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 behaviour 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 the
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
106 * A new command "systemctl revert" has been added that may be used to
107 revert to the vendor version of a unit file, in case local changes
108 have been made by adding drop-ins or overriding the unit file.
110 * "machinectl clean" gained a new verb to automatically remove all or
111 just hidden container images.
113 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
114 merged into the kernel in its current form.
116 * systemd-networkd gained support for configuring proxy ARP support for
117 each interface, via the ProxyArp= setting in .network files. It also
118 gained support for configuring the multicast querier feature of
119 bridge devices, via the new MulticastQuerier= setting in .netdev
120 files. A new setting PreferredLifetime= has been added for addresses
121 configured in .network file to configure the lifetime intended for an
124 * systemd-tmpfiles gained support for a new line type "e" for emptying
125 directories, if they exist, without creating them if they don't.
127 * journalctl learned a new output mode "-o short-unix" that outputs log
128 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
129 UTC). It also gained support for a new --no-hostname setting to
130 suppress the hostname column in the family of "short" output modes.
132 * systemd-nspawn gained support for automatically patching the UID/GIDs
133 of the owners and the ACLs of all files and directories in a
134 container tree to match the UID/GID user namespacing range selected
135 for the container invocation. This mode is enabled via the new
136 --private-user-chown switch. It also gained support for automatically
137 choosing a free, previously unused UID/GID range when starting a
138 container, via the new --private-users=pick setting (which implies
139 --private-user-chown). Together, these options for the first time
140 make user namespacing for nspawn containers fully automatic and thus
141 deployable. The systemd-nspaw@.service template unit file has been
142 changed to use this functionality by default.
144 * The default start timeout may now be configured on the kernel command
145 line via systemd.default_timeout_start_sec=. It was configurable
146 previously via the DefaultTimeoutStartSec= option in
147 /etc/systemd/system.conf already.
149 * Socket units gained a new TriggerLimitIntervalSec= and
150 TriggerLimitBurst= setting to configure a limit on the activation
151 rate of the socket unit.
153 * The LimitNICE= setting now optionally takes normal UNIX nice values
154 in addition to the raw integer limit value. If the specified
155 parameter is prefixed with "+" or "-" and is in the range -20..19 the
156 value is understood as UNIX nice value. If not prefixed like this it
157 is understood as raw RLIMIT_NICE limit.
159 Contributions from: Alban Crequy, Alexander Kuleshov, Alex Crawford,
160 Andrew Eikum, Beniamino Galvani, Benjamin Robin, Benjamin ROBIN, Biao
161 Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Colin Guthrie, Daniel
162 J Walsh, Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
163 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
164 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
165 Bui, frankheckenbach, Georgia Brikis, Harald Hoyer, Hendrik Brueckner,
166 Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo Puustinen, Jakub
167 Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth, kayrus, Klearchos
168 Chaloulos, Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukáš
169 Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt, Michael Biebl,
170 michaelolbrich, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletar,
171 Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin, mulkieran,
172 muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween, Nicolas
173 Braud-Santoni, Patrik Flykt, Peter Hutterer, Petr Lautrbach, Petros
174 Angelatos, Piotr Drąg, Rabin Vincent, Robert Węcławski, Ronny
175 Chevalier, Samuel Tardieu, Stefan Schallenberg, Steven Siloti, Susant
176 Sahani, Sylvain Plantefève, Taylor Smock, tblume, Tejun Heo, Thomas
177 Blume, Thomas Haller, Thomas Hindoe Paaboel Andersen, Thomas
178 H. P. Andersen, Tobias Klauser, Tom Gundersen, Torstein Husebø, Umut
179 Tezduyar Lindskog, Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam),
180 Vladimir Panteleev, Wieland Hoffmann, Wouter Verhelst, Yu Watanabe,
181 Zbigniew Jędrzejewski-Szmek
187 * The systemd-resolved DNS resolver service has gained a substantial
188 set of new features, most prominently it may now act as a DNSSEC
189 validating stub resolver. DNSSEC mode is currently turned off by
190 default, but is expected to be turned on by default in one of the
191 next releases. For now, we invite everybody to test the DNSSEC logic
192 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
193 service also gained a full set of D-Bus interfaces, including calls
194 to configure DNS and DNSSEC settings per link (for use by external
195 network management software). systemd-resolved and systemd-networkd
196 now distinguish between "search" and "routing" domains. The former
197 are used to qualify single-label names, the latter are used purely
198 for routing lookups within certain domains to specific links.
199 resolved now also synthesizes RRs for all entries from /etc/hosts.
201 * The systemd-resolve tool (which is a client utility for
202 systemd-resolved) has been improved considerably and is now fully
203 supported and documented. Hence it has moved from /usr/lib/systemd to
206 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
209 * The coredump collection logic has been reworked: when a coredump is
210 collected it is now written to disk, compressed and processed
211 (including stacktrace extraction) from a new instantiated service
212 systemd-coredump@.service, instead of directly from the
213 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
214 processing large coredumps can take up a substantial amount of
215 resources and time, and this previously happened entirely outside of
216 systemd's service supervision. With the new logic the core_pattern
217 hook only does minimal metadata collection before passing off control
218 to the new instantiated service, which is configured with a time
219 limit, a nice level and other settings to minimize negative impact on
220 the rest of the system. Also note that the new logic will honour the
221 RLIMIT_CORE setting of the crashed process, which now allows users
222 and processes to turn off coredumping for their processes by setting
225 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
226 and all forked processes by default. Previously, PID 1 would leave
227 the setting at "0" for all processes, as set by the kernel. Note that
228 the resource limit traditionally has no effect on the generated
229 coredumps on the system if the /proc/sys/kernel/core_pattern hook
230 logic is used. Since the limit is now honoured (see above) its
231 default has been changed so that the coredumping logic is enabled by
232 default for all processes, while allowing specific opt-out.
234 * When the stacktrace is extracted from processes of system users, this
235 is now done as "systemd-coredump" user, in order to sandbox this
236 potentially security sensitive parsing operation. (Note that when
237 processing coredumps of normal users this is done under the user ID
238 of process that crashed, as before.) Packagers should take notice
239 that it is now necessary to create the "systemd-coredump" system user
240 and group at package installation time.
242 * The systemd-activate socket activation testing tool gained support
243 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
244 and --seqpacket switches. It also has been extended to support both
245 new-style and inetd-style file descriptor passing. Use the new
246 --inetd switch to request inetd-style file descriptor passing.
248 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
249 variable, which takes a boolean value. If set to false, ANSI color
250 output is disabled in the tools even when run on a terminal that
253 * The VXLAN support in networkd now supports two new settings
254 DestinationPort= and PortRange=.
256 * A new systemd.machine_id= kernel command line switch has been added,
257 that may be used to set the machine ID in /etc/machine-id if it is
258 not initialized yet. This command line option has no effect if the
259 file is already initialized.
261 * systemd-nspawn gained a new --as-pid2 switch that invokes any
262 specified command line as PID 2 rather than PID 1 in the
263 container. In this mode PID 1 is a minimal stub init process that
264 implements the special POSIX and Linux semantics of PID 1 regarding
265 signal and child process management. Note that this stub init process
266 is implemented in nspawn itself and requires no support from the
267 container image. This new logic is useful to support running
268 arbitrary commands in the container, as normal processes are
269 generally not prepared to run as PID 1.
271 * systemd-nspawn gained a new --chdir= switch for setting the current
272 working directory for the process started in the container.
274 * "journalctl /dev/sda" will now output all kernel log messages for
275 specified device from the current boot, in addition to all devices
276 that are parents of it. This should make log output about devices
277 pretty useful, as long as kernel drivers attach enough metadata to
278 the log messages. (The usual SATA drivers do.)
280 * The sd-journal API gained two new calls
281 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
282 that report whether log data from /run or /var has been found.
284 * journalctl gained a new switch "--fields" that prints all journal
285 record field names currently in use in the journal. This is backed
286 by two new sd-journal API calls sd_journal_enumerate_fields() and
287 sd_journal_restart_fields().
289 * Most configurable timeouts in systemd now expect an argument of
290 "infinity" to turn them off, instead of "0" as before. The semantics
291 from now on is that a timeout of "0" means "now", and "infinity"
292 means "never". To maintain backwards compatibility, "0" continues to
293 turn off previously existing timeout settings.
295 * "systemctl reload-or-try-restart" has been renamed to "systemctl
296 try-reload-or-restart" to clarify what it actually does: the "try"
297 logic applies to both reloading and restarting, not just restarting.
298 The old name continues to be accepted for compatibility.
300 * On boot-up, when PID 1 detects that the system clock is behind the
301 release date of the systemd version in use, the clock is now set
302 to the latter. Previously, this was already done in timesyncd, in order
303 to avoid running with clocks set to the various clock epochs such as
304 1902, 1938 or 1970. With this change the logic is now done in PID 1
305 in addition to timesyncd during early boot-up, so that it is enforced
306 before the first process is spawned by systemd. Note that the logic
307 in timesyncd remains, as it is more comprehensive and ensures
308 clock monotonicity by maintaining a persistent timestamp file in
309 /var. Since /var is generally not available in earliest boot or the
310 initrd, this part of the logic remains in timesyncd, and is not done
313 * Support for tweaking details in net_cls.class_id through the
314 NetClass= configuration directive has been removed, as the kernel
315 people have decided to deprecate that controller in cgroup v2.
316 Userspace tools such as nftables are moving over to setting rules
317 that are specific to the full cgroup path of a task, which obsoletes
318 these controllers anyway. The NetClass= directive is kept around for
319 legacy compatibility reasons. For a more in-depth description of the
320 kernel change, please refer to the respective upstream commit:
322 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
324 * A new service setting RuntimeMaxSec= has been added that may be used
325 to specify a maximum runtime for a service. If the timeout is hit, the
326 service is terminated and put into a failure state.
328 * A new service setting AmbientCapabilities= has been added. It allows
329 configuration of additional Linux process capabilities that are
330 passed to the activated processes. This is only available on very
333 * The process resource limit settings in service units may now be used
334 to configure hard and soft limits individually.
336 * The various libsystemd APIs such as sd-bus or sd-event now publicly
337 expose support for gcc's __attribute__((cleanup())) C extension.
338 Specifically, for many object destructor functions alternative
339 versions have been added that have names suffixed with "p" and take a
340 pointer to a pointer to the object to destroy, instead of just a
341 pointer to the object itself. This is useful because these destructor
342 functions may be used directly as parameters to the cleanup
343 construct. Internally, systemd has been a heavy user of this GCC
344 extension for a long time, and with this change similar support is
345 now available to consumers of the library outside of systemd. Note
346 that by using this extension in your sources compatibility with old
347 and strictly ANSI compatible C compilers is lost. However, all gcc or
348 LLVM versions of recent years support this extension.
350 * Timer units gained support for a new setting RandomizedDelaySec= that
351 allows configuring some additional randomized delay to the configured
352 time. This is useful to spread out timer events to avoid load peaks in
353 clusters or larger setups.
355 * Calendar time specifications now support sub-second accuracy.
357 * Socket units now support listening on SCTP and UDP-lite protocol
360 * The sd-event API now comes with a full set of man pages.
362 * Older versions of systemd contained experimental support for
363 compressing journal files and coredumps with the LZ4 compressor that
364 was not compatible with the lz4 binary (due to API limitations of the
365 lz4 library). This support has been removed; only support for files
366 compatible with the lz4 binary remains. This LZ4 logic is now
367 officially supported and no longer considered experimental.
369 * The dkr image import logic has been removed again from importd. dkr's
370 micro-services focus doesn't fit into the machine image focus of
371 importd, and quickly got out of date with the upstream dkr API.
373 * Creation of the /run/lock/lockdev/ directory was dropped from
374 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
375 been available for many years. If you still need this, you need to
376 create your own tmpfiles.d config file with:
378 d /run/lock/lockdev 0775 root lock -
380 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
381 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
382 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
383 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
384 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
385 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
386 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
387 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
388 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
389 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
390 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
391 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
392 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
393 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
394 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
395 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
396 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
397 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
398 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
404 * A number of properties previously only settable in unit
405 files are now also available as properties to set when
406 creating transient units programmatically via the bus, as it
407 is exposed with systemd-run's --property=
408 setting. Specifically, these are: SyslogIdentifier=,
409 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
410 EnvironmentFile=, ReadWriteDirectories=,
411 ReadOnlyDirectories=, InaccessibleDirectories=,
412 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
414 * When creating transient services via the bus API it is now
415 possible to pass in a set of file descriptors to use as
416 STDIN/STDOUT/STDERR for the invoked process.
418 * Slice units may now be created transiently via the bus APIs,
419 similar to the way service and scope units may already be
422 * Wherever systemd expects a calendar timestamp specification
423 (like in journalctl's --since= and --until= switches) UTC
424 timestamps are now supported. Timestamps suffixed with "UTC"
425 are now considered to be in Universal Time Coordinated
426 instead of the local timezone. Also, timestamps may now
427 optionally be specified with sub-second accuracy. Both of
428 these additions also apply to recurring calendar event
429 specification, such as OnCalendar= in timer units.
431 * journalctl gained a new "--sync" switch that asks the
432 journal daemon to write all so far unwritten log messages to
433 disk and sync the files, before returning.
435 * systemd-tmpfiles learned two new line types "q" and "Q" that
436 operate like "v", but also set up a basic btrfs quota
437 hierarchy when used on a btrfs file system with quota
440 * tmpfiles' "v", "q" and "Q" will now create a plain directory
441 instead of a subvolume (even on a btrfs file system) if the
442 root directory is a plain directory, and not a
443 subvolume. This should simplify things with certain chroot()
444 environments which are not aware of the concept of btrfs
447 * systemd-detect-virt gained a new --chroot switch to detect
448 whether execution takes place in a chroot() environment.
450 * CPUAffinity= now takes CPU index ranges in addition to
453 * The various memory-related resource limit settings (such as
454 LimitAS=) now understand the usual K, M, G, ... suffixes to
455 the base of 1024 (IEC). Similar, the time-related resource
456 limit settings understand the usual min, h, day, ...
459 * There's a new system.conf setting DefaultTasksMax= to
460 control the default TasksMax= setting for services and
461 scopes running on the system. (TasksMax= is the primary
462 setting that exposes the "pids" cgroup controller on systemd
463 and was introduced in the previous systemd release.) The
464 setting now defaults to 512, which means services that are
465 not explicitly configured otherwise will only be able to
466 create 512 processes or threads at maximum, from this
467 version on. Note that this means that thread- or
468 process-heavy services might need to be reconfigured to set
469 TasksMax= to a higher value. It is sufficient to set
470 TasksMax= in these specific unit files to a higher value, or
471 even "infinity". Similar, there's now a logind.conf setting
472 UserTasksMax= that defaults to 4096 and limits the total
473 number of processes or tasks each user may own
474 concurrently. nspawn containers also have the TasksMax=
475 value set by default now, to 8192. Note that all of this
476 only has an effect if the "pids" cgroup controller is
477 enabled in the kernel. The general benefit of these changes
478 should be a more robust and safer system, that provides a
479 certain amount of per-service fork() bomb protection.
481 * systemd-nspawn gained the new --network-veth-extra= switch
482 to define additional and arbitrarily-named virtual Ethernet
483 links between the host and the container.
485 * A new service execution setting PassEnvironment= has been
486 added that allows importing select environment variables
487 from PID1's environment block into the environment block of
490 * Timer units gained support for a new RemainAfterElapse=
491 setting which takes a boolean argument. It defaults on on,
492 exposing behaviour unchanged to previous releases. If set to
493 off, timer units are unloaded after they elapsed if they
494 cannot elapse again. This is particularly useful for
495 transient timer units, which shall not stay around longer
496 than until they first elapse.
498 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
499 default now (the kernel default is 16). This is beneficial
500 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
501 allows substantially larger numbers of queued
502 datagrams. This should increase the capability of systemd to
503 parallelize boot-up, as logging and sd_notify() are unlikely
504 to stall execution anymore. If you need to change the value
505 from the new defaults, use the usual sysctl.d/ snippets.
507 * The compression framing format used by the journal or
508 coredump processing has changed to be in line with what the
509 official LZ4 tools generate. LZ4 compression support in
510 systemd was considered unsupported previously, as the format
511 was not compatible with the normal tools. With this release
512 this has changed now, and it is hence safe for downstream
513 distributions to turn it on. While not compressing as well
514 as the XZ, LZ4 is substantially faster, which makes
515 it a good default choice for the compression logic in the
516 journal and in coredump handling.
518 * Any reference to /etc/mtab has been dropped from
519 systemd. The file has been obsolete since a while, but
520 systemd refused to work on systems where it was incorrectly
521 set up (it should be a symlink or non-existent). Please make
522 sure to update to util-linux 2.27.1 or newer in conjunction
523 with this systemd release, which also drops any reference to
524 /etc/mtab. If you maintain a distribution make sure that no
525 software you package still references it, as this is a
526 likely source of bugs. There's also a glibc bug pending,
527 asking for removal of any reference to this obsolete file:
529 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
531 Note that only util-linux versions built with
532 --enable-libmount-force-mountinfo are supported.
534 * Support for the ".snapshot" unit type has been removed. This
535 feature turned out to be little useful and little used, and
536 has now been removed from the core and from systemctl.
538 * The dependency types RequiresOverridable= and
539 RequisiteOverridable= have been removed from systemd. They
540 have been used only very sparingly to our knowledge and
541 other options that provide a similar effect (such as
542 systemctl --mode=ignore-dependencies) are much more useful
543 and commonly used. Moreover, they were only half-way
544 implemented as the option to control behaviour regarding
545 these dependencies was never added to systemctl. By removing
546 these dependency types the execution engine becomes a bit
547 simpler. Unit files that use these dependencies should be
548 changed to use the non-Overridable dependency types
549 instead. In fact, when parsing unit files with these
550 options, that's what systemd will automatically convert them
551 too, but it will also warn, asking users to fix the unit
552 files accordingly. Removal of these dependency types should
553 only affect a negligible number of unit files in the wild.
555 * Behaviour of networkd's IPForward= option changed
556 (again). It will no longer maintain a per-interface setting,
557 but propagate one way from interfaces where this is enabled
558 to the global kernel setting. The global setting will be
559 enabled when requested by a network that is set up, but
560 never be disabled again. This change was made to make sure
561 IPv4 and IPv6 behaviour regarding packet forwarding is
562 similar (as the Linux IPv6 stack does not support
563 per-interface control of this setting) and to minimize
566 * In unit files the behaviour of %u, %U, %h, %s has
567 changed. These specifiers will now unconditionally resolve
568 to the various user database fields of the user that the
569 systemd instance is running as, instead of the user
570 configured in the specific unit via User=. Note that this
571 effectively doesn't change much, as resolving of these
572 specifiers was already turned off in the --system instance
573 of systemd, as we cannot do NSS lookups from PID 1. In the
574 --user instance of systemd these specifiers where correctly
575 resolved, but hardly made any sense, since the user instance
576 lacks privileges to do user switches anyway, and User= is
577 hence useless. Morever, even in the --user instance of
578 systemd behaviour was awkward as it would only take settings
579 from User= assignment placed before the specifier into
580 account. In order to unify and simplify the logic around
581 this the specifiers will now always resolve to the
582 credentials of the user invoking the manager (which in case
583 of PID 1 is the root user).
585 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
586 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
587 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
588 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
589 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
590 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
591 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
592 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
593 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
594 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
595 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
596 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
597 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
598 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
605 * systemd now depends on util-linux v2.27. More specifically,
606 the newly added mount monitor feature in libmount now
607 replaces systemd's former own implementation.
609 * libmount mandates /etc/mtab not to be regular file, and
610 systemd now enforces this condition at early boot.
611 /etc/mtab has been deprecated and warned about for a very
612 long time, so systems running systemd should already have
613 stopped having this file around as anything else than a
614 symlink to /proc/self/mounts.
616 * Support for the "pids" cgroup controller has been added. It
617 allows accounting the number of tasks in a cgroup and
618 enforcing limits on it. This adds two new setting
619 TasksAccounting= and TasksMax= to each unit, as well as a
620 global option DefaultTasksAccounting=.
622 * Support for the "net_cls" cgroup controller has been added.
623 It allows assigning a net class ID to each task in the
624 cgroup, which can then be used in firewall rules and traffic
625 shaping configurations. Note that the kernel netfilter net
626 class code does not currently work reliably for ingress
627 packets on unestablished sockets.
629 This adds a new config directive called NetClass= to CGroup
630 enabled units. Allowed values are positive numbers for fixed
631 assignments and "auto" for picking a free value
634 * 'systemctl is-system-running' now returns 'offline' if the
635 system is not booted with systemd. This command can now be
636 used as a substitute for 'systemd-notify --booted'.
638 * Watchdog timeouts have been increased to 3 minutes for all
639 in-tree service files. Apparently, disk IO issues are more
640 frequent than we hoped, and user reported >1 minute waiting
643 * 'machine-id-commit' functionality has been merged into
644 'machine-id-setup --commit'. The separate binary has been
647 * The WorkingDirectory= directive in unit files may now be set
648 to the special value '~'. In this case, the working
649 directory is set to the home directory of the user
652 * "machinectl shell" will now open the shell in the home
653 directory of the selected user by default.
655 * The CrashChVT= configuration file setting is renamed to
656 CrashChangeVT=, following our usual logic of not
657 abbreviating unnecessarily. The old directive is still
658 supported for compat reasons. Also, this directive now takes
659 an integer value between 1 and 63, or a boolean value. The
660 formerly supported '-1' value for disabling stays around for
663 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
664 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
665 RootDirectory= properties can now be set for transient
668 * The systemd-analyze tool gained a new "set-log-target" verb
669 to change the logging target the system manager logs to
670 dynamically during runtime. This is similar to how
671 "systemd-analyze set-log-level" already changes the log
674 * In nspawn /sys is now mounted as tmpfs, with only a selected
675 set of subdirectories mounted in from the real sysfs. This
676 enhances security slightly, and is useful for ensuring user
677 namespaces work correctly.
679 * Support for USB FunctionFS activation has been added. This
680 allows implementation of USB gadget services that are
681 activated as soon as they are requested, so that they don't
682 have to run continously, similar to classic socket
685 * The "systemctl exit" command now optionally takes an
686 additional parameter that sets the exit code to return from
687 the systemd manager when exiting. This is only relevant when
688 running the systemd user instance, or when running the
689 system instance in a container.
691 * sd-bus gained the new API calls sd_bus_path_encode_many()
692 and sd_bus_path_decode_many() that allow easy encoding and
693 decoding of multiple identifier strings inside a D-Bus
694 object path. Another new call sd_bus_default_flush_close()
695 has been added to flush and close per-thread default
698 * systemd-cgtop gained support for a -M/--machine= switch to
699 show the control groups within a certain container only.
701 * "systemctl kill" gained support for an optional --fail
702 switch. If specified the requested operation will fail of no
703 processes have been killed, because the unit had no
704 processes attached, or similar.
706 * A new systemd.crash_reboot=1 kernel command line option has
707 been added that triggers a reboot after crashing. This can
708 also be set through CrashReboot= in systemd.conf.
710 * The RuntimeDirectory= setting now understands unit
711 specifiers like %i or %f.
713 * A new (still internal) libary API sd-ipv4acd has been added,
714 that implements address conflict detection for IPv4. It's
715 based on code from sd-ipv4ll, and will be useful for
716 detecting DHCP address conflicts.
718 * File descriptors passed during socket activation may now be
719 named. A new API sd_listen_fds_with_names() is added to
720 access the names. The default names may be overridden,
721 either in the .socket file using the FileDescriptorName=
722 parameter, or by passing FDNAME= when storing the file
723 descriptors using sd_notify().
725 * systemd-networkd gained support for:
727 - Setting the IPv6 Router Advertisment settings via
728 IPv6AcceptRouterAdvertisements= in .network files.
730 - Configuring the HelloTimeSec=, MaxAgeSec= and
731 ForwardDelaySec= bridge parameters in .netdev files.
733 - Configuring PreferredSource= for static routes in
736 * The "ask-password" framework used to query for LUKS harddisk
737 passwords or SSL passwords during boot gained support for
738 caching passwords in the kernel keyring, if it is
739 available. This makes sure that the user only has to type in
740 a passphrase once if there are multiple objects to unlock
741 with the same one. Previously, such password caching was
742 available only when Plymouth was used; this moves the
743 caching logic into the systemd codebase itself. The
744 "systemd-ask-password" utility gained a new --keyname=
745 switch to control which kernel keyring key to use for
746 caching a password in. This functionality is also useful for
747 enabling display managers such as gdm to automatically
748 unlock the user's GNOME keyring if its passphrase, the
749 user's password and the harddisk password are the same, if
750 gdm-autologin is used.
752 * When downloading tar or raw images using "machinectl
753 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
754 file is now also downloaded, if it is available and stored
755 next to the image file.
757 * Units of type ".socket" gained a new boolean setting
758 Writable= which is only useful in conjunction with
759 ListenSpecial=. If true, enables opening the specified
760 special file in O_RDWR mode rather than O_RDONLY mode.
762 * systemd-rfkill has been reworked to become a singleton
763 service that is activated through /dev/rfkill on each rfkill
764 state change and saves the settings to disk. This way,
765 systemd-rfkill is now compatible with devices that exist
766 only intermittendly, and even restores state if the previous
767 system shutdown was abrupt rather than clean.
769 * The journal daemon gained support for vacuuming old journal
770 files controlled by the number of files that shall remain,
771 in addition to the already existing control by size and by
772 date. This is useful as journal interleaving performance
773 degrades with too many seperate journal files, and allows
774 putting an effective limit on them. The new setting defaults
775 to 100, but this may be changed by setting SystemMaxFiles=
776 and RuntimeMaxFiles= in journald.conf. Also, the
777 "journalctl" tool gained the new --vacuum-files= switch to
778 manually vacuum journal files to leave only the specified
779 number of files in place.
781 * udev will now create /dev/disk/by-path links for ATA devices
782 on kernels where that is supported.
784 * Galician, Serbian, Turkish and Korean translations were added.
786 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
787 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
788 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
789 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
790 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
791 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
792 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
793 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
794 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
795 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
796 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
797 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
798 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
799 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
800 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
801 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
802 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
803 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
809 * The DHCP implementation of systemd-networkd gained a set of
812 - The DHCP server now supports emitting DNS and NTP
813 information. It may be enabled and configured via
814 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
815 and NTP information is enabled, but no servers are
816 configured, the corresponding uplink information (if there
817 is any) is propagated.
819 - Server and client now support transmission and reception
820 of timezone information. It can be configured via the
821 newly introduced network options UseTimezone=,
822 EmitTimezone=, and Timezone=. Transmission of timezone
823 information is enabled between host and containers by
824 default now: the container will change its local timezone
825 to what the host has set.
827 - Lease timeouts can now be configured via
828 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
830 - The DHCP server improved on the stability of
831 leases. Clients are more likely to get the same lease
832 information back, even if the server loses state.
834 - The DHCP server supports two new configuration options to
835 control the lease address pool metrics, PoolOffset= and
838 * The encapsulation limit of tunnels in systemd-networkd may
839 now be configured via 'EncapsulationLimit='. It allows
840 modifying the maximum additional levels of encapsulation
841 that are permitted to be prepended to a packet.
843 * systemd now supports the concept of user buses replacing
844 session buses, if used with dbus-1.10 (and enabled via dbus
845 --enable-user-session). It previously only supported this on
846 kdbus-enabled systems, and this release expands this to
847 'dbus-daemon' systems.
849 * systemd-networkd now supports predictable interface names
852 * systemd now optionally supports the new Linux kernel
853 "unified" control group hierarchy. If enabled via the kernel
854 command-line option 'systemd.unified_cgroup_hierarchy=1',
855 systemd will try to mount the unified cgroup hierarchy
856 directly on /sys/fs/cgroup. If not enabled, or not
857 available, systemd will fall back to the legacy cgroup
858 hierarchy setup, as before. Host system and containers can
859 mix and match legacy and unified hierarchies as they
860 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
861 environment variable to individually select the hierarchy to
862 use for executed containers. By default, nspawn will use the
863 unified hierarchy for the containers if the host uses the
864 unified hierarchy, and the legacy hierarchy otherwise.
865 Please note that at this point the unified hierarchy is an
866 experimental kernel feature and is likely to change in one
867 of the next kernel releases. Therefore, it should not be
868 enabled by default in downstream distributions yet. The
869 minimum required kernel version for the unified hierarchy to
870 work is 4.2. Note that when the unified hierarchy is used
871 for the first time delegated access to controllers is
872 safe. Because of this systemd-nspawn containers will get
873 access to controllers now, as will systemd user
874 sessions. This means containers and user sessions may now
875 manage their own resources, partitioning up what the system
878 * A new special scope unit "init.scope" has been introduced
879 that encapsulates PID 1 of the system. It may be used to
880 determine resource usage and enforce resource limits on PID
881 1 itself. PID 1 hence moved out of the root of the control
884 * The cgtop tool gained support for filtering out kernel
885 threads when counting tasks in a control group. Also, the
886 count of processes is now recursively summed up by
887 default. Two options -k and --recursive= have been added to
888 revert to old behaviour. The tool has also been updated to
889 work correctly in containers now.
891 * systemd-nspawn's --bind= and --bind-ro= options have been
892 extended to allow creation of non-recursive bind mounts.
894 * libsystemd gained two new calls sd_pid_get_cgroup() and
895 sd_peer_get_cgroup() which return the control group path of
896 a process or peer of a connected AF_UNIX socket. This
897 function call is particularly useful when implementing
898 delegated subtrees support in the control group hierarchy.
900 * The "sd-event" event loop API of libsystemd now supports
901 correct dequeuing of real-time signals, without losing
904 * When systemd requests a PolicyKit decision when managing
905 units it will now add additional fields to the request,
906 including unit name and desired operation. This enables more
907 powerful PolicyKit policies, that make decisions depending
910 * nspawn learnt support for .nspawn settings files, that may
911 accompany the image files or directories of containers, and
912 may contain additional settings for the container. This is
913 an alternative to configuring container parameters via the
916 Contributions from: Cristian Rodríguez, Daniel Mack, David
917 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
918 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
919 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
920 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
921 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
922 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
923 Andersen, Tom Gundersen, Torstein Husebø
929 * machinectl gained a new verb 'shell' which opens a fresh
930 shell on the target container or the host. It is similar to
931 the existing 'login' command of machinectl, but spawns the
932 shell directly without prompting for username or
933 password. The pseudo machine '.host' now refers to the local
934 host and is used by default. Hence, 'machinectl shell' can
935 be used as replacement for 'su -' which spawns a session as
936 a fresh systemd unit in a way that is fully isolated from
937 the originating session.
939 * systemd-networkd learned to cope with private-zone DHCP
940 options and allows other programs to query the values.
942 * SELinux access control when enabling/disabling units is no
943 longer enforced with this release. The previous
944 implementation was incorrect, and a new corrected
945 implementation is not yet available. As unit file operations
946 are still protected via PolicyKit and D-Bus policy this is
947 not a security problem. Yet, distributions which care about
948 optimal SELinux support should probably not stabilize on
951 * sd-bus gained support for matches of type "arg0has=", that
952 test for membership of strings in string arrays sent in bus
955 * systemd-resolved now dumps the contents of its DNS and LLMNR
956 caches to the logs on reception of the SIGUSR1 signal. This
957 is useful to debug DNS behaviour.
959 * The coredumpctl tool gained a new --directory= option to
960 operate on journal files in a specific directory.
962 * "systemctl reboot" and related commands gained a new
963 "--message=" option which may be used to set a free-text
964 wall message when shutting down or rebooting the
965 system. This message is also logged, which is useful for
966 figuring out the reason for a reboot or shutdown a
969 * The "systemd-resolve-host" tool's -i switch now takes
970 network interface numbers as alternative to interface names.
972 * A new unit file setting for services has been introduced:
973 UtmpMode= allows configuration of how precisely systemd
974 handles utmp and wtmp entries for the service if this is
975 enabled. This allows writing services that appear similar to
976 user sessions in the output of the "w", "who", "last" and
979 * systemd-resolved will now locally synthesize DNS resource
980 records for the "localhost" and "gateway" domains as well as
981 the local hostname. This should ensure that clients querying
982 RRs via resolved will get similar results as those going via
983 NSS, if nss-myhostname is enabled.
985 Contributions from: Alastair Hughes, Alex Crawford, Daniel
986 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
987 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
988 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
989 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
990 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
991 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
992 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
993 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
994 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
995 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
996 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1002 * The systemd-efi-boot-generator functionality was merged into
1003 systemd-gpt-auto-generator.
1005 * systemd-networkd now supports Group Policy for vxlan
1006 devices. It can be enabled via the new boolean configuration
1007 option called 'GroupPolicyExtension='.
1009 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1010 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1011 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1013 — Berlin, 2015-07-31
1017 * The python-systemd code has been removed from the systemd repository.
1018 A new repository has been created which accommodates the code from
1019 now on, and we kindly ask distributions to create a separate package
1020 for this: https://github.com/systemd/python-systemd
1022 * The systemd daemon will now reload its main configuration
1023 (/etc/systemd/system.conf) on daemon-reload.
1025 * sd-dhcp now exposes vendor specific extensions via
1026 sd_dhcp_lease_get_vendor_specific().
1028 * systemd-networkd gained a number of new configuration options.
1030 - A new boolean configuration option for TAP devices called
1031 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1032 device, thus allowing to send and receive GSO packets.
1034 - A new tunnel configuration option called 'CopyDSCP='.
1035 If enabled, the DSCP field of ip6 tunnels is copied into the
1036 decapsulated packet.
1038 - A set of boolean bridge configuration options were added.
1039 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1040 and 'UnicastFlood=' are now parsed by networkd and applied to the
1041 respective bridge link device via the respective IFLA_BRPORT_*
1044 - A new string configuration option to override the hostname sent
1045 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1046 is true, networkd will use the configured hostname instead of the
1047 system hostname when sending DHCP requests.
1049 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1050 networkd will configure the IPv6 flow-label of the tunnel device
1051 according to RFC2460.
1053 - The 'macvtap' virtual network devices are now supported, similar to
1054 the already supported 'macvlan' devices.
1056 * systemd-resolved now implements RFC5452 to improve resilience against
1057 cache poisoning. Additionally, source port randomization is enabled
1058 by default to further protect against DNS spoofing attacks.
1060 * nss-mymachines now supports translating UIDs and GIDs of running
1061 containers with user-namespaces enabled. If a container 'foo'
1062 translates a host uid 'UID' to the container uid 'TUID', then
1063 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1064 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1065 mapped as 'vg-foo-TGID'.
1067 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1068 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1069 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1070 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1071 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1072 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1073 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1074 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1075 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1076 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1078 — Berlin, 2015-07-29
1082 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1083 There are no known issues with current sysfs, and udev does not need
1084 or should be used to work around such bugs.
1086 * udev does no longer enable USB HID power management. Several reports
1087 indicate, that some devices cannot handle that setting.
1089 * The udev accelerometer helper was removed. The functionality
1090 is now fully included in iio-sensor-proxy. But this means,
1091 older iio-sensor-proxy versions will no longer provide
1092 accelerometer/orientation data with this systemd version.
1093 Please upgrade iio-sensor-proxy to version 1.0.
1095 * networkd gained a new configuration option IPv6PrivacyExtensions=
1096 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1097 for Stateless Address") on selected networks.
1099 * For the sake of fewer build-time dependencies and less code in the
1100 main repository, the python bindings are about to be removed in the
1101 next release. A new repository has been created which accommodates
1102 the code from now on, and we kindly ask distributions to create a
1103 separate package for this. The removal will take place in v223.
1105 https://github.com/systemd/python-systemd
1107 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1108 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1109 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1110 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1111 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1112 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1113 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1114 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1115 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1116 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1118 — Berlin, 2015-07-07
1122 * The sd-bus.h and sd-event.h APIs have now been declared
1123 stable and have been added to the official interface of
1124 libsystemd.so. sd-bus implements an alternative D-Bus client
1125 library, that is relatively easy to use, very efficient and
1126 supports both classic D-Bus as well as kdbus as transport
1127 backend. sd-event is a generic event loop abstraction that
1128 is built around Linux epoll, but adds features such as event
1129 prioritization or efficient timer handling. Both APIs are good
1130 choices for C programs looking for a bus and/or event loop
1131 implementation that is minimal and does not have to be
1132 portable to other kernels.
1134 * kdbus support is no longer compile-time optional. It is now
1135 always built-in. However, it can still be disabled at
1136 runtime using the kdbus=0 kernel command line setting, and
1137 that setting may be changed to default to off, by specifying
1138 --disable-kdbus at build-time. Note though that the kernel
1139 command line setting has no effect if the kdbus.ko kernel
1140 module is not installed, in which case kdbus is (obviously)
1141 also disabled. We encourage all downstream distributions to
1142 begin testing kdbus by adding it to the kernel images in the
1143 development distributions, and leaving kdbus support in
1146 * The minimal required util-linux version has been bumped to
1149 * Support for chkconfig (--enable-chkconfig) was removed in
1150 favor of calling an abstraction tool
1151 /lib/systemd/systemd-sysv-install. This needs to be
1152 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1153 in README for details.
1155 * If there's a systemd unit and a SysV init script for the
1156 same service name, and the user executes "systemctl enable"
1157 for it (or a related call), then this will now enable both
1158 (or execute the related operation on both), not just the
1161 * The libudev API documentation has been converted from gtkdoc
1164 * gudev has been removed from the systemd tree, it is now an
1167 * The systemd-cgtop tool learnt a new --raw switch to generate
1168 "raw" (machine parsable) output.
1170 * networkd's IPForwarding= .network file setting learnt the
1171 new setting "kernel", which ensures that networkd does not
1172 change the IP forwarding sysctl from the default kernel
1175 * The systemd-logind bus API now exposes a new boolean
1176 property "Docked" that reports whether logind considers the
1177 system "docked", i.e. connected to a docking station or not.
1179 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1180 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1181 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1182 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1183 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1184 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1185 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1186 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1187 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1188 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1189 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1190 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1191 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1192 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1193 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1194 Fink, Zbigniew Jędrzejewski-Szmek
1196 — Berlin, 2015-06-19
1200 * The gudev library has been extracted into a separate repository
1201 available at: https://git.gnome.org/browse/libgudev/
1202 It is now managed as part of the Gnome project. Distributions
1203 are recommended to pass --disable-gudev to systemd and use
1204 gudev from the Gnome project instead. gudev is still included
1205 in systemd, for now. It will be removed soon, though. Please
1206 also see the announcement-thread on systemd-devel:
1207 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1209 * systemd now exposes a CPUUsageNSec= property for each
1210 service unit on the bus, that contains the overall consumed
1211 CPU time of a service (the sum of what each process of the
1212 service consumed). This value is only available if
1213 CPUAccounting= is turned on for a service, and is then shown
1214 in the "systemctl status" output.
1216 * Support for configuring alternative mappings of the old SysV
1217 runlevels to systemd targets has been removed. They are now
1218 hardcoded in a way that runlevels 2, 3, 4 all map to
1219 multi-user.target and 5 to graphical.target (which
1220 previously was already the default behaviour).
1222 * The auto-mounter logic gained support for mount point
1223 expiry, using a new TimeoutIdleSec= setting in .automount
1224 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1226 * The EFI System Partition (ESP) as mounted to /boot by
1227 systemd-efi-boot-generator will now be unmounted
1228 automatically after 2 minutes of not being used. This should
1229 minimize the risk of ESP corruptions.
1231 * New /etc/fstab options x-systemd.requires= and
1232 x-systemd.requires-mounts-for= are now supported to express
1233 additional dependencies for mounts. This is useful for
1234 journalling file systems that support external journal
1235 devices or overlay file systems that require underlying file
1236 systems to be mounted.
1238 * systemd does not support direct live-upgrades (via systemctl
1239 daemon-reexec) from versions older than v44 anymore. As no
1240 distribution we are aware of shipped such old versions in a
1241 stable release this should not be problematic.
1243 * When systemd forks off a new per-connection service instance
1244 it will now set the $REMOTE_ADDR environment variable to the
1245 remote IP address, and $REMOTE_PORT environment variable to
1246 the remote IP port. This behaviour is similar to the
1247 corresponding environment variables defined by CGI.
1249 * systemd-networkd gained support for uplink failure
1250 detection. The BindCarrier= option allows binding interface
1251 configuration dynamically to the link sense of other
1252 interfaces. This is useful to achieve behaviour like in
1255 * systemd-networkd gained support for configuring the DHCP
1256 client identifier to use when requesting leases.
1258 * systemd-networkd now has a per-network UseNTP= option to
1259 configure whether NTP server information acquired via DHCP
1260 is passed on to services like systemd-timesyncd.
1262 * systemd-networkd gained support for vti6 tunnels.
1264 * Note that systemd-networkd manages the sysctl variable
1265 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1266 it is configured for since v219. The variable controls IP
1267 forwarding, and is a per-interface alternative to the global
1268 /proc/sys/net/ipv[46]/ip_forward. This setting is
1269 configurable in the IPForward= option, which defaults to
1270 "no". This means if networkd is used for an interface it is
1271 no longer sufficient to set the global sysctl option to turn
1272 on IP forwarding! Instead, the .network file option
1273 IPForward= needs to be turned on! Note that the
1274 implementation of this behaviour was broken in v219 and has
1277 * Many bonding and vxlan options are now configurable in
1280 * systemd-nspawn gained a new --property= setting to set unit
1281 properties for the container scope. This is useful for
1282 setting resource parameters (e.g "CPUShares=500") on
1283 containers started from the command line.
1285 * systemd-nspawn gained a new --private-users= switch to make
1286 use of user namespacing available on recent Linux kernels.
1288 * systemd-nspawn may now be called as part of a shell pipeline
1289 in which case the pipes used for stdin and stdout are passed
1290 directly to the process invoked in the container, without
1291 indirection via a pseudo tty.
1293 * systemd-nspawn gained a new switch to control the UNIX
1294 signal to use when killing the init process of the container
1297 * systemd-nspawn gained a new --overlay= switch for mounting
1298 overlay file systems into the container using the new kernel
1301 * When a container image is imported via systemd-importd and
1302 the host file system is not btrfs, a loopback block device
1303 file is created in /var/lib/machines.raw with a btrfs file
1304 system inside. It is then mounted to /var/lib/machines to
1305 enable btrfs features for container management. The loopback
1306 file and btrfs file system is grown as needed when container
1307 images are imported via systemd-importd.
1309 * systemd-machined/systemd-importd gained support for btrfs
1310 quota, to enforce container disk space limits on disk. This
1311 is exposed in "machinectl set-limit".
1313 * systemd-importd now can import containers from local .tar,
1314 .raw and .qcow2 images, and export them to .tar and .raw. It
1315 can also import dkr v2 images now from the network (on top
1318 * systemd-importd gained support for verifying downloaded
1319 images with gpg2 (previously only gpg1 was supported).
1321 * systemd-machined, systemd-logind, systemd: most bus calls
1322 are now accessible to unprivileged processes via
1323 PolicyKit. Also, systemd-logind will now allow users to kill
1324 their own sessions without further privileges or
1327 * systemd-shutdownd has been removed. This service was
1328 previously responsible for implementing scheduled shutdowns
1329 as exposed in /usr/bin/shutdown's time parameter. This
1330 functionality has now been moved into systemd-logind and is
1331 accessible via a bus interface.
1333 * "systemctl reboot" gained a new switch --firmware-setup that
1334 can be used to reboot into the EFI firmware setup, if that
1335 is available. systemd-logind now exposes an API on the bus
1336 to trigger such reboots, in case graphical desktop UIs want
1337 to cover this functionality.
1339 * "systemctl enable", "systemctl disable" and "systemctl mask"
1340 now support a new "--now" switch. If specified the units
1341 that are enabled will also be started, and the ones
1342 disabled/masked also stopped.
1344 * The Gummiboot EFI boot loader tool has been merged into
1345 systemd, and renamed to "systemd-boot". The bootctl tool has been
1346 updated to support systemd-boot.
1348 * An EFI kernel stub has been added that may be used to create
1349 kernel EFI binaries that contain not only the actual kernel,
1350 but also an initrd, boot splash, command line and OS release
1351 information. This combined binary can then be signed as a
1352 single image, so that the firmware can verify it all in one
1353 step. systemd-boot has special support for EFI binaries created
1354 like this and can extract OS release information from them
1355 and show them in the boot menu. This functionality is useful
1356 to implement cryptographically verified boot schemes.
1358 * Optional support has been added to systemd-fsck to pass
1359 fsck's progress report to an AF_UNIX socket in the file
1362 * udev will no longer create device symlinks for all block
1363 devices by default. A blacklist for excluding special block
1364 devices from this logic has been turned into a whitelist
1365 that requires picking block devices explicitly that require
1368 * A new (currently still internal) API sd-device.h has been
1369 added to libsystemd. This modernized API is supposed to
1370 replace libudev eventually. In fact, already much of libudev
1371 is now just a wrapper around sd-device.h.
1373 * A new hwdb database for storing metadata about pointing
1374 stick devices has been added.
1376 * systemd-tmpfiles gained support for setting file attributes
1377 similar to the "chattr" tool with new 'h' and 'H' lines.
1379 * systemd-journald will no longer unconditionally set the
1380 btrfs NOCOW flag on new journal files. This is instead done
1381 with tmpfiles snippet using the new 'h' line type. This
1382 allows easy disabling of this logic, by masking the
1383 journal-nocow.conf tmpfiles file.
1385 * systemd-journald will now translate audit message types to
1386 human readable identifiers when writing them to the
1387 journal. This should improve readability of audit messages.
1389 * The LUKS logic gained support for the offset= and skip=
1390 options in /etc/crypttab, as previously implemented by
1393 * /usr/lib/os-release gained a new optional field VARIANT= for
1394 distributions that support multiple variants (such as a
1395 desktop edition, a server edition, ...)
1397 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1398 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1399 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1400 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1401 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1402 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1403 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1404 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1405 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1406 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1407 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1408 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1409 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1410 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1411 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1412 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1413 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1414 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1415 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1416 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1417 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1418 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1419 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1420 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1421 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1422 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1423 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1425 — Berlin, 2015-05-22
1429 * Introduce a new API "sd-hwdb.h" for querying the hardware
1430 metadata database. With this minimal interface one can query
1431 and enumerate the udev hwdb, decoupled from the old libudev
1432 library. libudev's interface for this is now only a wrapper
1433 around sd-hwdb. A new tool systemd-hwdb has been added to
1434 interface with and update the database.
1436 * When any of systemd's tools copies files (for example due to
1437 tmpfiles' C lines) a btrfs reflink will attempted first,
1438 before bytewise copying is done.
1440 * systemd-nspawn gained a new --ephemeral switch. When
1441 specified a btrfs snapshot is taken of the container's root
1442 directory, and immediately removed when the container
1443 terminates again. Thus, a container can be started whose
1444 changes never alter the container's root directory, and are
1445 lost on container termination. This switch can also be used
1446 for starting a container off the root file system of the
1447 host without affecting the host OS. This switch is only
1448 available on btrfs file systems.
1450 * systemd-nspawn gained a new --template= switch. It takes the
1451 path to a container tree to use as template for the tree
1452 specified via --directory=, should that directory be
1453 missing. This allows instantiating containers dynamically,
1454 on first run. This switch is only available on btrfs file
1457 * When a .mount unit refers to a mount point on which multiple
1458 mounts are stacked, and the .mount unit is stopped all of
1459 the stacked mount points will now be unmounted until no
1460 mount point remains.
1462 * systemd now has an explicit notion of supported and
1463 unsupported unit types. Jobs enqueued for unsupported unit
1464 types will now fail with an "unsupported" error code. More
1465 specifically .swap, .automount and .device units are not
1466 supported in containers, .busname units are not supported on
1467 non-kdbus systems. .swap and .automount are also not
1468 supported if their respective kernel compile time options
1471 * machinectl gained support for two new "copy-from" and
1472 "copy-to" commands for copying files from a running
1473 container to the host or vice versa.
1475 * machinectl gained support for a new "bind" command to bind
1476 mount host directories into local containers. This is
1477 currently only supported for nspawn containers.
1479 * networkd gained support for configuring bridge forwarding
1480 database entries (fdb) from .network files.
1482 * A new tiny daemon "systemd-importd" has been added that can
1483 download container images in tar, raw, qcow2 or dkr formats,
1484 and make them available locally in /var/lib/machines, so
1485 that they can run as nspawn containers. The daemon can GPG
1486 verify the downloads (not supported for dkr, since it has no
1487 provisions for verifying downloads). It will transparently
1488 decompress bz2, xz, gzip compressed downloads if necessary,
1489 and restore sparse files on disk. The daemon uses privilege
1490 separation to ensure the actual download logic runs with
1491 fewer privileges than the daemon itself. machinectl has
1492 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1493 make the functionality of importd available to the
1494 user. With this in place the Fedora and Ubuntu "Cloud"
1495 images can be downloaded and booted as containers unmodified
1496 (the Fedora images lack the appropriate GPG signature files
1497 currently, so they cannot be verified, but this will change
1498 soon, hopefully). Note that downloading images is currently
1499 only fully supported on btrfs.
1501 * machinectl is now able to list container images found in
1502 /var/lib/machines, along with some metadata about sizes of
1503 disk and similar. If the directory is located on btrfs and
1504 quota is enabled, this includes quota display. A new command
1505 "image-status" has been added that shows additional
1506 information about images.
1508 * machinectl is now able to clone container images
1509 efficiently, if the underlying file system (btrfs) supports
1510 it, with the new "machinectl clone" command. It also
1511 gained commands for renaming and removing images, as well as
1512 marking them read-only or read-write (supported also on
1513 legacy file systems).
1515 * networkd gained support for collecting LLDP network
1516 announcements, from hardware that supports this. This is
1517 shown in networkctl output.
1519 * systemd-run gained support for a new -t (--pty) switch for
1520 invoking a binary on a pty whose input and output is
1521 connected to the invoking terminal. This allows executing
1522 processes as system services while interactively
1523 communicating with them via the terminal. Most interestingly
1524 this is supported across container boundaries. Invoking
1525 "systemd-run -t /bin/bash" is an alternative to running a
1526 full login session, the difference being that the former
1527 will not register a session, nor go through the PAM session
1530 * tmpfiles gained support for a new "v" line type for creating
1531 btrfs subvolumes. If the underlying file system is a legacy
1532 file system, this automatically degrades to creating a
1533 normal directory. Among others /var/lib/machines is now
1534 created like this at boot, should it be missing.
1536 * The directory /var/lib/containers/ has been deprecated and
1537 been replaced by /var/lib/machines. The term "machines" has
1538 been used in the systemd context as generic term for both
1539 VMs and containers, and hence appears more appropriate for
1540 this, as the directory can also contain raw images bootable
1543 * systemd-nspawn when invoked with -M but without --directory=
1544 or --image= is now capable of searching for the container
1545 root directory, subvolume or disk image automatically, in
1546 /var/lib/machines. systemd-nspawn@.service has been updated
1547 to make use of this, thus allowing it to be used for raw
1550 * A new machines.target unit has been introduced that is
1551 supposed to group all containers/VMs invoked as services on
1552 the system. systemd-nspawn@.service has been updated to
1553 integrate with that.
1555 * machinectl gained a new "start" command, for invoking a
1556 container as a service. "machinectl start foo" is mostly
1557 equivalent to "systemctl start systemd-nspawn@foo.service",
1558 but handles escaping in a nicer way.
1560 * systemd-nspawn will now mount most of the cgroupfs tree
1561 read-only into each container, with the exception of the
1562 container's own subtree in the name=systemd hierarchy.
1564 * journald now sets the special FS_NOCOW file flag for its
1565 journal files. This should improve performance on btrfs, by
1566 avoiding heavy fragmentation when journald's write-pattern
1567 is used on COW file systems. It degrades btrfs' data
1568 integrity guarantees for the files to the same levels as for
1569 ext3/ext4 however. This should be OK though as journald does
1570 its own data integrity checks and all its objects are
1571 checksummed on disk. Also, journald should handle btrfs disk
1572 full events a lot more gracefully now, by processing SIGBUS
1573 errors, and not relying on fallocate() anymore.
1575 * When journald detects that journal files it is writing to
1576 have been deleted it will immediately start new journal
1579 * systemd now provides a way to store file descriptors
1580 per-service in PID 1.This is useful for daemons to ensure
1581 that fds they require are not lost during a daemon
1582 restart. The fds are passed to the daemon on the next
1583 invocation in the same way socket activation fds are
1584 passed. This is now used by journald to ensure that the
1585 various sockets connected to all the system's stdout/stderr
1586 are not lost when journald is restarted. File descriptors
1587 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1588 an extension to sd_notify(). Note that a limit is enforced
1589 on the number of fds a service can store in PID 1, and it
1590 defaults to 0, so that no fds may be stored, unless this is
1591 explicitly turned on.
1593 * The default TERM variable to use for units connected to a
1594 terminal, when no other value is explicitly is set is now
1595 vt220 rather than vt102. This should be fairly safe still,
1596 but allows PgUp/PgDn work.
1598 * The /etc/crypttab option header= as known from Debian is now
1601 * "loginctl user-status" and "loginctl session-status" will
1602 now show the last 10 lines of log messages of the
1603 user/session following the status output. Similar,
1604 "machinectl status" will show the last 10 log lines
1605 associated with a virtual machine or container
1606 service. (Note that this is usually not the log messages
1607 done in the VM/container itself, but simply what the
1608 container manager logs. For nspawn this includes all console
1611 * "loginctl session-status" without further argument will now
1612 show the status of the session of the caller. Similar,
1613 "lock-session", "unlock-session", "activate",
1614 "enable-linger", "disable-linger" may now be called without
1615 session/user parameter in which case they apply to the
1616 caller's session/user.
1618 * An X11 session scriptlet is now shipped that uploads
1619 $DISPLAY and $XAUTHORITY into the environment of the systemd
1620 --user daemon if a session begins. This should improve
1621 compatibility with X11 enabled applications run as systemd
1624 * Generators are now subject to masking via /etc and /run, the
1625 same way as unit files.
1627 * networkd .network files gained support for configuring
1628 per-link IPv4/IPv6 packet forwarding as well as IPv4
1629 masquerading. This is by default turned on for veth links to
1630 containers, as registered by systemd-nspawn. This means that
1631 nspawn containers run with --network-veth will now get
1632 automatic routed access to the host's networks without any
1633 further configuration or setup, as long as networkd runs on
1636 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1637 or UDP posts of a container on the host. With this in place
1638 it is possible to run containers with private veth links
1639 (--network-veth), and have their functionality exposed on
1640 the host as if their services were running directly on the
1643 * systemd-nspawn's --network-veth switch now gained a short
1644 version "-n", since with the changes above it is now truly
1645 useful out-of-the-box. The systemd-nspawn@.service has been
1646 updated to make use of it too by default.
1648 * systemd-nspawn will now maintain a per-image R/W lock, to
1649 ensure that the same image is not started more than once
1650 writable. (It's OK to run an image multiple times
1651 simultaneously in read-only mode.)
1653 * systemd-nspawn's --image= option is now capable of
1654 dissecting and booting MBR and GPT disk images that contain
1655 only a single active Linux partition. Previously it
1656 supported only GPT disk images with proper GPT type
1657 IDs. This allows running cloud images from major
1658 distributions directly with systemd-nspawn, without
1661 * In addition to collecting mouse dpi data in the udev
1662 hardware database, there's now support for collecting angle
1663 information for mouse scroll wheels. The database is
1664 supposed to guarantee similar scrolling behavior on mice
1665 that it knows about. There's also support for collecting
1666 information about Touchpad types.
1668 * udev's input_id built-in will now also collect touch screen
1669 dimension data and attach it to probed devices.
1671 * /etc/os-release gained support for a Distribution Privacy
1674 * networkd gained support for creating "ipvlan", "gretap",
1675 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1677 * systemd-tmpfiles gained support for "a" lines for setting
1680 * systemd-nspawn will now mount /tmp in the container to
1681 tmpfs, automatically.
1683 * systemd now exposes the memory.usage_in_bytes cgroup
1684 attribute and shows it for each service in the "systemctl
1685 status" output, if available.
1687 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1688 immediate reboot is triggered. This useful if shutdown is
1689 hung and is unable to complete, to expedite the
1690 operation. Note that this kind of reboot will still unmount
1691 all file systems, and hence should not result in fsck being
1694 * A .device unit for an optical block device will now be
1695 considered active only when a medium is in the drive. Also,
1696 mount units are now bound to their backing devices thus
1697 triggering automatic unmounting when devices become
1698 unavailable. With this in place systemd will now
1699 automatically unmount left-over mounts when a CD-ROM is
1700 ejected or an USB stick is yanked from the system.
1702 * networkd-wait-online now has support for waiting for
1703 specific interfaces only (with globbing), and for giving up
1704 after a configurable timeout.
1706 * networkd now exits when idle. It will be automatically
1707 restarted as soon as interfaces show up, are removed or
1708 change state. networkd will stay around as long as there is
1709 at least one DHCP state machine or similar around, that keep
1712 * networkd may now configure IPv6 link-local addressing in
1713 addition to IPv4 link-local addressing.
1715 * The IPv6 "token" for use in SLAAC may now be configured for
1716 each .network interface in networkd.
1718 * Routes configured with networkd may now be assigned a scope
1721 * networkd's [Match] sections now support globbing and lists
1722 of multiple space-separated matches per item.
1724 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1725 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1726 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1727 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1728 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1729 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1730 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1731 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1732 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1733 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1734 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1735 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1736 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1737 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1738 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1739 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1740 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1741 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1742 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1743 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1744 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1745 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1746 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1747 Hoffmann, Zbigniew Jędrzejewski-Szmek
1749 — Berlin, 2015-02-16
1753 * When querying unit file enablement status (for example via
1754 "systemctl is-enabled"), a new state "indirect" is now known
1755 which indicates that a unit might not be enabled itself, but
1756 another unit listed in its Also= setting might be.
1758 * Similar to the various existing ConditionXYZ= settings for
1759 units, there are now matching AssertXYZ= settings. While
1760 failing conditions cause a unit to be skipped, but its job
1761 to succeed, failing assertions declared like this will cause
1762 a unit start operation and its job to fail.
1764 * hostnamed now knows a new chassis type "embedded".
1766 * systemctl gained a new "edit" command. When used on a unit
1767 file, this allows extending unit files with .d/ drop-in
1768 configuration snippets or editing the full file (after
1769 copying it from /usr/lib to /etc). This will invoke the
1770 user's editor (as configured with $EDITOR), and reload the
1771 modified configuration after editing.
1773 * "systemctl status" now shows the suggested enablement state
1774 for a unit, as declared in the (usually vendor-supplied)
1775 system preset files.
1777 * nss-myhostname will now resolve the single-label host name
1778 "gateway" to the locally configured default IP routing
1779 gateways, ordered by their metrics. This assigns a stable
1780 name to the used gateways, regardless which ones are
1781 currently configured. Note that the name will only be
1782 resolved after all other name sources (if nss-myhostname is
1783 configured properly) and should hence not negatively impact
1784 systems that use the single-label host name "gateway" in
1787 * systemd-inhibit now allows filtering by mode when listing
1790 * Scope and service units gained a new "Delegate" boolean
1791 property, which, when set, allows processes running inside the
1792 unit to further partition resources. This is primarily
1793 useful for systemd user instances as well as container
1796 * journald will now pick up audit messages directly from
1797 the kernel, and log them like any other log message. The
1798 audit fields are split up and fully indexed. This means that
1799 journalctl in many ways is now a (nicer!) alternative to
1800 ausearch, the traditional audit client. Note that this
1801 implements only a minimal audit client. If you want the
1802 special audit modes like reboot-on-log-overflow, please use
1803 the traditional auditd instead, which can be used in
1804 parallel to journald.
1806 * The ConditionSecurity= unit file option now understands the
1807 special string "audit" to check whether auditing is
1810 * journalctl gained two new commands --vacuum-size= and
1811 --vacuum-time= to delete old journal files until the
1812 remaining ones take up no more than the specified size on disk,
1813 or are not older than the specified time.
1815 * A new, native PPPoE library has been added to sd-network,
1816 systemd's library of light-weight networking protocols. This
1817 library will be used in a future version of networkd to
1818 enable PPPoE communication without an external pppd daemon.
1820 * The busctl tool now understands a new "capture" verb that
1821 works similar to "monitor", but writes a packet capture
1822 trace to STDOUT that can be redirected to a file which is
1823 compatible with libcap's capture file format. This can then
1824 be loaded in Wireshark and similar tools to inspect bus
1827 * The busctl tool now understands a new "tree" verb that shows
1828 the object trees of a specific service on the bus, or of all
1831 * The busctl tool now understands a new "introspect" verb that
1832 shows all interfaces and members of objects on the bus,
1833 including their signature and values. This is particularly
1834 useful to get more information about bus objects shown by
1835 the new "busctl tree" command.
1837 * The busctl tool now understands new verbs "call",
1838 "set-property" and "get-property" for invoking bus method
1839 calls, setting and getting bus object properties in a
1842 * busctl gained a new --augment-creds= argument that controls
1843 whether the tool shall augment credential information it
1844 gets from the bus with data from /proc, in a possibly
1847 * nspawn's --link-journal= switch gained two new values
1848 "try-guest" and "try-host" that work like "guest" and
1849 "host", but do not fail if the host has no persistent
1850 journalling enabled. -j is now equivalent to
1851 --link-journal=try-guest.
1853 * macvlan network devices created by nspawn will now have
1854 stable MAC addresses.
1856 * A new SmackProcessLabel= unit setting has been added, which
1857 controls the SMACK security label processes forked off by
1858 the respective unit shall use.
1860 * If compiled with --enable-xkbcommon, systemd-localed will
1861 verify x11 keymap settings by compiling the given keymap. It
1862 will spew out warnings if the compilation fails. This
1863 requires libxkbcommon to be installed.
1865 * When a coredump is collected, a larger number of metadata
1866 fields is now collected and included in the journal records
1867 created for it. More specifically, control group membership,
1868 environment variables, memory maps, working directory,
1869 chroot directory, /proc/$PID/status, and a list of open file
1870 descriptors is now stored in the log entry.
1872 * The udev hwdb now contains DPI information for mice. For
1875 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1877 * All systemd programs that read standalone configuration
1878 files in /etc now also support a corresponding series of
1879 .conf.d configuration directories in /etc/, /run/,
1880 /usr/local/lib/, /usr/lib/, and (if configured with
1881 --enable-split-usr) /lib/. In particular, the following
1882 configuration files now have corresponding configuration
1883 directories: system.conf user.conf, logind.conf,
1884 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1885 resolved.conf, timesyncd.conf, journal-remote.conf, and
1886 journal-upload.conf. Note that distributions should use the
1887 configuration directories in /usr/lib/; the directories in
1888 /etc/ are reserved for the system administrator.
1890 * systemd-rfkill will no longer take the rfkill device name
1891 into account when storing rfkill state on disk, as the name
1892 might be dynamically assigned and not stable. Instead, the
1893 ID_PATH udev variable combined with the rfkill type (wlan,
1894 bluetooth, ...) is used.
1896 * A new service systemd-machine-id-commit.service has been
1897 added. When used on systems where /etc is read-only during
1898 boot, and /etc/machine-id is not initialized (but an empty
1899 file), this service will copy the temporary machine ID
1900 created as replacement into /etc after the system is fully
1901 booted up. This is useful for systems that are freshly
1902 installed with a non-initialized machine ID, but should get
1903 a fixed machine ID for subsequent boots.
1905 * networkd's .netdev files now provide a large set of
1906 configuration parameters for VXLAN devices. Similarly, the
1907 bridge port cost parameter is now configurable in .network
1908 files. There's also new support for configuring IP source
1909 routing. networkd .link files gained support for a new
1910 OriginalName= match that is useful to match against the
1911 original interface name the kernel assigned. .network files
1912 may include MTU= and MACAddress= fields for altering the MTU
1913 and MAC address while being connected to a specific network
1916 * The LUKS logic gained supported for configuring
1917 UUID-specific key files. There's also new support for naming
1918 LUKS device from the kernel command line, using the new
1919 luks.name= argument.
1921 * Timer units may now be transiently created via the bus API
1922 (this was previously already available for scope and service
1923 units). In addition it is now possible to create multiple
1924 transient units at the same time with a single bus call. The
1925 "systemd-run" tool has been updated to make use of this for
1926 running commands on a specified time, in at(1)-style.
1928 * tmpfiles gained support for "t" lines, for assigning
1929 extended attributes to files. Among other uses this may be
1930 used to assign SMACK labels to files.
1932 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1933 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1934 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1935 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1936 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1937 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1938 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1939 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1940 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1941 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1942 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1943 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1944 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1945 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1946 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1947 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1948 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1949 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1951 — Berlin, 2014-12-10
1955 * journalctl gained the new options -t/--identifier= to match
1956 on the syslog identifier (aka "tag"), as well as --utc to
1957 show log timestamps in the UTC timezone. journalctl now also
1958 accepts -n/--lines=all to disable line capping in a pager.
1960 * journalctl gained a new switch, --flush, that synchronously
1961 flushes logs from /run/log/journal to /var/log/journal if
1962 persistent storage is enabled. systemd-journal-flush.service
1963 now waits until the operation is complete.
1965 * Services can notify the manager before they start a reload
1966 (by sending RELOADING=1) or shutdown (by sending
1967 STOPPING=1). This allows the manager to track and show the
1968 internal state of daemons and closes a race condition when
1969 the process is still running but has closed its D-Bus
1972 * Services with Type=oneshot do not have to have any ExecStart
1975 * User units are now loaded also from
1976 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1977 /run/systemd/user directory that was already previously
1978 supported, but is under the control of the user.
1980 * Job timeouts (i.e. time-outs on the time a job that is
1981 queued stays in the run queue) can now optionally result in
1982 immediate reboot or power-off actions (JobTimeoutAction= and
1983 JobTimeoutRebootArgument=). This is useful on ".target"
1984 units, to limit the maximum time a target remains
1985 undispatched in the run queue, and to trigger an emergency
1986 operation in such a case. This is now used by default to
1987 turn off the system if boot-up (as defined by everything in
1988 basic.target) hangs and does not complete for at least
1989 15min. Also, if power-off or reboot hang for at least 30min
1990 an immediate power-off/reboot operation is triggered. This
1991 functionality is particularly useful to increase reliability
1992 on embedded devices, but also on laptops which might
1993 accidentally get powered on when carried in a backpack and
1994 whose boot stays stuck in a hard disk encryption passphrase
1997 * systemd-logind can be configured to also handle lid switch
1998 events even when the machine is docked or multiple displays
1999 are attached (HandleLidSwitchDocked= option).
2001 * A helper binary and a service have been added which can be
2002 used to resume from hibernation in the initramfs. A
2003 generator will parse the resume= option on the kernel
2004 command line to trigger resume.
2006 * A user console daemon systemd-consoled has been
2007 added. Currently, it is a preview, and will so far open a
2008 single terminal on each session of the user marked as
2009 Desktop=systemd-console.
2011 * Route metrics can be specified for DHCP routes added by
2014 * The SELinux context of socket-activated services can be set
2015 from the information provided by the networking stack
2016 (SELinuxContextFromNet= option).
2018 * Userspace firmware loading support has been removed and
2019 the minimum supported kernel version is thus bumped to 3.7.
2021 * Timeout for udev workers has been increased from 1 to 3
2022 minutes, but a warning will be printed after 1 minute to
2023 help diagnose kernel modules that take a long time to load.
2025 * Udev rules can now remove tags on devices with TAG-="foobar".
2027 * systemd's readahead implementation has been removed. In many
2028 circumstances it didn't give expected benefits even for
2029 rotational disk drives and was becoming less relevant in the
2030 age of SSDs. As none of the developers has been using
2031 rotating media anymore, and nobody stepped up to actively
2032 maintain this component of systemd it has now been removed.
2034 * Swap units can use Options= to specify discard options.
2035 Discard options specified for swaps in /etc/fstab are now
2038 * Docker containers are now detected as a separate type of
2041 * The Password Agent protocol gained support for queries where
2042 the user input is shown, useful e.g. for user names.
2043 systemd-ask-password gained a new --echo option to turn that
2046 * The default sysctl.d/ snippets will now set:
2048 net.core.default_qdisc = fq_codel
2050 This selects Fair Queuing Controlled Delay as the default
2051 queuing discipline for network interfaces. fq_codel helps
2052 fight the network bufferbloat problem. It is believed to be
2053 a good default with no tuning required for most workloads.
2054 Downstream distributions may override this choice. On 10Gbit
2055 servers that do not do forwarding, "fq" may perform better.
2056 Systems without a good clocksource should use "pfifo_fast".
2058 * If kdbus is enabled during build a new option BusPolicy= is
2059 available for service units, that allows locking all service
2060 processes into a stricter bus policy, in order to limit
2061 access to various bus services, or even hide most of them
2062 from the service's view entirely.
2064 * networkctl will now show the .network and .link file
2065 networkd has applied to a specific interface.
2067 * sd-login gained a new API call sd_session_get_desktop() to
2068 query which desktop environment has been selected for a
2071 * UNIX utmp support is now compile-time optional to support
2072 legacy-free systems.
2074 * systemctl gained two new commands "add-wants" and
2075 "add-requires" for pulling in units from specific targets
2078 * If the word "rescue" is specified on the kernel command line
2079 the system will now boot into rescue mode (aka
2080 rescue.target), which was previously available only by
2081 specifying "1" or "systemd.unit=rescue.target" on the kernel
2082 command line. This new kernel command line option nicely
2083 mirrors the already existing "emergency" kernel command line
2086 * New kernel command line options mount.usr=, mount.usrflags=,
2087 mount.usrfstype= have been added that match root=, rootflags=,
2088 rootfstype= but allow mounting a specific file system to
2091 * The $NOTIFY_SOCKET is now also passed to control processes of
2092 services, not only the main process.
2094 * This version reenables support for fsck's -l switch. This
2095 means at least version v2.25 of util-linux is required for
2096 operation, otherwise dead-locks on device nodes may
2097 occur. Again: you need to update util-linux to at least
2098 v2.25 when updating systemd to v217.
2100 * The "multi-seat-x" tool has been removed from systemd, as
2101 its functionality has been integrated into X servers 1.16,
2102 and the tool is hence redundant. It is recommended to update
2103 display managers invoking this tool to simply invoke X
2104 directly from now on, again.
2106 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2107 message flag has been added for all of systemd's PolicyKit
2108 authenticated method calls has been added. In particular
2109 this now allows optional interactive authorization via
2110 PolicyKit for many of PID1's privileged operations such as
2111 unit file enabling and disabling.
2113 * "udevadm hwdb --update" learnt a new switch "--usr" for
2114 placing the rebuilt hardware database in /usr instead of
2115 /etc. When used only hardware database entries stored in
2116 /usr will be used, and any user database entries in /etc are
2117 ignored. This functionality is useful for vendors to ship a
2118 pre-built database on systems where local configuration is
2119 unnecessary or unlikely.
2121 * Calendar time specifications in .timer units now also
2122 understand the strings "semi-annually", "quarterly" and
2123 "minutely" as shortcuts (in addition to the preexisting
2124 "anually", "hourly", ...).
2126 * systemd-tmpfiles will now correctly create files in /dev
2127 at boot which are marked for creation only at boot. It is
2128 recommended to always create static device nodes with 'c!'
2129 and 'b!', so that they are created only at boot and not
2130 overwritten at runtime.
2132 * When the watchdog logic is used for a service (WatchdogSec=)
2133 and the watchdog timeout is hit the service will now be
2134 terminated with SIGABRT (instead of just SIGTERM), in order
2135 to make sure a proper coredump and backtrace is
2136 generated. This ensures that hanging services will result in
2137 similar coredump/backtrace behaviour as services that hit a
2140 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2141 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2142 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2143 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2144 Herrmann, David Sommerseth, David Strauss, Emil Renner
2145 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2146 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2147 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2148 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2149 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2150 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2151 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2152 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2153 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2154 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2155 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2156 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2157 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2158 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2159 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2160 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2163 — Berlin, 2014-10-28
2167 * timedated no longer reads NTP implementation unit names from
2168 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2169 implementations should add a
2171 Conflicts=systemd-timesyncd.service
2173 to their unit files to take over and replace systemd's NTP
2174 default functionality.
2176 * systemd-sysusers gained a new line type "r" for configuring
2177 which UID/GID ranges to allocate system users/groups
2178 from. Lines of type "u" may now add an additional column
2179 that specifies the home directory for the system user to be
2180 created. Also, systemd-sysusers may now optionally read user
2181 information from STDIN instead of a file. This is useful for
2182 invoking it from RPM preinst scriptlets that need to create
2183 users before the first RPM file is installed since these
2184 files might need to be owned by them. A new
2185 %sysusers_create_inline RPM macro has been introduced to do
2186 just that. systemd-sysusers now updates the shadow files as
2187 well as the user/group databases, which should enhance
2188 compatibility with certain tools like grpck.
2190 * A number of bus APIs of PID 1 now optionally consult
2191 PolicyKit to permit access for otherwise unprivileged
2192 clients under certain conditions. Note that this currently
2193 doesn't support interactive authentication yet, but this is
2194 expected to be added eventually, too.
2196 * /etc/machine-info now has new fields for configuring the
2197 deployment environment of the machine, as well as the
2198 location of the machine. hostnamectl has been updated with
2199 new command to update these fields.
2201 * systemd-timesyncd has been updated to automatically acquire
2202 NTP server information from systemd-networkd, which might
2203 have been discovered via DHCP.
2205 * systemd-resolved now includes a caching DNS stub resolver
2206 and a complete LLMNR name resolution implementation. A new
2207 NSS module "nss-resolve" has been added which can be used
2208 instead of glibc's own "nss-dns" to resolve hostnames via
2209 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2210 be resolved via systemd-resolved D-Bus APIs. In contrast to
2211 the glibc internal resolver systemd-resolved is aware of
2212 multi-homed system, and keeps DNS server and caches separate
2213 and per-interface. Queries are sent simultaneously on all
2214 interfaces that have DNS servers configured, in order to
2215 properly handle VPNs and local LANs which might resolve
2216 separate sets of domain names. systemd-resolved may acquire
2217 DNS server information from systemd-networkd automatically,
2218 which in turn might have discovered them via DHCP. A tool
2219 "systemd-resolve-host" has been added that may be used to
2220 query the DNS logic in resolved. systemd-resolved implements
2221 IDNA and automatically uses IDNA or UTF-8 encoding depending
2222 on whether classic DNS or LLMNR is used as transport. In the
2223 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2224 implementation to systemd-resolved.
2226 * A new NSS module nss-mymachines has been added, that
2227 automatically resolves the names of all local registered
2228 containers to their respective IP addresses.
2230 * A new client tool "networkctl" for systemd-networkd has been
2231 added. It currently is entirely passive and will query
2232 networking configuration from udev, rtnetlink and networkd,
2233 and present it to the user in a very friendly
2234 way. Eventually, we hope to extend it to become a full
2235 control utility for networkd.
2237 * .socket units gained a new DeferAcceptSec= setting that
2238 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2239 TCP. Similarly, support for controlling TCP keep-alive
2240 settings has been added (KeepAliveTimeSec=,
2241 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2242 turning off Nagle's algorithm on TCP has been added
2245 * logind learned a new session type "web", for use in projects
2246 like Cockpit which register web clients as PAM sessions.
2248 * timer units with at least one OnCalendar= setting will now
2249 be started only after timer-sync.target has been
2250 reached. This way they will not elapse before the system
2251 clock has been corrected by a local NTP client or
2252 similar. This is particular useful on RTC-less embedded
2253 machines, that come up with an invalid system clock.
2255 * systemd-nspawn's --network-veth= switch should now result in
2256 stable MAC addresses for both the outer and the inner side
2259 * systemd-nspawn gained a new --volatile= switch for running
2260 container instances with /etc or /var unpopulated.
2262 * The kdbus client code has been updated to use the new Linux
2263 3.17 memfd subsystem instead of the old kdbus-specific one.
2265 * systemd-networkd's DHCP client and server now support
2266 FORCERENEW. There are also new configuration options to
2267 configure the vendor client identifier and broadcast mode
2270 * systemd will no longer inform the kernel about the current
2271 timezone, as this is necessarily incorrect and racy as the
2272 kernel has no understanding of DST and similar
2273 concepts. This hence means FAT timestamps will be always
2274 considered UTC, similar to what Android is already
2275 doing. Also, when the RTC is configured to the local time
2276 (rather than UTC) systemd will never synchronize back to it,
2277 as this might confuse Windows at a later boot.
2279 * systemd-analyze gained a new command "verify" for offline
2280 validation of unit files.
2282 * systemd-networkd gained support for a couple of additional
2283 settings for bonding networking setups. Also, the metric for
2284 statically configured routes may now be configured. For
2285 network interfaces where this is appropriate the peer IP
2286 address may now be configured.
2288 * systemd-networkd's DHCP client will no longer request
2289 broadcasting by default, as this tripped up some networks.
2290 For hardware where broadcast is required the feature should
2291 be switched back on using RequestBroadcast=yes.
2293 * systemd-networkd will now set up IPv4LL addresses (when
2294 enabled) even if DHCP is configured successfully.
2296 * udev will now default to respect network device names given
2297 by the kernel when the kernel indicates that these are
2298 predictable. This behavior can be tweaked by changing
2299 NamePolicy= in the relevant .link file.
2301 * A new library systemd-terminal has been added that
2302 implements full TTY stream parsing and rendering. This
2303 library is supposed to be used later on for implementing a
2304 full userspace VT subsystem, replacing the current kernel
2307 * A new tool systemd-journal-upload has been added to push
2308 journal data to a remote system running
2309 systemd-journal-remote.
2311 * journald will no longer forward all local data to another
2312 running syslog daemon. This change has been made because
2313 rsyslog (which appears to be the most commonly used syslog
2314 implementation these days) no longer makes use of this, and
2315 instead pulls the data out of the journal on its own. Since
2316 forwarding the messages to a non-existent syslog server is
2317 more expensive than we assumed we have now turned this
2318 off. If you run a syslog server that is not a recent rsyslog
2319 version, you have to turn this option on again
2320 (ForwardToSyslog= in journald.conf).
2322 * journald now optionally supports the LZ4 compressor for
2323 larger journal fields. This compressor should perform much
2324 better than XZ which was the previous default.
2326 * machinectl now shows the IP addresses of local containers,
2327 if it knows them, plus the interface name of the container.
2329 * A new tool "systemd-escape" has been added that makes it
2330 easy to escape strings to build unit names and similar.
2332 * sd_notify() messages may now include a new ERRNO= field
2333 which is parsed and collected by systemd and shown among the
2334 "systemctl status" output for a service.
2336 * A new component "systemd-firstboot" has been added that
2337 queries the most basic systemd information (timezone,
2338 hostname, root password) interactively on first
2339 boot. Alternatively it may also be used to provision these
2340 things offline on OS images installed into directories.
2342 * The default sysctl.d/ snippets will now set
2344 net.ipv4.conf.default.promote_secondaries=1
2346 This has the benefit of no flushing secondary IP addresses
2347 when primary addresses are removed.
2349 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2350 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2351 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2352 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2353 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2354 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2355 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2356 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2357 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2358 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2359 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2360 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2361 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2362 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2363 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2365 — Berlin, 2014-08-19
2369 * A new tool systemd-sysusers has been added. This tool
2370 creates system users and groups in /etc/passwd and
2371 /etc/group, based on static declarative system user/group
2372 definitions in /usr/lib/sysusers.d/. This is useful to
2373 enable factory resets and volatile systems that boot up with
2374 an empty /etc directory, and thus need system users and
2375 groups created during early boot. systemd now also ships
2376 with two default sysusers.d/ files for the most basic
2377 users and groups systemd and the core operating system
2380 * A new tmpfiles snippet has been added that rebuilds the
2381 essential files in /etc on boot, should they be missing.
2383 * A directive for ensuring automatic clean-up of
2384 /var/cache/man/ has been removed from the default
2385 configuration. This line should now be shipped by the man
2386 implementation. The necessary change has been made to the
2387 man-db implementation. Note that you need to update your man
2388 implementation to one that ships this line, otherwise no
2389 automatic clean-up of /var/cache/man will take place.
2391 * A new condition ConditionNeedsUpdate= has been added that
2392 may conditionalize services to only run when /etc or /var
2393 are "older" than the vendor operating system resources in
2394 /usr. This is useful for reconstructing or updating /etc
2395 after an offline update of /usr or a factory reset, on the
2396 next reboot. Services that want to run once after such an
2397 update or reset should use this condition and order
2398 themselves before the new systemd-update-done.service, which
2399 will mark the two directories as fully updated. A number of
2400 service files have been added making use of this, to rebuild
2401 the udev hardware database, the journald message catalog and
2402 dynamic loader cache (ldconfig). The systemd-sysusers tool
2403 described above also makes use of this now. With this in
2404 place it is now possible to start up a minimal operating
2405 system with /etc empty cleanly. For more information on the
2406 concepts involved see this recent blog story:
2408 http://0pointer.de/blog/projects/stateless.html
2410 * A new system group "input" has been introduced, and all
2411 input device nodes get this group assigned. This is useful
2412 for system-level software to get access to input devices. It
2413 complements what is already done for "audio" and "video".
2415 * systemd-networkd learnt minimal DHCPv4 server support in
2416 addition to the existing DHCPv4 client support. It also
2417 learnt DHCPv6 client and IPv6 Router Solicitation client
2418 support. The DHCPv4 client gained support for static routes
2419 passed in from the server. Note that the [DHCPv4] section
2420 known in older systemd-networkd versions has been renamed to
2421 [DHCP] and is now also used by the DHCPv6 client. Existing
2422 .network files using settings of this section should be
2423 updated, though compatibility is maintained. Optionally, the
2424 client hostname may now be sent to the DHCP server.
2426 * networkd gained support for vxlan virtual networks as well
2427 as tun/tap and dummy devices.
2429 * networkd gained support for automatic allocation of address
2430 ranges for interfaces from a system-wide pool of
2431 addresses. This is useful for dynamically managing a large
2432 number of interfaces with a single network configuration
2433 file. In particular this is useful to easily assign
2434 appropriate IP addresses to the veth links of a large number
2435 of nspawn instances.
2437 * RPM macros for processing sysusers, sysctl and binfmt
2438 drop-in snippets at package installation time have been
2441 * The /etc/os-release file should now be placed in
2442 /usr/lib/os-release. The old location is automatically
2443 created as symlink. /usr/lib is the more appropriate
2444 location of this file, since it shall actually describe the
2445 vendor operating system shipped in /usr, and not the
2446 configuration stored in /etc.
2448 * .mount units gained a new boolean SloppyOptions= setting
2449 that maps to mount(8)'s -s option which enables permissive
2450 parsing of unknown mount options.
2452 * tmpfiles learnt a new "L+" directive which creates a symlink
2453 but (unlike "L") deletes a pre-existing file first, should
2454 it already exist and not already be the correct
2455 symlink. Similarly, "b+", "c+" and "p+" directives have been
2456 added as well, which create block and character devices, as
2457 well as fifos in the filesystem, possibly removing any
2458 pre-existing files of different types.
2460 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2461 'argument' field (which so far specified the source to
2462 symlink/copy the files from) is now optional. If omitted the
2463 same file os copied from /usr/share/factory/ suffixed by the
2464 full destination path. This is useful for populating /etc
2465 with essential files, by copying them from vendor defaults
2466 shipped in /usr/share/factory/etc.
2468 * A new command "systemctl preset-all" has been added that
2469 applies the service preset settings to all installed unit
2470 files. A new switch --preset-mode= has been added that
2471 controls whether only enable or only disable operations
2474 * A new command "systemctl is-system-running" has been added
2475 that allows checking the overall state of the system, for
2476 example whether it is fully up and running.
2478 * When the system boots up with an empty /etc, the equivalent
2479 to "systemctl preset-all" is executed during early boot, to
2480 make sure all default services are enabled after a factory
2483 * systemd now contains a minimal preset file that enables the
2484 most basic services systemd ships by default.
2486 * Unit files' [Install] section gained a new DefaultInstance=
2487 field for defining the default instance to create if a
2488 template unit is enabled with no instance specified.
2490 * A new passive target cryptsetup-pre.target has been added
2491 that may be used by services that need to make they run and
2492 finish before the first LUKS cryptographic device is set up.
2494 * The /dev/loop-control and /dev/btrfs-control device nodes
2495 are now owned by the "disk" group by default, opening up
2496 access to this group.
2498 * systemd-coredump will now automatically generate a
2499 stack trace of all core dumps taking place on the system,
2500 based on elfutils' libdw library. This stack trace is logged
2503 * systemd-coredump may now optionally store coredumps directly
2504 on disk (in /var/lib/systemd/coredump, possibly compressed),
2505 instead of storing them unconditionally in the journal. This
2506 mode is the new default. A new configuration file
2507 /etc/systemd/coredump.conf has been added to configure this
2508 and other parameters of systemd-coredump.
2510 * coredumpctl gained a new "info" verb to show details about a
2511 specific coredump. A new switch "-1" has also been added
2512 that makes sure to only show information about the most
2513 recent entry instead of all entries. Also, as the tool is
2514 generally useful now the "systemd-" prefix of the binary
2515 name has been removed. Distributions that want to maintain
2516 compatibility with the old name should add a symlink from
2517 the old name to the new name.
2519 * journald's SplitMode= now defaults to "uid". This makes sure
2520 that unprivileged users can access their own coredumps with
2521 coredumpctl without restrictions.
2523 * New kernel command line options "systemd.wants=" (for
2524 pulling an additional unit during boot), "systemd.mask="
2525 (for masking a specific unit for the boot), and
2526 "systemd.debug-shell" (for enabling the debug shell on tty9)
2527 have been added. This is implemented in the new generator
2528 "systemd-debug-generator".
2530 * systemd-nspawn will now by default filter a couple of
2531 syscalls for containers, among them those required for
2532 kernel module loading, direct x86 IO port access, swap
2533 management, and kexec. Most importantly though
2534 open_by_handle_at() is now prohibited for containers,
2535 closing a hole similar to a recently discussed vulnerability
2536 in docker regarding access to files on file hierarchies the
2537 container should normally not have access to. Note that, for
2538 nspawn, we generally make no security claims anyway (and
2539 this is explicitly documented in the man page), so this is
2540 just a fix for one of the most obvious problems.
2542 * A new man page file-hierarchy(7) has been added that
2543 contains a minimized, modernized version of the file system
2544 layout systemd expects, similar in style to the FHS
2545 specification or hier(5). A new tool systemd-path(1) has
2546 been added to query many of these paths for the local
2549 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2550 longer done. Since the directory now has a per-user size
2551 limit, and is cleaned on logout this appears unnecessary,
2552 in particular since this now brings the lifecycle of this
2553 directory closer in line with how IPC objects are handled.
2555 * systemd.pc now exports a number of additional directories,
2556 including $libdir (which is useful to identify the library
2557 path for the primary architecture of the system), and a
2558 couple of drop-in directories.
2560 * udev's predictable network interface names now use the dev_port
2561 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2562 distinguish between ports of the same PCI function. dev_id should
2563 only be used for ports using the same HW address, hence the need
2566 * machined has been updated to export the OS version of a
2567 container (read from /etc/os-release and
2568 /usr/lib/os-release) on the bus. This is now shown in
2569 "machinectl status" for a machine.
2571 * A new service setting RestartForceExitStatus= has been
2572 added. If configured to a set of exit signals or process
2573 return values, the service will be restarted when the main
2574 daemon process exits with any of them, regardless of the
2577 * systemctl's -H switch for connecting to remote systemd
2578 machines has been extended so that it may be used to
2579 directly connect to a specific container on the
2580 host. "systemctl -H root@foobar:waldi" will now connect as
2581 user "root" to host "foobar", and then proceed directly to
2582 the container named "waldi". Note that currently you have to
2583 authenticate as user "root" for this to work, as entering
2584 containers is a privileged operation.
2586 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2587 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2588 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2589 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2590 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2591 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2592 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2593 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2594 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2595 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2596 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2597 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2599 — Berlin, 2014-07-03
2603 * As an experimental feature, udev now tries to lock the
2604 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2605 executes events for the disk or any of its partitions.
2606 Applications like partitioning programs can lock the
2607 disk device node (flock(LOCK_EX)) and claim temporary
2608 device ownership that way; udev will entirely skip all event
2609 handling for this disk and its partitions. If the disk
2610 was opened for writing, the close will trigger a partition
2611 table rescan in udev's "watch" facility, and if needed
2612 synthesize "change" events for the disk and all its partitions.
2613 This is now unconditionally enabled, and if it turns out to
2614 cause major problems, we might turn it on only for specific
2615 devices, or might need to disable it entirely. Device Mapper
2616 devices are excluded from this logic.
2618 * We temporarily dropped the "-l" switch for fsck invocations,
2619 since they collide with the flock() logic above. util-linux
2620 upstream has been changed already to avoid this conflict,
2621 and we will readd "-l" as soon as util-linux with this
2622 change has been released.
2624 * The dependency on libattr has been removed. Since a long
2625 time, the extended attribute calls have moved to glibc, and
2626 libattr is thus unnecessary.
2628 * Virtualization detection works without priviliges now. This
2629 means the systemd-detect-virt binary no longer requires
2630 CAP_SYS_PTRACE file capabilities, and our daemons can run
2631 with fewer privileges.
2633 * systemd-networkd now runs under its own "systemd-network"
2634 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2635 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2636 loses the ability to write to files owned by root this way.
2638 * Similarly, systemd-resolved now runs under its own
2639 "systemd-resolve" user with no capabilities remaining.
2641 * Similarly, systemd-bus-proxyd now runs under its own
2642 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2644 * systemd-networkd gained support for setting up "veth"
2645 virtual Ethernet devices for container connectivity, as well
2646 as GRE and VTI tunnels.
2648 * systemd-networkd will no longer automatically attempt to
2649 manually load kernel modules necessary for certain tunnel
2650 transports. Instead, it is assumed the kernel loads them
2651 automatically when required. This only works correctly on
2652 very new kernels. On older kernels, please consider adding
2653 the kernel modules to /etc/modules-load.d/ as a work-around.
2655 * The resolv.conf file systemd-resolved generates has been
2656 moved to /run/systemd/resolve/. If you have a symlink from
2657 /etc/resolv.conf, it might be necessary to correct it.
2659 * Two new service settings, ProtectHome= and ProtectSystem=,
2660 have been added. When enabled, they will make the user data
2661 (such as /home) inaccessible or read-only and the system
2662 (such as /usr) read-only, for specific services. This allows
2663 very light-weight per-service sandboxing to avoid
2664 modifications of user data or system files from
2665 services. These two new switches have been enabled for all
2666 of systemd's long-running services, where appropriate.
2668 * Socket units gained new SocketUser= and SocketGroup=
2669 settings to set the owner user and group of AF_UNIX sockets
2670 and FIFOs in the file system.
2672 * Socket units gained a new RemoveOnStop= setting. If enabled,
2673 all FIFOS and sockets in the file system will be removed
2674 when the specific socket unit is stopped.
2676 * Socket units gained a new Symlinks= setting. It takes a list
2677 of symlinks to create to file system sockets or FIFOs
2678 created by the specific Unix sockets. This is useful to
2679 manage symlinks to socket nodes with the same life-cycle as
2682 * The /dev/log socket and /dev/initctl FIFO have been moved to
2683 /run, and have been replaced by symlinks. This allows
2684 connecting to these facilities even if PrivateDevices=yes is
2685 used for a service (which makes /dev/log itself unavailable,
2686 but /run is left). This also has the benefit of ensuring
2687 that /dev only contains device nodes, directories and
2688 symlinks, and nothing else.
2690 * sd-daemon gained two new calls sd_pid_notify() and
2691 sd_pid_notifyf(). They are similar to sd_notify() and
2692 sd_notifyf(), but allow overriding of the source PID of
2693 notification messages if permissions permit this. This is
2694 useful to send notify messages on behalf of a different
2695 process (for example, the parent process). The
2696 systemd-notify tool has been updated to make use of this
2697 when sending messages (so that notification messages now
2698 originate from the shell script invoking systemd-notify and
2699 not the systemd-notify process itself. This should minimize
2700 a race where systemd fails to associate notification
2701 messages to services when the originating process already
2704 * A new "on-abnormal" setting for Restart= has been added. If
2705 set, it will result in automatic restarts on all "abnormal"
2706 reasons for a process to exit, which includes unclean
2707 signals, core dumps, timeouts and watchdog timeouts, but
2708 does not include clean and unclean exit codes or clean
2709 signals. Restart=on-abnormal is an alternative for
2710 Restart=on-failure for services that shall be able to
2711 terminate and avoid restarts on certain errors, by
2712 indicating so with an unclean exit code. Restart=on-failure
2713 or Restart=on-abnormal is now the recommended setting for
2714 all long-running services.
2716 * If the InaccessibleDirectories= service setting points to a
2717 mount point (or if there are any submounts contained within
2718 it), it is now attempted to completely unmount it, to make
2719 the file systems truly unavailable for the respective
2722 * The ReadOnlyDirectories= service setting and
2723 systemd-nspawn's --read-only parameter are now recursively
2724 applied to all submounts, too.
2726 * Mount units may now be created transiently via the bus APIs.
2728 * The support for SysV and LSB init scripts has been removed
2729 from the systemd daemon itself. Instead, it is now
2730 implemented as a generator that creates native systemd units
2731 from these scripts when needed. This enables us to remove a
2732 substantial amount of legacy code from PID 1, following the
2733 fact that many distributions only ship a very small number
2734 of LSB/SysV init scripts nowadays.
2736 * Privileged Xen (dom0) domains are not considered
2737 virtualization anymore by the virtualization detection
2738 logic. After all, they generally have unrestricted access to
2739 the hardware and usually are used to manage the unprivileged
2742 * systemd-tmpfiles gained a new "C" line type, for copying
2743 files or entire directories.
2745 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2746 lines. So far, they have been non-globbing versions of the
2747 latter, and have thus been redundant. In future, it is
2748 recommended to only use "z". "m" has hence been removed
2749 from the documentation, even though it stays supported.
2751 * A tmpfiles snippet to recreate the most basic structure in
2752 /var has been added. This is enough to create the /var/run →
2753 /run symlink and create a couple of structural
2754 directories. This allows systems to boot up with an empty or
2755 volatile /var. Of course, while with this change, the core OS
2756 now is capable with dealing with a volatile /var, not all
2757 user services are ready for it. However, we hope that sooner
2758 or later, many service daemons will be changed upstream so
2759 that they are able to automatically create their necessary
2760 directories in /var at boot, should they be missing. This is
2761 the first step to allow state-less systems that only require
2762 the vendor image for /usr to boot.
2764 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2765 empty tmpfs instance to a specific directory. This is
2766 particularly useful for making use of the automatic
2767 reconstruction of /var (see above), by passing --tmpfs=/var.
2769 * Access modes specified in tmpfiles snippets may now be
2770 prefixed with "~", which indicates that they shall be masked
2771 by whether the existing file or directory is currently
2772 writable, readable or executable at all. Also, if specified,
2773 the sgid/suid/sticky bits will be masked for all
2776 * A new passive target unit "network-pre.target" has been
2777 added which is useful for services that shall run before any
2778 network is configured, for example firewall scripts.
2780 * The "floppy" group that previously owned the /dev/fd*
2781 devices is no longer used. The "disk" group is now used
2782 instead. Distributions should probably deprecate usage of
2785 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2786 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2787 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2788 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2789 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2790 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2793 — Berlin, 2014-06-11
2797 * A new "systemd-timesyncd" daemon has been added for
2798 synchronizing the system clock across the network. It
2799 implements an SNTP client. In contrast to NTP
2800 implementations such as chrony or the NTP reference server,
2801 this only implements a client side, and does not bother with
2802 the full NTP complexity, focusing only on querying time from
2803 one remote server and synchronizing the local clock to
2804 it. Unless you intend to serve NTP to networked clients or
2805 want to connect to local hardware clocks, this simple NTP
2806 client should be more than appropriate for most
2807 installations. The daemon runs with minimal privileges, and
2808 has been hooked up with networkd to only operate when
2809 network connectivity is available. The daemon saves the
2810 current clock to disk every time a new NTP sync has been
2811 acquired, and uses this to possibly correct the system clock
2812 early at bootup, in order to accommodate for systems that
2813 lack an RTC such as the Raspberry Pi and embedded devices,
2814 and to make sure that time monotonically progresses on these
2815 systems, even if it is not always correct. To make use of
2816 this daemon, a new system user and group "systemd-timesync"
2817 needs to be created on installation of systemd.
2819 * The queue "seqnum" interface of libudev has been disabled, as
2820 it was generally incompatible with device namespacing as
2821 sequence numbers of devices go "missing" if the devices are
2822 part of a different namespace.
2824 * "systemctl list-timers" and "systemctl list-sockets" gained
2825 a --recursive switch for showing units of these types also
2826 for all local containers, similar in style to the already
2827 supported --recursive switch for "systemctl list-units".
2829 * A new RebootArgument= setting has been added for service
2830 units, which may be used to specify a kernel reboot argument
2831 to use when triggering reboots with StartLimitAction=.
2833 * A new FailureAction= setting has been added for service
2834 units which may be used to specify an operation to trigger
2835 when a service fails. This works similarly to
2836 StartLimitAction=, but unlike it, controls what is done
2837 immediately rather than only after several attempts to
2838 restart the service in question.
2840 * hostnamed got updated to also expose the kernel name,
2841 release, and version on the bus. This is useful for
2842 executing commands like hostnamectl with the -H switch.
2843 systemd-analyze makes use of this to properly display
2844 details when running non-locally.
2846 * The bootchart tool can now show cgroup information in the
2847 graphs it generates.
2849 * The CFS CPU quota cgroup attribute is now exposed for
2850 services. The new CPUQuota= switch has been added for this
2851 which takes a percentage value. Setting this will have the
2852 result that a service may never get more CPU time than the
2853 specified percentage, even if the machine is otherwise idle.
2855 * systemd-networkd learned IPIP and SIT tunnel support.
2857 * LSB init scripts exposing a dependency on $network will now
2858 get a dependency on network-online.target rather than simply
2859 network.target. This should bring LSB handling closer to
2860 what it was on SysV systems.
2862 * A new fsck.repair= kernel option has been added to control
2863 how fsck shall deal with unclean file systems at boot.
2865 * The (.ini) configuration file parser will now silently
2866 ignore sections whose name begins with "X-". This may be
2867 used to maintain application-specific extension sections in unit
2870 * machined gained a new API to query the IP addresses of
2871 registered containers. "machinectl status" has been updated
2872 to show these addresses in its output.
2874 * A new call sd_uid_get_display() has been added to the
2875 sd-login APIs for querying the "primary" session of a
2876 user. The "primary" session of the user is elected from the
2877 user's sessions and generally a graphical session is
2878 preferred over a text one.
2880 * A minimal systemd-resolved daemon has been added. It
2881 currently simply acts as a companion to systemd-networkd and
2882 manages resolv.conf based on per-interface DNS
2883 configuration, possibly supplied via DHCP. In the long run
2884 we hope to extend this into a local DNSSEC enabled DNS and
2887 * The systemd-networkd-wait-online tool is now enabled by
2888 default. It will delay network-online.target until a network
2889 connection has been configured. The tool primarily integrates
2890 with networkd, but will also make a best effort to make sense
2891 of network configuration performed in some other way.
2893 * Two new service options StartupCPUShares= and
2894 StartupBlockIOWeight= have been added that work similarly to
2895 CPUShares= and BlockIOWeight= however only apply during
2896 system startup. This is useful to prioritize certain services
2897 differently during bootup than during normal runtime.
2899 * hostnamed has been changed to prefer the statically
2900 configured hostname in /etc/hostname (unless set to
2901 'localhost' or empty) over any dynamic one supplied by
2902 dhcp. With this change, the rules for picking the hostname
2903 match more closely the rules of other configuration settings
2904 where the local administrator's configuration in /etc always
2905 overrides any other settings.
2907 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2908 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2909 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2910 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2911 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2912 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2913 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2914 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2915 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2916 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2917 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2918 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2919 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2920 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2921 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2922 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2925 — Beijing, 2014-05-28
2929 * When restoring the screen brightness at boot, stay away from
2930 the darkest setting or from the lowest 5% of the available
2931 range, depending on which is the larger value of both. This
2932 should effectively protect the user from rebooting into a
2933 black screen, should the brightness have been set to minimum
2936 * sd-login gained a new sd_machine_get_class() call to
2937 determine the class ("vm" or "container") of a machine
2938 registered with machined.
2940 * sd-login gained new calls
2941 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2942 to query the identity of the peer of a local AF_UNIX
2943 connection. They operate similarly to their sd_pid_get_xyz()
2946 * PID 1 will now maintain a system-wide system state engine
2947 with the states "starting", "running", "degraded",
2948 "maintenance", "stopping". These states are bound to system
2949 startup, normal runtime, runtime with at least one failed
2950 service, rescue/emergency mode and system shutdown. This
2951 state is shown in the "systemctl status" output when no unit
2952 name is passed. It is useful to determine system state, in
2953 particularly when doing so for many systems or containers at
2956 * A new command "list-machines" has been added to "systemctl"
2957 that lists all local OS containers and shows their system
2958 state (see above), if systemd runs inside of them.
2960 * systemctl gained a new "-r" switch to recursively enumerate
2961 units on all local containers, when used with the
2962 "list-unit" command (which is the default one that is
2963 executed when no parameters are specified).
2965 * The GPT automatic partition discovery logic will now honour
2966 two GPT partition flags: one may be set on a partition to
2967 cause it to be mounted read-only, and the other may be set
2968 on a partition to ignore it during automatic discovery.
2970 * Two new GPT type UUIDs have been added for automatic root
2971 partition discovery, for 32-bit and 64-bit ARM. This is not
2972 particularly useful for discovering the root directory on
2973 these architectures during bare-metal boots (since UEFI is
2974 not common there), but still very useful to allow booting of
2975 ARM disk images in nspawn with the -i option.
2977 * MAC addresses of interfaces created with nspawn's
2978 --network-interface= switch will now be generated from the
2979 machine name, and thus be stable between multiple invocations
2982 * logind will now automatically remove all IPC objects owned
2983 by a user if she or he fully logs out. This makes sure that
2984 users who are logged out cannot continue to consume IPC
2985 resources. This covers SysV memory, semaphores and message
2986 queues as well as POSIX shared memory and message
2987 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2988 limits. With this functionality, that is corrected. This may
2989 be turned off by using the RemoveIPC= switch of logind.conf.
2991 * The systemd-machine-id-setup and tmpfiles tools gained a
2992 --root= switch to operate on a specific root directory,
2995 * journald can now forward logged messages to the TTYs of all
2996 logged in users ("wall"). This is the default for all
2997 emergency messages now.
2999 * A new tool systemd-journal-remote has been added to stream
3000 journal log messages across the network.
3002 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3003 controller trees are mounted into it. Note that the
3004 directories mounted beneath it are not read-only. This is a
3005 security measure and is particularly useful because glibc
3006 actually includes a search logic to pick any tmpfs it can
3007 find to implement shm_open() if /dev/shm is not available
3008 (which it might very well be in namespaced setups).
3010 * machinectl gained a new "poweroff" command to cleanly power
3011 down a local OS container.
3013 * The PrivateDevices= unit file setting will now also drop the
3014 CAP_MKNOD capability from the capability bound set, and
3015 imply DevicePolicy=closed.
3017 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3018 comprehensively on all long-running systemd services where
3019 this is appropriate.
3021 * systemd-udevd will now run in a disassociated mount
3022 namespace. To mount directories from udev rules, make sure to
3023 pull in mount units via SYSTEMD_WANTS properties.
3025 * The kdbus support gained support for uploading policy into
3026 the kernel. sd-bus gained support for creating "monitoring"
3027 connections that can eavesdrop into all bus communication
3028 for debugging purposes.
3030 * Timestamps may now be specified in seconds since the UNIX
3031 epoch Jan 1st, 1970 by specifying "@" followed by the value
3034 * Native tcpwrap support in systemd has been removed. tcpwrap
3035 is old code, not really maintained anymore and has serious
3036 shortcomings, and better options such as firewalls
3037 exist. For setups that require tcpwrap usage, please
3038 consider invoking your socket-activated service via tcpd,
3039 like on traditional inetd.
3041 * A new system.conf configuration option
3042 DefaultTimerAccuracySec= has been added that controls the
3043 default AccuracySec= setting of .timer units.
3045 * Timer units gained a new WakeSystem= switch. If enabled,
3046 timers configured this way will cause the system to resume
3047 from system suspend (if the system supports that, which most
3050 * Timer units gained a new Persistent= switch. If enabled,
3051 timers configured this way will save to disk when they have
3052 been last triggered. This information is then used on next
3053 reboot to possible execute overdue timer events, that
3054 could not take place because the system was powered off.
3055 This enables simple anacron-like behaviour for timer units.
3057 * systemctl's "list-timers" will now also list the time a
3058 timer unit was last triggered in addition to the next time
3059 it will be triggered.
3061 * systemd-networkd will now assign predictable IPv4LL
3062 addresses to its local interfaces.
3064 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3065 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3066 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3067 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3068 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3069 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3070 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3071 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3074 — Berlin, 2014-03-25
3078 * A new unit file setting RestrictAddressFamilies= has been
3079 added to restrict which socket address families unit
3080 processes gain access to. This takes address family names
3081 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3082 attack surface of services via exotic protocol stacks. This
3083 is built on seccomp system call filters.
3085 * Two new unit file settings RuntimeDirectory= and
3086 RuntimeDirectoryMode= have been added that may be used to
3087 manage a per-daemon runtime directories below /run. This is
3088 an alternative for setting up directory permissions with
3089 tmpfiles snippets, and has the advantage that the runtime
3090 directory's lifetime is bound to the daemon runtime and that
3091 the daemon starts up with an empty directory each time. This
3092 is particularly useful when writing services that drop
3093 privileges using the User= or Group= setting.
3095 * The DeviceAllow= unit setting now supports globbing for
3096 matching against device group names.
3098 * The systemd configuration file system.conf gained new
3099 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3100 DefaultMemoryAccounting= to globally turn on/off accounting
3101 for specific resources (cgroups) for all units. These
3102 settings may still be overridden individually in each unit
3105 * systemd-gpt-auto-generator is now able to discover /srv and
3106 root partitions in addition to /home and swap partitions. It
3107 also supports LUKS-encrypted partitions now. With this in
3108 place, automatic discovery of partitions to mount following
3109 the Discoverable Partitions Specification
3110 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3111 is now a lot more complete. This allows booting without
3112 /etc/fstab and without root= on the kernel command line on
3113 systems prepared appropriately.
3115 * systemd-nspawn gained a new --image= switch which allows
3116 booting up disk images and Linux installations on any block
3117 device that follow the Discoverable Partitions Specification
3118 (see above). This means that installations made with
3119 appropriately updated installers may now be started and
3120 deployed using container managers, completely
3121 unmodified. (We hope that libvirt-lxc will add support for
3122 this feature soon, too.)
3124 * systemd-nspawn gained a new --network-macvlan= setting to
3125 set up a private macvlan interface for the
3126 container. Similarly, systemd-networkd gained a new
3127 Kind=macvlan setting in .netdev files.
3129 * systemd-networkd now supports configuring local addresses
3132 * A new tool systemd-network-wait-online has been added to
3133 synchronously wait for network connectivity using
3136 * The sd-bus.h bus API gained a new sd_bus_track object for
3137 tracking the life-cycle of bus peers. Note that sd-bus.h is
3138 still not a public API though (unless you specify
3139 --enable-kdbus on the configure command line, which however
3140 voids your warranty and you get no API stability guarantee).
3142 * The $XDG_RUNTIME_DIR runtime directories for each user are
3143 now individual tmpfs instances, which has the benefit of
3144 introducing separate pools for each user, with individual
3145 size limits, and thus making sure that unprivileged clients
3146 can no longer negatively impact the system or other users by
3147 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3148 RuntimeDirectorySize= has been introduced that allows
3149 controlling the default size limit for all users. It
3150 defaults to 10% of the available physical memory. This is no
3151 replacement for quotas on tmpfs though (which the kernel
3152 still does not support), as /dev/shm and /tmp are still
3153 shared resources used by both the system and unprivileged
3156 * logind will now automatically turn off automatic suspending
3157 on laptop lid close when more than one display is
3158 connected. This was previously expected to be implemented
3159 individually in desktop environments (such as GNOME),
3160 however has been added to logind now, in order to fix a
3161 boot-time race where a desktop environment might not have
3162 been started yet and thus not been able to take an inhibitor
3163 lock at the time where logind already suspends the system
3164 due to a closed lid.
3166 * logind will now wait at least 30s after each system
3167 suspend/resume cycle, and 3min after system boot before
3168 suspending the system due to a closed laptop lid. This
3169 should give USB docking stations and similar enough time to
3170 be probed and configured after system resume and boot in
3171 order to then act as suspend blocker.
3173 * systemd-run gained a new --property= setting which allows
3174 initialization of resource control properties (and others)
3175 for the created scope or service unit. Example: "systemd-run
3176 --property=BlockIOWeight=10 updatedb" may be used to run
3177 updatedb at a low block IO scheduling weight.
3179 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3180 now also work in --scope mode.
3182 * When systemd is compiled with kdbus support, basic support
3183 for enforced policies is now in place. (Note that enabling
3184 kdbus still voids your warranty and no API compatibility
3187 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3188 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3189 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3190 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3191 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3192 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3193 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3194 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3195 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3196 Zbigniew Jędrzejewski-Szmek
3198 — Berlin, 2014-03-12
3202 * systemd will now relabel /dev after loading the SMACK policy
3203 according to SMACK rules.
3205 * A new unit file option AppArmorProfile= has been added to
3206 set the AppArmor profile for the processes of a unit.
3208 * A new condition check ConditionArchitecture= has been added
3209 to conditionalize units based on the system architecture, as
3210 reported by uname()'s "machine" field.
3212 * systemd-networkd now supports matching on the system
3213 virtualization, architecture, kernel command line, host name
3216 * logind is now a lot more aggressive when suspending the
3217 machine due to a closed laptop lid. Instead of acting only
3218 on the lid close action, it will continuously watch the lid
3219 status and act on it. This is useful for laptops where the
3220 power button is on the outside of the chassis so that it can
3221 be reached without opening the lid (such as the Lenovo
3222 Yoga). On those machines, logind will now immediately
3223 re-suspend the machine if the power button has been
3224 accidentally pressed while the laptop was suspended and in a
3225 backpack or similar.
3227 * logind will now watch SW_DOCK switches and inhibit reaction
3228 to the lid switch if it is pressed. This means that logind
3229 will not suspend the machine anymore if the lid is closed
3230 and the system is docked, if the laptop supports SW_DOCK
3231 notifications via the input layer. Note that ACPI docking
3232 stations do not generate this currently. Also note that this
3233 logic is usually not fully sufficient and Desktop
3234 Environments should take a lid switch inhibitor lock when an
3235 external display is connected, as systemd will not watch
3238 * nspawn will now make use of the devices cgroup controller by
3239 default, and only permit creation of and access to the usual
3240 API device nodes like /dev/null or /dev/random, as well as
3241 access to (but not creation of) the pty devices.
3243 * We will now ship a default .network file for
3244 systemd-networkd that automatically configures DHCP for
3245 network interfaces created by nspawn's --network-veth or
3246 --network-bridge= switches.
3248 * systemd will now understand the usual M, K, G, T suffixes
3249 according to SI conventions (i.e. to the base 1000) when
3250 referring to throughput and hardware metrics. It will stay
3251 with IEC conventions (i.e. to the base 1024) for software
3252 metrics, according to what is customary according to
3253 Wikipedia. We explicitly document which base applies for
3254 each configuration option.
3256 * The DeviceAllow= setting in unit files now supports a syntax
3257 to whitelist an entire group of devices node majors at once,
3258 based on the /proc/devices listing. For example, with the
3259 string "char-pts", it is now possible to whitelist all
3260 current and future pseudo-TTYs at once.
3262 * sd-event learned a new "post" event source. Event sources of
3263 this type are triggered by the dispatching of any event
3264 source of a type that is not "post". This is useful for
3265 implementing clean-up and check event sources that are
3266 triggered by other work being done in the program.
3268 * systemd-networkd is no longer statically enabled, but uses
3269 the usual [Install] sections so that it can be
3270 enabled/disabled using systemctl. It still is enabled by
3273 * When creating a veth interface pair with systemd-nspawn, the
3274 host side will now be prefixed with "vb-" if
3275 --network-bridge= is used, and with "ve-" if --network-veth
3276 is used. This way, it is easy to distinguish these cases on
3277 the host, for example to apply different configuration to
3278 them with systemd-networkd.
3280 * The compatibility libraries for libsystemd-journal.so,
3281 libsystem-id128.so, libsystemd-login.so and
3282 libsystemd-daemon.so do not make use of IFUNC
3283 anymore. Instead, we now build libsystemd.so multiple times
3284 under these alternative names. This means that the footprint
3285 is drastically increased, but given that these are
3286 transitional compatibility libraries, this should not matter
3287 much. This change has been made necessary to support the ARM
3288 platform for these compatibility libraries, as the ARM
3289 toolchain is not really at the same level as the toolchain
3290 for other architectures like x86 and does not support
3291 IFUNC. Please make sure to use --enable-compat-libs only
3292 during a transitional period!
3294 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3295 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3296 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3297 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3298 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3299 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3300 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3301 Zbigniew Jędrzejewski-Szmek
3303 — Berlin, 2014-02-24
3307 * A new component "systemd-networkd" has been added that can
3308 be used to configure local network interfaces statically or
3309 via DHCP. It is capable of bringing up bridges, VLANs, and
3310 bonding. Currently, no hook-ups for interactive network
3311 configuration are provided. Use this for your initrd,
3312 container, embedded, or server setup if you need a simple,
3313 yet powerful, network configuration solution. This
3314 configuration subsystem is quite nifty, as it allows wildcard
3315 hotplug matching in interfaces. For example, with a single
3316 configuration snippet, you can configure that all Ethernet
3317 interfaces showing up are automatically added to a bridge,
3318 or similar. It supports link-sensing and more.
3320 * A new tool "systemd-socket-proxyd" has been added which can
3321 act as a bidirectional proxy for TCP sockets. This is
3322 useful for adding socket activation support to services that
3323 do not actually support socket activation, including virtual
3324 machines and the like.
3326 * Add a new tool to save/restore rfkill state on
3329 * Save/restore state of keyboard backlights in addition to
3330 display backlights on shutdown/boot.
3332 * udev learned a new SECLABEL{} construct to label device
3333 nodes with a specific security label when they appear. For
3334 now, only SECLABEL{selinux} is supported, but the syntax is
3335 prepared for additional security frameworks.
3337 * udev gained a new scheme to configure link-level attributes
3338 from files in /etc/systemd/network/*.link. These files can
3339 match against MAC address, device path, driver name and type,
3340 and will apply attributes like the naming policy, link speed,
3341 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3342 address assignment policy (randomized, ...).
3344 * The configuration of network interface naming rules for
3345 "permanent interface names" has changed: a new NamePolicy=
3346 setting in the [Link] section of .link files determines the
3347 priority of possible naming schemes (onboard, slot, MAC,
3348 path). The default value of this setting is determined by
3349 /usr/lib/net/links/99-default.link. Old
3350 80-net-name-slot.rules udev configuration file has been
3351 removed, so local configuration overriding this file should
3352 be adapated to override 99-default.link instead.
3354 * When the User= switch is used in a unit file, also
3355 initialize $SHELL= based on the user database entry.
3357 * systemd no longer depends on libdbus. All communication is
3358 now done with sd-bus, systemd's low-level bus library
3361 * kdbus support has been added to PID 1 itself. When kdbus is
3362 enabled, this causes PID 1 to set up the system bus and
3363 enable support for a new ".busname" unit type that
3364 encapsulates bus name activation on kdbus. It works a little
3365 bit like ".socket" units, except for bus names. A new
3366 generator has been added that converts classic dbus1 service
3367 activation files automatically into native systemd .busname
3370 * sd-bus: add a light-weight vtable implementation that allows
3371 defining objects on the bus with a simple static const
3372 vtable array of its methods, signals and properties.
3374 * systemd will not generate or install static dbus
3375 introspection data anymore to /usr/share/dbus-1/interfaces,
3376 as the precise format of these files is unclear, and
3377 nothing makes use of it.
3379 * A proxy daemon is now provided to proxy clients connecting
3380 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3381 compatibility with classic D-Bus.
3383 * A bus driver implementation has been added that supports the
3384 classic D-Bus bus driver calls on kdbus, also for
3385 compatibility purposes.
3387 * A new API "sd-event.h" has been added that implements a
3388 minimal event loop API built around epoll. It provides a
3389 couple of features that direct epoll usage is lacking:
3390 prioritization of events, scales to large numbers of timer
3391 events, per-event timer slack (accuracy), system-wide
3392 coalescing of timer events, exit handlers, watchdog
3393 supervision support using systemd's sd_notify() API, child
3396 * A new API "sd-rntl.h" has been added that provides an API
3397 around the route netlink interface of the kernel, similar in
3398 style to "sd-bus.h".
3400 * A new API "sd-dhcp-client.h" has been added that provides a
3401 small DHCPv4 client-side implementation. This is used by
3404 * There is a new kernel command line option
3405 "systemd.restore_state=0|1". When set to "0", none of the
3406 systemd tools will restore saved runtime state to hardware
3407 devices. More specifically, the rfkill and backlight states
3410 * The FsckPassNo= compatibility option in mount/service units
3411 has been removed. The fstab generator will now add the
3412 necessary dependencies automatically, and does not require
3413 PID1's support for that anymore.
3415 * journalctl gained a new switch, --list-boots, that lists
3416 recent boots with their times and boot IDs.
3418 * The various tools like systemctl, loginctl, timedatectl,
3419 busctl, systemd-run, ... have gained a new switch "-M" to
3420 connect to a specific, local OS container (as direct
3421 connection, without requiring SSH). This works on any
3422 container that is registered with machined, such as those
3423 created by libvirt-lxc or nspawn.
3425 * systemd-run and systemd-analyze also gained support for "-H"
3426 to connect to remote hosts via SSH. This is particularly
3427 useful for systemd-run because it enables queuing of jobs
3428 onto remote systems.
3430 * machinectl gained a new command "login" to open a getty
3431 login in any local container. This works with any container
3432 that is registered with machined (such as those created by
3433 libvirt-lxc or nspawn), and which runs systemd inside.
3435 * machinectl gained a new "reboot" command that may be used to
3436 trigger a reboot on a specific container that is registered
3437 with machined. This works on any container that runs an init
3438 system of some kind.
3440 * systemctl gained a new "list-timers" command to print a nice
3441 listing of installed timer units with the times they elapse
3444 * Alternative reboot() parameters may now be specified on the
3445 "systemctl reboot" command line and are passed to the
3446 reboot() system call.
3448 * systemctl gained a new --job-mode= switch to configure the
3449 mode to queue a job with. This is a more generic version of
3450 --fail, --irreversible, and --ignore-dependencies, which are
3451 still available but not advertised anymore.
3453 * /etc/systemd/system.conf gained new settings to configure
3454 various default timeouts of units, as well as the default
3455 start limit interval and burst. These may still be overridden
3458 * PID1 will now export on the bus profile data of the security
3459 policy upload process (such as the SELinux policy upload to
3462 * journald: when forwarding logs to the console, include
3463 timestamps (following the setting in
3464 /sys/module/printk/parameters/time).
3466 * OnCalendar= in timer units now understands the special
3467 strings "yearly" and "annually". (Both are equivalent)
3469 * The accuracy of timer units is now configurable with the new
3470 AccuracySec= setting. It defaults to 1min.
3472 * A new dependency type JoinsNamespaceOf= has been added that
3473 allows running two services within the same /tmp and network
3474 namespace, if PrivateNetwork= or PrivateTmp= are used.
3476 * A new command "cat" has been added to systemctl. It outputs
3477 the original unit file of a unit, and concatenates the
3478 contents of additional "drop-in" unit file snippets, so that
3479 the full configuration is shown.
3481 * systemctl now supports globbing on the various "list-xyz"
3482 commands, like "list-units" or "list-sockets", as well as on
3483 those commands which take multiple unit names.
3485 * journalctl's --unit= switch gained support for globbing.
3487 * All systemd daemons now make use of the watchdog logic so
3488 that systemd automatically notices when they hang.
3490 * If the $container_ttys environment variable is set,
3491 getty-generator will automatically spawn a getty for each
3492 listed tty. This is useful for container managers to request
3493 login gettys to be spawned on as many ttys as needed.
3495 * %h, %s, %U specifier support is not available anymore when
3496 used in unit files for PID 1. This is because NSS calls are
3497 not safe from PID 1. They stay available for --user
3498 instances of systemd, and as special case for the root user.
3500 * loginctl gained a new "--no-legend" switch to turn off output
3503 * The "sd-login.h" API gained three new calls:
3504 sd_session_is_remote(), sd_session_get_remote_user(),
3505 sd_session_get_remote_host() to query information about
3508 * The udev hardware database now also carries vendor/product
3509 information of SDIO devices.
3511 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3512 determine whether watchdog notifications are requested by
3515 * Socket-activated per-connection services now include a
3516 short description of the connection parameters in the
3519 * tmpfiles gained a new "--boot" option. When this is not used,
3520 only lines where the command character is not suffixed with
3521 "!" are executed. When this option is specified, those
3522 options are executed too. This partitions tmpfiles
3523 directives into those that can be safely executed at any
3524 time, and those which should be run only at boot (for
3525 example, a line that creates /run/nologin).
3527 * A new API "sd-resolve.h" has been added which provides a simple
3528 asynchronous wrapper around glibc NSS host name resolution
3529 calls, such as getaddrinfo(). In contrast to glibc's
3530 getaddrinfo_a(), it does not use signals. In contrast to most
3531 other asynchronous name resolution libraries, this one does
3532 not reimplement DNS, but reuses NSS, so that alternate
3533 host name resolution systems continue to work, such as mDNS,
3534 LDAP, etc. This API is based on libasyncns, but it has been
3535 cleaned up for inclusion in systemd.
3537 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3538 "sd-daemon.h" are no longer found in individual libraries
3539 libsystemd-journal.so, libsystemd-login.so,
3540 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3541 merged them into a single library, libsystemd.so, which
3542 provides all symbols. The reason for this is cyclic
3543 dependencies, as these libraries tend to use each other's
3544 symbols. So far, we have managed to workaround that by linking
3545 a copy of a good part of our code into each of these
3546 libraries again and again, which, however, makes certain
3547 things hard to do, like sharing static variables. Also, it
3548 substantially increases footprint. With this change, there
3549 is only one library for the basic APIs systemd
3550 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3551 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3552 library as well, however are subject to the --enable-kdbus
3553 switch (see below). Note that "sd-dhcp-client.h" is not part
3554 of this library (this is because it only consumes, never
3555 provides, services of/to other APIs). To make the transition
3556 easy from the separate libraries to the unified one, we
3557 provide the --enable-compat-libs compile-time switch which
3558 will generate stub libraries that are compatible with the
3559 old ones but redirect all calls to the new one.
3561 * All of the kdbus logic and the new APIs "sd-bus.h",
3562 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3563 and "sd-utf8.h" are compile-time optional via the
3564 "--enable-kdbus" switch, and they are not compiled in by
3565 default. To make use of kdbus, you have to explicitly enable
3566 the switch. Note however, that neither the kernel nor the
3567 userspace API for all of this is considered stable yet. We
3568 want to maintain the freedom to still change the APIs for
3569 now. By specifying this build-time switch, you acknowledge
3570 that you are aware of the instability of the current
3573 * Also, note that while kdbus is pretty much complete,
3574 it lacks one thing: proper policy support. This means you
3575 can build a fully working system with all features; however,
3576 it will be highly insecure. Policy support will be added in
3577 one of the next releases, at the same time that we will
3578 declare the APIs stable.
3580 * When the kernel command line argument "kdbus" is specified,
3581 systemd will automatically load the kdbus.ko kernel module. At
3582 this stage of development, it is only useful for testing kdbus
3583 and should not be used in production. Note: if "--enable-kdbus"
3584 is specified, and the kdbus.ko kernel module is available, and
3585 "kdbus" is added to the kernel command line, the entire system
3586 runs with kdbus instead of dbus-daemon, with the above mentioned
3587 problem of missing the system policy enforcement. Also a future
3588 version of kdbus.ko or a newer systemd will not be compatible with
3589 each other, and will unlikely be able to boot the machine if only
3590 one of them is updated.
3592 * systemctl gained a new "import-environment" command which
3593 uploads the caller's environment (or parts thereof) into the
3594 service manager so that it is inherited by services started
3595 by the manager. This is useful to upload variables like
3596 $DISPLAY into the user service manager.
3598 * A new PrivateDevices= switch has been added to service units
3599 which allows running a service with a namespaced /dev
3600 directory that does not contain any device nodes for
3601 physical devices. More specifically, it only includes devices
3602 such as /dev/null, /dev/urandom, and /dev/zero which are API
3605 * logind has been extended to support behaviour like VT
3606 switching on seats that do not support a VT. This makes
3607 multi-session available on seats that are not the first seat
3608 (seat0), and on systems where kernel support for VTs has
3609 been disabled at compile-time.
3611 * If a process holds a delay lock for system sleep or shutdown
3612 and fails to release it in time, we will now log its
3613 identity. This makes it easier to identify processes that
3614 cause slow suspends or power-offs.
3616 * When parsing /etc/crypttab, support for a new key-slot=
3617 option as supported by Debian is added. It allows indicating
3618 which LUKS slot to use on disk, speeding up key loading.
3620 * The sd_journald_sendv() API call has been checked and
3621 officially declared to be async-signal-safe so that it may
3622 be invoked from signal handlers for logging purposes.
3624 * Boot-time status output is now enabled automatically after a
3625 short timeout if boot does not progress, in order to give
3626 the user an indication what she or he is waiting for.
3628 * The boot-time output has been improved to show how much time
3629 remains until jobs expire.
3631 * The KillMode= switch in service units gained a new possible
3632 value "mixed". If set, and the unit is shut down, then the
3633 initial SIGTERM signal is sent only to the main daemon
3634 process, while the following SIGKILL signal is sent to
3635 all remaining processes of the service.
3637 * When a scope unit is registered, a new property "Controller"
3638 may be set. If set to a valid bus name, systemd will send a
3639 RequestStop() signal to this name when it would like to shut
3640 down the scope. This may be used to hook manager logic into
3641 the shutdown logic of scope units. Also, scope units may now
3642 be put in a special "abandoned" state, in which case the
3643 manager process which created them takes no further
3644 responsibilities for it.
3646 * When reading unit files, systemd will now verify
3647 the access mode of these files, and warn about certain
3648 suspicious combinations. This has been added to make it
3649 easier to track down packaging bugs where unit files are
3650 marked executable or world-writable.
3652 * systemd-nspawn gained a new "--setenv=" switch to set
3653 container-wide environment variables. The similar option in
3654 systemd-activate was renamed from "--environment=" to
3655 "--setenv=" for consistency.
3657 * systemd-nspawn has been updated to create a new kdbus domain
3658 for each container that is invoked, thus allowing each
3659 container to have its own set of system and user buses,
3660 independent of the host.
3662 * systemd-nspawn gained a new --drop-capability= switch to run
3663 the container with less capabilities than the default. Both
3664 --drop-capability= and --capability= now take the special
3665 string "all" for dropping or keeping all capabilities.
3667 * systemd-nspawn gained new switches for executing containers
3668 with specific SELinux labels set.
3670 * systemd-nspawn gained a new --quiet switch to not generate
3671 any additional output but the container's own console
3674 * systemd-nspawn gained a new --share-system switch to run a
3675 container without PID namespacing enabled.
3677 * systemd-nspawn gained a new --register= switch to control
3678 whether the container is registered with systemd-machined or
3679 not. This is useful for containers that do not run full
3680 OS images, but only specific apps.
3682 * systemd-nspawn gained a new --keep-unit which may be used
3683 when invoked as the only program from a service unit, and
3684 results in registration of the unit service itself in
3685 systemd-machined, instead of a newly opened scope unit.
3687 * systemd-nspawn gained a new --network-interface= switch for
3688 moving arbitrary interfaces to the container. The new
3689 --network-veth switch creates a virtual Ethernet connection
3690 between host and container. The new --network-bridge=
3691 switch then allows assigning the host side of this virtual
3692 Ethernet connection to a bridge device.
3694 * systemd-nspawn gained a new --personality= switch for
3695 setting the kernel personality for the container. This is
3696 useful when running a 32-bit container on a 64-bit host. A
3697 similar option Personality= is now also available for service
3700 * logind will now also track a "Desktop" identifier for each
3701 session which encodes the desktop environment of it. This is
3702 useful for desktop environments that want to identify
3703 multiple running sessions of itself easily.
3705 * A new SELinuxContext= setting for service units has been
3706 added that allows setting a specific SELinux execution
3707 context for a service.
3709 * Most systemd client tools will now honour $SYSTEMD_LESS for
3710 settings of the "less" pager. By default, these tools will
3711 override $LESS to allow certain operations to work, such as
3712 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3713 influence this logic.
3715 * systemd's "seccomp" hook-up has been changed to make use of
3716 the libseccomp library instead of using its own
3717 implementation. This has benefits for portability among
3720 * For usage together with SystemCallFilter=, a new
3721 SystemCallErrorNumber= setting has been introduced that
3722 allows configuration of a system error number to be returned
3723 on filtered system calls, instead of immediately killing the
3724 process. Also, SystemCallArchitectures= has been added to
3725 limit access to system calls of a particular architecture
3726 (in order to turn off support for unused secondary
3727 architectures). There is also a global
3728 SystemCallArchitectures= setting in system.conf now to turn
3729 off support for non-native system calls system-wide.
3731 * systemd requires a kernel with a working name_to_handle_at(),
3732 please see the kernel config requirements in the README file.
3734 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3735 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3736 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3737 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3738 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3739 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3740 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3741 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3742 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3743 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3744 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3745 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3746 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3747 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3748 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3749 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3750 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3751 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3752 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3753 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3754 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3755 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3756 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3757 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3759 — Berlin, 2014-02-20
3763 * logind has gained support for facilitating privileged input
3764 and drm device access for unprivileged clients. This work is
3765 useful to allow Wayland display servers (and similar
3766 programs, such as kmscon) to run under the user's ID and
3767 access input and drm devices which are normally
3768 protected. When this is used (and the kernel is new enough)
3769 logind will "mute" IO on the file descriptors passed to
3770 Wayland as long as it is in the background and "unmute" it
3771 if it returns into the foreground. This allows secure
3772 session switching without allowing background sessions to
3773 eavesdrop on input and display data. This also introduces
3774 session switching support if VT support is turned off in the
3775 kernel, and on seats that are not seat0.
3777 * A new kernel command line option luks.options= is understood
3778 now which allows specifying LUKS options for usage for LUKS
3779 encrypted partitions specified with luks.uuid=.
3781 * tmpfiles.d(5) snippets may now use specifier expansion in
3782 path names. More specifically %m, %b, %H, %v, are now
3783 replaced by the local machine id, boot id, hostname, and
3784 kernel version number.
3786 * A new tmpfiles.d(5) command "m" has been introduced which
3787 may be used to change the owner/group/access mode of a file
3788 or directory if it exists, but do nothing if it does not.
3790 * This release removes high-level support for the
3791 MemorySoftLimit= cgroup setting. The underlying kernel
3792 cgroup attribute memory.soft_limit= is currently badly
3793 designed and likely to be removed from the kernel API in its
3794 current form, hence we should not expose it for now.
3796 * The memory.use_hierarchy cgroup attribute is now enabled for
3797 all cgroups systemd creates in the memory cgroup
3798 hierarchy. This option is likely to be come the built-in
3799 default in the kernel anyway, and the non-hierarchical mode
3800 never made much sense in the intrinsically hierarchical
3803 * A new field _SYSTEMD_SLICE= is logged along with all journal
3804 messages containing the slice a message was generated
3805 from. This is useful to allow easy per-customer filtering of
3806 logs among other things.
3808 * systemd-journald will no longer adjust the group of journal
3809 files it creates to the "systemd-journal" group. Instead we
3810 rely on the journal directory to be owned by the
3811 "systemd-journal" group, and its setgid bit set, so that the
3812 kernel file system layer will automatically enforce that
3813 journal files inherit this group assignment. The reason for
3814 this change is that we cannot allow NSS look-ups from
3815 journald which would be necessary to resolve
3816 "systemd-journal" to a numeric GID, because this might
3817 create deadlocks if NSS involves synchronous queries to
3818 other daemons (such as nscd, or sssd) which in turn are
3819 logging clients of journald and might block on it, which
3820 would then dead lock. A tmpfiles.d(5) snippet included in
3821 systemd will make sure the setgid bit and group are
3822 properly set on the journal directory if it exists on every
3823 boot. However, we recommend adjusting it manually after
3824 upgrades too (or from RPM scriptlets), so that the change is
3825 not delayed until next reboot.
3827 * Backlight and random seed files in /var/lib/ have moved into
3828 the /var/lib/systemd/ directory, in order to centralize all
3829 systemd generated files in one directory.
3831 * Boot time performance measurements (as displayed by
3832 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3833 performance information if that's available to determine how
3834 much time BIOS and boot loader initialization required. With
3835 a sufficiently new BIOS you hence no longer need to boot
3836 with Gummiboot to get access to such information.
3838 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3839 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3840 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3841 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3842 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3843 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3844 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3846 — Berlin, 2013-10-02
3850 * The Restart= option for services now understands a new
3851 on-watchdog setting, which will restart the service
3852 automatically if the service stops sending out watchdog keep
3853 alive messages (as configured with WatchdogSec=).
3855 * The getty generator (which is responsible for bringing up a
3856 getty on configured serial consoles) will no longer only
3857 start a getty on the primary kernel console but on all
3858 others, too. This makes the order in which console= is
3859 specified on the kernel command line less important.
3861 * libsystemd-logind gained a new sd_session_get_vt() call to
3862 retrieve the VT number of a session.
3864 * If the option "tries=0" is set for an entry of /etc/crypttab
3865 its passphrase is queried indefinitely instead of any
3866 maximum number of tries.
3868 * If a service with a configure PID file terminates its PID
3869 file will now be removed automatically if it still exists
3870 afterwards. This should put an end to stale PID files.
3872 * systemd-run will now also take relative binary path names
3873 for execution and no longer insists on absolute paths.
3875 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3876 paths that are optionally prefixed with "-" to indicate that
3877 it should not be considered a failure if they do not exist.
3879 * journalctl -o (and similar commands) now understands a new
3880 output mode "short-precise", it is similar to "short" but
3881 shows timestamps with usec accuracy.
3883 * The option "discard" (as known from Debian) is now
3884 synonymous to "allow-discards" in /etc/crypttab. In fact,
3885 "discard" is preferred now (since it is easier to remember
3888 * Some licensing clean-ups were made, so that more code is now
3889 LGPL-2.1 licensed than before.
3891 * A minimal tool to save/restore the display backlight
3892 brightness across reboots has been added. It will store the
3893 backlight setting as late as possible at shutdown, and
3894 restore it as early as possible during reboot.
3896 * A logic to automatically discover and enable home and swap
3897 partitions on GPT disks has been added. With this in place
3898 /etc/fstab becomes optional for many setups as systemd can
3899 discover certain partitions located on the root disk
3900 automatically. Home partitions are recognized under their
3901 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3902 partitions are recognized under their GPT type ID
3903 0657fd6da4ab43c484e50933c84b4f4f.
3905 * systemd will no longer pass any environment from the kernel
3906 or initrd to system services. If you want to set an
3907 environment for all services, do so via the kernel command
3908 line systemd.setenv= assignment.
3910 * The systemd-sysctl tool no longer natively reads the file
3911 /etc/sysctl.conf. If desired, the file should be symlinked
3912 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3913 legacy support by a symlink rather than built-in code, it
3914 also makes the otherwise hidden order of application of the
3915 different files visible. (Note that this partly reverts to a
3916 pre-198 application order of sysctl knobs!)
3918 * The "systemctl set-log-level" and "systemctl dump" commands
3919 have been moved to systemd-analyze.
3921 * systemd-run learned the new --remain-after-exit switch,
3922 which causes the scope unit not to be cleaned up
3923 automatically after the process terminated.
3925 * tmpfiles learned a new --exclude-prefix= switch to exclude
3926 certain paths from operation.
3928 * journald will now automatically flush all messages to disk
3929 as soon as a message at the log level CRIT, ALERT or EMERG
3932 Contributions from: Andrew Cook, Brandon Philips, Christian
3933 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3934 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3935 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3936 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3937 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3938 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3939 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3940 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3941 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3942 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3943 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3944 William Giokas, Zbigniew Jędrzejewski-Szmek
3946 — Berlin, 2013-09-13
3950 * The documentation has been updated to cover the various new
3951 concepts introduced with 205.
3953 * Unit files now understand the new %v specifier which
3954 resolves to the kernel version string as returned by "uname
3957 * systemctl now supports filtering the unit list output by
3958 load state, active state and sub state, using the new
3961 * "systemctl status" will now show the results of the
3962 condition checks (like ConditionPathExists= and similar) of
3963 the last start attempts of the unit. They are also logged to
3966 * "journalctl -b" may now be used to look for boot output of a
3967 specific boot. Try "journalctl -b -1" for the previous boot,
3968 but the syntax is substantially more powerful.
3970 * "journalctl --show-cursor" has been added which prints the
3971 cursor string the last shown log line. This may then be used
3972 with the new "journalctl --after-cursor=" switch to continue
3973 browsing logs from that point on.
3975 * "journalctl --force" may now be used to force regeneration
3978 * Creation of "dead" device nodes has been moved from udev
3979 into kmod and tmpfiles. Previously, udev would read the kmod
3980 databases to pre-generate dead device nodes based on meta
3981 information contained in kernel modules, so that these would
3982 be auto-loaded on access rather then at boot. As this
3983 does not really have much to do with the exposing actual
3984 kernel devices to userspace this has always been slightly
3985 alien in the udev codebase. Following the new scheme kmod
3986 will now generate a runtime snippet for tmpfiles from the
3987 module meta information and it now is tmpfiles' job to the
3988 create the nodes. This also allows overriding access and
3989 other parameters for the nodes using the usual tmpfiles
3990 facilities. As side effect this allows us to remove the
3991 CAP_SYS_MKNOD capability bit from udevd entirely.
3993 * logind's device ACLs may now be applied to these "dead"
3994 devices nodes too, thus finally allowing managed access to
3995 devices such as /dev/snd/sequencer whithout loading the
3996 backing module right-away.
3998 * A new RPM macro has been added that may be used to apply
3999 tmpfiles configuration during package installation.
4001 * systemd-detect-virt and ConditionVirtualization= now can
4002 detect User-Mode-Linux machines (UML).
4004 * journald will now implicitly log the effective capabilities
4005 set of processes in the message metadata.
4007 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4009 * The initrd interface has been simplified (more specifically,
4010 support for passing performance data via environment
4011 variables and fsck results via files in /run has been
4012 removed). These features were non-essential, and are
4013 nowadays available in a much nicer way by having systemd in
4014 the initrd serialize its state and have the hosts systemd
4015 deserialize it again.
4017 * The udev "keymap" data files and tools to apply keyboard
4018 specific mappings of scan to key codes, and force-release
4019 scan code lists have been entirely replaced by a udev
4020 "keyboard" builtin and a hwdb data file.
4022 * systemd will now honour the kernel's "quiet" command line
4023 argument also during late shutdown, resulting in a
4024 completely silent shutdown when used.
4026 * There's now an option to control the SO_REUSEPORT socket
4027 option in .socket units.
4029 * Instance units will now automatically get a per-template
4030 subslice of system.slice unless something else is explicitly
4031 configured. For example, instances of sshd@.service will now
4032 implicitly be placed in system-sshd.slice rather than
4033 system.slice as before.
4035 * Test coverage support may now be enabled at build time.
4037 Contributions from: Dave Reisner, Frederic Crozat, Harald
4038 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4039 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4040 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4041 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4042 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4043 Giokas, Zbigniew Jędrzejewski-Szmek
4045 — Berlin, 2013-07-23
4049 * Two new unit types have been introduced:
4051 Scope units are very similar to service units, however, are
4052 created out of pre-existing processes — instead of PID 1
4053 forking off the processes. By using scope units it is
4054 possible for system services and applications to group their
4055 own child processes (worker processes) in a powerful way
4056 which then maybe used to organize them, or kill them
4057 together, or apply resource limits on them.
4059 Slice units may be used to partition system resources in an
4060 hierarchical fashion and then assign other units to them. By
4061 default there are now three slices: system.slice (for all
4062 system services), user.slice (for all user sessions),
4063 machine.slice (for VMs and containers).
4065 Slices and scopes have been introduced primarily in
4066 context of the work to move cgroup handling to a
4067 single-writer scheme, where only PID 1
4068 creates/removes/manages cgroups.
4070 * There's a new concept of "transient" units. In contrast to
4071 normal units these units are created via an API at runtime,
4072 not from configuration from disk. More specifically this
4073 means it is now possible to run arbitrary programs as
4074 independent services, with all execution parameters passed
4075 in via bus APIs rather than read from disk. Transient units
4076 make systemd substantially more dynamic then it ever was,
4077 and useful as a general batch manager.
4079 * logind has been updated to make use of scope and slice units
4080 for managing user sessions. As a user logs in he will get
4081 his own private slice unit, to which all sessions are added
4082 as scope units. We also added support for automatically
4083 adding an instance of user@.service for the user into the
4084 slice. Effectively logind will no longer create cgroup
4085 hierarchies on its own now, it will defer entirely to PID 1
4086 for this by means of scope, service and slice units. Since
4087 user sessions this way become entities managed by PID 1
4088 the output of "systemctl" is now a lot more comprehensive.
4090 * A new mini-daemon "systemd-machined" has been added which
4091 may be used by virtualization managers to register local
4092 VMs/containers. nspawn has been updated accordingly, and
4093 libvirt will be updated shortly. machined will collect a bit
4094 of meta information about the VMs/containers, and assign
4095 them their own scope unit (see above). The collected
4096 meta-data is then made available via the "machinectl" tool,
4097 and exposed in "ps" and similar tools. machined/machinectl
4098 is compile-time optional.
4100 * As discussed earlier, the low-level cgroup configuration
4101 options ControlGroup=, ControlGroupModify=,
4102 ControlGroupPersistent=, ControlGroupAttribute= have been
4103 removed. Please use high-level attribute settings instead as
4104 well as slice units.
4106 * A new bus call SetUnitProperties() has been added to alter
4107 various runtime parameters of a unit. This is primarily
4108 useful to alter cgroup parameters dynamically in a nice way,
4109 but will be extended later on to make more properties
4110 modifiable at runtime. systemctl gained a new set-properties
4111 command that wraps this call.
4113 * A new tool "systemd-run" has been added which can be used to
4114 run arbitrary command lines as transient services or scopes,
4115 while configuring a number of settings via the command
4116 line. This tool is currently very basic, however already
4117 very useful. We plan to extend this tool to even allow
4118 queuing of execution jobs with time triggers from the
4119 command line, similar in fashion to "at".
4121 * nspawn will now inform the user explicitly that kernels with
4122 audit enabled break containers, and suggest the user to turn
4125 * Support for detecting the IMA and AppArmor security
4126 frameworks with ConditionSecurity= has been added.
4128 * journalctl gained a new "-k" switch for showing only kernel
4129 messages, mimicking dmesg output; in addition to "--user"
4130 and "--system" switches for showing only user's own logs
4133 * systemd-delta can now show information about drop-in
4134 snippets extending unit files.
4136 * libsystemd-bus has been substantially updated but is still
4137 not available as public API.
4139 * systemd will now look for the "debug" argument on the kernel
4140 command line and enable debug logging, similar to what
4141 "systemd.log_level=debug" already did before.
4143 * "systemctl set-default", "systemctl get-default" has been
4144 added to configure the default.target symlink, which
4145 controls what to boot into by default.
4147 * "systemctl set-log-level" has been added as a convenient
4148 way to raise and lower systemd logging threshold.
4150 * "systemd-analyze plot" will now show the time the various
4151 generators needed for execution, as well as information
4152 about the unit file loading.
4154 * libsystemd-journal gained a new sd_journal_open_files() call
4155 for opening specific journal files. journactl also gained a
4156 new switch to expose this new functionality. Previously we
4157 only supported opening all files from a directory, or all
4158 files from the system, as opening individual files only is
4159 racy due to journal file rotation.
4161 * systemd gained the new DefaultEnvironment= setting in
4162 /etc/systemd/system.conf to set environment variables for
4165 * If a privileged process logs a journal message with the
4166 OBJECT_PID= field set, then journald will automatically
4167 augment this with additional OBJECT_UID=, OBJECT_GID=,
4168 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4169 system services want to log events about specific client
4170 processes. journactl/systemctl has been updated to make use
4171 of this information if all log messages regarding a specific
4174 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4175 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4176 Reisner, David Coppa, David King, David Strauss, Eelco
4177 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4178 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4179 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4180 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4181 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4182 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4183 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4184 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4185 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4186 Łukasz Stelmach, 장동준
4190 * The Python bindings gained some minimal support for the APIs
4191 exposed by libsystemd-logind.
4193 * ConditionSecurity= gained support for detecting SMACK. Since
4194 this condition already supports SELinux and AppArmor we only
4195 miss IMA for this. Patches welcome!
4197 Contributions from: Karol Lewandowski, Lennart Poettering,
4198 Zbigniew Jędrzejewski-Szmek
4202 * systemd-nspawn will now create /etc/resolv.conf if
4203 necessary, before bind-mounting the host's file onto it.
4205 * systemd-nspawn will now store meta information about a
4206 container on the container's cgroup as extended attribute
4207 fields, including the root directory.
4209 * The cgroup hierarchy has been reworked in many ways. All
4210 objects any of the components systemd creates in the cgroup
4211 tree are now suffixed. More specifically, user sessions are
4212 now placed in cgroups suffixed with ".session", users in
4213 cgroups suffixed with ".user", and nspawn containers in
4214 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4215 names are now escaped in a simple scheme to avoid collision
4216 of userspace object names with kernel filenames. This work
4217 is preparation for making these objects relocatable in the
4218 cgroup tree, in order to allow easy resource partitioning of
4219 these objects without causing naming conflicts.
4221 * systemctl list-dependencies gained the new switches
4222 --plain, --reverse, --after and --before.
4224 * systemd-inhibit now shows the process name of processes that
4225 have taken an inhibitor lock.
4227 * nss-myhostname will now also resolve "localhost"
4228 implicitly. This makes /etc/hosts an optional file and
4229 nicely handles that on IPv6 ::1 maps to both "localhost" and
4232 * libsystemd-logind.so gained a new call
4233 sd_get_machine_names() to enumerate running containers and
4234 VMs (currently only supported by very new libvirt and
4235 nspawn). sd_login_monitor can now be used to watch
4236 VMs/containers coming and going.
4238 * .include is not allowed recursively anymore, and only in
4239 unit files. Usually it is better to use drop-in snippets in
4240 .d/*.conf anyway, as introduced with systemd 198.
4242 * systemd-analyze gained a new "critical-chain" command that
4243 determines the slowest chain of units run during system
4244 boot-up. It is very useful for tracking down where
4245 optimizing boot time is the most beneficial.
4247 * systemd will no longer allow manipulating service paths in
4248 the name=systemd:/system cgroup tree using ControlGroup= in
4249 units. (But is still fine with it in all other dirs.)
4251 * There's a new systemd-nspawn@.service service file that may
4252 be used to easily run nspawn containers as system
4253 services. With the container's root directory in
4254 /var/lib/container/foobar it is now sufficient to run
4255 "systemctl start systemd-nspawn@foobar.service" to boot it.
4257 * systemd-cgls gained a new parameter "--machine" to list only
4258 the processes within a certain container.
4260 * ConditionSecurity= now can check for "apparmor". We still
4261 are lacking checks for SMACK and IMA for this condition
4262 check though. Patches welcome!
4264 * A new configuration file /etc/systemd/sleep.conf has been
4265 added that may be used to configure which kernel operation
4266 systemd is supposed to execute when "suspend", "hibernate"
4267 or "hybrid-sleep" is requested. This makes the new kernel
4268 "freeze" state accessible to the user.
4270 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4271 the passed argument if applicable.
4273 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4274 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4275 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4276 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4277 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4278 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4279 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4284 * The output of 'systemctl list-jobs' got some polishing. The
4285 '--type=' argument may now be passed more than once. A new
4286 command 'systemctl list-sockets' has been added which shows
4287 a list of kernel sockets systemd is listening on with the
4288 socket units they belong to, plus the units these socket
4291 * The experimental libsystemd-bus library got substantial
4292 updates to work in conjunction with the (also experimental)
4293 kdbus kernel project. It works well enough to exchange
4294 messages with some sophistication. Note that kdbus is not
4295 ready yet, and the library is mostly an elaborate test case
4296 for now, and not installable.
4298 * systemd gained a new unit 'systemd-static-nodes.service'
4299 that generates static device nodes earlier during boot, and
4300 can run in conjunction with udev.
4302 * libsystemd-login gained a new call sd_pid_get_user_unit()
4303 to retrieve the user systemd unit a process is running
4304 in. This is useful for systems where systemd is used as
4307 * systemd-nspawn now places all containers in the new /machine
4308 top-level cgroup directory in the name=systemd
4309 hierarchy. libvirt will soon do the same, so that we get a
4310 uniform separation of /system, /user and /machine for system
4311 services, user processes and containers/virtual
4312 machines. This new cgroup hierarchy is also useful to stick
4313 stable names to specific container instances, which can be
4314 recognized later this way (this name may be controlled
4315 via systemd-nspawn's new -M switch). libsystemd-login also
4316 gained a new call sd_pid_get_machine_name() to retrieve the
4317 name of the container/VM a specific process belongs to.
4319 * bootchart can now store its data in the journal.
4321 * libsystemd-journal gained a new call
4322 sd_journal_add_conjunction() for AND expressions to the
4323 matching logic. This can be used to express more complex
4324 logical expressions.
4326 * journactl can now take multiple --unit= and --user-unit=
4329 * The cryptsetup logic now understands the "luks.key=" kernel
4330 command line switch for specifying a file to read the
4331 decryption key from. Also, if a configured key file is not
4332 found the tool will now automatically fall back to prompting
4335 * Python systemd.journal module was updated to wrap recently
4336 added functions from libsystemd-journal. The interface was
4337 changed to bring the low level interface in s.j._Reader
4338 closer to the C API, and the high level interface in
4339 s.j.Reader was updated to wrap and convert all data about
4342 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4343 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4344 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4345 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4346 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4347 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4351 * journalctl --update-catalog now understands a new --root=
4352 option to operate on catalogs found in a different root
4355 * During shutdown after systemd has terminated all running
4356 services a final killing loop kills all remaining left-over
4357 processes. We will now print the name of these processes
4358 when we send SIGKILL to them, since this usually indicates a
4361 * If /etc/crypttab refers to password files stored on
4362 configured mount points automatic dependencies will now be
4363 generated to ensure the specific mount is established first
4364 before the key file is attempted to be read.
4366 * 'systemctl status' will now show information about the
4367 network sockets a socket unit is listening on.
4369 * 'systemctl status' will also shown information about any
4370 drop-in configuration file for units. (Drop-In configuration
4371 files in this context are files such as
4372 /etc/systemd/systemd/foobar.service.d/*.conf)
4374 * systemd-cgtop now optionally shows summed up CPU times of
4375 cgroups. Press '%' while running cgtop to switch between
4376 percentage and absolute mode. This is useful to determine
4377 which cgroups use up the most CPU time over the entire
4378 runtime of the system. systemd-cgtop has also been updated
4379 to be 'pipeable' for processing with further shell tools.
4381 * 'hostnamectl set-hostname' will now allow setting of FQDN
4384 * The formatting and parsing of time span values has been
4385 changed. The parser now understands fractional expressions
4386 such as "5.5h". The formatter will now output fractional
4387 expressions for all time spans under 1min, i.e. "5.123456s"
4388 rather than "5s 123ms 456us". For time spans under 1s
4389 millisecond values are shown, for those under 1ms
4390 microsecond values are shown. This should greatly improve
4391 all time-related output of systemd.
4393 * libsystemd-login and libsystemd-journal gained new
4394 functions for querying the poll() events mask and poll()
4395 timeout value for integration into arbitrary event
4398 * localectl gained the ability to list available X11 keymaps
4399 (models, layouts, variants, options).
4401 * 'systemd-analyze dot' gained the ability to filter for
4402 specific units via shell-style globs, to create smaller,
4403 more useful graphs. I.e. it is now possible to create simple
4404 graphs of all the dependencies between only target units, or
4405 of all units that Avahi has dependencies with.
4407 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4408 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4409 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4410 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4411 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4412 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4413 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4417 * The boot-time readahead implementation for rotating media
4418 will now read the read-ahead data in multiple passes which
4419 consist of all read requests made in equidistant time
4420 intervals. This means instead of strictly reading read-ahead
4421 data in its physical order on disk we now try to find a
4422 middle ground between physical and access time order.
4424 * /etc/os-release files gained a new BUILD_ID= field for usage
4425 on operating systems that provide continuous builds of OS
4428 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4429 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4430 William Douglas, Zbigniew Jędrzejewski-Szmek
4434 * systemd-python gained an API exposing libsystemd-daemon.
4436 * The SMACK setup logic gained support for uploading CIPSO
4439 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4440 ReadOnlyDirectories= and InaccessibleDirectories= has
4441 changed. The private /tmp and /var/tmp directories are now
4442 shared by all processes of a service (which means
4443 ExecStartPre= may now leave data in /tmp that ExecStart= of
4444 the same service can still access). When a service is
4445 stopped its temporary directories are immediately deleted
4446 (normal clean-up with tmpfiles is still done in addition to
4449 * By default, systemd will now set a couple of sysctl
4450 variables in the kernel: the safe sysrq options are turned
4451 on, IP route verification is turned on, and source routing
4452 disabled. The recently added hardlink and softlink
4453 protection of the kernel is turned on. These settings should
4454 be reasonably safe, and good defaults for all new systems.
4456 * The predictable network naming logic may now be turned off
4457 with a new kernel command line switch: net.ifnames=0.
4459 * A new libsystemd-bus module has been added that implements a
4460 pretty complete D-Bus client library. For details see:
4462 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4464 * journald will now explicitly flush the journal files to disk
4465 at the latest 5min after each write. The file will then also
4466 be marked offline until the next write. This should increase
4467 reliability in case of a crash. The synchronization delay
4468 can be configured via SyncIntervalSec= in journald.conf.
4470 * There's a new remote-fs-setup.target unit that can be used
4471 to pull in specific services when at least one remote file
4472 system is to be mounted.
4474 * There are new targets timers.target and paths.target as
4475 canonical targets to pull user timer and path units in
4476 from. This complements sockets.target with a similar
4477 purpose for socket units.
4479 * libudev gained a new call udev_device_set_attribute_value()
4480 to set sysfs attributes of a device.
4482 * The udev daemon now sets the default number of worker
4483 processes executed in parallel based on the number of available
4484 CPUs instead of the amount of available RAM. This is supposed
4485 to provide a more reliable default and limit a too aggressive
4486 paralellism for setups with 1000s of devices connected.
4488 Contributions from: Auke Kok, Colin Walters, Cristian
4489 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4490 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4491 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4492 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4493 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4494 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4495 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4496 Zbigniew Jędrzejewski-Szmek
4500 * Configuration of unit files may now be extended via drop-in
4501 files without having to edit/override the unit files
4502 themselves. More specifically, if the administrator wants to
4503 change one value for a service file foobar.service he can
4504 now do so by dropping in a configuration snippet into
4505 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4506 will load all these snippets and apply them on top of the
4507 main unit configuration file, possibly extending or
4508 overriding its settings. Using these drop-in snippets is
4509 generally nicer than the two earlier options for changing
4510 unit files locally: copying the files from
4511 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4512 them there; or creating a new file in /etc/systemd/system/
4513 that incorporates the original one via ".include". Drop-in
4514 snippets into these .d/ directories can be placed in any
4515 directory systemd looks for units in, and the usual
4516 overriding semantics between /usr/lib, /etc and /run apply
4519 * Most unit file settings which take lists of items can now be
4520 reset by assigning the empty string to them. For example,
4521 normally, settings such as Environment=FOO=BAR append a new
4522 environment variable assignment to the environment block,
4523 each time they are used. By assigning Environment= the empty
4524 string the environment block can be reset to empty. This is
4525 particularly useful with the .d/*.conf drop-in snippets
4526 mentioned above, since this adds the ability to reset list
4527 settings from vendor unit files via these drop-ins.
4529 * systemctl gained a new "list-dependencies" command for
4530 listing the dependencies of a unit recursively.
4532 * Inhibitors are now honored and listed by "systemctl
4533 suspend", "systemctl poweroff" (and similar) too, not only
4534 GNOME. These commands will also list active sessions by
4537 * Resource limits (as exposed by the various control group
4538 controllers) can now be controlled dynamically at runtime
4539 for all units. More specifically, you can now use a command
4540 like "systemctl set-cgroup-attr foobar.service cpu.shares
4541 2000" to alter the CPU shares a specific service gets. These
4542 settings are stored persistently on disk, and thus allow the
4543 administrator to easily adjust the resource usage of
4544 services with a few simple commands. This dynamic resource
4545 management logic is also available to other programs via the
4546 bus. Almost any kernel cgroup attribute and controller is
4549 * systemd-vconsole-setup will now copy all font settings to
4550 all allocated VTs, where it previously applied them only to
4553 * libsystemd-login gained the new sd_session_get_tty() API
4556 * This release drops support for a few legacy or
4557 distribution-specific LSB facility names when parsing init
4558 scripts: $x-display-manager, $mail-transfer-agent,
4559 $mail-transport-agent, $mail-transfer-agent, $smtp,
4560 $null. Also, the mail-transfer-agent.target unit backing
4561 this has been removed. Distributions which want to retain
4562 compatibility with this should carry the burden for
4563 supporting this themselves and patch support for these back
4564 in, if they really need to. Also, the facilities $syslog and
4565 $local_fs are now ignored, since systemd does not support
4566 early-boot LSB init scripts anymore, and these facilities
4567 are implied anyway for normal services. syslog.target has
4570 * There are new bus calls on PID1's Manager object for
4571 cancelling jobs, and removing snapshot units. Previously,
4572 both calls were only available on the Job and Snapshot
4575 * systemd-journal-gatewayd gained SSL support.
4577 * The various "environment" files, such as /etc/locale.conf
4578 now support continuation lines with a backslash ("\") as
4579 last character in the line, similarly in style (but different)
4580 to how this is supported in shells.
4582 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4583 now implicitly appended to every log entry logged. systemctl
4584 has been updated to filter by this field when operating on a
4585 user systemd instance.
4587 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4588 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4589 the container. This makes it easier to boot unmodified
4590 Fedora systems in a container, which however still requires
4591 audit=0 to be passed on the kernel command line. Auditing in
4592 kernel and userspace is unfortunately still too broken in
4593 context of containers, hence we recommend compiling it out
4594 of the kernel or using audit=0. Hopefully this will be fixed
4595 one day for good in the kernel.
4597 * nspawn gained the new --bind= and --bind-ro= parameters to
4598 bind mount specific directories from the host into the
4601 * nspawn will now mount its own devpts file system instance
4602 into the container, in order not to leak pty devices from
4603 the host into the container.
4605 * systemd will now read the firmware boot time performance
4606 information from the EFI variables, if the used boot loader
4607 supports this, and takes it into account for boot performance
4608 analysis via "systemd-analyze". This is currently supported
4609 only in conjunction with Gummiboot, but could be supported
4610 by other boot loaders too. For details see:
4612 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4614 * A new generator has been added that automatically mounts the
4615 EFI System Partition (ESP) to /boot, if that directory
4616 exists, is empty, and no other file system has been
4617 configured to be mounted there.
4619 * logind will now send out PrepareForSleep(false) out
4620 unconditionally, after coming back from suspend. This may be
4621 used by applications as asynchronous notification for
4622 system resume events.
4624 * "systemctl unlock-sessions" has been added, that allows
4625 unlocking the screens of all user sessions at once, similar
4626 to how "systemctl lock-sessions" already locked all users
4627 sessions. This is backed by a new D-Bus call UnlockSessions().
4629 * "loginctl seat-status" will now show the master device of a
4630 seat. (i.e. the device of a seat that needs to be around for
4631 the seat to be considered available, usually the graphics
4634 * tmpfiles gained a new "X" line type, that allows
4635 configuration of files and directories (with wildcards) that
4636 shall be excluded from automatic cleanup ("aging").
4638 * udev default rules set the device node permissions now only
4639 at "add" events, and do not change them any longer with a
4640 later "change" event.
4642 * The log messages for lid events and power/sleep keypresses
4643 now carry a message ID.
4645 * We now have a substantially larger unit test suite, but this
4646 continues to be work in progress.
4648 * udevadm hwdb gained a new --root= parameter to change the
4649 root directory to operate relative to.
4651 * logind will now issue a background sync() request to the kernel
4652 early at shutdown, so that dirty buffers are flushed to disk early
4653 instead of at the last moment, in order to optimize shutdown
4656 * A new bootctl tool has been added that is an interface for
4657 certain boot loader operations. This is currently a preview
4658 and is likely to be extended into a small mechanism daemon
4659 like timedated, localed, hostnamed, and can be used by
4660 graphical UIs to enumerate available boot options, and
4661 request boot into firmware operations.
4663 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4664 the rest of the package. It also has been updated to work
4665 correctly in initrds.
4667 * Policykit previously has been runtime optional, and is now
4668 also compile time optional via a configure switch.
4670 * systemd-analyze has been reimplemented in C. Also "systemctl
4671 dot" has moved into systemd-analyze.
4673 * "systemctl status" with no further parameters will now print
4674 the status of all active or failed units.
4676 * Operations such as "systemctl start" can now be executed
4677 with a new mode "--irreversible" which may be used to queue
4678 operations that cannot accidentally be reversed by a later
4679 job queuing. This is by default used to make shutdown
4680 requests more robust.
4682 * The Python API of systemd now gained a new module for
4683 reading journal files.
4685 * A new tool kernel-install has been added that can install
4686 kernel images according to the Boot Loader Specification:
4688 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4690 * Boot time console output has been improved to provide
4691 animated boot time output for hanging jobs.
4693 * A new tool systemd-activate has been added which can be used
4694 to test socket activation with, directly from the command
4695 line. This should make it much easier to test and debug
4696 socket activation in daemons.
4698 * journalctl gained a new "--reverse" (or -r) option to show
4699 journal output in reverse order (i.e. newest line first).
4701 * journalctl gained a new "--pager-end" (or -e) option to jump
4702 to immediately jump to the end of the journal in the
4703 pager. This is only supported in conjunction with "less".
4705 * journalctl gained a new "--user-unit=" option, that works
4706 similarly to "--unit=" but filters for user units rather than
4709 * A number of unit files to ease adoption of systemd in
4710 initrds has been added. This moves some minimal logic from
4711 the various initrd implementations into systemd proper.
4713 * The journal files are now owned by a new group
4714 "systemd-journal", which exists specifically to allow access
4715 to the journal, and nothing else. Previously, we used the
4716 "adm" group for that, which however possibly covers more
4717 than just journal/log file access. This new group is now
4718 already used by systemd-journal-gatewayd to ensure this
4719 daemon gets access to the journal files and as little else
4720 as possible. Note that "make install" will also set FS ACLs
4721 up for /var/log/journal to give "adm" and "wheel" read
4722 access to it, in addition to "systemd-journal" which owns
4723 the journal files. We recommend that packaging scripts also
4724 add read access to "adm" + "wheel" to /var/log/journal, and
4725 all existing/future journal files. To normal users and
4726 administrators little changes, however packagers need to
4727 ensure to create the "systemd-journal" system group at
4728 package installation time.
4730 * The systemd-journal-gatewayd now runs as unprivileged user
4731 systemd-journal-gateway:systemd-journal-gateway. Packaging
4732 scripts need to create these system user/group at
4735 * timedated now exposes a new boolean property CanNTP that
4736 indicates whether a local NTP service is available or not.
4738 * systemd-detect-virt will now also detect xen PVs
4740 * The pstore file system is now mounted by default, if it is
4743 * In addition to the SELinux and IMA policies we will now also
4744 load SMACK policies at early boot.
4746 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4747 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4748 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4749 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4750 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4751 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4752 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4753 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4754 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4755 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4756 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4757 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4758 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4759 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4763 * Timer units now support calendar time events in addition to
4764 monotonic time events. That means you can now trigger a unit
4765 based on a calendar time specification such as "Thu,Fri
4766 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4767 or fifth day of any month of the year 2013, given that it is
4768 a thursday or friday. This brings timer event support
4769 considerably closer to cron's capabilities. For details on
4770 the supported calendar time specification language see
4773 * udev now supports a number of different naming policies for
4774 network interfaces for predictable names, and a combination
4775 of these policies is now the default. Please see this wiki
4776 document for details:
4778 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4780 * Auke Kok's bootchart implementation has been added to the
4781 systemd tree. It is an optional component that can graph the
4782 boot in quite some detail. It is one of the best bootchart
4783 implementations around and minimal in its code and
4786 * nss-myhostname has been integrated into the systemd source
4787 tree. nss-myhostname guarantees that the local hostname
4788 always stays resolvable via NSS. It has been a weak
4789 requirement of systemd-hostnamed since a long time, and
4790 since its code is actually trivial we decided to just
4791 include it in systemd's source tree. It can be turned off
4792 with a configure switch.
4794 * The read-ahead logic is now capable of properly detecting
4795 whether a btrfs file system is on SSD or rotating media, in
4796 order to optimize the read-ahead scheme. Previously, it was
4797 only capable of detecting this on traditional file systems
4800 * In udev, additional device properties are now read from the
4801 IAB in addition to the OUI database. Also, Bluetooth company
4802 identities are attached to the devices as well.
4804 * In service files %U may be used as specifier that is
4805 replaced by the configured user name of the service.
4807 * nspawn may now be invoked without a controlling TTY. This
4808 makes it suitable for invocation as its own service. This
4809 may be used to set up a simple containerized server system
4810 using only core OS tools.
4812 * systemd and nspawn can now accept socket file descriptors
4813 when they are started for socket activation. This enables
4814 implementation of socket activated nspawn
4815 containers. i.e. think about autospawning an entire OS image
4816 when the first SSH or HTTP connection is received. We expect
4817 that similar functionality will also be added to libvirt-lxc
4820 * journalctl will now suppress ANSI color codes when
4821 presenting log data.
4823 * systemctl will no longer show control group information for
4824 a unit if a the control group is empty anyway.
4826 * logind can now automatically suspend/hibernate/shutdown the
4829 * /etc/machine-info and hostnamed now also expose the chassis
4830 type of the system. This can be used to determine whether
4831 the local system is a laptop, desktop, handset or
4832 tablet. This information may either be configured by the
4833 user/vendor or is automatically determined from ACPI and DMI
4834 information if possible.
4836 * A number of PolicyKit actions are now bound together with
4837 "imply" rules. This should simplify creating UIs because
4838 many actions will now authenticate similar ones as well.
4840 * Unit files learnt a new condition ConditionACPower= which
4841 may be used to conditionalize a unit depending on whether an
4842 AC power source is connected or not, of whether the system
4843 is running on battery power.
4845 * systemctl gained a new "is-failed" verb that may be used in
4846 shell scripts and suchlike to check whether a specific unit
4847 is in the "failed" state.
4849 * The EnvironmentFile= setting in unit files now supports file
4850 globbing, and can hence be used to easily read a number of
4851 environment files at once.
4853 * systemd will no longer detect and recognize specific
4854 distributions. All distribution-specific #ifdeffery has been
4855 removed, systemd is now fully generic and
4856 distribution-agnostic. Effectively, not too much is lost as
4857 a lot of the code is still accessible via explicit configure
4858 switches. However, support for some distribution specific
4859 legacy configuration file formats has been dropped. We
4860 recommend distributions to simply adopt the configuration
4861 files everybody else uses now and convert the old
4862 configuration from packaging scripts. Most distributions
4863 already did that. If that's not possible or desirable,
4864 distributions are welcome to forward port the specific
4865 pieces of code locally from the git history.
4867 * When logging a message about a unit systemd will now always
4868 log the unit name in the message meta data.
4870 * localectl will now also discover system locale data that is
4871 not stored in locale archives, but directly unpacked.
4873 * logind will no longer unconditionally use framebuffer
4874 devices as seat masters, i.e. as devices that are required
4875 to be existing before a seat is considered preset. Instead,
4876 it will now look for all devices that are tagged as
4877 "seat-master" in udev. By default, framebuffer devices will
4878 be marked as such, but depending on local systems, other
4879 devices might be marked as well. This may be used to
4880 integrate graphics cards using closed source drivers (such
4881 as NVidia ones) more nicely into logind. Note however, that
4882 we recommend using the open source NVidia drivers instead,
4883 and no udev rules for the closed-source drivers will be
4884 shipped from us upstream.
4886 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4887 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4888 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4889 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4890 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4891 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4892 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4893 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4894 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4895 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4896 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4897 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4902 * udev gained support for loading additional device properties
4903 from an indexed database that is keyed by vendor/product IDs
4904 and similar device identifiers. For the beginning this
4905 "hwdb" is populated with data from the well-known PCI and
4906 USB database, but also includes PNP, ACPI and OID data. In
4907 the longer run this indexed database shall grow into
4908 becoming the one central database for non-essential
4909 userspace device metadata. Previously, data from the PCI/USB
4910 database was only attached to select devices, since the
4911 lookup was a relatively expensive operation due to O(n) time
4912 complexity (with n being the number of entries in the
4913 database). Since this is now O(1), we decided to add in this
4914 data for all devices where this is available, by
4915 default. Note that the indexed database needs to be rebuilt
4916 when new data files are installed. To achieve this you need
4917 to update your packaging scripts to invoke "udevadm hwdb
4918 --update" after installation of hwdb data files. For
4919 RPM-based distributions we introduced the new
4920 %udev_hwdb_update macro for this purpose.
4922 * The Journal gained support for the "Message Catalog", an
4923 indexed database to link up additional information with
4924 journal entries. For further details please check:
4926 http://www.freedesktop.org/wiki/Software/systemd/catalog
4928 The indexed message catalog database also needs to be
4929 rebuilt after installation of message catalog files. Use
4930 "journalctl --update-catalog" for this. For RPM-based
4931 distributions we introduced the %journal_catalog_update
4932 macro for this purpose.
4934 * The Python Journal bindings gained support for the standard
4935 Python logging framework.
4937 * The Journal API gained new functions for checking whether
4938 the underlying file system of a journal file is capable of
4939 properly reporting file change notifications, or whether
4940 applications that want to reflect journal changes "live"
4941 need to recheck journal files continuously in appropriate
4944 * It is now possible to set the "age" field for tmpfiles
4945 entries to 0, indicating that files matching this entry
4946 shall always be removed when the directories are cleaned up.
4948 * coredumpctl gained a new "gdb" verb which invokes gdb
4949 right-away on the selected coredump.
4951 * There's now support for "hybrid sleep" on kernels that
4952 support this, in addition to "suspend" and "hibernate". Use
4953 "systemctl hybrid-sleep" to make use of this.
4955 * logind's HandleSuspendKey= setting (and related settings)
4956 now gained support for a new "lock" setting to simply
4957 request the screen lock on all local sessions, instead of
4958 actually executing a suspend or hibernation.
4960 * systemd will now mount the EFI variables file system by
4963 * Socket units now gained support for configuration of the
4964 SMACK security label.
4966 * timedatectl will now output the time of the last and next
4967 daylight saving change.
4969 * We dropped support for various legacy and distro-specific
4970 concepts, such as insserv, early-boot SysV services
4971 (i.e. those for non-standard runlevels such as 'b' or 'S')
4972 or ArchLinux /etc/rc.conf support. We recommend the
4973 distributions who still need support this to either continue
4974 to maintain the necessary patches downstream, or find a
4975 different solution. (Talk to us if you have questions!)
4977 * Various systemd components will now bypass PolicyKit checks
4978 for root and otherwise handle properly if PolicyKit is not
4979 found to be around. This should fix most issues for
4980 PolicyKit-less systems. Quite frankly this should have been
4981 this way since day one. It is absolutely our intention to
4982 make systemd work fine on PolicyKit-less systems, and we
4983 consider it a bug if something does not work as it should if
4984 PolicyKit is not around.
4986 * For embedded systems it is now possible to build udev and
4987 systemd without blkid and/or kmod support.
4989 * "systemctl switch-root" is now capable of switching root
4990 more than once. I.e. in addition to transitions from the
4991 initrd to the host OS it is now possible to transition to
4992 further OS images from the host. This is useful to implement
4993 offline updating tools.
4995 * Various other additions have been made to the RPM macros
4996 shipped with systemd. Use %udev_rules_update() after
4997 installing new udev rules files. %_udevhwdbdir,
4998 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4999 %_sysctldir are now available which resolve to the right
5000 directories for packages to place various data files in.
5002 * journalctl gained the new --full switch (in addition to
5003 --all, to disable ellipsation for long messages.
5005 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5006 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5007 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5008 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5009 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5010 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5011 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5012 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5013 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5017 * journalctl gained new --since= and --until= switches to
5018 filter by time. It also now supports nice filtering for
5019 units via --unit=/-u.
5021 * Type=oneshot services may use ExecReload= and do the
5024 * The journal daemon now supports time-based rotation and
5025 vacuuming, in addition to the usual disk-space based
5028 * The journal will now index the available field values for
5029 each field name. This enables clients to show pretty drop
5030 downs of available match values when filtering. The bash
5031 completion of journalctl has been updated
5032 accordingly. journalctl gained a new switch -F to list all
5033 values a certain field takes in the journal database.
5035 * More service events are now written as structured messages
5036 to the journal, and made recognizable via message IDs.
5038 * The timedated, localed and hostnamed mini-services which
5039 previously only provided support for changing time, locale
5040 and hostname settings from graphical DEs such as GNOME now
5041 also have a minimal (but very useful) text-based client
5042 utility each. This is probably the nicest way to changing
5043 these settings from the command line now, especially since
5044 it lists available options and is fully integrated with bash
5047 * There's now a new tool "systemd-coredumpctl" to list and
5048 extract coredumps from the journal.
5050 * We now install a README each in /var/log/ and
5051 /etc/rc.d/init.d explaining where the system logs and init
5052 scripts went. This hopefully should help folks who go to
5053 that dirs and look into the otherwise now empty void and
5054 scratch their heads.
5056 * When user-services are invoked (by systemd --user) the
5057 $MANAGERPID env var is set to the PID of systemd.
5059 * SIGRTMIN+24 when sent to a --user instance will now result
5060 in immediate termination of systemd.
5062 * gatewayd received numerous feature additions such as a
5063 "follow" mode, for live syncing and filtering.
5065 * browse.html now allows filtering and showing detailed
5066 information on specific entries. Keyboard navigation and
5067 mouse screen support has been added.
5069 * gatewayd/journalctl now supports HTML5/JSON
5070 Server-Sent-Events as output.
5072 * The SysV init script compatibility logic will now
5073 heuristically determine whether a script supports the
5074 "reload" verb, and only then make this available as
5077 * "systemctl status --follow" has been removed, use "journalctl
5080 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5081 have been removed since they are hardly useful to be
5084 * And I'd like to take the opportunity to specifically mention
5085 Zbigniew for his great contributions. Zbigniew, you rock!
5087 Contributions from: Andrew Eikum, Christian Hesse, Colin
5088 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5089 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5090 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5091 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5092 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5093 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5097 * If /etc/vconsole.conf is non-existent or empty we will no
5098 longer load any console font or key map at boot by
5099 default. Instead the kernel defaults will be left
5100 intact. This is definitely the right thing to do, as no
5101 configuration should mean no configuration, and hard-coding
5102 font names that are different on all archs is probably a bad
5103 idea. Also, the kernel default key map and font should be
5104 good enough for most cases anyway, and mostly identical to
5105 the userspace fonts/key maps we previously overloaded them
5106 with. If distributions want to continue to default to a
5107 non-kernel font or key map they should ship a default
5108 /etc/vconsole.conf with the appropriate contents.
5110 Contributions from: Colin Walters, Daniel J Walsh, Dave
5111 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5112 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5116 * journalctl gained a new --cursor= switch to show entries
5117 starting from the specified location in the journal.
5119 * We now enforce a size limit on journal entry fields exported
5120 with "-o json" in journalctl. Fields larger than 4K will be
5121 assigned null. This can be turned off with --all.
5123 * An (optional) journal gateway daemon is now available as
5124 "systemd-journal-gatewayd.service". This service provides
5125 access to the journal via HTTP and JSON. This functionality
5126 will be used to implement live log synchronization in both
5127 pull and push modes, but has various other users too, such
5128 as easy log access for debugging of embedded devices. Right
5129 now it is already useful to retrieve the journal via HTTP:
5131 # systemctl start systemd-journal-gatewayd.service
5132 # wget http://localhost:19531/entries
5134 This will download the journal contents in a
5135 /var/log/messages compatible format. The same as JSON:
5137 # curl -H"Accept: application/json" http://localhost:19531/entries
5139 This service is also accessible via a web browser where a
5140 single static HTML5 app is served that uses the JSON logic
5141 to enable the user to do some basic browsing of the
5142 journal. This will be extended later on. Here's an example
5143 screenshot of this app in its current state:
5145 http://0pointer.de/public/journal-gatewayd
5147 Contributions from: Kay Sievers, Lennart Poettering, Robert
5148 Milasan, Tom Gundersen
5152 * The bash completion logic is now available for journalctl
5155 * We do not mount the "cpuset" controller anymore together with
5156 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5157 started if no parameters are assigned to it. "cpuset" hence
5158 broke code that assumed it it could create "cpu" groups and
5161 * journalctl -f will now subscribe to terminal size changes,
5162 and line break accordingly.
5164 Contributions from: Dave Reisner, Kay Sievers, Lennart
5165 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5169 * nspawn will now create a symlink /etc/localtime in the
5170 container environment, copying the host's timezone
5171 setting. Previously this has been done via a bind mount, but
5172 since symlinks cannot be bind mounted this has now been
5173 changed to create/update the appropriate symlink.
5175 * journalctl -n's line number argument is now optional, and
5176 will default to 10 if omitted.
5178 * journald will now log the maximum size the journal files may
5179 take up on disk. This is particularly useful if the default
5180 built-in logic of determining this parameter from the file
5181 system size is used. Use "systemctl status
5182 systemd-journald.service" to see this information.
5184 * The multi-seat X wrapper tool has been stripped down. As X
5185 is now capable of enumerating graphics devices via udev in a
5186 seat-aware way the wrapper is not strictly necessary
5187 anymore. A stripped down temporary stop-gap is still shipped
5188 until the upstream display managers have been updated to
5189 fully support the new X logic. Expect this wrapper to be
5190 removed entirely in one of the next releases.
5192 * HandleSleepKey= in logind.conf has been split up into
5193 HandleSuspendKey= and HandleHibernateKey=. The old setting
5194 is not available anymore. X11 and the kernel are
5195 distinguishing between these keys and we should too. This
5196 also means the inhibition lock for these keys has been split
5199 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5200 Poettering, Lukas Nykryn, Václav Pavlín
5204 * Whenever a unit changes state we will now log this to the
5205 journal and show along the unit's own log output in
5208 * ConditionPathIsMountPoint= can now properly detect bind
5209 mount points too. (Previously, a bind mount of one file
5210 system to another place in the same file system could not be
5211 detected as mount, since they shared struct stat's st_dev
5214 * We will now mount the cgroup controllers cpu, cpuacct,
5215 cpuset and the controllers net_cls, net_prio together by
5218 * nspawn containers will now have a virtualized boot
5219 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5220 over with a randomized ID at container initialization). This
5221 has the effect of making "journalctl -b" do the right thing
5224 * The JSON output journal serialization has been updated not
5225 to generate "endless" list objects anymore, but rather one
5226 JSON object per line. This is more in line how most JSON
5227 parsers expect JSON objects. The new output mode
5228 "json-pretty" has been added to provide similar output, but
5229 neatly aligned for readability by humans.
5231 * We dropped all explicit sync() invocations in the shutdown
5232 code. The kernel does this implicitly anyway in the kernel
5233 reboot() syscall. halt(8)'s -n option is now a compatibility
5236 * We now support virtualized reboot() in containers, as
5237 supported by newer kernels. We will fall back to exit() if
5238 CAP_SYS_REBOOT is not available to the container. Also,
5239 nspawn makes use of this now and will actually reboot the
5240 container if the containerized OS asks for that.
5242 * journalctl will only show local log output by default
5243 now. Use --merge (-m) to show remote log output, too.
5245 * libsystemd-journal gained the new sd_journal_get_usage()
5246 call to determine the current disk usage of all journal
5247 files. This is exposed in the new "journalctl --disk-usage"
5250 * journald gained a new configuration setting SplitMode= in
5251 journald.conf which may be used to control how user journals
5252 are split off. See journald.conf(5) for details.
5254 * A new condition type ConditionFileNotEmpty= has been added.
5256 * tmpfiles' "w" lines now support file globbing, to write
5257 multiple files at once.
5259 * We added Python bindings for the journal submission
5260 APIs. More Python APIs for a number of selected APIs will
5261 likely follow. Note that we intend to add native bindings
5262 only for the Python language, as we consider it common
5263 enough to deserve bindings shipped within systemd. There are
5264 various projects outside of systemd that provide bindings
5265 for languages such as PHP or Lua.
5267 * Many conditions will now resolve specifiers such as %i. In
5268 addition, PathChanged= and related directives of .path units
5269 now support specifiers as well.
5271 * There's now a new RPM macro definition for the system preset
5274 * journald will now warn if it ca not forward a message to the
5275 syslog daemon because its socket is full.
5277 * timedated will no longer write or process /etc/timezone,
5278 except on Debian. As we do not support late mounted /usr
5279 anymore /etc/localtime always being a symlink is now safe,
5280 and hence the information in /etc/timezone is not necessary
5283 * logind will now always reserve one VT for a text getty (VT6
5284 by default). Previously if more than 6 X sessions where
5285 started they took up all the VTs with auto-spawned gettys,
5286 so that no text gettys were available anymore.
5288 * udev will now automatically inform the btrfs kernel logic
5289 about btrfs RAID components showing up. This should make
5290 simple hotplug based btrfs RAID assembly work.
5292 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5293 (but not for its children which will stay at the kernel
5294 default). This should allow setups with a lot more listening
5297 * systemd will now always pass the configured timezone to the
5298 kernel at boot. timedated will do the same when the timezone
5301 * logind's inhibition logic has been updated. By default,
5302 logind will now handle the lid switch, the power and sleep
5303 keys all the time, even in graphical sessions. If DEs want
5304 to handle these events on their own they should take the new
5305 handle-power-key, handle-sleep-key and handle-lid-switch
5306 inhibitors during their runtime. A simple way to achieve
5307 that is to invoke the DE wrapped in an invocation of:
5309 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5311 * Access to unit operations is now checked via SELinux taking
5312 the unit file label and client process label into account.
5314 * systemd will now notify the administrator in the journal
5315 when he over-mounts a non-empty directory.
5317 * There are new specifiers that are resolved in unit files,
5318 for the host name (%H), the machine ID (%m) and the boot ID
5321 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5322 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5323 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5324 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5325 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5326 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5327 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5331 * Support for reading structured kernel messages from
5332 /dev/kmsg has now been added and is enabled by default.
5334 * Support for reading kernel messages from /proc/kmsg has now
5335 been removed. If you want kernel messages in the journal
5336 make sure to run a recent kernel (>= 3.5) that supports
5337 reading structured messages from /dev/kmsg (see
5338 above). /proc/kmsg is now exclusive property of classic
5339 syslog daemons again.
5341 * The libudev API gained the new
5342 udev_device_new_from_device_id() call.
5344 * The logic for file system namespace (ReadOnlyDirectory=,
5345 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5346 require pivot_root() anymore. This means fewer temporary
5347 directories are created below /tmp for this feature.
5349 * nspawn containers will now see and receive all submounts
5350 made on the host OS below the root file system of the
5353 * Forward Secure Sealing is now supported for Journal files,
5354 which provide cryptographical sealing of journal files so
5355 that attackers cannot alter log history anymore without this
5356 being detectable. Lennart will soon post a blog story about
5357 this explaining it in more detail.
5359 * There are two new service settings RestartPreventExitStatus=
5360 and SuccessExitStatus= which allow configuration of exit
5361 status (exit code or signal) which will be excepted from the
5362 restart logic, resp. consider successful.
5364 * journalctl gained the new --verify switch that can be used
5365 to check the integrity of the structure of journal files and
5366 (if Forward Secure Sealing is enabled) the contents of
5369 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5370 and similar symlinks pre-created. This makes running shells
5371 as container init process a lot more fun.
5373 * The fstab support can now handle PARTUUID= and PARTLABEL=
5376 * A new ConditionHost= condition has been added to match
5377 against the hostname (with globs) and machine ID. This is
5378 useful for clusters where a single OS image is used to
5379 provision a large number of hosts which shall run slightly
5380 different sets of services.
5382 * Services which hit the restart limit will now be placed in a
5385 Contributions from: Bertram Poettering, Dave Reisner, Huang
5386 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5387 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5391 * When running in --user mode systemd will now become a
5392 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5393 tree a lot more organized.
5395 * A new PartOf= unit dependency type has been introduced that
5396 may be used to group services in a natural way.
5398 * "systemctl enable" may now be used to enable instances of
5401 * journalctl now prints error log levels in red, and
5402 warning/notice log levels in bright white. It also supports
5403 filtering by log level now.
5405 * cgtop gained a new -n switch (similar to top), to configure
5406 the maximum number of iterations to run for. It also gained
5407 -b, to run in batch mode (accepting no input).
5409 * The suffix ".service" may now be omitted on most systemctl
5410 command lines involving service unit names.
5412 * There's a new bus call in logind to lock all sessions, as
5413 well as a loginctl verb for it "lock-sessions".
5415 * libsystemd-logind.so gained a new call sd_journal_perror()
5416 that works similar to libc perror() but logs to the journal
5417 and encodes structured information about the error number.
5419 * /etc/crypttab entries now understand the new keyfile-size=
5422 * shutdown(8) now can send a (configurable) wall message when
5423 a shutdown is cancelled.
5425 * The mount propagation mode for the root file system will now
5426 default to "shared", which is useful to make containers work
5427 nicely out-of-the-box so that they receive new mounts from
5428 the host. This can be undone locally by running "mount
5429 --make-rprivate /" if needed.
5431 * The prefdm.service file has been removed. Distributions
5432 should maintain this unit downstream if they intend to keep
5433 it around. However, we recommend writing normal unit files
5434 for display managers instead.
5436 * Since systemd is a crucial part of the OS we will now
5437 default to a number of compiler switches that improve
5438 security (hardening) such as read-only relocations, stack
5439 protection, and suchlike.
5441 * The TimeoutSec= setting for services is now split into
5442 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5443 of individual time outs for the start and the stop phase of
5446 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5447 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5448 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5449 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5450 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5451 Gundersen, Zbigniew Jędrzejewski-Szmek
5455 * The journal and id128 C APIs are now fully documented as man
5458 * Extra safety checks have been added when transitioning from
5459 the initial RAM disk to the main system to avoid accidental
5462 * /etc/crypttab entries now understand the new keyfile-offset=
5465 * systemctl -t can now be used to filter by unit load state.
5467 * The journal C API gained the new sd_journal_wait() call to
5468 make writing synchronous journal clients easier.
5470 * journalctl gained the new -D switch to show journals from a
5473 * journalctl now displays a special marker between log
5474 messages of two different boots.
5476 * The journal is now explicitly flushed to /var via a service
5477 systemd-journal-flush.service, rather than implicitly simply
5478 by seeing /var/log/journal to be writable.
5480 * journalctl (and the journal C APIs) can now match for much
5481 more complex expressions, with alternatives and
5484 * When transitioning from the initial RAM disk to the main
5485 system we will now kill all processes in a killing spree to
5486 ensure no processes stay around by accident.
5488 * Three new specifiers may be used in unit files: %u, %h, %s
5489 resolve to the user name, user home directory resp. user
5490 shell. This is useful for running systemd user instances.
5492 * We now automatically rotate journal files if their data
5493 object hash table gets a fill level > 75%. We also size the
5494 hash table based on the configured maximum file size. This
5495 together should lower hash collisions drastically and thus
5496 speed things up a bit.
5498 * journalctl gained the new "--header" switch to introspect
5499 header data of journal files.
5501 * A new setting SystemCallFilters= has been added to services
5502 which may be used to apply blacklists or whitelists to
5503 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5505 * nspawn gained a new --link-journal= switch (and quicker: -j)
5506 to link the container journal with the host. This makes it
5507 very easy to centralize log viewing on the host for all
5508 guests while still keeping the journal files separated.
5510 * Many bugfixes and optimizations
5512 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5513 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5514 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5519 * Several tools now understand kernel command line arguments,
5520 which are only read when run in an initial RAM disk. They
5521 usually follow closely their normal counterparts, but are
5524 * There's a new tool to analyze the readahead files that are
5525 automatically generated at boot. Use:
5527 /usr/lib/systemd/systemd-readahead analyze /.readahead
5529 * We now provide an early debug shell on tty9 if this enabled. Use:
5531 systemctl enable debug-shell.service
5533 * All plymouth related units have been moved into the Plymouth
5534 package. Please make sure to upgrade your Plymouth version
5537 * systemd-tmpfiles now supports getting passed the basename of
5538 a configuration file only, in which case it will look for it
5539 in all appropriate directories automatically.
5541 * udevadm info now takes a /dev or /sys path as argument, and
5542 does the right thing. Example:
5544 udevadm info /dev/sda
5545 udevadm info /sys/class/block/sda
5547 * systemctl now prints a warning if a unit is stopped but a
5548 unit that might trigger it continues to run. Example: a
5549 service is stopped but the socket that activates it is left
5552 * "systemctl status" will now mention if the log output was
5553 shortened due to rotation since a service has been started.
5555 * The journal API now exposes functions to determine the
5556 "cutoff" times due to rotation.
5558 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5559 immediately flushing of runtime logs to /var if possible,
5560 resp. for triggering immediate rotation of the journal
5563 * It is now considered an error if a service is attempted to
5564 be stopped that is not loaded.
5566 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5568 * systemd-analyze now supports Python 3
5570 * tmpfiles now supports cleaning up directories via aging
5571 where the first level dirs are always kept around but
5572 directories beneath it automatically aged. This is enabled
5573 by prefixing the age field with '~'.
5575 * Seat objects now expose CanGraphical, CanTTY properties
5576 which is required to deal with very fast bootups where the
5577 display manager might be running before the graphics drivers
5578 completed initialization.
5580 * Seat objects now expose a State property.
5582 * We now include RPM macros for service enabling/disabling
5583 based on the preset logic. We recommend RPM based
5584 distributions to make use of these macros if possible. This
5585 makes it simpler to reuse RPM spec files across
5588 * We now make sure that the collected systemd unit name is
5589 always valid when services log to the journal via
5592 * There's a new man page kernel-command-line(7) detailing all
5593 command line options we understand.
5595 * The fstab generator may now be disabled at boot by passing
5596 fstab=0 on the kernel command line.
5598 * A new kernel command line option modules-load= is now understood
5599 to load a specific kernel module statically, early at boot.
5601 * Unit names specified on the systemctl command line are now
5602 automatically escaped as needed. Also, if file system or
5603 device paths are specified they are automatically turned
5604 into the appropriate mount or device unit names. Example:
5606 systemctl status /home
5607 systemctl status /dev/sda
5609 * The SysVConsole= configuration option has been removed from
5610 system.conf parsing.
5612 * The SysV search path is no longer exported on the D-Bus
5615 * The Names= option is been removed from unit file parsing.
5617 * There's a new man page bootup(7) detailing the boot process.
5619 * Every unit and every generator we ship with systemd now
5620 comes with full documentation. The self-explanatory boot is
5623 * A couple of services gained "systemd-" prefixes in their
5624 name if they wrap systemd code, rather than only external
5625 code. Among them fsck@.service which is now
5626 systemd-fsck@.service.
5628 * The HaveWatchdog property has been removed from the D-Bus
5631 * systemd.confirm_spawn= on the kernel command line should now
5634 * There's a new man page crypttab(5) which details all options
5635 we actually understand.
5637 * systemd-nspawn gained a new --capability= switch to pass
5638 additional capabilities to the container.
5640 * timedated will now read known NTP implementation unit names
5641 from /usr/lib/systemd/ntp-units.d/*.list,
5642 systemd-timedated-ntp.target has been removed.
5644 * journalctl gained a new switch "-b" that lists log data of
5645 the current boot only.
5647 * The notify socket is in the abstract namespace again, in
5648 order to support daemons which chroot() at start-up.
5650 * There is a new Storage= configuration option for journald
5651 which allows configuration of where log data should go. This
5652 also provides a way to disable journal logging entirely, so
5653 that data collected is only forwarded to the console, the
5654 kernel log buffer or another syslog implementation.
5656 * Many bugfixes and optimizations
5658 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5659 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5660 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5661 Shawn Landden, Tom Gundersen
5665 * "systemctl help <unit>" now shows the man page if one is
5668 * Several new man pages have been added.
5670 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5671 MaxLevelConsole= can now be specified in
5672 journald.conf. These options allow reducing the amount of
5673 data stored on disk or forwarded by the log level.
5675 * TimerSlackNSec= can now be specified in system.conf for
5676 PID1. This allows system-wide power savings.
5678 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5679 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5684 * logind is now capable of (optionally) handling power and
5685 sleep keys as well as the lid switch.
5687 * journalctl now understands the syntax "journalctl
5688 /usr/bin/avahi-daemon" to get all log output of a specific
5691 * CapabilityBoundingSet= in system.conf now also influences
5692 the capability bound set of usermode helpers of the kernel.
5694 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5695 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5696 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5697 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5701 * Note that we skipped 139 releases here in order to set the
5702 new version to something that is greater than both udev's
5703 and systemd's most recent version number.
5705 * udev: all udev sources are merged into the systemd source tree now.
5706 All future udev development will happen in the systemd tree. It
5707 is still fully supported to use the udev daemon and tools without
5708 systemd running, like in initramfs or other init systems. Building
5709 udev though, will require the *build* of the systemd tree, but
5710 udev can be properly *run* without systemd.
5712 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5713 should be used to create dead device nodes as workarounds for broken
5716 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5717 no longer supported. udev_monitor_new_from_netlink() needs to be
5718 used to subscribe to events.
5720 * udev: when udevd is started by systemd, processes which are left
5721 behind by forking them off of udev rules, are unconditionally cleaned
5722 up and killed now after the event handling has finished. Services or
5723 daemons must be started as systemd services. Services can be
5724 pulled-in by udev to get started, but they can no longer be directly
5725 forked by udev rules.
5727 * udev: the daemon binary is called systemd-udevd now and installed
5728 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5729 to adapt to that, create symlink, or rename the binary after building
5732 * libudev no longer provides these symbols:
5733 udev_monitor_from_socket()
5734 udev_queue_get_failed_list_entry()
5735 udev_get_{dev,sys,run}_path()
5736 The versions number was bumped and symbol versioning introduced.
5738 * systemd-loginctl and systemd-journalctl have been renamed
5739 to loginctl and journalctl to match systemctl.
5741 * The config files: /etc/systemd/systemd-logind.conf and
5742 /etc/systemd/systemd-journald.conf have been renamed to
5743 logind.conf and journald.conf. Package updates should rename
5744 the files to the new names on upgrade.
5746 * For almost all files the license is now LGPL2.1+, changed
5747 from the previous GPL2.0+. Exceptions are some minor stuff
5748 of udev (which will be changed to LGPL2.1 eventually, too),
5749 and the MIT licensed sd-daemon.[ch] library that is suitable
5750 to be used as drop-in files.
5752 * systemd and logind now handle system sleep states, in
5753 particular suspending and hibernating.
5755 * logind now implements a sleep/shutdown/idle inhibiting logic
5756 suitable for a variety of uses. Soonishly Lennart will blog
5757 about this in more detail.
5759 * var-run.mount and var-lock.mount are no longer provided
5760 (which prevously bind mounted these directories to their new
5761 places). Distributions which have not converted these
5762 directories to symlinks should consider stealing these files
5763 from git history and add them downstream.
5765 * We introduced the Documentation= field for units and added
5766 this to all our shipped units. This is useful to make it
5767 easier to explore the boot and the purpose of the various
5770 * All smaller setup units (such as
5771 systemd-vconsole-setup.service) now detect properly if they
5772 are run in a container and are skipped when
5773 appropriate. This guarantees an entirely noise-free boot in
5774 Linux container environments such as systemd-nspawn.
5776 * A framework for implementing offline system updates is now
5777 integrated, for details see:
5778 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5780 * A new service type Type=idle is available now which helps us
5781 avoiding ugly interleaving of getty output and boot status
5784 * There's now a system-wide CapabilityBoundingSet= option to
5785 globally reduce the set of capabilities for the
5786 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5787 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5788 even CAP_NET_ADMIN system-wide for secure systems.
5790 * There are now system-wide DefaultLimitXXX= options to
5791 globally change the defaults of the various resource limits
5792 for all units started by PID 1.
5794 * Harald Hoyer's systemd test suite has been integrated into
5795 systemd which allows easy testing of systemd builds in qemu
5796 and nspawn. (This is really awesome! Ask us for details!)
5798 * The fstab parser is now implemented as generator, not inside
5801 * systemctl will now warn you if .mount units generated from
5802 /etc/fstab are out of date due to changes in fstab that
5803 have not been read by systemd yet.
5805 * systemd is now suitable for usage in initrds. Dracut has
5806 already been updated to make use of this. With this in place
5807 initrds get a slight bit faster but primarily are much
5808 easier to introspect and debug since "systemctl status" in
5809 the host system can be used to introspect initrd services,
5810 and the journal from the initrd is kept around too.
5812 * systemd-delta has been added, a tool to explore differences
5813 between user/admin configuration and vendor defaults.
5815 * PrivateTmp= now affects both /tmp and /var/tmp.
5817 * Boot time status messages are now much prettier and feature
5818 proper english language. Booting up systemd has never been
5821 * Read-ahead pack files now include the inode number of all
5822 files to pre-cache. When the inode changes the pre-caching
5823 is not attempted. This should be nicer to deal with updated
5824 packages which might result in changes of read-ahead
5827 * We now temporaritly lower the kernel's read_ahead_kb variable
5828 when collecting read-ahead data to ensure the kernel's
5829 built-in read-ahead does not add noise to our measurements
5830 of necessary blocks to pre-cache.
5832 * There's now RequiresMountsFor= to add automatic dependencies
5833 for all mounts necessary for a specific file system path.
5835 * MountAuto= and SwapAuto= have been removed from
5836 system.conf. Mounting file systems at boot has to take place
5839 * nspawn now learned a new switch --uuid= to set the machine
5840 ID on the command line.
5842 * nspawn now learned the -b switch to automatically search
5845 * vt102 is now the default TERM for serial TTYs, upgraded from
5848 * systemd-logind now works on VT-less systems.
5850 * The build tree has been reorganized. The individual
5851 components now have directories of their own.
5853 * A new condition type ConditionPathIsReadWrite= is now available.
5855 * nspawn learned the new -C switch to create cgroups for the
5856 container in other hierarchies.
5858 * We now have support for hardware watchdogs, configurable in
5861 * The scheduled shutdown logic now has a public API.
5863 * We now mount /tmp as tmpfs by default, but this can be
5864 masked and /etc/fstab can override it.
5866 * Since udisks does not make use of /media anymore we are not
5867 mounting a tmpfs on it anymore.
5869 * journalctl gained a new --local switch to only interleave
5870 locally generated journal files.
5872 * We can now load the IMA policy at boot automatically.
5874 * The GTK tools have been split off into a systemd-ui.
5876 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5877 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5878 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5879 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5880 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5881 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5882 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5883 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5884 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5889 * This is mostly a bugfix release
5891 * Support optional initialization of the machine ID from the
5892 KVM or container configured UUID.
5894 * Support immediate reboots with "systemctl reboot -ff"
5896 * Show /etc/os-release data in systemd-analyze output
5898 * Many bugfixes for the journal, including endianness fixes and
5899 ensuring that disk space enforcement works
5901 * sd-login.h is C++ comptaible again
5903 * Extend the /etc/os-release format on request of the Debian
5906 * We now refuse non-UTF8 strings used in various configuration
5907 and unit files. This is done to ensure we do not pass invalid
5908 data over D-Bus or expose it elsewhere.
5910 * Register Mimo USB Screens as suitable for automatic seat
5913 * Read SELinux client context from journal clients in a race
5916 * Reorder configuration file lookup order. /etc now always
5917 overrides /run in order to allow the administrator to always
5918 and unconditionally override vendor-supplied or
5919 automatically generated data.
5921 * The various user visible bits of the journal now have man
5922 pages. We still lack man pages for the journal API calls
5925 * We now ship all man pages in HTML format again in the
5928 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5929 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5930 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5935 * This is mostly a bugfix release
5937 * systems lacking /etc/os-release are no longer supported.
5939 * Various functionality updates to libsystemd-login.so
5941 * Track class of PAM logins to distinguish greeters from
5944 Contributions from: Kay Sievers, Lennart Poettering, Michael
5949 * This is an important bugfix release for v41.
5951 * Building man pages is now optional which should be useful
5952 for those building systemd from git but unwilling to install
5955 * Watchdog support for supervising services is now usable. In
5956 a future release support for hardware watchdogs
5957 (i.e. /dev/watchdog) will be added building on this.
5959 * Service start rate limiting is now configurable and can be
5960 turned off per service. When a start rate limit is hit a
5961 reboot can automatically be triggered.
5963 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5965 Contributions from: Benjamin Franzke, Bill Nottingham,
5966 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5967 Schmidt, Michał Górny, Piotr Drąg
5971 * The systemd binary is installed /usr/lib/systemd/systemd now;
5972 An existing /sbin/init symlink needs to be adapted with the
5975 * The code that loads kernel modules has been ported to invoke
5976 libkmod directly, instead of modprobe. This means we do not
5977 support systems with module-init-tools anymore.
5979 * Watchdog support is now already useful, but still not
5982 * A new kernel command line option systemd.setenv= is
5983 understood to set system wide environment variables
5984 dynamically at boot.
5986 * We now limit the set of capabilities of systemd-journald.
5988 * We now set SIGPIPE to ignore by default, since it only is
5989 useful in shell pipelines, and has little use in general
5990 code. This can be disabled with IgnoreSIPIPE=no in unit
5993 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5994 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5999 * This is mostly a bugfix release
6001 * We now expose the reason why a service failed in the
6002 "Result" D-Bus property.
6004 * Rudimentary service watchdog support (will be completed over
6005 the next few releases.)
6007 * When systemd forks off in order execute some service we will
6008 now immediately changes its argv[0] to reflect which process
6009 it will execute. This is useful to minimize the time window
6010 with a generic argv[0], which makes bootcharts more useful
6012 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6013 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6014 Mike Kazantsev, Ray Strode
6018 * This is mostly a test release, but incorporates many
6021 * New systemd-cgtop tool to show control groups by their
6024 * Linking against libacl for ACLs is optional again. If
6025 disabled, support tracking device access for active logins
6026 goes becomes unavailable, and so does access to the user
6027 journals by the respective users.
6029 * If a group "adm" exists, journal files are automatically
6030 owned by them, thus allow members of this group full access
6031 to the system journal as well as all user journals.
6033 * The journal now stores the SELinux context of the logging
6034 client for all entries.
6036 * Add C++ inclusion guards to all public headers
6038 * New output mode "cat" in the journal to print only text
6039 messages, without any meta data like date or time.
6041 * Include tiny X server wrapper as a temporary stop-gap to
6042 teach XOrg udev display enumeration. This is used by display
6043 managers such as gdm, and will go away as soon as XOrg
6044 learned native udev hotplugging for display devices.
6046 * Add new systemd-cat tool for executing arbitrary programs
6047 with STDERR/STDOUT connected to the journal. Can also act as
6048 BSD logger replacement, and does so by default.
6050 * Optionally store all locally generated coredumps in the
6051 journal along with meta data.
6053 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6054 writing short strings to files (for usage for /sys), and for
6055 creating symlinks, character and block device nodes.
6057 * New unit file option ControlGroupPersistent= to make cgroups
6058 persistent, following the mechanisms outlined in
6059 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6061 * Support multiple local RTCs in a sane way
6063 * No longer monopolize IO when replaying readahead data on
6064 rotating disks, since we might starve non-file-system IO to
6065 death, since fanotify() will not see accesses done by blkid,
6068 * Do not show kernel threads in systemd-cgls anymore, unless
6069 requested with new -k switch.
6071 Contributions from: Dan Horák, Kay Sievers, Lennart
6072 Poettering, Michal Schmidt
6076 * This is mostly a test release, but incorporates many
6079 * The git repository moved to:
6080 git://anongit.freedesktop.org/systemd/systemd
6081 ssh://git.freedesktop.org/git/systemd/systemd
6083 * First release with the journal
6084 http://0pointer.de/blog/projects/the-journal.html
6086 * The journal replaces both systemd-kmsg-syslogd and
6087 systemd-stdout-bridge.
6089 * New sd_pid_get_unit() API call in libsystemd-logind
6091 * Many systemadm clean-ups
6093 * Introduce remote-fs-pre.target which is ordered before all
6094 remote mounts and may be used to start services before all
6097 * Added Mageia support
6099 * Add bash completion for systemd-loginctl
6101 * Actively monitor PID file creation for daemons which exit in
6102 the parent process before having finished writing the PID
6103 file in the daemon process. Daemons which do this need to be
6104 fixed (i.e. PID file creation must have finished before the
6105 parent exits), but we now react a bit more gracefully to them.
6107 * Add colourful boot output, mimicking the well-known output
6108 of existing distributions.
6110 * New option PassCredentials= for socket units, for
6111 compatibility with a recent kernel ABI breakage.
6113 * /etc/rc.local is now hooked in via a generator binary, and
6114 thus will no longer act as synchronization point during
6117 * systemctl list-unit-files now supports --root=.
6119 * systemd-tmpfiles now understands two new commands: z, Z for
6120 relabelling files according to the SELinux database. This is
6121 useful to apply SELinux labels to specific files in /sys,
6124 * Output of SysV services is now forwarded to both the console
6125 and the journal by default, not only just the console.
6127 * New man pages for all APIs from libsystemd-login.
6129 * The build tree got reorganized and a the build system is a
6130 lot more modular allowing embedded setups to specifically
6131 select the components of systemd they are interested in.
6133 * Support for Linux systems lacking the kernel VT subsystem is
6136 * configure's --with-rootdir= got renamed to
6137 --with-rootprefix= to follow the naming used by udev and
6140 * Unless specified otherwise we will now install to /usr instead
6141 of /usr/local by default.
6143 * Processes with '@' in argv[0][0] are now excluded from the
6144 final shut-down killing spree, following the logic explained
6146 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6148 * All processes remaining in a service cgroup when we enter
6149 the START or START_PRE states are now killed with
6150 SIGKILL. That means it is no longer possible to spawn
6151 background processes from ExecStart= lines (which was never
6152 supported anyway, and bad style).
6154 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6155 reloading of units together.
6157 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6158 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6159 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6160 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6161 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob