
703 is any) is propagated.
705 - Server and client now support transmission and reception
706 of timezone information. It can be configured via the
707 newly introduced network options UseTimezone=,
708 EmitTimezone=, and Timezone=. Transmission of timezone
709 information is enabled between host and containers by
710 default now: the container will change its local timezone
711 to what the host has set.
713 - Lease timeouts can now be configured via
714 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
716 - The DHCP server improved on the stability of
717 leases. Clients are more likely to get the same lease
718 information back, even if the server loses state.
720 - The DHCP server supports two new configuration options to
721 control the lease address pool metrics, PoolOffset= and
724 * The encapsulation limit of tunnels in systemd-networkd may
725 now be configured via 'EncapsulationLimit='. It allows
726 modifying the maximum additional levels of encapsulation
727 that are permitted to be prepended to a packet.
729 * systemd now supports the concept of user buses replacing
730 session buses, if used with dbus-1.10 (and enabled via dbus
731 --enable-user-session). It previously only supported this on
732 kdbus-enabled systems, and this release expands this to
733 'dbus-daemon' systems.
735 * systemd-networkd now supports predictable interface names
738 * systemd now optionally supports the new Linux kernel
739 "unified" control group hierarchy. If enabled via the kernel
740 command-line option 'systemd.unified_cgroup_hierarchy=1',
741 systemd will try to mount the unified cgroup hierarchy
742 directly on /sys/fs/cgroup. If not enabled, or not
743 available, systemd will fall back to the legacy cgroup
744 hierarchy setup, as before. Host system and containers can
745 mix and match legacy and unified hierarchies as they
746 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
747 environment variable to individually select the hierarchy to
748 use for executed containers. By default, nspawn will use the
749 unified hierarchy for the containers if the host uses the
750 unified hierarchy, and the legacy hierarchy otherwise.
751 Please note that at this point the unified hierarchy is an
752 experimental kernel feature and is likely to change in one
753 of the next kernel releases. Therefore, it should not be
754 enabled by default in downstream distributions yet. The
755 minimum required kernel version for the unified hierarchy to
756 work is 4.2. Note that when the unified hierarchy is used
757 for the first time delegated access to controllers is
758 safe. Because of this systemd-nspawn containers will get
759 access to controllers now, as will systemd user
760 sessions. This means containers and user sessions may now
761 manage their own resources, partitioning up what the system
764 * A new special scope unit "init.scope" has been introduced
765 that encapsulates PID 1 of the system. It may be used to
766 determine resource usage and enforce resource limits on PID
767 1 itself. PID 1 hence moved out of the root of the control
770 * The cgtop tool gained support for filtering out kernel
771 threads when counting tasks in a control group. Also, the
772 count of processes is now recursively summed up by
773 default. Two options -k and --recursive= have been added to
774 revert to old behaviour. The tool has also been updated to
775 work correctly in containers now.
777 * systemd-nspawn's --bind= and --bind-ro= options have been
778 extended to allow creation of non-recursive bind mounts.
780 * libsystemd gained two new calls sd_pid_get_cgroup() and
781 sd_peer_get_cgroup() which return the control group path of
782 a process or peer of a connected AF_UNIX socket. This
783 function call is particularly useful when implementing
784 delegated subtrees support in the control group hierarchy.
786 * The "sd-event" event loop API of libsystemd now supports
787 correct dequeuing of real-time signals, without losing
790 * When systemd requests a PolicyKit decision when managing
791 units it will now add additional fields to the request,
792 including unit name and desired operation. This enables more
793 powerful PolicyKit policies, that make decisions depending
796 * nspawn learnt support for .nspawn settings files, that may
797 accompany the image files or directories of containers, and
798 may contain additional settings for the container. This is
799 an alternative to configuring container parameters via the
802 Contributions from: Cristian Rodríguez, Daniel Mack, David
803 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
804 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
805 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
806 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
807 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
808 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
809 Andersen, Tom Gundersen, Torstein Husebø
811 -- Berlin, 2015-09-08
815 * machinectl gained a new verb 'shell' which opens a fresh
816 shell on the target container or the host. It is similar to
817 the existing 'login' command of machinectl, but spawns the
818 shell directly without prompting for username or
819 password. The pseudo machine '.host' now refers to the local
820 host and is used by default. Hence, 'machinectl shell' can
821 be used as replacement for 'su -' which spawns a session as
822 a fresh systemd unit in a way that is fully isolated from
823 the originating session.
825 * systemd-networkd learned to cope with private-zone DHCP
826 options and allows other programs to query the values.
828 * SELinux access control when enabling/disabling units is no
829 longer enforced with this release. The previous
830 implementation was incorrect, and a new corrected
831 implementation is not yet available. As unit file operations
832 are still protected via PolicyKit and D-Bus policy this is
833 not a security problem. Yet, distributions which care about
834 optimal SELinux support should probably not stabilize on
837 * sd-bus gained support for matches of type "arg0has=", that
838 test for membership of strings in string arrays sent in bus
841 * systemd-resolved now dumps the contents of its DNS and LLMNR
842 caches to the logs on reception of the SIGUSR1 signal. This
843 is useful to debug DNS behaviour.
845 * The coredumpctl tool gained a new --directory= option to
846 operate on journal files in a specific directory.
848 * "systemctl reboot" and related commands gained a new
849 "--message=" option which may be used to set a free-text
850 wall message when shutting down or rebooting the
851 system. This message is also logged, which is useful for
852 figuring out the reason for a reboot or shutdown a
855 * The "systemd-resolve-host" tool's -i switch now takes
856 network interface numbers as alternative to interface names.
858 * A new unit file setting for services has been introduced:
859 UtmpMode= allows configuration of how precisely systemd
860 handles utmp and wtmp entries for the service if this is
861 enabled. This allows writing services that appear similar to
862 user sessions in the output of the "w", "who", "last" and
865 * systemd-resolved will now locally synthesize DNS resource
866 records for the "localhost" and "gateway" domains as well as
867 the local hostname. This should ensure that clients querying
868 RRs via resolved will get similar results as those going via
869 NSS, if nss-myhostname is enabled.
871 Contributions from: Alastair Hughes, Alex Crawford, Daniel
872 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
873 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
874 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
875 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
876 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
877 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
878 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
879 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
880 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
881 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
882 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
884 -- Berlin, 2015-08-27
888 * The systemd-efi-boot-generator functionality was merged into
889 systemd-gpt-auto-generator.
891 * systemd-networkd now supports Group Policy for vxlan
892 devices. It can be enabled via the new boolean configuration
893 option called 'GroupPolicyExtension='.
895 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
896 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
897 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
899 -- Berlin, 2015-07-31
903 * The python-systemd code has been removed from the systemd repository.
904 A new repository has been created which accommodates the code from
905 now on, and we kindly ask distributions to create a separate package
906 for this: https://github.com/systemd/python-systemd
908 * The systemd daemon will now reload its main configuration
909 (/etc/systemd/system.conf) on daemon-reload.
911 * sd-dhcp now exposes vendor specific extensions via
912 sd_dhcp_lease_get_vendor_specific().
914 * systemd-networkd gained a number of new configuration options.
916 - A new boolean configuration option for TAP devices called
917 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
918 device, thus allowing to send and receive GSO packets.
920 - A new tunnel configuration option called 'CopyDSCP='.
921 If enabled, the DSCP field of ip6 tunnels is copied into the
924 - A set of boolean bridge configuration options were added.
925 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
926 and 'UnicastFlood=' are now parsed by networkd and applied to the
927 respective bridge link device via the respective IFLA_BRPORT_*
930 - A new string configuration option to override the hostname sent
931 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
932 is true, networkd will use the configured hostname instead of the
933 system hostname when sending DHCP requests.
935 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
936 networkd will configure the IPv6 flow-label of the tunnel device
937 according to RFC2460.
939 - The 'macvtap' virtual network devices are now supported, similar to
940 the already supported 'macvlan' devices.
942 * systemd-resolved now implements RFC5452 to improve resilience against
943 cache poisoning. Additionally, source port randomization is enabled
944 by default to further protect against DNS spoofing attacks.
946 * nss-mymachines now supports translating UIDs and GIDs of running
947 containers with user-namespaces enabled. If a container 'foo'
948 translates a host uid 'UID' to the container uid 'TUID', then
949 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
950 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
951 mapped as 'vg-foo-TGID'.
953 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
954 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
955 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
956 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
957 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
958 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
959 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
960 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
961 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
962 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
964 -- Berlin, 2015-07-29
968 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
969 There are no known issues with current sysfs, and udev does not need
970 or should be used to work around such bugs.
972 * udev does no longer enable USB HID power management. Several reports
973 indicate, that some devices cannot handle that setting.
975 * The udev accelerometer helper was removed. The functionality
976 is now fully included in iio-sensor-proxy. But this means,
977 older iio-sensor-proxy versions will no longer provide
978 accelerometer/orientation data with this systemd version.
979 Please upgrade iio-sensor-proxy to version 1.0.
981 * networkd gained a new configuration option IPv6PrivacyExtensions=
982 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
983 for Stateless Address") on selected networks.
985 * For the sake of fewer build-time dependencies and less code in the
986 main repository, the python bindings are about to be removed in the
987 next release. A new repository has been created which accommodates
988 the code from now on, and we kindly ask distributions to create a
989 separate package for this. The removal will take place in v223.
991 https://github.com/systemd/python-systemd
993 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
994 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
995 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
996 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
997 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
998 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
999 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1000 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1001 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1002 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1004 -- Berlin, 2015-07-07
1008 * The sd-bus.h and sd-event.h APIs have now been declared
1009 stable and have been added to the official interface of
1010 libsystemd.so. sd-bus implements an alternative D-Bus client
1011 library, that is relatively easy to use, very efficient and
1012 supports both classic D-Bus as well as kdbus as transport
1013 backend. sd-event is a generic event loop abstraction that
1014 is built around Linux epoll, but adds features such as event
1015 prioritization or efficient timer handling. Both APIs are good
1016 choices for C programs looking for a bus and/or event loop
1017 implementation that is minimal and does not have to be
1018 portable to other kernels.
1020 * kdbus support is no longer compile-time optional. It is now
1021 always built-in. However, it can still be disabled at
1022 runtime using the kdbus=0 kernel command line setting, and
1023 that setting may be changed to default to off, by specifying
1024 --disable-kdbus at build-time. Note though that the kernel
1025 command line setting has no effect if the kdbus.ko kernel
1026 module is not installed, in which case kdbus is (obviously)
1027 also disabled. We encourage all downstream distributions to
1028 begin testing kdbus by adding it to the kernel images in the
1029 development distributions, and leaving kdbus support in
1032 * The minimal required util-linux version has been bumped to
1035 * Support for chkconfig (--enable-chkconfig) was removed in
1036 favor of calling an abstraction tool
1037 /lib/systemd/systemd-sysv-install. This needs to be
1038 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1039 in README for details.
1041 * If there's a systemd unit and a SysV init script for the
1042 same service name, and the user executes "systemctl enable"
1043 for it (or a related call), then this will now enable both
1044 (or execute the related operation on both), not just the
1047 * The libudev API documentation has been converted from gtkdoc
1050 * gudev has been removed from the systemd tree, it is now an
1053 * The systemd-cgtop tool learnt a new --raw switch to generate
1054 "raw" (machine parsable) output.
1056 * networkd's IPForwarding= .network file setting learnt the
1057 new setting "kernel", which ensures that networkd does not
1058 change the IP forwarding sysctl from the default kernel
1061 * The systemd-logind bus API now exposes a new boolean
1062 property "Docked" that reports whether logind considers the
1063 system "docked", i.e. connected to a docking station or not.
1065 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1066 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1067 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1068 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1069 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1070 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1071 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1072 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1073 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1074 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1075 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1076 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1077 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1078 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1079 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1080 Fink, Zbigniew Jędrzejewski-Szmek
1082 -- Berlin, 2015-06-19
1086 * The gudev library has been extracted into a separate repository
1087 available at: https://git.gnome.org/browse/libgudev/
1088 It is now managed as part of the Gnome project. Distributions
1089 are recommended to pass --disable-gudev to systemd and use
1090 gudev from the Gnome project instead. gudev is still included
1091 in systemd, for now. It will be removed soon, though. Please
1092 also see the announcement-thread on systemd-devel:
1093 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1095 * systemd now exposes a CPUUsageNSec= property for each
1096 service unit on the bus, that contains the overall consumed
1097 CPU time of a service (the sum of what each process of the
1098 service consumed). This value is only available if
1099 CPUAccounting= is turned on for a service, and is then shown
1100 in the "systemctl status" output.
1102 * Support for configuring alternative mappings of the old SysV
1103 runlevels to systemd targets has been removed. They are now
1104 hardcoded in a way that runlevels 2, 3, 4 all map to
1105 multi-user.target and 5 to graphical.target (which
1106 previously was already the default behaviour).
1108 * The auto-mounter logic gained support for mount point
1109 expiry, using a new TimeoutIdleSec= setting in .automount
1110 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1112 * The EFI System Partition (ESP) as mounted to /boot by
1113 systemd-efi-boot-generator will now be unmounted
1114 automatically after 2 minutes of not being used. This should
1115 minimize the risk of ESP corruptions.
1117 * New /etc/fstab options x-systemd.requires= and
1118 x-systemd.requires-mounts-for= are now supported to express
1119 additional dependencies for mounts. This is useful for
1120 journalling file systems that support external journal
1121 devices or overlay file systems that require underlying file
1122 systems to be mounted.
1124 * systemd does not support direct live-upgrades (via systemctl
1125 daemon-reexec) from versions older than v44 anymore. As no
1126 distribution we are aware of shipped such old versions in a
1127 stable release this should not be problematic.
1129 * When systemd forks off a new per-connection service instance
1130 it will now set the $REMOTE_ADDR environment variable to the
1131 remote IP address, and $REMOTE_PORT environment variable to
1132 the remote IP port. This behaviour is similar to the
1133 corresponding environment variables defined by CGI.
1135 * systemd-networkd gained support for uplink failure
1136 detection. The BindCarrier= option allows binding interface
1137 configuration dynamically to the link sense of other
1138 interfaces. This is useful to achieve behaviour like in
1141 * systemd-networkd gained support for configuring the DHCP
1142 client identifier to use when requesting leases.
1144 * systemd-networkd now has a per-network UseNTP= option to
1145 configure whether NTP server information acquired via DHCP
1146 is passed on to services like systemd-timesyncd.
1148 * systemd-networkd gained support for vti6 tunnels.
1150 * Note that systemd-networkd manages the sysctl variable
1151 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1152 it is configured for since v219. The variable controls IP
1153 forwarding, and is a per-interface alternative to the global
1154 /proc/sys/net/ipv[46]/ip_forward. This setting is
1155 configurable in the IPForward= option, which defaults to
1156 "no". This means if networkd is used for an interface it is
1157 no longer sufficient to set the global sysctl option to turn
1158 on IP forwarding! Instead, the .network file option
1159 IPForward= needs to be turned on! Note that the
1160 implementation of this behaviour was broken in v219 and has
1163 * Many bonding and vxlan options are now configurable in
1166 * systemd-nspawn gained a new --property= setting to set unit
1167 properties for the container scope. This is useful for
1168 setting resource parameters (e.g "CPUShares=500") on
1169 containers started from the command line.
1171 * systemd-nspawn gained a new --private-users= switch to make
1172 use of user namespacing available on recent Linux kernels.
1174 * systemd-nspawn may now be called as part of a shell pipeline
1175 in which case the pipes used for stdin and stdout are passed
1176 directly to the process invoked in the container, without
1177 indirection via a pseudo tty.
1179 * systemd-nspawn gained a new switch to control the UNIX
1180 signal to use when killing the init process of the container
1183 * systemd-nspawn gained a new --overlay= switch for mounting
1184 overlay file systems into the container using the new kernel
1187 * When a container image is imported via systemd-importd and
1188 the host file system is not btrfs, a loopback block device
1189 file is created in /var/lib/machines.raw with a btrfs file
1190 system inside. It is then mounted to /var/lib/machines to
1191 enable btrfs features for container management. The loopback
1192 file and btrfs file system is grown as needed when container
1193 images are imported via systemd-importd.
1195 * systemd-machined/systemd-importd gained support for btrfs
1196 quota, to enforce container disk space limits on disk. This
1197 is exposed in "machinectl set-limit".
1199 * systemd-importd now can import containers from local .tar,
1200 .raw and .qcow2 images, and export them to .tar and .raw. It
1201 can also import dkr v2 images now from the network (on top
1204 * systemd-importd gained support for verifying downloaded
1205 images with gpg2 (previously only gpg1 was supported).
1207 * systemd-machined, systemd-logind, systemd: most bus calls
1208 are now accessible to unprivileged processes via
1209 PolicyKit. Also, systemd-logind will now allow users to kill
1210 their own sessions without further privileges or
1213 * systemd-shutdownd has been removed. This service was
1214 previously responsible for implementing scheduled shutdowns
1215 as exposed in /usr/bin/shutdown's time parameter. This
1216 functionality has now been moved into systemd-logind and is
1217 accessible via a bus interface.
1219 * "systemctl reboot" gained a new switch --firmware-setup that
1220 can be used to reboot into the EFI firmware setup, if that
1221 is available. systemd-logind now exposes an API on the bus
1222 to trigger such reboots, in case graphical desktop UIs want
1223 to cover this functionality.
1225 * "systemctl enable", "systemctl disable" and "systemctl mask"
1226 now support a new "--now" switch. If specified the units
1227 that are enabled will also be started, and the ones
1228 disabled/masked also stopped.
1230 * The Gummiboot EFI boot loader tool has been merged into
1231 systemd, and renamed to "systemd-boot". The bootctl tool has been
1232 updated to support systemd-boot.
1234 * An EFI kernel stub has been added that may be used to create
1235 kernel EFI binaries that contain not only the actual kernel,
1236 but also an initrd, boot splash, command line and OS release
1237 information. This combined binary can then be signed as a
1238 single image, so that the firmware can verify it all in one
1239 step. systemd-boot has special support for EFI binaries created
1240 like this and can extract OS release information from them
1241 and show them in the boot menu. This functionality is useful
1242 to implement cryptographically verified boot schemes.
1244 * Optional support has been added to systemd-fsck to pass
1245 fsck's progress report to an AF_UNIX socket in the file
1248 * udev will no longer create device symlinks for all block
1249 devices by default. A blacklist for excluding special block
1250 devices from this logic has been turned into a whitelist
1251 that requires picking block devices explicitly that require
1254 * A new (currently still internal) API sd-device.h has been
1255 added to libsystemd. This modernized API is supposed to
1256 replace libudev eventually. In fact, already much of libudev
1257 is now just a wrapper around sd-device.h.
1259 * A new hwdb database for storing metadata about pointing
1260 stick devices has been added.
1262 * systemd-tmpfiles gained support for setting file attributes
1263 similar to the "chattr" tool with new 'h' and 'H' lines.
1265 * systemd-journald will no longer unconditionally set the
1266 btrfs NOCOW flag on new journal files. This is instead done
1267 with tmpfiles snippet using the new 'h' line type. This
1268 allows easy disabling of this logic, by masking the
1269 journal-nocow.conf tmpfiles file.
1271 * systemd-journald will now translate audit message types to
1272 human readable identifiers when writing them to the
1273 journal. This should improve readability of audit messages.
1275 * The LUKS logic gained support for the offset= and skip=
1276 options in /etc/crypttab, as previously implemented by
1279 * /usr/lib/os-release gained a new optional field VARIANT= for
1280 distributions that support multiple variants (such as a
1281 desktop edition, a server edition, ...)
1283 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1284 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1285 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1286 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1287 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1288 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1289 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1290 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1291 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1292 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1293 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1294 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1295 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1296 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1297 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1298 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1299 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1300 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1301 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1302 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1303 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1304 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1305 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1306 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1307 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1308 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1309 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1311 -- Berlin, 2015-05-22
1315 * Introduce a new API "sd-hwdb.h" for querying the hardware
1316 metadata database. With this minimal interface one can query
1317 and enumerate the udev hwdb, decoupled from the old libudev
1318 library. libudev's interface for this is now only a wrapper
1319 around sd-hwdb. A new tool systemd-hwdb has been added to
1320 interface with and update the database.
1322 * When any of systemd's tools copies files (for example due to
1323 tmpfiles' C lines) a btrfs reflink will attempted first,
1324 before bytewise copying is done.
1326 * systemd-nspawn gained a new --ephemeral switch. When
1327 specified a btrfs snapshot is taken of the container's root
1328 directory, and immediately removed when the container
1329 terminates again. Thus, a container can be started whose
1330 changes never alter the container's root directory, and are
1331 lost on container termination. This switch can also be used
1332 for starting a container off the root file system of the
1333 host without affecting the host OS. This switch is only
1334 available on btrfs file systems.
1336 * systemd-nspawn gained a new --template= switch. It takes the
1337 path to a container tree to use as template for the tree
1338 specified via --directory=, should that directory be
1339 missing. This allows instantiating containers dynamically,
1340 on first run. This switch is only available on btrfs file
1343 * When a .mount unit refers to a mount point on which multiple
1344 mounts are stacked, and the .mount unit is stopped all of
1345 the stacked mount points will now be unmounted until no
1346 mount point remains.
1348 * systemd now has an explicit notion of supported and
1349 unsupported unit types. Jobs enqueued for unsupported unit
1350 types will now fail with an "unsupported" error code. More
1351 specifically .swap, .automount and .device units are not
1352 supported in containers, .busname units are not supported on
1353 non-kdbus systems. .swap and .automount are also not
1354 supported if their respective kernel compile time options
1357 * machinectl gained support for two new "copy-from" and
1358 "copy-to" commands for copying files from a running
1359 container to the host or vice versa.
1361 * machinectl gained support for a new "bind" command to bind
1362 mount host directories into local containers. This is
1363 currently only supported for nspawn containers.
1365 * networkd gained support for configuring bridge forwarding
1366 database entries (fdb) from .network files.
1368 * A new tiny daemon "systemd-importd" has been added that can
1369 download container images in tar, raw, qcow2 or dkr formats,
1370 and make them available locally in /var/lib/machines, so
1371 that they can run as nspawn containers. The daemon can GPG
1372 verify the downloads (not supported for dkr, since it has no
1373 provisions for verifying downloads). It will transparently
1374 decompress bz2, xz, gzip compressed downloads if necessary,
1375 and restore sparse files on disk. The daemon uses privilege
1376 separation to ensure the actual download logic runs with
1377 fewer privileges than the daemon itself. machinectl has
1378 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1379 make the functionality of importd available to the
1380 user. With this in place the Fedora and Ubuntu "Cloud"
1381 images can be downloaded and booted as containers unmodified
1382 (the Fedora images lack the appropriate GPG signature files
1383 currently, so they cannot be verified, but this will change
1384 soon, hopefully). Note that downloading images is currently
1385 only fully supported on btrfs.
1387 * machinectl is now able to list container images found in
1388 /var/lib/machines, along with some metadata about sizes of
1389 disk and similar. If the directory is located on btrfs and
1390 quota is enabled, this includes quota display. A new command
1391 "image-status" has been added that shows additional
1392 information about images.
1394 * machinectl is now able to clone container images
1395 efficiently, if the underlying file system (btrfs) supports
1396 it, with the new "machinectl clone" command. It also
1397 gained commands for renaming and removing images, as well as
1398 marking them read-only or read-write (supported also on
1399 legacy file systems).
1401 * networkd gained support for collecting LLDP network
1402 announcements, from hardware that supports this. This is
1403 shown in networkctl output.
1405 * systemd-run gained support for a new -t (--pty) switch for
1406 invoking a binary on a pty whose input and output is
1407 connected to the invoking terminal. This allows executing
1408 processes as system services while interactively
1409 communicating with them via the terminal. Most interestingly
1410 this is supported across container boundaries. Invoking
1411 "systemd-run -t /bin/bash" is an alternative to running a
1412 full login session, the difference being that the former
1413 will not register a session, nor go through the PAM session
1416 * tmpfiles gained support for a new "v" line type for creating
1417 btrfs subvolumes. If the underlying file system is a legacy
1418 file system, this automatically degrades to creating a
1419 normal directory. Among others /var/lib/machines is now
1420 created like this at boot, should it be missing.
1422 * The directory /var/lib/containers/ has been deprecated and
1423 been replaced by /var/lib/machines. The term "machines" has
1424 been used in the systemd context as generic term for both
1425 VMs and containers, and hence appears more appropriate for
1426 this, as the directory can also contain raw images bootable
1429 * systemd-nspawn when invoked with -M but without --directory=
1430 or --image= is now capable of searching for the container
1431 root directory, subvolume or disk image automatically, in
1432 /var/lib/machines. systemd-nspawn@.service has been updated
1433 to make use of this, thus allowing it to be used for raw
1436 * A new machines.target unit has been introduced that is
1437 supposed to group all containers/VMs invoked as services on
1438 the system. systemd-nspawn@.service has been updated to
1439 integrate with that.
1441 * machinectl gained a new "start" command, for invoking a
1442 container as a service. "machinectl start foo" is mostly
1443 equivalent to "systemctl start systemd-nspawn@foo.service",
1444 but handles escaping in a nicer way.
1446 * systemd-nspawn will now mount most of the cgroupfs tree
1447 read-only into each container, with the exception of the
1448 container's own subtree in the name=systemd hierarchy.
1450 * journald now sets the special FS_NOCOW file flag for its
1451 journal files. This should improve performance on btrfs, by
1452 avoiding heavy fragmentation when journald's write-pattern
1453 is used on COW file systems. It degrades btrfs' data
1454 integrity guarantees for the files to the same levels as for
1455 ext3/ext4 however. This should be OK though as journald does
1456 its own data integrity checks and all its objects are
1457 checksummed on disk. Also, journald should handle btrfs disk
1458 full events a lot more gracefully now, by processing SIGBUS
1459 errors, and not relying on fallocate() anymore.
1461 * When journald detects that journal files it is writing to
1462 have been deleted it will immediately start new journal
1465 * systemd now provides a way to store file descriptors
1466 per-service in PID 1.This is useful for daemons to ensure
1467 that fds they require are not lost during a daemon
1468 restart. The fds are passed to the daemon on the next
1469 invocation in the same way socket activation fds are
1470 passed. This is now used by journald to ensure that the
1471 various sockets connected to all the system's stdout/stderr
1472 are not lost when journald is restarted. File descriptors
1473 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1474 an extension to sd_notify(). Note that a limit is enforced
1475 on the number of fds a service can store in PID 1, and it
1476 defaults to 0, so that no fds may be stored, unless this is
1477 explicitly turned on.
1479 * The default TERM variable to use for units connected to a
1480 terminal, when no other value is explicitly is set is now
1481 vt220 rather than vt102. This should be fairly safe still,
1482 but allows PgUp/PgDn work.
1484 * The /etc/crypttab option header= as known from Debian is now
1487 * "loginctl user-status" and "loginctl session-status" will
1488 now show the last 10 lines of log messages of the
1489 user/session following the status output. Similar,
1490 "machinectl status" will show the last 10 log lines
1491 associated with a virtual machine or container
1492 service. (Note that this is usually not the log messages
1493 done in the VM/container itself, but simply what the
1494 container manager logs. For nspawn this includes all console
1497 * "loginctl session-status" without further argument will now
1498 show the status of the session of the caller. Similar,
1499 "lock-session", "unlock-session", "activate",
1500 "enable-linger", "disable-linger" may now be called without
1501 session/user parameter in which case they apply to the
1502 caller's session/user.
1504 * An X11 session scriptlet is now shipped that uploads
1505 $DISPLAY and $XAUTHORITY into the environment of the systemd
1506 --user daemon if a session begins. This should improve
1507 compatibility with X11 enabled applications run as systemd
1510 * Generators are now subject to masking via /etc and /run, the
1511 same way as unit files.
1513 * networkd .network files gained support for configuring
1514 per-link IPv4/IPv6 packet forwarding as well as IPv4
1515 masquerading. This is by default turned on for veth links to
1516 containers, as registered by systemd-nspawn. This means that
1517 nspawn containers run with --network-veth will now get
1518 automatic routed access to the host's networks without any
1519 further configuration or setup, as long as networkd runs on
1522 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1523 or UDP posts of a container on the host. With this in place
1524 it is possible to run containers with private veth links
1525 (--network-veth), and have their functionality exposed on
1526 the host as if their services were running directly on the
1529 * systemd-nspawn's --network-veth switch now gained a short
1530 version "-n", since with the changes above it is now truly
1531 useful out-of-the-box. The systemd-nspawn@.service has been
1532 updated to make use of it too by default.
1534 * systemd-nspawn will now maintain a per-image R/W lock, to
1535 ensure that the same image is not started more than once
1536 writable. (It's OK to run an image multiple times
1537 simultaneously in read-only mode.)
1539 * systemd-nspawn's --image= option is now capable of
1540 dissecting and booting MBR and GPT disk images that contain
1541 only a single active Linux partition. Previously it
1542 supported only GPT disk images with proper GPT type
1543 IDs. This allows running cloud images from major
1544 distributions directly with systemd-nspawn, without
1547 * In addition to collecting mouse dpi data in the udev
1548 hardware database, there's now support for collecting angle
1549 information for mouse scroll wheels. The database is
1550 supposed to guarantee similar scrolling behavior on mice
1551 that it knows about. There's also support for collecting
1552 information about Touchpad types.
1554 * udev's input_id built-in will now also collect touch screen
1555 dimension data and attach it to probed devices.
1557 * /etc/os-release gained support for a Distribution Privacy
1560 * networkd gained support for creating "ipvlan", "gretap",
1561 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1563 * systemd-tmpfiles gained support for "a" lines for setting
1566 * systemd-nspawn will now mount /tmp in the container to
1567 tmpfs, automatically.
1569 * systemd now exposes the memory.usage_in_bytes cgroup
1570 attribute and shows it for each service in the "systemctl
1571 status" output, if available.
1573 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1574 immediate reboot is triggered. This useful if shutdown is
1575 hung and is unable to complete, to expedite the
1576 operation. Note that this kind of reboot will still unmount
1577 all file systems, and hence should not result in fsck being
1580 * A .device unit for an optical block device will now be
1581 considered active only when a medium is in the drive. Also,
1582 mount units are now bound to their backing devices thus
1583 triggering automatic unmounting when devices become
1584 unavailable. With this in place systemd will now
1585 automatically unmount left-over mounts when a CD-ROM is
1586 ejected or an USB stick is yanked from the system.
1588 * networkd-wait-online now has support for waiting for
1589 specific interfaces only (with globbing), and for giving up
1590 after a configurable timeout.
1592 * networkd now exits when idle. It will be automatically
1593 restarted as soon as interfaces show up, are removed or
1594 change state. networkd will stay around as long as there is
1595 at least one DHCP state machine or similar around, that keep
1598 * networkd may now configure IPv6 link-local addressing in
1599 addition to IPv4 link-local addressing.
1601 * The IPv6 "token" for use in SLAAC may now be configured for
1602 each .network interface in networkd.
1604 * Routes configured with networkd may now be assigned a scope
1607 * networkd's [Match] sections now support globbing and lists
1608 of multiple space-separated matches per item.
1610 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1611 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1612 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1613 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1614 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1615 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1616 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1617 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1618 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1619 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1620 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1621 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1622 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1623 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1624 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1625 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1626 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1627 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1628 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1629 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1630 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1631 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1632 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1633 Hoffmann, Zbigniew Jędrzejewski-Szmek
1635 -- Berlin, 2015-02-16
1639 * When querying unit file enablement status (for example via
1640 "systemctl is-enabled"), a new state "indirect" is now known
1641 which indicates that a unit might not be enabled itself, but
1642 another unit listed in its Also= setting might be.
1644 * Similar to the various existing ConditionXYZ= settings for
1645 units, there are now matching AssertXYZ= settings. While
1646 failing conditions cause a unit to be skipped, but its job
1647 to succeed, failing assertions declared like this will cause
1648 a unit start operation and its job to fail.
1650 * hostnamed now knows a new chassis type "embedded".
1652 * systemctl gained a new "edit" command. When used on a unit
1653 file, this allows extending unit files with .d/ drop-in
1654 configuration snippets or editing the full file (after
1655 copying it from /usr/lib to /etc). This will invoke the
1656 user's editor (as configured with $EDITOR), and reload the
1657 modified configuration after editing.
1659 * "systemctl status" now shows the suggested enablement state
1660 for a unit, as declared in the (usually vendor-supplied)
1661 system preset files.
1663 * nss-myhostname will now resolve the single-label host name
1664 "gateway" to the locally configured default IP routing
1665 gateways, ordered by their metrics. This assigns a stable
1666 name to the used gateways, regardless which ones are
1667 currently configured. Note that the name will only be
1668 resolved after all other name sources (if nss-myhostname is
1669 configured properly) and should hence not negatively impact
1670 systems that use the single-label host name "gateway" in
1673 * systemd-inhibit now allows filtering by mode when listing
1676 * Scope and service units gained a new "Delegate" boolean
1677 property, which, when set, allows processes running inside the
1678 unit to further partition resources. This is primarily
1679 useful for systemd user instances as well as container
1682 * journald will now pick up audit messages directly from
1683 the kernel, and log them like any other log message. The
1684 audit fields are split up and fully indexed. This means that
1685 journalctl in many ways is now a (nicer!) alternative to
1686 ausearch, the traditional audit client. Note that this
1687 implements only a minimal audit client. If you want the
1688 special audit modes like reboot-on-log-overflow, please use
1689 the traditional auditd instead, which can be used in
1690 parallel to journald.
1692 * The ConditionSecurity= unit file option now understands the
1693 special string "audit" to check whether auditing is
1696 * journalctl gained two new commands --vacuum-size= and
1697 --vacuum-time= to delete old journal files until the
1698 remaining ones take up no more than the specified size on disk,
1699 or are not older than the specified time.
1701 * A new, native PPPoE library has been added to sd-network,
1702 systemd's library of light-weight networking protocols. This
1703 library will be used in a future version of networkd to
1704 enable PPPoE communication without an external pppd daemon.
1706 * The busctl tool now understands a new "capture" verb that
1707 works similar to "monitor", but writes a packet capture
1708 trace to STDOUT that can be redirected to a file which is
1709 compatible with libcap's capture file format. This can then
1710 be loaded in Wireshark and similar tools to inspect bus
1713 * The busctl tool now understands a new "tree" verb that shows
1714 the object trees of a specific service on the bus, or of all
1717 * The busctl tool now understands a new "introspect" verb that
1718 shows all interfaces and members of objects on the bus,
1719 including their signature and values. This is particularly
1720 useful to get more information about bus objects shown by
1721 the new "busctl tree" command.
1723 * The busctl tool now understands new verbs "call",
1724 "set-property" and "get-property" for invoking bus method
1725 calls, setting and getting bus object properties in a
1728 * busctl gained a new --augment-creds= argument that controls
1729 whether the tool shall augment credential information it
1730 gets from the bus with data from /proc, in a possibly
1733 * nspawn's --link-journal= switch gained two new values
1734 "try-guest" and "try-host" that work like "guest" and
1735 "host", but do not fail if the host has no persistent
1736 journalling enabled. -j is now equivalent to
1737 --link-journal=try-guest.
1739 * macvlan network devices created by nspawn will now have
1740 stable MAC addresses.
1742 * A new SmackProcessLabel= unit setting has been added, which
1743 controls the SMACK security label processes forked off by
1744 the respective unit shall use.
1746 * If compiled with --enable-xkbcommon, systemd-localed will
1747 verify x11 keymap settings by compiling the given keymap. It
1748 will spew out warnings if the compilation fails. This
1749 requires libxkbcommon to be installed.
1751 * When a coredump is collected, a larger number of metadata
1752 fields is now collected and included in the journal records
1753 created for it. More specifically, control group membership,
1754 environment variables, memory maps, working directory,
1755 chroot directory, /proc/$PID/status, and a list of open file
1756 descriptors is now stored in the log entry.
1758 * The udev hwdb now contains DPI information for mice. For
1761 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1763 * All systemd programs that read standalone configuration
1764 files in /etc now also support a corresponding series of
1765 .conf.d configuration directories in /etc/, /run/,
1766 /usr/local/lib/, /usr/lib/, and (if configured with
1767 --enable-split-usr) /lib/. In particular, the following
1768 configuration files now have corresponding configuration
1769 directories: system.conf user.conf, logind.conf,
1770 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1771 resolved.conf, timesyncd.conf, journal-remote.conf, and
1772 journal-upload.conf. Note that distributions should use the
1773 configuration directories in /usr/lib/; the directories in
1774 /etc/ are reserved for the system administrator.
1776 * systemd-rfkill will no longer take the rfkill device name
1777 into account when storing rfkill state on disk, as the name
1778 might be dynamically assigned and not stable. Instead, the
1779 ID_PATH udev variable combined with the rfkill type (wlan,
1780 bluetooth, ...) is used.
1782 * A new service systemd-machine-id-commit.service has been
1783 added. When used on systems where /etc is read-only during
1784 boot, and /etc/machine-id is not initialized (but an empty
1785 file), this service will copy the temporary machine ID
1786 created as replacement into /etc after the system is fully
1787 booted up. This is useful for systems that are freshly
1788 installed with a non-initialized machine ID, but should get
1789 a fixed machine ID for subsequent boots.
1791 * networkd's .netdev files now provide a large set of
1792 configuration parameters for VXLAN devices. Similarly, the
1793 bridge port cost parameter is now configurable in .network
1794 files. There's also new support for configuring IP source
1795 routing. networkd .link files gained support for a new
1796 OriginalName= match that is useful to match against the
1797 original interface name the kernel assigned. .network files
1798 may include MTU= and MACAddress= fields for altering the MTU
1799 and MAC address while being connected to a specific network
1802 * The LUKS logic gained supported for configuring
1803 UUID-specific key files. There's also new support for naming
1804 LUKS device from the kernel command line, using the new
1805 luks.name= argument.
1807 * Timer units may now be transiently created via the bus API
1808 (this was previously already available for scope and service
1809 units). In addition it is now possible to create multiple
1810 transient units at the same time with a single bus call. The
1811 "systemd-run" tool has been updated to make use of this for
1812 running commands on a specified time, in at(1)-style.
1814 * tmpfiles gained support for "t" lines, for assigning
1815 extended attributes to files. Among other uses this may be
1816 used to assign SMACK labels to files.
1818 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1819 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1820 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1821 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1822 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1823 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1824 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1825 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1826 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1827 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1828 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1829 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1830 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1831 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1832 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1833 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1834 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1835 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1837 -- Berlin, 2014-12-10
1841 * journalctl gained the new options -t/--identifier= to match
1842 on the syslog identifier (aka "tag"), as well as --utc to
1843 show log timestamps in the UTC timezone. journalctl now also
1844 accepts -n/--lines=all to disable line capping in a pager.
1846 * journalctl gained a new switch, --flush, that synchronously
1847 flushes logs from /run/log/journal to /var/log/journal if
1848 persistent storage is enabled. systemd-journal-flush.service
1849 now waits until the operation is complete.
1851 * Services can notify the manager before they start a reload
1852 (by sending RELOADING=1) or shutdown (by sending
1853 STOPPING=1). This allows the manager to track and show the
1854 internal state of daemons and closes a race condition when
1855 the process is still running but has closed its D-Bus
1858 * Services with Type=oneshot do not have to have any ExecStart
1861 * User units are now loaded also from
1862 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1863 /run/systemd/user directory that was already previously
1864 supported, but is under the control of the user.
1866 * Job timeouts (i.e. time-outs on the time a job that is
1867 queued stays in the run queue) can now optionally result in
1868 immediate reboot or power-off actions (JobTimeoutAction= and
1869 JobTimeoutRebootArgument=). This is useful on ".target"
1870 units, to limit the maximum time a target remains
1871 undispatched in the run queue, and to trigger an emergency
1872 operation in such a case. This is now used by default to
1873 turn off the system if boot-up (as defined by everything in
1874 basic.target) hangs and does not complete for at least
1875 15min. Also, if power-off or reboot hang for at least 30min
1876 an immediate power-off/reboot operation is triggered. This
1877 functionality is particularly useful to increase reliability
1878 on embedded devices, but also on laptops which might
1879 accidentally get powered on when carried in a backpack and
1880 whose boot stays stuck in a hard disk encryption passphrase
1883 * systemd-logind can be configured to also handle lid switch
1884 events even when the machine is docked or multiple displays
1885 are attached (HandleLidSwitchDocked= option).
1887 * A helper binary and a service have been added which can be
1888 used to resume from hibernation in the initramfs. A
1889 generator will parse the resume= option on the kernel
1890 command line to trigger resume.
1892 * A user console daemon systemd-consoled has been
1893 added. Currently, it is a preview, and will so far open a
1894 single terminal on each session of the user marked as
1895 Desktop=systemd-console.
1897 * Route metrics can be specified for DHCP routes added by
1900 * The SELinux context of socket-activated services can be set
1901 from the information provided by the networking stack
1902 (SELinuxContextFromNet= option).
1904 * Userspace firmware loading support has been removed and
1905 the minimum supported kernel version is thus bumped to 3.7.
1907 * Timeout for udev workers has been increased from 1 to 3
1908 minutes, but a warning will be printed after 1 minute to
1909 help diagnose kernel modules that take a long time to load.
1911 * Udev rules can now remove tags on devices with TAG-="foobar".
1913 * systemd's readahead implementation has been removed. In many
1914 circumstances it didn't give expected benefits even for
1915 rotational disk drives and was becoming less relevant in the
1916 age of SSDs. As none of the developers has been using
1917 rotating media anymore, and nobody stepped up to actively
1918 maintain this component of systemd it has now been removed.
1920 * Swap units can use Options= to specify discard options.
1921 Discard options specified for swaps in /etc/fstab are now
1924 * Docker containers are now detected as a separate type of
1927 * The Password Agent protocol gained support for queries where
1928 the user input is shown, useful e.g. for user names.
1929 systemd-ask-password gained a new --echo option to turn that
1932 * The default sysctl.d/ snippets will now set:
1934 net.core.default_qdisc = fq_codel
1936 This selects Fair Queuing Controlled Delay as the default
1937 queuing discipline for network interfaces. fq_codel helps
1938 fight the network bufferbloat problem. It is believed to be
1939 a good default with no tuning required for most workloads.
1940 Downstream distributions may override this choice. On 10Gbit
1941 servers that do not do forwarding, "fq" may perform better.
1942 Systems without a good clocksource should use "pfifo_fast".
1944 * If kdbus is enabled during build a new option BusPolicy= is
1945 available for service units, that allows locking all service
1946 processes into a stricter bus policy, in order to limit
1947 access to various bus services, or even hide most of them
1948 from the service's view entirely.
1950 * networkctl will now show the .network and .link file
1951 networkd has applied to a specific interface.
1953 * sd-login gained a new API call sd_session_get_desktop() to
1954 query which desktop environment has been selected for a
1957 * UNIX utmp support is now compile-time optional to support
1958 legacy-free systems.
1960 * systemctl gained two new commands "add-wants" and
1961 "add-requires" for pulling in units from specific targets
1964 * If the word "rescue" is specified on the kernel command line
1965 the system will now boot into rescue mode (aka
1966 rescue.target), which was previously available only by
1967 specifying "1" or "systemd.unit=rescue.target" on the kernel
1968 command line. This new kernel command line option nicely
1969 mirrors the already existing "emergency" kernel command line
1972 * New kernel command line options mount.usr=, mount.usrflags=,
1973 mount.usrfstype= have been added that match root=, rootflags=,
1974 rootfstype= but allow mounting a specific file system to
1977 * The $NOTIFY_SOCKET is now also passed to control processes of
1978 services, not only the main process.
1980 * This version reenables support for fsck's -l switch. This
1981 means at least version v2.25 of util-linux is required for
1982 operation, otherwise dead-locks on device nodes may
1983 occur. Again: you need to update util-linux to at least
1984 v2.25 when updating systemd to v217.
1986 * The "multi-seat-x" tool has been removed from systemd, as
1987 its functionality has been integrated into X servers 1.16,
1988 and the tool is hence redundant. It is recommended to update
1989 display managers invoking this tool to simply invoke X
1990 directly from now on, again.
1992 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1993 message flag has been added for all of systemd's PolicyKit
1994 authenticated method calls has been added. In particular
1995 this now allows optional interactive authorization via
1996 PolicyKit for many of PID1's privileged operations such as
1997 unit file enabling and disabling.
1999 * "udevadm hwdb --update" learnt a new switch "--usr" for
2000 placing the rebuilt hardware database in /usr instead of
2001 /etc. When used only hardware database entries stored in
2002 /usr will be used, and any user database entries in /etc are
2003 ignored. This functionality is useful for vendors to ship a
2004 pre-built database on systems where local configuration is
2005 unnecessary or unlikely.
2007 * Calendar time specifications in .timer units now also
2008 understand the strings "semi-annually", "quarterly" and
2009 "minutely" as shortcuts (in addition to the preexisting
2010 "anually", "hourly", ...).
2012 * systemd-tmpfiles will now correctly create files in /dev
2013 at boot which are marked for creation only at boot. It is
2014 recommended to always create static device nodes with 'c!'
2015 and 'b!', so that they are created only at boot and not
2016 overwritten at runtime.
2018 * When the watchdog logic is used for a service (WatchdogSec=)
2019 and the watchdog timeout is hit the service will now be
2020 terminated with SIGABRT (instead of just SIGTERM), in order
2021 to make sure a proper coredump and backtrace is
2022 generated. This ensures that hanging services will result in
2023 similar coredump/backtrace behaviour as services that hit a
2026 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2027 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2028 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2029 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2030 Herrmann, David Sommerseth, David Strauss, Emil Renner
2031 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2032 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2033 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2034 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2035 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2036 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2037 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2038 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2039 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2040 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2041 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2042 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2043 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2044 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2045 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2046 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2049 -- Berlin, 2014-10-28
2053 * timedated no longer reads NTP implementation unit names from
2054 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2055 implementations should add a
2057 Conflicts=systemd-timesyncd.service
2059 to their unit files to take over and replace systemd's NTP
2060 default functionality.
2062 * systemd-sysusers gained a new line type "r" for configuring
2063 which UID/GID ranges to allocate system users/groups
2064 from. Lines of type "u" may now add an additional column
2065 that specifies the home directory for the system user to be
2066 created. Also, systemd-sysusers may now optionally read user
2067 information from STDIN instead of a file. This is useful for
2068 invoking it from RPM preinst scriptlets that need to create
2069 users before the first RPM file is installed since these
2070 files might need to be owned by them. A new
2071 %sysusers_create_inline RPM macro has been introduced to do
2072 just that. systemd-sysusers now updates the shadow files as
2073 well as the user/group databases, which should enhance
2074 compatibility with certain tools like grpck.
2076 * A number of bus APIs of PID 1 now optionally consult
2077 PolicyKit to permit access for otherwise unprivileged
2078 clients under certain conditions. Note that this currently
2079 doesn't support interactive authentication yet, but this is
2080 expected to be added eventually, too.
2082 * /etc/machine-info now has new fields for configuring the
2083 deployment environment of the machine, as well as the
2084 location of the machine. hostnamectl has been updated with
2085 new command to update these fields.
2087 * systemd-timesyncd has been updated to automatically acquire
2088 NTP server information from systemd-networkd, which might
2089 have been discovered via DHCP.
2091 * systemd-resolved now includes a caching DNS stub resolver
2092 and a complete LLMNR name resolution implementation. A new
2093 NSS module "nss-resolve" has been added which can be used
2094 instead of glibc's own "nss-dns" to resolve hostnames via
2095 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2096 be resolved via systemd-resolved D-Bus APIs. In contrast to
2097 the glibc internal resolver systemd-resolved is aware of
2098 multi-homed system, and keeps DNS server and caches separate
2099 and per-interface. Queries are sent simultaneously on all
2100 interfaces that have DNS servers configured, in order to
2101 properly handle VPNs and local LANs which might resolve
2102 separate sets of domain names. systemd-resolved may acquire
2103 DNS server information from systemd-networkd automatically,
2104 which in turn might have discovered them via DHCP. A tool
2105 "systemd-resolve-host" has been added that may be used to
2106 query the DNS logic in resolved. systemd-resolved implements
2107 IDNA and automatically uses IDNA or UTF-8 encoding depending
2108 on whether classic DNS or LLMNR is used as transport. In the
2109 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2110 implementation to systemd-resolved.
2112 * A new NSS module nss-mymachines has been added, that
2113 automatically resolves the names of all local registered
2114 containers to their respective IP addresses.
2116 * A new client tool "networkctl" for systemd-networkd has been
2117 added. It currently is entirely passive and will query
2118 networking configuration from udev, rtnetlink and networkd,
2119 and present it to the user in a very friendly
2120 way. Eventually, we hope to extend it to become a full
2121 control utility for networkd.
2123 * .socket units gained a new DeferAcceptSec= setting that
2124 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2125 TCP. Similarly, support for controlling TCP keep-alive
2126 settings has been added (KeepAliveTimeSec=,
2127 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2128 turning off Nagle's algorithm on TCP has been added
2131 * logind learned a new session type "web", for use in projects
2132 like Cockpit which register web clients as PAM sessions.
2134 * timer units with at least one OnCalendar= setting will now
2135 be started only after timer-sync.target has been
2136 reached. This way they will not elapse before the system
2137 clock has been corrected by a local NTP client or
2138 similar. This is particular useful on RTC-less embedded
2139 machines, that come up with an invalid system clock.
2141 * systemd-nspawn's --network-veth= switch should now result in
2142 stable MAC addresses for both the outer and the inner side
2145 * systemd-nspawn gained a new --volatile= switch for running
2146 container instances with /etc or /var unpopulated.
2148 * The kdbus client code has been updated to use the new Linux
2149 3.17 memfd subsystem instead of the old kdbus-specific one.
2151 * systemd-networkd's DHCP client and server now support
2152 FORCERENEW. There are also new configuration options to
2153 configure the vendor client identifier and broadcast mode
2156 * systemd will no longer inform the kernel about the current
2157 timezone, as this is necessarily incorrect and racy as the
2158 kernel has no understanding of DST and similar
2159 concepts. This hence means FAT timestamps will be always
2160 considered UTC, similar to what Android is already
2161 doing. Also, when the RTC is configured to the local time
2162 (rather than UTC) systemd will never synchronize back to it,
2163 as this might confuse Windows at a later boot.
2165 * systemd-analyze gained a new command "verify" for offline
2166 validation of unit files.
2168 * systemd-networkd gained support for a couple of additional
2169 settings for bonding networking setups. Also, the metric for
2170 statically configured routes may now be configured. For
2171 network interfaces where this is appropriate the peer IP
2172 address may now be configured.
2174 * systemd-networkd's DHCP client will no longer request
2175 broadcasting by default, as this tripped up some networks.
2176 For hardware where broadcast is required the feature should
2177 be switched back on using RequestBroadcast=yes.
2179 * systemd-networkd will now set up IPv4LL addresses (when
2180 enabled) even if DHCP is configured successfully.
2182 * udev will now default to respect network device names given
2183 by the kernel when the kernel indicates that these are
2184 predictable. This behavior can be tweaked by changing
2185 NamePolicy= in the relevant .link file.
2187 * A new library systemd-terminal has been added that
2188 implements full TTY stream parsing and rendering. This
2189 library is supposed to be used later on for implementing a
2190 full userspace VT subsystem, replacing the current kernel
2193 * A new tool systemd-journal-upload has been added to push
2194 journal data to a remote system running
2195 systemd-journal-remote.
2197 * journald will no longer forward all local data to another
2198 running syslog daemon. This change has been made because
2199 rsyslog (which appears to be the most commonly used syslog
2200 implementation these days) no longer makes use of this, and
2201 instead pulls the data out of the journal on its own. Since
2202 forwarding the messages to a non-existent syslog server is
2203 more expensive than we assumed we have now turned this
2204 off. If you run a syslog server that is not a recent rsyslog
2205 version, you have to turn this option on again
2206 (ForwardToSyslog= in journald.conf).
2208 * journald now optionally supports the LZ4 compressor for
2209 larger journal fields. This compressor should perform much
2210 better than XZ which was the previous default.
2212 * machinectl now shows the IP addresses of local containers,
2213 if it knows them, plus the interface name of the container.
2215 * A new tool "systemd-escape" has been added that makes it
2216 easy to escape strings to build unit names and similar.
2218 * sd_notify() messages may now include a new ERRNO= field
2219 which is parsed and collected by systemd and shown among the
2220 "systemctl status" output for a service.
2222 * A new component "systemd-firstboot" has been added that
2223 queries the most basic systemd information (timezone,
2224 hostname, root password) interactively on first
2225 boot. Alternatively it may also be used to provision these
2226 things offline on OS images installed into directories.
2228 * The default sysctl.d/ snippets will now set
2230 net.ipv4.conf.default.promote_secondaries=1
2232 This has the benefit of no flushing secondary IP addresses
2233 when primary addresses are removed.
2235 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2236 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2237 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2238 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2239 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2240 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2241 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2242 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2243 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2244 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2245 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2246 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2247 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2248 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2249 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2251 -- Berlin, 2014-08-19
2255 * A new tool systemd-sysusers has been added. This tool
2256 creates system users and groups in /etc/passwd and
2257 /etc/group, based on static declarative system user/group
2258 definitions in /usr/lib/sysusers.d/. This is useful to
2259 enable factory resets and volatile systems that boot up with
2260 an empty /etc directory, and thus need system users and
2261 groups created during early boot. systemd now also ships
2262 with two default sysusers.d/ files for the most basic
2263 users and groups systemd and the core operating system
2266 * A new tmpfiles snippet has been added that rebuilds the
2267 essential files in /etc on boot, should they be missing.
2269 * A directive for ensuring automatic clean-up of
2270 /var/cache/man/ has been removed from the default
2271 configuration. This line should now be shipped by the man
2272 implementation. The necessary change has been made to the
2273 man-db implementation. Note that you need to update your man
2274 implementation to one that ships this line, otherwise no
2275 automatic clean-up of /var/cache/man will take place.
2277 * A new condition ConditionNeedsUpdate= has been added that
2278 may conditionalize services to only run when /etc or /var
2279 are "older" than the vendor operating system resources in
2280 /usr. This is useful for reconstructing or updating /etc
2281 after an offline update of /usr or a factory reset, on the
2282 next reboot. Services that want to run once after such an
2283 update or reset should use this condition and order
2284 themselves before the new systemd-update-done.service, which
2285 will mark the two directories as fully updated. A number of
2286 service files have been added making use of this, to rebuild
2287 the udev hardware database, the journald message catalog and
2288 dynamic loader cache (ldconfig). The systemd-sysusers tool
2289 described above also makes use of this now. With this in
2290 place it is now possible to start up a minimal operating
2291 system with /etc empty cleanly. For more information on the
2292 concepts involved see this recent blog story:
2294 http://0pointer.de/blog/projects/stateless.html
2296 * A new system group "input" has been introduced, and all
2297 input device nodes get this group assigned. This is useful
2298 for system-level software to get access to input devices. It
2299 complements what is already done for "audio" and "video".
2301 * systemd-networkd learnt minimal DHCPv4 server support in
2302 addition to the existing DHCPv4 client support. It also
2303 learnt DHCPv6 client and IPv6 Router Solicitation client
2304 support. The DHCPv4 client gained support for static routes
2305 passed in from the server. Note that the [DHCPv4] section
2306 known in older systemd-networkd versions has been renamed to
2307 [DHCP] and is now also used by the DHCPv6 client. Existing
2308 .network files using settings of this section should be
2309 updated, though compatibility is maintained. Optionally, the
2310 client hostname may now be sent to the DHCP server.
2312 * networkd gained support for vxlan virtual networks as well
2313 as tun/tap and dummy devices.
2315 * networkd gained support for automatic allocation of address
2316 ranges for interfaces from a system-wide pool of
2317 addresses. This is useful for dynamically managing a large
2318 number of interfaces with a single network configuration
2319 file. In particular this is useful to easily assign
2320 appropriate IP addresses to the veth links of a large number
2321 of nspawn instances.
2323 * RPM macros for processing sysusers, sysctl and binfmt
2324 drop-in snippets at package installation time have been
2327 * The /etc/os-release file should now be placed in
2328 /usr/lib/os-release. The old location is automatically
2329 created as symlink. /usr/lib is the more appropriate
2330 location of this file, since it shall actually describe the
2331 vendor operating system shipped in /usr, and not the
2332 configuration stored in /etc.
2334 * .mount units gained a new boolean SloppyOptions= setting
2335 that maps to mount(8)'s -s option which enables permissive
2336 parsing of unknown mount options.
2338 * tmpfiles learnt a new "L+" directive which creates a symlink
2339 but (unlike "L") deletes a pre-existing file first, should
2340 it already exist and not already be the correct
2341 symlink. Similarly, "b+", "c+" and "p+" directives have been
2342 added as well, which create block and character devices, as
2343 well as fifos in the filesystem, possibly removing any
2344 pre-existing files of different types.
2346 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2347 'argument' field (which so far specified the source to
2348 symlink/copy the files from) is now optional. If omitted the
2349 same file os copied from /usr/share/factory/ suffixed by the
2350 full destination path. This is useful for populating /etc
2351 with essential files, by copying them from vendor defaults
2352 shipped in /usr/share/factory/etc.
2354 * A new command "systemctl preset-all" has been added that
2355 applies the service preset settings to all installed unit
2356 files. A new switch --preset-mode= has been added that
2357 controls whether only enable or only disable operations
2360 * A new command "systemctl is-system-running" has been added
2361 that allows checking the overall state of the system, for
2362 example whether it is fully up and running.
2364 * When the system boots up with an empty /etc, the equivalent
2365 to "systemctl preset-all" is executed during early boot, to
2366 make sure all default services are enabled after a factory
2369 * systemd now contains a minimal preset file that enables the
2370 most basic services systemd ships by default.
2372 * Unit files' [Install] section gained a new DefaultInstance=
2373 field for defining the default instance to create if a
2374 template unit is enabled with no instance specified.
2376 * A new passive target cryptsetup-pre.target has been added
2377 that may be used by services that need to make they run and
2378 finish before the first LUKS cryptographic device is set up.
2380 * The /dev/loop-control and /dev/btrfs-control device nodes
2381 are now owned by the "disk" group by default, opening up
2382 access to this group.
2384 * systemd-coredump will now automatically generate a
2385 stack trace of all core dumps taking place on the system,
2386 based on elfutils' libdw library. This stack trace is logged
2389 * systemd-coredump may now optionally store coredumps directly
2390 on disk (in /var/lib/systemd/coredump, possibly compressed),
2391 instead of storing them unconditionally in the journal. This
2392 mode is the new default. A new configuration file
2393 /etc/systemd/coredump.conf has been added to configure this
2394 and other parameters of systemd-coredump.
2396 * coredumpctl gained a new "info" verb to show details about a
2397 specific coredump. A new switch "-1" has also been added
2398 that makes sure to only show information about the most
2399 recent entry instead of all entries. Also, as the tool is
2400 generally useful now the "systemd-" prefix of the binary
2401 name has been removed. Distributions that want to maintain
2402 compatibility with the old name should add a symlink from
2403 the old name to the new name.
2405 * journald's SplitMode= now defaults to "uid". This makes sure
2406 that unprivileged users can access their own coredumps with
2407 coredumpctl without restrictions.
2409 * New kernel command line options "systemd.wants=" (for
2410 pulling an additional unit during boot), "systemd.mask="
2411 (for masking a specific unit for the boot), and
2412 "systemd.debug-shell" (for enabling the debug shell on tty9)
2413 have been added. This is implemented in the new generator
2414 "systemd-debug-generator".
2416 * systemd-nspawn will now by default filter a couple of
2417 syscalls for containers, among them those required for
2418 kernel module loading, direct x86 IO port access, swap
2419 management, and kexec. Most importantly though
2420 open_by_handle_at() is now prohibited for containers,
2421 closing a hole similar to a recently discussed vulnerability
2422 in docker regarding access to files on file hierarchies the
2423 container should normally not have access to. Note that, for
2424 nspawn, we generally make no security claims anyway (and
2425 this is explicitly documented in the man page), so this is
2426 just a fix for one of the most obvious problems.
2428 * A new man page file-hierarchy(7) has been added that
2429 contains a minimized, modernized version of the file system
2430 layout systemd expects, similar in style to the FHS
2431 specification or hier(5). A new tool systemd-path(1) has
2432 been added to query many of these paths for the local
2435 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2436 longer done. Since the directory now has a per-user size
2437 limit, and is cleaned on logout this appears unnecessary,
2438 in particular since this now brings the lifecycle of this
2439 directory closer in line with how IPC objects are handled.
2441 * systemd.pc now exports a number of additional directories,
2442 including $libdir (which is useful to identify the library
2443 path for the primary architecture of the system), and a
2444 couple of drop-in directories.
2446 * udev's predictable network interface names now use the dev_port
2447 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2448 distinguish between ports of the same PCI function. dev_id should
2449 only be used for ports using the same HW address, hence the need
2452 * machined has been updated to export the OS version of a
2453 container (read from /etc/os-release and
2454 /usr/lib/os-release) on the bus. This is now shown in
2455 "machinectl status" for a machine.
2457 * A new service setting RestartForceExitStatus= has been
2458 added. If configured to a set of exit signals or process
2459 return values, the service will be restarted when the main
2460 daemon process exits with any of them, regardless of the
2463 * systemctl's -H switch for connecting to remote systemd
2464 machines has been extended so that it may be used to
2465 directly connect to a specific container on the
2466 host. "systemctl -H root@foobar:waldi" will now connect as
2467 user "root" to host "foobar", and then proceed directly to
2468 the container named "waldi". Note that currently you have to
2469 authenticate as user "root" for this to work, as entering
2470 containers is a privileged operation.
2472 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2473 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2474 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2475 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2476 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2477 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2478 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2479 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2480 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2481 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2482 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2483 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2485 -- Berlin, 2014-07-03
2489 * As an experimental feature, udev now tries to lock the
2490 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2491 executes events for the disk or any of its partitions.
2492 Applications like partitioning programs can lock the
2493 disk device node (flock(LOCK_EX)) and claim temporary
2494 device ownership that way; udev will entirely skip all event
2495 handling for this disk and its partitions. If the disk
2496 was opened for writing, the close will trigger a partition
2497 table rescan in udev's "watch" facility, and if needed
2498 synthesize "change" events for the disk and all its partitions.
2499 This is now unconditionally enabled, and if it turns out to
2500 cause major problems, we might turn it on only for specific
2501 devices, or might need to disable it entirely. Device Mapper
2502 devices are excluded from this logic.
2504 * We temporarily dropped the "-l" switch for fsck invocations,
2505 since they collide with the flock() logic above. util-linux
2506 upstream has been changed already to avoid this conflict,
2507 and we will readd "-l" as soon as util-linux with this
2508 change has been released.
2510 * The dependency on libattr has been removed. Since a long
2511 time, the extended attribute calls have moved to glibc, and
2512 libattr is thus unnecessary.
2514 * Virtualization detection works without priviliges now. This
2515 means the systemd-detect-virt binary no longer requires
2516 CAP_SYS_PTRACE file capabilities, and our daemons can run
2517 with fewer privileges.
2519 * systemd-networkd now runs under its own "systemd-network"
2520 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2521 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2522 loses the ability to write to files owned by root this way.
2524 * Similarly, systemd-resolved now runs under its own
2525 "systemd-resolve" user with no capabilities remaining.
2527 * Similarly, systemd-bus-proxyd now runs under its own
2528 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2530 * systemd-networkd gained support for setting up "veth"
2531 virtual Ethernet devices for container connectivity, as well
2532 as GRE and VTI tunnels.
2534 * systemd-networkd will no longer automatically attempt to
2535 manually load kernel modules necessary for certain tunnel
2536 transports. Instead, it is assumed the kernel loads them
2537 automatically when required. This only works correctly on
2538 very new kernels. On older kernels, please consider adding
2539 the kernel modules to /etc/modules-load.d/ as a work-around.
2541 * The resolv.conf file systemd-resolved generates has been
2542 moved to /run/systemd/resolve/. If you have a symlink from
2543 /etc/resolv.conf, it might be necessary to correct it.
2545 * Two new service settings, ProtectHome= and ProtectSystem=,
2546 have been added. When enabled, they will make the user data
2547 (such as /home) inaccessible or read-only and the system
2548 (such as /usr) read-only, for specific services. This allows
2549 very light-weight per-service sandboxing to avoid
2550 modifications of user data or system files from
2551 services. These two new switches have been enabled for all
2552 of systemd's long-running services, where appropriate.
2554 * Socket units gained new SocketUser= and SocketGroup=
2555 settings to set the owner user and group of AF_UNIX sockets
2556 and FIFOs in the file system.
2558 * Socket units gained a new RemoveOnStop= setting. If enabled,
2559 all FIFOS and sockets in the file system will be removed
2560 when the specific socket unit is stopped.
2562 * Socket units gained a new Symlinks= setting. It takes a list
2563 of symlinks to create to file system sockets or FIFOs
2564 created by the specific Unix sockets. This is useful to
2565 manage symlinks to socket nodes with the same life-cycle as
2568 * The /dev/log socket and /dev/initctl FIFO have been moved to
2569 /run, and have been replaced by symlinks. This allows
2570 connecting to these facilities even if PrivateDevices=yes is
2571 used for a service (which makes /dev/log itself unavailable,
2572 but /run is left). This also has the benefit of ensuring
2573 that /dev only contains device nodes, directories and
2574 symlinks, and nothing else.
2576 * sd-daemon gained two new calls sd_pid_notify() and
2577 sd_pid_notifyf(). They are similar to sd_notify() and
2578 sd_notifyf(), but allow overriding of the source PID of
2579 notification messages if permissions permit this. This is
2580 useful to send notify messages on behalf of a different
2581 process (for example, the parent process). The
2582 systemd-notify tool has been updated to make use of this
2583 when sending messages (so that notification messages now
2584 originate from the shell script invoking systemd-notify and
2585 not the systemd-notify process itself. This should minimize
2586 a race where systemd fails to associate notification
2587 messages to services when the originating process already
2590 * A new "on-abnormal" setting for Restart= has been added. If
2591 set, it will result in automatic restarts on all "abnormal"
2592 reasons for a process to exit, which includes unclean
2593 signals, core dumps, timeouts and watchdog timeouts, but
2594 does not include clean and unclean exit codes or clean
2595 signals. Restart=on-abnormal is an alternative for
2596 Restart=on-failure for services that shall be able to
2597 terminate and avoid restarts on certain errors, by
2598 indicating so with an unclean exit code. Restart=on-failure
2599 or Restart=on-abnormal is now the recommended setting for
2600 all long-running services.
2602 * If the InaccessibleDirectories= service setting points to a
2603 mount point (or if there are any submounts contained within
2604 it), it is now attempted to completely unmount it, to make
2605 the file systems truly unavailable for the respective
2608 * The ReadOnlyDirectories= service setting and
2609 systemd-nspawn's --read-only parameter are now recursively
2610 applied to all submounts, too.
2612 * Mount units may now be created transiently via the bus APIs.
2614 * The support for SysV and LSB init scripts has been removed
2615 from the systemd daemon itself. Instead, it is now
2616 implemented as a generator that creates native systemd units
2617 from these scripts when needed. This enables us to remove a
2618 substantial amount of legacy code from PID 1, following the
2619 fact that many distributions only ship a very small number
2620 of LSB/SysV init scripts nowadays.
2622 * Privileged Xen (dom0) domains are not considered
2623 virtualization anymore by the virtualization detection
2624 logic. After all, they generally have unrestricted access to
2625 the hardware and usually are used to manage the unprivileged
2628 * systemd-tmpfiles gained a new "C" line type, for copying
2629 files or entire directories.
2631 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2632 lines. So far, they have been non-globbing versions of the
2633 latter, and have thus been redundant. In future, it is
2634 recommended to only use "z". "m" has hence been removed
2635 from the documentation, even though it stays supported.
2637 * A tmpfiles snippet to recreate the most basic structure in
2638 /var has been added. This is enough to create the /var/run →
2639 /run symlink and create a couple of structural
2640 directories. This allows systems to boot up with an empty or
2641 volatile /var. Of course, while with this change, the core OS
2642 now is capable with dealing with a volatile /var, not all
2643 user services are ready for it. However, we hope that sooner
2644 or later, many service daemons will be changed upstream so
2645 that they are able to automatically create their necessary
2646 directories in /var at boot, should they be missing. This is
2647 the first step to allow state-less systems that only require
2648 the vendor image for /usr to boot.
2650 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2651 empty tmpfs instance to a specific directory. This is
2652 particularly useful for making use of the automatic
2653 reconstruction of /var (see above), by passing --tmpfs=/var.
2655 * Access modes specified in tmpfiles snippets may now be
2656 prefixed with "~", which indicates that they shall be masked
2657 by whether the existing file or directory is currently
2658 writable, readable or executable at all. Also, if specified,
2659 the sgid/suid/sticky bits will be masked for all
2662 * A new passive target unit "network-pre.target" has been
2663 added which is useful for services that shall run before any
2664 network is configured, for example firewall scripts.
2666 * The "floppy" group that previously owned the /dev/fd*
2667 devices is no longer used. The "disk" group is now used
2668 instead. Distributions should probably deprecate usage of
2671 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2672 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2673 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2674 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2675 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2676 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2679 -- Berlin, 2014-06-11
2683 * A new "systemd-timesyncd" daemon has been added for
2684 synchronizing the system clock across the network. It
2685 implements an SNTP client. In contrast to NTP
2686 implementations such as chrony or the NTP reference server,
2687 this only implements a client side, and does not bother with
2688 the full NTP complexity, focusing only on querying time from
2689 one remote server and synchronizing the local clock to
2690 it. Unless you intend to serve NTP to networked clients or
2691 want to connect to local hardware clocks, this simple NTP
2692 client should be more than appropriate for most
2693 installations. The daemon runs with minimal privileges, and
2694 has been hooked up with networkd to only operate when
2695 network connectivity is available. The daemon saves the
2696 current clock to disk every time a new NTP sync has been
2697 acquired, and uses this to possibly correct the system clock
2698 early at bootup, in order to accommodate for systems that
2699 lack an RTC such as the Raspberry Pi and embedded devices,
2700 and to make sure that time monotonically progresses on these
2701 systems, even if it is not always correct. To make use of
2702 this daemon, a new system user and group "systemd-timesync"
2703 needs to be created on installation of systemd.
2705 * The queue "seqnum" interface of libudev has been disabled, as
2706 it was generally incompatible with device namespacing as
2707 sequence numbers of devices go "missing" if the devices are
2708 part of a different namespace.
2710 * "systemctl list-timers" and "systemctl list-sockets" gained
2711 a --recursive switch for showing units of these types also
2712 for all local containers, similar in style to the already
2713 supported --recursive switch for "systemctl list-units".
2715 * A new RebootArgument= setting has been added for service
2716 units, which may be used to specify a kernel reboot argument
2717 to use when triggering reboots with StartLimitAction=.
2719 * A new FailureAction= setting has been added for service
2720 units which may be used to specify an operation to trigger
2721 when a service fails. This works similarly to
2722 StartLimitAction=, but unlike it, controls what is done
2723 immediately rather than only after several attempts to
2724 restart the service in question.
2726 * hostnamed got updated to also expose the kernel name,
2727 release, and version on the bus. This is useful for
2728 executing commands like hostnamectl with the -H switch.
2729 systemd-analyze makes use of this to properly display
2730 details when running non-locally.
2732 * The bootchart tool can now show cgroup information in the
2733 graphs it generates.
2735 * The CFS CPU quota cgroup attribute is now exposed for
2736 services. The new CPUQuota= switch has been added for this
2737 which takes a percentage value. Setting this will have the
2738 result that a service may never get more CPU time than the
2739 specified percentage, even if the machine is otherwise idle.
2741 * systemd-networkd learned IPIP and SIT tunnel support.
2743 * LSB init scripts exposing a dependency on $network will now
2744 get a dependency on network-online.target rather than simply
2745 network.target. This should bring LSB handling closer to
2746 what it was on SysV systems.
2748 * A new fsck.repair= kernel option has been added to control
2749 how fsck shall deal with unclean file systems at boot.
2751 * The (.ini) configuration file parser will now silently
2752 ignore sections whose name begins with "X-". This may be
2753 used to maintain application-specific extension sections in unit
2756 * machined gained a new API to query the IP addresses of
2757 registered containers. "machinectl status" has been updated
2758 to show these addresses in its output.
2760 * A new call sd_uid_get_display() has been added to the
2761 sd-login APIs for querying the "primary" session of a
2762 user. The "primary" session of the user is elected from the
2763 user's sessions and generally a graphical session is
2764 preferred over a text one.
2766 * A minimal systemd-resolved daemon has been added. It
2767 currently simply acts as a companion to systemd-networkd and
2768 manages resolv.conf based on per-interface DNS
2769 configuration, possibly supplied via DHCP. In the long run
2770 we hope to extend this into a local DNSSEC enabled DNS and
2773 * The systemd-networkd-wait-online tool is now enabled by
2774 default. It will delay network-online.target until a network
2775 connection has been configured. The tool primarily integrates
2776 with networkd, but will also make a best effort to make sense
2777 of network configuration performed in some other way.
2779 * Two new service options StartupCPUShares= and
2780 StartupBlockIOWeight= have been added that work similarly to
2781 CPUShares= and BlockIOWeight= however only apply during
2782 system startup. This is useful to prioritize certain services
2783 differently during bootup than during normal runtime.
2785 * hostnamed has been changed to prefer the statically
2786 configured hostname in /etc/hostname (unless set to
2787 'localhost' or empty) over any dynamic one supplied by
2788 dhcp. With this change, the rules for picking the hostname
2789 match more closely the rules of other configuration settings
2790 where the local administrator's configuration in /etc always
2791 overrides any other settings.
2793 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2794 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2795 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2796 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2797 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2798 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2799 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2800 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2801 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2802 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2803 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2804 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2805 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2806 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2807 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2808 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2811 -- Beijing, 2014-05-28
2815 * When restoring the screen brightness at boot, stay away from
2816 the darkest setting or from the lowest 5% of the available
2817 range, depending on which is the larger value of both. This
2818 should effectively protect the user from rebooting into a
2819 black screen, should the brightness have been set to minimum
2822 * sd-login gained a new sd_machine_get_class() call to
2823 determine the class ("vm" or "container") of a machine
2824 registered with machined.
2826 * sd-login gained new calls
2827 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2828 to query the identity of the peer of a local AF_UNIX
2829 connection. They operate similarly to their sd_pid_get_xyz()
2832 * PID 1 will now maintain a system-wide system state engine
2833 with the states "starting", "running", "degraded",
2834 "maintenance", "stopping". These states are bound to system
2835 startup, normal runtime, runtime with at least one failed
2836 service, rescue/emergency mode and system shutdown. This
2837 state is shown in the "systemctl status" output when no unit
2838 name is passed. It is useful to determine system state, in
2839 particularly when doing so for many systems or containers at
2842 * A new command "list-machines" has been added to "systemctl"
2843 that lists all local OS containers and shows their system
2844 state (see above), if systemd runs inside of them.
2846 * systemctl gained a new "-r" switch to recursively enumerate
2847 units on all local containers, when used with the
2848 "list-unit" command (which is the default one that is
2849 executed when no parameters are specified).
2851 * The GPT automatic partition discovery logic will now honour
2852 two GPT partition flags: one may be set on a partition to
2853 cause it to be mounted read-only, and the other may be set
2854 on a partition to ignore it during automatic discovery.
2856 * Two new GPT type UUIDs have been added for automatic root
2857 partition discovery, for 32-bit and 64-bit ARM. This is not
2858 particularly useful for discovering the root directory on
2859 these architectures during bare-metal boots (since UEFI is
2860 not common there), but still very useful to allow booting of
2861 ARM disk images in nspawn with the -i option.
2863 * MAC addresses of interfaces created with nspawn's
2864 --network-interface= switch will now be generated from the
2865 machine name, and thus be stable between multiple invocations
2868 * logind will now automatically remove all IPC objects owned
2869 by a user if she or he fully logs out. This makes sure that
2870 users who are logged out cannot continue to consume IPC
2871 resources. This covers SysV memory, semaphores and message
2872 queues as well as POSIX shared memory and message
2873 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2874 limits. With this functionality, that is corrected. This may
2875 be turned off by using the RemoveIPC= switch of logind.conf.
2877 * The systemd-machine-id-setup and tmpfiles tools gained a
2878 --root= switch to operate on a specific root directory,
2881 * journald can now forward logged messages to the TTYs of all
2882 logged in users ("wall"). This is the default for all
2883 emergency messages now.
2885 * A new tool systemd-journal-remote has been added to stream
2886 journal log messages across the network.
2888 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2889 controller trees are mounted into it. Note that the
2890 directories mounted beneath it are not read-only. This is a
2891 security measure and is particularly useful because glibc
2892 actually includes a search logic to pick any tmpfs it can
2893 find to implement shm_open() if /dev/shm is not available
2894 (which it might very well be in namespaced setups).
2896 * machinectl gained a new "poweroff" command to cleanly power
2897 down a local OS container.
2899 * The PrivateDevices= unit file setting will now also drop the
2900 CAP_MKNOD capability from the capability bound set, and
2901 imply DevicePolicy=closed.
2903 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2904 comprehensively on all long-running systemd services where
2905 this is appropriate.
2907 * systemd-udevd will now run in a disassociated mount
2908 namespace. To mount directories from udev rules, make sure to
2909 pull in mount units via SYSTEMD_WANTS properties.
2911 * The kdbus support gained support for uploading policy into
2912 the kernel. sd-bus gained support for creating "monitoring"
2913 connections that can eavesdrop into all bus communication
2914 for debugging purposes.
2916 * Timestamps may now be specified in seconds since the UNIX
2917 epoch Jan 1st, 1970 by specifying "@" followed by the value
2920 * Native tcpwrap support in systemd has been removed. tcpwrap
2921 is old code, not really maintained anymore and has serious
2922 shortcomings, and better options such as firewalls
2923 exist. For setups that require tcpwrap usage, please
2924 consider invoking your socket-activated service via tcpd,
2925 like on traditional inetd.
2927 * A new system.conf configuration option
2928 DefaultTimerAccuracySec= has been added that controls the
2929 default AccuracySec= setting of .timer units.
2931 * Timer units gained a new WakeSystem= switch. If enabled,
2932 timers configured this way will cause the system to resume
2933 from system suspend (if the system supports that, which most
2936 * Timer units gained a new Persistent= switch. If enabled,
2937 timers configured this way will save to disk when they have
2938 been last triggered. This information is then used on next
2939 reboot to possible execute overdue timer events, that
2940 could not take place because the system was powered off.
2941 This enables simple anacron-like behaviour for timer units.
2943 * systemctl's "list-timers" will now also list the time a
2944 timer unit was last triggered in addition to the next time
2945 it will be triggered.
2947 * systemd-networkd will now assign predictable IPv4LL
2948 addresses to its local interfaces.
2950 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2951 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2952 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2953 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2954 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2955 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2956 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2957 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2960 -- Berlin, 2014-03-25
2964 * A new unit file setting RestrictAddressFamilies= has been
2965 added to restrict which socket address families unit
2966 processes gain access to. This takes address family names
2967 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2968 attack surface of services via exotic protocol stacks. This
2969 is built on seccomp system call filters.
2971 * Two new unit file settings RuntimeDirectory= and
2972 RuntimeDirectoryMode= have been added that may be used to
2973 manage a per-daemon runtime directories below /run. This is
2974 an alternative for setting up directory permissions with
2975 tmpfiles snippets, and has the advantage that the runtime
2976 directory's lifetime is bound to the daemon runtime and that
2977 the daemon starts up with an empty directory each time. This
2978 is particularly useful when writing services that drop
2979 privileges using the User= or Group= setting.
2981 * The DeviceAllow= unit setting now supports globbing for
2982 matching against device group names.
2984 * The systemd configuration file system.conf gained new
2985 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2986 DefaultMemoryAccounting= to globally turn on/off accounting
2987 for specific resources (cgroups) for all units. These
2988 settings may still be overridden individually in each unit
2991 * systemd-gpt-auto-generator is now able to discover /srv and
2992 root partitions in addition to /home and swap partitions. It
2993 also supports LUKS-encrypted partitions now. With this in
2994 place, automatic discovery of partitions to mount following
2995 the Discoverable Partitions Specification
2996 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2997 is now a lot more complete. This allows booting without
2998 /etc/fstab and without root= on the kernel command line on
2999 systems prepared appropriately.
3001 * systemd-nspawn gained a new --image= switch which allows
3002 booting up disk images and Linux installations on any block
3003 device that follow the Discoverable Partitions Specification
3004 (see above). This means that installations made with
3005 appropriately updated installers may now be started and
3006 deployed using container managers, completely
3007 unmodified. (We hope that libvirt-lxc will add support for
3008 this feature soon, too.)
3010 * systemd-nspawn gained a new --network-macvlan= setting to
3011 set up a private macvlan interface for the
3012 container. Similarly, systemd-networkd gained a new
3013 Kind=macvlan setting in .netdev files.
3015 * systemd-networkd now supports configuring local addresses
3018 * A new tool systemd-network-wait-online has been added to
3019 synchronously wait for network connectivity using
3022 * The sd-bus.h bus API gained a new sd_bus_track object for
3023 tracking the life-cycle of bus peers. Note that sd-bus.h is
3024 still not a public API though (unless you specify
3025 --enable-kdbus on the configure command line, which however
3026 voids your warranty and you get no API stability guarantee).
3028 * The $XDG_RUNTIME_DIR runtime directories for each user are
3029 now individual tmpfs instances, which has the benefit of
3030 introducing separate pools for each user, with individual
3031 size limits, and thus making sure that unprivileged clients
3032 can no longer negatively impact the system or other users by
3033 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3034 RuntimeDirectorySize= has been introduced that allows
3035 controlling the default size limit for all users. It
3036 defaults to 10% of the available physical memory. This is no
3037 replacement for quotas on tmpfs though (which the kernel
3038 still does not support), as /dev/shm and /tmp are still
3039 shared resources used by both the system and unprivileged
3042 * logind will now automatically turn off automatic suspending
3043 on laptop lid close when more than one display is
3044 connected. This was previously expected to be implemented
3045 individually in desktop environments (such as GNOME),
3046 however has been added to logind now, in order to fix a
3047 boot-time race where a desktop environment might not have
3048 been started yet and thus not been able to take an inhibitor
3049 lock at the time where logind already suspends the system
3050 due to a closed lid.
3052 * logind will now wait at least 30s after each system
3053 suspend/resume cycle, and 3min after system boot before
3054 suspending the system due to a closed laptop lid. This
3055 should give USB docking stations and similar enough time to
3056 be probed and configured after system resume and boot in
3057 order to then act as suspend blocker.
3059 * systemd-run gained a new --property= setting which allows
3060 initialization of resource control properties (and others)
3061 for the created scope or service unit. Example: "systemd-run
3062 --property=BlockIOWeight=10 updatedb" may be used to run
3063 updatedb at a low block IO scheduling weight.
3065 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3066 now also work in --scope mode.
3068 * When systemd is compiled with kdbus support, basic support
3069 for enforced policies is now in place. (Note that enabling
3070 kdbus still voids your warranty and no API compatibility
3073 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3074 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3075 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3076 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3077 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3078 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3079 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3080 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3081 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3082 Zbigniew Jędrzejewski-Szmek
3084 -- Berlin, 2014-03-12
3088 * systemd will now relabel /dev after loading the SMACK policy
3089 according to SMACK rules.
3091 * A new unit file option AppArmorProfile= has been added to
3092 set the AppArmor profile for the processes of a unit.
3094 * A new condition check ConditionArchitecture= has been added
3095 to conditionalize units based on the system architecture, as
3096 reported by uname()'s "machine" field.
3098 * systemd-networkd now supports matching on the system
3099 virtualization, architecture, kernel command line, host name
3102 * logind is now a lot more aggressive when suspending the
3103 machine due to a closed laptop lid. Instead of acting only
3104 on the lid close action, it will continuously watch the lid
3105 status and act on it. This is useful for laptops where the
3106 power button is on the outside of the chassis so that it can
3107 be reached without opening the lid (such as the Lenovo
3108 Yoga). On those machines, logind will now immediately
3109 re-suspend the machine if the power button has been
3110 accidentally pressed while the laptop was suspended and in a
3111 backpack or similar.
3113 * logind will now watch SW_DOCK switches and inhibit reaction
3114 to the lid switch if it is pressed. This means that logind
3115 will not suspend the machine anymore if the lid is closed
3116 and the system is docked, if the laptop supports SW_DOCK
3117 notifications via the input layer. Note that ACPI docking
3118 stations do not generate this currently. Also note that this
3119 logic is usually not fully sufficient and Desktop
3120 Environments should take a lid switch inhibitor lock when an
3121 external display is connected, as systemd will not watch
3124 * nspawn will now make use of the devices cgroup controller by
3125 default, and only permit creation of and access to the usual
3126 API device nodes like /dev/null or /dev/random, as well as
3127 access to (but not creation of) the pty devices.
3129 * We will now ship a default .network file for
3130 systemd-networkd that automatically configures DHCP for
3131 network interfaces created by nspawn's --network-veth or
3132 --network-bridge= switches.
3134 * systemd will now understand the usual M, K, G, T suffixes
3135 according to SI conventions (i.e. to the base 1000) when
3136 referring to throughput and hardware metrics. It will stay
3137 with IEC conventions (i.e. to the base 1024) for software
3138 metrics, according to what is customary according to
3139 Wikipedia. We explicitly document which base applies for
3140 each configuration option.
3142 * The DeviceAllow= setting in unit files now supports a syntax
3143 to whitelist an entire group of devices node majors at once,
3144 based on the /proc/devices listing. For example, with the
3145 string "char-pts", it is now possible to whitelist all
3146 current and future pseudo-TTYs at once.
3148 * sd-event learned a new "post" event source. Event sources of
3149 this type are triggered by the dispatching of any event
3150 source of a type that is not "post". This is useful for
3151 implementing clean-up and check event sources that are
3152 triggered by other work being done in the program.
3154 * systemd-networkd is no longer statically enabled, but uses
3155 the usual [Install] sections so that it can be
3156 enabled/disabled using systemctl. It still is enabled by
3159 * When creating a veth interface pair with systemd-nspawn, the
3160 host side will now be prefixed with "vb-" if
3161 --network-bridge= is used, and with "ve-" if --network-veth
3162 is used. This way, it is easy to distinguish these cases on
3163 the host, for example to apply different configuration to
3164 them with systemd-networkd.
3166 * The compatibility libraries for libsystemd-journal.so,
3167 libsystem-id128.so, libsystemd-login.so and
3168 libsystemd-daemon.so do not make use of IFUNC
3169 anymore. Instead, we now build libsystemd.so multiple times
3170 under these alternative names. This means that the footprint
3171 is drastically increased, but given that these are
3172 transitional compatibility libraries, this should not matter
3173 much. This change has been made necessary to support the ARM
3174 platform for these compatibility libraries, as the ARM
3175 toolchain is not really at the same level as the toolchain
3176 for other architectures like x86 and does not support
3177 IFUNC. Please make sure to use --enable-compat-libs only
3178 during a transitional period!
3180 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3181 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3182 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3183 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3184 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3185 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3186 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3187 Zbigniew Jędrzejewski-Szmek
3189 -- Berlin, 2014-02-24
3193 * A new component "systemd-networkd" has been added that can
3194 be used to configure local network interfaces statically or
3195 via DHCP. It is capable of bringing up bridges, VLANs, and
3196 bonding. Currently, no hook-ups for interactive network
3197 configuration are provided. Use this for your initrd,
3198 container, embedded, or server setup if you need a simple,
3199 yet powerful, network configuration solution. This
3200 configuration subsystem is quite nifty, as it allows wildcard
3201 hotplug matching in interfaces. For example, with a single
3202 configuration snippet, you can configure that all Ethernet
3203 interfaces showing up are automatically added to a bridge,
3204 or similar. It supports link-sensing and more.
3206 * A new tool "systemd-socket-proxyd" has been added which can
3207 act as a bidirectional proxy for TCP sockets. This is
3208 useful for adding socket activation support to services that
3209 do not actually support socket activation, including virtual
3210 machines and the like.
3212 * Add a new tool to save/restore rfkill state on
3215 * Save/restore state of keyboard backlights in addition to
3216 display backlights on shutdown/boot.
3218 * udev learned a new SECLABEL{} construct to label device
3219 nodes with a specific security label when they appear. For
3220 now, only SECLABEL{selinux} is supported, but the syntax is
3221 prepared for additional security frameworks.
3223 * udev gained a new scheme to configure link-level attributes
3224 from files in /etc/systemd/network/*.link. These files can
3225 match against MAC address, device path, driver name and type,
3226 and will apply attributes like the naming policy, link speed,
3227 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3228 address assignment policy (randomized, ...).
3230 * The configuration of network interface naming rules for
3231 "permanent interface names" has changed: a new NamePolicy=
3232 setting in the [Link] section of .link files determines the
3233 priority of possible naming schemes (onboard, slot, MAC,
3234 path). The default value of this setting is determined by
3235 /usr/lib/net/links/99-default.link. Old
3236 80-net-name-slot.rules udev configuration file has been
3237 removed, so local configuration overriding this file should
3238 be adapated to override 99-default.link instead.
3240 * When the User= switch is used in a unit file, also
3241 initialize $SHELL= based on the user database entry.
3243 * systemd no longer depends on libdbus. All communication is
3244 now done with sd-bus, systemd's low-level bus library
3247 * kdbus support has been added to PID 1 itself. When kdbus is
3248 enabled, this causes PID 1 to set up the system bus and
3249 enable support for a new ".busname" unit type that
3250 encapsulates bus name activation on kdbus. It works a little
3251 bit like ".socket" units, except for bus names. A new
3252 generator has been added that converts classic dbus1 service
3253 activation files automatically into native systemd .busname
3256 * sd-bus: add a light-weight vtable implementation that allows
3257 defining objects on the bus with a simple static const
3258 vtable array of its methods, signals and properties.
3260 * systemd will not generate or install static dbus
3261 introspection data anymore to /usr/share/dbus-1/interfaces,
3262 as the precise format of these files is unclear, and
3263 nothing makes use of it.
3265 * A proxy daemon is now provided to proxy clients connecting
3266 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3267 compatibility with classic D-Bus.
3269 * A bus driver implementation has been added that supports the
3270 classic D-Bus bus driver calls on kdbus, also for
3271 compatibility purposes.
3273 * A new API "sd-event.h" has been added that implements a
3274 minimal event loop API built around epoll. It provides a
3275 couple of features that direct epoll usage is lacking:
3276 prioritization of events, scales to large numbers of timer
3277 events, per-event timer slack (accuracy), system-wide
3278 coalescing of timer events, exit handlers, watchdog
3279 supervision support using systemd's sd_notify() API, child
3282 * A new API "sd-rntl.h" has been added that provides an API
3283 around the route netlink interface of the kernel, similar in
3284 style to "sd-bus.h".
3286 * A new API "sd-dhcp-client.h" has been added that provides a
3287 small DHCPv4 client-side implementation. This is used by
3290 * There is a new kernel command line option
3291 "systemd.restore_state=0|1". When set to "0", none of the
3292 systemd tools will restore saved runtime state to hardware
3293 devices. More specifically, the rfkill and backlight states
3296 * The FsckPassNo= compatibility option in mount/service units
3297 has been removed. The fstab generator will now add the
3298 necessary dependencies automatically, and does not require
3299 PID1's support for that anymore.
3301 * journalctl gained a new switch, --list-boots, that lists
3302 recent boots with their times and boot IDs.
3304 * The various tools like systemctl, loginctl, timedatectl,
3305 busctl, systemd-run, ... have gained a new switch "-M" to
3306 connect to a specific, local OS container (as direct
3307 connection, without requiring SSH). This works on any
3308 container that is registered with machined, such as those
3309 created by libvirt-lxc or nspawn.
3311 * systemd-run and systemd-analyze also gained support for "-H"
3312 to connect to remote hosts via SSH. This is particularly
3313 useful for systemd-run because it enables queuing of jobs
3314 onto remote systems.
3316 * machinectl gained a new command "login" to open a getty
3317 login in any local container. This works with any container
3318 that is registered with machined (such as those created by
3319 libvirt-lxc or nspawn), and which runs systemd inside.
3321 * machinectl gained a new "reboot" command that may be used to
3322 trigger a reboot on a specific container that is registered
3323 with machined. This works on any container that runs an init
3324 system of some kind.
3326 * systemctl gained a new "list-timers" command to print a nice
3327 listing of installed timer units with the times they elapse
3330 * Alternative reboot() parameters may now be specified on the
3331 "systemctl reboot" command line and are passed to the
3332 reboot() system call.
3334 * systemctl gained a new --job-mode= switch to configure the
3335 mode to queue a job with. This is a more generic version of
3336 --fail, --irreversible, and --ignore-dependencies, which are
3337 still available but not advertised anymore.
3339 * /etc/systemd/system.conf gained new settings to configure
3340 various default timeouts of units, as well as the default
3341 start limit interval and burst. These may still be overridden
3344 * PID1 will now export on the bus profile data of the security
3345 policy upload process (such as the SELinux policy upload to
3348 * journald: when forwarding logs to the console, include
3349 timestamps (following the setting in
3350 /sys/module/printk/parameters/time).
3352 * OnCalendar= in timer units now understands the special
3353 strings "yearly" and "annually". (Both are equivalent)
3355 * The accuracy of timer units is now configurable with the new
3356 AccuracySec= setting. It defaults to 1min.
3358 * A new dependency type JoinsNamespaceOf= has been added that
3359 allows running two services within the same /tmp and network
3360 namespace, if PrivateNetwork= or PrivateTmp= are used.
3362 * A new command "cat" has been added to systemctl. It outputs
3363 the original unit file of a unit, and concatenates the
3364 contents of additional "drop-in" unit file snippets, so that
3365 the full configuration is shown.
3367 * systemctl now supports globbing on the various "list-xyz"
3368 commands, like "list-units" or "list-sockets", as well as on
3369 those commands which take multiple unit names.
3371 * journalctl's --unit= switch gained support for globbing.
3373 * All systemd daemons now make use of the watchdog logic so
3374 that systemd automatically notices when they hang.
3376 * If the $container_ttys environment variable is set,
3377 getty-generator will automatically spawn a getty for each
3378 listed tty. This is useful for container managers to request
3379 login gettys to be spawned on as many ttys as needed.
3381 * %h, %s, %U specifier support is not available anymore when
3382 used in unit files for PID 1. This is because NSS calls are
3383 not safe from PID 1. They stay available for --user
3384 instances of systemd, and as special case for the root user.
3386 * loginctl gained a new "--no-legend" switch to turn off output
3389 * The "sd-login.h" API gained three new calls:
3390 sd_session_is_remote(), sd_session_get_remote_user(),
3391 sd_session_get_remote_host() to query information about
3394 * The udev hardware database now also carries vendor/product
3395 information of SDIO devices.
3397 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3398 determine whether watchdog notifications are requested by
3401 * Socket-activated per-connection services now include a
3402 short description of the connection parameters in the
3405 * tmpfiles gained a new "--boot" option. When this is not used,
3406 only lines where the command character is not suffixed with
3407 "!" are executed. When this option is specified, those
3408 options are executed too. This partitions tmpfiles
3409 directives into those that can be safely executed at any
3410 time, and those which should be run only at boot (for
3411 example, a line that creates /run/nologin).
3413 * A new API "sd-resolve.h" has been added which provides a simple
3414 asynchronous wrapper around glibc NSS host name resolution
3415 calls, such as getaddrinfo(). In contrast to glibc's
3416 getaddrinfo_a(), it does not use signals. In contrast to most
3417 other asynchronous name resolution libraries, this one does
3418 not reimplement DNS, but reuses NSS, so that alternate
3419 host name resolution systems continue to work, such as mDNS,
3420 LDAP, etc. This API is based on libasyncns, but it has been
3421 cleaned up for inclusion in systemd.
3423 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3424 "sd-daemon.h" are no longer found in individual libraries
3425 libsystemd-journal.so, libsystemd-login.so,
3426 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3427 merged them into a single library, libsystemd.so, which
3428 provides all symbols. The reason for this is cyclic
3429 dependencies, as these libraries tend to use each other's
3430 symbols. So far, we have managed to workaround that by linking
3431 a copy of a good part of our code into each of these
3432 libraries again and again, which, however, makes certain
3433 things hard to do, like sharing static variables. Also, it
3434 substantially increases footprint. With this change, there
3435 is only one library for the basic APIs systemd
3436 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3437 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3438 library as well, however are subject to the --enable-kdbus
3439 switch (see below). Note that "sd-dhcp-client.h" is not part
3440 of this library (this is because it only consumes, never
3441 provides, services of/to other APIs). To make the transition
3442 easy from the separate libraries to the unified one, we
3443 provide the --enable-compat-libs compile-time switch which
3444 will generate stub libraries that are compatible with the
3445 old ones but redirect all calls to the new one.
3447 * All of the kdbus logic and the new APIs "sd-bus.h",
3448 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3449 and "sd-utf8.h" are compile-time optional via the
3450 "--enable-kdbus" switch, and they are not compiled in by
3451 default. To make use of kdbus, you have to explicitly enable
3452 the switch. Note however, that neither the kernel nor the
3453 userspace API for all of this is considered stable yet. We
3454 want to maintain the freedom to still change the APIs for
3455 now. By specifying this build-time switch, you acknowledge
3456 that you are aware of the instability of the current
3459 * Also, note that while kdbus is pretty much complete,
3460 it lacks one thing: proper policy support. This means you
3461 can build a fully working system with all features; however,
3462 it will be highly insecure. Policy support will be added in
3463 one of the next releases, at the same time that we will
3464 declare the APIs stable.
3466 * When the kernel command line argument "kdbus" is specified,
3467 systemd will automatically load the kdbus.ko kernel module. At
3468 this stage of development, it is only useful for testing kdbus
3469 and should not be used in production. Note: if "--enable-kdbus"
3470 is specified, and the kdbus.ko kernel module is available, and
3471 "kdbus" is added to the kernel command line, the entire system
3472 runs with kdbus instead of dbus-daemon, with the above mentioned
3473 problem of missing the system policy enforcement. Also a future
3474 version of kdbus.ko or a newer systemd will not be compatible with
3475 each other, and will unlikely be able to boot the machine if only
3476 one of them is updated.
3478 * systemctl gained a new "import-environment" command which
3479 uploads the caller's environment (or parts thereof) into the
3480 service manager so that it is inherited by services started
3481 by the manager. This is useful to upload variables like
3482 $DISPLAY into the user service manager.
3484 * A new PrivateDevices= switch has been added to service units
3485 which allows running a service with a namespaced /dev
3486 directory that does not contain any device nodes for
3487 physical devices. More specifically, it only includes devices
3488 such as /dev/null, /dev/urandom, and /dev/zero which are API
3491 * logind has been extended to support behaviour like VT
3492 switching on seats that do not support a VT. This makes
3493 multi-session available on seats that are not the first seat
3494 (seat0), and on systems where kernel support for VTs has
3495 been disabled at compile-time.
3497 * If a process holds a delay lock for system sleep or shutdown
3498 and fails to release it in time, we will now log its
3499 identity. This makes it easier to identify processes that
3500 cause slow suspends or power-offs.
3502 * When parsing /etc/crypttab, support for a new key-slot=
3503 option as supported by Debian is added. It allows indicating
3504 which LUKS slot to use on disk, speeding up key loading.
3506 * The sd_journald_sendv() API call has been checked and
3507 officially declared to be async-signal-safe so that it may
3508 be invoked from signal handlers for logging purposes.
3510 * Boot-time status output is now enabled automatically after a
3511 short timeout if boot does not progress, in order to give
3512 the user an indication what she or he is waiting for.
3514 * The boot-time output has been improved to show how much time
3515 remains until jobs expire.
3517 * The KillMode= switch in service units gained a new possible
3518 value "mixed". If set, and the unit is shut down, then the
3519 initial SIGTERM signal is sent only to the main daemon
3520 process, while the following SIGKILL signal is sent to
3521 all remaining processes of the service.
3523 * When a scope unit is registered, a new property "Controller"
3524 may be set. If set to a valid bus name, systemd will send a
3525 RequestStop() signal to this name when it would like to shut
3526 down the scope. This may be used to hook manager logic into
3527 the shutdown logic of scope units. Also, scope units may now
3528 be put in a special "abandoned" state, in which case the
3529 manager process which created them takes no further
3530 responsibilities for it.
3532 * When reading unit files, systemd will now verify
3533 the access mode of these files, and warn about certain
3534 suspicious combinations. This has been added to make it
3535 easier to track down packaging bugs where unit files are
3536 marked executable or world-writable.
3538 * systemd-nspawn gained a new "--setenv=" switch to set
3539 container-wide environment variables. The similar option in
3540 systemd-activate was renamed from "--environment=" to
3541 "--setenv=" for consistency.
3543 * systemd-nspawn has been updated to create a new kdbus domain
3544 for each container that is invoked, thus allowing each
3545 container to have its own set of system and user buses,
3546 independent of the host.
3548 * systemd-nspawn gained a new --drop-capability= switch to run
3549 the container with less capabilities than the default. Both
3550 --drop-capability= and --capability= now take the special
3551 string "all" for dropping or keeping all capabilities.
3553 * systemd-nspawn gained new switches for executing containers
3554 with specific SELinux labels set.
3556 * systemd-nspawn gained a new --quiet switch to not generate
3557 any additional output but the container's own console
3560 * systemd-nspawn gained a new --share-system switch to run a
3561 container without PID namespacing enabled.
3563 * systemd-nspawn gained a new --register= switch to control
3564 whether the container is registered with systemd-machined or
3565 not. This is useful for containers that do not run full
3566 OS images, but only specific apps.
3568 * systemd-nspawn gained a new --keep-unit which may be used
3569 when invoked as the only program from a service unit, and
3570 results in registration of the unit service itself in
3571 systemd-machined, instead of a newly opened scope unit.
3573 * systemd-nspawn gained a new --network-interface= switch for
3574 moving arbitrary interfaces to the container. The new
3575 --network-veth switch creates a virtual Ethernet connection
3576 between host and container. The new --network-bridge=
3577 switch then allows assigning the host side of this virtual
3578 Ethernet connection to a bridge device.
3580 * systemd-nspawn gained a new --personality= switch for
3581 setting the kernel personality for the container. This is
3582 useful when running a 32-bit container on a 64-bit host. A
3583 similar option Personality= is now also available for service
3586 * logind will now also track a "Desktop" identifier for each
3587 session which encodes the desktop environment of it. This is
3588 useful for desktop environments that want to identify
3589 multiple running sessions of itself easily.
3591 * A new SELinuxContext= setting for service units has been
3592 added that allows setting a specific SELinux execution
3593 context for a service.
3595 * Most systemd client tools will now honour $SYSTEMD_LESS for
3596 settings of the "less" pager. By default, these tools will
3597 override $LESS to allow certain operations to work, such as
3598 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3599 influence this logic.
3601 * systemd's "seccomp" hook-up has been changed to make use of
3602 the libseccomp library instead of using its own
3603 implementation. This has benefits for portability among
3606 * For usage together with SystemCallFilter=, a new
3607 SystemCallErrorNumber= setting has been introduced that
3608 allows configuration of a system error number to be returned
3609 on filtered system calls, instead of immediately killing the
3610 process. Also, SystemCallArchitectures= has been added to
3611 limit access to system calls of a particular architecture
3612 (in order to turn off support for unused secondary
3613 architectures). There is also a global
3614 SystemCallArchitectures= setting in system.conf now to turn
3615 off support for non-native system calls system-wide.
3617 * systemd requires a kernel with a working name_to_handle_at(),
3618 please see the kernel config requirements in the README file.
3620 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3621 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3622 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3623 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3624 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3625 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3626 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3627 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3628 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3629 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3630 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3631 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3632 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3633 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3634 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3635 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3636 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3637 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3638 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3639 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3640 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3641 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3642 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3643 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3645 -- Berlin, 2014-02-20
3649 * logind has gained support for facilitating privileged input
3650 and drm device access for unprivileged clients. This work is
3651 useful to allow Wayland display servers (and similar
3652 programs, such as kmscon) to run under the user's ID and
3653 access input and drm devices which are normally
3654 protected. When this is used (and the kernel is new enough)
3655 logind will "mute" IO on the file descriptors passed to
3656 Wayland as long as it is in the background and "unmute" it
3657 if it returns into the foreground. This allows secure
3658 session switching without allowing background sessions to
3659 eavesdrop on input and display data. This also introduces
3660 session switching support if VT support is turned off in the
3661 kernel, and on seats that are not seat0.
3663 * A new kernel command line option luks.options= is understood
3664 now which allows specifying LUKS options for usage for LUKS
3665 encrypted partitions specified with luks.uuid=.
3667 * tmpfiles.d(5) snippets may now use specifier expansion in
3668 path names. More specifically %m, %b, %H, %v, are now
3669 replaced by the local machine id, boot id, hostname, and
3670 kernel version number.
3672 * A new tmpfiles.d(5) command "m" has been introduced which
3673 may be used to change the owner/group/access mode of a file
3674 or directory if it exists, but do nothing if it does not.
3676 * This release removes high-level support for the
3677 MemorySoftLimit= cgroup setting. The underlying kernel
3678 cgroup attribute memory.soft_limit= is currently badly
3679 designed and likely to be removed from the kernel API in its
3680 current form, hence we should not expose it for now.
3682 * The memory.use_hierarchy cgroup attribute is now enabled for
3683 all cgroups systemd creates in the memory cgroup
3684 hierarchy. This option is likely to be come the built-in
3685 default in the kernel anyway, and the non-hierarchical mode
3686 never made much sense in the intrinsically hierarchical
3689 * A new field _SYSTEMD_SLICE= is logged along with all journal
3690 messages containing the slice a message was generated
3691 from. This is useful to allow easy per-customer filtering of
3692 logs among other things.
3694 * systemd-journald will no longer adjust the group of journal
3695 files it creates to the "systemd-journal" group. Instead we
3696 rely on the journal directory to be owned by the
3697 "systemd-journal" group, and its setgid bit set, so that the
3698 kernel file system layer will automatically enforce that
3699 journal files inherit this group assignment. The reason for
3700 this change is that we cannot allow NSS look-ups from
3701 journald which would be necessary to resolve
3702 "systemd-journal" to a numeric GID, because this might
3703 create deadlocks if NSS involves synchronous queries to
3704 other daemons (such as nscd, or sssd) which in turn are
3705 logging clients of journald and might block on it, which
3706 would then dead lock. A tmpfiles.d(5) snippet included in
3707 systemd will make sure the setgid bit and group are
3708 properly set on the journal directory if it exists on every
3709 boot. However, we recommend adjusting it manually after
3710 upgrades too (or from RPM scriptlets), so that the change is
3711 not delayed until next reboot.
3713 * Backlight and random seed files in /var/lib/ have moved into
3714 the /var/lib/systemd/ directory, in order to centralize all
3715 systemd generated files in one directory.
3717 * Boot time performance measurements (as displayed by
3718 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3719 performance information if that's available to determine how
3720 much time BIOS and boot loader initialization required. With
3721 a sufficiently new BIOS you hence no longer need to boot
3722 with Gummiboot to get access to such information.
3724 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3725 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3726 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3727 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3728 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3729 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3730 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3732 -- Berlin, 2013-10-02
3736 * The Restart= option for services now understands a new
3737 on-watchdog setting, which will restart the service
3738 automatically if the service stops sending out watchdog keep
3739 alive messages (as configured with WatchdogSec=).
3741 * The getty generator (which is responsible for bringing up a
3742 getty on configured serial consoles) will no longer only
3743 start a getty on the primary kernel console but on all
3744 others, too. This makes the order in which console= is
3745 specified on the kernel command line less important.
3747 * libsystemd-logind gained a new sd_session_get_vt() call to
3748 retrieve the VT number of a session.
3750 * If the option "tries=0" is set for an entry of /etc/crypttab
3751 its passphrase is queried indefinitely instead of any
3752 maximum number of tries.
3754 * If a service with a configure PID file terminates its PID
3755 file will now be removed automatically if it still exists
3756 afterwards. This should put an end to stale PID files.
3758 * systemd-run will now also take relative binary path names
3759 for execution and no longer insists on absolute paths.
3761 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3762 paths that are optionally prefixed with "-" to indicate that
3763 it should not be considered a failure if they do not exist.
3765 * journalctl -o (and similar commands) now understands a new
3766 output mode "short-precise", it is similar to "short" but
3767 shows timestamps with usec accuracy.
3769 * The option "discard" (as known from Debian) is now
3770 synonymous to "allow-discards" in /etc/crypttab. In fact,
3771 "discard" is preferred now (since it is easier to remember
3774 * Some licensing clean-ups were made, so that more code is now
3775 LGPL-2.1 licensed than before.
3777 * A minimal tool to save/restore the display backlight
3778 brightness across reboots has been added. It will store the
3779 backlight setting as late as possible at shutdown, and
3780 restore it as early as possible during reboot.
3782 * A logic to automatically discover and enable home and swap
3783 partitions on GPT disks has been added. With this in place
3784 /etc/fstab becomes optional for many setups as systemd can
3785 discover certain partitions located on the root disk
3786 automatically. Home partitions are recognized under their
3787 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3788 partitions are recognized under their GPT type ID
3789 0657fd6da4ab43c484e50933c84b4f4f.
3791 * systemd will no longer pass any environment from the kernel
3792 or initrd to system services. If you want to set an
3793 environment for all services, do so via the kernel command
3794 line systemd.setenv= assignment.
3796 * The systemd-sysctl tool no longer natively reads the file
3797 /etc/sysctl.conf. If desired, the file should be symlinked
3798 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3799 legacy support by a symlink rather than built-in code, it
3800 also makes the otherwise hidden order of application of the
3801 different files visible. (Note that this partly reverts to a
3802 pre-198 application order of sysctl knobs!)
3804 * The "systemctl set-log-level" and "systemctl dump" commands
3805 have been moved to systemd-analyze.
3807 * systemd-run learned the new --remain-after-exit switch,
3808 which causes the scope unit not to be cleaned up
3809 automatically after the process terminated.
3811 * tmpfiles learned a new --exclude-prefix= switch to exclude
3812 certain paths from operation.
3814 * journald will now automatically flush all messages to disk
3815 as soon as a message at the log level CRIT, ALERT or EMERG
3818 Contributions from: Andrew Cook, Brandon Philips, Christian
3819 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3820 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3821 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3822 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3823 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3824 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3825 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3826 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3827 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3828 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3829 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3830 William Giokas, Zbigniew Jędrzejewski-Szmek
3832 -- Berlin, 2013-09-13
3836 * The documentation has been updated to cover the various new
3837 concepts introduced with 205.
3839 * Unit files now understand the new %v specifier which
3840 resolves to the kernel version string as returned by "uname
3843 * systemctl now supports filtering the unit list output by
3844 load state, active state and sub state, using the new
3847 * "systemctl status" will now show the results of the
3848 condition checks (like ConditionPathExists= and similar) of
3849 the last start attempts of the unit. They are also logged to
3852 * "journalctl -b" may now be used to look for boot output of a
3853 specific boot. Try "journalctl -b -1" for the previous boot,
3854 but the syntax is substantially more powerful.
3856 * "journalctl --show-cursor" has been added which prints the
3857 cursor string the last shown log line. This may then be used
3858 with the new "journalctl --after-cursor=" switch to continue
3859 browsing logs from that point on.
3861 * "journalctl --force" may now be used to force regeneration
3864 * Creation of "dead" device nodes has been moved from udev
3865 into kmod and tmpfiles. Previously, udev would read the kmod
3866 databases to pre-generate dead device nodes based on meta
3867 information contained in kernel modules, so that these would
3868 be auto-loaded on access rather then at boot. As this
3869 does not really have much to do with the exposing actual
3870 kernel devices to userspace this has always been slightly
3871 alien in the udev codebase. Following the new scheme kmod
3872 will now generate a runtime snippet for tmpfiles from the
3873 module meta information and it now is tmpfiles' job to the
3874 create the nodes. This also allows overriding access and
3875 other parameters for the nodes using the usual tmpfiles
3876 facilities. As side effect this allows us to remove the
3877 CAP_SYS_MKNOD capability bit from udevd entirely.
3879 * logind's device ACLs may now be applied to these "dead"
3880 devices nodes too, thus finally allowing managed access to
3881 devices such as /dev/snd/sequencer whithout loading the
3882 backing module right-away.
3884 * A new RPM macro has been added that may be used to apply
3885 tmpfiles configuration during package installation.
3887 * systemd-detect-virt and ConditionVirtualization= now can
3888 detect User-Mode-Linux machines (UML).
3890 * journald will now implicitly log the effective capabilities
3891 set of processes in the message metadata.
3893 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3895 * The initrd interface has been simplified (more specifically,
3896 support for passing performance data via environment
3897 variables and fsck results via files in /run has been
3898 removed). These features were non-essential, and are
3899 nowadays available in a much nicer way by having systemd in
3900 the initrd serialize its state and have the hosts systemd
3901 deserialize it again.
3903 * The udev "keymap" data files and tools to apply keyboard
3904 specific mappings of scan to key codes, and force-release
3905 scan code lists have been entirely replaced by a udev
3906 "keyboard" builtin and a hwdb data file.
3908 * systemd will now honour the kernel's "quiet" command line
3909 argument also during late shutdown, resulting in a
3910 completely silent shutdown when used.
3912 * There's now an option to control the SO_REUSEPORT socket
3913 option in .socket units.
3915 * Instance units will now automatically get a per-template
3916 subslice of system.slice unless something else is explicitly
3917 configured. For example, instances of sshd@.service will now
3918 implicitly be placed in system-sshd.slice rather than
3919 system.slice as before.
3921 * Test coverage support may now be enabled at build time.
3923 Contributions from: Dave Reisner, Frederic Crozat, Harald
3924 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3925 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3926 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3927 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3928 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3929 Giokas, Zbigniew Jędrzejewski-Szmek
3931 -- Berlin, 2013-07-23
3935 * Two new unit types have been introduced:
3937 Scope units are very similar to service units, however, are
3938 created out of pre-existing processes -- instead of PID 1
3939 forking off the processes. By using scope units it is
3940 possible for system services and applications to group their
3941 own child processes (worker processes) in a powerful way
3942 which then maybe used to organize them, or kill them
3943 together, or apply resource limits on them.
3945 Slice units may be used to partition system resources in an
3946 hierarchical fashion and then assign other units to them. By
3947 default there are now three slices: system.slice (for all
3948 system services), user.slice (for all user sessions),
3949 machine.slice (for VMs and containers).
3951 Slices and scopes have been introduced primarily in
3952 context of the work to move cgroup handling to a
3953 single-writer scheme, where only PID 1
3954 creates/removes/manages cgroups.
3956 * There's a new concept of "transient" units. In contrast to
3957 normal units these units are created via an API at runtime,
3958 not from configuration from disk. More specifically this
3959 means it is now possible to run arbitrary programs as
3960 independent services, with all execution parameters passed
3961 in via bus APIs rather than read from disk. Transient units
3962 make systemd substantially more dynamic then it ever was,
3963 and useful as a general batch manager.
3965 * logind has been updated to make use of scope and slice units
3966 for managing user sessions. As a user logs in he will get
3967 his own private slice unit, to which all sessions are added
3968 as scope units. We also added support for automatically
3969 adding an instance of user@.service for the user into the
3970 slice. Effectively logind will no longer create cgroup
3971 hierarchies on its own now, it will defer entirely to PID 1
3972 for this by means of scope, service and slice units. Since
3973 user sessions this way become entities managed by PID 1
3974 the output of "systemctl" is now a lot more comprehensive.
3976 * A new mini-daemon "systemd-machined" has been added which
3977 may be used by virtualization managers to register local
3978 VMs/containers. nspawn has been updated accordingly, and
3979 libvirt will be updated shortly. machined will collect a bit
3980 of meta information about the VMs/containers, and assign
3981 them their own scope unit (see above). The collected
3982 meta-data is then made available via the "machinectl" tool,
3983 and exposed in "ps" and similar tools. machined/machinectl
3984 is compile-time optional.
3986 * As discussed earlier, the low-level cgroup configuration
3987 options ControlGroup=, ControlGroupModify=,
3988 ControlGroupPersistent=, ControlGroupAttribute= have been
3989 removed. Please use high-level attribute settings instead as
3990 well as slice units.
3992 * A new bus call SetUnitProperties() has been added to alter
3993 various runtime parameters of a unit. This is primarily
3994 useful to alter cgroup parameters dynamically in a nice way,
3995 but will be extended later on to make more properties
3996 modifiable at runtime. systemctl gained a new set-properties
3997 command that wraps this call.
3999 * A new tool "systemd-run" has been added which can be used to
4000 run arbitrary command lines as transient services or scopes,
4001 while configuring a number of settings via the command
4002 line. This tool is currently very basic, however already
4003 very useful. We plan to extend this tool to even allow
4004 queuing of execution jobs with time triggers from the
4005 command line, similar in fashion to "at".
4007 * nspawn will now inform the user explicitly that kernels with
4008 audit enabled break containers, and suggest the user to turn
4011 * Support for detecting the IMA and AppArmor security
4012 frameworks with ConditionSecurity= has been added.
4014 * journalctl gained a new "-k" switch for showing only kernel
4015 messages, mimicking dmesg output; in addition to "--user"
4016 and "--system" switches for showing only user's own logs
4019 * systemd-delta can now show information about drop-in
4020 snippets extending unit files.
4022 * libsystemd-bus has been substantially updated but is still
4023 not available as public API.
4025 * systemd will now look for the "debug" argument on the kernel
4026 command line and enable debug logging, similar to what
4027 "systemd.log_level=debug" already did before.
4029 * "systemctl set-default", "systemctl get-default" has been
4030 added to configure the default.target symlink, which
4031 controls what to boot into by default.
4033 * "systemctl set-log-level" has been added as a convenient
4034 way to raise and lower systemd logging threshold.
4036 * "systemd-analyze plot" will now show the time the various
4037 generators needed for execution, as well as information
4038 about the unit file loading.
4040 * libsystemd-journal gained a new sd_journal_open_files() call
4041 for opening specific journal files. journactl also gained a
4042 new switch to expose this new functionality. Previously we
4043 only supported opening all files from a directory, or all
4044 files from the system, as opening individual files only is
4045 racy due to journal file rotation.
4047 * systemd gained the new DefaultEnvironment= setting in
4048 /etc/systemd/system.conf to set environment variables for
4051 * If a privileged process logs a journal message with the
4052 OBJECT_PID= field set, then journald will automatically
4053 augment this with additional OBJECT_UID=, OBJECT_GID=,
4054 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4055 system services want to log events about specific client
4056 processes. journactl/systemctl has been updated to make use
4057 of this information if all log messages regarding a specific
4060 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4061 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4062 Reisner, David Coppa, David King, David Strauss, Eelco
4063 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4064 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4065 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4066 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4067 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4068 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4069 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4070 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4071 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4072 Łukasz Stelmach, 장동준
4076 * The Python bindings gained some minimal support for the APIs
4077 exposed by libsystemd-logind.
4079 * ConditionSecurity= gained support for detecting SMACK. Since
4080 this condition already supports SELinux and AppArmor we only
4081 miss IMA for this. Patches welcome!
4083 Contributions from: Karol Lewandowski, Lennart Poettering,
4084 Zbigniew Jędrzejewski-Szmek
4088 * systemd-nspawn will now create /etc/resolv.conf if
4089 necessary, before bind-mounting the host's file onto it.
4091 * systemd-nspawn will now store meta information about a
4092 container on the container's cgroup as extended attribute
4093 fields, including the root directory.
4095 * The cgroup hierarchy has been reworked in many ways. All
4096 objects any of the components systemd creates in the cgroup
4097 tree are now suffixed. More specifically, user sessions are
4098 now placed in cgroups suffixed with ".session", users in
4099 cgroups suffixed with ".user", and nspawn containers in
4100 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4101 names are now escaped in a simple scheme to avoid collision
4102 of userspace object names with kernel filenames. This work
4103 is preparation for making these objects relocatable in the
4104 cgroup tree, in order to allow easy resource partitioning of
4105 these objects without causing naming conflicts.
4107 * systemctl list-dependencies gained the new switches
4108 --plain, --reverse, --after and --before.
4110 * systemd-inhibit now shows the process name of processes that
4111 have taken an inhibitor lock.
4113 * nss-myhostname will now also resolve "localhost"
4114 implicitly. This makes /etc/hosts an optional file and
4115 nicely handles that on IPv6 ::1 maps to both "localhost" and
4118 * libsystemd-logind.so gained a new call
4119 sd_get_machine_names() to enumerate running containers and
4120 VMs (currently only supported by very new libvirt and
4121 nspawn). sd_login_monitor can now be used to watch
4122 VMs/containers coming and going.
4124 * .include is not allowed recursively anymore, and only in
4125 unit files. Usually it is better to use drop-in snippets in
4126 .d/*.conf anyway, as introduced with systemd 198.
4128 * systemd-analyze gained a new "critical-chain" command that
4129 determines the slowest chain of units run during system
4130 boot-up. It is very useful for tracking down where
4131 optimizing boot time is the most beneficial.
4133 * systemd will no longer allow manipulating service paths in
4134 the name=systemd:/system cgroup tree using ControlGroup= in
4135 units. (But is still fine with it in all other dirs.)
4137 * There's a new systemd-nspawn@.service service file that may
4138 be used to easily run nspawn containers as system
4139 services. With the container's root directory in
4140 /var/lib/container/foobar it is now sufficient to run
4141 "systemctl start systemd-nspawn@foobar.service" to boot it.
4143 * systemd-cgls gained a new parameter "--machine" to list only
4144 the processes within a certain container.
4146 * ConditionSecurity= now can check for "apparmor". We still
4147 are lacking checks for SMACK and IMA for this condition
4148 check though. Patches welcome!
4150 * A new configuration file /etc/systemd/sleep.conf has been
4151 added that may be used to configure which kernel operation
4152 systemd is supposed to execute when "suspend", "hibernate"
4153 or "hybrid-sleep" is requested. This makes the new kernel
4154 "freeze" state accessible to the user.
4156 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4157 the passed argument if applicable.
4159 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4160 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4161 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4162 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4163 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4164 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4165 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4170 * The output of 'systemctl list-jobs' got some polishing. The
4171 '--type=' argument may now be passed more than once. A new
4172 command 'systemctl list-sockets' has been added which shows
4173 a list of kernel sockets systemd is listening on with the
4174 socket units they belong to, plus the units these socket
4177 * The experimental libsystemd-bus library got substantial
4178 updates to work in conjunction with the (also experimental)
4179 kdbus kernel project. It works well enough to exchange
4180 messages with some sophistication. Note that kdbus is not
4181 ready yet, and the library is mostly an elaborate test case
4182 for now, and not installable.
4184 * systemd gained a new unit 'systemd-static-nodes.service'
4185 that generates static device nodes earlier during boot, and
4186 can run in conjunction with udev.
4188 * libsystemd-login gained a new call sd_pid_get_user_unit()
4189 to retrieve the user systemd unit a process is running
4190 in. This is useful for systems where systemd is used as
4193 * systemd-nspawn now places all containers in the new /machine
4194 top-level cgroup directory in the name=systemd
4195 hierarchy. libvirt will soon do the same, so that we get a
4196 uniform separation of /system, /user and /machine for system
4197 services, user processes and containers/virtual
4198 machines. This new cgroup hierarchy is also useful to stick
4199 stable names to specific container instances, which can be
4200 recognized later this way (this name may be controlled
4201 via systemd-nspawn's new -M switch). libsystemd-login also
4202 gained a new call sd_pid_get_machine_name() to retrieve the
4203 name of the container/VM a specific process belongs to.
4205 * bootchart can now store its data in the journal.
4207 * libsystemd-journal gained a new call
4208 sd_journal_add_conjunction() for AND expressions to the
4209 matching logic. This can be used to express more complex
4210 logical expressions.
4212 * journactl can now take multiple --unit= and --user-unit=
4215 * The cryptsetup logic now understands the "luks.key=" kernel
4216 command line switch for specifying a file to read the
4217 decryption key from. Also, if a configured key file is not
4218 found the tool will now automatically fall back to prompting
4221 * Python systemd.journal module was updated to wrap recently
4222 added functions from libsystemd-journal. The interface was
4223 changed to bring the low level interface in s.j._Reader
4224 closer to the C API, and the high level interface in
4225 s.j.Reader was updated to wrap and convert all data about
4228 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4229 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4230 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4231 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4232 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4233 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4237 * journalctl --update-catalog now understands a new --root=
4238 option to operate on catalogs found in a different root
4241 * During shutdown after systemd has terminated all running
4242 services a final killing loop kills all remaining left-over
4243 processes. We will now print the name of these processes
4244 when we send SIGKILL to them, since this usually indicates a
4247 * If /etc/crypttab refers to password files stored on
4248 configured mount points automatic dependencies will now be
4249 generated to ensure the specific mount is established first
4250 before the key file is attempted to be read.
4252 * 'systemctl status' will now show information about the
4253 network sockets a socket unit is listening on.
4255 * 'systemctl status' will also shown information about any
4256 drop-in configuration file for units. (Drop-In configuration
4257 files in this context are files such as
4258 /etc/systemd/systemd/foobar.service.d/*.conf)
4260 * systemd-cgtop now optionally shows summed up CPU times of
4261 cgroups. Press '%' while running cgtop to switch between
4262 percentage and absolute mode. This is useful to determine
4263 which cgroups use up the most CPU time over the entire
4264 runtime of the system. systemd-cgtop has also been updated
4265 to be 'pipeable' for processing with further shell tools.
4267 * 'hostnamectl set-hostname' will now allow setting of FQDN
4270 * The formatting and parsing of time span values has been
4271 changed. The parser now understands fractional expressions
4272 such as "5.5h". The formatter will now output fractional
4273 expressions for all time spans under 1min, i.e. "5.123456s"
4274 rather than "5s 123ms 456us". For time spans under 1s
4275 millisecond values are shown, for those under 1ms
4276 microsecond values are shown. This should greatly improve
4277 all time-related output of systemd.
4279 * libsystemd-login and libsystemd-journal gained new
4280 functions for querying the poll() events mask and poll()
4281 timeout value for integration into arbitrary event
4284 * localectl gained the ability to list available X11 keymaps
4285 (models, layouts, variants, options).
4287 * 'systemd-analyze dot' gained the ability to filter for
4288 specific units via shell-style globs, to create smaller,
4289 more useful graphs. I.e. it is now possible to create simple
4290 graphs of all the dependencies between only target units, or
4291 of all units that Avahi has dependencies with.
4293 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4294 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4295 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4296 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4297 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4298 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4299 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4303 * The boot-time readahead implementation for rotating media
4304 will now read the read-ahead data in multiple passes which
4305 consist of all read requests made in equidistant time
4306 intervals. This means instead of strictly reading read-ahead
4307 data in its physical order on disk we now try to find a
4308 middle ground between physical and access time order.
4310 * /etc/os-release files gained a new BUILD_ID= field for usage
4311 on operating systems that provide continuous builds of OS
4314 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4315 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4316 William Douglas, Zbigniew Jędrzejewski-Szmek
4320 * systemd-python gained an API exposing libsystemd-daemon.
4322 * The SMACK setup logic gained support for uploading CIPSO
4325 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4326 ReadOnlyDirectories= and InaccessibleDirectories= has
4327 changed. The private /tmp and /var/tmp directories are now
4328 shared by all processes of a service (which means
4329 ExecStartPre= may now leave data in /tmp that ExecStart= of
4330 the same service can still access). When a service is
4331 stopped its temporary directories are immediately deleted
4332 (normal clean-up with tmpfiles is still done in addition to
4335 * By default, systemd will now set a couple of sysctl
4336 variables in the kernel: the safe sysrq options are turned
4337 on, IP route verification is turned on, and source routing
4338 disabled. The recently added hardlink and softlink
4339 protection of the kernel is turned on. These settings should
4340 be reasonably safe, and good defaults for all new systems.
4342 * The predictable network naming logic may now be turned off
4343 with a new kernel command line switch: net.ifnames=0.
4345 * A new libsystemd-bus module has been added that implements a
4346 pretty complete D-Bus client library. For details see:
4348 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4350 * journald will now explicitly flush the journal files to disk
4351 at the latest 5min after each write. The file will then also
4352 be marked offline until the next write. This should increase
4353 reliability in case of a crash. The synchronization delay
4354 can be configured via SyncIntervalSec= in journald.conf.
4356 * There's a new remote-fs-setup.target unit that can be used
4357 to pull in specific services when at least one remote file
4358 system is to be mounted.
4360 * There are new targets timers.target and paths.target as
4361 canonical targets to pull user timer and path units in
4362 from. This complements sockets.target with a similar
4363 purpose for socket units.
4365 * libudev gained a new call udev_device_set_attribute_value()
4366 to set sysfs attributes of a device.
4368 * The udev daemon now sets the default number of worker
4369 processes executed in parallel based on the number of available
4370 CPUs instead of the amount of available RAM. This is supposed
4371 to provide a more reliable default and limit a too aggressive
4372 paralellism for setups with 1000s of devices connected.
4374 Contributions from: Auke Kok, Colin Walters, Cristian
4375 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4376 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4377 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4378 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4379 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4380 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4381 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4382 Zbigniew Jędrzejewski-Szmek
4386 * Configuration of unit files may now be extended via drop-in
4387 files without having to edit/override the unit files
4388 themselves. More specifically, if the administrator wants to
4389 change one value for a service file foobar.service he can
4390 now do so by dropping in a configuration snippet into
4391 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4392 will load all these snippets and apply them on top of the
4393 main unit configuration file, possibly extending or
4394 overriding its settings. Using these drop-in snippets is
4395 generally nicer than the two earlier options for changing
4396 unit files locally: copying the files from
4397 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4398 them there; or creating a new file in /etc/systemd/system/
4399 that incorporates the original one via ".include". Drop-in
4400 snippets into these .d/ directories can be placed in any
4401 directory systemd looks for units in, and the usual
4402 overriding semantics between /usr/lib, /etc and /run apply
4405 * Most unit file settings which take lists of items can now be
4406 reset by assigning the empty string to them. For example,
4407 normally, settings such as Environment=FOO=BAR append a new
4408 environment variable assignment to the environment block,
4409 each time they are used. By assigning Environment= the empty
4410 string the environment block can be reset to empty. This is
4411 particularly useful with the .d/*.conf drop-in snippets
4412 mentioned above, since this adds the ability to reset list
4413 settings from vendor unit files via these drop-ins.
4415 * systemctl gained a new "list-dependencies" command for
4416 listing the dependencies of a unit recursively.
4418 * Inhibitors are now honored and listed by "systemctl
4419 suspend", "systemctl poweroff" (and similar) too, not only
4420 GNOME. These commands will also list active sessions by
4423 * Resource limits (as exposed by the various control group
4424 controllers) can now be controlled dynamically at runtime
4425 for all units. More specifically, you can now use a command
4426 like "systemctl set-cgroup-attr foobar.service cpu.shares
4427 2000" to alter the CPU shares a specific service gets. These
4428 settings are stored persistently on disk, and thus allow the
4429 administrator to easily adjust the resource usage of
4430 services with a few simple commands. This dynamic resource
4431 management logic is also available to other programs via the
4432 bus. Almost any kernel cgroup attribute and controller is
4435 * systemd-vconsole-setup will now copy all font settings to
4436 all allocated VTs, where it previously applied them only to
4439 * libsystemd-login gained the new sd_session_get_tty() API
4442 * This release drops support for a few legacy or
4443 distribution-specific LSB facility names when parsing init
4444 scripts: $x-display-manager, $mail-transfer-agent,
4445 $mail-transport-agent, $mail-transfer-agent, $smtp,
4446 $null. Also, the mail-transfer-agent.target unit backing
4447 this has been removed. Distributions which want to retain
4448 compatibility with this should carry the burden for
4449 supporting this themselves and patch support for these back
4450 in, if they really need to. Also, the facilities $syslog and
4451 $local_fs are now ignored, since systemd does not support
4452 early-boot LSB init scripts anymore, and these facilities
4453 are implied anyway for normal services. syslog.target has
4456 * There are new bus calls on PID1's Manager object for
4457 cancelling jobs, and removing snapshot units. Previously,
4458 both calls were only available on the Job and Snapshot
4461 * systemd-journal-gatewayd gained SSL support.
4463 * The various "environment" files, such as /etc/locale.conf
4464 now support continuation lines with a backslash ("\") as
4465 last character in the line, similarly in style (but different)
4466 to how this is supported in shells.
4468 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4469 now implicitly appended to every log entry logged. systemctl
4470 has been updated to filter by this field when operating on a
4471 user systemd instance.
4473 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4474 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4475 the container. This makes it easier to boot unmodified
4476 Fedora systems in a container, which however still requires
4477 audit=0 to be passed on the kernel command line. Auditing in
4478 kernel and userspace is unfortunately still too broken in
4479 context of containers, hence we recommend compiling it out
4480 of the kernel or using audit=0. Hopefully this will be fixed
4481 one day for good in the kernel.
4483 * nspawn gained the new --bind= and --bind-ro= parameters to
4484 bind mount specific directories from the host into the
4487 * nspawn will now mount its own devpts file system instance
4488 into the container, in order not to leak pty devices from
4489 the host into the container.
4491 * systemd will now read the firmware boot time performance
4492 information from the EFI variables, if the used boot loader
4493 supports this, and takes it into account for boot performance
4494 analysis via "systemd-analyze". This is currently supported
4495 only in conjunction with Gummiboot, but could be supported
4496 by other boot loaders too. For details see:
4498 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4500 * A new generator has been added that automatically mounts the
4501 EFI System Partition (ESP) to /boot, if that directory
4502 exists, is empty, and no other file system has been
4503 configured to be mounted there.
4505 * logind will now send out PrepareForSleep(false) out
4506 unconditionally, after coming back from suspend. This may be
4507 used by applications as asynchronous notification for
4508 system resume events.
4510 * "systemctl unlock-sessions" has been added, that allows
4511 unlocking the screens of all user sessions at once, similar
4512 to how "systemctl lock-sessions" already locked all users
4513 sessions. This is backed by a new D-Bus call UnlockSessions().
4515 * "loginctl seat-status" will now show the master device of a
4516 seat. (i.e. the device of a seat that needs to be around for
4517 the seat to be considered available, usually the graphics
4520 * tmpfiles gained a new "X" line type, that allows
4521 configuration of files and directories (with wildcards) that
4522 shall be excluded from automatic cleanup ("aging").
4524 * udev default rules set the device node permissions now only
4525 at "add" events, and do not change them any longer with a
4526 later "change" event.
4528 * The log messages for lid events and power/sleep keypresses
4529 now carry a message ID.
4531 * We now have a substantially larger unit test suite, but this
4532 continues to be work in progress.
4534 * udevadm hwdb gained a new --root= parameter to change the
4535 root directory to operate relative to.
4537 * logind will now issue a background sync() request to the kernel
4538 early at shutdown, so that dirty buffers are flushed to disk early
4539 instead of at the last moment, in order to optimize shutdown
4542 * A new bootctl tool has been added that is an interface for
4543 certain boot loader operations. This is currently a preview
4544 and is likely to be extended into a small mechanism daemon
4545 like timedated, localed, hostnamed, and can be used by
4546 graphical UIs to enumerate available boot options, and
4547 request boot into firmware operations.
4549 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4550 the rest of the package. It also has been updated to work
4551 correctly in initrds.
4553 * Policykit previously has been runtime optional, and is now
4554 also compile time optional via a configure switch.
4556 * systemd-analyze has been reimplemented in C. Also "systemctl
4557 dot" has moved into systemd-analyze.
4559 * "systemctl status" with no further parameters will now print
4560 the status of all active or failed units.
4562 * Operations such as "systemctl start" can now be executed
4563 with a new mode "--irreversible" which may be used to queue
4564 operations that cannot accidentally be reversed by a later
4565 job queuing. This is by default used to make shutdown
4566 requests more robust.
4568 * The Python API of systemd now gained a new module for
4569 reading journal files.
4571 * A new tool kernel-install has been added that can install
4572 kernel images according to the Boot Loader Specification:
4574 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4576 * Boot time console output has been improved to provide
4577 animated boot time output for hanging jobs.
4579 * A new tool systemd-activate has been added which can be used
4580 to test socket activation with, directly from the command
4581 line. This should make it much easier to test and debug
4582 socket activation in daemons.
4584 * journalctl gained a new "--reverse" (or -r) option to show
4585 journal output in reverse order (i.e. newest line first).
4587 * journalctl gained a new "--pager-end" (or -e) option to jump
4588 to immediately jump to the end of the journal in the
4589 pager. This is only supported in conjunction with "less".
4591 * journalctl gained a new "--user-unit=" option, that works
4592 similarly to "--unit=" but filters for user units rather than
4595 * A number of unit files to ease adoption of systemd in
4596 initrds has been added. This moves some minimal logic from
4597 the various initrd implementations into systemd proper.
4599 * The journal files are now owned by a new group
4600 "systemd-journal", which exists specifically to allow access
4601 to the journal, and nothing else. Previously, we used the
4602 "adm" group for that, which however possibly covers more
4603 than just journal/log file access. This new group is now
4604 already used by systemd-journal-gatewayd to ensure this
4605 daemon gets access to the journal files and as little else
4606 as possible. Note that "make install" will also set FS ACLs
4607 up for /var/log/journal to give "adm" and "wheel" read
4608 access to it, in addition to "systemd-journal" which owns
4609 the journal files. We recommend that packaging scripts also
4610 add read access to "adm" + "wheel" to /var/log/journal, and
4611 all existing/future journal files. To normal users and
4612 administrators little changes, however packagers need to
4613 ensure to create the "systemd-journal" system group at
4614 package installation time.
4616 * The systemd-journal-gatewayd now runs as unprivileged user
4617 systemd-journal-gateway:systemd-journal-gateway. Packaging
4618 scripts need to create these system user/group at
4621 * timedated now exposes a new boolean property CanNTP that
4622 indicates whether a local NTP service is available or not.
4624 * systemd-detect-virt will now also detect xen PVs
4626 * The pstore file system is now mounted by default, if it is
4629 * In addition to the SELinux and IMA policies we will now also
4630 load SMACK policies at early boot.
4632 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4633 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4634 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4635 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4636 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4637 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4638 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4639 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4640 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4641 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4642 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4643 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4644 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4645 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4649 * Timer units now support calendar time events in addition to
4650 monotonic time events. That means you can now trigger a unit
4651 based on a calendar time specification such as "Thu,Fri
4652 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4653 or fifth day of any month of the year 2013, given that it is
4654 a thursday or friday. This brings timer event support
4655 considerably closer to cron's capabilities. For details on
4656 the supported calendar time specification language see
4659 * udev now supports a number of different naming policies for
4660 network interfaces for predictable names, and a combination
4661 of these policies is now the default. Please see this wiki
4662 document for details:
4664 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4666 * Auke Kok's bootchart implementation has been added to the
4667 systemd tree. It is an optional component that can graph the
4668 boot in quite some detail. It is one of the best bootchart
4669 implementations around and minimal in its code and
4672 * nss-myhostname has been integrated into the systemd source
4673 tree. nss-myhostname guarantees that the local hostname
4674 always stays resolvable via NSS. It has been a weak
4675 requirement of systemd-hostnamed since a long time, and
4676 since its code is actually trivial we decided to just
4677 include it in systemd's source tree. It can be turned off
4678 with a configure switch.
4680 * The read-ahead logic is now capable of properly detecting
4681 whether a btrfs file system is on SSD or rotating media, in
4682 order to optimize the read-ahead scheme. Previously, it was
4683 only capable of detecting this on traditional file systems
4686 * In udev, additional device properties are now read from the
4687 IAB in addition to the OUI database. Also, Bluetooth company
4688 identities are attached to the devices as well.
4690 * In service files %U may be used as specifier that is
4691 replaced by the configured user name of the service.
4693 * nspawn may now be invoked without a controlling TTY. This
4694 makes it suitable for invocation as its own service. This
4695 may be used to set up a simple containerized server system
4696 using only core OS tools.
4698 * systemd and nspawn can now accept socket file descriptors
4699 when they are started for socket activation. This enables
4700 implementation of socket activated nspawn
4701 containers. i.e. think about autospawning an entire OS image
4702 when the first SSH or HTTP connection is received. We expect
4703 that similar functionality will also be added to libvirt-lxc
4706 * journalctl will now suppress ANSI color codes when
4707 presenting log data.
4709 * systemctl will no longer show control group information for
4710 a unit if a the control group is empty anyway.
4712 * logind can now automatically suspend/hibernate/shutdown the
4715 * /etc/machine-info and hostnamed now also expose the chassis
4716 type of the system. This can be used to determine whether
4717 the local system is a laptop, desktop, handset or
4718 tablet. This information may either be configured by the
4719 user/vendor or is automatically determined from ACPI and DMI
4720 information if possible.
4722 * A number of PolicyKit actions are now bound together with
4723 "imply" rules. This should simplify creating UIs because
4724 many actions will now authenticate similar ones as well.
4726 * Unit files learnt a new condition ConditionACPower= which
4727 may be used to conditionalize a unit depending on whether an
4728 AC power source is connected or not, of whether the system
4729 is running on battery power.
4731 * systemctl gained a new "is-failed" verb that may be used in
4732 shell scripts and suchlike to check whether a specific unit
4733 is in the "failed" state.
4735 * The EnvironmentFile= setting in unit files now supports file
4736 globbing, and can hence be used to easily read a number of
4737 environment files at once.
4739 * systemd will no longer detect and recognize specific
4740 distributions. All distribution-specific #ifdeffery has been
4741 removed, systemd is now fully generic and
4742 distribution-agnostic. Effectively, not too much is lost as
4743 a lot of the code is still accessible via explicit configure
4744 switches. However, support for some distribution specific
4745 legacy configuration file formats has been dropped. We
4746 recommend distributions to simply adopt the configuration
4747 files everybody else uses now and convert the old
4748 configuration from packaging scripts. Most distributions
4749 already did that. If that's not possible or desirable,
4750 distributions are welcome to forward port the specific
4751 pieces of code locally from the git history.
4753 * When logging a message about a unit systemd will now always
4754 log the unit name in the message meta data.
4756 * localectl will now also discover system locale data that is
4757 not stored in locale archives, but directly unpacked.
4759 * logind will no longer unconditionally use framebuffer
4760 devices as seat masters, i.e. as devices that are required
4761 to be existing before a seat is considered preset. Instead,
4762 it will now look for all devices that are tagged as
4763 "seat-master" in udev. By default, framebuffer devices will
4764 be marked as such, but depending on local systems, other
4765 devices might be marked as well. This may be used to
4766 integrate graphics cards using closed source drivers (such
4767 as NVidia ones) more nicely into logind. Note however, that
4768 we recommend using the open source NVidia drivers instead,
4769 and no udev rules for the closed-source drivers will be
4770 shipped from us upstream.
4772 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4773 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4774 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4775 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4776 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4777 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4778 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4779 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4780 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4781 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4782 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4783 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4788 * udev gained support for loading additional device properties
4789 from an indexed database that is keyed by vendor/product IDs
4790 and similar device identifiers. For the beginning this
4791 "hwdb" is populated with data from the well-known PCI and
4792 USB database, but also includes PNP, ACPI and OID data. In
4793 the longer run this indexed database shall grow into
4794 becoming the one central database for non-essential
4795 userspace device metadata. Previously, data from the PCI/USB
4796 database was only attached to select devices, since the
4797 lookup was a relatively expensive operation due to O(n) time
4798 complexity (with n being the number of entries in the
4799 database). Since this is now O(1), we decided to add in this
4800 data for all devices where this is available, by
4801 default. Note that the indexed database needs to be rebuilt
4802 when new data files are installed. To achieve this you need
4803 to update your packaging scripts to invoke "udevadm hwdb
4804 --update" after installation of hwdb data files. For
4805 RPM-based distributions we introduced the new
4806 %udev_hwdb_update macro for this purpose.
4808 * The Journal gained support for the "Message Catalog", an
4809 indexed database to link up additional information with
4810 journal entries. For further details please check:
4812 http://www.freedesktop.org/wiki/Software/systemd/catalog
4814 The indexed message catalog database also needs to be
4815 rebuilt after installation of message catalog files. Use
4816 "journalctl --update-catalog" for this. For RPM-based
4817 distributions we introduced the %journal_catalog_update
4818 macro for this purpose.
4820 * The Python Journal bindings gained support for the standard
4821 Python logging framework.
4823 * The Journal API gained new functions for checking whether
4824 the underlying file system of a journal file is capable of
4825 properly reporting file change notifications, or whether
4826 applications that want to reflect journal changes "live"
4827 need to recheck journal files continuously in appropriate
4830 * It is now possible to set the "age" field for tmpfiles
4831 entries to 0, indicating that files matching this entry
4832 shall always be removed when the directories are cleaned up.
4834 * coredumpctl gained a new "gdb" verb which invokes gdb
4835 right-away on the selected coredump.
4837 * There's now support for "hybrid sleep" on kernels that
4838 support this, in addition to "suspend" and "hibernate". Use
4839 "systemctl hybrid-sleep" to make use of this.
4841 * logind's HandleSuspendKey= setting (and related settings)
4842 now gained support for a new "lock" setting to simply
4843 request the screen lock on all local sessions, instead of
4844 actually executing a suspend or hibernation.
4846 * systemd will now mount the EFI variables file system by
4849 * Socket units now gained support for configuration of the
4850 SMACK security label.
4852 * timedatectl will now output the time of the last and next
4853 daylight saving change.
4855 * We dropped support for various legacy and distro-specific
4856 concepts, such as insserv, early-boot SysV services
4857 (i.e. those for non-standard runlevels such as 'b' or 'S')
4858 or ArchLinux /etc/rc.conf support. We recommend the
4859 distributions who still need support this to either continue
4860 to maintain the necessary patches downstream, or find a
4861 different solution. (Talk to us if you have questions!)
4863 * Various systemd components will now bypass PolicyKit checks
4864 for root and otherwise handle properly if PolicyKit is not
4865 found to be around. This should fix most issues for
4866 PolicyKit-less systems. Quite frankly this should have been
4867 this way since day one. It is absolutely our intention to
4868 make systemd work fine on PolicyKit-less systems, and we
4869 consider it a bug if something does not work as it should if
4870 PolicyKit is not around.
4872 * For embedded systems it is now possible to build udev and
4873 systemd without blkid and/or kmod support.
4875 * "systemctl switch-root" is now capable of switching root
4876 more than once. I.e. in addition to transitions from the
4877 initrd to the host OS it is now possible to transition to
4878 further OS images from the host. This is useful to implement
4879 offline updating tools.
4881 * Various other additions have been made to the RPM macros
4882 shipped with systemd. Use %udev_rules_update() after
4883 installing new udev rules files. %_udevhwdbdir,
4884 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4885 %_sysctldir are now available which resolve to the right
4886 directories for packages to place various data files in.
4888 * journalctl gained the new --full switch (in addition to
4889 --all, to disable ellipsation for long messages.
4891 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4892 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4893 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4894 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4895 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4896 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4897 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4898 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4899 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4903 * journalctl gained new --since= and --until= switches to
4904 filter by time. It also now supports nice filtering for
4905 units via --unit=/-u.
4907 * Type=oneshot services may use ExecReload= and do the
4910 * The journal daemon now supports time-based rotation and
4911 vacuuming, in addition to the usual disk-space based
4914 * The journal will now index the available field values for
4915 each field name. This enables clients to show pretty drop
4916 downs of available match values when filtering. The bash
4917 completion of journalctl has been updated
4918 accordingly. journalctl gained a new switch -F to list all
4919 values a certain field takes in the journal database.
4921 * More service events are now written as structured messages
4922 to the journal, and made recognizable via message IDs.
4924 * The timedated, localed and hostnamed mini-services which
4925 previously only provided support for changing time, locale
4926 and hostname settings from graphical DEs such as GNOME now
4927 also have a minimal (but very useful) text-based client
4928 utility each. This is probably the nicest way to changing
4929 these settings from the command line now, especially since
4930 it lists available options and is fully integrated with bash
4933 * There's now a new tool "systemd-coredumpctl" to list and
4934 extract coredumps from the journal.
4936 * We now install a README each in /var/log/ and
4937 /etc/rc.d/init.d explaining where the system logs and init
4938 scripts went. This hopefully should help folks who go to
4939 that dirs and look into the otherwise now empty void and
4940 scratch their heads.
4942 * When user-services are invoked (by systemd --user) the
4943 $MANAGERPID env var is set to the PID of systemd.
4945 * SIGRTMIN+24 when sent to a --user instance will now result
4946 in immediate termination of systemd.
4948 * gatewayd received numerous feature additions such as a
4949 "follow" mode, for live syncing and filtering.
4951 * browse.html now allows filtering and showing detailed
4952 information on specific entries. Keyboard navigation and
4953 mouse screen support has been added.
4955 * gatewayd/journalctl now supports HTML5/JSON
4956 Server-Sent-Events as output.
4958 * The SysV init script compatibility logic will now
4959 heuristically determine whether a script supports the
4960 "reload" verb, and only then make this available as
4963 * "systemctl status --follow" has been removed, use "journalctl
4966 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4967 have been removed since they are hardly useful to be
4970 * And I'd like to take the opportunity to specifically mention
4971 Zbigniew for his great contributions. Zbigniew, you rock!
4973 Contributions from: Andrew Eikum, Christian Hesse, Colin
4974 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4975 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4976 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4977 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4978 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4979 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4983 * If /etc/vconsole.conf is non-existent or empty we will no
4984 longer load any console font or key map at boot by
4985 default. Instead the kernel defaults will be left
4986 intact. This is definitely the right thing to do, as no
4987 configuration should mean no configuration, and hard-coding
4988 font names that are different on all archs is probably a bad
4989 idea. Also, the kernel default key map and font should be
4990 good enough for most cases anyway, and mostly identical to
4991 the userspace fonts/key maps we previously overloaded them
4992 with. If distributions want to continue to default to a
4993 non-kernel font or key map they should ship a default
4994 /etc/vconsole.conf with the appropriate contents.
4996 Contributions from: Colin Walters, Daniel J Walsh, Dave
4997 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4998 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5002 * journalctl gained a new --cursor= switch to show entries
5003 starting from the specified location in the journal.
5005 * We now enforce a size limit on journal entry fields exported
5006 with "-o json" in journalctl. Fields larger than 4K will be
5007 assigned null. This can be turned off with --all.
5009 * An (optional) journal gateway daemon is now available as
5010 "systemd-journal-gatewayd.service". This service provides
5011 access to the journal via HTTP and JSON. This functionality
5012 will be used to implement live log synchronization in both
5013 pull and push modes, but has various other users too, such
5014 as easy log access for debugging of embedded devices. Right
5015 now it is already useful to retrieve the journal via HTTP:
5017 # systemctl start systemd-journal-gatewayd.service
5018 # wget http://localhost:19531/entries
5020 This will download the journal contents in a
5021 /var/log/messages compatible format. The same as JSON:
5023 # curl -H"Accept: application/json" http://localhost:19531/entries
5025 This service is also accessible via a web browser where a
5026 single static HTML5 app is served that uses the JSON logic
5027 to enable the user to do some basic browsing of the
5028 journal. This will be extended later on. Here's an example
5029 screenshot of this app in its current state:
5031 http://0pointer.de/public/journal-gatewayd
5033 Contributions from: Kay Sievers, Lennart Poettering, Robert
5034 Milasan, Tom Gundersen
5038 * The bash completion logic is now available for journalctl
5041 * We do not mount the "cpuset" controller anymore together with
5042 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5043 started if no parameters are assigned to it. "cpuset" hence
5044 broke code that assumed it it could create "cpu" groups and
5047 * journalctl -f will now subscribe to terminal size changes,
5048 and line break accordingly.
5050 Contributions from: Dave Reisner, Kay Sievers, Lennart
5051 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5055 * nspawn will now create a symlink /etc/localtime in the
5056 container environment, copying the host's timezone
5057 setting. Previously this has been done via a bind mount, but
5058 since symlinks cannot be bind mounted this has now been
5059 changed to create/update the appropriate symlink.
5061 * journalctl -n's line number argument is now optional, and
5062 will default to 10 if omitted.
5064 * journald will now log the maximum size the journal files may
5065 take up on disk. This is particularly useful if the default
5066 built-in logic of determining this parameter from the file
5067 system size is used. Use "systemctl status
5068 systemd-journald.service" to see this information.
5070 * The multi-seat X wrapper tool has been stripped down. As X
5071 is now capable of enumerating graphics devices via udev in a
5072 seat-aware way the wrapper is not strictly necessary
5073 anymore. A stripped down temporary stop-gap is still shipped
5074 until the upstream display managers have been updated to
5075 fully support the new X logic. Expect this wrapper to be
5076 removed entirely in one of the next releases.
5078 * HandleSleepKey= in logind.conf has been split up into
5079 HandleSuspendKey= and HandleHibernateKey=. The old setting
5080 is not available anymore. X11 and the kernel are
5081 distinguishing between these keys and we should too. This
5082 also means the inhibition lock for these keys has been split
5085 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5086 Poettering, Lukas Nykryn, Václav Pavlín
5090 * Whenever a unit changes state we will now log this to the
5091 journal and show along the unit's own log output in
5094 * ConditionPathIsMountPoint= can now properly detect bind
5095 mount points too. (Previously, a bind mount of one file
5096 system to another place in the same file system could not be
5097 detected as mount, since they shared struct stat's st_dev
5100 * We will now mount the cgroup controllers cpu, cpuacct,
5101 cpuset and the controllers net_cls, net_prio together by
5104 * nspawn containers will now have a virtualized boot
5105 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5106 over with a randomized ID at container initialization). This
5107 has the effect of making "journalctl -b" do the right thing
5110 * The JSON output journal serialization has been updated not
5111 to generate "endless" list objects anymore, but rather one
5112 JSON object per line. This is more in line how most JSON
5113 parsers expect JSON objects. The new output mode
5114 "json-pretty" has been added to provide similar output, but
5115 neatly aligned for readability by humans.
5117 * We dropped all explicit sync() invocations in the shutdown
5118 code. The kernel does this implicitly anyway in the kernel
5119 reboot() syscall. halt(8)'s -n option is now a compatibility
5122 * We now support virtualized reboot() in containers, as
5123 supported by newer kernels. We will fall back to exit() if
5124 CAP_SYS_REBOOT is not available to the container. Also,
5125 nspawn makes use of this now and will actually reboot the
5126 container if the containerized OS asks for that.
5128 * journalctl will only show local log output by default
5129 now. Use --merge (-m) to show remote log output, too.
5131 * libsystemd-journal gained the new sd_journal_get_usage()
5132 call to determine the current disk usage of all journal
5133 files. This is exposed in the new "journalctl --disk-usage"
5136 * journald gained a new configuration setting SplitMode= in
5137 journald.conf which may be used to control how user journals
5138 are split off. See journald.conf(5) for details.
5140 * A new condition type ConditionFileNotEmpty= has been added.
5142 * tmpfiles' "w" lines now support file globbing, to write
5143 multiple files at once.
5145 * We added Python bindings for the journal submission
5146 APIs. More Python APIs for a number of selected APIs will
5147 likely follow. Note that we intend to add native bindings
5148 only for the Python language, as we consider it common
5149 enough to deserve bindings shipped within systemd. There are
5150 various projects outside of systemd that provide bindings
5151 for languages such as PHP or Lua.
5153 * Many conditions will now resolve specifiers such as %i. In
5154 addition, PathChanged= and related directives of .path units
5155 now support specifiers as well.
5157 * There's now a new RPM macro definition for the system preset
5160 * journald will now warn if it ca not forward a message to the
5161 syslog daemon because its socket is full.
5163 * timedated will no longer write or process /etc/timezone,
5164 except on Debian. As we do not support late mounted /usr
5165 anymore /etc/localtime always being a symlink is now safe,
5166 and hence the information in /etc/timezone is not necessary
5169 * logind will now always reserve one VT for a text getty (VT6
5170 by default). Previously if more than 6 X sessions where
5171 started they took up all the VTs with auto-spawned gettys,
5172 so that no text gettys were available anymore.
5174 * udev will now automatically inform the btrfs kernel logic
5175 about btrfs RAID components showing up. This should make
5176 simple hotplug based btrfs RAID assembly work.
5178 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5179 (but not for its children which will stay at the kernel
5180 default). This should allow setups with a lot more listening
5183 * systemd will now always pass the configured timezone to the
5184 kernel at boot. timedated will do the same when the timezone
5187 * logind's inhibition logic has been updated. By default,
5188 logind will now handle the lid switch, the power and sleep
5189 keys all the time, even in graphical sessions. If DEs want
5190 to handle these events on their own they should take the new
5191 handle-power-key, handle-sleep-key and handle-lid-switch
5192 inhibitors during their runtime. A simple way to achieve
5193 that is to invoke the DE wrapped in an invocation of:
5195 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5197 * Access to unit operations is now checked via SELinux taking
5198 the unit file label and client process label into account.
5200 * systemd will now notify the administrator in the journal
5201 when he over-mounts a non-empty directory.
5203 * There are new specifiers that are resolved in unit files,
5204 for the host name (%H), the machine ID (%m) and the boot ID
5207 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5208 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5209 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5210 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5211 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5212 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5213 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5217 * Support for reading structured kernel messages from
5218 /dev/kmsg has now been added and is enabled by default.
5220 * Support for reading kernel messages from /proc/kmsg has now
5221 been removed. If you want kernel messages in the journal
5222 make sure to run a recent kernel (>= 3.5) that supports
5223 reading structured messages from /dev/kmsg (see
5224 above). /proc/kmsg is now exclusive property of classic
5225 syslog daemons again.
5227 * The libudev API gained the new
5228 udev_device_new_from_device_id() call.
5230 * The logic for file system namespace (ReadOnlyDirectory=,
5231 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5232 require pivot_root() anymore. This means fewer temporary
5233 directories are created below /tmp for this feature.
5235 * nspawn containers will now see and receive all submounts
5236 made on the host OS below the root file system of the
5239 * Forward Secure Sealing is now supported for Journal files,
5240 which provide cryptographical sealing of journal files so
5241 that attackers cannot alter log history anymore without this
5242 being detectable. Lennart will soon post a blog story about
5243 this explaining it in more detail.
5245 * There are two new service settings RestartPreventExitStatus=
5246 and SuccessExitStatus= which allow configuration of exit
5247 status (exit code or signal) which will be excepted from the
5248 restart logic, resp. consider successful.
5250 * journalctl gained the new --verify switch that can be used
5251 to check the integrity of the structure of journal files and
5252 (if Forward Secure Sealing is enabled) the contents of
5255 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5256 and similar symlinks pre-created. This makes running shells
5257 as container init process a lot more fun.
5259 * The fstab support can now handle PARTUUID= and PARTLABEL=
5262 * A new ConditionHost= condition has been added to match
5263 against the hostname (with globs) and machine ID. This is
5264 useful for clusters where a single OS image is used to
5265 provision a large number of hosts which shall run slightly
5266 different sets of services.
5268 * Services which hit the restart limit will now be placed in a
5271 Contributions from: Bertram Poettering, Dave Reisner, Huang
5272 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5273 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5277 * When running in --user mode systemd will now become a
5278 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5279 tree a lot more organized.
5281 * A new PartOf= unit dependency type has been introduced that
5282 may be used to group services in a natural way.
5284 * "systemctl enable" may now be used to enable instances of
5287 * journalctl now prints error log levels in red, and
5288 warning/notice log levels in bright white. It also supports
5289 filtering by log level now.
5291 * cgtop gained a new -n switch (similar to top), to configure
5292 the maximum number of iterations to run for. It also gained
5293 -b, to run in batch mode (accepting no input).
5295 * The suffix ".service" may now be omitted on most systemctl
5296 command lines involving service unit names.
5298 * There's a new bus call in logind to lock all sessions, as
5299 well as a loginctl verb for it "lock-sessions".
5301 * libsystemd-logind.so gained a new call sd_journal_perror()
5302 that works similar to libc perror() but logs to the journal
5303 and encodes structured information about the error number.
5305 * /etc/crypttab entries now understand the new keyfile-size=
5308 * shutdown(8) now can send a (configurable) wall message when
5309 a shutdown is cancelled.
5311 * The mount propagation mode for the root file system will now
5312 default to "shared", which is useful to make containers work
5313 nicely out-of-the-box so that they receive new mounts from
5314 the host. This can be undone locally by running "mount
5315 --make-rprivate /" if needed.
5317 * The prefdm.service file has been removed. Distributions
5318 should maintain this unit downstream if they intend to keep
5319 it around. However, we recommend writing normal unit files
5320 for display managers instead.
5322 * Since systemd is a crucial part of the OS we will now
5323 default to a number of compiler switches that improve
5324 security (hardening) such as read-only relocations, stack
5325 protection, and suchlike.
5327 * The TimeoutSec= setting for services is now split into
5328 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5329 of individual time outs for the start and the stop phase of
5332 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5333 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5334 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5335 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5336 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5337 Gundersen, Zbigniew Jędrzejewski-Szmek
5341 * The journal and id128 C APIs are now fully documented as man
5344 * Extra safety checks have been added when transitioning from
5345 the initial RAM disk to the main system to avoid accidental
5348 * /etc/crypttab entries now understand the new keyfile-offset=
5351 * systemctl -t can now be used to filter by unit load state.
5353 * The journal C API gained the new sd_journal_wait() call to
5354 make writing synchronous journal clients easier.
5356 * journalctl gained the new -D switch to show journals from a
5359 * journalctl now displays a special marker between log
5360 messages of two different boots.
5362 * The journal is now explicitly flushed to /var via a service
5363 systemd-journal-flush.service, rather than implicitly simply
5364 by seeing /var/log/journal to be writable.
5366 * journalctl (and the journal C APIs) can now match for much
5367 more complex expressions, with alternatives and
5370 * When transitioning from the initial RAM disk to the main
5371 system we will now kill all processes in a killing spree to
5372 ensure no processes stay around by accident.
5374 * Three new specifiers may be used in unit files: %u, %h, %s
5375 resolve to the user name, user home directory resp. user
5376 shell. This is useful for running systemd user instances.
5378 * We now automatically rotate journal files if their data
5379 object hash table gets a fill level > 75%. We also size the
5380 hash table based on the configured maximum file size. This
5381 together should lower hash collisions drastically and thus
5382 speed things up a bit.
5384 * journalctl gained the new "--header" switch to introspect
5385 header data of journal files.
5387 * A new setting SystemCallFilters= has been added to services
5388 which may be used to apply blacklists or whitelists to
5389 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5391 * nspawn gained a new --link-journal= switch (and quicker: -j)
5392 to link the container journal with the host. This makes it
5393 very easy to centralize log viewing on the host for all
5394 guests while still keeping the journal files separated.
5396 * Many bugfixes and optimizations
5398 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5399 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5400 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5405 * Several tools now understand kernel command line arguments,
5406 which are only read when run in an initial RAM disk. They
5407 usually follow closely their normal counterparts, but are
5410 * There's a new tool to analyze the readahead files that are
5411 automatically generated at boot. Use:
5413 /usr/lib/systemd/systemd-readahead analyze /.readahead
5415 * We now provide an early debug shell on tty9 if this enabled. Use:
5417 systemctl enable debug-shell.service
5419 * All plymouth related units have been moved into the Plymouth
5420 package. Please make sure to upgrade your Plymouth version
5423 * systemd-tmpfiles now supports getting passed the basename of
5424 a configuration file only, in which case it will look for it
5425 in all appropriate directories automatically.
5427 * udevadm info now takes a /dev or /sys path as argument, and
5428 does the right thing. Example:
5430 udevadm info /dev/sda
5431 udevadm info /sys/class/block/sda
5433 * systemctl now prints a warning if a unit is stopped but a
5434 unit that might trigger it continues to run. Example: a
5435 service is stopped but the socket that activates it is left
5438 * "systemctl status" will now mention if the log output was
5439 shortened due to rotation since a service has been started.
5441 * The journal API now exposes functions to determine the
5442 "cutoff" times due to rotation.
5444 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5445 immediately flushing of runtime logs to /var if possible,
5446 resp. for triggering immediate rotation of the journal
5449 * It is now considered an error if a service is attempted to
5450 be stopped that is not loaded.
5452 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5454 * systemd-analyze now supports Python 3
5456 * tmpfiles now supports cleaning up directories via aging
5457 where the first level dirs are always kept around but
5458 directories beneath it automatically aged. This is enabled
5459 by prefixing the age field with '~'.
5461 * Seat objects now expose CanGraphical, CanTTY properties
5462 which is required to deal with very fast bootups where the
5463 display manager might be running before the graphics drivers
5464 completed initialization.
5466 * Seat objects now expose a State property.
5468 * We now include RPM macros for service enabling/disabling
5469 based on the preset logic. We recommend RPM based
5470 distributions to make use of these macros if possible. This
5471 makes it simpler to reuse RPM spec files across
5474 * We now make sure that the collected systemd unit name is
5475 always valid when services log to the journal via
5478 * There's a new man page kernel-command-line(7) detailing all
5479 command line options we understand.
5481 * The fstab generator may now be disabled at boot by passing
5482 fstab=0 on the kernel command line.
5484 * A new kernel command line option modules-load= is now understood
5485 to load a specific kernel module statically, early at boot.
5487 * Unit names specified on the systemctl command line are now
5488 automatically escaped as needed. Also, if file system or
5489 device paths are specified they are automatically turned
5490 into the appropriate mount or device unit names. Example:
5492 systemctl status /home
5493 systemctl status /dev/sda
5495 * The SysVConsole= configuration option has been removed from
5496 system.conf parsing.
5498 * The SysV search path is no longer exported on the D-Bus
5501 * The Names= option is been removed from unit file parsing.
5503 * There's a new man page bootup(7) detailing the boot process.
5505 * Every unit and every generator we ship with systemd now
5506 comes with full documentation. The self-explanatory boot is
5509 * A couple of services gained "systemd-" prefixes in their
5510 name if they wrap systemd code, rather than only external
5511 code. Among them fsck@.service which is now
5512 systemd-fsck@.service.
5514 * The HaveWatchdog property has been removed from the D-Bus
5517 * systemd.confirm_spawn= on the kernel command line should now
5520 * There's a new man page crypttab(5) which details all options
5521 we actually understand.
5523 * systemd-nspawn gained a new --capability= switch to pass
5524 additional capabilities to the container.
5526 * timedated will now read known NTP implementation unit names
5527 from /usr/lib/systemd/ntp-units.d/*.list,
5528 systemd-timedated-ntp.target has been removed.
5530 * journalctl gained a new switch "-b" that lists log data of
5531 the current boot only.
5533 * The notify socket is in the abstract namespace again, in
5534 order to support daemons which chroot() at start-up.
5536 * There is a new Storage= configuration option for journald
5537 which allows configuration of where log data should go. This
5538 also provides a way to disable journal logging entirely, so
5539 that data collected is only forwarded to the console, the
5540 kernel log buffer or another syslog implementation.
5542 * Many bugfixes and optimizations
5544 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5545 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5546 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5547 Shawn Landden, Tom Gundersen
5551 * "systemctl help <unit>" now shows the man page if one is
5554 * Several new man pages have been added.
5556 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5557 MaxLevelConsole= can now be specified in
5558 journald.conf. These options allow reducing the amount of
5559 data stored on disk or forwarded by the log level.
5561 * TimerSlackNSec= can now be specified in system.conf for
5562 PID1. This allows system-wide power savings.
5564 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5565 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5570 * logind is now capable of (optionally) handling power and
5571 sleep keys as well as the lid switch.
5573 * journalctl now understands the syntax "journalctl
5574 /usr/bin/avahi-daemon" to get all log output of a specific
5577 * CapabilityBoundingSet= in system.conf now also influences
5578 the capability bound set of usermode helpers of the kernel.
5580 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5581 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5582 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5583 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5587 * Note that we skipped 139 releases here in order to set the
5588 new version to something that is greater than both udev's
5589 and systemd's most recent version number.
5591 * udev: all udev sources are merged into the systemd source tree now.
5592 All future udev development will happen in the systemd tree. It
5593 is still fully supported to use the udev daemon and tools without
5594 systemd running, like in initramfs or other init systems. Building
5595 udev though, will require the *build* of the systemd tree, but
5596 udev can be properly *run* without systemd.
5598 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5599 should be used to create dead device nodes as workarounds for broken
5602 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5603 no longer supported. udev_monitor_new_from_netlink() needs to be
5604 used to subscribe to events.
5606 * udev: when udevd is started by systemd, processes which are left
5607 behind by forking them off of udev rules, are unconditionally cleaned
5608 up and killed now after the event handling has finished. Services or
5609 daemons must be started as systemd services. Services can be
5610 pulled-in by udev to get started, but they can no longer be directly
5611 forked by udev rules.
5613 * udev: the daemon binary is called systemd-udevd now and installed
5614 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5615 to adapt to that, create symlink, or rename the binary after building
5618 * libudev no longer provides these symbols:
5619 udev_monitor_from_socket()
5620 udev_queue_get_failed_list_entry()
5621 udev_get_{dev,sys,run}_path()
5622 The versions number was bumped and symbol versioning introduced.
5624 * systemd-loginctl and systemd-journalctl have been renamed
5625 to loginctl and journalctl to match systemctl.
5627 * The config files: /etc/systemd/systemd-logind.conf and
5628 /etc/systemd/systemd-journald.conf have been renamed to
5629 logind.conf and journald.conf. Package updates should rename
5630 the files to the new names on upgrade.
5632 * For almost all files the license is now LGPL2.1+, changed
5633 from the previous GPL2.0+. Exceptions are some minor stuff
5634 of udev (which will be changed to LGPL2.1 eventually, too),
5635 and the MIT licensed sd-daemon.[ch] library that is suitable
5636 to be used as drop-in files.
5638 * systemd and logind now handle system sleep states, in
5639 particular suspending and hibernating.
5641 * logind now implements a sleep/shutdown/idle inhibiting logic
5642 suitable for a variety of uses. Soonishly Lennart will blog
5643 about this in more detail.
5645 * var-run.mount and var-lock.mount are no longer provided
5646 (which prevously bind mounted these directories to their new
5647 places). Distributions which have not converted these
5648 directories to symlinks should consider stealing these files
5649 from git history and add them downstream.
5651 * We introduced the Documentation= field for units and added
5652 this to all our shipped units. This is useful to make it
5653 easier to explore the boot and the purpose of the various
5656 * All smaller setup units (such as
5657 systemd-vconsole-setup.service) now detect properly if they
5658 are run in a container and are skipped when
5659 appropriate. This guarantees an entirely noise-free boot in
5660 Linux container environments such as systemd-nspawn.
5662 * A framework for implementing offline system updates is now
5663 integrated, for details see:
5664 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5666 * A new service type Type=idle is available now which helps us
5667 avoiding ugly interleaving of getty output and boot status
5670 * There's now a system-wide CapabilityBoundingSet= option to
5671 globally reduce the set of capabilities for the
5672 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5673 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5674 even CAP_NET_ADMIN system-wide for secure systems.
5676 * There are now system-wide DefaultLimitXXX= options to
5677 globally change the defaults of the various resource limits
5678 for all units started by PID 1.
5680 * Harald Hoyer's systemd test suite has been integrated into
5681 systemd which allows easy testing of systemd builds in qemu
5682 and nspawn. (This is really awesome! Ask us for details!)
5684 * The fstab parser is now implemented as generator, not inside
5687 * systemctl will now warn you if .mount units generated from
5688 /etc/fstab are out of date due to changes in fstab that
5689 have not been read by systemd yet.
5691 * systemd is now suitable for usage in initrds. Dracut has
5692 already been updated to make use of this. With this in place
5693 initrds get a slight bit faster but primarily are much
5694 easier to introspect and debug since "systemctl status" in
5695 the host system can be used to introspect initrd services,
5696 and the journal from the initrd is kept around too.
5698 * systemd-delta has been added, a tool to explore differences
5699 between user/admin configuration and vendor defaults.
5701 * PrivateTmp= now affects both /tmp and /var/tmp.
5703 * Boot time status messages are now much prettier and feature
5704 proper english language. Booting up systemd has never been
5707 * Read-ahead pack files now include the inode number of all
5708 files to pre-cache. When the inode changes the pre-caching
5709 is not attempted. This should be nicer to deal with updated
5710 packages which might result in changes of read-ahead
5713 * We now temporaritly lower the kernel's read_ahead_kb variable
5714 when collecting read-ahead data to ensure the kernel's
5715 built-in read-ahead does not add noise to our measurements
5716 of necessary blocks to pre-cache.
5718 * There's now RequiresMountsFor= to add automatic dependencies
5719 for all mounts necessary for a specific file system path.
5721 * MountAuto= and SwapAuto= have been removed from
5722 system.conf. Mounting file systems at boot has to take place
5725 * nspawn now learned a new switch --uuid= to set the machine
5726 ID on the command line.
5728 * nspawn now learned the -b switch to automatically search
5731 * vt102 is now the default TERM for serial TTYs, upgraded from
5734 * systemd-logind now works on VT-less systems.
5736 * The build tree has been reorganized. The individual
5737 components now have directories of their own.
5739 * A new condition type ConditionPathIsReadWrite= is now available.
5741 * nspawn learned the new -C switch to create cgroups for the
5742 container in other hierarchies.
5744 * We now have support for hardware watchdogs, configurable in
5747 * The scheduled shutdown logic now has a public API.
5749 * We now mount /tmp as tmpfs by default, but this can be
5750 masked and /etc/fstab can override it.
5752 * Since udisks does not make use of /media anymore we are not
5753 mounting a tmpfs on it anymore.
5755 * journalctl gained a new --local switch to only interleave
5756 locally generated journal files.
5758 * We can now load the IMA policy at boot automatically.
5760 * The GTK tools have been split off into a systemd-ui.
5762 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5763 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5764 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5765 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5766 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5767 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5768 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5769 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5770 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5775 * This is mostly a bugfix release
5777 * Support optional initialization of the machine ID from the
5778 KVM or container configured UUID.
5780 * Support immediate reboots with "systemctl reboot -ff"
5782 * Show /etc/os-release data in systemd-analyze output
5784 * Many bugfixes for the journal, including endianness fixes and
5785 ensuring that disk space enforcement works
5787 * sd-login.h is C++ comptaible again
5789 * Extend the /etc/os-release format on request of the Debian
5792 * We now refuse non-UTF8 strings used in various configuration
5793 and unit files. This is done to ensure we do not pass invalid
5794 data over D-Bus or expose it elsewhere.
5796 * Register Mimo USB Screens as suitable for automatic seat
5799 * Read SELinux client context from journal clients in a race
5802 * Reorder configuration file lookup order. /etc now always
5803 overrides /run in order to allow the administrator to always
5804 and unconditionally override vendor-supplied or
5805 automatically generated data.
5807 * The various user visible bits of the journal now have man
5808 pages. We still lack man pages for the journal API calls
5811 * We now ship all man pages in HTML format again in the
5814 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5815 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5816 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5821 * This is mostly a bugfix release
5823 * systems lacking /etc/os-release are no longer supported.
5825 * Various functionality updates to libsystemd-login.so
5827 * Track class of PAM logins to distinguish greeters from
5830 Contributions from: Kay Sievers, Lennart Poettering, Michael
5835 * This is an important bugfix release for v41.
5837 * Building man pages is now optional which should be useful
5838 for those building systemd from git but unwilling to install
5841 * Watchdog support for supervising services is now usable. In
5842 a future release support for hardware watchdogs
5843 (i.e. /dev/watchdog) will be added building on this.
5845 * Service start rate limiting is now configurable and can be
5846 turned off per service. When a start rate limit is hit a
5847 reboot can automatically be triggered.
5849 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5851 Contributions from: Benjamin Franzke, Bill Nottingham,
5852 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5853 Schmidt, Michał Górny, Piotr Drąg
5857 * The systemd binary is installed /usr/lib/systemd/systemd now;
5858 An existing /sbin/init symlink needs to be adapted with the
5861 * The code that loads kernel modules has been ported to invoke
5862 libkmod directly, instead of modprobe. This means we do not
5863 support systems with module-init-tools anymore.
5865 * Watchdog support is now already useful, but still not
5868 * A new kernel command line option systemd.setenv= is
5869 understood to set system wide environment variables
5870 dynamically at boot.
5872 * We now limit the set of capabilities of systemd-journald.
5874 * We now set SIGPIPE to ignore by default, since it only is
5875 useful in shell pipelines, and has little use in general
5876 code. This can be disabled with IgnoreSIPIPE=no in unit
5879 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5880 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5885 * This is mostly a bugfix release
5887 * We now expose the reason why a service failed in the
5888 "Result" D-Bus property.
5890 * Rudimentary service watchdog support (will be completed over
5891 the next few releases.)
5893 * When systemd forks off in order execute some service we will
5894 now immediately changes its argv[0] to reflect which process
5895 it will execute. This is useful to minimize the time window
5896 with a generic argv[0], which makes bootcharts more useful
5898 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5899 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5900 Mike Kazantsev, Ray Strode
5904 * This is mostly a test release, but incorporates many
5907 * New systemd-cgtop tool to show control groups by their
5910 * Linking against libacl for ACLs is optional again. If
5911 disabled, support tracking device access for active logins
5912 goes becomes unavailable, and so does access to the user
5913 journals by the respective users.
5915 * If a group "adm" exists, journal files are automatically
5916 owned by them, thus allow members of this group full access
5917 to the system journal as well as all user journals.
5919 * The journal now stores the SELinux context of the logging
5920 client for all entries.
5922 * Add C++ inclusion guards to all public headers
5924 * New output mode "cat" in the journal to print only text
5925 messages, without any meta data like date or time.
5927 * Include tiny X server wrapper as a temporary stop-gap to
5928 teach XOrg udev display enumeration. This is used by display
5929 managers such as gdm, and will go away as soon as XOrg
5930 learned native udev hotplugging for display devices.
5932 * Add new systemd-cat tool for executing arbitrary programs
5933 with STDERR/STDOUT connected to the journal. Can also act as
5934 BSD logger replacement, and does so by default.
5936 * Optionally store all locally generated coredumps in the
5937 journal along with meta data.
5939 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5940 writing short strings to files (for usage for /sys), and for
5941 creating symlinks, character and block device nodes.
5943 * New unit file option ControlGroupPersistent= to make cgroups
5944 persistent, following the mechanisms outlined in
5945 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5947 * Support multiple local RTCs in a sane way
5949 * No longer monopolize IO when replaying readahead data on
5950 rotating disks, since we might starve non-file-system IO to
5951 death, since fanotify() will not see accesses done by blkid,
5954 * Do not show kernel threads in systemd-cgls anymore, unless
5955 requested with new -k switch.
5957 Contributions from: Dan Horák, Kay Sievers, Lennart
5958 Poettering, Michal Schmidt
5962 * This is mostly a test release, but incorporates many
5965 * The git repository moved to:
5966 git://anongit.freedesktop.org/systemd/systemd
5967 ssh://git.freedesktop.org/git/systemd/systemd
5969 * First release with the journal
5970 http://0pointer.de/blog/projects/the-journal.html
5972 * The journal replaces both systemd-kmsg-syslogd and
5973 systemd-stdout-bridge.
5975 * New sd_pid_get_unit() API call in libsystemd-logind
5977 * Many systemadm clean-ups
5979 * Introduce remote-fs-pre.target which is ordered before all
5980 remote mounts and may be used to start services before all
5983 * Added Mageia support
5985 * Add bash completion for systemd-loginctl
5987 * Actively monitor PID file creation for daemons which exit in
5988 the parent process before having finished writing the PID
5989 file in the daemon process. Daemons which do this need to be
5990 fixed (i.e. PID file creation must have finished before the
5991 parent exits), but we now react a bit more gracefully to them.
5993 * Add colourful boot output, mimicking the well-known output
5994 of existing distributions.
5996 * New option PassCredentials= for socket units, for
5997 compatibility with a recent kernel ABI breakage.
5999 * /etc/rc.local is now hooked in via a generator binary, and
6000 thus will no longer act as synchronization point during
6003 * systemctl list-unit-files now supports --root=.
6005 * systemd-tmpfiles now understands two new commands: z, Z for
6006 relabelling files according to the SELinux database. This is
6007 useful to apply SELinux labels to specific files in /sys,
6010 * Output of SysV services is now forwarded to both the console
6011 and the journal by default, not only just the console.
6013 * New man pages for all APIs from libsystemd-login.
6015 * The build tree got reorganized and a the build system is a
6016 lot more modular allowing embedded setups to specifically
6017 select the components of systemd they are interested in.
6019 * Support for Linux systems lacking the kernel VT subsystem is
6022 * configure's --with-rootdir= got renamed to
6023 --with-rootprefix= to follow the naming used by udev and
6026 * Unless specified otherwise we will now install to /usr instead
6027 of /usr/local by default.
6029 * Processes with '@' in argv[0][0] are now excluded from the
6030 final shut-down killing spree, following the logic explained
6032 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6034 * All processes remaining in a service cgroup when we enter
6035 the START or START_PRE states are now killed with
6036 SIGKILL. That means it is no longer possible to spawn
6037 background processes from ExecStart= lines (which was never
6038 supported anyway, and bad style).
6040 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6041 reloading of units together.
6043 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6044 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6045 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6046 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6047 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob