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. (Note that you have to enable
21 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
22 and its DNSSEC mode for host name resolution from local
25 * systemd-resolve conveniently resolves DANE records with the --tlsa
26 option and OPENPGPKEY records with the --openpgp option. It also
27 supports dumping raw DNS record data via the new --raw= switch now.
29 * systemd-logind will now by default terminate user processes that are
30 part of the user session scope unit (session-XX.scope) when the user
31 logs out. This behavior is controlled by the KillUserProcesses=
32 setting in logind.conf, and the previous default of "no" is now
33 changed to "yes". This means that user sessions will be properly
34 cleaned up after, but additional steps are necessary to allow
35 intentionally long-running processes to survive logout.
37 While the user is logged in at least once, user@.service is running,
38 and any service that should survive the end of any individual login
39 session can be started at a user service or scope using systemd-run.
40 systemd-run(1) man page has been extended with an example which shows
41 how to run screen in a scope unit underneath user@.service. The same
42 command works for tmux.
44 After the user logs out of all sessions, user@.service will be
45 terminated too, by default, unless the user has "lingering" enabled.
46 To effectively allow users to run long-term tasks even if they are
47 logged out, lingering must be enabled for them. See loginctl(1) for
48 details. The default polkit policy was modified to allow users to
49 set lingering for themselves without authentication.
51 Previous defaults can be restored at compile time by the
52 --without-kill-user-processes option to "configure".
54 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
55 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
58 WARNING: it is not possible to use previous systemd versions with
59 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
60 is necessary to also update systemd in the initramfs if using the
61 unified hierarchy. An updated SELinux policy is also required.
63 * LLDP support has been extended, and both passive (receive-only) and
64 active (sender) modes are supported. Passive mode ("routers-only") is
65 enabled by default in systemd-networkd. Active LLDP mode is enabled
66 by default for containers on the internal network. The "networkctl
67 lldp" command may be used to list information gathered. "networkctl
68 status" will also show basic LLDP information on connected peers now.
70 * The IAID and DUID unique identifier sent in DHCP requests may now be
71 configured for the system and each .network file managed by
74 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
75 systemd-socket-activate and installed into /usr/bin. It is now fully
78 * systemd-journald now uses separate threads to flush changes to disk
79 when closing journal files, thus reducing impact of slow disk I/O on
82 * systemd-ask-password now optionally skips printing of the password to
83 stdout with --no-output which can be useful in scripts.
85 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
86 (devices tagged with ID_MAKER_TOOL) are now tagged with
87 "uaccess" and are available to logged in users.
89 * systemd-bootchart has been split out to a separate repository:
90 https://github.com/systemd/systemd-bootchart
92 * The compatibility libraries libsystemd-daemon.so,
93 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
94 which have been deprecated since systemd-209 have been removed along
95 with the corresponding pkg-config files. All symbols provided by
96 those libraries are provided by libsystemd.so.
98 * The Capabilities= unit file setting has been removed (it is ignored
99 for backwards compatibility). AmbientCapabilities= and
100 CapabilityBoundingSet= should be used instead.
102 * "systemctl show" gained a new --value switch, which allows print a
103 only the contents of a specific unit property, without also printing
104 the property's name. Similar support was added to "show*" verbs
105 of loginctl and machinectl that output "key=value" lists.
107 * A new command "systemctl revert" has been added that may be used to
108 revert to the vendor version of a unit file, in case local changes
109 have been made by adding drop-ins or overriding the unit file.
111 * "machinectl clean" gained a new verb to automatically remove all or
112 just hidden container images.
114 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
115 merged into the kernel in its current form.
117 * systemd-networkd gained support for configuring proxy ARP support for
118 each interface, via the ProxyArp= setting in .network files. It also
119 gained support for configuring the multicast querier feature of
120 bridge devices, via the new MulticastQuerier= setting in .netdev
121 files. A new setting PreferredLifetime= has been added for addresses
122 configured in .network file to configure the lifetime intended for an
125 * systemd-tmpfiles gained support for a new line type "e" for emptying
126 directories, if they exist, without creating them if they don't.
128 * journalctl learned a new output mode "-o short-unix" that outputs log
129 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
130 UTC). It also gained support for a new --no-hostname setting to
131 suppress the hostname column in the family of "short" output modes.
133 * systemd-nspawn gained support for automatically patching the UID/GIDs
134 of the owners and the ACLs of all files and directories in a
135 container tree to match the UID/GID user namespacing range selected
136 for the container invocation. This mode is enabled via the new
137 --private-user-chown switch. It also gained support for automatically
138 choosing a free, previously unused UID/GID range when starting a
139 container, via the new --private-users=pick setting (which implies
140 --private-user-chown). Together, these options for the first time
141 make user namespacing for nspawn containers fully automatic and thus
142 deployable. The systemd-nspaw@.service template unit file has been
143 changed to use this functionality by default.
145 * The default start timeout may now be configured on the kernel command
146 line via systemd.default_timeout_start_sec=. It was already
147 configurable via the DefaultTimeoutStartSec= option in
148 /etc/systemd/system.conf.
150 * Socket units gained a new TriggerLimitIntervalSec= and
151 TriggerLimitBurst= setting to configure a limit on the activation
152 rate of the socket unit.
154 * The LimitNICE= setting now optionally takes normal UNIX nice values
155 in addition to the raw integer limit value. If the specified
156 parameter is prefixed with "+" or "-" and is in the range -20..19 the
157 value is understood as UNIX nice value. If not prefixed like this it
158 is understood as raw RLIMIT_NICE limit.
160 Contributions from: Alban Crequy, Alexander Kuleshov, Alex Crawford,
161 Andrew Eikum, Beniamino Galvani, Benjamin Robin, Benjamin ROBIN, Biao
162 Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Colin Guthrie, Daniel
163 J Walsh, Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
164 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
165 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
166 Bui, frankheckenbach, Georgia Brikis, Harald Hoyer, Hendrik Brueckner,
167 Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo Puustinen, Jakub
168 Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth, kayrus, Klearchos
169 Chaloulos, Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukáš
170 Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt, Michael Biebl,
171 michaelolbrich, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletar,
172 Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin, mulkieran,
173 muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween, Nicolas
174 Braud-Santoni, Patrik Flykt, Peter Hutterer, Petr Lautrbach, Petros
175 Angelatos, Piotr Drąg, Rabin Vincent, Robert Węcławski, Ronny
176 Chevalier, Samuel Tardieu, Stefan Schallenberg, Steven Siloti, Susant
177 Sahani, Sylvain Plantefève, Taylor Smock, tblume, Tejun Heo, Thomas
178 Blume, Thomas Haller, Thomas Hindoe Paaboel Andersen, Thomas
179 H. P. Andersen, Tobias Klauser, Tom Gundersen, Torstein Husebø, Umut
180 Tezduyar Lindskog, Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam),
181 Vladimir Panteleev, Wieland Hoffmann, Wouter Verhelst, Yu Watanabe,
182 Zbigniew Jędrzejewski-Szmek
188 * The systemd-resolved DNS resolver service has gained a substantial
189 set of new features, most prominently it may now act as a DNSSEC
190 validating stub resolver. DNSSEC mode is currently turned off by
191 default, but is expected to be turned on by default in one of the
192 next releases. For now, we invite everybody to test the DNSSEC logic
193 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
194 service also gained a full set of D-Bus interfaces, including calls
195 to configure DNS and DNSSEC settings per link (for use by external
196 network management software). systemd-resolved and systemd-networkd
197 now distinguish between "search" and "routing" domains. The former
198 are used to qualify single-label names, the latter are used purely
199 for routing lookups within certain domains to specific links.
200 resolved now also synthesizes RRs for all entries from /etc/hosts.
202 * The systemd-resolve tool (which is a client utility for
203 systemd-resolved) has been improved considerably and is now fully
204 supported and documented. Hence it has moved from /usr/lib/systemd to
207 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
210 * The coredump collection logic has been reworked: when a coredump is
211 collected it is now written to disk, compressed and processed
212 (including stacktrace extraction) from a new instantiated service
213 systemd-coredump@.service, instead of directly from the
214 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
215 processing large coredumps can take up a substantial amount of
216 resources and time, and this previously happened entirely outside of
217 systemd's service supervision. With the new logic the core_pattern
218 hook only does minimal metadata collection before passing off control
219 to the new instantiated service, which is configured with a time
220 limit, a nice level and other settings to minimize negative impact on
221 the rest of the system. Also note that the new logic will honour the
222 RLIMIT_CORE setting of the crashed process, which now allows users
223 and processes to turn off coredumping for their processes by setting
226 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
227 and all forked processes by default. Previously, PID 1 would leave
228 the setting at "0" for all processes, as set by the kernel. Note that
229 the resource limit traditionally has no effect on the generated
230 coredumps on the system if the /proc/sys/kernel/core_pattern hook
231 logic is used. Since the limit is now honoured (see above) its
232 default has been changed so that the coredumping logic is enabled by
233 default for all processes, while allowing specific opt-out.
235 * When the stacktrace is extracted from processes of system users, this
236 is now done as "systemd-coredump" user, in order to sandbox this
237 potentially security sensitive parsing operation. (Note that when
238 processing coredumps of normal users this is done under the user ID
239 of process that crashed, as before.) Packagers should take notice
240 that it is now necessary to create the "systemd-coredump" system user
241 and group at package installation time.
243 * The systemd-activate socket activation testing tool gained support
244 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
245 and --seqpacket switches. It also has been extended to support both
246 new-style and inetd-style file descriptor passing. Use the new
247 --inetd switch to request inetd-style file descriptor passing.
249 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
250 variable, which takes a boolean value. If set to false, ANSI color
251 output is disabled in the tools even when run on a terminal that
254 * The VXLAN support in networkd now supports two new settings
255 DestinationPort= and PortRange=.
257 * A new systemd.machine_id= kernel command line switch has been added,
258 that may be used to set the machine ID in /etc/machine-id if it is
259 not initialized yet. This command line option has no effect if the
260 file is already initialized.
262 * systemd-nspawn gained a new --as-pid2 switch that invokes any
263 specified command line as PID 2 rather than PID 1 in the
264 container. In this mode PID 1 is a minimal stub init process that
265 implements the special POSIX and Linux semantics of PID 1 regarding
266 signal and child process management. Note that this stub init process
267 is implemented in nspawn itself and requires no support from the
268 container image. This new logic is useful to support running
269 arbitrary commands in the container, as normal processes are
270 generally not prepared to run as PID 1.
272 * systemd-nspawn gained a new --chdir= switch for setting the current
273 working directory for the process started in the container.
275 * "journalctl /dev/sda" will now output all kernel log messages for
276 specified device from the current boot, in addition to all devices
277 that are parents of it. This should make log output about devices
278 pretty useful, as long as kernel drivers attach enough metadata to
279 the log messages. (The usual SATA drivers do.)
281 * The sd-journal API gained two new calls
282 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
283 that report whether log data from /run or /var has been found.
285 * journalctl gained a new switch "--fields" that prints all journal
286 record field names currently in use in the journal. This is backed
287 by two new sd-journal API calls sd_journal_enumerate_fields() and
288 sd_journal_restart_fields().
290 * Most configurable timeouts in systemd now expect an argument of
291 "infinity" to turn them off, instead of "0" as before. The semantics
292 from now on is that a timeout of "0" means "now", and "infinity"
293 means "never". To maintain backwards compatibility, "0" continues to
294 turn off previously existing timeout settings.
296 * "systemctl reload-or-try-restart" has been renamed to "systemctl
297 try-reload-or-restart" to clarify what it actually does: the "try"
298 logic applies to both reloading and restarting, not just restarting.
299 The old name continues to be accepted for compatibility.
301 * On boot-up, when PID 1 detects that the system clock is behind the
302 release date of the systemd version in use, the clock is now set
303 to the latter. Previously, this was already done in timesyncd, in order
304 to avoid running with clocks set to the various clock epochs such as
305 1902, 1938 or 1970. With this change the logic is now done in PID 1
306 in addition to timesyncd during early boot-up, so that it is enforced
307 before the first process is spawned by systemd. Note that the logic
308 in timesyncd remains, as it is more comprehensive and ensures
309 clock monotonicity by maintaining a persistent timestamp file in
310 /var. Since /var is generally not available in earliest boot or the
311 initrd, this part of the logic remains in timesyncd, and is not done
314 * Support for tweaking details in net_cls.class_id through the
315 NetClass= configuration directive has been removed, as the kernel
316 people have decided to deprecate that controller in cgroup v2.
317 Userspace tools such as nftables are moving over to setting rules
318 that are specific to the full cgroup path of a task, which obsoletes
319 these controllers anyway. The NetClass= directive is kept around for
320 legacy compatibility reasons. For a more in-depth description of the
321 kernel change, please refer to the respective upstream commit:
323 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
325 * A new service setting RuntimeMaxSec= has been added that may be used
326 to specify a maximum runtime for a service. If the timeout is hit, the
327 service is terminated and put into a failure state.
329 * A new service setting AmbientCapabilities= has been added. It allows
330 configuration of additional Linux process capabilities that are
331 passed to the activated processes. This is only available on very
334 * The process resource limit settings in service units may now be used
335 to configure hard and soft limits individually.
337 * The various libsystemd APIs such as sd-bus or sd-event now publicly
338 expose support for gcc's __attribute__((cleanup())) C extension.
339 Specifically, for many object destructor functions alternative
340 versions have been added that have names suffixed with "p" and take a
341 pointer to a pointer to the object to destroy, instead of just a
342 pointer to the object itself. This is useful because these destructor
343 functions may be used directly as parameters to the cleanup
344 construct. Internally, systemd has been a heavy user of this GCC
345 extension for a long time, and with this change similar support is
346 now available to consumers of the library outside of systemd. Note
347 that by using this extension in your sources compatibility with old
348 and strictly ANSI compatible C compilers is lost. However, all gcc or
349 LLVM versions of recent years support this extension.
351 * Timer units gained support for a new setting RandomizedDelaySec= that
352 allows configuring some additional randomized delay to the configured
353 time. This is useful to spread out timer events to avoid load peaks in
354 clusters or larger setups.
356 * Calendar time specifications now support sub-second accuracy.
358 * Socket units now support listening on SCTP and UDP-lite protocol
361 * The sd-event API now comes with a full set of man pages.
363 * Older versions of systemd contained experimental support for
364 compressing journal files and coredumps with the LZ4 compressor that
365 was not compatible with the lz4 binary (due to API limitations of the
366 lz4 library). This support has been removed; only support for files
367 compatible with the lz4 binary remains. This LZ4 logic is now
368 officially supported and no longer considered experimental.
370 * The dkr image import logic has been removed again from importd. dkr's
371 micro-services focus doesn't fit into the machine image focus of
372 importd, and quickly got out of date with the upstream dkr API.
374 * Creation of the /run/lock/lockdev/ directory was dropped from
375 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
376 been available for many years. If you still need this, you need to
377 create your own tmpfiles.d config file with:
379 d /run/lock/lockdev 0775 root lock -
381 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
382 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
383 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
384 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
385 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
386 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
387 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
388 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
389 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
390 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
391 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
392 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
393 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
394 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
395 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
396 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
397 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
398 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
399 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
405 * A number of properties previously only settable in unit
406 files are now also available as properties to set when
407 creating transient units programmatically via the bus, as it
408 is exposed with systemd-run's --property=
409 setting. Specifically, these are: SyslogIdentifier=,
410 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
411 EnvironmentFile=, ReadWriteDirectories=,
412 ReadOnlyDirectories=, InaccessibleDirectories=,
413 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
415 * When creating transient services via the bus API it is now
416 possible to pass in a set of file descriptors to use as
417 STDIN/STDOUT/STDERR for the invoked process.
419 * Slice units may now be created transiently via the bus APIs,
420 similar to the way service and scope units may already be
423 * Wherever systemd expects a calendar timestamp specification
424 (like in journalctl's --since= and --until= switches) UTC
425 timestamps are now supported. Timestamps suffixed with "UTC"
426 are now considered to be in Universal Time Coordinated
427 instead of the local timezone. Also, timestamps may now
428 optionally be specified with sub-second accuracy. Both of
429 these additions also apply to recurring calendar event
430 specification, such as OnCalendar= in timer units.
432 * journalctl gained a new "--sync" switch that asks the
433 journal daemon to write all so far unwritten log messages to
434 disk and sync the files, before returning.
436 * systemd-tmpfiles learned two new line types "q" and "Q" that
437 operate like "v", but also set up a basic btrfs quota
438 hierarchy when used on a btrfs file system with quota
441 * tmpfiles' "v", "q" and "Q" will now create a plain directory
442 instead of a subvolume (even on a btrfs file system) if the
443 root directory is a plain directory, and not a
444 subvolume. This should simplify things with certain chroot()
445 environments which are not aware of the concept of btrfs
448 * systemd-detect-virt gained a new --chroot switch to detect
449 whether execution takes place in a chroot() environment.
451 * CPUAffinity= now takes CPU index ranges in addition to
454 * The various memory-related resource limit settings (such as
455 LimitAS=) now understand the usual K, M, G, ... suffixes to
456 the base of 1024 (IEC). Similar, the time-related resource
457 limit settings understand the usual min, h, day, ...
460 * There's a new system.conf setting DefaultTasksMax= to
461 control the default TasksMax= setting for services and
462 scopes running on the system. (TasksMax= is the primary
463 setting that exposes the "pids" cgroup controller on systemd
464 and was introduced in the previous systemd release.) The
465 setting now defaults to 512, which means services that are
466 not explicitly configured otherwise will only be able to
467 create 512 processes or threads at maximum, from this
468 version on. Note that this means that thread- or
469 process-heavy services might need to be reconfigured to set
470 TasksMax= to a higher value. It is sufficient to set
471 TasksMax= in these specific unit files to a higher value, or
472 even "infinity". Similar, there's now a logind.conf setting
473 UserTasksMax= that defaults to 4096 and limits the total
474 number of processes or tasks each user may own
475 concurrently. nspawn containers also have the TasksMax=
476 value set by default now, to 8192. Note that all of this
477 only has an effect if the "pids" cgroup controller is
478 enabled in the kernel. The general benefit of these changes
479 should be a more robust and safer system, that provides a
480 certain amount of per-service fork() bomb protection.
482 * systemd-nspawn gained the new --network-veth-extra= switch
483 to define additional and arbitrarily-named virtual Ethernet
484 links between the host and the container.
486 * A new service execution setting PassEnvironment= has been
487 added that allows importing select environment variables
488 from PID1's environment block into the environment block of
491 * Timer units gained support for a new RemainAfterElapse=
492 setting which takes a boolean argument. It defaults on on,
493 exposing behaviour unchanged to previous releases. If set to
494 off, timer units are unloaded after they elapsed if they
495 cannot elapse again. This is particularly useful for
496 transient timer units, which shall not stay around longer
497 than until they first elapse.
499 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
500 default now (the kernel default is 16). This is beneficial
501 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
502 allows substantially larger numbers of queued
503 datagrams. This should increase the capability of systemd to
504 parallelize boot-up, as logging and sd_notify() are unlikely
505 to stall execution anymore. If you need to change the value
506 from the new defaults, use the usual sysctl.d/ snippets.
508 * The compression framing format used by the journal or
509 coredump processing has changed to be in line with what the
510 official LZ4 tools generate. LZ4 compression support in
511 systemd was considered unsupported previously, as the format
512 was not compatible with the normal tools. With this release
513 this has changed now, and it is hence safe for downstream
514 distributions to turn it on. While not compressing as well
515 as the XZ, LZ4 is substantially faster, which makes
516 it a good default choice for the compression logic in the
517 journal and in coredump handling.
519 * Any reference to /etc/mtab has been dropped from
520 systemd. The file has been obsolete since a while, but
521 systemd refused to work on systems where it was incorrectly
522 set up (it should be a symlink or non-existent). Please make
523 sure to update to util-linux 2.27.1 or newer in conjunction
524 with this systemd release, which also drops any reference to
525 /etc/mtab. If you maintain a distribution make sure that no
526 software you package still references it, as this is a
527 likely source of bugs. There's also a glibc bug pending,
528 asking for removal of any reference to this obsolete file:
530 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
532 Note that only util-linux versions built with
533 --enable-libmount-force-mountinfo are supported.
535 * Support for the ".snapshot" unit type has been removed. This
536 feature turned out to be little useful and little used, and
537 has now been removed from the core and from systemctl.
539 * The dependency types RequiresOverridable= and
540 RequisiteOverridable= have been removed from systemd. They
541 have been used only very sparingly to our knowledge and
542 other options that provide a similar effect (such as
543 systemctl --mode=ignore-dependencies) are much more useful
544 and commonly used. Moreover, they were only half-way
545 implemented as the option to control behaviour regarding
546 these dependencies was never added to systemctl. By removing
547 these dependency types the execution engine becomes a bit
548 simpler. Unit files that use these dependencies should be
549 changed to use the non-Overridable dependency types
550 instead. In fact, when parsing unit files with these
551 options, that's what systemd will automatically convert them
552 too, but it will also warn, asking users to fix the unit
553 files accordingly. Removal of these dependency types should
554 only affect a negligible number of unit files in the wild.
556 * Behaviour of networkd's IPForward= option changed
557 (again). It will no longer maintain a per-interface setting,
558 but propagate one way from interfaces where this is enabled
559 to the global kernel setting. The global setting will be
560 enabled when requested by a network that is set up, but
561 never be disabled again. This change was made to make sure
562 IPv4 and IPv6 behaviour regarding packet forwarding is
563 similar (as the Linux IPv6 stack does not support
564 per-interface control of this setting) and to minimize
567 * In unit files the behaviour of %u, %U, %h, %s has
568 changed. These specifiers will now unconditionally resolve
569 to the various user database fields of the user that the
570 systemd instance is running as, instead of the user
571 configured in the specific unit via User=. Note that this
572 effectively doesn't change much, as resolving of these
573 specifiers was already turned off in the --system instance
574 of systemd, as we cannot do NSS lookups from PID 1. In the
575 --user instance of systemd these specifiers where correctly
576 resolved, but hardly made any sense, since the user instance
577 lacks privileges to do user switches anyway, and User= is
578 hence useless. Morever, even in the --user instance of
579 systemd behaviour was awkward as it would only take settings
580 from User= assignment placed before the specifier into
581 account. In order to unify and simplify the logic around
582 this the specifiers will now always resolve to the
583 credentials of the user invoking the manager (which in case
584 of PID 1 is the root user).
586 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
587 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
588 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
589 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
590 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
591 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
592 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
593 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
594 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
595 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
596 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
597 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
598 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
599 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
606 * systemd now depends on util-linux v2.27. More specifically,
607 the newly added mount monitor feature in libmount now
608 replaces systemd's former own implementation.
610 * libmount mandates /etc/mtab not to be regular file, and
611 systemd now enforces this condition at early boot.
612 /etc/mtab has been deprecated and warned about for a very
613 long time, so systems running systemd should already have
614 stopped having this file around as anything else than a
615 symlink to /proc/self/mounts.
617 * Support for the "pids" cgroup controller has been added. It
618 allows accounting the number of tasks in a cgroup and
619 enforcing limits on it. This adds two new setting
620 TasksAccounting= and TasksMax= to each unit, as well as a
621 global option DefaultTasksAccounting=.
623 * Support for the "net_cls" cgroup controller has been added.
624 It allows assigning a net class ID to each task in the
625 cgroup, which can then be used in firewall rules and traffic
626 shaping configurations. Note that the kernel netfilter net
627 class code does not currently work reliably for ingress
628 packets on unestablished sockets.
630 This adds a new config directive called NetClass= to CGroup
631 enabled units. Allowed values are positive numbers for fixed
632 assignments and "auto" for picking a free value
635 * 'systemctl is-system-running' now returns 'offline' if the
636 system is not booted with systemd. This command can now be
637 used as a substitute for 'systemd-notify --booted'.
639 * Watchdog timeouts have been increased to 3 minutes for all
640 in-tree service files. Apparently, disk IO issues are more
641 frequent than we hoped, and user reported >1 minute waiting
644 * 'machine-id-commit' functionality has been merged into
645 'machine-id-setup --commit'. The separate binary has been
648 * The WorkingDirectory= directive in unit files may now be set
649 to the special value '~'. In this case, the working
650 directory is set to the home directory of the user
653 * "machinectl shell" will now open the shell in the home
654 directory of the selected user by default.
656 * The CrashChVT= configuration file setting is renamed to
657 CrashChangeVT=, following our usual logic of not
658 abbreviating unnecessarily. The old directive is still
659 supported for compat reasons. Also, this directive now takes
660 an integer value between 1 and 63, or a boolean value. The
661 formerly supported '-1' value for disabling stays around for
664 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
665 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
666 RootDirectory= properties can now be set for transient
669 * The systemd-analyze tool gained a new "set-log-target" verb
670 to change the logging target the system manager logs to
671 dynamically during runtime. This is similar to how
672 "systemd-analyze set-log-level" already changes the log
675 * In nspawn /sys is now mounted as tmpfs, with only a selected
676 set of subdirectories mounted in from the real sysfs. This
677 enhances security slightly, and is useful for ensuring user
678 namespaces work correctly.
680 * Support for USB FunctionFS activation has been added. This
681 allows implementation of USB gadget services that are
682 activated as soon as they are requested, so that they don't
683 have to run continously, similar to classic socket
686 * The "systemctl exit" command now optionally takes an
687 additional parameter that sets the exit code to return from
688 the systemd manager when exiting. This is only relevant when
689 running the systemd user instance, or when running the
690 system instance in a container.
692 * sd-bus gained the new API calls sd_bus_path_encode_many()
693 and sd_bus_path_decode_many() that allow easy encoding and
694 decoding of multiple identifier strings inside a D-Bus
695 object path. Another new call sd_bus_default_flush_close()
696 has been added to flush and close per-thread default
699 * systemd-cgtop gained support for a -M/--machine= switch to
700 show the control groups within a certain container only.
702 * "systemctl kill" gained support for an optional --fail
703 switch. If specified the requested operation will fail of no
704 processes have been killed, because the unit had no
705 processes attached, or similar.
707 * A new systemd.crash_reboot=1 kernel command line option has
708 been added that triggers a reboot after crashing. This can
709 also be set through CrashReboot= in systemd.conf.
711 * The RuntimeDirectory= setting now understands unit
712 specifiers like %i or %f.
714 * A new (still internal) libary API sd-ipv4acd has been added,
715 that implements address conflict detection for IPv4. It's
716 based on code from sd-ipv4ll, and will be useful for
717 detecting DHCP address conflicts.
719 * File descriptors passed during socket activation may now be
720 named. A new API sd_listen_fds_with_names() is added to
721 access the names. The default names may be overridden,
722 either in the .socket file using the FileDescriptorName=
723 parameter, or by passing FDNAME= when storing the file
724 descriptors using sd_notify().
726 * systemd-networkd gained support for:
728 - Setting the IPv6 Router Advertisment settings via
729 IPv6AcceptRouterAdvertisements= in .network files.
731 - Configuring the HelloTimeSec=, MaxAgeSec= and
732 ForwardDelaySec= bridge parameters in .netdev files.
734 - Configuring PreferredSource= for static routes in
737 * The "ask-password" framework used to query for LUKS harddisk
738 passwords or SSL passwords during boot gained support for
739 caching passwords in the kernel keyring, if it is
740 available. This makes sure that the user only has to type in
741 a passphrase once if there are multiple objects to unlock
742 with the same one. Previously, such password caching was
743 available only when Plymouth was used; this moves the
744 caching logic into the systemd codebase itself. The
745 "systemd-ask-password" utility gained a new --keyname=
746 switch to control which kernel keyring key to use for
747 caching a password in. This functionality is also useful for
748 enabling display managers such as gdm to automatically
749 unlock the user's GNOME keyring if its passphrase, the
750 user's password and the harddisk password are the same, if
751 gdm-autologin is used.
753 * When downloading tar or raw images using "machinectl
754 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
755 file is now also downloaded, if it is available and stored
756 next to the image file.
758 * Units of type ".socket" gained a new boolean setting
759 Writable= which is only useful in conjunction with
760 ListenSpecial=. If true, enables opening the specified
761 special file in O_RDWR mode rather than O_RDONLY mode.
763 * systemd-rfkill has been reworked to become a singleton
764 service that is activated through /dev/rfkill on each rfkill
765 state change and saves the settings to disk. This way,
766 systemd-rfkill is now compatible with devices that exist
767 only intermittendly, and even restores state if the previous
768 system shutdown was abrupt rather than clean.
770 * The journal daemon gained support for vacuuming old journal
771 files controlled by the number of files that shall remain,
772 in addition to the already existing control by size and by
773 date. This is useful as journal interleaving performance
774 degrades with too many seperate journal files, and allows
775 putting an effective limit on them. The new setting defaults
776 to 100, but this may be changed by setting SystemMaxFiles=
777 and RuntimeMaxFiles= in journald.conf. Also, the
778 "journalctl" tool gained the new --vacuum-files= switch to
779 manually vacuum journal files to leave only the specified
780 number of files in place.
782 * udev will now create /dev/disk/by-path links for ATA devices
783 on kernels where that is supported.
785 * Galician, Serbian, Turkish and Korean translations were added.
787 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
788 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
789 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
790 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
791 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
792 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
793 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
794 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
795 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
796 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
797 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
798 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
799 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
800 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
801 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
802 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
803 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
804 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
810 * The DHCP implementation of systemd-networkd gained a set of
813 - The DHCP server now supports emitting DNS and NTP
814 information. It may be enabled and configured via
815 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
816 and NTP information is enabled, but no servers are
817 configured, the corresponding uplink information (if there
818 is any) is propagated.
820 - Server and client now support transmission and reception
821 of timezone information. It can be configured via the
822 newly introduced network options UseTimezone=,
823 EmitTimezone=, and Timezone=. Transmission of timezone
824 information is enabled between host and containers by
825 default now: the container will change its local timezone
826 to what the host has set.
828 - Lease timeouts can now be configured via
829 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
831 - The DHCP server improved on the stability of
832 leases. Clients are more likely to get the same lease
833 information back, even if the server loses state.
835 - The DHCP server supports two new configuration options to
836 control the lease address pool metrics, PoolOffset= and
839 * The encapsulation limit of tunnels in systemd-networkd may
840 now be configured via 'EncapsulationLimit='. It allows
841 modifying the maximum additional levels of encapsulation
842 that are permitted to be prepended to a packet.
844 * systemd now supports the concept of user buses replacing
845 session buses, if used with dbus-1.10 (and enabled via dbus
846 --enable-user-session). It previously only supported this on
847 kdbus-enabled systems, and this release expands this to
848 'dbus-daemon' systems.
850 * systemd-networkd now supports predictable interface names
853 * systemd now optionally supports the new Linux kernel
854 "unified" control group hierarchy. If enabled via the kernel
855 command-line option 'systemd.unified_cgroup_hierarchy=1',
856 systemd will try to mount the unified cgroup hierarchy
857 directly on /sys/fs/cgroup. If not enabled, or not
858 available, systemd will fall back to the legacy cgroup
859 hierarchy setup, as before. Host system and containers can
860 mix and match legacy and unified hierarchies as they
861 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
862 environment variable to individually select the hierarchy to
863 use for executed containers. By default, nspawn will use the
864 unified hierarchy for the containers if the host uses the
865 unified hierarchy, and the legacy hierarchy otherwise.
866 Please note that at this point the unified hierarchy is an
867 experimental kernel feature and is likely to change in one
868 of the next kernel releases. Therefore, it should not be
869 enabled by default in downstream distributions yet. The
870 minimum required kernel version for the unified hierarchy to
871 work is 4.2. Note that when the unified hierarchy is used
872 for the first time delegated access to controllers is
873 safe. Because of this systemd-nspawn containers will get
874 access to controllers now, as will systemd user
875 sessions. This means containers and user sessions may now
876 manage their own resources, partitioning up what the system
879 * A new special scope unit "init.scope" has been introduced
880 that encapsulates PID 1 of the system. It may be used to
881 determine resource usage and enforce resource limits on PID
882 1 itself. PID 1 hence moved out of the root of the control
885 * The cgtop tool gained support for filtering out kernel
886 threads when counting tasks in a control group. Also, the
887 count of processes is now recursively summed up by
888 default. Two options -k and --recursive= have been added to
889 revert to old behaviour. The tool has also been updated to
890 work correctly in containers now.
892 * systemd-nspawn's --bind= and --bind-ro= options have been
893 extended to allow creation of non-recursive bind mounts.
895 * libsystemd gained two new calls sd_pid_get_cgroup() and
896 sd_peer_get_cgroup() which return the control group path of
897 a process or peer of a connected AF_UNIX socket. This
898 function call is particularly useful when implementing
899 delegated subtrees support in the control group hierarchy.
901 * The "sd-event" event loop API of libsystemd now supports
902 correct dequeuing of real-time signals, without losing
905 * When systemd requests a PolicyKit decision when managing
906 units it will now add additional fields to the request,
907 including unit name and desired operation. This enables more
908 powerful PolicyKit policies, that make decisions depending
911 * nspawn learnt support for .nspawn settings files, that may
912 accompany the image files or directories of containers, and
913 may contain additional settings for the container. This is
914 an alternative to configuring container parameters via the
917 Contributions from: Cristian Rodríguez, Daniel Mack, David
918 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
919 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
920 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
921 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
922 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
923 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
924 Andersen, Tom Gundersen, Torstein Husebø
930 * machinectl gained a new verb 'shell' which opens a fresh
931 shell on the target container or the host. It is similar to
932 the existing 'login' command of machinectl, but spawns the
933 shell directly without prompting for username or
934 password. The pseudo machine '.host' now refers to the local
935 host and is used by default. Hence, 'machinectl shell' can
936 be used as replacement for 'su -' which spawns a session as
937 a fresh systemd unit in a way that is fully isolated from
938 the originating session.
940 * systemd-networkd learned to cope with private-zone DHCP
941 options and allows other programs to query the values.
943 * SELinux access control when enabling/disabling units is no
944 longer enforced with this release. The previous
945 implementation was incorrect, and a new corrected
946 implementation is not yet available. As unit file operations
947 are still protected via PolicyKit and D-Bus policy this is
948 not a security problem. Yet, distributions which care about
949 optimal SELinux support should probably not stabilize on
952 * sd-bus gained support for matches of type "arg0has=", that
953 test for membership of strings in string arrays sent in bus
956 * systemd-resolved now dumps the contents of its DNS and LLMNR
957 caches to the logs on reception of the SIGUSR1 signal. This
958 is useful to debug DNS behaviour.
960 * The coredumpctl tool gained a new --directory= option to
961 operate on journal files in a specific directory.
963 * "systemctl reboot" and related commands gained a new
964 "--message=" option which may be used to set a free-text
965 wall message when shutting down or rebooting the
966 system. This message is also logged, which is useful for
967 figuring out the reason for a reboot or shutdown a
970 * The "systemd-resolve-host" tool's -i switch now takes
971 network interface numbers as alternative to interface names.
973 * A new unit file setting for services has been introduced:
974 UtmpMode= allows configuration of how precisely systemd
975 handles utmp and wtmp entries for the service if this is
976 enabled. This allows writing services that appear similar to
977 user sessions in the output of the "w", "who", "last" and
980 * systemd-resolved will now locally synthesize DNS resource
981 records for the "localhost" and "gateway" domains as well as
982 the local hostname. This should ensure that clients querying
983 RRs via resolved will get similar results as those going via
984 NSS, if nss-myhostname is enabled.
986 Contributions from: Alastair Hughes, Alex Crawford, Daniel
987 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
988 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
989 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
990 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
991 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
992 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
993 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
994 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
995 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
996 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
997 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1003 * The systemd-efi-boot-generator functionality was merged into
1004 systemd-gpt-auto-generator.
1006 * systemd-networkd now supports Group Policy for vxlan
1007 devices. It can be enabled via the new boolean configuration
1008 option called 'GroupPolicyExtension='.
1010 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1011 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1012 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1014 — Berlin, 2015-07-31
1018 * The python-systemd code has been removed from the systemd repository.
1019 A new repository has been created which accommodates the code from
1020 now on, and we kindly ask distributions to create a separate package
1021 for this: https://github.com/systemd/python-systemd
1023 * The systemd daemon will now reload its main configuration
1024 (/etc/systemd/system.conf) on daemon-reload.
1026 * sd-dhcp now exposes vendor specific extensions via
1027 sd_dhcp_lease_get_vendor_specific().
1029 * systemd-networkd gained a number of new configuration options.
1031 - A new boolean configuration option for TAP devices called
1032 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1033 device, thus allowing to send and receive GSO packets.
1035 - A new tunnel configuration option called 'CopyDSCP='.
1036 If enabled, the DSCP field of ip6 tunnels is copied into the
1037 decapsulated packet.
1039 - A set of boolean bridge configuration options were added.
1040 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1041 and 'UnicastFlood=' are now parsed by networkd and applied to the
1042 respective bridge link device via the respective IFLA_BRPORT_*
1045 - A new string configuration option to override the hostname sent
1046 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1047 is true, networkd will use the configured hostname instead of the
1048 system hostname when sending DHCP requests.
1050 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1051 networkd will configure the IPv6 flow-label of the tunnel device
1052 according to RFC2460.
1054 - The 'macvtap' virtual network devices are now supported, similar to
1055 the already supported 'macvlan' devices.
1057 * systemd-resolved now implements RFC5452 to improve resilience against
1058 cache poisoning. Additionally, source port randomization is enabled
1059 by default to further protect against DNS spoofing attacks.
1061 * nss-mymachines now supports translating UIDs and GIDs of running
1062 containers with user-namespaces enabled. If a container 'foo'
1063 translates a host uid 'UID' to the container uid 'TUID', then
1064 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1065 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1066 mapped as 'vg-foo-TGID'.
1068 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1069 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1070 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1071 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1072 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1073 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1074 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1075 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1076 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1077 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1079 — Berlin, 2015-07-29
1083 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1084 There are no known issues with current sysfs, and udev does not need
1085 or should be used to work around such bugs.
1087 * udev does no longer enable USB HID power management. Several reports
1088 indicate, that some devices cannot handle that setting.
1090 * The udev accelerometer helper was removed. The functionality
1091 is now fully included in iio-sensor-proxy. But this means,
1092 older iio-sensor-proxy versions will no longer provide
1093 accelerometer/orientation data with this systemd version.
1094 Please upgrade iio-sensor-proxy to version 1.0.
1096 * networkd gained a new configuration option IPv6PrivacyExtensions=
1097 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1098 for Stateless Address") on selected networks.
1100 * For the sake of fewer build-time dependencies and less code in the
1101 main repository, the python bindings are about to be removed in the
1102 next release. A new repository has been created which accommodates
1103 the code from now on, and we kindly ask distributions to create a
1104 separate package for this. The removal will take place in v223.
1106 https://github.com/systemd/python-systemd
1108 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1109 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1110 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1111 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1112 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1113 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1114 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1115 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1116 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1117 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1119 — Berlin, 2015-07-07
1123 * The sd-bus.h and sd-event.h APIs have now been declared
1124 stable and have been added to the official interface of
1125 libsystemd.so. sd-bus implements an alternative D-Bus client
1126 library, that is relatively easy to use, very efficient and
1127 supports both classic D-Bus as well as kdbus as transport
1128 backend. sd-event is a generic event loop abstraction that
1129 is built around Linux epoll, but adds features such as event
1130 prioritization or efficient timer handling. Both APIs are good
1131 choices for C programs looking for a bus and/or event loop
1132 implementation that is minimal and does not have to be
1133 portable to other kernels.
1135 * kdbus support is no longer compile-time optional. It is now
1136 always built-in. However, it can still be disabled at
1137 runtime using the kdbus=0 kernel command line setting, and
1138 that setting may be changed to default to off, by specifying
1139 --disable-kdbus at build-time. Note though that the kernel
1140 command line setting has no effect if the kdbus.ko kernel
1141 module is not installed, in which case kdbus is (obviously)
1142 also disabled. We encourage all downstream distributions to
1143 begin testing kdbus by adding it to the kernel images in the
1144 development distributions, and leaving kdbus support in
1147 * The minimal required util-linux version has been bumped to
1150 * Support for chkconfig (--enable-chkconfig) was removed in
1151 favor of calling an abstraction tool
1152 /lib/systemd/systemd-sysv-install. This needs to be
1153 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1154 in README for details.
1156 * If there's a systemd unit and a SysV init script for the
1157 same service name, and the user executes "systemctl enable"
1158 for it (or a related call), then this will now enable both
1159 (or execute the related operation on both), not just the
1162 * The libudev API documentation has been converted from gtkdoc
1165 * gudev has been removed from the systemd tree, it is now an
1168 * The systemd-cgtop tool learnt a new --raw switch to generate
1169 "raw" (machine parsable) output.
1171 * networkd's IPForwarding= .network file setting learnt the
1172 new setting "kernel", which ensures that networkd does not
1173 change the IP forwarding sysctl from the default kernel
1176 * The systemd-logind bus API now exposes a new boolean
1177 property "Docked" that reports whether logind considers the
1178 system "docked", i.e. connected to a docking station or not.
1180 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1181 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1182 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1183 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1184 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1185 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1186 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1187 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1188 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1189 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1190 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1191 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1192 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1193 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1194 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1195 Fink, Zbigniew Jędrzejewski-Szmek
1197 — Berlin, 2015-06-19
1201 * The gudev library has been extracted into a separate repository
1202 available at: https://git.gnome.org/browse/libgudev/
1203 It is now managed as part of the Gnome project. Distributions
1204 are recommended to pass --disable-gudev to systemd and use
1205 gudev from the Gnome project instead. gudev is still included
1206 in systemd, for now. It will be removed soon, though. Please
1207 also see the announcement-thread on systemd-devel:
1208 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1210 * systemd now exposes a CPUUsageNSec= property for each
1211 service unit on the bus, that contains the overall consumed
1212 CPU time of a service (the sum of what each process of the
1213 service consumed). This value is only available if
1214 CPUAccounting= is turned on for a service, and is then shown
1215 in the "systemctl status" output.
1217 * Support for configuring alternative mappings of the old SysV
1218 runlevels to systemd targets has been removed. They are now
1219 hardcoded in a way that runlevels 2, 3, 4 all map to
1220 multi-user.target and 5 to graphical.target (which
1221 previously was already the default behaviour).
1223 * The auto-mounter logic gained support for mount point
1224 expiry, using a new TimeoutIdleSec= setting in .automount
1225 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1227 * The EFI System Partition (ESP) as mounted to /boot by
1228 systemd-efi-boot-generator will now be unmounted
1229 automatically after 2 minutes of not being used. This should
1230 minimize the risk of ESP corruptions.
1232 * New /etc/fstab options x-systemd.requires= and
1233 x-systemd.requires-mounts-for= are now supported to express
1234 additional dependencies for mounts. This is useful for
1235 journalling file systems that support external journal
1236 devices or overlay file systems that require underlying file
1237 systems to be mounted.
1239 * systemd does not support direct live-upgrades (via systemctl
1240 daemon-reexec) from versions older than v44 anymore. As no
1241 distribution we are aware of shipped such old versions in a
1242 stable release this should not be problematic.
1244 * When systemd forks off a new per-connection service instance
1245 it will now set the $REMOTE_ADDR environment variable to the
1246 remote IP address, and $REMOTE_PORT environment variable to
1247 the remote IP port. This behaviour is similar to the
1248 corresponding environment variables defined by CGI.
1250 * systemd-networkd gained support for uplink failure
1251 detection. The BindCarrier= option allows binding interface
1252 configuration dynamically to the link sense of other
1253 interfaces. This is useful to achieve behaviour like in
1256 * systemd-networkd gained support for configuring the DHCP
1257 client identifier to use when requesting leases.
1259 * systemd-networkd now has a per-network UseNTP= option to
1260 configure whether NTP server information acquired via DHCP
1261 is passed on to services like systemd-timesyncd.
1263 * systemd-networkd gained support for vti6 tunnels.
1265 * Note that systemd-networkd manages the sysctl variable
1266 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1267 it is configured for since v219. The variable controls IP
1268 forwarding, and is a per-interface alternative to the global
1269 /proc/sys/net/ipv[46]/ip_forward. This setting is
1270 configurable in the IPForward= option, which defaults to
1271 "no". This means if networkd is used for an interface it is
1272 no longer sufficient to set the global sysctl option to turn
1273 on IP forwarding! Instead, the .network file option
1274 IPForward= needs to be turned on! Note that the
1275 implementation of this behaviour was broken in v219 and has
1278 * Many bonding and vxlan options are now configurable in
1281 * systemd-nspawn gained a new --property= setting to set unit
1282 properties for the container scope. This is useful for
1283 setting resource parameters (e.g "CPUShares=500") on
1284 containers started from the command line.
1286 * systemd-nspawn gained a new --private-users= switch to make
1287 use of user namespacing available on recent Linux kernels.
1289 * systemd-nspawn may now be called as part of a shell pipeline
1290 in which case the pipes used for stdin and stdout are passed
1291 directly to the process invoked in the container, without
1292 indirection via a pseudo tty.
1294 * systemd-nspawn gained a new switch to control the UNIX
1295 signal to use when killing the init process of the container
1298 * systemd-nspawn gained a new --overlay= switch for mounting
1299 overlay file systems into the container using the new kernel
1302 * When a container image is imported via systemd-importd and
1303 the host file system is not btrfs, a loopback block device
1304 file is created in /var/lib/machines.raw with a btrfs file
1305 system inside. It is then mounted to /var/lib/machines to
1306 enable btrfs features for container management. The loopback
1307 file and btrfs file system is grown as needed when container
1308 images are imported via systemd-importd.
1310 * systemd-machined/systemd-importd gained support for btrfs
1311 quota, to enforce container disk space limits on disk. This
1312 is exposed in "machinectl set-limit".
1314 * systemd-importd now can import containers from local .tar,
1315 .raw and .qcow2 images, and export them to .tar and .raw. It
1316 can also import dkr v2 images now from the network (on top
1319 * systemd-importd gained support for verifying downloaded
1320 images with gpg2 (previously only gpg1 was supported).
1322 * systemd-machined, systemd-logind, systemd: most bus calls
1323 are now accessible to unprivileged processes via
1324 PolicyKit. Also, systemd-logind will now allow users to kill
1325 their own sessions without further privileges or
1328 * systemd-shutdownd has been removed. This service was
1329 previously responsible for implementing scheduled shutdowns
1330 as exposed in /usr/bin/shutdown's time parameter. This
1331 functionality has now been moved into systemd-logind and is
1332 accessible via a bus interface.
1334 * "systemctl reboot" gained a new switch --firmware-setup that
1335 can be used to reboot into the EFI firmware setup, if that
1336 is available. systemd-logind now exposes an API on the bus
1337 to trigger such reboots, in case graphical desktop UIs want
1338 to cover this functionality.
1340 * "systemctl enable", "systemctl disable" and "systemctl mask"
1341 now support a new "--now" switch. If specified the units
1342 that are enabled will also be started, and the ones
1343 disabled/masked also stopped.
1345 * The Gummiboot EFI boot loader tool has been merged into
1346 systemd, and renamed to "systemd-boot". The bootctl tool has been
1347 updated to support systemd-boot.
1349 * An EFI kernel stub has been added that may be used to create
1350 kernel EFI binaries that contain not only the actual kernel,
1351 but also an initrd, boot splash, command line and OS release
1352 information. This combined binary can then be signed as a
1353 single image, so that the firmware can verify it all in one
1354 step. systemd-boot has special support for EFI binaries created
1355 like this and can extract OS release information from them
1356 and show them in the boot menu. This functionality is useful
1357 to implement cryptographically verified boot schemes.
1359 * Optional support has been added to systemd-fsck to pass
1360 fsck's progress report to an AF_UNIX socket in the file
1363 * udev will no longer create device symlinks for all block
1364 devices by default. A blacklist for excluding special block
1365 devices from this logic has been turned into a whitelist
1366 that requires picking block devices explicitly that require
1369 * A new (currently still internal) API sd-device.h has been
1370 added to libsystemd. This modernized API is supposed to
1371 replace libudev eventually. In fact, already much of libudev
1372 is now just a wrapper around sd-device.h.
1374 * A new hwdb database for storing metadata about pointing
1375 stick devices has been added.
1377 * systemd-tmpfiles gained support for setting file attributes
1378 similar to the "chattr" tool with new 'h' and 'H' lines.
1380 * systemd-journald will no longer unconditionally set the
1381 btrfs NOCOW flag on new journal files. This is instead done
1382 with tmpfiles snippet using the new 'h' line type. This
1383 allows easy disabling of this logic, by masking the
1384 journal-nocow.conf tmpfiles file.
1386 * systemd-journald will now translate audit message types to
1387 human readable identifiers when writing them to the
1388 journal. This should improve readability of audit messages.
1390 * The LUKS logic gained support for the offset= and skip=
1391 options in /etc/crypttab, as previously implemented by
1394 * /usr/lib/os-release gained a new optional field VARIANT= for
1395 distributions that support multiple variants (such as a
1396 desktop edition, a server edition, ...)
1398 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1399 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1400 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1401 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1402 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1403 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1404 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1405 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1406 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1407 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1408 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1409 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1410 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1411 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1412 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1413 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1414 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1415 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1416 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1417 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1418 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1419 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1420 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1421 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1422 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1423 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1424 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1426 — Berlin, 2015-05-22
1430 * Introduce a new API "sd-hwdb.h" for querying the hardware
1431 metadata database. With this minimal interface one can query
1432 and enumerate the udev hwdb, decoupled from the old libudev
1433 library. libudev's interface for this is now only a wrapper
1434 around sd-hwdb. A new tool systemd-hwdb has been added to
1435 interface with and update the database.
1437 * When any of systemd's tools copies files (for example due to
1438 tmpfiles' C lines) a btrfs reflink will attempted first,
1439 before bytewise copying is done.
1441 * systemd-nspawn gained a new --ephemeral switch. When
1442 specified a btrfs snapshot is taken of the container's root
1443 directory, and immediately removed when the container
1444 terminates again. Thus, a container can be started whose
1445 changes never alter the container's root directory, and are
1446 lost on container termination. This switch can also be used
1447 for starting a container off the root file system of the
1448 host without affecting the host OS. This switch is only
1449 available on btrfs file systems.
1451 * systemd-nspawn gained a new --template= switch. It takes the
1452 path to a container tree to use as template for the tree
1453 specified via --directory=, should that directory be
1454 missing. This allows instantiating containers dynamically,
1455 on first run. This switch is only available on btrfs file
1458 * When a .mount unit refers to a mount point on which multiple
1459 mounts are stacked, and the .mount unit is stopped all of
1460 the stacked mount points will now be unmounted until no
1461 mount point remains.
1463 * systemd now has an explicit notion of supported and
1464 unsupported unit types. Jobs enqueued for unsupported unit
1465 types will now fail with an "unsupported" error code. More
1466 specifically .swap, .automount and .device units are not
1467 supported in containers, .busname units are not supported on
1468 non-kdbus systems. .swap and .automount are also not
1469 supported if their respective kernel compile time options
1472 * machinectl gained support for two new "copy-from" and
1473 "copy-to" commands for copying files from a running
1474 container to the host or vice versa.
1476 * machinectl gained support for a new "bind" command to bind
1477 mount host directories into local containers. This is
1478 currently only supported for nspawn containers.
1480 * networkd gained support for configuring bridge forwarding
1481 database entries (fdb) from .network files.
1483 * A new tiny daemon "systemd-importd" has been added that can
1484 download container images in tar, raw, qcow2 or dkr formats,
1485 and make them available locally in /var/lib/machines, so
1486 that they can run as nspawn containers. The daemon can GPG
1487 verify the downloads (not supported for dkr, since it has no
1488 provisions for verifying downloads). It will transparently
1489 decompress bz2, xz, gzip compressed downloads if necessary,
1490 and restore sparse files on disk. The daemon uses privilege
1491 separation to ensure the actual download logic runs with
1492 fewer privileges than the daemon itself. machinectl has
1493 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1494 make the functionality of importd available to the
1495 user. With this in place the Fedora and Ubuntu "Cloud"
1496 images can be downloaded and booted as containers unmodified
1497 (the Fedora images lack the appropriate GPG signature files
1498 currently, so they cannot be verified, but this will change
1499 soon, hopefully). Note that downloading images is currently
1500 only fully supported on btrfs.
1502 * machinectl is now able to list container images found in
1503 /var/lib/machines, along with some metadata about sizes of
1504 disk and similar. If the directory is located on btrfs and
1505 quota is enabled, this includes quota display. A new command
1506 "image-status" has been added that shows additional
1507 information about images.
1509 * machinectl is now able to clone container images
1510 efficiently, if the underlying file system (btrfs) supports
1511 it, with the new "machinectl clone" command. It also
1512 gained commands for renaming and removing images, as well as
1513 marking them read-only or read-write (supported also on
1514 legacy file systems).
1516 * networkd gained support for collecting LLDP network
1517 announcements, from hardware that supports this. This is
1518 shown in networkctl output.
1520 * systemd-run gained support for a new -t (--pty) switch for
1521 invoking a binary on a pty whose input and output is
1522 connected to the invoking terminal. This allows executing
1523 processes as system services while interactively
1524 communicating with them via the terminal. Most interestingly
1525 this is supported across container boundaries. Invoking
1526 "systemd-run -t /bin/bash" is an alternative to running a
1527 full login session, the difference being that the former
1528 will not register a session, nor go through the PAM session
1531 * tmpfiles gained support for a new "v" line type for creating
1532 btrfs subvolumes. If the underlying file system is a legacy
1533 file system, this automatically degrades to creating a
1534 normal directory. Among others /var/lib/machines is now
1535 created like this at boot, should it be missing.
1537 * The directory /var/lib/containers/ has been deprecated and
1538 been replaced by /var/lib/machines. The term "machines" has
1539 been used in the systemd context as generic term for both
1540 VMs and containers, and hence appears more appropriate for
1541 this, as the directory can also contain raw images bootable
1544 * systemd-nspawn when invoked with -M but without --directory=
1545 or --image= is now capable of searching for the container
1546 root directory, subvolume or disk image automatically, in
1547 /var/lib/machines. systemd-nspawn@.service has been updated
1548 to make use of this, thus allowing it to be used for raw
1551 * A new machines.target unit has been introduced that is
1552 supposed to group all containers/VMs invoked as services on
1553 the system. systemd-nspawn@.service has been updated to
1554 integrate with that.
1556 * machinectl gained a new "start" command, for invoking a
1557 container as a service. "machinectl start foo" is mostly
1558 equivalent to "systemctl start systemd-nspawn@foo.service",
1559 but handles escaping in a nicer way.
1561 * systemd-nspawn will now mount most of the cgroupfs tree
1562 read-only into each container, with the exception of the
1563 container's own subtree in the name=systemd hierarchy.
1565 * journald now sets the special FS_NOCOW file flag for its
1566 journal files. This should improve performance on btrfs, by
1567 avoiding heavy fragmentation when journald's write-pattern
1568 is used on COW file systems. It degrades btrfs' data
1569 integrity guarantees for the files to the same levels as for
1570 ext3/ext4 however. This should be OK though as journald does
1571 its own data integrity checks and all its objects are
1572 checksummed on disk. Also, journald should handle btrfs disk
1573 full events a lot more gracefully now, by processing SIGBUS
1574 errors, and not relying on fallocate() anymore.
1576 * When journald detects that journal files it is writing to
1577 have been deleted it will immediately start new journal
1580 * systemd now provides a way to store file descriptors
1581 per-service in PID 1.This is useful for daemons to ensure
1582 that fds they require are not lost during a daemon
1583 restart. The fds are passed to the daemon on the next
1584 invocation in the same way socket activation fds are
1585 passed. This is now used by journald to ensure that the
1586 various sockets connected to all the system's stdout/stderr
1587 are not lost when journald is restarted. File descriptors
1588 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1589 an extension to sd_notify(). Note that a limit is enforced
1590 on the number of fds a service can store in PID 1, and it
1591 defaults to 0, so that no fds may be stored, unless this is
1592 explicitly turned on.
1594 * The default TERM variable to use for units connected to a
1595 terminal, when no other value is explicitly is set is now
1596 vt220 rather than vt102. This should be fairly safe still,
1597 but allows PgUp/PgDn work.
1599 * The /etc/crypttab option header= as known from Debian is now
1602 * "loginctl user-status" and "loginctl session-status" will
1603 now show the last 10 lines of log messages of the
1604 user/session following the status output. Similar,
1605 "machinectl status" will show the last 10 log lines
1606 associated with a virtual machine or container
1607 service. (Note that this is usually not the log messages
1608 done in the VM/container itself, but simply what the
1609 container manager logs. For nspawn this includes all console
1612 * "loginctl session-status" without further argument will now
1613 show the status of the session of the caller. Similar,
1614 "lock-session", "unlock-session", "activate",
1615 "enable-linger", "disable-linger" may now be called without
1616 session/user parameter in which case they apply to the
1617 caller's session/user.
1619 * An X11 session scriptlet is now shipped that uploads
1620 $DISPLAY and $XAUTHORITY into the environment of the systemd
1621 --user daemon if a session begins. This should improve
1622 compatibility with X11 enabled applications run as systemd
1625 * Generators are now subject to masking via /etc and /run, the
1626 same way as unit files.
1628 * networkd .network files gained support for configuring
1629 per-link IPv4/IPv6 packet forwarding as well as IPv4
1630 masquerading. This is by default turned on for veth links to
1631 containers, as registered by systemd-nspawn. This means that
1632 nspawn containers run with --network-veth will now get
1633 automatic routed access to the host's networks without any
1634 further configuration or setup, as long as networkd runs on
1637 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1638 or UDP posts of a container on the host. With this in place
1639 it is possible to run containers with private veth links
1640 (--network-veth), and have their functionality exposed on
1641 the host as if their services were running directly on the
1644 * systemd-nspawn's --network-veth switch now gained a short
1645 version "-n", since with the changes above it is now truly
1646 useful out-of-the-box. The systemd-nspawn@.service has been
1647 updated to make use of it too by default.
1649 * systemd-nspawn will now maintain a per-image R/W lock, to
1650 ensure that the same image is not started more than once
1651 writable. (It's OK to run an image multiple times
1652 simultaneously in read-only mode.)
1654 * systemd-nspawn's --image= option is now capable of
1655 dissecting and booting MBR and GPT disk images that contain
1656 only a single active Linux partition. Previously it
1657 supported only GPT disk images with proper GPT type
1658 IDs. This allows running cloud images from major
1659 distributions directly with systemd-nspawn, without
1662 * In addition to collecting mouse dpi data in the udev
1663 hardware database, there's now support for collecting angle
1664 information for mouse scroll wheels. The database is
1665 supposed to guarantee similar scrolling behavior on mice
1666 that it knows about. There's also support for collecting
1667 information about Touchpad types.
1669 * udev's input_id built-in will now also collect touch screen
1670 dimension data and attach it to probed devices.
1672 * /etc/os-release gained support for a Distribution Privacy
1675 * networkd gained support for creating "ipvlan", "gretap",
1676 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1678 * systemd-tmpfiles gained support for "a" lines for setting
1681 * systemd-nspawn will now mount /tmp in the container to
1682 tmpfs, automatically.
1684 * systemd now exposes the memory.usage_in_bytes cgroup
1685 attribute and shows it for each service in the "systemctl
1686 status" output, if available.
1688 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1689 immediate reboot is triggered. This useful if shutdown is
1690 hung and is unable to complete, to expedite the
1691 operation. Note that this kind of reboot will still unmount
1692 all file systems, and hence should not result in fsck being
1695 * A .device unit for an optical block device will now be
1696 considered active only when a medium is in the drive. Also,
1697 mount units are now bound to their backing devices thus
1698 triggering automatic unmounting when devices become
1699 unavailable. With this in place systemd will now
1700 automatically unmount left-over mounts when a CD-ROM is
1701 ejected or an USB stick is yanked from the system.
1703 * networkd-wait-online now has support for waiting for
1704 specific interfaces only (with globbing), and for giving up
1705 after a configurable timeout.
1707 * networkd now exits when idle. It will be automatically
1708 restarted as soon as interfaces show up, are removed or
1709 change state. networkd will stay around as long as there is
1710 at least one DHCP state machine or similar around, that keep
1713 * networkd may now configure IPv6 link-local addressing in
1714 addition to IPv4 link-local addressing.
1716 * The IPv6 "token" for use in SLAAC may now be configured for
1717 each .network interface in networkd.
1719 * Routes configured with networkd may now be assigned a scope
1722 * networkd's [Match] sections now support globbing and lists
1723 of multiple space-separated matches per item.
1725 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1726 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1727 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1728 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1729 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1730 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1731 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1732 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1733 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1734 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1735 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1736 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1737 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1738 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1739 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1740 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1741 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1742 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1743 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1744 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1745 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1746 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1747 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1748 Hoffmann, Zbigniew Jędrzejewski-Szmek
1750 — Berlin, 2015-02-16
1754 * When querying unit file enablement status (for example via
1755 "systemctl is-enabled"), a new state "indirect" is now known
1756 which indicates that a unit might not be enabled itself, but
1757 another unit listed in its Also= setting might be.
1759 * Similar to the various existing ConditionXYZ= settings for
1760 units, there are now matching AssertXYZ= settings. While
1761 failing conditions cause a unit to be skipped, but its job
1762 to succeed, failing assertions declared like this will cause
1763 a unit start operation and its job to fail.
1765 * hostnamed now knows a new chassis type "embedded".
1767 * systemctl gained a new "edit" command. When used on a unit
1768 file, this allows extending unit files with .d/ drop-in
1769 configuration snippets or editing the full file (after
1770 copying it from /usr/lib to /etc). This will invoke the
1771 user's editor (as configured with $EDITOR), and reload the
1772 modified configuration after editing.
1774 * "systemctl status" now shows the suggested enablement state
1775 for a unit, as declared in the (usually vendor-supplied)
1776 system preset files.
1778 * nss-myhostname will now resolve the single-label host name
1779 "gateway" to the locally configured default IP routing
1780 gateways, ordered by their metrics. This assigns a stable
1781 name to the used gateways, regardless which ones are
1782 currently configured. Note that the name will only be
1783 resolved after all other name sources (if nss-myhostname is
1784 configured properly) and should hence not negatively impact
1785 systems that use the single-label host name "gateway" in
1788 * systemd-inhibit now allows filtering by mode when listing
1791 * Scope and service units gained a new "Delegate" boolean
1792 property, which, when set, allows processes running inside the
1793 unit to further partition resources. This is primarily
1794 useful for systemd user instances as well as container
1797 * journald will now pick up audit messages directly from
1798 the kernel, and log them like any other log message. The
1799 audit fields are split up and fully indexed. This means that
1800 journalctl in many ways is now a (nicer!) alternative to
1801 ausearch, the traditional audit client. Note that this
1802 implements only a minimal audit client. If you want the
1803 special audit modes like reboot-on-log-overflow, please use
1804 the traditional auditd instead, which can be used in
1805 parallel to journald.
1807 * The ConditionSecurity= unit file option now understands the
1808 special string "audit" to check whether auditing is
1811 * journalctl gained two new commands --vacuum-size= and
1812 --vacuum-time= to delete old journal files until the
1813 remaining ones take up no more than the specified size on disk,
1814 or are not older than the specified time.
1816 * A new, native PPPoE library has been added to sd-network,
1817 systemd's library of light-weight networking protocols. This
1818 library will be used in a future version of networkd to
1819 enable PPPoE communication without an external pppd daemon.
1821 * The busctl tool now understands a new "capture" verb that
1822 works similar to "monitor", but writes a packet capture
1823 trace to STDOUT that can be redirected to a file which is
1824 compatible with libcap's capture file format. This can then
1825 be loaded in Wireshark and similar tools to inspect bus
1828 * The busctl tool now understands a new "tree" verb that shows
1829 the object trees of a specific service on the bus, or of all
1832 * The busctl tool now understands a new "introspect" verb that
1833 shows all interfaces and members of objects on the bus,
1834 including their signature and values. This is particularly
1835 useful to get more information about bus objects shown by
1836 the new "busctl tree" command.
1838 * The busctl tool now understands new verbs "call",
1839 "set-property" and "get-property" for invoking bus method
1840 calls, setting and getting bus object properties in a
1843 * busctl gained a new --augment-creds= argument that controls
1844 whether the tool shall augment credential information it
1845 gets from the bus with data from /proc, in a possibly
1848 * nspawn's --link-journal= switch gained two new values
1849 "try-guest" and "try-host" that work like "guest" and
1850 "host", but do not fail if the host has no persistent
1851 journalling enabled. -j is now equivalent to
1852 --link-journal=try-guest.
1854 * macvlan network devices created by nspawn will now have
1855 stable MAC addresses.
1857 * A new SmackProcessLabel= unit setting has been added, which
1858 controls the SMACK security label processes forked off by
1859 the respective unit shall use.
1861 * If compiled with --enable-xkbcommon, systemd-localed will
1862 verify x11 keymap settings by compiling the given keymap. It
1863 will spew out warnings if the compilation fails. This
1864 requires libxkbcommon to be installed.
1866 * When a coredump is collected, a larger number of metadata
1867 fields is now collected and included in the journal records
1868 created for it. More specifically, control group membership,
1869 environment variables, memory maps, working directory,
1870 chroot directory, /proc/$PID/status, and a list of open file
1871 descriptors is now stored in the log entry.
1873 * The udev hwdb now contains DPI information for mice. For
1876 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1878 * All systemd programs that read standalone configuration
1879 files in /etc now also support a corresponding series of
1880 .conf.d configuration directories in /etc/, /run/,
1881 /usr/local/lib/, /usr/lib/, and (if configured with
1882 --enable-split-usr) /lib/. In particular, the following
1883 configuration files now have corresponding configuration
1884 directories: system.conf user.conf, logind.conf,
1885 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1886 resolved.conf, timesyncd.conf, journal-remote.conf, and
1887 journal-upload.conf. Note that distributions should use the
1888 configuration directories in /usr/lib/; the directories in
1889 /etc/ are reserved for the system administrator.
1891 * systemd-rfkill will no longer take the rfkill device name
1892 into account when storing rfkill state on disk, as the name
1893 might be dynamically assigned and not stable. Instead, the
1894 ID_PATH udev variable combined with the rfkill type (wlan,
1895 bluetooth, ...) is used.
1897 * A new service systemd-machine-id-commit.service has been
1898 added. When used on systems where /etc is read-only during
1899 boot, and /etc/machine-id is not initialized (but an empty
1900 file), this service will copy the temporary machine ID
1901 created as replacement into /etc after the system is fully
1902 booted up. This is useful for systems that are freshly
1903 installed with a non-initialized machine ID, but should get
1904 a fixed machine ID for subsequent boots.
1906 * networkd's .netdev files now provide a large set of
1907 configuration parameters for VXLAN devices. Similarly, the
1908 bridge port cost parameter is now configurable in .network
1909 files. There's also new support for configuring IP source
1910 routing. networkd .link files gained support for a new
1911 OriginalName= match that is useful to match against the
1912 original interface name the kernel assigned. .network files
1913 may include MTU= and MACAddress= fields for altering the MTU
1914 and MAC address while being connected to a specific network
1917 * The LUKS logic gained supported for configuring
1918 UUID-specific key files. There's also new support for naming
1919 LUKS device from the kernel command line, using the new
1920 luks.name= argument.
1922 * Timer units may now be transiently created via the bus API
1923 (this was previously already available for scope and service
1924 units). In addition it is now possible to create multiple
1925 transient units at the same time with a single bus call. The
1926 "systemd-run" tool has been updated to make use of this for
1927 running commands on a specified time, in at(1)-style.
1929 * tmpfiles gained support for "t" lines, for assigning
1930 extended attributes to files. Among other uses this may be
1931 used to assign SMACK labels to files.
1933 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1934 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1935 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1936 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1937 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1938 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1939 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1940 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1941 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1942 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1943 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1944 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1945 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1946 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1947 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1948 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1949 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1950 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1952 — Berlin, 2014-12-10
1956 * journalctl gained the new options -t/--identifier= to match
1957 on the syslog identifier (aka "tag"), as well as --utc to
1958 show log timestamps in the UTC timezone. journalctl now also
1959 accepts -n/--lines=all to disable line capping in a pager.
1961 * journalctl gained a new switch, --flush, that synchronously
1962 flushes logs from /run/log/journal to /var/log/journal if
1963 persistent storage is enabled. systemd-journal-flush.service
1964 now waits until the operation is complete.
1966 * Services can notify the manager before they start a reload
1967 (by sending RELOADING=1) or shutdown (by sending
1968 STOPPING=1). This allows the manager to track and show the
1969 internal state of daemons and closes a race condition when
1970 the process is still running but has closed its D-Bus
1973 * Services with Type=oneshot do not have to have any ExecStart
1976 * User units are now loaded also from
1977 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1978 /run/systemd/user directory that was already previously
1979 supported, but is under the control of the user.
1981 * Job timeouts (i.e. time-outs on the time a job that is
1982 queued stays in the run queue) can now optionally result in
1983 immediate reboot or power-off actions (JobTimeoutAction= and
1984 JobTimeoutRebootArgument=). This is useful on ".target"
1985 units, to limit the maximum time a target remains
1986 undispatched in the run queue, and to trigger an emergency
1987 operation in such a case. This is now used by default to
1988 turn off the system if boot-up (as defined by everything in
1989 basic.target) hangs and does not complete for at least
1990 15min. Also, if power-off or reboot hang for at least 30min
1991 an immediate power-off/reboot operation is triggered. This
1992 functionality is particularly useful to increase reliability
1993 on embedded devices, but also on laptops which might
1994 accidentally get powered on when carried in a backpack and
1995 whose boot stays stuck in a hard disk encryption passphrase
1998 * systemd-logind can be configured to also handle lid switch
1999 events even when the machine is docked or multiple displays
2000 are attached (HandleLidSwitchDocked= option).
2002 * A helper binary and a service have been added which can be
2003 used to resume from hibernation in the initramfs. A
2004 generator will parse the resume= option on the kernel
2005 command line to trigger resume.
2007 * A user console daemon systemd-consoled has been
2008 added. Currently, it is a preview, and will so far open a
2009 single terminal on each session of the user marked as
2010 Desktop=systemd-console.
2012 * Route metrics can be specified for DHCP routes added by
2015 * The SELinux context of socket-activated services can be set
2016 from the information provided by the networking stack
2017 (SELinuxContextFromNet= option).
2019 * Userspace firmware loading support has been removed and
2020 the minimum supported kernel version is thus bumped to 3.7.
2022 * Timeout for udev workers has been increased from 1 to 3
2023 minutes, but a warning will be printed after 1 minute to
2024 help diagnose kernel modules that take a long time to load.
2026 * Udev rules can now remove tags on devices with TAG-="foobar".
2028 * systemd's readahead implementation has been removed. In many
2029 circumstances it didn't give expected benefits even for
2030 rotational disk drives and was becoming less relevant in the
2031 age of SSDs. As none of the developers has been using
2032 rotating media anymore, and nobody stepped up to actively
2033 maintain this component of systemd it has now been removed.
2035 * Swap units can use Options= to specify discard options.
2036 Discard options specified for swaps in /etc/fstab are now
2039 * Docker containers are now detected as a separate type of
2042 * The Password Agent protocol gained support for queries where
2043 the user input is shown, useful e.g. for user names.
2044 systemd-ask-password gained a new --echo option to turn that
2047 * The default sysctl.d/ snippets will now set:
2049 net.core.default_qdisc = fq_codel
2051 This selects Fair Queuing Controlled Delay as the default
2052 queuing discipline for network interfaces. fq_codel helps
2053 fight the network bufferbloat problem. It is believed to be
2054 a good default with no tuning required for most workloads.
2055 Downstream distributions may override this choice. On 10Gbit
2056 servers that do not do forwarding, "fq" may perform better.
2057 Systems without a good clocksource should use "pfifo_fast".
2059 * If kdbus is enabled during build a new option BusPolicy= is
2060 available for service units, that allows locking all service
2061 processes into a stricter bus policy, in order to limit
2062 access to various bus services, or even hide most of them
2063 from the service's view entirely.
2065 * networkctl will now show the .network and .link file
2066 networkd has applied to a specific interface.
2068 * sd-login gained a new API call sd_session_get_desktop() to
2069 query which desktop environment has been selected for a
2072 * UNIX utmp support is now compile-time optional to support
2073 legacy-free systems.
2075 * systemctl gained two new commands "add-wants" and
2076 "add-requires" for pulling in units from specific targets
2079 * If the word "rescue" is specified on the kernel command line
2080 the system will now boot into rescue mode (aka
2081 rescue.target), which was previously available only by
2082 specifying "1" or "systemd.unit=rescue.target" on the kernel
2083 command line. This new kernel command line option nicely
2084 mirrors the already existing "emergency" kernel command line
2087 * New kernel command line options mount.usr=, mount.usrflags=,
2088 mount.usrfstype= have been added that match root=, rootflags=,
2089 rootfstype= but allow mounting a specific file system to
2092 * The $NOTIFY_SOCKET is now also passed to control processes of
2093 services, not only the main process.
2095 * This version reenables support for fsck's -l switch. This
2096 means at least version v2.25 of util-linux is required for
2097 operation, otherwise dead-locks on device nodes may
2098 occur. Again: you need to update util-linux to at least
2099 v2.25 when updating systemd to v217.
2101 * The "multi-seat-x" tool has been removed from systemd, as
2102 its functionality has been integrated into X servers 1.16,
2103 and the tool is hence redundant. It is recommended to update
2104 display managers invoking this tool to simply invoke X
2105 directly from now on, again.
2107 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2108 message flag has been added for all of systemd's PolicyKit
2109 authenticated method calls has been added. In particular
2110 this now allows optional interactive authorization via
2111 PolicyKit for many of PID1's privileged operations such as
2112 unit file enabling and disabling.
2114 * "udevadm hwdb --update" learnt a new switch "--usr" for
2115 placing the rebuilt hardware database in /usr instead of
2116 /etc. When used only hardware database entries stored in
2117 /usr will be used, and any user database entries in /etc are
2118 ignored. This functionality is useful for vendors to ship a
2119 pre-built database on systems where local configuration is
2120 unnecessary or unlikely.
2122 * Calendar time specifications in .timer units now also
2123 understand the strings "semi-annually", "quarterly" and
2124 "minutely" as shortcuts (in addition to the preexisting
2125 "anually", "hourly", ...).
2127 * systemd-tmpfiles will now correctly create files in /dev
2128 at boot which are marked for creation only at boot. It is
2129 recommended to always create static device nodes with 'c!'
2130 and 'b!', so that they are created only at boot and not
2131 overwritten at runtime.
2133 * When the watchdog logic is used for a service (WatchdogSec=)
2134 and the watchdog timeout is hit the service will now be
2135 terminated with SIGABRT (instead of just SIGTERM), in order
2136 to make sure a proper coredump and backtrace is
2137 generated. This ensures that hanging services will result in
2138 similar coredump/backtrace behaviour as services that hit a
2141 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2142 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2143 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2144 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2145 Herrmann, David Sommerseth, David Strauss, Emil Renner
2146 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2147 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2148 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2149 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2150 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2151 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2152 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2153 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2154 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2155 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2156 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2157 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2158 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2159 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2160 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2161 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2164 — Berlin, 2014-10-28
2168 * timedated no longer reads NTP implementation unit names from
2169 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2170 implementations should add a
2172 Conflicts=systemd-timesyncd.service
2174 to their unit files to take over and replace systemd's NTP
2175 default functionality.
2177 * systemd-sysusers gained a new line type "r" for configuring
2178 which UID/GID ranges to allocate system users/groups
2179 from. Lines of type "u" may now add an additional column
2180 that specifies the home directory for the system user to be
2181 created. Also, systemd-sysusers may now optionally read user
2182 information from STDIN instead of a file. This is useful for
2183 invoking it from RPM preinst scriptlets that need to create
2184 users before the first RPM file is installed since these
2185 files might need to be owned by them. A new
2186 %sysusers_create_inline RPM macro has been introduced to do
2187 just that. systemd-sysusers now updates the shadow files as
2188 well as the user/group databases, which should enhance
2189 compatibility with certain tools like grpck.
2191 * A number of bus APIs of PID 1 now optionally consult
2192 PolicyKit to permit access for otherwise unprivileged
2193 clients under certain conditions. Note that this currently
2194 doesn't support interactive authentication yet, but this is
2195 expected to be added eventually, too.
2197 * /etc/machine-info now has new fields for configuring the
2198 deployment environment of the machine, as well as the
2199 location of the machine. hostnamectl has been updated with
2200 new command to update these fields.
2202 * systemd-timesyncd has been updated to automatically acquire
2203 NTP server information from systemd-networkd, which might
2204 have been discovered via DHCP.
2206 * systemd-resolved now includes a caching DNS stub resolver
2207 and a complete LLMNR name resolution implementation. A new
2208 NSS module "nss-resolve" has been added which can be used
2209 instead of glibc's own "nss-dns" to resolve hostnames via
2210 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2211 be resolved via systemd-resolved D-Bus APIs. In contrast to
2212 the glibc internal resolver systemd-resolved is aware of
2213 multi-homed system, and keeps DNS server and caches separate
2214 and per-interface. Queries are sent simultaneously on all
2215 interfaces that have DNS servers configured, in order to
2216 properly handle VPNs and local LANs which might resolve
2217 separate sets of domain names. systemd-resolved may acquire
2218 DNS server information from systemd-networkd automatically,
2219 which in turn might have discovered them via DHCP. A tool
2220 "systemd-resolve-host" has been added that may be used to
2221 query the DNS logic in resolved. systemd-resolved implements
2222 IDNA and automatically uses IDNA or UTF-8 encoding depending
2223 on whether classic DNS or LLMNR is used as transport. In the
2224 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2225 implementation to systemd-resolved.
2227 * A new NSS module nss-mymachines has been added, that
2228 automatically resolves the names of all local registered
2229 containers to their respective IP addresses.
2231 * A new client tool "networkctl" for systemd-networkd has been
2232 added. It currently is entirely passive and will query
2233 networking configuration from udev, rtnetlink and networkd,
2234 and present it to the user in a very friendly
2235 way. Eventually, we hope to extend it to become a full
2236 control utility for networkd.
2238 * .socket units gained a new DeferAcceptSec= setting that
2239 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2240 TCP. Similarly, support for controlling TCP keep-alive
2241 settings has been added (KeepAliveTimeSec=,
2242 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2243 turning off Nagle's algorithm on TCP has been added
2246 * logind learned a new session type "web", for use in projects
2247 like Cockpit which register web clients as PAM sessions.
2249 * timer units with at least one OnCalendar= setting will now
2250 be started only after timer-sync.target has been
2251 reached. This way they will not elapse before the system
2252 clock has been corrected by a local NTP client or
2253 similar. This is particular useful on RTC-less embedded
2254 machines, that come up with an invalid system clock.
2256 * systemd-nspawn's --network-veth= switch should now result in
2257 stable MAC addresses for both the outer and the inner side
2260 * systemd-nspawn gained a new --volatile= switch for running
2261 container instances with /etc or /var unpopulated.
2263 * The kdbus client code has been updated to use the new Linux
2264 3.17 memfd subsystem instead of the old kdbus-specific one.
2266 * systemd-networkd's DHCP client and server now support
2267 FORCERENEW. There are also new configuration options to
2268 configure the vendor client identifier and broadcast mode
2271 * systemd will no longer inform the kernel about the current
2272 timezone, as this is necessarily incorrect and racy as the
2273 kernel has no understanding of DST and similar
2274 concepts. This hence means FAT timestamps will be always
2275 considered UTC, similar to what Android is already
2276 doing. Also, when the RTC is configured to the local time
2277 (rather than UTC) systemd will never synchronize back to it,
2278 as this might confuse Windows at a later boot.
2280 * systemd-analyze gained a new command "verify" for offline
2281 validation of unit files.
2283 * systemd-networkd gained support for a couple of additional
2284 settings for bonding networking setups. Also, the metric for
2285 statically configured routes may now be configured. For
2286 network interfaces where this is appropriate the peer IP
2287 address may now be configured.
2289 * systemd-networkd's DHCP client will no longer request
2290 broadcasting by default, as this tripped up some networks.
2291 For hardware where broadcast is required the feature should
2292 be switched back on using RequestBroadcast=yes.
2294 * systemd-networkd will now set up IPv4LL addresses (when
2295 enabled) even if DHCP is configured successfully.
2297 * udev will now default to respect network device names given
2298 by the kernel when the kernel indicates that these are
2299 predictable. This behavior can be tweaked by changing
2300 NamePolicy= in the relevant .link file.
2302 * A new library systemd-terminal has been added that
2303 implements full TTY stream parsing and rendering. This
2304 library is supposed to be used later on for implementing a
2305 full userspace VT subsystem, replacing the current kernel
2308 * A new tool systemd-journal-upload has been added to push
2309 journal data to a remote system running
2310 systemd-journal-remote.
2312 * journald will no longer forward all local data to another
2313 running syslog daemon. This change has been made because
2314 rsyslog (which appears to be the most commonly used syslog
2315 implementation these days) no longer makes use of this, and
2316 instead pulls the data out of the journal on its own. Since
2317 forwarding the messages to a non-existent syslog server is
2318 more expensive than we assumed we have now turned this
2319 off. If you run a syslog server that is not a recent rsyslog
2320 version, you have to turn this option on again
2321 (ForwardToSyslog= in journald.conf).
2323 * journald now optionally supports the LZ4 compressor for
2324 larger journal fields. This compressor should perform much
2325 better than XZ which was the previous default.
2327 * machinectl now shows the IP addresses of local containers,
2328 if it knows them, plus the interface name of the container.
2330 * A new tool "systemd-escape" has been added that makes it
2331 easy to escape strings to build unit names and similar.
2333 * sd_notify() messages may now include a new ERRNO= field
2334 which is parsed and collected by systemd and shown among the
2335 "systemctl status" output for a service.
2337 * A new component "systemd-firstboot" has been added that
2338 queries the most basic systemd information (timezone,
2339 hostname, root password) interactively on first
2340 boot. Alternatively it may also be used to provision these
2341 things offline on OS images installed into directories.
2343 * The default sysctl.d/ snippets will now set
2345 net.ipv4.conf.default.promote_secondaries=1
2347 This has the benefit of no flushing secondary IP addresses
2348 when primary addresses are removed.
2350 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2351 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2352 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2353 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2354 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2355 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2356 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2357 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2358 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2359 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2360 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2361 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2362 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2363 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2364 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2366 — Berlin, 2014-08-19
2370 * A new tool systemd-sysusers has been added. This tool
2371 creates system users and groups in /etc/passwd and
2372 /etc/group, based on static declarative system user/group
2373 definitions in /usr/lib/sysusers.d/. This is useful to
2374 enable factory resets and volatile systems that boot up with
2375 an empty /etc directory, and thus need system users and
2376 groups created during early boot. systemd now also ships
2377 with two default sysusers.d/ files for the most basic
2378 users and groups systemd and the core operating system
2381 * A new tmpfiles snippet has been added that rebuilds the
2382 essential files in /etc on boot, should they be missing.
2384 * A directive for ensuring automatic clean-up of
2385 /var/cache/man/ has been removed from the default
2386 configuration. This line should now be shipped by the man
2387 implementation. The necessary change has been made to the
2388 man-db implementation. Note that you need to update your man
2389 implementation to one that ships this line, otherwise no
2390 automatic clean-up of /var/cache/man will take place.
2392 * A new condition ConditionNeedsUpdate= has been added that
2393 may conditionalize services to only run when /etc or /var
2394 are "older" than the vendor operating system resources in
2395 /usr. This is useful for reconstructing or updating /etc
2396 after an offline update of /usr or a factory reset, on the
2397 next reboot. Services that want to run once after such an
2398 update or reset should use this condition and order
2399 themselves before the new systemd-update-done.service, which
2400 will mark the two directories as fully updated. A number of
2401 service files have been added making use of this, to rebuild
2402 the udev hardware database, the journald message catalog and
2403 dynamic loader cache (ldconfig). The systemd-sysusers tool
2404 described above also makes use of this now. With this in
2405 place it is now possible to start up a minimal operating
2406 system with /etc empty cleanly. For more information on the
2407 concepts involved see this recent blog story:
2409 http://0pointer.de/blog/projects/stateless.html
2411 * A new system group "input" has been introduced, and all
2412 input device nodes get this group assigned. This is useful
2413 for system-level software to get access to input devices. It
2414 complements what is already done for "audio" and "video".
2416 * systemd-networkd learnt minimal DHCPv4 server support in
2417 addition to the existing DHCPv4 client support. It also
2418 learnt DHCPv6 client and IPv6 Router Solicitation client
2419 support. The DHCPv4 client gained support for static routes
2420 passed in from the server. Note that the [DHCPv4] section
2421 known in older systemd-networkd versions has been renamed to
2422 [DHCP] and is now also used by the DHCPv6 client. Existing
2423 .network files using settings of this section should be
2424 updated, though compatibility is maintained. Optionally, the
2425 client hostname may now be sent to the DHCP server.
2427 * networkd gained support for vxlan virtual networks as well
2428 as tun/tap and dummy devices.
2430 * networkd gained support for automatic allocation of address
2431 ranges for interfaces from a system-wide pool of
2432 addresses. This is useful for dynamically managing a large
2433 number of interfaces with a single network configuration
2434 file. In particular this is useful to easily assign
2435 appropriate IP addresses to the veth links of a large number
2436 of nspawn instances.
2438 * RPM macros for processing sysusers, sysctl and binfmt
2439 drop-in snippets at package installation time have been
2442 * The /etc/os-release file should now be placed in
2443 /usr/lib/os-release. The old location is automatically
2444 created as symlink. /usr/lib is the more appropriate
2445 location of this file, since it shall actually describe the
2446 vendor operating system shipped in /usr, and not the
2447 configuration stored in /etc.
2449 * .mount units gained a new boolean SloppyOptions= setting
2450 that maps to mount(8)'s -s option which enables permissive
2451 parsing of unknown mount options.
2453 * tmpfiles learnt a new "L+" directive which creates a symlink
2454 but (unlike "L") deletes a pre-existing file first, should
2455 it already exist and not already be the correct
2456 symlink. Similarly, "b+", "c+" and "p+" directives have been
2457 added as well, which create block and character devices, as
2458 well as fifos in the filesystem, possibly removing any
2459 pre-existing files of different types.
2461 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2462 'argument' field (which so far specified the source to
2463 symlink/copy the files from) is now optional. If omitted the
2464 same file os copied from /usr/share/factory/ suffixed by the
2465 full destination path. This is useful for populating /etc
2466 with essential files, by copying them from vendor defaults
2467 shipped in /usr/share/factory/etc.
2469 * A new command "systemctl preset-all" has been added that
2470 applies the service preset settings to all installed unit
2471 files. A new switch --preset-mode= has been added that
2472 controls whether only enable or only disable operations
2475 * A new command "systemctl is-system-running" has been added
2476 that allows checking the overall state of the system, for
2477 example whether it is fully up and running.
2479 * When the system boots up with an empty /etc, the equivalent
2480 to "systemctl preset-all" is executed during early boot, to
2481 make sure all default services are enabled after a factory
2484 * systemd now contains a minimal preset file that enables the
2485 most basic services systemd ships by default.
2487 * Unit files' [Install] section gained a new DefaultInstance=
2488 field for defining the default instance to create if a
2489 template unit is enabled with no instance specified.
2491 * A new passive target cryptsetup-pre.target has been added
2492 that may be used by services that need to make they run and
2493 finish before the first LUKS cryptographic device is set up.
2495 * The /dev/loop-control and /dev/btrfs-control device nodes
2496 are now owned by the "disk" group by default, opening up
2497 access to this group.
2499 * systemd-coredump will now automatically generate a
2500 stack trace of all core dumps taking place on the system,
2501 based on elfutils' libdw library. This stack trace is logged
2504 * systemd-coredump may now optionally store coredumps directly
2505 on disk (in /var/lib/systemd/coredump, possibly compressed),
2506 instead of storing them unconditionally in the journal. This
2507 mode is the new default. A new configuration file
2508 /etc/systemd/coredump.conf has been added to configure this
2509 and other parameters of systemd-coredump.
2511 * coredumpctl gained a new "info" verb to show details about a
2512 specific coredump. A new switch "-1" has also been added
2513 that makes sure to only show information about the most
2514 recent entry instead of all entries. Also, as the tool is
2515 generally useful now the "systemd-" prefix of the binary
2516 name has been removed. Distributions that want to maintain
2517 compatibility with the old name should add a symlink from
2518 the old name to the new name.
2520 * journald's SplitMode= now defaults to "uid". This makes sure
2521 that unprivileged users can access their own coredumps with
2522 coredumpctl without restrictions.
2524 * New kernel command line options "systemd.wants=" (for
2525 pulling an additional unit during boot), "systemd.mask="
2526 (for masking a specific unit for the boot), and
2527 "systemd.debug-shell" (for enabling the debug shell on tty9)
2528 have been added. This is implemented in the new generator
2529 "systemd-debug-generator".
2531 * systemd-nspawn will now by default filter a couple of
2532 syscalls for containers, among them those required for
2533 kernel module loading, direct x86 IO port access, swap
2534 management, and kexec. Most importantly though
2535 open_by_handle_at() is now prohibited for containers,
2536 closing a hole similar to a recently discussed vulnerability
2537 in docker regarding access to files on file hierarchies the
2538 container should normally not have access to. Note that, for
2539 nspawn, we generally make no security claims anyway (and
2540 this is explicitly documented in the man page), so this is
2541 just a fix for one of the most obvious problems.
2543 * A new man page file-hierarchy(7) has been added that
2544 contains a minimized, modernized version of the file system
2545 layout systemd expects, similar in style to the FHS
2546 specification or hier(5). A new tool systemd-path(1) has
2547 been added to query many of these paths for the local
2550 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2551 longer done. Since the directory now has a per-user size
2552 limit, and is cleaned on logout this appears unnecessary,
2553 in particular since this now brings the lifecycle of this
2554 directory closer in line with how IPC objects are handled.
2556 * systemd.pc now exports a number of additional directories,
2557 including $libdir (which is useful to identify the library
2558 path for the primary architecture of the system), and a
2559 couple of drop-in directories.
2561 * udev's predictable network interface names now use the dev_port
2562 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2563 distinguish between ports of the same PCI function. dev_id should
2564 only be used for ports using the same HW address, hence the need
2567 * machined has been updated to export the OS version of a
2568 container (read from /etc/os-release and
2569 /usr/lib/os-release) on the bus. This is now shown in
2570 "machinectl status" for a machine.
2572 * A new service setting RestartForceExitStatus= has been
2573 added. If configured to a set of exit signals or process
2574 return values, the service will be restarted when the main
2575 daemon process exits with any of them, regardless of the
2578 * systemctl's -H switch for connecting to remote systemd
2579 machines has been extended so that it may be used to
2580 directly connect to a specific container on the
2581 host. "systemctl -H root@foobar:waldi" will now connect as
2582 user "root" to host "foobar", and then proceed directly to
2583 the container named "waldi". Note that currently you have to
2584 authenticate as user "root" for this to work, as entering
2585 containers is a privileged operation.
2587 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2588 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2589 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2590 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2591 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2592 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2593 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2594 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2595 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2596 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2597 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2598 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2600 — Berlin, 2014-07-03
2604 * As an experimental feature, udev now tries to lock the
2605 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2606 executes events for the disk or any of its partitions.
2607 Applications like partitioning programs can lock the
2608 disk device node (flock(LOCK_EX)) and claim temporary
2609 device ownership that way; udev will entirely skip all event
2610 handling for this disk and its partitions. If the disk
2611 was opened for writing, the close will trigger a partition
2612 table rescan in udev's "watch" facility, and if needed
2613 synthesize "change" events for the disk and all its partitions.
2614 This is now unconditionally enabled, and if it turns out to
2615 cause major problems, we might turn it on only for specific
2616 devices, or might need to disable it entirely. Device Mapper
2617 devices are excluded from this logic.
2619 * We temporarily dropped the "-l" switch for fsck invocations,
2620 since they collide with the flock() logic above. util-linux
2621 upstream has been changed already to avoid this conflict,
2622 and we will readd "-l" as soon as util-linux with this
2623 change has been released.
2625 * The dependency on libattr has been removed. Since a long
2626 time, the extended attribute calls have moved to glibc, and
2627 libattr is thus unnecessary.
2629 * Virtualization detection works without priviliges now. This
2630 means the systemd-detect-virt binary no longer requires
2631 CAP_SYS_PTRACE file capabilities, and our daemons can run
2632 with fewer privileges.
2634 * systemd-networkd now runs under its own "systemd-network"
2635 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2636 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2637 loses the ability to write to files owned by root this way.
2639 * Similarly, systemd-resolved now runs under its own
2640 "systemd-resolve" user with no capabilities remaining.
2642 * Similarly, systemd-bus-proxyd now runs under its own
2643 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2645 * systemd-networkd gained support for setting up "veth"
2646 virtual Ethernet devices for container connectivity, as well
2647 as GRE and VTI tunnels.
2649 * systemd-networkd will no longer automatically attempt to
2650 manually load kernel modules necessary for certain tunnel
2651 transports. Instead, it is assumed the kernel loads them
2652 automatically when required. This only works correctly on
2653 very new kernels. On older kernels, please consider adding
2654 the kernel modules to /etc/modules-load.d/ as a work-around.
2656 * The resolv.conf file systemd-resolved generates has been
2657 moved to /run/systemd/resolve/. If you have a symlink from
2658 /etc/resolv.conf, it might be necessary to correct it.
2660 * Two new service settings, ProtectHome= and ProtectSystem=,
2661 have been added. When enabled, they will make the user data
2662 (such as /home) inaccessible or read-only and the system
2663 (such as /usr) read-only, for specific services. This allows
2664 very light-weight per-service sandboxing to avoid
2665 modifications of user data or system files from
2666 services. These two new switches have been enabled for all
2667 of systemd's long-running services, where appropriate.
2669 * Socket units gained new SocketUser= and SocketGroup=
2670 settings to set the owner user and group of AF_UNIX sockets
2671 and FIFOs in the file system.
2673 * Socket units gained a new RemoveOnStop= setting. If enabled,
2674 all FIFOS and sockets in the file system will be removed
2675 when the specific socket unit is stopped.
2677 * Socket units gained a new Symlinks= setting. It takes a list
2678 of symlinks to create to file system sockets or FIFOs
2679 created by the specific Unix sockets. This is useful to
2680 manage symlinks to socket nodes with the same life-cycle as
2683 * The /dev/log socket and /dev/initctl FIFO have been moved to
2684 /run, and have been replaced by symlinks. This allows
2685 connecting to these facilities even if PrivateDevices=yes is
2686 used for a service (which makes /dev/log itself unavailable,
2687 but /run is left). This also has the benefit of ensuring
2688 that /dev only contains device nodes, directories and
2689 symlinks, and nothing else.
2691 * sd-daemon gained two new calls sd_pid_notify() and
2692 sd_pid_notifyf(). They are similar to sd_notify() and
2693 sd_notifyf(), but allow overriding of the source PID of
2694 notification messages if permissions permit this. This is
2695 useful to send notify messages on behalf of a different
2696 process (for example, the parent process). The
2697 systemd-notify tool has been updated to make use of this
2698 when sending messages (so that notification messages now
2699 originate from the shell script invoking systemd-notify and
2700 not the systemd-notify process itself. This should minimize
2701 a race where systemd fails to associate notification
2702 messages to services when the originating process already
2705 * A new "on-abnormal" setting for Restart= has been added. If
2706 set, it will result in automatic restarts on all "abnormal"
2707 reasons for a process to exit, which includes unclean
2708 signals, core dumps, timeouts and watchdog timeouts, but
2709 does not include clean and unclean exit codes or clean
2710 signals. Restart=on-abnormal is an alternative for
2711 Restart=on-failure for services that shall be able to
2712 terminate and avoid restarts on certain errors, by
2713 indicating so with an unclean exit code. Restart=on-failure
2714 or Restart=on-abnormal is now the recommended setting for
2715 all long-running services.
2717 * If the InaccessibleDirectories= service setting points to a
2718 mount point (or if there are any submounts contained within
2719 it), it is now attempted to completely unmount it, to make
2720 the file systems truly unavailable for the respective
2723 * The ReadOnlyDirectories= service setting and
2724 systemd-nspawn's --read-only parameter are now recursively
2725 applied to all submounts, too.
2727 * Mount units may now be created transiently via the bus APIs.
2729 * The support for SysV and LSB init scripts has been removed
2730 from the systemd daemon itself. Instead, it is now
2731 implemented as a generator that creates native systemd units
2732 from these scripts when needed. This enables us to remove a
2733 substantial amount of legacy code from PID 1, following the
2734 fact that many distributions only ship a very small number
2735 of LSB/SysV init scripts nowadays.
2737 * Privileged Xen (dom0) domains are not considered
2738 virtualization anymore by the virtualization detection
2739 logic. After all, they generally have unrestricted access to
2740 the hardware and usually are used to manage the unprivileged
2743 * systemd-tmpfiles gained a new "C" line type, for copying
2744 files or entire directories.
2746 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2747 lines. So far, they have been non-globbing versions of the
2748 latter, and have thus been redundant. In future, it is
2749 recommended to only use "z". "m" has hence been removed
2750 from the documentation, even though it stays supported.
2752 * A tmpfiles snippet to recreate the most basic structure in
2753 /var has been added. This is enough to create the /var/run →
2754 /run symlink and create a couple of structural
2755 directories. This allows systems to boot up with an empty or
2756 volatile /var. Of course, while with this change, the core OS
2757 now is capable with dealing with a volatile /var, not all
2758 user services are ready for it. However, we hope that sooner
2759 or later, many service daemons will be changed upstream so
2760 that they are able to automatically create their necessary
2761 directories in /var at boot, should they be missing. This is
2762 the first step to allow state-less systems that only require
2763 the vendor image for /usr to boot.
2765 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2766 empty tmpfs instance to a specific directory. This is
2767 particularly useful for making use of the automatic
2768 reconstruction of /var (see above), by passing --tmpfs=/var.
2770 * Access modes specified in tmpfiles snippets may now be
2771 prefixed with "~", which indicates that they shall be masked
2772 by whether the existing file or directory is currently
2773 writable, readable or executable at all. Also, if specified,
2774 the sgid/suid/sticky bits will be masked for all
2777 * A new passive target unit "network-pre.target" has been
2778 added which is useful for services that shall run before any
2779 network is configured, for example firewall scripts.
2781 * The "floppy" group that previously owned the /dev/fd*
2782 devices is no longer used. The "disk" group is now used
2783 instead. Distributions should probably deprecate usage of
2786 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2787 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2788 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2789 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2790 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2791 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2794 — Berlin, 2014-06-11
2798 * A new "systemd-timesyncd" daemon has been added for
2799 synchronizing the system clock across the network. It
2800 implements an SNTP client. In contrast to NTP
2801 implementations such as chrony or the NTP reference server,
2802 this only implements a client side, and does not bother with
2803 the full NTP complexity, focusing only on querying time from
2804 one remote server and synchronizing the local clock to
2805 it. Unless you intend to serve NTP to networked clients or
2806 want to connect to local hardware clocks, this simple NTP
2807 client should be more than appropriate for most
2808 installations. The daemon runs with minimal privileges, and
2809 has been hooked up with networkd to only operate when
2810 network connectivity is available. The daemon saves the
2811 current clock to disk every time a new NTP sync has been
2812 acquired, and uses this to possibly correct the system clock
2813 early at bootup, in order to accommodate for systems that
2814 lack an RTC such as the Raspberry Pi and embedded devices,
2815 and to make sure that time monotonically progresses on these
2816 systems, even if it is not always correct. To make use of
2817 this daemon, a new system user and group "systemd-timesync"
2818 needs to be created on installation of systemd.
2820 * The queue "seqnum" interface of libudev has been disabled, as
2821 it was generally incompatible with device namespacing as
2822 sequence numbers of devices go "missing" if the devices are
2823 part of a different namespace.
2825 * "systemctl list-timers" and "systemctl list-sockets" gained
2826 a --recursive switch for showing units of these types also
2827 for all local containers, similar in style to the already
2828 supported --recursive switch for "systemctl list-units".
2830 * A new RebootArgument= setting has been added for service
2831 units, which may be used to specify a kernel reboot argument
2832 to use when triggering reboots with StartLimitAction=.
2834 * A new FailureAction= setting has been added for service
2835 units which may be used to specify an operation to trigger
2836 when a service fails. This works similarly to
2837 StartLimitAction=, but unlike it, controls what is done
2838 immediately rather than only after several attempts to
2839 restart the service in question.
2841 * hostnamed got updated to also expose the kernel name,
2842 release, and version on the bus. This is useful for
2843 executing commands like hostnamectl with the -H switch.
2844 systemd-analyze makes use of this to properly display
2845 details when running non-locally.
2847 * The bootchart tool can now show cgroup information in the
2848 graphs it generates.
2850 * The CFS CPU quota cgroup attribute is now exposed for
2851 services. The new CPUQuota= switch has been added for this
2852 which takes a percentage value. Setting this will have the
2853 result that a service may never get more CPU time than the
2854 specified percentage, even if the machine is otherwise idle.
2856 * systemd-networkd learned IPIP and SIT tunnel support.
2858 * LSB init scripts exposing a dependency on $network will now
2859 get a dependency on network-online.target rather than simply
2860 network.target. This should bring LSB handling closer to
2861 what it was on SysV systems.
2863 * A new fsck.repair= kernel option has been added to control
2864 how fsck shall deal with unclean file systems at boot.
2866 * The (.ini) configuration file parser will now silently
2867 ignore sections whose name begins with "X-". This may be
2868 used to maintain application-specific extension sections in unit
2871 * machined gained a new API to query the IP addresses of
2872 registered containers. "machinectl status" has been updated
2873 to show these addresses in its output.
2875 * A new call sd_uid_get_display() has been added to the
2876 sd-login APIs for querying the "primary" session of a
2877 user. The "primary" session of the user is elected from the
2878 user's sessions and generally a graphical session is
2879 preferred over a text one.
2881 * A minimal systemd-resolved daemon has been added. It
2882 currently simply acts as a companion to systemd-networkd and
2883 manages resolv.conf based on per-interface DNS
2884 configuration, possibly supplied via DHCP. In the long run
2885 we hope to extend this into a local DNSSEC enabled DNS and
2888 * The systemd-networkd-wait-online tool is now enabled by
2889 default. It will delay network-online.target until a network
2890 connection has been configured. The tool primarily integrates
2891 with networkd, but will also make a best effort to make sense
2892 of network configuration performed in some other way.
2894 * Two new service options StartupCPUShares= and
2895 StartupBlockIOWeight= have been added that work similarly to
2896 CPUShares= and BlockIOWeight= however only apply during
2897 system startup. This is useful to prioritize certain services
2898 differently during bootup than during normal runtime.
2900 * hostnamed has been changed to prefer the statically
2901 configured hostname in /etc/hostname (unless set to
2902 'localhost' or empty) over any dynamic one supplied by
2903 dhcp. With this change, the rules for picking the hostname
2904 match more closely the rules of other configuration settings
2905 where the local administrator's configuration in /etc always
2906 overrides any other settings.
2908 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2909 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2910 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2911 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2912 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2913 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2914 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2915 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2916 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2917 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2918 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2919 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2920 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2921 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2922 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2923 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2926 — Beijing, 2014-05-28
2930 * When restoring the screen brightness at boot, stay away from
2931 the darkest setting or from the lowest 5% of the available
2932 range, depending on which is the larger value of both. This
2933 should effectively protect the user from rebooting into a
2934 black screen, should the brightness have been set to minimum
2937 * sd-login gained a new sd_machine_get_class() call to
2938 determine the class ("vm" or "container") of a machine
2939 registered with machined.
2941 * sd-login gained new calls
2942 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2943 to query the identity of the peer of a local AF_UNIX
2944 connection. They operate similarly to their sd_pid_get_xyz()
2947 * PID 1 will now maintain a system-wide system state engine
2948 with the states "starting", "running", "degraded",
2949 "maintenance", "stopping". These states are bound to system
2950 startup, normal runtime, runtime with at least one failed
2951 service, rescue/emergency mode and system shutdown. This
2952 state is shown in the "systemctl status" output when no unit
2953 name is passed. It is useful to determine system state, in
2954 particularly when doing so for many systems or containers at
2957 * A new command "list-machines" has been added to "systemctl"
2958 that lists all local OS containers and shows their system
2959 state (see above), if systemd runs inside of them.
2961 * systemctl gained a new "-r" switch to recursively enumerate
2962 units on all local containers, when used with the
2963 "list-unit" command (which is the default one that is
2964 executed when no parameters are specified).
2966 * The GPT automatic partition discovery logic will now honour
2967 two GPT partition flags: one may be set on a partition to
2968 cause it to be mounted read-only, and the other may be set
2969 on a partition to ignore it during automatic discovery.
2971 * Two new GPT type UUIDs have been added for automatic root
2972 partition discovery, for 32-bit and 64-bit ARM. This is not
2973 particularly useful for discovering the root directory on
2974 these architectures during bare-metal boots (since UEFI is
2975 not common there), but still very useful to allow booting of
2976 ARM disk images in nspawn with the -i option.
2978 * MAC addresses of interfaces created with nspawn's
2979 --network-interface= switch will now be generated from the
2980 machine name, and thus be stable between multiple invocations
2983 * logind will now automatically remove all IPC objects owned
2984 by a user if she or he fully logs out. This makes sure that
2985 users who are logged out cannot continue to consume IPC
2986 resources. This covers SysV memory, semaphores and message
2987 queues as well as POSIX shared memory and message
2988 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2989 limits. With this functionality, that is corrected. This may
2990 be turned off by using the RemoveIPC= switch of logind.conf.
2992 * The systemd-machine-id-setup and tmpfiles tools gained a
2993 --root= switch to operate on a specific root directory,
2996 * journald can now forward logged messages to the TTYs of all
2997 logged in users ("wall"). This is the default for all
2998 emergency messages now.
3000 * A new tool systemd-journal-remote has been added to stream
3001 journal log messages across the network.
3003 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3004 controller trees are mounted into it. Note that the
3005 directories mounted beneath it are not read-only. This is a
3006 security measure and is particularly useful because glibc
3007 actually includes a search logic to pick any tmpfs it can
3008 find to implement shm_open() if /dev/shm is not available
3009 (which it might very well be in namespaced setups).
3011 * machinectl gained a new "poweroff" command to cleanly power
3012 down a local OS container.
3014 * The PrivateDevices= unit file setting will now also drop the
3015 CAP_MKNOD capability from the capability bound set, and
3016 imply DevicePolicy=closed.
3018 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3019 comprehensively on all long-running systemd services where
3020 this is appropriate.
3022 * systemd-udevd will now run in a disassociated mount
3023 namespace. To mount directories from udev rules, make sure to
3024 pull in mount units via SYSTEMD_WANTS properties.
3026 * The kdbus support gained support for uploading policy into
3027 the kernel. sd-bus gained support for creating "monitoring"
3028 connections that can eavesdrop into all bus communication
3029 for debugging purposes.
3031 * Timestamps may now be specified in seconds since the UNIX
3032 epoch Jan 1st, 1970 by specifying "@" followed by the value
3035 * Native tcpwrap support in systemd has been removed. tcpwrap
3036 is old code, not really maintained anymore and has serious
3037 shortcomings, and better options such as firewalls
3038 exist. For setups that require tcpwrap usage, please
3039 consider invoking your socket-activated service via tcpd,
3040 like on traditional inetd.
3042 * A new system.conf configuration option
3043 DefaultTimerAccuracySec= has been added that controls the
3044 default AccuracySec= setting of .timer units.
3046 * Timer units gained a new WakeSystem= switch. If enabled,
3047 timers configured this way will cause the system to resume
3048 from system suspend (if the system supports that, which most
3051 * Timer units gained a new Persistent= switch. If enabled,
3052 timers configured this way will save to disk when they have
3053 been last triggered. This information is then used on next
3054 reboot to possible execute overdue timer events, that
3055 could not take place because the system was powered off.
3056 This enables simple anacron-like behaviour for timer units.
3058 * systemctl's "list-timers" will now also list the time a
3059 timer unit was last triggered in addition to the next time
3060 it will be triggered.
3062 * systemd-networkd will now assign predictable IPv4LL
3063 addresses to its local interfaces.
3065 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3066 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3067 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3068 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3069 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3070 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3071 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3072 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3075 — Berlin, 2014-03-25
3079 * A new unit file setting RestrictAddressFamilies= has been
3080 added to restrict which socket address families unit
3081 processes gain access to. This takes address family names
3082 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3083 attack surface of services via exotic protocol stacks. This
3084 is built on seccomp system call filters.
3086 * Two new unit file settings RuntimeDirectory= and
3087 RuntimeDirectoryMode= have been added that may be used to
3088 manage a per-daemon runtime directories below /run. This is
3089 an alternative for setting up directory permissions with
3090 tmpfiles snippets, and has the advantage that the runtime
3091 directory's lifetime is bound to the daemon runtime and that
3092 the daemon starts up with an empty directory each time. This
3093 is particularly useful when writing services that drop
3094 privileges using the User= or Group= setting.
3096 * The DeviceAllow= unit setting now supports globbing for
3097 matching against device group names.
3099 * The systemd configuration file system.conf gained new
3100 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3101 DefaultMemoryAccounting= to globally turn on/off accounting
3102 for specific resources (cgroups) for all units. These
3103 settings may still be overridden individually in each unit
3106 * systemd-gpt-auto-generator is now able to discover /srv and
3107 root partitions in addition to /home and swap partitions. It
3108 also supports LUKS-encrypted partitions now. With this in
3109 place, automatic discovery of partitions to mount following
3110 the Discoverable Partitions Specification
3111 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3112 is now a lot more complete. This allows booting without
3113 /etc/fstab and without root= on the kernel command line on
3114 systems prepared appropriately.
3116 * systemd-nspawn gained a new --image= switch which allows
3117 booting up disk images and Linux installations on any block
3118 device that follow the Discoverable Partitions Specification
3119 (see above). This means that installations made with
3120 appropriately updated installers may now be started and
3121 deployed using container managers, completely
3122 unmodified. (We hope that libvirt-lxc will add support for
3123 this feature soon, too.)
3125 * systemd-nspawn gained a new --network-macvlan= setting to
3126 set up a private macvlan interface for the
3127 container. Similarly, systemd-networkd gained a new
3128 Kind=macvlan setting in .netdev files.
3130 * systemd-networkd now supports configuring local addresses
3133 * A new tool systemd-network-wait-online has been added to
3134 synchronously wait for network connectivity using
3137 * The sd-bus.h bus API gained a new sd_bus_track object for
3138 tracking the life-cycle of bus peers. Note that sd-bus.h is
3139 still not a public API though (unless you specify
3140 --enable-kdbus on the configure command line, which however
3141 voids your warranty and you get no API stability guarantee).
3143 * The $XDG_RUNTIME_DIR runtime directories for each user are
3144 now individual tmpfs instances, which has the benefit of
3145 introducing separate pools for each user, with individual
3146 size limits, and thus making sure that unprivileged clients
3147 can no longer negatively impact the system or other users by
3148 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3149 RuntimeDirectorySize= has been introduced that allows
3150 controlling the default size limit for all users. It
3151 defaults to 10% of the available physical memory. This is no
3152 replacement for quotas on tmpfs though (which the kernel
3153 still does not support), as /dev/shm and /tmp are still
3154 shared resources used by both the system and unprivileged
3157 * logind will now automatically turn off automatic suspending
3158 on laptop lid close when more than one display is
3159 connected. This was previously expected to be implemented
3160 individually in desktop environments (such as GNOME),
3161 however has been added to logind now, in order to fix a
3162 boot-time race where a desktop environment might not have
3163 been started yet and thus not been able to take an inhibitor
3164 lock at the time where logind already suspends the system
3165 due to a closed lid.
3167 * logind will now wait at least 30s after each system
3168 suspend/resume cycle, and 3min after system boot before
3169 suspending the system due to a closed laptop lid. This
3170 should give USB docking stations and similar enough time to
3171 be probed and configured after system resume and boot in
3172 order to then act as suspend blocker.
3174 * systemd-run gained a new --property= setting which allows
3175 initialization of resource control properties (and others)
3176 for the created scope or service unit. Example: "systemd-run
3177 --property=BlockIOWeight=10 updatedb" may be used to run
3178 updatedb at a low block IO scheduling weight.
3180 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3181 now also work in --scope mode.
3183 * When systemd is compiled with kdbus support, basic support
3184 for enforced policies is now in place. (Note that enabling
3185 kdbus still voids your warranty and no API compatibility
3188 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3189 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3190 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3191 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3192 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3193 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3194 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3195 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3196 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3197 Zbigniew Jędrzejewski-Szmek
3199 — Berlin, 2014-03-12
3203 * systemd will now relabel /dev after loading the SMACK policy
3204 according to SMACK rules.
3206 * A new unit file option AppArmorProfile= has been added to
3207 set the AppArmor profile for the processes of a unit.
3209 * A new condition check ConditionArchitecture= has been added
3210 to conditionalize units based on the system architecture, as
3211 reported by uname()'s "machine" field.
3213 * systemd-networkd now supports matching on the system
3214 virtualization, architecture, kernel command line, host name
3217 * logind is now a lot more aggressive when suspending the
3218 machine due to a closed laptop lid. Instead of acting only
3219 on the lid close action, it will continuously watch the lid
3220 status and act on it. This is useful for laptops where the
3221 power button is on the outside of the chassis so that it can
3222 be reached without opening the lid (such as the Lenovo
3223 Yoga). On those machines, logind will now immediately
3224 re-suspend the machine if the power button has been
3225 accidentally pressed while the laptop was suspended and in a
3226 backpack or similar.
3228 * logind will now watch SW_DOCK switches and inhibit reaction
3229 to the lid switch if it is pressed. This means that logind
3230 will not suspend the machine anymore if the lid is closed
3231 and the system is docked, if the laptop supports SW_DOCK
3232 notifications via the input layer. Note that ACPI docking
3233 stations do not generate this currently. Also note that this
3234 logic is usually not fully sufficient and Desktop
3235 Environments should take a lid switch inhibitor lock when an
3236 external display is connected, as systemd will not watch
3239 * nspawn will now make use of the devices cgroup controller by
3240 default, and only permit creation of and access to the usual
3241 API device nodes like /dev/null or /dev/random, as well as
3242 access to (but not creation of) the pty devices.
3244 * We will now ship a default .network file for
3245 systemd-networkd that automatically configures DHCP for
3246 network interfaces created by nspawn's --network-veth or
3247 --network-bridge= switches.
3249 * systemd will now understand the usual M, K, G, T suffixes
3250 according to SI conventions (i.e. to the base 1000) when
3251 referring to throughput and hardware metrics. It will stay
3252 with IEC conventions (i.e. to the base 1024) for software
3253 metrics, according to what is customary according to
3254 Wikipedia. We explicitly document which base applies for
3255 each configuration option.
3257 * The DeviceAllow= setting in unit files now supports a syntax
3258 to whitelist an entire group of devices node majors at once,
3259 based on the /proc/devices listing. For example, with the
3260 string "char-pts", it is now possible to whitelist all
3261 current and future pseudo-TTYs at once.
3263 * sd-event learned a new "post" event source. Event sources of
3264 this type are triggered by the dispatching of any event
3265 source of a type that is not "post". This is useful for
3266 implementing clean-up and check event sources that are
3267 triggered by other work being done in the program.
3269 * systemd-networkd is no longer statically enabled, but uses
3270 the usual [Install] sections so that it can be
3271 enabled/disabled using systemctl. It still is enabled by
3274 * When creating a veth interface pair with systemd-nspawn, the
3275 host side will now be prefixed with "vb-" if
3276 --network-bridge= is used, and with "ve-" if --network-veth
3277 is used. This way, it is easy to distinguish these cases on
3278 the host, for example to apply different configuration to
3279 them with systemd-networkd.
3281 * The compatibility libraries for libsystemd-journal.so,
3282 libsystem-id128.so, libsystemd-login.so and
3283 libsystemd-daemon.so do not make use of IFUNC
3284 anymore. Instead, we now build libsystemd.so multiple times
3285 under these alternative names. This means that the footprint
3286 is drastically increased, but given that these are
3287 transitional compatibility libraries, this should not matter
3288 much. This change has been made necessary to support the ARM
3289 platform for these compatibility libraries, as the ARM
3290 toolchain is not really at the same level as the toolchain
3291 for other architectures like x86 and does not support
3292 IFUNC. Please make sure to use --enable-compat-libs only
3293 during a transitional period!
3295 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3296 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3297 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3298 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3299 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3300 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3301 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3302 Zbigniew Jędrzejewski-Szmek
3304 — Berlin, 2014-02-24
3308 * A new component "systemd-networkd" has been added that can
3309 be used to configure local network interfaces statically or
3310 via DHCP. It is capable of bringing up bridges, VLANs, and
3311 bonding. Currently, no hook-ups for interactive network
3312 configuration are provided. Use this for your initrd,
3313 container, embedded, or server setup if you need a simple,
3314 yet powerful, network configuration solution. This
3315 configuration subsystem is quite nifty, as it allows wildcard
3316 hotplug matching in interfaces. For example, with a single
3317 configuration snippet, you can configure that all Ethernet
3318 interfaces showing up are automatically added to a bridge,
3319 or similar. It supports link-sensing and more.
3321 * A new tool "systemd-socket-proxyd" has been added which can
3322 act as a bidirectional proxy for TCP sockets. This is
3323 useful for adding socket activation support to services that
3324 do not actually support socket activation, including virtual
3325 machines and the like.
3327 * Add a new tool to save/restore rfkill state on
3330 * Save/restore state of keyboard backlights in addition to
3331 display backlights on shutdown/boot.
3333 * udev learned a new SECLABEL{} construct to label device
3334 nodes with a specific security label when they appear. For
3335 now, only SECLABEL{selinux} is supported, but the syntax is
3336 prepared for additional security frameworks.
3338 * udev gained a new scheme to configure link-level attributes
3339 from files in /etc/systemd/network/*.link. These files can
3340 match against MAC address, device path, driver name and type,
3341 and will apply attributes like the naming policy, link speed,
3342 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3343 address assignment policy (randomized, ...).
3345 * The configuration of network interface naming rules for
3346 "permanent interface names" has changed: a new NamePolicy=
3347 setting in the [Link] section of .link files determines the
3348 priority of possible naming schemes (onboard, slot, MAC,
3349 path). The default value of this setting is determined by
3350 /usr/lib/net/links/99-default.link. Old
3351 80-net-name-slot.rules udev configuration file has been
3352 removed, so local configuration overriding this file should
3353 be adapated to override 99-default.link instead.
3355 * When the User= switch is used in a unit file, also
3356 initialize $SHELL= based on the user database entry.
3358 * systemd no longer depends on libdbus. All communication is
3359 now done with sd-bus, systemd's low-level bus library
3362 * kdbus support has been added to PID 1 itself. When kdbus is
3363 enabled, this causes PID 1 to set up the system bus and
3364 enable support for a new ".busname" unit type that
3365 encapsulates bus name activation on kdbus. It works a little
3366 bit like ".socket" units, except for bus names. A new
3367 generator has been added that converts classic dbus1 service
3368 activation files automatically into native systemd .busname
3371 * sd-bus: add a light-weight vtable implementation that allows
3372 defining objects on the bus with a simple static const
3373 vtable array of its methods, signals and properties.
3375 * systemd will not generate or install static dbus
3376 introspection data anymore to /usr/share/dbus-1/interfaces,
3377 as the precise format of these files is unclear, and
3378 nothing makes use of it.
3380 * A proxy daemon is now provided to proxy clients connecting
3381 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3382 compatibility with classic D-Bus.
3384 * A bus driver implementation has been added that supports the
3385 classic D-Bus bus driver calls on kdbus, also for
3386 compatibility purposes.
3388 * A new API "sd-event.h" has been added that implements a
3389 minimal event loop API built around epoll. It provides a
3390 couple of features that direct epoll usage is lacking:
3391 prioritization of events, scales to large numbers of timer
3392 events, per-event timer slack (accuracy), system-wide
3393 coalescing of timer events, exit handlers, watchdog
3394 supervision support using systemd's sd_notify() API, child
3397 * A new API "sd-rntl.h" has been added that provides an API
3398 around the route netlink interface of the kernel, similar in
3399 style to "sd-bus.h".
3401 * A new API "sd-dhcp-client.h" has been added that provides a
3402 small DHCPv4 client-side implementation. This is used by
3405 * There is a new kernel command line option
3406 "systemd.restore_state=0|1". When set to "0", none of the
3407 systemd tools will restore saved runtime state to hardware
3408 devices. More specifically, the rfkill and backlight states
3411 * The FsckPassNo= compatibility option in mount/service units
3412 has been removed. The fstab generator will now add the
3413 necessary dependencies automatically, and does not require
3414 PID1's support for that anymore.
3416 * journalctl gained a new switch, --list-boots, that lists
3417 recent boots with their times and boot IDs.
3419 * The various tools like systemctl, loginctl, timedatectl,
3420 busctl, systemd-run, ... have gained a new switch "-M" to
3421 connect to a specific, local OS container (as direct
3422 connection, without requiring SSH). This works on any
3423 container that is registered with machined, such as those
3424 created by libvirt-lxc or nspawn.
3426 * systemd-run and systemd-analyze also gained support for "-H"
3427 to connect to remote hosts via SSH. This is particularly
3428 useful for systemd-run because it enables queuing of jobs
3429 onto remote systems.
3431 * machinectl gained a new command "login" to open a getty
3432 login in any local container. This works with any container
3433 that is registered with machined (such as those created by
3434 libvirt-lxc or nspawn), and which runs systemd inside.
3436 * machinectl gained a new "reboot" command that may be used to
3437 trigger a reboot on a specific container that is registered
3438 with machined. This works on any container that runs an init
3439 system of some kind.
3441 * systemctl gained a new "list-timers" command to print a nice
3442 listing of installed timer units with the times they elapse
3445 * Alternative reboot() parameters may now be specified on the
3446 "systemctl reboot" command line and are passed to the
3447 reboot() system call.
3449 * systemctl gained a new --job-mode= switch to configure the
3450 mode to queue a job with. This is a more generic version of
3451 --fail, --irreversible, and --ignore-dependencies, which are
3452 still available but not advertised anymore.
3454 * /etc/systemd/system.conf gained new settings to configure
3455 various default timeouts of units, as well as the default
3456 start limit interval and burst. These may still be overridden
3459 * PID1 will now export on the bus profile data of the security
3460 policy upload process (such as the SELinux policy upload to
3463 * journald: when forwarding logs to the console, include
3464 timestamps (following the setting in
3465 /sys/module/printk/parameters/time).
3467 * OnCalendar= in timer units now understands the special
3468 strings "yearly" and "annually". (Both are equivalent)
3470 * The accuracy of timer units is now configurable with the new
3471 AccuracySec= setting. It defaults to 1min.
3473 * A new dependency type JoinsNamespaceOf= has been added that
3474 allows running two services within the same /tmp and network
3475 namespace, if PrivateNetwork= or PrivateTmp= are used.
3477 * A new command "cat" has been added to systemctl. It outputs
3478 the original unit file of a unit, and concatenates the
3479 contents of additional "drop-in" unit file snippets, so that
3480 the full configuration is shown.
3482 * systemctl now supports globbing on the various "list-xyz"
3483 commands, like "list-units" or "list-sockets", as well as on
3484 those commands which take multiple unit names.
3486 * journalctl's --unit= switch gained support for globbing.
3488 * All systemd daemons now make use of the watchdog logic so
3489 that systemd automatically notices when they hang.
3491 * If the $container_ttys environment variable is set,
3492 getty-generator will automatically spawn a getty for each
3493 listed tty. This is useful for container managers to request
3494 login gettys to be spawned on as many ttys as needed.
3496 * %h, %s, %U specifier support is not available anymore when
3497 used in unit files for PID 1. This is because NSS calls are
3498 not safe from PID 1. They stay available for --user
3499 instances of systemd, and as special case for the root user.
3501 * loginctl gained a new "--no-legend" switch to turn off output
3504 * The "sd-login.h" API gained three new calls:
3505 sd_session_is_remote(), sd_session_get_remote_user(),
3506 sd_session_get_remote_host() to query information about
3509 * The udev hardware database now also carries vendor/product
3510 information of SDIO devices.
3512 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3513 determine whether watchdog notifications are requested by
3516 * Socket-activated per-connection services now include a
3517 short description of the connection parameters in the
3520 * tmpfiles gained a new "--boot" option. When this is not used,
3521 only lines where the command character is not suffixed with
3522 "!" are executed. When this option is specified, those
3523 options are executed too. This partitions tmpfiles
3524 directives into those that can be safely executed at any
3525 time, and those which should be run only at boot (for
3526 example, a line that creates /run/nologin).
3528 * A new API "sd-resolve.h" has been added which provides a simple
3529 asynchronous wrapper around glibc NSS host name resolution
3530 calls, such as getaddrinfo(). In contrast to glibc's
3531 getaddrinfo_a(), it does not use signals. In contrast to most
3532 other asynchronous name resolution libraries, this one does
3533 not reimplement DNS, but reuses NSS, so that alternate
3534 host name resolution systems continue to work, such as mDNS,
3535 LDAP, etc. This API is based on libasyncns, but it has been
3536 cleaned up for inclusion in systemd.
3538 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3539 "sd-daemon.h" are no longer found in individual libraries
3540 libsystemd-journal.so, libsystemd-login.so,
3541 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3542 merged them into a single library, libsystemd.so, which
3543 provides all symbols. The reason for this is cyclic
3544 dependencies, as these libraries tend to use each other's
3545 symbols. So far, we have managed to workaround that by linking
3546 a copy of a good part of our code into each of these
3547 libraries again and again, which, however, makes certain
3548 things hard to do, like sharing static variables. Also, it
3549 substantially increases footprint. With this change, there
3550 is only one library for the basic APIs systemd
3551 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3552 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3553 library as well, however are subject to the --enable-kdbus
3554 switch (see below). Note that "sd-dhcp-client.h" is not part
3555 of this library (this is because it only consumes, never
3556 provides, services of/to other APIs). To make the transition
3557 easy from the separate libraries to the unified one, we
3558 provide the --enable-compat-libs compile-time switch which
3559 will generate stub libraries that are compatible with the
3560 old ones but redirect all calls to the new one.
3562 * All of the kdbus logic and the new APIs "sd-bus.h",
3563 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3564 and "sd-utf8.h" are compile-time optional via the
3565 "--enable-kdbus" switch, and they are not compiled in by
3566 default. To make use of kdbus, you have to explicitly enable
3567 the switch. Note however, that neither the kernel nor the
3568 userspace API for all of this is considered stable yet. We
3569 want to maintain the freedom to still change the APIs for
3570 now. By specifying this build-time switch, you acknowledge
3571 that you are aware of the instability of the current
3574 * Also, note that while kdbus is pretty much complete,
3575 it lacks one thing: proper policy support. This means you
3576 can build a fully working system with all features; however,
3577 it will be highly insecure. Policy support will be added in
3578 one of the next releases, at the same time that we will
3579 declare the APIs stable.
3581 * When the kernel command line argument "kdbus" is specified,
3582 systemd will automatically load the kdbus.ko kernel module. At
3583 this stage of development, it is only useful for testing kdbus
3584 and should not be used in production. Note: if "--enable-kdbus"
3585 is specified, and the kdbus.ko kernel module is available, and
3586 "kdbus" is added to the kernel command line, the entire system
3587 runs with kdbus instead of dbus-daemon, with the above mentioned
3588 problem of missing the system policy enforcement. Also a future
3589 version of kdbus.ko or a newer systemd will not be compatible with
3590 each other, and will unlikely be able to boot the machine if only
3591 one of them is updated.
3593 * systemctl gained a new "import-environment" command which
3594 uploads the caller's environment (or parts thereof) into the
3595 service manager so that it is inherited by services started
3596 by the manager. This is useful to upload variables like
3597 $DISPLAY into the user service manager.
3599 * A new PrivateDevices= switch has been added to service units
3600 which allows running a service with a namespaced /dev
3601 directory that does not contain any device nodes for
3602 physical devices. More specifically, it only includes devices
3603 such as /dev/null, /dev/urandom, and /dev/zero which are API
3606 * logind has been extended to support behaviour like VT
3607 switching on seats that do not support a VT. This makes
3608 multi-session available on seats that are not the first seat
3609 (seat0), and on systems where kernel support for VTs has
3610 been disabled at compile-time.
3612 * If a process holds a delay lock for system sleep or shutdown
3613 and fails to release it in time, we will now log its
3614 identity. This makes it easier to identify processes that
3615 cause slow suspends or power-offs.
3617 * When parsing /etc/crypttab, support for a new key-slot=
3618 option as supported by Debian is added. It allows indicating
3619 which LUKS slot to use on disk, speeding up key loading.
3621 * The sd_journald_sendv() API call has been checked and
3622 officially declared to be async-signal-safe so that it may
3623 be invoked from signal handlers for logging purposes.
3625 * Boot-time status output is now enabled automatically after a
3626 short timeout if boot does not progress, in order to give
3627 the user an indication what she or he is waiting for.
3629 * The boot-time output has been improved to show how much time
3630 remains until jobs expire.
3632 * The KillMode= switch in service units gained a new possible
3633 value "mixed". If set, and the unit is shut down, then the
3634 initial SIGTERM signal is sent only to the main daemon
3635 process, while the following SIGKILL signal is sent to
3636 all remaining processes of the service.
3638 * When a scope unit is registered, a new property "Controller"
3639 may be set. If set to a valid bus name, systemd will send a
3640 RequestStop() signal to this name when it would like to shut
3641 down the scope. This may be used to hook manager logic into
3642 the shutdown logic of scope units. Also, scope units may now
3643 be put in a special "abandoned" state, in which case the
3644 manager process which created them takes no further
3645 responsibilities for it.
3647 * When reading unit files, systemd will now verify
3648 the access mode of these files, and warn about certain
3649 suspicious combinations. This has been added to make it
3650 easier to track down packaging bugs where unit files are
3651 marked executable or world-writable.
3653 * systemd-nspawn gained a new "--setenv=" switch to set
3654 container-wide environment variables. The similar option in
3655 systemd-activate was renamed from "--environment=" to
3656 "--setenv=" for consistency.
3658 * systemd-nspawn has been updated to create a new kdbus domain
3659 for each container that is invoked, thus allowing each
3660 container to have its own set of system and user buses,
3661 independent of the host.
3663 * systemd-nspawn gained a new --drop-capability= switch to run
3664 the container with less capabilities than the default. Both
3665 --drop-capability= and --capability= now take the special
3666 string "all" for dropping or keeping all capabilities.
3668 * systemd-nspawn gained new switches for executing containers
3669 with specific SELinux labels set.
3671 * systemd-nspawn gained a new --quiet switch to not generate
3672 any additional output but the container's own console
3675 * systemd-nspawn gained a new --share-system switch to run a
3676 container without PID namespacing enabled.
3678 * systemd-nspawn gained a new --register= switch to control
3679 whether the container is registered with systemd-machined or
3680 not. This is useful for containers that do not run full
3681 OS images, but only specific apps.
3683 * systemd-nspawn gained a new --keep-unit which may be used
3684 when invoked as the only program from a service unit, and
3685 results in registration of the unit service itself in
3686 systemd-machined, instead of a newly opened scope unit.
3688 * systemd-nspawn gained a new --network-interface= switch for
3689 moving arbitrary interfaces to the container. The new
3690 --network-veth switch creates a virtual Ethernet connection
3691 between host and container. The new --network-bridge=
3692 switch then allows assigning the host side of this virtual
3693 Ethernet connection to a bridge device.
3695 * systemd-nspawn gained a new --personality= switch for
3696 setting the kernel personality for the container. This is
3697 useful when running a 32-bit container on a 64-bit host. A
3698 similar option Personality= is now also available for service
3701 * logind will now also track a "Desktop" identifier for each
3702 session which encodes the desktop environment of it. This is
3703 useful for desktop environments that want to identify
3704 multiple running sessions of itself easily.
3706 * A new SELinuxContext= setting for service units has been
3707 added that allows setting a specific SELinux execution
3708 context for a service.
3710 * Most systemd client tools will now honour $SYSTEMD_LESS for
3711 settings of the "less" pager. By default, these tools will
3712 override $LESS to allow certain operations to work, such as
3713 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3714 influence this logic.
3716 * systemd's "seccomp" hook-up has been changed to make use of
3717 the libseccomp library instead of using its own
3718 implementation. This has benefits for portability among
3721 * For usage together with SystemCallFilter=, a new
3722 SystemCallErrorNumber= setting has been introduced that
3723 allows configuration of a system error number to be returned
3724 on filtered system calls, instead of immediately killing the
3725 process. Also, SystemCallArchitectures= has been added to
3726 limit access to system calls of a particular architecture
3727 (in order to turn off support for unused secondary
3728 architectures). There is also a global
3729 SystemCallArchitectures= setting in system.conf now to turn
3730 off support for non-native system calls system-wide.
3732 * systemd requires a kernel with a working name_to_handle_at(),
3733 please see the kernel config requirements in the README file.
3735 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3736 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3737 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3738 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3739 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3740 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3741 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3742 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3743 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3744 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3745 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3746 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3747 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3748 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3749 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3750 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3751 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3752 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3753 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3754 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3755 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3756 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3757 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3758 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3760 — Berlin, 2014-02-20
3764 * logind has gained support for facilitating privileged input
3765 and drm device access for unprivileged clients. This work is
3766 useful to allow Wayland display servers (and similar
3767 programs, such as kmscon) to run under the user's ID and
3768 access input and drm devices which are normally
3769 protected. When this is used (and the kernel is new enough)
3770 logind will "mute" IO on the file descriptors passed to
3771 Wayland as long as it is in the background and "unmute" it
3772 if it returns into the foreground. This allows secure
3773 session switching without allowing background sessions to
3774 eavesdrop on input and display data. This also introduces
3775 session switching support if VT support is turned off in the
3776 kernel, and on seats that are not seat0.
3778 * A new kernel command line option luks.options= is understood
3779 now which allows specifying LUKS options for usage for LUKS
3780 encrypted partitions specified with luks.uuid=.
3782 * tmpfiles.d(5) snippets may now use specifier expansion in
3783 path names. More specifically %m, %b, %H, %v, are now
3784 replaced by the local machine id, boot id, hostname, and
3785 kernel version number.
3787 * A new tmpfiles.d(5) command "m" has been introduced which
3788 may be used to change the owner/group/access mode of a file
3789 or directory if it exists, but do nothing if it does not.
3791 * This release removes high-level support for the
3792 MemorySoftLimit= cgroup setting. The underlying kernel
3793 cgroup attribute memory.soft_limit= is currently badly
3794 designed and likely to be removed from the kernel API in its
3795 current form, hence we should not expose it for now.
3797 * The memory.use_hierarchy cgroup attribute is now enabled for
3798 all cgroups systemd creates in the memory cgroup
3799 hierarchy. This option is likely to be come the built-in
3800 default in the kernel anyway, and the non-hierarchical mode
3801 never made much sense in the intrinsically hierarchical
3804 * A new field _SYSTEMD_SLICE= is logged along with all journal
3805 messages containing the slice a message was generated
3806 from. This is useful to allow easy per-customer filtering of
3807 logs among other things.
3809 * systemd-journald will no longer adjust the group of journal
3810 files it creates to the "systemd-journal" group. Instead we
3811 rely on the journal directory to be owned by the
3812 "systemd-journal" group, and its setgid bit set, so that the
3813 kernel file system layer will automatically enforce that
3814 journal files inherit this group assignment. The reason for
3815 this change is that we cannot allow NSS look-ups from
3816 journald which would be necessary to resolve
3817 "systemd-journal" to a numeric GID, because this might
3818 create deadlocks if NSS involves synchronous queries to
3819 other daemons (such as nscd, or sssd) which in turn are
3820 logging clients of journald and might block on it, which
3821 would then dead lock. A tmpfiles.d(5) snippet included in
3822 systemd will make sure the setgid bit and group are
3823 properly set on the journal directory if it exists on every
3824 boot. However, we recommend adjusting it manually after
3825 upgrades too (or from RPM scriptlets), so that the change is
3826 not delayed until next reboot.
3828 * Backlight and random seed files in /var/lib/ have moved into
3829 the /var/lib/systemd/ directory, in order to centralize all
3830 systemd generated files in one directory.
3832 * Boot time performance measurements (as displayed by
3833 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3834 performance information if that's available to determine how
3835 much time BIOS and boot loader initialization required. With
3836 a sufficiently new BIOS you hence no longer need to boot
3837 with Gummiboot to get access to such information.
3839 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3840 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3841 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3842 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3843 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3844 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3845 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3847 — Berlin, 2013-10-02
3851 * The Restart= option for services now understands a new
3852 on-watchdog setting, which will restart the service
3853 automatically if the service stops sending out watchdog keep
3854 alive messages (as configured with WatchdogSec=).
3856 * The getty generator (which is responsible for bringing up a
3857 getty on configured serial consoles) will no longer only
3858 start a getty on the primary kernel console but on all
3859 others, too. This makes the order in which console= is
3860 specified on the kernel command line less important.
3862 * libsystemd-logind gained a new sd_session_get_vt() call to
3863 retrieve the VT number of a session.
3865 * If the option "tries=0" is set for an entry of /etc/crypttab
3866 its passphrase is queried indefinitely instead of any
3867 maximum number of tries.
3869 * If a service with a configure PID file terminates its PID
3870 file will now be removed automatically if it still exists
3871 afterwards. This should put an end to stale PID files.
3873 * systemd-run will now also take relative binary path names
3874 for execution and no longer insists on absolute paths.
3876 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3877 paths that are optionally prefixed with "-" to indicate that
3878 it should not be considered a failure if they do not exist.
3880 * journalctl -o (and similar commands) now understands a new
3881 output mode "short-precise", it is similar to "short" but
3882 shows timestamps with usec accuracy.
3884 * The option "discard" (as known from Debian) is now
3885 synonymous to "allow-discards" in /etc/crypttab. In fact,
3886 "discard" is preferred now (since it is easier to remember
3889 * Some licensing clean-ups were made, so that more code is now
3890 LGPL-2.1 licensed than before.
3892 * A minimal tool to save/restore the display backlight
3893 brightness across reboots has been added. It will store the
3894 backlight setting as late as possible at shutdown, and
3895 restore it as early as possible during reboot.
3897 * A logic to automatically discover and enable home and swap
3898 partitions on GPT disks has been added. With this in place
3899 /etc/fstab becomes optional for many setups as systemd can
3900 discover certain partitions located on the root disk
3901 automatically. Home partitions are recognized under their
3902 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3903 partitions are recognized under their GPT type ID
3904 0657fd6da4ab43c484e50933c84b4f4f.
3906 * systemd will no longer pass any environment from the kernel
3907 or initrd to system services. If you want to set an
3908 environment for all services, do so via the kernel command
3909 line systemd.setenv= assignment.
3911 * The systemd-sysctl tool no longer natively reads the file
3912 /etc/sysctl.conf. If desired, the file should be symlinked
3913 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3914 legacy support by a symlink rather than built-in code, it
3915 also makes the otherwise hidden order of application of the
3916 different files visible. (Note that this partly reverts to a
3917 pre-198 application order of sysctl knobs!)
3919 * The "systemctl set-log-level" and "systemctl dump" commands
3920 have been moved to systemd-analyze.
3922 * systemd-run learned the new --remain-after-exit switch,
3923 which causes the scope unit not to be cleaned up
3924 automatically after the process terminated.
3926 * tmpfiles learned a new --exclude-prefix= switch to exclude
3927 certain paths from operation.
3929 * journald will now automatically flush all messages to disk
3930 as soon as a message at the log level CRIT, ALERT or EMERG
3933 Contributions from: Andrew Cook, Brandon Philips, Christian
3934 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3935 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3936 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3937 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3938 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3939 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3940 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3941 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3942 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3943 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3944 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3945 William Giokas, Zbigniew Jędrzejewski-Szmek
3947 — Berlin, 2013-09-13
3951 * The documentation has been updated to cover the various new
3952 concepts introduced with 205.
3954 * Unit files now understand the new %v specifier which
3955 resolves to the kernel version string as returned by "uname
3958 * systemctl now supports filtering the unit list output by
3959 load state, active state and sub state, using the new
3962 * "systemctl status" will now show the results of the
3963 condition checks (like ConditionPathExists= and similar) of
3964 the last start attempts of the unit. They are also logged to
3967 * "journalctl -b" may now be used to look for boot output of a
3968 specific boot. Try "journalctl -b -1" for the previous boot,
3969 but the syntax is substantially more powerful.
3971 * "journalctl --show-cursor" has been added which prints the
3972 cursor string the last shown log line. This may then be used
3973 with the new "journalctl --after-cursor=" switch to continue
3974 browsing logs from that point on.
3976 * "journalctl --force" may now be used to force regeneration
3979 * Creation of "dead" device nodes has been moved from udev
3980 into kmod and tmpfiles. Previously, udev would read the kmod
3981 databases to pre-generate dead device nodes based on meta
3982 information contained in kernel modules, so that these would
3983 be auto-loaded on access rather then at boot. As this
3984 does not really have much to do with the exposing actual
3985 kernel devices to userspace this has always been slightly
3986 alien in the udev codebase. Following the new scheme kmod
3987 will now generate a runtime snippet for tmpfiles from the
3988 module meta information and it now is tmpfiles' job to the
3989 create the nodes. This also allows overriding access and
3990 other parameters for the nodes using the usual tmpfiles
3991 facilities. As side effect this allows us to remove the
3992 CAP_SYS_MKNOD capability bit from udevd entirely.
3994 * logind's device ACLs may now be applied to these "dead"
3995 devices nodes too, thus finally allowing managed access to
3996 devices such as /dev/snd/sequencer whithout loading the
3997 backing module right-away.
3999 * A new RPM macro has been added that may be used to apply
4000 tmpfiles configuration during package installation.
4002 * systemd-detect-virt and ConditionVirtualization= now can
4003 detect User-Mode-Linux machines (UML).
4005 * journald will now implicitly log the effective capabilities
4006 set of processes in the message metadata.
4008 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4010 * The initrd interface has been simplified (more specifically,
4011 support for passing performance data via environment
4012 variables and fsck results via files in /run has been
4013 removed). These features were non-essential, and are
4014 nowadays available in a much nicer way by having systemd in
4015 the initrd serialize its state and have the hosts systemd
4016 deserialize it again.
4018 * The udev "keymap" data files and tools to apply keyboard
4019 specific mappings of scan to key codes, and force-release
4020 scan code lists have been entirely replaced by a udev
4021 "keyboard" builtin and a hwdb data file.
4023 * systemd will now honour the kernel's "quiet" command line
4024 argument also during late shutdown, resulting in a
4025 completely silent shutdown when used.
4027 * There's now an option to control the SO_REUSEPORT socket
4028 option in .socket units.
4030 * Instance units will now automatically get a per-template
4031 subslice of system.slice unless something else is explicitly
4032 configured. For example, instances of sshd@.service will now
4033 implicitly be placed in system-sshd.slice rather than
4034 system.slice as before.
4036 * Test coverage support may now be enabled at build time.
4038 Contributions from: Dave Reisner, Frederic Crozat, Harald
4039 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4040 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4041 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4042 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4043 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4044 Giokas, Zbigniew Jędrzejewski-Szmek
4046 — Berlin, 2013-07-23
4050 * Two new unit types have been introduced:
4052 Scope units are very similar to service units, however, are
4053 created out of pre-existing processes — instead of PID 1
4054 forking off the processes. By using scope units it is
4055 possible for system services and applications to group their
4056 own child processes (worker processes) in a powerful way
4057 which then maybe used to organize them, or kill them
4058 together, or apply resource limits on them.
4060 Slice units may be used to partition system resources in an
4061 hierarchical fashion and then assign other units to them. By
4062 default there are now three slices: system.slice (for all
4063 system services), user.slice (for all user sessions),
4064 machine.slice (for VMs and containers).
4066 Slices and scopes have been introduced primarily in
4067 context of the work to move cgroup handling to a
4068 single-writer scheme, where only PID 1
4069 creates/removes/manages cgroups.
4071 * There's a new concept of "transient" units. In contrast to
4072 normal units these units are created via an API at runtime,
4073 not from configuration from disk. More specifically this
4074 means it is now possible to run arbitrary programs as
4075 independent services, with all execution parameters passed
4076 in via bus APIs rather than read from disk. Transient units
4077 make systemd substantially more dynamic then it ever was,
4078 and useful as a general batch manager.
4080 * logind has been updated to make use of scope and slice units
4081 for managing user sessions. As a user logs in he will get
4082 his own private slice unit, to which all sessions are added
4083 as scope units. We also added support for automatically
4084 adding an instance of user@.service for the user into the
4085 slice. Effectively logind will no longer create cgroup
4086 hierarchies on its own now, it will defer entirely to PID 1
4087 for this by means of scope, service and slice units. Since
4088 user sessions this way become entities managed by PID 1
4089 the output of "systemctl" is now a lot more comprehensive.
4091 * A new mini-daemon "systemd-machined" has been added which
4092 may be used by virtualization managers to register local
4093 VMs/containers. nspawn has been updated accordingly, and
4094 libvirt will be updated shortly. machined will collect a bit
4095 of meta information about the VMs/containers, and assign
4096 them their own scope unit (see above). The collected
4097 meta-data is then made available via the "machinectl" tool,
4098 and exposed in "ps" and similar tools. machined/machinectl
4099 is compile-time optional.
4101 * As discussed earlier, the low-level cgroup configuration
4102 options ControlGroup=, ControlGroupModify=,
4103 ControlGroupPersistent=, ControlGroupAttribute= have been
4104 removed. Please use high-level attribute settings instead as
4105 well as slice units.
4107 * A new bus call SetUnitProperties() has been added to alter
4108 various runtime parameters of a unit. This is primarily
4109 useful to alter cgroup parameters dynamically in a nice way,
4110 but will be extended later on to make more properties
4111 modifiable at runtime. systemctl gained a new set-properties
4112 command that wraps this call.
4114 * A new tool "systemd-run" has been added which can be used to
4115 run arbitrary command lines as transient services or scopes,
4116 while configuring a number of settings via the command
4117 line. This tool is currently very basic, however already
4118 very useful. We plan to extend this tool to even allow
4119 queuing of execution jobs with time triggers from the
4120 command line, similar in fashion to "at".
4122 * nspawn will now inform the user explicitly that kernels with
4123 audit enabled break containers, and suggest the user to turn
4126 * Support for detecting the IMA and AppArmor security
4127 frameworks with ConditionSecurity= has been added.
4129 * journalctl gained a new "-k" switch for showing only kernel
4130 messages, mimicking dmesg output; in addition to "--user"
4131 and "--system" switches for showing only user's own logs
4134 * systemd-delta can now show information about drop-in
4135 snippets extending unit files.
4137 * libsystemd-bus has been substantially updated but is still
4138 not available as public API.
4140 * systemd will now look for the "debug" argument on the kernel
4141 command line and enable debug logging, similar to what
4142 "systemd.log_level=debug" already did before.
4144 * "systemctl set-default", "systemctl get-default" has been
4145 added to configure the default.target symlink, which
4146 controls what to boot into by default.
4148 * "systemctl set-log-level" has been added as a convenient
4149 way to raise and lower systemd logging threshold.
4151 * "systemd-analyze plot" will now show the time the various
4152 generators needed for execution, as well as information
4153 about the unit file loading.
4155 * libsystemd-journal gained a new sd_journal_open_files() call
4156 for opening specific journal files. journactl also gained a
4157 new switch to expose this new functionality. Previously we
4158 only supported opening all files from a directory, or all
4159 files from the system, as opening individual files only is
4160 racy due to journal file rotation.
4162 * systemd gained the new DefaultEnvironment= setting in
4163 /etc/systemd/system.conf to set environment variables for
4166 * If a privileged process logs a journal message with the
4167 OBJECT_PID= field set, then journald will automatically
4168 augment this with additional OBJECT_UID=, OBJECT_GID=,
4169 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4170 system services want to log events about specific client
4171 processes. journactl/systemctl has been updated to make use
4172 of this information if all log messages regarding a specific
4175 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4176 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4177 Reisner, David Coppa, David King, David Strauss, Eelco
4178 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4179 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4180 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4181 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4182 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4183 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4184 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4185 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4186 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4187 Łukasz Stelmach, 장동준
4191 * The Python bindings gained some minimal support for the APIs
4192 exposed by libsystemd-logind.
4194 * ConditionSecurity= gained support for detecting SMACK. Since
4195 this condition already supports SELinux and AppArmor we only
4196 miss IMA for this. Patches welcome!
4198 Contributions from: Karol Lewandowski, Lennart Poettering,
4199 Zbigniew Jędrzejewski-Szmek
4203 * systemd-nspawn will now create /etc/resolv.conf if
4204 necessary, before bind-mounting the host's file onto it.
4206 * systemd-nspawn will now store meta information about a
4207 container on the container's cgroup as extended attribute
4208 fields, including the root directory.
4210 * The cgroup hierarchy has been reworked in many ways. All
4211 objects any of the components systemd creates in the cgroup
4212 tree are now suffixed. More specifically, user sessions are
4213 now placed in cgroups suffixed with ".session", users in
4214 cgroups suffixed with ".user", and nspawn containers in
4215 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4216 names are now escaped in a simple scheme to avoid collision
4217 of userspace object names with kernel filenames. This work
4218 is preparation for making these objects relocatable in the
4219 cgroup tree, in order to allow easy resource partitioning of
4220 these objects without causing naming conflicts.
4222 * systemctl list-dependencies gained the new switches
4223 --plain, --reverse, --after and --before.
4225 * systemd-inhibit now shows the process name of processes that
4226 have taken an inhibitor lock.
4228 * nss-myhostname will now also resolve "localhost"
4229 implicitly. This makes /etc/hosts an optional file and
4230 nicely handles that on IPv6 ::1 maps to both "localhost" and
4233 * libsystemd-logind.so gained a new call
4234 sd_get_machine_names() to enumerate running containers and
4235 VMs (currently only supported by very new libvirt and
4236 nspawn). sd_login_monitor can now be used to watch
4237 VMs/containers coming and going.
4239 * .include is not allowed recursively anymore, and only in
4240 unit files. Usually it is better to use drop-in snippets in
4241 .d/*.conf anyway, as introduced with systemd 198.
4243 * systemd-analyze gained a new "critical-chain" command that
4244 determines the slowest chain of units run during system
4245 boot-up. It is very useful for tracking down where
4246 optimizing boot time is the most beneficial.
4248 * systemd will no longer allow manipulating service paths in
4249 the name=systemd:/system cgroup tree using ControlGroup= in
4250 units. (But is still fine with it in all other dirs.)
4252 * There's a new systemd-nspawn@.service service file that may
4253 be used to easily run nspawn containers as system
4254 services. With the container's root directory in
4255 /var/lib/container/foobar it is now sufficient to run
4256 "systemctl start systemd-nspawn@foobar.service" to boot it.
4258 * systemd-cgls gained a new parameter "--machine" to list only
4259 the processes within a certain container.
4261 * ConditionSecurity= now can check for "apparmor". We still
4262 are lacking checks for SMACK and IMA for this condition
4263 check though. Patches welcome!
4265 * A new configuration file /etc/systemd/sleep.conf has been
4266 added that may be used to configure which kernel operation
4267 systemd is supposed to execute when "suspend", "hibernate"
4268 or "hybrid-sleep" is requested. This makes the new kernel
4269 "freeze" state accessible to the user.
4271 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4272 the passed argument if applicable.
4274 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4275 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4276 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4277 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4278 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4279 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4280 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4285 * The output of 'systemctl list-jobs' got some polishing. The
4286 '--type=' argument may now be passed more than once. A new
4287 command 'systemctl list-sockets' has been added which shows
4288 a list of kernel sockets systemd is listening on with the
4289 socket units they belong to, plus the units these socket
4292 * The experimental libsystemd-bus library got substantial
4293 updates to work in conjunction with the (also experimental)
4294 kdbus kernel project. It works well enough to exchange
4295 messages with some sophistication. Note that kdbus is not
4296 ready yet, and the library is mostly an elaborate test case
4297 for now, and not installable.
4299 * systemd gained a new unit 'systemd-static-nodes.service'
4300 that generates static device nodes earlier during boot, and
4301 can run in conjunction with udev.
4303 * libsystemd-login gained a new call sd_pid_get_user_unit()
4304 to retrieve the user systemd unit a process is running
4305 in. This is useful for systems where systemd is used as
4308 * systemd-nspawn now places all containers in the new /machine
4309 top-level cgroup directory in the name=systemd
4310 hierarchy. libvirt will soon do the same, so that we get a
4311 uniform separation of /system, /user and /machine for system
4312 services, user processes and containers/virtual
4313 machines. This new cgroup hierarchy is also useful to stick
4314 stable names to specific container instances, which can be
4315 recognized later this way (this name may be controlled
4316 via systemd-nspawn's new -M switch). libsystemd-login also
4317 gained a new call sd_pid_get_machine_name() to retrieve the
4318 name of the container/VM a specific process belongs to.
4320 * bootchart can now store its data in the journal.
4322 * libsystemd-journal gained a new call
4323 sd_journal_add_conjunction() for AND expressions to the
4324 matching logic. This can be used to express more complex
4325 logical expressions.
4327 * journactl can now take multiple --unit= and --user-unit=
4330 * The cryptsetup logic now understands the "luks.key=" kernel
4331 command line switch for specifying a file to read the
4332 decryption key from. Also, if a configured key file is not
4333 found the tool will now automatically fall back to prompting
4336 * Python systemd.journal module was updated to wrap recently
4337 added functions from libsystemd-journal. The interface was
4338 changed to bring the low level interface in s.j._Reader
4339 closer to the C API, and the high level interface in
4340 s.j.Reader was updated to wrap and convert all data about
4343 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4344 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4345 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4346 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4347 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4348 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4352 * journalctl --update-catalog now understands a new --root=
4353 option to operate on catalogs found in a different root
4356 * During shutdown after systemd has terminated all running
4357 services a final killing loop kills all remaining left-over
4358 processes. We will now print the name of these processes
4359 when we send SIGKILL to them, since this usually indicates a
4362 * If /etc/crypttab refers to password files stored on
4363 configured mount points automatic dependencies will now be
4364 generated to ensure the specific mount is established first
4365 before the key file is attempted to be read.
4367 * 'systemctl status' will now show information about the
4368 network sockets a socket unit is listening on.
4370 * 'systemctl status' will also shown information about any
4371 drop-in configuration file for units. (Drop-In configuration
4372 files in this context are files such as
4373 /etc/systemd/systemd/foobar.service.d/*.conf)
4375 * systemd-cgtop now optionally shows summed up CPU times of
4376 cgroups. Press '%' while running cgtop to switch between
4377 percentage and absolute mode. This is useful to determine
4378 which cgroups use up the most CPU time over the entire
4379 runtime of the system. systemd-cgtop has also been updated
4380 to be 'pipeable' for processing with further shell tools.
4382 * 'hostnamectl set-hostname' will now allow setting of FQDN
4385 * The formatting and parsing of time span values has been
4386 changed. The parser now understands fractional expressions
4387 such as "5.5h". The formatter will now output fractional
4388 expressions for all time spans under 1min, i.e. "5.123456s"
4389 rather than "5s 123ms 456us". For time spans under 1s
4390 millisecond values are shown, for those under 1ms
4391 microsecond values are shown. This should greatly improve
4392 all time-related output of systemd.
4394 * libsystemd-login and libsystemd-journal gained new
4395 functions for querying the poll() events mask and poll()
4396 timeout value for integration into arbitrary event
4399 * localectl gained the ability to list available X11 keymaps
4400 (models, layouts, variants, options).
4402 * 'systemd-analyze dot' gained the ability to filter for
4403 specific units via shell-style globs, to create smaller,
4404 more useful graphs. I.e. it is now possible to create simple
4405 graphs of all the dependencies between only target units, or
4406 of all units that Avahi has dependencies with.
4408 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4409 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4410 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4411 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4412 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4413 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4414 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4418 * The boot-time readahead implementation for rotating media
4419 will now read the read-ahead data in multiple passes which
4420 consist of all read requests made in equidistant time
4421 intervals. This means instead of strictly reading read-ahead
4422 data in its physical order on disk we now try to find a
4423 middle ground between physical and access time order.
4425 * /etc/os-release files gained a new BUILD_ID= field for usage
4426 on operating systems that provide continuous builds of OS
4429 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4430 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4431 William Douglas, Zbigniew Jędrzejewski-Szmek
4435 * systemd-python gained an API exposing libsystemd-daemon.
4437 * The SMACK setup logic gained support for uploading CIPSO
4440 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4441 ReadOnlyDirectories= and InaccessibleDirectories= has
4442 changed. The private /tmp and /var/tmp directories are now
4443 shared by all processes of a service (which means
4444 ExecStartPre= may now leave data in /tmp that ExecStart= of
4445 the same service can still access). When a service is
4446 stopped its temporary directories are immediately deleted
4447 (normal clean-up with tmpfiles is still done in addition to
4450 * By default, systemd will now set a couple of sysctl
4451 variables in the kernel: the safe sysrq options are turned
4452 on, IP route verification is turned on, and source routing
4453 disabled. The recently added hardlink and softlink
4454 protection of the kernel is turned on. These settings should
4455 be reasonably safe, and good defaults for all new systems.
4457 * The predictable network naming logic may now be turned off
4458 with a new kernel command line switch: net.ifnames=0.
4460 * A new libsystemd-bus module has been added that implements a
4461 pretty complete D-Bus client library. For details see:
4463 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4465 * journald will now explicitly flush the journal files to disk
4466 at the latest 5min after each write. The file will then also
4467 be marked offline until the next write. This should increase
4468 reliability in case of a crash. The synchronization delay
4469 can be configured via SyncIntervalSec= in journald.conf.
4471 * There's a new remote-fs-setup.target unit that can be used
4472 to pull in specific services when at least one remote file
4473 system is to be mounted.
4475 * There are new targets timers.target and paths.target as
4476 canonical targets to pull user timer and path units in
4477 from. This complements sockets.target with a similar
4478 purpose for socket units.
4480 * libudev gained a new call udev_device_set_attribute_value()
4481 to set sysfs attributes of a device.
4483 * The udev daemon now sets the default number of worker
4484 processes executed in parallel based on the number of available
4485 CPUs instead of the amount of available RAM. This is supposed
4486 to provide a more reliable default and limit a too aggressive
4487 paralellism for setups with 1000s of devices connected.
4489 Contributions from: Auke Kok, Colin Walters, Cristian
4490 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4491 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4492 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4493 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4494 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4495 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4496 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4497 Zbigniew Jędrzejewski-Szmek
4501 * Configuration of unit files may now be extended via drop-in
4502 files without having to edit/override the unit files
4503 themselves. More specifically, if the administrator wants to
4504 change one value for a service file foobar.service he can
4505 now do so by dropping in a configuration snippet into
4506 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4507 will load all these snippets and apply them on top of the
4508 main unit configuration file, possibly extending or
4509 overriding its settings. Using these drop-in snippets is
4510 generally nicer than the two earlier options for changing
4511 unit files locally: copying the files from
4512 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4513 them there; or creating a new file in /etc/systemd/system/
4514 that incorporates the original one via ".include". Drop-in
4515 snippets into these .d/ directories can be placed in any
4516 directory systemd looks for units in, and the usual
4517 overriding semantics between /usr/lib, /etc and /run apply
4520 * Most unit file settings which take lists of items can now be
4521 reset by assigning the empty string to them. For example,
4522 normally, settings such as Environment=FOO=BAR append a new
4523 environment variable assignment to the environment block,
4524 each time they are used. By assigning Environment= the empty
4525 string the environment block can be reset to empty. This is
4526 particularly useful with the .d/*.conf drop-in snippets
4527 mentioned above, since this adds the ability to reset list
4528 settings from vendor unit files via these drop-ins.
4530 * systemctl gained a new "list-dependencies" command for
4531 listing the dependencies of a unit recursively.
4533 * Inhibitors are now honored and listed by "systemctl
4534 suspend", "systemctl poweroff" (and similar) too, not only
4535 GNOME. These commands will also list active sessions by
4538 * Resource limits (as exposed by the various control group
4539 controllers) can now be controlled dynamically at runtime
4540 for all units. More specifically, you can now use a command
4541 like "systemctl set-cgroup-attr foobar.service cpu.shares
4542 2000" to alter the CPU shares a specific service gets. These
4543 settings are stored persistently on disk, and thus allow the
4544 administrator to easily adjust the resource usage of
4545 services with a few simple commands. This dynamic resource
4546 management logic is also available to other programs via the
4547 bus. Almost any kernel cgroup attribute and controller is
4550 * systemd-vconsole-setup will now copy all font settings to
4551 all allocated VTs, where it previously applied them only to
4554 * libsystemd-login gained the new sd_session_get_tty() API
4557 * This release drops support for a few legacy or
4558 distribution-specific LSB facility names when parsing init
4559 scripts: $x-display-manager, $mail-transfer-agent,
4560 $mail-transport-agent, $mail-transfer-agent, $smtp,
4561 $null. Also, the mail-transfer-agent.target unit backing
4562 this has been removed. Distributions which want to retain
4563 compatibility with this should carry the burden for
4564 supporting this themselves and patch support for these back
4565 in, if they really need to. Also, the facilities $syslog and
4566 $local_fs are now ignored, since systemd does not support
4567 early-boot LSB init scripts anymore, and these facilities
4568 are implied anyway for normal services. syslog.target has
4571 * There are new bus calls on PID1's Manager object for
4572 cancelling jobs, and removing snapshot units. Previously,
4573 both calls were only available on the Job and Snapshot
4576 * systemd-journal-gatewayd gained SSL support.
4578 * The various "environment" files, such as /etc/locale.conf
4579 now support continuation lines with a backslash ("\") as
4580 last character in the line, similarly in style (but different)
4581 to how this is supported in shells.
4583 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4584 now implicitly appended to every log entry logged. systemctl
4585 has been updated to filter by this field when operating on a
4586 user systemd instance.
4588 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4589 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4590 the container. This makes it easier to boot unmodified
4591 Fedora systems in a container, which however still requires
4592 audit=0 to be passed on the kernel command line. Auditing in
4593 kernel and userspace is unfortunately still too broken in
4594 context of containers, hence we recommend compiling it out
4595 of the kernel or using audit=0. Hopefully this will be fixed
4596 one day for good in the kernel.
4598 * nspawn gained the new --bind= and --bind-ro= parameters to
4599 bind mount specific directories from the host into the
4602 * nspawn will now mount its own devpts file system instance
4603 into the container, in order not to leak pty devices from
4604 the host into the container.
4606 * systemd will now read the firmware boot time performance
4607 information from the EFI variables, if the used boot loader
4608 supports this, and takes it into account for boot performance
4609 analysis via "systemd-analyze". This is currently supported
4610 only in conjunction with Gummiboot, but could be supported
4611 by other boot loaders too. For details see:
4613 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4615 * A new generator has been added that automatically mounts the
4616 EFI System Partition (ESP) to /boot, if that directory
4617 exists, is empty, and no other file system has been
4618 configured to be mounted there.
4620 * logind will now send out PrepareForSleep(false) out
4621 unconditionally, after coming back from suspend. This may be
4622 used by applications as asynchronous notification for
4623 system resume events.
4625 * "systemctl unlock-sessions" has been added, that allows
4626 unlocking the screens of all user sessions at once, similar
4627 to how "systemctl lock-sessions" already locked all users
4628 sessions. This is backed by a new D-Bus call UnlockSessions().
4630 * "loginctl seat-status" will now show the master device of a
4631 seat. (i.e. the device of a seat that needs to be around for
4632 the seat to be considered available, usually the graphics
4635 * tmpfiles gained a new "X" line type, that allows
4636 configuration of files and directories (with wildcards) that
4637 shall be excluded from automatic cleanup ("aging").
4639 * udev default rules set the device node permissions now only
4640 at "add" events, and do not change them any longer with a
4641 later "change" event.
4643 * The log messages for lid events and power/sleep keypresses
4644 now carry a message ID.
4646 * We now have a substantially larger unit test suite, but this
4647 continues to be work in progress.
4649 * udevadm hwdb gained a new --root= parameter to change the
4650 root directory to operate relative to.
4652 * logind will now issue a background sync() request to the kernel
4653 early at shutdown, so that dirty buffers are flushed to disk early
4654 instead of at the last moment, in order to optimize shutdown
4657 * A new bootctl tool has been added that is an interface for
4658 certain boot loader operations. This is currently a preview
4659 and is likely to be extended into a small mechanism daemon
4660 like timedated, localed, hostnamed, and can be used by
4661 graphical UIs to enumerate available boot options, and
4662 request boot into firmware operations.
4664 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4665 the rest of the package. It also has been updated to work
4666 correctly in initrds.
4668 * Policykit previously has been runtime optional, and is now
4669 also compile time optional via a configure switch.
4671 * systemd-analyze has been reimplemented in C. Also "systemctl
4672 dot" has moved into systemd-analyze.
4674 * "systemctl status" with no further parameters will now print
4675 the status of all active or failed units.
4677 * Operations such as "systemctl start" can now be executed
4678 with a new mode "--irreversible" which may be used to queue
4679 operations that cannot accidentally be reversed by a later
4680 job queuing. This is by default used to make shutdown
4681 requests more robust.
4683 * The Python API of systemd now gained a new module for
4684 reading journal files.
4686 * A new tool kernel-install has been added that can install
4687 kernel images according to the Boot Loader Specification:
4689 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4691 * Boot time console output has been improved to provide
4692 animated boot time output for hanging jobs.
4694 * A new tool systemd-activate has been added which can be used
4695 to test socket activation with, directly from the command
4696 line. This should make it much easier to test and debug
4697 socket activation in daemons.
4699 * journalctl gained a new "--reverse" (or -r) option to show
4700 journal output in reverse order (i.e. newest line first).
4702 * journalctl gained a new "--pager-end" (or -e) option to jump
4703 to immediately jump to the end of the journal in the
4704 pager. This is only supported in conjunction with "less".
4706 * journalctl gained a new "--user-unit=" option, that works
4707 similarly to "--unit=" but filters for user units rather than
4710 * A number of unit files to ease adoption of systemd in
4711 initrds has been added. This moves some minimal logic from
4712 the various initrd implementations into systemd proper.
4714 * The journal files are now owned by a new group
4715 "systemd-journal", which exists specifically to allow access
4716 to the journal, and nothing else. Previously, we used the
4717 "adm" group for that, which however possibly covers more
4718 than just journal/log file access. This new group is now
4719 already used by systemd-journal-gatewayd to ensure this
4720 daemon gets access to the journal files and as little else
4721 as possible. Note that "make install" will also set FS ACLs
4722 up for /var/log/journal to give "adm" and "wheel" read
4723 access to it, in addition to "systemd-journal" which owns
4724 the journal files. We recommend that packaging scripts also
4725 add read access to "adm" + "wheel" to /var/log/journal, and
4726 all existing/future journal files. To normal users and
4727 administrators little changes, however packagers need to
4728 ensure to create the "systemd-journal" system group at
4729 package installation time.
4731 * The systemd-journal-gatewayd now runs as unprivileged user
4732 systemd-journal-gateway:systemd-journal-gateway. Packaging
4733 scripts need to create these system user/group at
4736 * timedated now exposes a new boolean property CanNTP that
4737 indicates whether a local NTP service is available or not.
4739 * systemd-detect-virt will now also detect xen PVs
4741 * The pstore file system is now mounted by default, if it is
4744 * In addition to the SELinux and IMA policies we will now also
4745 load SMACK policies at early boot.
4747 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4748 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4749 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4750 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4751 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4752 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4753 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4754 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4755 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4756 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4757 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4758 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4759 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4760 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4764 * Timer units now support calendar time events in addition to
4765 monotonic time events. That means you can now trigger a unit
4766 based on a calendar time specification such as "Thu,Fri
4767 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4768 or fifth day of any month of the year 2013, given that it is
4769 a thursday or friday. This brings timer event support
4770 considerably closer to cron's capabilities. For details on
4771 the supported calendar time specification language see
4774 * udev now supports a number of different naming policies for
4775 network interfaces for predictable names, and a combination
4776 of these policies is now the default. Please see this wiki
4777 document for details:
4779 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4781 * Auke Kok's bootchart implementation has been added to the
4782 systemd tree. It is an optional component that can graph the
4783 boot in quite some detail. It is one of the best bootchart
4784 implementations around and minimal in its code and
4787 * nss-myhostname has been integrated into the systemd source
4788 tree. nss-myhostname guarantees that the local hostname
4789 always stays resolvable via NSS. It has been a weak
4790 requirement of systemd-hostnamed since a long time, and
4791 since its code is actually trivial we decided to just
4792 include it in systemd's source tree. It can be turned off
4793 with a configure switch.
4795 * The read-ahead logic is now capable of properly detecting
4796 whether a btrfs file system is on SSD or rotating media, in
4797 order to optimize the read-ahead scheme. Previously, it was
4798 only capable of detecting this on traditional file systems
4801 * In udev, additional device properties are now read from the
4802 IAB in addition to the OUI database. Also, Bluetooth company
4803 identities are attached to the devices as well.
4805 * In service files %U may be used as specifier that is
4806 replaced by the configured user name of the service.
4808 * nspawn may now be invoked without a controlling TTY. This
4809 makes it suitable for invocation as its own service. This
4810 may be used to set up a simple containerized server system
4811 using only core OS tools.
4813 * systemd and nspawn can now accept socket file descriptors
4814 when they are started for socket activation. This enables
4815 implementation of socket activated nspawn
4816 containers. i.e. think about autospawning an entire OS image
4817 when the first SSH or HTTP connection is received. We expect
4818 that similar functionality will also be added to libvirt-lxc
4821 * journalctl will now suppress ANSI color codes when
4822 presenting log data.
4824 * systemctl will no longer show control group information for
4825 a unit if a the control group is empty anyway.
4827 * logind can now automatically suspend/hibernate/shutdown the
4830 * /etc/machine-info and hostnamed now also expose the chassis
4831 type of the system. This can be used to determine whether
4832 the local system is a laptop, desktop, handset or
4833 tablet. This information may either be configured by the
4834 user/vendor or is automatically determined from ACPI and DMI
4835 information if possible.
4837 * A number of PolicyKit actions are now bound together with
4838 "imply" rules. This should simplify creating UIs because
4839 many actions will now authenticate similar ones as well.
4841 * Unit files learnt a new condition ConditionACPower= which
4842 may be used to conditionalize a unit depending on whether an
4843 AC power source is connected or not, of whether the system
4844 is running on battery power.
4846 * systemctl gained a new "is-failed" verb that may be used in
4847 shell scripts and suchlike to check whether a specific unit
4848 is in the "failed" state.
4850 * The EnvironmentFile= setting in unit files now supports file
4851 globbing, and can hence be used to easily read a number of
4852 environment files at once.
4854 * systemd will no longer detect and recognize specific
4855 distributions. All distribution-specific #ifdeffery has been
4856 removed, systemd is now fully generic and
4857 distribution-agnostic. Effectively, not too much is lost as
4858 a lot of the code is still accessible via explicit configure
4859 switches. However, support for some distribution specific
4860 legacy configuration file formats has been dropped. We
4861 recommend distributions to simply adopt the configuration
4862 files everybody else uses now and convert the old
4863 configuration from packaging scripts. Most distributions
4864 already did that. If that's not possible or desirable,
4865 distributions are welcome to forward port the specific
4866 pieces of code locally from the git history.
4868 * When logging a message about a unit systemd will now always
4869 log the unit name in the message meta data.
4871 * localectl will now also discover system locale data that is
4872 not stored in locale archives, but directly unpacked.
4874 * logind will no longer unconditionally use framebuffer
4875 devices as seat masters, i.e. as devices that are required
4876 to be existing before a seat is considered preset. Instead,
4877 it will now look for all devices that are tagged as
4878 "seat-master" in udev. By default, framebuffer devices will
4879 be marked as such, but depending on local systems, other
4880 devices might be marked as well. This may be used to
4881 integrate graphics cards using closed source drivers (such
4882 as NVidia ones) more nicely into logind. Note however, that
4883 we recommend using the open source NVidia drivers instead,
4884 and no udev rules for the closed-source drivers will be
4885 shipped from us upstream.
4887 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4888 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4889 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4890 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4891 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4892 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4893 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4894 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4895 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4896 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4897 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4898 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4903 * udev gained support for loading additional device properties
4904 from an indexed database that is keyed by vendor/product IDs
4905 and similar device identifiers. For the beginning this
4906 "hwdb" is populated with data from the well-known PCI and
4907 USB database, but also includes PNP, ACPI and OID data. In
4908 the longer run this indexed database shall grow into
4909 becoming the one central database for non-essential
4910 userspace device metadata. Previously, data from the PCI/USB
4911 database was only attached to select devices, since the
4912 lookup was a relatively expensive operation due to O(n) time
4913 complexity (with n being the number of entries in the
4914 database). Since this is now O(1), we decided to add in this
4915 data for all devices where this is available, by
4916 default. Note that the indexed database needs to be rebuilt
4917 when new data files are installed. To achieve this you need
4918 to update your packaging scripts to invoke "udevadm hwdb
4919 --update" after installation of hwdb data files. For
4920 RPM-based distributions we introduced the new
4921 %udev_hwdb_update macro for this purpose.
4923 * The Journal gained support for the "Message Catalog", an
4924 indexed database to link up additional information with
4925 journal entries. For further details please check:
4927 http://www.freedesktop.org/wiki/Software/systemd/catalog
4929 The indexed message catalog database also needs to be
4930 rebuilt after installation of message catalog files. Use
4931 "journalctl --update-catalog" for this. For RPM-based
4932 distributions we introduced the %journal_catalog_update
4933 macro for this purpose.
4935 * The Python Journal bindings gained support for the standard
4936 Python logging framework.
4938 * The Journal API gained new functions for checking whether
4939 the underlying file system of a journal file is capable of
4940 properly reporting file change notifications, or whether
4941 applications that want to reflect journal changes "live"
4942 need to recheck journal files continuously in appropriate
4945 * It is now possible to set the "age" field for tmpfiles
4946 entries to 0, indicating that files matching this entry
4947 shall always be removed when the directories are cleaned up.
4949 * coredumpctl gained a new "gdb" verb which invokes gdb
4950 right-away on the selected coredump.
4952 * There's now support for "hybrid sleep" on kernels that
4953 support this, in addition to "suspend" and "hibernate". Use
4954 "systemctl hybrid-sleep" to make use of this.
4956 * logind's HandleSuspendKey= setting (and related settings)
4957 now gained support for a new "lock" setting to simply
4958 request the screen lock on all local sessions, instead of
4959 actually executing a suspend or hibernation.
4961 * systemd will now mount the EFI variables file system by
4964 * Socket units now gained support for configuration of the
4965 SMACK security label.
4967 * timedatectl will now output the time of the last and next
4968 daylight saving change.
4970 * We dropped support for various legacy and distro-specific
4971 concepts, such as insserv, early-boot SysV services
4972 (i.e. those for non-standard runlevels such as 'b' or 'S')
4973 or ArchLinux /etc/rc.conf support. We recommend the
4974 distributions who still need support this to either continue
4975 to maintain the necessary patches downstream, or find a
4976 different solution. (Talk to us if you have questions!)
4978 * Various systemd components will now bypass PolicyKit checks
4979 for root and otherwise handle properly if PolicyKit is not
4980 found to be around. This should fix most issues for
4981 PolicyKit-less systems. Quite frankly this should have been
4982 this way since day one. It is absolutely our intention to
4983 make systemd work fine on PolicyKit-less systems, and we
4984 consider it a bug if something does not work as it should if
4985 PolicyKit is not around.
4987 * For embedded systems it is now possible to build udev and
4988 systemd without blkid and/or kmod support.
4990 * "systemctl switch-root" is now capable of switching root
4991 more than once. I.e. in addition to transitions from the
4992 initrd to the host OS it is now possible to transition to
4993 further OS images from the host. This is useful to implement
4994 offline updating tools.
4996 * Various other additions have been made to the RPM macros
4997 shipped with systemd. Use %udev_rules_update() after
4998 installing new udev rules files. %_udevhwdbdir,
4999 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5000 %_sysctldir are now available which resolve to the right
5001 directories for packages to place various data files in.
5003 * journalctl gained the new --full switch (in addition to
5004 --all, to disable ellipsation for long messages.
5006 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5007 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5008 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5009 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5010 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5011 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5012 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5013 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5014 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5018 * journalctl gained new --since= and --until= switches to
5019 filter by time. It also now supports nice filtering for
5020 units via --unit=/-u.
5022 * Type=oneshot services may use ExecReload= and do the
5025 * The journal daemon now supports time-based rotation and
5026 vacuuming, in addition to the usual disk-space based
5029 * The journal will now index the available field values for
5030 each field name. This enables clients to show pretty drop
5031 downs of available match values when filtering. The bash
5032 completion of journalctl has been updated
5033 accordingly. journalctl gained a new switch -F to list all
5034 values a certain field takes in the journal database.
5036 * More service events are now written as structured messages
5037 to the journal, and made recognizable via message IDs.
5039 * The timedated, localed and hostnamed mini-services which
5040 previously only provided support for changing time, locale
5041 and hostname settings from graphical DEs such as GNOME now
5042 also have a minimal (but very useful) text-based client
5043 utility each. This is probably the nicest way to changing
5044 these settings from the command line now, especially since
5045 it lists available options and is fully integrated with bash
5048 * There's now a new tool "systemd-coredumpctl" to list and
5049 extract coredumps from the journal.
5051 * We now install a README each in /var/log/ and
5052 /etc/rc.d/init.d explaining where the system logs and init
5053 scripts went. This hopefully should help folks who go to
5054 that dirs and look into the otherwise now empty void and
5055 scratch their heads.
5057 * When user-services are invoked (by systemd --user) the
5058 $MANAGERPID env var is set to the PID of systemd.
5060 * SIGRTMIN+24 when sent to a --user instance will now result
5061 in immediate termination of systemd.
5063 * gatewayd received numerous feature additions such as a
5064 "follow" mode, for live syncing and filtering.
5066 * browse.html now allows filtering and showing detailed
5067 information on specific entries. Keyboard navigation and
5068 mouse screen support has been added.
5070 * gatewayd/journalctl now supports HTML5/JSON
5071 Server-Sent-Events as output.
5073 * The SysV init script compatibility logic will now
5074 heuristically determine whether a script supports the
5075 "reload" verb, and only then make this available as
5078 * "systemctl status --follow" has been removed, use "journalctl
5081 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5082 have been removed since they are hardly useful to be
5085 * And I'd like to take the opportunity to specifically mention
5086 Zbigniew for his great contributions. Zbigniew, you rock!
5088 Contributions from: Andrew Eikum, Christian Hesse, Colin
5089 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5090 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5091 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5092 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5093 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5094 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5098 * If /etc/vconsole.conf is non-existent or empty we will no
5099 longer load any console font or key map at boot by
5100 default. Instead the kernel defaults will be left
5101 intact. This is definitely the right thing to do, as no
5102 configuration should mean no configuration, and hard-coding
5103 font names that are different on all archs is probably a bad
5104 idea. Also, the kernel default key map and font should be
5105 good enough for most cases anyway, and mostly identical to
5106 the userspace fonts/key maps we previously overloaded them
5107 with. If distributions want to continue to default to a
5108 non-kernel font or key map they should ship a default
5109 /etc/vconsole.conf with the appropriate contents.
5111 Contributions from: Colin Walters, Daniel J Walsh, Dave
5112 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5113 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5117 * journalctl gained a new --cursor= switch to show entries
5118 starting from the specified location in the journal.
5120 * We now enforce a size limit on journal entry fields exported
5121 with "-o json" in journalctl. Fields larger than 4K will be
5122 assigned null. This can be turned off with --all.
5124 * An (optional) journal gateway daemon is now available as
5125 "systemd-journal-gatewayd.service". This service provides
5126 access to the journal via HTTP and JSON. This functionality
5127 will be used to implement live log synchronization in both
5128 pull and push modes, but has various other users too, such
5129 as easy log access for debugging of embedded devices. Right
5130 now it is already useful to retrieve the journal via HTTP:
5132 # systemctl start systemd-journal-gatewayd.service
5133 # wget http://localhost:19531/entries
5135 This will download the journal contents in a
5136 /var/log/messages compatible format. The same as JSON:
5138 # curl -H"Accept: application/json" http://localhost:19531/entries
5140 This service is also accessible via a web browser where a
5141 single static HTML5 app is served that uses the JSON logic
5142 to enable the user to do some basic browsing of the
5143 journal. This will be extended later on. Here's an example
5144 screenshot of this app in its current state:
5146 http://0pointer.de/public/journal-gatewayd
5148 Contributions from: Kay Sievers, Lennart Poettering, Robert
5149 Milasan, Tom Gundersen
5153 * The bash completion logic is now available for journalctl
5156 * We do not mount the "cpuset" controller anymore together with
5157 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5158 started if no parameters are assigned to it. "cpuset" hence
5159 broke code that assumed it it could create "cpu" groups and
5162 * journalctl -f will now subscribe to terminal size changes,
5163 and line break accordingly.
5165 Contributions from: Dave Reisner, Kay Sievers, Lennart
5166 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5170 * nspawn will now create a symlink /etc/localtime in the
5171 container environment, copying the host's timezone
5172 setting. Previously this has been done via a bind mount, but
5173 since symlinks cannot be bind mounted this has now been
5174 changed to create/update the appropriate symlink.
5176 * journalctl -n's line number argument is now optional, and
5177 will default to 10 if omitted.
5179 * journald will now log the maximum size the journal files may
5180 take up on disk. This is particularly useful if the default
5181 built-in logic of determining this parameter from the file
5182 system size is used. Use "systemctl status
5183 systemd-journald.service" to see this information.
5185 * The multi-seat X wrapper tool has been stripped down. As X
5186 is now capable of enumerating graphics devices via udev in a
5187 seat-aware way the wrapper is not strictly necessary
5188 anymore. A stripped down temporary stop-gap is still shipped
5189 until the upstream display managers have been updated to
5190 fully support the new X logic. Expect this wrapper to be
5191 removed entirely in one of the next releases.
5193 * HandleSleepKey= in logind.conf has been split up into
5194 HandleSuspendKey= and HandleHibernateKey=. The old setting
5195 is not available anymore. X11 and the kernel are
5196 distinguishing between these keys and we should too. This
5197 also means the inhibition lock for these keys has been split
5200 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5201 Poettering, Lukas Nykryn, Václav Pavlín
5205 * Whenever a unit changes state we will now log this to the
5206 journal and show along the unit's own log output in
5209 * ConditionPathIsMountPoint= can now properly detect bind
5210 mount points too. (Previously, a bind mount of one file
5211 system to another place in the same file system could not be
5212 detected as mount, since they shared struct stat's st_dev
5215 * We will now mount the cgroup controllers cpu, cpuacct,
5216 cpuset and the controllers net_cls, net_prio together by
5219 * nspawn containers will now have a virtualized boot
5220 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5221 over with a randomized ID at container initialization). This
5222 has the effect of making "journalctl -b" do the right thing
5225 * The JSON output journal serialization has been updated not
5226 to generate "endless" list objects anymore, but rather one
5227 JSON object per line. This is more in line how most JSON
5228 parsers expect JSON objects. The new output mode
5229 "json-pretty" has been added to provide similar output, but
5230 neatly aligned for readability by humans.
5232 * We dropped all explicit sync() invocations in the shutdown
5233 code. The kernel does this implicitly anyway in the kernel
5234 reboot() syscall. halt(8)'s -n option is now a compatibility
5237 * We now support virtualized reboot() in containers, as
5238 supported by newer kernels. We will fall back to exit() if
5239 CAP_SYS_REBOOT is not available to the container. Also,
5240 nspawn makes use of this now and will actually reboot the
5241 container if the containerized OS asks for that.
5243 * journalctl will only show local log output by default
5244 now. Use --merge (-m) to show remote log output, too.
5246 * libsystemd-journal gained the new sd_journal_get_usage()
5247 call to determine the current disk usage of all journal
5248 files. This is exposed in the new "journalctl --disk-usage"
5251 * journald gained a new configuration setting SplitMode= in
5252 journald.conf which may be used to control how user journals
5253 are split off. See journald.conf(5) for details.
5255 * A new condition type ConditionFileNotEmpty= has been added.
5257 * tmpfiles' "w" lines now support file globbing, to write
5258 multiple files at once.
5260 * We added Python bindings for the journal submission
5261 APIs. More Python APIs for a number of selected APIs will
5262 likely follow. Note that we intend to add native bindings
5263 only for the Python language, as we consider it common
5264 enough to deserve bindings shipped within systemd. There are
5265 various projects outside of systemd that provide bindings
5266 for languages such as PHP or Lua.
5268 * Many conditions will now resolve specifiers such as %i. In
5269 addition, PathChanged= and related directives of .path units
5270 now support specifiers as well.
5272 * There's now a new RPM macro definition for the system preset
5275 * journald will now warn if it ca not forward a message to the
5276 syslog daemon because its socket is full.
5278 * timedated will no longer write or process /etc/timezone,
5279 except on Debian. As we do not support late mounted /usr
5280 anymore /etc/localtime always being a symlink is now safe,
5281 and hence the information in /etc/timezone is not necessary
5284 * logind will now always reserve one VT for a text getty (VT6
5285 by default). Previously if more than 6 X sessions where
5286 started they took up all the VTs with auto-spawned gettys,
5287 so that no text gettys were available anymore.
5289 * udev will now automatically inform the btrfs kernel logic
5290 about btrfs RAID components showing up. This should make
5291 simple hotplug based btrfs RAID assembly work.
5293 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5294 (but not for its children which will stay at the kernel
5295 default). This should allow setups with a lot more listening
5298 * systemd will now always pass the configured timezone to the
5299 kernel at boot. timedated will do the same when the timezone
5302 * logind's inhibition logic has been updated. By default,
5303 logind will now handle the lid switch, the power and sleep
5304 keys all the time, even in graphical sessions. If DEs want
5305 to handle these events on their own they should take the new
5306 handle-power-key, handle-sleep-key and handle-lid-switch
5307 inhibitors during their runtime. A simple way to achieve
5308 that is to invoke the DE wrapped in an invocation of:
5310 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5312 * Access to unit operations is now checked via SELinux taking
5313 the unit file label and client process label into account.
5315 * systemd will now notify the administrator in the journal
5316 when he over-mounts a non-empty directory.
5318 * There are new specifiers that are resolved in unit files,
5319 for the host name (%H), the machine ID (%m) and the boot ID
5322 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5323 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5324 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5325 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5326 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5327 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5328 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5332 * Support for reading structured kernel messages from
5333 /dev/kmsg has now been added and is enabled by default.
5335 * Support for reading kernel messages from /proc/kmsg has now
5336 been removed. If you want kernel messages in the journal
5337 make sure to run a recent kernel (>= 3.5) that supports
5338 reading structured messages from /dev/kmsg (see
5339 above). /proc/kmsg is now exclusive property of classic
5340 syslog daemons again.
5342 * The libudev API gained the new
5343 udev_device_new_from_device_id() call.
5345 * The logic for file system namespace (ReadOnlyDirectory=,
5346 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5347 require pivot_root() anymore. This means fewer temporary
5348 directories are created below /tmp for this feature.
5350 * nspawn containers will now see and receive all submounts
5351 made on the host OS below the root file system of the
5354 * Forward Secure Sealing is now supported for Journal files,
5355 which provide cryptographical sealing of journal files so
5356 that attackers cannot alter log history anymore without this
5357 being detectable. Lennart will soon post a blog story about
5358 this explaining it in more detail.
5360 * There are two new service settings RestartPreventExitStatus=
5361 and SuccessExitStatus= which allow configuration of exit
5362 status (exit code or signal) which will be excepted from the
5363 restart logic, resp. consider successful.
5365 * journalctl gained the new --verify switch that can be used
5366 to check the integrity of the structure of journal files and
5367 (if Forward Secure Sealing is enabled) the contents of
5370 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5371 and similar symlinks pre-created. This makes running shells
5372 as container init process a lot more fun.
5374 * The fstab support can now handle PARTUUID= and PARTLABEL=
5377 * A new ConditionHost= condition has been added to match
5378 against the hostname (with globs) and machine ID. This is
5379 useful for clusters where a single OS image is used to
5380 provision a large number of hosts which shall run slightly
5381 different sets of services.
5383 * Services which hit the restart limit will now be placed in a
5386 Contributions from: Bertram Poettering, Dave Reisner, Huang
5387 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5388 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5392 * When running in --user mode systemd will now become a
5393 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5394 tree a lot more organized.
5396 * A new PartOf= unit dependency type has been introduced that
5397 may be used to group services in a natural way.
5399 * "systemctl enable" may now be used to enable instances of
5402 * journalctl now prints error log levels in red, and
5403 warning/notice log levels in bright white. It also supports
5404 filtering by log level now.
5406 * cgtop gained a new -n switch (similar to top), to configure
5407 the maximum number of iterations to run for. It also gained
5408 -b, to run in batch mode (accepting no input).
5410 * The suffix ".service" may now be omitted on most systemctl
5411 command lines involving service unit names.
5413 * There's a new bus call in logind to lock all sessions, as
5414 well as a loginctl verb for it "lock-sessions".
5416 * libsystemd-logind.so gained a new call sd_journal_perror()
5417 that works similar to libc perror() but logs to the journal
5418 and encodes structured information about the error number.
5420 * /etc/crypttab entries now understand the new keyfile-size=
5423 * shutdown(8) now can send a (configurable) wall message when
5424 a shutdown is cancelled.
5426 * The mount propagation mode for the root file system will now
5427 default to "shared", which is useful to make containers work
5428 nicely out-of-the-box so that they receive new mounts from
5429 the host. This can be undone locally by running "mount
5430 --make-rprivate /" if needed.
5432 * The prefdm.service file has been removed. Distributions
5433 should maintain this unit downstream if they intend to keep
5434 it around. However, we recommend writing normal unit files
5435 for display managers instead.
5437 * Since systemd is a crucial part of the OS we will now
5438 default to a number of compiler switches that improve
5439 security (hardening) such as read-only relocations, stack
5440 protection, and suchlike.
5442 * The TimeoutSec= setting for services is now split into
5443 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5444 of individual time outs for the start and the stop phase of
5447 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5448 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5449 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5450 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5451 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5452 Gundersen, Zbigniew Jędrzejewski-Szmek
5456 * The journal and id128 C APIs are now fully documented as man
5459 * Extra safety checks have been added when transitioning from
5460 the initial RAM disk to the main system to avoid accidental
5463 * /etc/crypttab entries now understand the new keyfile-offset=
5466 * systemctl -t can now be used to filter by unit load state.
5468 * The journal C API gained the new sd_journal_wait() call to
5469 make writing synchronous journal clients easier.
5471 * journalctl gained the new -D switch to show journals from a
5474 * journalctl now displays a special marker between log
5475 messages of two different boots.
5477 * The journal is now explicitly flushed to /var via a service
5478 systemd-journal-flush.service, rather than implicitly simply
5479 by seeing /var/log/journal to be writable.
5481 * journalctl (and the journal C APIs) can now match for much
5482 more complex expressions, with alternatives and
5485 * When transitioning from the initial RAM disk to the main
5486 system we will now kill all processes in a killing spree to
5487 ensure no processes stay around by accident.
5489 * Three new specifiers may be used in unit files: %u, %h, %s
5490 resolve to the user name, user home directory resp. user
5491 shell. This is useful for running systemd user instances.
5493 * We now automatically rotate journal files if their data
5494 object hash table gets a fill level > 75%. We also size the
5495 hash table based on the configured maximum file size. This
5496 together should lower hash collisions drastically and thus
5497 speed things up a bit.
5499 * journalctl gained the new "--header" switch to introspect
5500 header data of journal files.
5502 * A new setting SystemCallFilters= has been added to services
5503 which may be used to apply blacklists or whitelists to
5504 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5506 * nspawn gained a new --link-journal= switch (and quicker: -j)
5507 to link the container journal with the host. This makes it
5508 very easy to centralize log viewing on the host for all
5509 guests while still keeping the journal files separated.
5511 * Many bugfixes and optimizations
5513 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5514 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5515 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5520 * Several tools now understand kernel command line arguments,
5521 which are only read when run in an initial RAM disk. They
5522 usually follow closely their normal counterparts, but are
5525 * There's a new tool to analyze the readahead files that are
5526 automatically generated at boot. Use:
5528 /usr/lib/systemd/systemd-readahead analyze /.readahead
5530 * We now provide an early debug shell on tty9 if this enabled. Use:
5532 systemctl enable debug-shell.service
5534 * All plymouth related units have been moved into the Plymouth
5535 package. Please make sure to upgrade your Plymouth version
5538 * systemd-tmpfiles now supports getting passed the basename of
5539 a configuration file only, in which case it will look for it
5540 in all appropriate directories automatically.
5542 * udevadm info now takes a /dev or /sys path as argument, and
5543 does the right thing. Example:
5545 udevadm info /dev/sda
5546 udevadm info /sys/class/block/sda
5548 * systemctl now prints a warning if a unit is stopped but a
5549 unit that might trigger it continues to run. Example: a
5550 service is stopped but the socket that activates it is left
5553 * "systemctl status" will now mention if the log output was
5554 shortened due to rotation since a service has been started.
5556 * The journal API now exposes functions to determine the
5557 "cutoff" times due to rotation.
5559 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5560 immediately flushing of runtime logs to /var if possible,
5561 resp. for triggering immediate rotation of the journal
5564 * It is now considered an error if a service is attempted to
5565 be stopped that is not loaded.
5567 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5569 * systemd-analyze now supports Python 3
5571 * tmpfiles now supports cleaning up directories via aging
5572 where the first level dirs are always kept around but
5573 directories beneath it automatically aged. This is enabled
5574 by prefixing the age field with '~'.
5576 * Seat objects now expose CanGraphical, CanTTY properties
5577 which is required to deal with very fast bootups where the
5578 display manager might be running before the graphics drivers
5579 completed initialization.
5581 * Seat objects now expose a State property.
5583 * We now include RPM macros for service enabling/disabling
5584 based on the preset logic. We recommend RPM based
5585 distributions to make use of these macros if possible. This
5586 makes it simpler to reuse RPM spec files across
5589 * We now make sure that the collected systemd unit name is
5590 always valid when services log to the journal via
5593 * There's a new man page kernel-command-line(7) detailing all
5594 command line options we understand.
5596 * The fstab generator may now be disabled at boot by passing
5597 fstab=0 on the kernel command line.
5599 * A new kernel command line option modules-load= is now understood
5600 to load a specific kernel module statically, early at boot.
5602 * Unit names specified on the systemctl command line are now
5603 automatically escaped as needed. Also, if file system or
5604 device paths are specified they are automatically turned
5605 into the appropriate mount or device unit names. Example:
5607 systemctl status /home
5608 systemctl status /dev/sda
5610 * The SysVConsole= configuration option has been removed from
5611 system.conf parsing.
5613 * The SysV search path is no longer exported on the D-Bus
5616 * The Names= option is been removed from unit file parsing.
5618 * There's a new man page bootup(7) detailing the boot process.
5620 * Every unit and every generator we ship with systemd now
5621 comes with full documentation. The self-explanatory boot is
5624 * A couple of services gained "systemd-" prefixes in their
5625 name if they wrap systemd code, rather than only external
5626 code. Among them fsck@.service which is now
5627 systemd-fsck@.service.
5629 * The HaveWatchdog property has been removed from the D-Bus
5632 * systemd.confirm_spawn= on the kernel command line should now
5635 * There's a new man page crypttab(5) which details all options
5636 we actually understand.
5638 * systemd-nspawn gained a new --capability= switch to pass
5639 additional capabilities to the container.
5641 * timedated will now read known NTP implementation unit names
5642 from /usr/lib/systemd/ntp-units.d/*.list,
5643 systemd-timedated-ntp.target has been removed.
5645 * journalctl gained a new switch "-b" that lists log data of
5646 the current boot only.
5648 * The notify socket is in the abstract namespace again, in
5649 order to support daemons which chroot() at start-up.
5651 * There is a new Storage= configuration option for journald
5652 which allows configuration of where log data should go. This
5653 also provides a way to disable journal logging entirely, so
5654 that data collected is only forwarded to the console, the
5655 kernel log buffer or another syslog implementation.
5657 * Many bugfixes and optimizations
5659 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5660 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5661 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5662 Shawn Landden, Tom Gundersen
5666 * "systemctl help <unit>" now shows the man page if one is
5669 * Several new man pages have been added.
5671 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5672 MaxLevelConsole= can now be specified in
5673 journald.conf. These options allow reducing the amount of
5674 data stored on disk or forwarded by the log level.
5676 * TimerSlackNSec= can now be specified in system.conf for
5677 PID1. This allows system-wide power savings.
5679 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5680 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5685 * logind is now capable of (optionally) handling power and
5686 sleep keys as well as the lid switch.
5688 * journalctl now understands the syntax "journalctl
5689 /usr/bin/avahi-daemon" to get all log output of a specific
5692 * CapabilityBoundingSet= in system.conf now also influences
5693 the capability bound set of usermode helpers of the kernel.
5695 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5696 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5697 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5698 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5702 * Note that we skipped 139 releases here in order to set the
5703 new version to something that is greater than both udev's
5704 and systemd's most recent version number.
5706 * udev: all udev sources are merged into the systemd source tree now.
5707 All future udev development will happen in the systemd tree. It
5708 is still fully supported to use the udev daemon and tools without
5709 systemd running, like in initramfs or other init systems. Building
5710 udev though, will require the *build* of the systemd tree, but
5711 udev can be properly *run* without systemd.
5713 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5714 should be used to create dead device nodes as workarounds for broken
5717 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5718 no longer supported. udev_monitor_new_from_netlink() needs to be
5719 used to subscribe to events.
5721 * udev: when udevd is started by systemd, processes which are left
5722 behind by forking them off of udev rules, are unconditionally cleaned
5723 up and killed now after the event handling has finished. Services or
5724 daemons must be started as systemd services. Services can be
5725 pulled-in by udev to get started, but they can no longer be directly
5726 forked by udev rules.
5728 * udev: the daemon binary is called systemd-udevd now and installed
5729 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5730 to adapt to that, create symlink, or rename the binary after building
5733 * libudev no longer provides these symbols:
5734 udev_monitor_from_socket()
5735 udev_queue_get_failed_list_entry()
5736 udev_get_{dev,sys,run}_path()
5737 The versions number was bumped and symbol versioning introduced.
5739 * systemd-loginctl and systemd-journalctl have been renamed
5740 to loginctl and journalctl to match systemctl.
5742 * The config files: /etc/systemd/systemd-logind.conf and
5743 /etc/systemd/systemd-journald.conf have been renamed to
5744 logind.conf and journald.conf. Package updates should rename
5745 the files to the new names on upgrade.
5747 * For almost all files the license is now LGPL2.1+, changed
5748 from the previous GPL2.0+. Exceptions are some minor stuff
5749 of udev (which will be changed to LGPL2.1 eventually, too),
5750 and the MIT licensed sd-daemon.[ch] library that is suitable
5751 to be used as drop-in files.
5753 * systemd and logind now handle system sleep states, in
5754 particular suspending and hibernating.
5756 * logind now implements a sleep/shutdown/idle inhibiting logic
5757 suitable for a variety of uses. Soonishly Lennart will blog
5758 about this in more detail.
5760 * var-run.mount and var-lock.mount are no longer provided
5761 (which prevously bind mounted these directories to their new
5762 places). Distributions which have not converted these
5763 directories to symlinks should consider stealing these files
5764 from git history and add them downstream.
5766 * We introduced the Documentation= field for units and added
5767 this to all our shipped units. This is useful to make it
5768 easier to explore the boot and the purpose of the various
5771 * All smaller setup units (such as
5772 systemd-vconsole-setup.service) now detect properly if they
5773 are run in a container and are skipped when
5774 appropriate. This guarantees an entirely noise-free boot in
5775 Linux container environments such as systemd-nspawn.
5777 * A framework for implementing offline system updates is now
5778 integrated, for details see:
5779 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5781 * A new service type Type=idle is available now which helps us
5782 avoiding ugly interleaving of getty output and boot status
5785 * There's now a system-wide CapabilityBoundingSet= option to
5786 globally reduce the set of capabilities for the
5787 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5788 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5789 even CAP_NET_ADMIN system-wide for secure systems.
5791 * There are now system-wide DefaultLimitXXX= options to
5792 globally change the defaults of the various resource limits
5793 for all units started by PID 1.
5795 * Harald Hoyer's systemd test suite has been integrated into
5796 systemd which allows easy testing of systemd builds in qemu
5797 and nspawn. (This is really awesome! Ask us for details!)
5799 * The fstab parser is now implemented as generator, not inside
5802 * systemctl will now warn you if .mount units generated from
5803 /etc/fstab are out of date due to changes in fstab that
5804 have not been read by systemd yet.
5806 * systemd is now suitable for usage in initrds. Dracut has
5807 already been updated to make use of this. With this in place
5808 initrds get a slight bit faster but primarily are much
5809 easier to introspect and debug since "systemctl status" in
5810 the host system can be used to introspect initrd services,
5811 and the journal from the initrd is kept around too.
5813 * systemd-delta has been added, a tool to explore differences
5814 between user/admin configuration and vendor defaults.
5816 * PrivateTmp= now affects both /tmp and /var/tmp.
5818 * Boot time status messages are now much prettier and feature
5819 proper english language. Booting up systemd has never been
5822 * Read-ahead pack files now include the inode number of all
5823 files to pre-cache. When the inode changes the pre-caching
5824 is not attempted. This should be nicer to deal with updated
5825 packages which might result in changes of read-ahead
5828 * We now temporaritly lower the kernel's read_ahead_kb variable
5829 when collecting read-ahead data to ensure the kernel's
5830 built-in read-ahead does not add noise to our measurements
5831 of necessary blocks to pre-cache.
5833 * There's now RequiresMountsFor= to add automatic dependencies
5834 for all mounts necessary for a specific file system path.
5836 * MountAuto= and SwapAuto= have been removed from
5837 system.conf. Mounting file systems at boot has to take place
5840 * nspawn now learned a new switch --uuid= to set the machine
5841 ID on the command line.
5843 * nspawn now learned the -b switch to automatically search
5846 * vt102 is now the default TERM for serial TTYs, upgraded from
5849 * systemd-logind now works on VT-less systems.
5851 * The build tree has been reorganized. The individual
5852 components now have directories of their own.
5854 * A new condition type ConditionPathIsReadWrite= is now available.
5856 * nspawn learned the new -C switch to create cgroups for the
5857 container in other hierarchies.
5859 * We now have support for hardware watchdogs, configurable in
5862 * The scheduled shutdown logic now has a public API.
5864 * We now mount /tmp as tmpfs by default, but this can be
5865 masked and /etc/fstab can override it.
5867 * Since udisks does not make use of /media anymore we are not
5868 mounting a tmpfs on it anymore.
5870 * journalctl gained a new --local switch to only interleave
5871 locally generated journal files.
5873 * We can now load the IMA policy at boot automatically.
5875 * The GTK tools have been split off into a systemd-ui.
5877 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5878 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5879 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5880 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5881 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5882 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5883 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5884 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5885 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5890 * This is mostly a bugfix release
5892 * Support optional initialization of the machine ID from the
5893 KVM or container configured UUID.
5895 * Support immediate reboots with "systemctl reboot -ff"
5897 * Show /etc/os-release data in systemd-analyze output
5899 * Many bugfixes for the journal, including endianness fixes and
5900 ensuring that disk space enforcement works
5902 * sd-login.h is C++ comptaible again
5904 * Extend the /etc/os-release format on request of the Debian
5907 * We now refuse non-UTF8 strings used in various configuration
5908 and unit files. This is done to ensure we do not pass invalid
5909 data over D-Bus or expose it elsewhere.
5911 * Register Mimo USB Screens as suitable for automatic seat
5914 * Read SELinux client context from journal clients in a race
5917 * Reorder configuration file lookup order. /etc now always
5918 overrides /run in order to allow the administrator to always
5919 and unconditionally override vendor-supplied or
5920 automatically generated data.
5922 * The various user visible bits of the journal now have man
5923 pages. We still lack man pages for the journal API calls
5926 * We now ship all man pages in HTML format again in the
5929 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5930 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5931 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5936 * This is mostly a bugfix release
5938 * systems lacking /etc/os-release are no longer supported.
5940 * Various functionality updates to libsystemd-login.so
5942 * Track class of PAM logins to distinguish greeters from
5945 Contributions from: Kay Sievers, Lennart Poettering, Michael
5950 * This is an important bugfix release for v41.
5952 * Building man pages is now optional which should be useful
5953 for those building systemd from git but unwilling to install
5956 * Watchdog support for supervising services is now usable. In
5957 a future release support for hardware watchdogs
5958 (i.e. /dev/watchdog) will be added building on this.
5960 * Service start rate limiting is now configurable and can be
5961 turned off per service. When a start rate limit is hit a
5962 reboot can automatically be triggered.
5964 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5966 Contributions from: Benjamin Franzke, Bill Nottingham,
5967 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5968 Schmidt, Michał Górny, Piotr Drąg
5972 * The systemd binary is installed /usr/lib/systemd/systemd now;
5973 An existing /sbin/init symlink needs to be adapted with the
5976 * The code that loads kernel modules has been ported to invoke
5977 libkmod directly, instead of modprobe. This means we do not
5978 support systems with module-init-tools anymore.
5980 * Watchdog support is now already useful, but still not
5983 * A new kernel command line option systemd.setenv= is
5984 understood to set system wide environment variables
5985 dynamically at boot.
5987 * We now limit the set of capabilities of systemd-journald.
5989 * We now set SIGPIPE to ignore by default, since it only is
5990 useful in shell pipelines, and has little use in general
5991 code. This can be disabled with IgnoreSIPIPE=no in unit
5994 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5995 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6000 * This is mostly a bugfix release
6002 * We now expose the reason why a service failed in the
6003 "Result" D-Bus property.
6005 * Rudimentary service watchdog support (will be completed over
6006 the next few releases.)
6008 * When systemd forks off in order execute some service we will
6009 now immediately changes its argv[0] to reflect which process
6010 it will execute. This is useful to minimize the time window
6011 with a generic argv[0], which makes bootcharts more useful
6013 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6014 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6015 Mike Kazantsev, Ray Strode
6019 * This is mostly a test release, but incorporates many
6022 * New systemd-cgtop tool to show control groups by their
6025 * Linking against libacl for ACLs is optional again. If
6026 disabled, support tracking device access for active logins
6027 goes becomes unavailable, and so does access to the user
6028 journals by the respective users.
6030 * If a group "adm" exists, journal files are automatically
6031 owned by them, thus allow members of this group full access
6032 to the system journal as well as all user journals.
6034 * The journal now stores the SELinux context of the logging
6035 client for all entries.
6037 * Add C++ inclusion guards to all public headers
6039 * New output mode "cat" in the journal to print only text
6040 messages, without any meta data like date or time.
6042 * Include tiny X server wrapper as a temporary stop-gap to
6043 teach XOrg udev display enumeration. This is used by display
6044 managers such as gdm, and will go away as soon as XOrg
6045 learned native udev hotplugging for display devices.
6047 * Add new systemd-cat tool for executing arbitrary programs
6048 with STDERR/STDOUT connected to the journal. Can also act as
6049 BSD logger replacement, and does so by default.
6051 * Optionally store all locally generated coredumps in the
6052 journal along with meta data.
6054 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6055 writing short strings to files (for usage for /sys), and for
6056 creating symlinks, character and block device nodes.
6058 * New unit file option ControlGroupPersistent= to make cgroups
6059 persistent, following the mechanisms outlined in
6060 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6062 * Support multiple local RTCs in a sane way
6064 * No longer monopolize IO when replaying readahead data on
6065 rotating disks, since we might starve non-file-system IO to
6066 death, since fanotify() will not see accesses done by blkid,
6069 * Do not show kernel threads in systemd-cgls anymore, unless
6070 requested with new -k switch.
6072 Contributions from: Dan Horák, Kay Sievers, Lennart
6073 Poettering, Michal Schmidt
6077 * This is mostly a test release, but incorporates many
6080 * The git repository moved to:
6081 git://anongit.freedesktop.org/systemd/systemd
6082 ssh://git.freedesktop.org/git/systemd/systemd
6084 * First release with the journal
6085 http://0pointer.de/blog/projects/the-journal.html
6087 * The journal replaces both systemd-kmsg-syslogd and
6088 systemd-stdout-bridge.
6090 * New sd_pid_get_unit() API call in libsystemd-logind
6092 * Many systemadm clean-ups
6094 * Introduce remote-fs-pre.target which is ordered before all
6095 remote mounts and may be used to start services before all
6098 * Added Mageia support
6100 * Add bash completion for systemd-loginctl
6102 * Actively monitor PID file creation for daemons which exit in
6103 the parent process before having finished writing the PID
6104 file in the daemon process. Daemons which do this need to be
6105 fixed (i.e. PID file creation must have finished before the
6106 parent exits), but we now react a bit more gracefully to them.
6108 * Add colourful boot output, mimicking the well-known output
6109 of existing distributions.
6111 * New option PassCredentials= for socket units, for
6112 compatibility with a recent kernel ABI breakage.
6114 * /etc/rc.local is now hooked in via a generator binary, and
6115 thus will no longer act as synchronization point during
6118 * systemctl list-unit-files now supports --root=.
6120 * systemd-tmpfiles now understands two new commands: z, Z for
6121 relabelling files according to the SELinux database. This is
6122 useful to apply SELinux labels to specific files in /sys,
6125 * Output of SysV services is now forwarded to both the console
6126 and the journal by default, not only just the console.
6128 * New man pages for all APIs from libsystemd-login.
6130 * The build tree got reorganized and a the build system is a
6131 lot more modular allowing embedded setups to specifically
6132 select the components of systemd they are interested in.
6134 * Support for Linux systems lacking the kernel VT subsystem is
6137 * configure's --with-rootdir= got renamed to
6138 --with-rootprefix= to follow the naming used by udev and
6141 * Unless specified otherwise we will now install to /usr instead
6142 of /usr/local by default.
6144 * Processes with '@' in argv[0][0] are now excluded from the
6145 final shut-down killing spree, following the logic explained
6147 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6149 * All processes remaining in a service cgroup when we enter
6150 the START or START_PRE states are now killed with
6151 SIGKILL. That means it is no longer possible to spawn
6152 background processes from ExecStart= lines (which was never
6153 supported anyway, and bad style).
6155 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6156 reloading of units together.
6158 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6159 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6160 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6161 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6162 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob