1 systemd System and Service Manager
3 CHANGES WITH 230 in spe:
5 * DNSSEC is now turned on by default in systemd-resolved (in
6 "allow-downgrade" mode), but may be turned off during compile time by
7 passing "--with-default-dnssec=no" to "configure" (and of course,
8 during runtime with DNSSEC= in resolved.conf). We recommend
9 downstreams to leave this on at least during development cycles and
10 report any issues with the DNSSEC logic upstream. We are very
11 interested in collecting feedback about the DNSSEC validator and its
12 limitations in the wild. Note however, that DNSSEC support is
13 probably nothing downstreams should turn on in stable distros just
14 yet, as it might create incompatibilities with a few DNS servers and
15 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
16 automatically whenever we detect such incompatible setups, but there
17 might be systems we do not cover yet. Hence: please help us testing
18 the DNSSEC code, leave this on where you can, report back, but then
19 again don't consider turning this on in your stable, LTS or
20 production release just yet. (Note that you have to enable
21 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
22 and its DNSSEC mode for host name resolution from local
25 * systemd-resolve conveniently resolves DANE records with the --tlsa
26 option and OPENPGPKEY records with the --openpgp option. It also
27 supports dumping raw DNS record data via the new --raw= switch now.
29 * systemd-logind will now by default terminate user processes that are
30 part of the user session scope unit (session-XX.scope) when the user
31 logs out. This behavior is controlled by the KillUserProcesses=
32 setting in logind.conf, and the previous default of "no" is now
33 changed to "yes". This means that user sessions will be properly
34 cleaned up after, but additional steps are necessary to allow
35 intentionally long-running processes to survive logout.
37 While the user is logged in at least once, user@.service is running,
38 and any service that should survive the end of any individual login
39 session can be started at a user service or scope using systemd-run.
40 systemd-run(1) man page has been extended with an example which shows
41 how to run screen in a scope unit underneath user@.service. The same
42 command works for tmux.
44 After the user logs out of all sessions, user@.service will be
45 terminated too, by default, unless the user has "lingering" enabled.
46 To effectively allow users to run long-term tasks even if they are
47 logged out, lingering must be enabled for them. See loginctl(1) for
48 details. The default polkit policy was modified to allow users to
49 set lingering for themselves without authentication.
51 Previous defaults can be restored at compile time by the
52 --without-kill-user-processes option to "configure".
54 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
55 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
56 enable. Also, support for the "io" cgroup controller in the unified
57 hierarchy has been added, so that the "memory", "pids" and "io" are
58 now the controllers that are supported on the unified hierarchy.
60 WARNING: it is not possible to use previous systemd versions with
61 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
62 is necessary to also update systemd in the initramfs if using the
63 unified hierarchy. An updated SELinux policy is also required.
65 * LLDP support has been extended, and both passive (receive-only) and
66 active (sender) modes are supported. Passive mode ("routers-only") is
67 enabled by default in systemd-networkd. Active LLDP mode is enabled
68 by default for containers on the internal network. The "networkctl
69 lldp" command may be used to list information gathered. "networkctl
70 status" will also show basic LLDP information on connected peers now.
72 * The IAID and DUID unique identifier sent in DHCP requests may now be
73 configured for the system and each .network file managed by
76 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
77 systemd-socket-activate and installed into /usr/bin. It is now fully
80 * systemd-journald now uses separate threads to flush changes to disk
81 when closing journal files, thus reducing impact of slow disk I/O on
84 * systemd-ask-password now optionally skips printing of the password to
85 stdout with --no-output which can be useful in scripts.
87 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
88 (devices tagged with ID_MAKER_TOOL) are now tagged with
89 "uaccess" and are available to logged in users.
91 * systemd-bootchart has been split out to a separate repository:
92 https://github.com/systemd/systemd-bootchart
94 * The compatibility libraries libsystemd-daemon.so,
95 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
96 which have been deprecated since systemd-209 have been removed along
97 with the corresponding pkg-config files. All symbols provided by
98 those libraries are provided by libsystemd.so.
100 * The Capabilities= unit file setting has been removed (it is ignored
101 for backwards compatibility). AmbientCapabilities= and
102 CapabilityBoundingSet= should be used instead.
104 * "systemctl show" gained a new --value switch, which allows print a
105 only the contents of a specific unit property, without also printing
106 the property's name. Similar support was added to "show*" verbs
107 of loginctl and machinectl that output "key=value" lists.
109 * A new command "systemctl revert" has been added that may be used to
110 revert to the vendor version of a unit file, in case local changes
111 have been made by adding drop-ins or overriding the unit file.
113 * "machinectl clean" gained a new verb to automatically remove all or
114 just hidden container images.
116 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
117 merged into the kernel in its current form.
119 * systemd-networkd gained support for configuring proxy ARP support for
120 each interface, via the ProxyArp= setting in .network files. It also
121 gained support for configuring the multicast querier feature of
122 bridge devices, via the new MulticastQuerier= setting in .netdev
123 files. A new setting PreferredLifetime= has been added for addresses
124 configured in .network file to configure the lifetime intended for an
127 * systemd-tmpfiles gained support for a new line type "e" for emptying
128 directories, if they exist, without creating them if they don't.
130 * journalctl learned a new output mode "-o short-unix" that outputs log
131 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
132 UTC). It also gained support for a new --no-hostname setting to
133 suppress the hostname column in the family of "short" output modes.
135 * systemd-nspawn gained support for automatically patching the UID/GIDs
136 of the owners and the ACLs of all files and directories in a
137 container tree to match the UID/GID user namespacing range selected
138 for the container invocation. This mode is enabled via the new
139 --private-user-chown switch. It also gained support for automatically
140 choosing a free, previously unused UID/GID range when starting a
141 container, via the new --private-users=pick setting (which implies
142 --private-user-chown). Together, these options for the first time
143 make user namespacing for nspawn containers fully automatic and thus
144 deployable. The systemd-nspaw@.service template unit file has been
145 changed to use this functionality by default.
147 * systemd-nspawn gained a new --network-zone= switch, that allows
148 creating ad-hoc virtual Ethernet links between multiple containers,
149 that only exist as long as at least one container referencing them is
150 running. This allows easy connecting of multiple containers with a
151 common link that implements an Ethernet broadcast domain. Each of
152 these network "zones" may be named relatively freely by the user, and
153 may be referenced by any number of containers, but each container may
154 only reference one of these "zones". On the lower level, this is
155 implemented by an automatically managed bridge network interface for
156 each zone, that is created when the first container referencing its
157 zone is created and removed when the last one referencing its zone
160 * The default start timeout may now be configured on the kernel command
161 line via systemd.default_timeout_start_sec=. It was already
162 configurable via the DefaultTimeoutStartSec= option in
163 /etc/systemd/system.conf.
165 * Socket units gained a new TriggerLimitIntervalSec= and
166 TriggerLimitBurst= setting to configure a limit on the activation
167 rate of the socket unit.
169 * The LimitNICE= setting now optionally takes normal UNIX nice values
170 in addition to the raw integer limit value. If the specified
171 parameter is prefixed with "+" or "-" and is in the range -20..19 the
172 value is understood as UNIX nice value. If not prefixed like this it
173 is understood as raw RLIMIT_NICE limit.
175 * Note that the effect of the PrivateDevices= unit file setting changed
176 slightly with this release: the per-device /dev file system will be
177 mounted read-only from this version on, and will have "noexec"
178 set. This (minor) change of behaviour might cause some (exceptional)
179 legacy software to break, when PrivateDevices=yes is set for its
180 service. Please leave PrivateDevices= off if you run into problems
183 * The systemd-networkd DHCP server gained the option EmitRouter=, which
184 defaults to yes, to configure if the DHCP Option 3 (Router) should be
187 Contributions from: Alban Crequy, Alexander Kuleshov, Alex Crawford,
188 Andrew Eikum, Beniamino Galvani, Benjamin Robin, Benjamin ROBIN, Biao
189 Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens Gruber, Colin
190 Guthrie, Daniel J Walsh, Daniel Mack, Dan Nicholson, daurnimator, David
191 Herrmann, David R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko,
192 Evgeny Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger,
193 Franck Bui, frankheckenbach, Georgia Brikis, Harald Hoyer, Hendrik
194 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
195 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
196 kayrus, Klearchos Chaloulos, Lennart Poettering, Lubomir Rintel, Lukas
197 Nykryn, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
198 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
199 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
200 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
201 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Petr Lautrbach,
202 Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert Węcławski, Ronny
203 Chevalier, Samuel Tardieu, Stefan Schallenberg, Steven Siloti, Susant
204 Sahani, Sylvain Plantefève, Taylor Smock, tblume, Tejun Heo, Thomas
205 Blume, Thomas Haller, Thomas Hindoe Paaboel Andersen, Thomas
206 H. P. Andersen, Tobias Klauser, Tom Gundersen, Torstein Husebø, Umut
207 Tezduyar Lindskog, Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam),
208 Vladimir Panteleev, Wieland Hoffmann, Wouter Verhelst, Yu Watanabe,
209 Zbigniew Jędrzejewski-Szmek
215 * The systemd-resolved DNS resolver service has gained a substantial
216 set of new features, most prominently it may now act as a DNSSEC
217 validating stub resolver. DNSSEC mode is currently turned off by
218 default, but is expected to be turned on by default in one of the
219 next releases. For now, we invite everybody to test the DNSSEC logic
220 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
221 service also gained a full set of D-Bus interfaces, including calls
222 to configure DNS and DNSSEC settings per link (for use by external
223 network management software). systemd-resolved and systemd-networkd
224 now distinguish between "search" and "routing" domains. The former
225 are used to qualify single-label names, the latter are used purely
226 for routing lookups within certain domains to specific links.
227 resolved now also synthesizes RRs for all entries from /etc/hosts.
229 * The systemd-resolve tool (which is a client utility for
230 systemd-resolved) has been improved considerably and is now fully
231 supported and documented. Hence it has moved from /usr/lib/systemd to
234 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
237 * The coredump collection logic has been reworked: when a coredump is
238 collected it is now written to disk, compressed and processed
239 (including stacktrace extraction) from a new instantiated service
240 systemd-coredump@.service, instead of directly from the
241 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
242 processing large coredumps can take up a substantial amount of
243 resources and time, and this previously happened entirely outside of
244 systemd's service supervision. With the new logic the core_pattern
245 hook only does minimal metadata collection before passing off control
246 to the new instantiated service, which is configured with a time
247 limit, a nice level and other settings to minimize negative impact on
248 the rest of the system. Also note that the new logic will honour the
249 RLIMIT_CORE setting of the crashed process, which now allows users
250 and processes to turn off coredumping for their processes by setting
253 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
254 and all forked processes by default. Previously, PID 1 would leave
255 the setting at "0" for all processes, as set by the kernel. Note that
256 the resource limit traditionally has no effect on the generated
257 coredumps on the system if the /proc/sys/kernel/core_pattern hook
258 logic is used. Since the limit is now honoured (see above) its
259 default has been changed so that the coredumping logic is enabled by
260 default for all processes, while allowing specific opt-out.
262 * When the stacktrace is extracted from processes of system users, this
263 is now done as "systemd-coredump" user, in order to sandbox this
264 potentially security sensitive parsing operation. (Note that when
265 processing coredumps of normal users this is done under the user ID
266 of process that crashed, as before.) Packagers should take notice
267 that it is now necessary to create the "systemd-coredump" system user
268 and group at package installation time.
270 * The systemd-activate socket activation testing tool gained support
271 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
272 and --seqpacket switches. It also has been extended to support both
273 new-style and inetd-style file descriptor passing. Use the new
274 --inetd switch to request inetd-style file descriptor passing.
276 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
277 variable, which takes a boolean value. If set to false, ANSI color
278 output is disabled in the tools even when run on a terminal that
281 * The VXLAN support in networkd now supports two new settings
282 DestinationPort= and PortRange=.
284 * A new systemd.machine_id= kernel command line switch has been added,
285 that may be used to set the machine ID in /etc/machine-id if it is
286 not initialized yet. This command line option has no effect if the
287 file is already initialized.
289 * systemd-nspawn gained a new --as-pid2 switch that invokes any
290 specified command line as PID 2 rather than PID 1 in the
291 container. In this mode PID 1 is a minimal stub init process that
292 implements the special POSIX and Linux semantics of PID 1 regarding
293 signal and child process management. Note that this stub init process
294 is implemented in nspawn itself and requires no support from the
295 container image. This new logic is useful to support running
296 arbitrary commands in the container, as normal processes are
297 generally not prepared to run as PID 1.
299 * systemd-nspawn gained a new --chdir= switch for setting the current
300 working directory for the process started in the container.
302 * "journalctl /dev/sda" will now output all kernel log messages for
303 specified device from the current boot, in addition to all devices
304 that are parents of it. This should make log output about devices
305 pretty useful, as long as kernel drivers attach enough metadata to
306 the log messages. (The usual SATA drivers do.)
308 * The sd-journal API gained two new calls
309 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
310 that report whether log data from /run or /var has been found.
312 * journalctl gained a new switch "--fields" that prints all journal
313 record field names currently in use in the journal. This is backed
314 by two new sd-journal API calls sd_journal_enumerate_fields() and
315 sd_journal_restart_fields().
317 * Most configurable timeouts in systemd now expect an argument of
318 "infinity" to turn them off, instead of "0" as before. The semantics
319 from now on is that a timeout of "0" means "now", and "infinity"
320 means "never". To maintain backwards compatibility, "0" continues to
321 turn off previously existing timeout settings.
323 * "systemctl reload-or-try-restart" has been renamed to "systemctl
324 try-reload-or-restart" to clarify what it actually does: the "try"
325 logic applies to both reloading and restarting, not just restarting.
326 The old name continues to be accepted for compatibility.
328 * On boot-up, when PID 1 detects that the system clock is behind the
329 release date of the systemd version in use, the clock is now set
330 to the latter. Previously, this was already done in timesyncd, in order
331 to avoid running with clocks set to the various clock epochs such as
332 1902, 1938 or 1970. With this change the logic is now done in PID 1
333 in addition to timesyncd during early boot-up, so that it is enforced
334 before the first process is spawned by systemd. Note that the logic
335 in timesyncd remains, as it is more comprehensive and ensures
336 clock monotonicity by maintaining a persistent timestamp file in
337 /var. Since /var is generally not available in earliest boot or the
338 initrd, this part of the logic remains in timesyncd, and is not done
341 * Support for tweaking details in net_cls.class_id through the
342 NetClass= configuration directive has been removed, as the kernel
343 people have decided to deprecate that controller in cgroup v2.
344 Userspace tools such as nftables are moving over to setting rules
345 that are specific to the full cgroup path of a task, which obsoletes
346 these controllers anyway. The NetClass= directive is kept around for
347 legacy compatibility reasons. For a more in-depth description of the
348 kernel change, please refer to the respective upstream commit:
350 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
352 * A new service setting RuntimeMaxSec= has been added that may be used
353 to specify a maximum runtime for a service. If the timeout is hit, the
354 service is terminated and put into a failure state.
356 * A new service setting AmbientCapabilities= has been added. It allows
357 configuration of additional Linux process capabilities that are
358 passed to the activated processes. This is only available on very
361 * The process resource limit settings in service units may now be used
362 to configure hard and soft limits individually.
364 * The various libsystemd APIs such as sd-bus or sd-event now publicly
365 expose support for gcc's __attribute__((cleanup())) C extension.
366 Specifically, for many object destructor functions alternative
367 versions have been added that have names suffixed with "p" and take a
368 pointer to a pointer to the object to destroy, instead of just a
369 pointer to the object itself. This is useful because these destructor
370 functions may be used directly as parameters to the cleanup
371 construct. Internally, systemd has been a heavy user of this GCC
372 extension for a long time, and with this change similar support is
373 now available to consumers of the library outside of systemd. Note
374 that by using this extension in your sources compatibility with old
375 and strictly ANSI compatible C compilers is lost. However, all gcc or
376 LLVM versions of recent years support this extension.
378 * Timer units gained support for a new setting RandomizedDelaySec= that
379 allows configuring some additional randomized delay to the configured
380 time. This is useful to spread out timer events to avoid load peaks in
381 clusters or larger setups.
383 * Calendar time specifications now support sub-second accuracy.
385 * Socket units now support listening on SCTP and UDP-lite protocol
388 * The sd-event API now comes with a full set of man pages.
390 * Older versions of systemd contained experimental support for
391 compressing journal files and coredumps with the LZ4 compressor that
392 was not compatible with the lz4 binary (due to API limitations of the
393 lz4 library). This support has been removed; only support for files
394 compatible with the lz4 binary remains. This LZ4 logic is now
395 officially supported and no longer considered experimental.
397 * The dkr image import logic has been removed again from importd. dkr's
398 micro-services focus doesn't fit into the machine image focus of
399 importd, and quickly got out of date with the upstream dkr API.
401 * Creation of the /run/lock/lockdev/ directory was dropped from
402 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
403 been available for many years. If you still need this, you need to
404 create your own tmpfiles.d config file with:
406 d /run/lock/lockdev 0775 root lock -
408 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
409 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
410 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
411 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
412 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
413 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
414 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
415 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
416 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
417 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
418 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
419 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
420 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
421 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
422 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
423 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
424 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
425 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
426 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
432 * A number of properties previously only settable in unit
433 files are now also available as properties to set when
434 creating transient units programmatically via the bus, as it
435 is exposed with systemd-run's --property=
436 setting. Specifically, these are: SyslogIdentifier=,
437 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
438 EnvironmentFile=, ReadWriteDirectories=,
439 ReadOnlyDirectories=, InaccessibleDirectories=,
440 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
442 * When creating transient services via the bus API it is now
443 possible to pass in a set of file descriptors to use as
444 STDIN/STDOUT/STDERR for the invoked process.
446 * Slice units may now be created transiently via the bus APIs,
447 similar to the way service and scope units may already be
450 * Wherever systemd expects a calendar timestamp specification
451 (like in journalctl's --since= and --until= switches) UTC
452 timestamps are now supported. Timestamps suffixed with "UTC"
453 are now considered to be in Universal Time Coordinated
454 instead of the local timezone. Also, timestamps may now
455 optionally be specified with sub-second accuracy. Both of
456 these additions also apply to recurring calendar event
457 specification, such as OnCalendar= in timer units.
459 * journalctl gained a new "--sync" switch that asks the
460 journal daemon to write all so far unwritten log messages to
461 disk and sync the files, before returning.
463 * systemd-tmpfiles learned two new line types "q" and "Q" that
464 operate like "v", but also set up a basic btrfs quota
465 hierarchy when used on a btrfs file system with quota
468 * tmpfiles' "v", "q" and "Q" will now create a plain directory
469 instead of a subvolume (even on a btrfs file system) if the
470 root directory is a plain directory, and not a
471 subvolume. This should simplify things with certain chroot()
472 environments which are not aware of the concept of btrfs
475 * systemd-detect-virt gained a new --chroot switch to detect
476 whether execution takes place in a chroot() environment.
478 * CPUAffinity= now takes CPU index ranges in addition to
481 * The various memory-related resource limit settings (such as
482 LimitAS=) now understand the usual K, M, G, ... suffixes to
483 the base of 1024 (IEC). Similar, the time-related resource
484 limit settings understand the usual min, h, day, ...
487 * There's a new system.conf setting DefaultTasksMax= to
488 control the default TasksMax= setting for services and
489 scopes running on the system. (TasksMax= is the primary
490 setting that exposes the "pids" cgroup controller on systemd
491 and was introduced in the previous systemd release.) The
492 setting now defaults to 512, which means services that are
493 not explicitly configured otherwise will only be able to
494 create 512 processes or threads at maximum, from this
495 version on. Note that this means that thread- or
496 process-heavy services might need to be reconfigured to set
497 TasksMax= to a higher value. It is sufficient to set
498 TasksMax= in these specific unit files to a higher value, or
499 even "infinity". Similar, there's now a logind.conf setting
500 UserTasksMax= that defaults to 4096 and limits the total
501 number of processes or tasks each user may own
502 concurrently. nspawn containers also have the TasksMax=
503 value set by default now, to 8192. Note that all of this
504 only has an effect if the "pids" cgroup controller is
505 enabled in the kernel. The general benefit of these changes
506 should be a more robust and safer system, that provides a
507 certain amount of per-service fork() bomb protection.
509 * systemd-nspawn gained the new --network-veth-extra= switch
510 to define additional and arbitrarily-named virtual Ethernet
511 links between the host and the container.
513 * A new service execution setting PassEnvironment= has been
514 added that allows importing select environment variables
515 from PID1's environment block into the environment block of
518 * Timer units gained support for a new RemainAfterElapse=
519 setting which takes a boolean argument. It defaults on on,
520 exposing behaviour unchanged to previous releases. If set to
521 off, timer units are unloaded after they elapsed if they
522 cannot elapse again. This is particularly useful for
523 transient timer units, which shall not stay around longer
524 than until they first elapse.
526 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
527 default now (the kernel default is 16). This is beneficial
528 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
529 allows substantially larger numbers of queued
530 datagrams. This should increase the capability of systemd to
531 parallelize boot-up, as logging and sd_notify() are unlikely
532 to stall execution anymore. If you need to change the value
533 from the new defaults, use the usual sysctl.d/ snippets.
535 * The compression framing format used by the journal or
536 coredump processing has changed to be in line with what the
537 official LZ4 tools generate. LZ4 compression support in
538 systemd was considered unsupported previously, as the format
539 was not compatible with the normal tools. With this release
540 this has changed now, and it is hence safe for downstream
541 distributions to turn it on. While not compressing as well
542 as the XZ, LZ4 is substantially faster, which makes
543 it a good default choice for the compression logic in the
544 journal and in coredump handling.
546 * Any reference to /etc/mtab has been dropped from
547 systemd. The file has been obsolete since a while, but
548 systemd refused to work on systems where it was incorrectly
549 set up (it should be a symlink or non-existent). Please make
550 sure to update to util-linux 2.27.1 or newer in conjunction
551 with this systemd release, which also drops any reference to
552 /etc/mtab. If you maintain a distribution make sure that no
553 software you package still references it, as this is a
554 likely source of bugs. There's also a glibc bug pending,
555 asking for removal of any reference to this obsolete file:
557 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
559 Note that only util-linux versions built with
560 --enable-libmount-force-mountinfo are supported.
562 * Support for the ".snapshot" unit type has been removed. This
563 feature turned out to be little useful and little used, and
564 has now been removed from the core and from systemctl.
566 * The dependency types RequiresOverridable= and
567 RequisiteOverridable= have been removed from systemd. They
568 have been used only very sparingly to our knowledge and
569 other options that provide a similar effect (such as
570 systemctl --mode=ignore-dependencies) are much more useful
571 and commonly used. Moreover, they were only half-way
572 implemented as the option to control behaviour regarding
573 these dependencies was never added to systemctl. By removing
574 these dependency types the execution engine becomes a bit
575 simpler. Unit files that use these dependencies should be
576 changed to use the non-Overridable dependency types
577 instead. In fact, when parsing unit files with these
578 options, that's what systemd will automatically convert them
579 too, but it will also warn, asking users to fix the unit
580 files accordingly. Removal of these dependency types should
581 only affect a negligible number of unit files in the wild.
583 * Behaviour of networkd's IPForward= option changed
584 (again). It will no longer maintain a per-interface setting,
585 but propagate one way from interfaces where this is enabled
586 to the global kernel setting. The global setting will be
587 enabled when requested by a network that is set up, but
588 never be disabled again. This change was made to make sure
589 IPv4 and IPv6 behaviour regarding packet forwarding is
590 similar (as the Linux IPv6 stack does not support
591 per-interface control of this setting) and to minimize
594 * In unit files the behaviour of %u, %U, %h, %s has
595 changed. These specifiers will now unconditionally resolve
596 to the various user database fields of the user that the
597 systemd instance is running as, instead of the user
598 configured in the specific unit via User=. Note that this
599 effectively doesn't change much, as resolving of these
600 specifiers was already turned off in the --system instance
601 of systemd, as we cannot do NSS lookups from PID 1. In the
602 --user instance of systemd these specifiers where correctly
603 resolved, but hardly made any sense, since the user instance
604 lacks privileges to do user switches anyway, and User= is
605 hence useless. Morever, even in the --user instance of
606 systemd behaviour was awkward as it would only take settings
607 from User= assignment placed before the specifier into
608 account. In order to unify and simplify the logic around
609 this the specifiers will now always resolve to the
610 credentials of the user invoking the manager (which in case
611 of PID 1 is the root user).
613 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
614 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
615 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
616 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
617 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
618 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
619 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
620 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
621 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
622 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
623 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
624 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
625 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
626 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
633 * systemd now depends on util-linux v2.27. More specifically,
634 the newly added mount monitor feature in libmount now
635 replaces systemd's former own implementation.
637 * libmount mandates /etc/mtab not to be regular file, and
638 systemd now enforces this condition at early boot.
639 /etc/mtab has been deprecated and warned about for a very
640 long time, so systems running systemd should already have
641 stopped having this file around as anything else than a
642 symlink to /proc/self/mounts.
644 * Support for the "pids" cgroup controller has been added. It
645 allows accounting the number of tasks in a cgroup and
646 enforcing limits on it. This adds two new setting
647 TasksAccounting= and TasksMax= to each unit, as well as a
648 global option DefaultTasksAccounting=.
650 * Support for the "net_cls" cgroup controller has been added.
651 It allows assigning a net class ID to each task in the
652 cgroup, which can then be used in firewall rules and traffic
653 shaping configurations. Note that the kernel netfilter net
654 class code does not currently work reliably for ingress
655 packets on unestablished sockets.
657 This adds a new config directive called NetClass= to CGroup
658 enabled units. Allowed values are positive numbers for fixed
659 assignments and "auto" for picking a free value
662 * 'systemctl is-system-running' now returns 'offline' if the
663 system is not booted with systemd. This command can now be
664 used as a substitute for 'systemd-notify --booted'.
666 * Watchdog timeouts have been increased to 3 minutes for all
667 in-tree service files. Apparently, disk IO issues are more
668 frequent than we hoped, and user reported >1 minute waiting
671 * 'machine-id-commit' functionality has been merged into
672 'machine-id-setup --commit'. The separate binary has been
675 * The WorkingDirectory= directive in unit files may now be set
676 to the special value '~'. In this case, the working
677 directory is set to the home directory of the user
680 * "machinectl shell" will now open the shell in the home
681 directory of the selected user by default.
683 * The CrashChVT= configuration file setting is renamed to
684 CrashChangeVT=, following our usual logic of not
685 abbreviating unnecessarily. The old directive is still
686 supported for compat reasons. Also, this directive now takes
687 an integer value between 1 and 63, or a boolean value. The
688 formerly supported '-1' value for disabling stays around for
691 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
692 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
693 RootDirectory= properties can now be set for transient
696 * The systemd-analyze tool gained a new "set-log-target" verb
697 to change the logging target the system manager logs to
698 dynamically during runtime. This is similar to how
699 "systemd-analyze set-log-level" already changes the log
702 * In nspawn /sys is now mounted as tmpfs, with only a selected
703 set of subdirectories mounted in from the real sysfs. This
704 enhances security slightly, and is useful for ensuring user
705 namespaces work correctly.
707 * Support for USB FunctionFS activation has been added. This
708 allows implementation of USB gadget services that are
709 activated as soon as they are requested, so that they don't
710 have to run continously, similar to classic socket
713 * The "systemctl exit" command now optionally takes an
714 additional parameter that sets the exit code to return from
715 the systemd manager when exiting. This is only relevant when
716 running the systemd user instance, or when running the
717 system instance in a container.
719 * sd-bus gained the new API calls sd_bus_path_encode_many()
720 and sd_bus_path_decode_many() that allow easy encoding and
721 decoding of multiple identifier strings inside a D-Bus
722 object path. Another new call sd_bus_default_flush_close()
723 has been added to flush and close per-thread default
726 * systemd-cgtop gained support for a -M/--machine= switch to
727 show the control groups within a certain container only.
729 * "systemctl kill" gained support for an optional --fail
730 switch. If specified the requested operation will fail of no
731 processes have been killed, because the unit had no
732 processes attached, or similar.
734 * A new systemd.crash_reboot=1 kernel command line option has
735 been added that triggers a reboot after crashing. This can
736 also be set through CrashReboot= in systemd.conf.
738 * The RuntimeDirectory= setting now understands unit
739 specifiers like %i or %f.
741 * A new (still internal) libary API sd-ipv4acd has been added,
742 that implements address conflict detection for IPv4. It's
743 based on code from sd-ipv4ll, and will be useful for
744 detecting DHCP address conflicts.
746 * File descriptors passed during socket activation may now be
747 named. A new API sd_listen_fds_with_names() is added to
748 access the names. The default names may be overridden,
749 either in the .socket file using the FileDescriptorName=
750 parameter, or by passing FDNAME= when storing the file
751 descriptors using sd_notify().
753 * systemd-networkd gained support for:
755 - Setting the IPv6 Router Advertisment settings via
756 IPv6AcceptRouterAdvertisements= in .network files.
758 - Configuring the HelloTimeSec=, MaxAgeSec= and
759 ForwardDelaySec= bridge parameters in .netdev files.
761 - Configuring PreferredSource= for static routes in
764 * The "ask-password" framework used to query for LUKS harddisk
765 passwords or SSL passwords during boot gained support for
766 caching passwords in the kernel keyring, if it is
767 available. This makes sure that the user only has to type in
768 a passphrase once if there are multiple objects to unlock
769 with the same one. Previously, such password caching was
770 available only when Plymouth was used; this moves the
771 caching logic into the systemd codebase itself. The
772 "systemd-ask-password" utility gained a new --keyname=
773 switch to control which kernel keyring key to use for
774 caching a password in. This functionality is also useful for
775 enabling display managers such as gdm to automatically
776 unlock the user's GNOME keyring if its passphrase, the
777 user's password and the harddisk password are the same, if
778 gdm-autologin is used.
780 * When downloading tar or raw images using "machinectl
781 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
782 file is now also downloaded, if it is available and stored
783 next to the image file.
785 * Units of type ".socket" gained a new boolean setting
786 Writable= which is only useful in conjunction with
787 ListenSpecial=. If true, enables opening the specified
788 special file in O_RDWR mode rather than O_RDONLY mode.
790 * systemd-rfkill has been reworked to become a singleton
791 service that is activated through /dev/rfkill on each rfkill
792 state change and saves the settings to disk. This way,
793 systemd-rfkill is now compatible with devices that exist
794 only intermittendly, and even restores state if the previous
795 system shutdown was abrupt rather than clean.
797 * The journal daemon gained support for vacuuming old journal
798 files controlled by the number of files that shall remain,
799 in addition to the already existing control by size and by
800 date. This is useful as journal interleaving performance
801 degrades with too many seperate journal files, and allows
802 putting an effective limit on them. The new setting defaults
803 to 100, but this may be changed by setting SystemMaxFiles=
804 and RuntimeMaxFiles= in journald.conf. Also, the
805 "journalctl" tool gained the new --vacuum-files= switch to
806 manually vacuum journal files to leave only the specified
807 number of files in place.
809 * udev will now create /dev/disk/by-path links for ATA devices
810 on kernels where that is supported.
812 * Galician, Serbian, Turkish and Korean translations were added.
814 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
815 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
816 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
817 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
818 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
819 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
820 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
821 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
822 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
823 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
824 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
825 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
826 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
827 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
828 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
829 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
830 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
831 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
837 * The DHCP implementation of systemd-networkd gained a set of
840 - The DHCP server now supports emitting DNS and NTP
841 information. It may be enabled and configured via
842 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
843 and NTP information is enabled, but no servers are
844 configured, the corresponding uplink information (if there
845 is any) is propagated.
847 - Server and client now support transmission and reception
848 of timezone information. It can be configured via the
849 newly introduced network options UseTimezone=,
850 EmitTimezone=, and Timezone=. Transmission of timezone
851 information is enabled between host and containers by
852 default now: the container will change its local timezone
853 to what the host has set.
855 - Lease timeouts can now be configured via
856 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
858 - The DHCP server improved on the stability of
859 leases. Clients are more likely to get the same lease
860 information back, even if the server loses state.
862 - The DHCP server supports two new configuration options to
863 control the lease address pool metrics, PoolOffset= and
866 * The encapsulation limit of tunnels in systemd-networkd may
867 now be configured via 'EncapsulationLimit='. It allows
868 modifying the maximum additional levels of encapsulation
869 that are permitted to be prepended to a packet.
871 * systemd now supports the concept of user buses replacing
872 session buses, if used with dbus-1.10 (and enabled via dbus
873 --enable-user-session). It previously only supported this on
874 kdbus-enabled systems, and this release expands this to
875 'dbus-daemon' systems.
877 * systemd-networkd now supports predictable interface names
880 * systemd now optionally supports the new Linux kernel
881 "unified" control group hierarchy. If enabled via the kernel
882 command-line option 'systemd.unified_cgroup_hierarchy=1',
883 systemd will try to mount the unified cgroup hierarchy
884 directly on /sys/fs/cgroup. If not enabled, or not
885 available, systemd will fall back to the legacy cgroup
886 hierarchy setup, as before. Host system and containers can
887 mix and match legacy and unified hierarchies as they
888 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
889 environment variable to individually select the hierarchy to
890 use for executed containers. By default, nspawn will use the
891 unified hierarchy for the containers if the host uses the
892 unified hierarchy, and the legacy hierarchy otherwise.
893 Please note that at this point the unified hierarchy is an
894 experimental kernel feature and is likely to change in one
895 of the next kernel releases. Therefore, it should not be
896 enabled by default in downstream distributions yet. The
897 minimum required kernel version for the unified hierarchy to
898 work is 4.2. Note that when the unified hierarchy is used
899 for the first time delegated access to controllers is
900 safe. Because of this systemd-nspawn containers will get
901 access to controllers now, as will systemd user
902 sessions. This means containers and user sessions may now
903 manage their own resources, partitioning up what the system
906 * A new special scope unit "init.scope" has been introduced
907 that encapsulates PID 1 of the system. It may be used to
908 determine resource usage and enforce resource limits on PID
909 1 itself. PID 1 hence moved out of the root of the control
912 * The cgtop tool gained support for filtering out kernel
913 threads when counting tasks in a control group. Also, the
914 count of processes is now recursively summed up by
915 default. Two options -k and --recursive= have been added to
916 revert to old behaviour. The tool has also been updated to
917 work correctly in containers now.
919 * systemd-nspawn's --bind= and --bind-ro= options have been
920 extended to allow creation of non-recursive bind mounts.
922 * libsystemd gained two new calls sd_pid_get_cgroup() and
923 sd_peer_get_cgroup() which return the control group path of
924 a process or peer of a connected AF_UNIX socket. This
925 function call is particularly useful when implementing
926 delegated subtrees support in the control group hierarchy.
928 * The "sd-event" event loop API of libsystemd now supports
929 correct dequeuing of real-time signals, without losing
932 * When systemd requests a PolicyKit decision when managing
933 units it will now add additional fields to the request,
934 including unit name and desired operation. This enables more
935 powerful PolicyKit policies, that make decisions depending
938 * nspawn learnt support for .nspawn settings files, that may
939 accompany the image files or directories of containers, and
940 may contain additional settings for the container. This is
941 an alternative to configuring container parameters via the
944 Contributions from: Cristian Rodríguez, Daniel Mack, David
945 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
946 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
947 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
948 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
949 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
950 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
951 Andersen, Tom Gundersen, Torstein Husebø
957 * machinectl gained a new verb 'shell' which opens a fresh
958 shell on the target container or the host. It is similar to
959 the existing 'login' command of machinectl, but spawns the
960 shell directly without prompting for username or
961 password. The pseudo machine '.host' now refers to the local
962 host and is used by default. Hence, 'machinectl shell' can
963 be used as replacement for 'su -' which spawns a session as
964 a fresh systemd unit in a way that is fully isolated from
965 the originating session.
967 * systemd-networkd learned to cope with private-zone DHCP
968 options and allows other programs to query the values.
970 * SELinux access control when enabling/disabling units is no
971 longer enforced with this release. The previous
972 implementation was incorrect, and a new corrected
973 implementation is not yet available. As unit file operations
974 are still protected via PolicyKit and D-Bus policy this is
975 not a security problem. Yet, distributions which care about
976 optimal SELinux support should probably not stabilize on
979 * sd-bus gained support for matches of type "arg0has=", that
980 test for membership of strings in string arrays sent in bus
983 * systemd-resolved now dumps the contents of its DNS and LLMNR
984 caches to the logs on reception of the SIGUSR1 signal. This
985 is useful to debug DNS behaviour.
987 * The coredumpctl tool gained a new --directory= option to
988 operate on journal files in a specific directory.
990 * "systemctl reboot" and related commands gained a new
991 "--message=" option which may be used to set a free-text
992 wall message when shutting down or rebooting the
993 system. This message is also logged, which is useful for
994 figuring out the reason for a reboot or shutdown a
997 * The "systemd-resolve-host" tool's -i switch now takes
998 network interface numbers as alternative to interface names.
1000 * A new unit file setting for services has been introduced:
1001 UtmpMode= allows configuration of how precisely systemd
1002 handles utmp and wtmp entries for the service if this is
1003 enabled. This allows writing services that appear similar to
1004 user sessions in the output of the "w", "who", "last" and
1007 * systemd-resolved will now locally synthesize DNS resource
1008 records for the "localhost" and "gateway" domains as well as
1009 the local hostname. This should ensure that clients querying
1010 RRs via resolved will get similar results as those going via
1011 NSS, if nss-myhostname is enabled.
1013 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1014 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1015 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1016 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1017 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1018 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1019 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1020 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1021 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1022 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1023 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1024 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1026 — Berlin, 2015-08-27
1030 * The systemd-efi-boot-generator functionality was merged into
1031 systemd-gpt-auto-generator.
1033 * systemd-networkd now supports Group Policy for vxlan
1034 devices. It can be enabled via the new boolean configuration
1035 option called 'GroupPolicyExtension='.
1037 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1038 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1039 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1041 — Berlin, 2015-07-31
1045 * The python-systemd code has been removed from the systemd repository.
1046 A new repository has been created which accommodates the code from
1047 now on, and we kindly ask distributions to create a separate package
1048 for this: https://github.com/systemd/python-systemd
1050 * The systemd daemon will now reload its main configuration
1051 (/etc/systemd/system.conf) on daemon-reload.
1053 * sd-dhcp now exposes vendor specific extensions via
1054 sd_dhcp_lease_get_vendor_specific().
1056 * systemd-networkd gained a number of new configuration options.
1058 - A new boolean configuration option for TAP devices called
1059 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1060 device, thus allowing to send and receive GSO packets.
1062 - A new tunnel configuration option called 'CopyDSCP='.
1063 If enabled, the DSCP field of ip6 tunnels is copied into the
1064 decapsulated packet.
1066 - A set of boolean bridge configuration options were added.
1067 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1068 and 'UnicastFlood=' are now parsed by networkd and applied to the
1069 respective bridge link device via the respective IFLA_BRPORT_*
1072 - A new string configuration option to override the hostname sent
1073 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1074 is true, networkd will use the configured hostname instead of the
1075 system hostname when sending DHCP requests.
1077 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1078 networkd will configure the IPv6 flow-label of the tunnel device
1079 according to RFC2460.
1081 - The 'macvtap' virtual network devices are now supported, similar to
1082 the already supported 'macvlan' devices.
1084 * systemd-resolved now implements RFC5452 to improve resilience against
1085 cache poisoning. Additionally, source port randomization is enabled
1086 by default to further protect against DNS spoofing attacks.
1088 * nss-mymachines now supports translating UIDs and GIDs of running
1089 containers with user-namespaces enabled. If a container 'foo'
1090 translates a host uid 'UID' to the container uid 'TUID', then
1091 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1092 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1093 mapped as 'vg-foo-TGID'.
1095 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1096 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1097 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1098 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1099 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1100 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1101 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1102 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1103 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1104 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1106 — Berlin, 2015-07-29
1110 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1111 There are no known issues with current sysfs, and udev does not need
1112 or should be used to work around such bugs.
1114 * udev does no longer enable USB HID power management. Several reports
1115 indicate, that some devices cannot handle that setting.
1117 * The udev accelerometer helper was removed. The functionality
1118 is now fully included in iio-sensor-proxy. But this means,
1119 older iio-sensor-proxy versions will no longer provide
1120 accelerometer/orientation data with this systemd version.
1121 Please upgrade iio-sensor-proxy to version 1.0.
1123 * networkd gained a new configuration option IPv6PrivacyExtensions=
1124 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1125 for Stateless Address") on selected networks.
1127 * For the sake of fewer build-time dependencies and less code in the
1128 main repository, the python bindings are about to be removed in the
1129 next release. A new repository has been created which accommodates
1130 the code from now on, and we kindly ask distributions to create a
1131 separate package for this. The removal will take place in v223.
1133 https://github.com/systemd/python-systemd
1135 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1136 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1137 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1138 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1139 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1140 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1141 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1142 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1143 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1144 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1146 — Berlin, 2015-07-07
1150 * The sd-bus.h and sd-event.h APIs have now been declared
1151 stable and have been added to the official interface of
1152 libsystemd.so. sd-bus implements an alternative D-Bus client
1153 library, that is relatively easy to use, very efficient and
1154 supports both classic D-Bus as well as kdbus as transport
1155 backend. sd-event is a generic event loop abstraction that
1156 is built around Linux epoll, but adds features such as event
1157 prioritization or efficient timer handling. Both APIs are good
1158 choices for C programs looking for a bus and/or event loop
1159 implementation that is minimal and does not have to be
1160 portable to other kernels.
1162 * kdbus support is no longer compile-time optional. It is now
1163 always built-in. However, it can still be disabled at
1164 runtime using the kdbus=0 kernel command line setting, and
1165 that setting may be changed to default to off, by specifying
1166 --disable-kdbus at build-time. Note though that the kernel
1167 command line setting has no effect if the kdbus.ko kernel
1168 module is not installed, in which case kdbus is (obviously)
1169 also disabled. We encourage all downstream distributions to
1170 begin testing kdbus by adding it to the kernel images in the
1171 development distributions, and leaving kdbus support in
1174 * The minimal required util-linux version has been bumped to
1177 * Support for chkconfig (--enable-chkconfig) was removed in
1178 favor of calling an abstraction tool
1179 /lib/systemd/systemd-sysv-install. This needs to be
1180 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1181 in README for details.
1183 * If there's a systemd unit and a SysV init script for the
1184 same service name, and the user executes "systemctl enable"
1185 for it (or a related call), then this will now enable both
1186 (or execute the related operation on both), not just the
1189 * The libudev API documentation has been converted from gtkdoc
1192 * gudev has been removed from the systemd tree, it is now an
1195 * The systemd-cgtop tool learnt a new --raw switch to generate
1196 "raw" (machine parsable) output.
1198 * networkd's IPForwarding= .network file setting learnt the
1199 new setting "kernel", which ensures that networkd does not
1200 change the IP forwarding sysctl from the default kernel
1203 * The systemd-logind bus API now exposes a new boolean
1204 property "Docked" that reports whether logind considers the
1205 system "docked", i.e. connected to a docking station or not.
1207 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1208 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1209 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1210 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1211 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1212 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1213 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1214 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1215 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1216 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1217 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1218 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1219 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1220 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1221 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1222 Fink, Zbigniew Jędrzejewski-Szmek
1224 — Berlin, 2015-06-19
1228 * The gudev library has been extracted into a separate repository
1229 available at: https://git.gnome.org/browse/libgudev/
1230 It is now managed as part of the Gnome project. Distributions
1231 are recommended to pass --disable-gudev to systemd and use
1232 gudev from the Gnome project instead. gudev is still included
1233 in systemd, for now. It will be removed soon, though. Please
1234 also see the announcement-thread on systemd-devel:
1235 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1237 * systemd now exposes a CPUUsageNSec= property for each
1238 service unit on the bus, that contains the overall consumed
1239 CPU time of a service (the sum of what each process of the
1240 service consumed). This value is only available if
1241 CPUAccounting= is turned on for a service, and is then shown
1242 in the "systemctl status" output.
1244 * Support for configuring alternative mappings of the old SysV
1245 runlevels to systemd targets has been removed. They are now
1246 hardcoded in a way that runlevels 2, 3, 4 all map to
1247 multi-user.target and 5 to graphical.target (which
1248 previously was already the default behaviour).
1250 * The auto-mounter logic gained support for mount point
1251 expiry, using a new TimeoutIdleSec= setting in .automount
1252 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1254 * The EFI System Partition (ESP) as mounted to /boot by
1255 systemd-efi-boot-generator will now be unmounted
1256 automatically after 2 minutes of not being used. This should
1257 minimize the risk of ESP corruptions.
1259 * New /etc/fstab options x-systemd.requires= and
1260 x-systemd.requires-mounts-for= are now supported to express
1261 additional dependencies for mounts. This is useful for
1262 journalling file systems that support external journal
1263 devices or overlay file systems that require underlying file
1264 systems to be mounted.
1266 * systemd does not support direct live-upgrades (via systemctl
1267 daemon-reexec) from versions older than v44 anymore. As no
1268 distribution we are aware of shipped such old versions in a
1269 stable release this should not be problematic.
1271 * When systemd forks off a new per-connection service instance
1272 it will now set the $REMOTE_ADDR environment variable to the
1273 remote IP address, and $REMOTE_PORT environment variable to
1274 the remote IP port. This behaviour is similar to the
1275 corresponding environment variables defined by CGI.
1277 * systemd-networkd gained support for uplink failure
1278 detection. The BindCarrier= option allows binding interface
1279 configuration dynamically to the link sense of other
1280 interfaces. This is useful to achieve behaviour like in
1283 * systemd-networkd gained support for configuring the DHCP
1284 client identifier to use when requesting leases.
1286 * systemd-networkd now has a per-network UseNTP= option to
1287 configure whether NTP server information acquired via DHCP
1288 is passed on to services like systemd-timesyncd.
1290 * systemd-networkd gained support for vti6 tunnels.
1292 * Note that systemd-networkd manages the sysctl variable
1293 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1294 it is configured for since v219. The variable controls IP
1295 forwarding, and is a per-interface alternative to the global
1296 /proc/sys/net/ipv[46]/ip_forward. This setting is
1297 configurable in the IPForward= option, which defaults to
1298 "no". This means if networkd is used for an interface it is
1299 no longer sufficient to set the global sysctl option to turn
1300 on IP forwarding! Instead, the .network file option
1301 IPForward= needs to be turned on! Note that the
1302 implementation of this behaviour was broken in v219 and has
1305 * Many bonding and vxlan options are now configurable in
1308 * systemd-nspawn gained a new --property= setting to set unit
1309 properties for the container scope. This is useful for
1310 setting resource parameters (e.g "CPUShares=500") on
1311 containers started from the command line.
1313 * systemd-nspawn gained a new --private-users= switch to make
1314 use of user namespacing available on recent Linux kernels.
1316 * systemd-nspawn may now be called as part of a shell pipeline
1317 in which case the pipes used for stdin and stdout are passed
1318 directly to the process invoked in the container, without
1319 indirection via a pseudo tty.
1321 * systemd-nspawn gained a new switch to control the UNIX
1322 signal to use when killing the init process of the container
1325 * systemd-nspawn gained a new --overlay= switch for mounting
1326 overlay file systems into the container using the new kernel
1329 * When a container image is imported via systemd-importd and
1330 the host file system is not btrfs, a loopback block device
1331 file is created in /var/lib/machines.raw with a btrfs file
1332 system inside. It is then mounted to /var/lib/machines to
1333 enable btrfs features for container management. The loopback
1334 file and btrfs file system is grown as needed when container
1335 images are imported via systemd-importd.
1337 * systemd-machined/systemd-importd gained support for btrfs
1338 quota, to enforce container disk space limits on disk. This
1339 is exposed in "machinectl set-limit".
1341 * systemd-importd now can import containers from local .tar,
1342 .raw and .qcow2 images, and export them to .tar and .raw. It
1343 can also import dkr v2 images now from the network (on top
1346 * systemd-importd gained support for verifying downloaded
1347 images with gpg2 (previously only gpg1 was supported).
1349 * systemd-machined, systemd-logind, systemd: most bus calls
1350 are now accessible to unprivileged processes via
1351 PolicyKit. Also, systemd-logind will now allow users to kill
1352 their own sessions without further privileges or
1355 * systemd-shutdownd has been removed. This service was
1356 previously responsible for implementing scheduled shutdowns
1357 as exposed in /usr/bin/shutdown's time parameter. This
1358 functionality has now been moved into systemd-logind and is
1359 accessible via a bus interface.
1361 * "systemctl reboot" gained a new switch --firmware-setup that
1362 can be used to reboot into the EFI firmware setup, if that
1363 is available. systemd-logind now exposes an API on the bus
1364 to trigger such reboots, in case graphical desktop UIs want
1365 to cover this functionality.
1367 * "systemctl enable", "systemctl disable" and "systemctl mask"
1368 now support a new "--now" switch. If specified the units
1369 that are enabled will also be started, and the ones
1370 disabled/masked also stopped.
1372 * The Gummiboot EFI boot loader tool has been merged into
1373 systemd, and renamed to "systemd-boot". The bootctl tool has been
1374 updated to support systemd-boot.
1376 * An EFI kernel stub has been added that may be used to create
1377 kernel EFI binaries that contain not only the actual kernel,
1378 but also an initrd, boot splash, command line and OS release
1379 information. This combined binary can then be signed as a
1380 single image, so that the firmware can verify it all in one
1381 step. systemd-boot has special support for EFI binaries created
1382 like this and can extract OS release information from them
1383 and show them in the boot menu. This functionality is useful
1384 to implement cryptographically verified boot schemes.
1386 * Optional support has been added to systemd-fsck to pass
1387 fsck's progress report to an AF_UNIX socket in the file
1390 * udev will no longer create device symlinks for all block
1391 devices by default. A blacklist for excluding special block
1392 devices from this logic has been turned into a whitelist
1393 that requires picking block devices explicitly that require
1396 * A new (currently still internal) API sd-device.h has been
1397 added to libsystemd. This modernized API is supposed to
1398 replace libudev eventually. In fact, already much of libudev
1399 is now just a wrapper around sd-device.h.
1401 * A new hwdb database for storing metadata about pointing
1402 stick devices has been added.
1404 * systemd-tmpfiles gained support for setting file attributes
1405 similar to the "chattr" tool with new 'h' and 'H' lines.
1407 * systemd-journald will no longer unconditionally set the
1408 btrfs NOCOW flag on new journal files. This is instead done
1409 with tmpfiles snippet using the new 'h' line type. This
1410 allows easy disabling of this logic, by masking the
1411 journal-nocow.conf tmpfiles file.
1413 * systemd-journald will now translate audit message types to
1414 human readable identifiers when writing them to the
1415 journal. This should improve readability of audit messages.
1417 * The LUKS logic gained support for the offset= and skip=
1418 options in /etc/crypttab, as previously implemented by
1421 * /usr/lib/os-release gained a new optional field VARIANT= for
1422 distributions that support multiple variants (such as a
1423 desktop edition, a server edition, ...)
1425 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1426 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1427 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1428 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1429 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1430 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1431 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1432 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1433 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1434 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1435 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1436 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1437 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1438 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1439 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1440 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1441 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1442 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1443 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1444 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1445 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1446 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1447 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1448 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1449 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1450 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1451 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1453 — Berlin, 2015-05-22
1457 * Introduce a new API "sd-hwdb.h" for querying the hardware
1458 metadata database. With this minimal interface one can query
1459 and enumerate the udev hwdb, decoupled from the old libudev
1460 library. libudev's interface for this is now only a wrapper
1461 around sd-hwdb. A new tool systemd-hwdb has been added to
1462 interface with and update the database.
1464 * When any of systemd's tools copies files (for example due to
1465 tmpfiles' C lines) a btrfs reflink will attempted first,
1466 before bytewise copying is done.
1468 * systemd-nspawn gained a new --ephemeral switch. When
1469 specified a btrfs snapshot is taken of the container's root
1470 directory, and immediately removed when the container
1471 terminates again. Thus, a container can be started whose
1472 changes never alter the container's root directory, and are
1473 lost on container termination. This switch can also be used
1474 for starting a container off the root file system of the
1475 host without affecting the host OS. This switch is only
1476 available on btrfs file systems.
1478 * systemd-nspawn gained a new --template= switch. It takes the
1479 path to a container tree to use as template for the tree
1480 specified via --directory=, should that directory be
1481 missing. This allows instantiating containers dynamically,
1482 on first run. This switch is only available on btrfs file
1485 * When a .mount unit refers to a mount point on which multiple
1486 mounts are stacked, and the .mount unit is stopped all of
1487 the stacked mount points will now be unmounted until no
1488 mount point remains.
1490 * systemd now has an explicit notion of supported and
1491 unsupported unit types. Jobs enqueued for unsupported unit
1492 types will now fail with an "unsupported" error code. More
1493 specifically .swap, .automount and .device units are not
1494 supported in containers, .busname units are not supported on
1495 non-kdbus systems. .swap and .automount are also not
1496 supported if their respective kernel compile time options
1499 * machinectl gained support for two new "copy-from" and
1500 "copy-to" commands for copying files from a running
1501 container to the host or vice versa.
1503 * machinectl gained support for a new "bind" command to bind
1504 mount host directories into local containers. This is
1505 currently only supported for nspawn containers.
1507 * networkd gained support for configuring bridge forwarding
1508 database entries (fdb) from .network files.
1510 * A new tiny daemon "systemd-importd" has been added that can
1511 download container images in tar, raw, qcow2 or dkr formats,
1512 and make them available locally in /var/lib/machines, so
1513 that they can run as nspawn containers. The daemon can GPG
1514 verify the downloads (not supported for dkr, since it has no
1515 provisions for verifying downloads). It will transparently
1516 decompress bz2, xz, gzip compressed downloads if necessary,
1517 and restore sparse files on disk. The daemon uses privilege
1518 separation to ensure the actual download logic runs with
1519 fewer privileges than the daemon itself. machinectl has
1520 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1521 make the functionality of importd available to the
1522 user. With this in place the Fedora and Ubuntu "Cloud"
1523 images can be downloaded and booted as containers unmodified
1524 (the Fedora images lack the appropriate GPG signature files
1525 currently, so they cannot be verified, but this will change
1526 soon, hopefully). Note that downloading images is currently
1527 only fully supported on btrfs.
1529 * machinectl is now able to list container images found in
1530 /var/lib/machines, along with some metadata about sizes of
1531 disk and similar. If the directory is located on btrfs and
1532 quota is enabled, this includes quota display. A new command
1533 "image-status" has been added that shows additional
1534 information about images.
1536 * machinectl is now able to clone container images
1537 efficiently, if the underlying file system (btrfs) supports
1538 it, with the new "machinectl clone" command. It also
1539 gained commands for renaming and removing images, as well as
1540 marking them read-only or read-write (supported also on
1541 legacy file systems).
1543 * networkd gained support for collecting LLDP network
1544 announcements, from hardware that supports this. This is
1545 shown in networkctl output.
1547 * systemd-run gained support for a new -t (--pty) switch for
1548 invoking a binary on a pty whose input and output is
1549 connected to the invoking terminal. This allows executing
1550 processes as system services while interactively
1551 communicating with them via the terminal. Most interestingly
1552 this is supported across container boundaries. Invoking
1553 "systemd-run -t /bin/bash" is an alternative to running a
1554 full login session, the difference being that the former
1555 will not register a session, nor go through the PAM session
1558 * tmpfiles gained support for a new "v" line type for creating
1559 btrfs subvolumes. If the underlying file system is a legacy
1560 file system, this automatically degrades to creating a
1561 normal directory. Among others /var/lib/machines is now
1562 created like this at boot, should it be missing.
1564 * The directory /var/lib/containers/ has been deprecated and
1565 been replaced by /var/lib/machines. The term "machines" has
1566 been used in the systemd context as generic term for both
1567 VMs and containers, and hence appears more appropriate for
1568 this, as the directory can also contain raw images bootable
1571 * systemd-nspawn when invoked with -M but without --directory=
1572 or --image= is now capable of searching for the container
1573 root directory, subvolume or disk image automatically, in
1574 /var/lib/machines. systemd-nspawn@.service has been updated
1575 to make use of this, thus allowing it to be used for raw
1578 * A new machines.target unit has been introduced that is
1579 supposed to group all containers/VMs invoked as services on
1580 the system. systemd-nspawn@.service has been updated to
1581 integrate with that.
1583 * machinectl gained a new "start" command, for invoking a
1584 container as a service. "machinectl start foo" is mostly
1585 equivalent to "systemctl start systemd-nspawn@foo.service",
1586 but handles escaping in a nicer way.
1588 * systemd-nspawn will now mount most of the cgroupfs tree
1589 read-only into each container, with the exception of the
1590 container's own subtree in the name=systemd hierarchy.
1592 * journald now sets the special FS_NOCOW file flag for its
1593 journal files. This should improve performance on btrfs, by
1594 avoiding heavy fragmentation when journald's write-pattern
1595 is used on COW file systems. It degrades btrfs' data
1596 integrity guarantees for the files to the same levels as for
1597 ext3/ext4 however. This should be OK though as journald does
1598 its own data integrity checks and all its objects are
1599 checksummed on disk. Also, journald should handle btrfs disk
1600 full events a lot more gracefully now, by processing SIGBUS
1601 errors, and not relying on fallocate() anymore.
1603 * When journald detects that journal files it is writing to
1604 have been deleted it will immediately start new journal
1607 * systemd now provides a way to store file descriptors
1608 per-service in PID 1.This is useful for daemons to ensure
1609 that fds they require are not lost during a daemon
1610 restart. The fds are passed to the daemon on the next
1611 invocation in the same way socket activation fds are
1612 passed. This is now used by journald to ensure that the
1613 various sockets connected to all the system's stdout/stderr
1614 are not lost when journald is restarted. File descriptors
1615 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1616 an extension to sd_notify(). Note that a limit is enforced
1617 on the number of fds a service can store in PID 1, and it
1618 defaults to 0, so that no fds may be stored, unless this is
1619 explicitly turned on.
1621 * The default TERM variable to use for units connected to a
1622 terminal, when no other value is explicitly is set is now
1623 vt220 rather than vt102. This should be fairly safe still,
1624 but allows PgUp/PgDn work.
1626 * The /etc/crypttab option header= as known from Debian is now
1629 * "loginctl user-status" and "loginctl session-status" will
1630 now show the last 10 lines of log messages of the
1631 user/session following the status output. Similar,
1632 "machinectl status" will show the last 10 log lines
1633 associated with a virtual machine or container
1634 service. (Note that this is usually not the log messages
1635 done in the VM/container itself, but simply what the
1636 container manager logs. For nspawn this includes all console
1639 * "loginctl session-status" without further argument will now
1640 show the status of the session of the caller. Similar,
1641 "lock-session", "unlock-session", "activate",
1642 "enable-linger", "disable-linger" may now be called without
1643 session/user parameter in which case they apply to the
1644 caller's session/user.
1646 * An X11 session scriptlet is now shipped that uploads
1647 $DISPLAY and $XAUTHORITY into the environment of the systemd
1648 --user daemon if a session begins. This should improve
1649 compatibility with X11 enabled applications run as systemd
1652 * Generators are now subject to masking via /etc and /run, the
1653 same way as unit files.
1655 * networkd .network files gained support for configuring
1656 per-link IPv4/IPv6 packet forwarding as well as IPv4
1657 masquerading. This is by default turned on for veth links to
1658 containers, as registered by systemd-nspawn. This means that
1659 nspawn containers run with --network-veth will now get
1660 automatic routed access to the host's networks without any
1661 further configuration or setup, as long as networkd runs on
1664 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1665 or UDP posts of a container on the host. With this in place
1666 it is possible to run containers with private veth links
1667 (--network-veth), and have their functionality exposed on
1668 the host as if their services were running directly on the
1671 * systemd-nspawn's --network-veth switch now gained a short
1672 version "-n", since with the changes above it is now truly
1673 useful out-of-the-box. The systemd-nspawn@.service has been
1674 updated to make use of it too by default.
1676 * systemd-nspawn will now maintain a per-image R/W lock, to
1677 ensure that the same image is not started more than once
1678 writable. (It's OK to run an image multiple times
1679 simultaneously in read-only mode.)
1681 * systemd-nspawn's --image= option is now capable of
1682 dissecting and booting MBR and GPT disk images that contain
1683 only a single active Linux partition. Previously it
1684 supported only GPT disk images with proper GPT type
1685 IDs. This allows running cloud images from major
1686 distributions directly with systemd-nspawn, without
1689 * In addition to collecting mouse dpi data in the udev
1690 hardware database, there's now support for collecting angle
1691 information for mouse scroll wheels. The database is
1692 supposed to guarantee similar scrolling behavior on mice
1693 that it knows about. There's also support for collecting
1694 information about Touchpad types.
1696 * udev's input_id built-in will now also collect touch screen
1697 dimension data and attach it to probed devices.
1699 * /etc/os-release gained support for a Distribution Privacy
1702 * networkd gained support for creating "ipvlan", "gretap",
1703 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1705 * systemd-tmpfiles gained support for "a" lines for setting
1708 * systemd-nspawn will now mount /tmp in the container to
1709 tmpfs, automatically.
1711 * systemd now exposes the memory.usage_in_bytes cgroup
1712 attribute and shows it for each service in the "systemctl
1713 status" output, if available.
1715 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1716 immediate reboot is triggered. This useful if shutdown is
1717 hung and is unable to complete, to expedite the
1718 operation. Note that this kind of reboot will still unmount
1719 all file systems, and hence should not result in fsck being
1722 * A .device unit for an optical block device will now be
1723 considered active only when a medium is in the drive. Also,
1724 mount units are now bound to their backing devices thus
1725 triggering automatic unmounting when devices become
1726 unavailable. With this in place systemd will now
1727 automatically unmount left-over mounts when a CD-ROM is
1728 ejected or an USB stick is yanked from the system.
1730 * networkd-wait-online now has support for waiting for
1731 specific interfaces only (with globbing), and for giving up
1732 after a configurable timeout.
1734 * networkd now exits when idle. It will be automatically
1735 restarted as soon as interfaces show up, are removed or
1736 change state. networkd will stay around as long as there is
1737 at least one DHCP state machine or similar around, that keep
1740 * networkd may now configure IPv6 link-local addressing in
1741 addition to IPv4 link-local addressing.
1743 * The IPv6 "token" for use in SLAAC may now be configured for
1744 each .network interface in networkd.
1746 * Routes configured with networkd may now be assigned a scope
1749 * networkd's [Match] sections now support globbing and lists
1750 of multiple space-separated matches per item.
1752 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1753 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1754 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1755 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1756 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1757 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1758 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1759 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1760 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1761 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1762 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1763 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1764 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1765 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1766 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1767 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1768 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1769 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1770 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1771 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1772 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1773 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1774 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1775 Hoffmann, Zbigniew Jędrzejewski-Szmek
1777 — Berlin, 2015-02-16
1781 * When querying unit file enablement status (for example via
1782 "systemctl is-enabled"), a new state "indirect" is now known
1783 which indicates that a unit might not be enabled itself, but
1784 another unit listed in its Also= setting might be.
1786 * Similar to the various existing ConditionXYZ= settings for
1787 units, there are now matching AssertXYZ= settings. While
1788 failing conditions cause a unit to be skipped, but its job
1789 to succeed, failing assertions declared like this will cause
1790 a unit start operation and its job to fail.
1792 * hostnamed now knows a new chassis type "embedded".
1794 * systemctl gained a new "edit" command. When used on a unit
1795 file, this allows extending unit files with .d/ drop-in
1796 configuration snippets or editing the full file (after
1797 copying it from /usr/lib to /etc). This will invoke the
1798 user's editor (as configured with $EDITOR), and reload the
1799 modified configuration after editing.
1801 * "systemctl status" now shows the suggested enablement state
1802 for a unit, as declared in the (usually vendor-supplied)
1803 system preset files.
1805 * nss-myhostname will now resolve the single-label host name
1806 "gateway" to the locally configured default IP routing
1807 gateways, ordered by their metrics. This assigns a stable
1808 name to the used gateways, regardless which ones are
1809 currently configured. Note that the name will only be
1810 resolved after all other name sources (if nss-myhostname is
1811 configured properly) and should hence not negatively impact
1812 systems that use the single-label host name "gateway" in
1815 * systemd-inhibit now allows filtering by mode when listing
1818 * Scope and service units gained a new "Delegate" boolean
1819 property, which, when set, allows processes running inside the
1820 unit to further partition resources. This is primarily
1821 useful for systemd user instances as well as container
1824 * journald will now pick up audit messages directly from
1825 the kernel, and log them like any other log message. The
1826 audit fields are split up and fully indexed. This means that
1827 journalctl in many ways is now a (nicer!) alternative to
1828 ausearch, the traditional audit client. Note that this
1829 implements only a minimal audit client. If you want the
1830 special audit modes like reboot-on-log-overflow, please use
1831 the traditional auditd instead, which can be used in
1832 parallel to journald.
1834 * The ConditionSecurity= unit file option now understands the
1835 special string "audit" to check whether auditing is
1838 * journalctl gained two new commands --vacuum-size= and
1839 --vacuum-time= to delete old journal files until the
1840 remaining ones take up no more than the specified size on disk,
1841 or are not older than the specified time.
1843 * A new, native PPPoE library has been added to sd-network,
1844 systemd's library of light-weight networking protocols. This
1845 library will be used in a future version of networkd to
1846 enable PPPoE communication without an external pppd daemon.
1848 * The busctl tool now understands a new "capture" verb that
1849 works similar to "monitor", but writes a packet capture
1850 trace to STDOUT that can be redirected to a file which is
1851 compatible with libcap's capture file format. This can then
1852 be loaded in Wireshark and similar tools to inspect bus
1855 * The busctl tool now understands a new "tree" verb that shows
1856 the object trees of a specific service on the bus, or of all
1859 * The busctl tool now understands a new "introspect" verb that
1860 shows all interfaces and members of objects on the bus,
1861 including their signature and values. This is particularly
1862 useful to get more information about bus objects shown by
1863 the new "busctl tree" command.
1865 * The busctl tool now understands new verbs "call",
1866 "set-property" and "get-property" for invoking bus method
1867 calls, setting and getting bus object properties in a
1870 * busctl gained a new --augment-creds= argument that controls
1871 whether the tool shall augment credential information it
1872 gets from the bus with data from /proc, in a possibly
1875 * nspawn's --link-journal= switch gained two new values
1876 "try-guest" and "try-host" that work like "guest" and
1877 "host", but do not fail if the host has no persistent
1878 journalling enabled. -j is now equivalent to
1879 --link-journal=try-guest.
1881 * macvlan network devices created by nspawn will now have
1882 stable MAC addresses.
1884 * A new SmackProcessLabel= unit setting has been added, which
1885 controls the SMACK security label processes forked off by
1886 the respective unit shall use.
1888 * If compiled with --enable-xkbcommon, systemd-localed will
1889 verify x11 keymap settings by compiling the given keymap. It
1890 will spew out warnings if the compilation fails. This
1891 requires libxkbcommon to be installed.
1893 * When a coredump is collected, a larger number of metadata
1894 fields is now collected and included in the journal records
1895 created for it. More specifically, control group membership,
1896 environment variables, memory maps, working directory,
1897 chroot directory, /proc/$PID/status, and a list of open file
1898 descriptors is now stored in the log entry.
1900 * The udev hwdb now contains DPI information for mice. For
1903 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1905 * All systemd programs that read standalone configuration
1906 files in /etc now also support a corresponding series of
1907 .conf.d configuration directories in /etc/, /run/,
1908 /usr/local/lib/, /usr/lib/, and (if configured with
1909 --enable-split-usr) /lib/. In particular, the following
1910 configuration files now have corresponding configuration
1911 directories: system.conf user.conf, logind.conf,
1912 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1913 resolved.conf, timesyncd.conf, journal-remote.conf, and
1914 journal-upload.conf. Note that distributions should use the
1915 configuration directories in /usr/lib/; the directories in
1916 /etc/ are reserved for the system administrator.
1918 * systemd-rfkill will no longer take the rfkill device name
1919 into account when storing rfkill state on disk, as the name
1920 might be dynamically assigned and not stable. Instead, the
1921 ID_PATH udev variable combined with the rfkill type (wlan,
1922 bluetooth, ...) is used.
1924 * A new service systemd-machine-id-commit.service has been
1925 added. When used on systems where /etc is read-only during
1926 boot, and /etc/machine-id is not initialized (but an empty
1927 file), this service will copy the temporary machine ID
1928 created as replacement into /etc after the system is fully
1929 booted up. This is useful for systems that are freshly
1930 installed with a non-initialized machine ID, but should get
1931 a fixed machine ID for subsequent boots.
1933 * networkd's .netdev files now provide a large set of
1934 configuration parameters for VXLAN devices. Similarly, the
1935 bridge port cost parameter is now configurable in .network
1936 files. There's also new support for configuring IP source
1937 routing. networkd .link files gained support for a new
1938 OriginalName= match that is useful to match against the
1939 original interface name the kernel assigned. .network files
1940 may include MTU= and MACAddress= fields for altering the MTU
1941 and MAC address while being connected to a specific network
1944 * The LUKS logic gained supported for configuring
1945 UUID-specific key files. There's also new support for naming
1946 LUKS device from the kernel command line, using the new
1947 luks.name= argument.
1949 * Timer units may now be transiently created via the bus API
1950 (this was previously already available for scope and service
1951 units). In addition it is now possible to create multiple
1952 transient units at the same time with a single bus call. The
1953 "systemd-run" tool has been updated to make use of this for
1954 running commands on a specified time, in at(1)-style.
1956 * tmpfiles gained support for "t" lines, for assigning
1957 extended attributes to files. Among other uses this may be
1958 used to assign SMACK labels to files.
1960 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1961 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1962 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1963 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1964 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1965 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1966 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1967 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1968 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1969 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1970 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1971 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1972 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1973 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1974 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1975 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1976 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1977 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1979 — Berlin, 2014-12-10
1983 * journalctl gained the new options -t/--identifier= to match
1984 on the syslog identifier (aka "tag"), as well as --utc to
1985 show log timestamps in the UTC timezone. journalctl now also
1986 accepts -n/--lines=all to disable line capping in a pager.
1988 * journalctl gained a new switch, --flush, that synchronously
1989 flushes logs from /run/log/journal to /var/log/journal if
1990 persistent storage is enabled. systemd-journal-flush.service
1991 now waits until the operation is complete.
1993 * Services can notify the manager before they start a reload
1994 (by sending RELOADING=1) or shutdown (by sending
1995 STOPPING=1). This allows the manager to track and show the
1996 internal state of daemons and closes a race condition when
1997 the process is still running but has closed its D-Bus
2000 * Services with Type=oneshot do not have to have any ExecStart
2003 * User units are now loaded also from
2004 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2005 /run/systemd/user directory that was already previously
2006 supported, but is under the control of the user.
2008 * Job timeouts (i.e. time-outs on the time a job that is
2009 queued stays in the run queue) can now optionally result in
2010 immediate reboot or power-off actions (JobTimeoutAction= and
2011 JobTimeoutRebootArgument=). This is useful on ".target"
2012 units, to limit the maximum time a target remains
2013 undispatched in the run queue, and to trigger an emergency
2014 operation in such a case. This is now used by default to
2015 turn off the system if boot-up (as defined by everything in
2016 basic.target) hangs and does not complete for at least
2017 15min. Also, if power-off or reboot hang for at least 30min
2018 an immediate power-off/reboot operation is triggered. This
2019 functionality is particularly useful to increase reliability
2020 on embedded devices, but also on laptops which might
2021 accidentally get powered on when carried in a backpack and
2022 whose boot stays stuck in a hard disk encryption passphrase
2025 * systemd-logind can be configured to also handle lid switch
2026 events even when the machine is docked or multiple displays
2027 are attached (HandleLidSwitchDocked= option).
2029 * A helper binary and a service have been added which can be
2030 used to resume from hibernation in the initramfs. A
2031 generator will parse the resume= option on the kernel
2032 command line to trigger resume.
2034 * A user console daemon systemd-consoled has been
2035 added. Currently, it is a preview, and will so far open a
2036 single terminal on each session of the user marked as
2037 Desktop=systemd-console.
2039 * Route metrics can be specified for DHCP routes added by
2042 * The SELinux context of socket-activated services can be set
2043 from the information provided by the networking stack
2044 (SELinuxContextFromNet= option).
2046 * Userspace firmware loading support has been removed and
2047 the minimum supported kernel version is thus bumped to 3.7.
2049 * Timeout for udev workers has been increased from 1 to 3
2050 minutes, but a warning will be printed after 1 minute to
2051 help diagnose kernel modules that take a long time to load.
2053 * Udev rules can now remove tags on devices with TAG-="foobar".
2055 * systemd's readahead implementation has been removed. In many
2056 circumstances it didn't give expected benefits even for
2057 rotational disk drives and was becoming less relevant in the
2058 age of SSDs. As none of the developers has been using
2059 rotating media anymore, and nobody stepped up to actively
2060 maintain this component of systemd it has now been removed.
2062 * Swap units can use Options= to specify discard options.
2063 Discard options specified for swaps in /etc/fstab are now
2066 * Docker containers are now detected as a separate type of
2069 * The Password Agent protocol gained support for queries where
2070 the user input is shown, useful e.g. for user names.
2071 systemd-ask-password gained a new --echo option to turn that
2074 * The default sysctl.d/ snippets will now set:
2076 net.core.default_qdisc = fq_codel
2078 This selects Fair Queuing Controlled Delay as the default
2079 queuing discipline for network interfaces. fq_codel helps
2080 fight the network bufferbloat problem. It is believed to be
2081 a good default with no tuning required for most workloads.
2082 Downstream distributions may override this choice. On 10Gbit
2083 servers that do not do forwarding, "fq" may perform better.
2084 Systems without a good clocksource should use "pfifo_fast".
2086 * If kdbus is enabled during build a new option BusPolicy= is
2087 available for service units, that allows locking all service
2088 processes into a stricter bus policy, in order to limit
2089 access to various bus services, or even hide most of them
2090 from the service's view entirely.
2092 * networkctl will now show the .network and .link file
2093 networkd has applied to a specific interface.
2095 * sd-login gained a new API call sd_session_get_desktop() to
2096 query which desktop environment has been selected for a
2099 * UNIX utmp support is now compile-time optional to support
2100 legacy-free systems.
2102 * systemctl gained two new commands "add-wants" and
2103 "add-requires" for pulling in units from specific targets
2106 * If the word "rescue" is specified on the kernel command line
2107 the system will now boot into rescue mode (aka
2108 rescue.target), which was previously available only by
2109 specifying "1" or "systemd.unit=rescue.target" on the kernel
2110 command line. This new kernel command line option nicely
2111 mirrors the already existing "emergency" kernel command line
2114 * New kernel command line options mount.usr=, mount.usrflags=,
2115 mount.usrfstype= have been added that match root=, rootflags=,
2116 rootfstype= but allow mounting a specific file system to
2119 * The $NOTIFY_SOCKET is now also passed to control processes of
2120 services, not only the main process.
2122 * This version reenables support for fsck's -l switch. This
2123 means at least version v2.25 of util-linux is required for
2124 operation, otherwise dead-locks on device nodes may
2125 occur. Again: you need to update util-linux to at least
2126 v2.25 when updating systemd to v217.
2128 * The "multi-seat-x" tool has been removed from systemd, as
2129 its functionality has been integrated into X servers 1.16,
2130 and the tool is hence redundant. It is recommended to update
2131 display managers invoking this tool to simply invoke X
2132 directly from now on, again.
2134 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2135 message flag has been added for all of systemd's PolicyKit
2136 authenticated method calls has been added. In particular
2137 this now allows optional interactive authorization via
2138 PolicyKit for many of PID1's privileged operations such as
2139 unit file enabling and disabling.
2141 * "udevadm hwdb --update" learnt a new switch "--usr" for
2142 placing the rebuilt hardware database in /usr instead of
2143 /etc. When used only hardware database entries stored in
2144 /usr will be used, and any user database entries in /etc are
2145 ignored. This functionality is useful for vendors to ship a
2146 pre-built database on systems where local configuration is
2147 unnecessary or unlikely.
2149 * Calendar time specifications in .timer units now also
2150 understand the strings "semi-annually", "quarterly" and
2151 "minutely" as shortcuts (in addition to the preexisting
2152 "anually", "hourly", ...).
2154 * systemd-tmpfiles will now correctly create files in /dev
2155 at boot which are marked for creation only at boot. It is
2156 recommended to always create static device nodes with 'c!'
2157 and 'b!', so that they are created only at boot and not
2158 overwritten at runtime.
2160 * When the watchdog logic is used for a service (WatchdogSec=)
2161 and the watchdog timeout is hit the service will now be
2162 terminated with SIGABRT (instead of just SIGTERM), in order
2163 to make sure a proper coredump and backtrace is
2164 generated. This ensures that hanging services will result in
2165 similar coredump/backtrace behaviour as services that hit a
2168 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2169 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2170 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2171 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2172 Herrmann, David Sommerseth, David Strauss, Emil Renner
2173 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2174 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2175 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2176 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2177 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2178 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2179 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2180 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2181 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2182 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2183 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2184 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2185 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2186 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2187 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2188 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2191 — Berlin, 2014-10-28
2195 * timedated no longer reads NTP implementation unit names from
2196 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2197 implementations should add a
2199 Conflicts=systemd-timesyncd.service
2201 to their unit files to take over and replace systemd's NTP
2202 default functionality.
2204 * systemd-sysusers gained a new line type "r" for configuring
2205 which UID/GID ranges to allocate system users/groups
2206 from. Lines of type "u" may now add an additional column
2207 that specifies the home directory for the system user to be
2208 created. Also, systemd-sysusers may now optionally read user
2209 information from STDIN instead of a file. This is useful for
2210 invoking it from RPM preinst scriptlets that need to create
2211 users before the first RPM file is installed since these
2212 files might need to be owned by them. A new
2213 %sysusers_create_inline RPM macro has been introduced to do
2214 just that. systemd-sysusers now updates the shadow files as
2215 well as the user/group databases, which should enhance
2216 compatibility with certain tools like grpck.
2218 * A number of bus APIs of PID 1 now optionally consult
2219 PolicyKit to permit access for otherwise unprivileged
2220 clients under certain conditions. Note that this currently
2221 doesn't support interactive authentication yet, but this is
2222 expected to be added eventually, too.
2224 * /etc/machine-info now has new fields for configuring the
2225 deployment environment of the machine, as well as the
2226 location of the machine. hostnamectl has been updated with
2227 new command to update these fields.
2229 * systemd-timesyncd has been updated to automatically acquire
2230 NTP server information from systemd-networkd, which might
2231 have been discovered via DHCP.
2233 * systemd-resolved now includes a caching DNS stub resolver
2234 and a complete LLMNR name resolution implementation. A new
2235 NSS module "nss-resolve" has been added which can be used
2236 instead of glibc's own "nss-dns" to resolve hostnames via
2237 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2238 be resolved via systemd-resolved D-Bus APIs. In contrast to
2239 the glibc internal resolver systemd-resolved is aware of
2240 multi-homed system, and keeps DNS server and caches separate
2241 and per-interface. Queries are sent simultaneously on all
2242 interfaces that have DNS servers configured, in order to
2243 properly handle VPNs and local LANs which might resolve
2244 separate sets of domain names. systemd-resolved may acquire
2245 DNS server information from systemd-networkd automatically,
2246 which in turn might have discovered them via DHCP. A tool
2247 "systemd-resolve-host" has been added that may be used to
2248 query the DNS logic in resolved. systemd-resolved implements
2249 IDNA and automatically uses IDNA or UTF-8 encoding depending
2250 on whether classic DNS or LLMNR is used as transport. In the
2251 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2252 implementation to systemd-resolved.
2254 * A new NSS module nss-mymachines has been added, that
2255 automatically resolves the names of all local registered
2256 containers to their respective IP addresses.
2258 * A new client tool "networkctl" for systemd-networkd has been
2259 added. It currently is entirely passive and will query
2260 networking configuration from udev, rtnetlink and networkd,
2261 and present it to the user in a very friendly
2262 way. Eventually, we hope to extend it to become a full
2263 control utility for networkd.
2265 * .socket units gained a new DeferAcceptSec= setting that
2266 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2267 TCP. Similarly, support for controlling TCP keep-alive
2268 settings has been added (KeepAliveTimeSec=,
2269 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2270 turning off Nagle's algorithm on TCP has been added
2273 * logind learned a new session type "web", for use in projects
2274 like Cockpit which register web clients as PAM sessions.
2276 * timer units with at least one OnCalendar= setting will now
2277 be started only after timer-sync.target has been
2278 reached. This way they will not elapse before the system
2279 clock has been corrected by a local NTP client or
2280 similar. This is particular useful on RTC-less embedded
2281 machines, that come up with an invalid system clock.
2283 * systemd-nspawn's --network-veth= switch should now result in
2284 stable MAC addresses for both the outer and the inner side
2287 * systemd-nspawn gained a new --volatile= switch for running
2288 container instances with /etc or /var unpopulated.
2290 * The kdbus client code has been updated to use the new Linux
2291 3.17 memfd subsystem instead of the old kdbus-specific one.
2293 * systemd-networkd's DHCP client and server now support
2294 FORCERENEW. There are also new configuration options to
2295 configure the vendor client identifier and broadcast mode
2298 * systemd will no longer inform the kernel about the current
2299 timezone, as this is necessarily incorrect and racy as the
2300 kernel has no understanding of DST and similar
2301 concepts. This hence means FAT timestamps will be always
2302 considered UTC, similar to what Android is already
2303 doing. Also, when the RTC is configured to the local time
2304 (rather than UTC) systemd will never synchronize back to it,
2305 as this might confuse Windows at a later boot.
2307 * systemd-analyze gained a new command "verify" for offline
2308 validation of unit files.
2310 * systemd-networkd gained support for a couple of additional
2311 settings for bonding networking setups. Also, the metric for
2312 statically configured routes may now be configured. For
2313 network interfaces where this is appropriate the peer IP
2314 address may now be configured.
2316 * systemd-networkd's DHCP client will no longer request
2317 broadcasting by default, as this tripped up some networks.
2318 For hardware where broadcast is required the feature should
2319 be switched back on using RequestBroadcast=yes.
2321 * systemd-networkd will now set up IPv4LL addresses (when
2322 enabled) even if DHCP is configured successfully.
2324 * udev will now default to respect network device names given
2325 by the kernel when the kernel indicates that these are
2326 predictable. This behavior can be tweaked by changing
2327 NamePolicy= in the relevant .link file.
2329 * A new library systemd-terminal has been added that
2330 implements full TTY stream parsing and rendering. This
2331 library is supposed to be used later on for implementing a
2332 full userspace VT subsystem, replacing the current kernel
2335 * A new tool systemd-journal-upload has been added to push
2336 journal data to a remote system running
2337 systemd-journal-remote.
2339 * journald will no longer forward all local data to another
2340 running syslog daemon. This change has been made because
2341 rsyslog (which appears to be the most commonly used syslog
2342 implementation these days) no longer makes use of this, and
2343 instead pulls the data out of the journal on its own. Since
2344 forwarding the messages to a non-existent syslog server is
2345 more expensive than we assumed we have now turned this
2346 off. If you run a syslog server that is not a recent rsyslog
2347 version, you have to turn this option on again
2348 (ForwardToSyslog= in journald.conf).
2350 * journald now optionally supports the LZ4 compressor for
2351 larger journal fields. This compressor should perform much
2352 better than XZ which was the previous default.
2354 * machinectl now shows the IP addresses of local containers,
2355 if it knows them, plus the interface name of the container.
2357 * A new tool "systemd-escape" has been added that makes it
2358 easy to escape strings to build unit names and similar.
2360 * sd_notify() messages may now include a new ERRNO= field
2361 which is parsed and collected by systemd and shown among the
2362 "systemctl status" output for a service.
2364 * A new component "systemd-firstboot" has been added that
2365 queries the most basic systemd information (timezone,
2366 hostname, root password) interactively on first
2367 boot. Alternatively it may also be used to provision these
2368 things offline on OS images installed into directories.
2370 * The default sysctl.d/ snippets will now set
2372 net.ipv4.conf.default.promote_secondaries=1
2374 This has the benefit of no flushing secondary IP addresses
2375 when primary addresses are removed.
2377 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2378 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2379 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2380 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2381 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2382 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2383 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2384 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2385 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2386 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2387 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2388 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2389 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2390 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2391 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2393 — Berlin, 2014-08-19
2397 * A new tool systemd-sysusers has been added. This tool
2398 creates system users and groups in /etc/passwd and
2399 /etc/group, based on static declarative system user/group
2400 definitions in /usr/lib/sysusers.d/. This is useful to
2401 enable factory resets and volatile systems that boot up with
2402 an empty /etc directory, and thus need system users and
2403 groups created during early boot. systemd now also ships
2404 with two default sysusers.d/ files for the most basic
2405 users and groups systemd and the core operating system
2408 * A new tmpfiles snippet has been added that rebuilds the
2409 essential files in /etc on boot, should they be missing.
2411 * A directive for ensuring automatic clean-up of
2412 /var/cache/man/ has been removed from the default
2413 configuration. This line should now be shipped by the man
2414 implementation. The necessary change has been made to the
2415 man-db implementation. Note that you need to update your man
2416 implementation to one that ships this line, otherwise no
2417 automatic clean-up of /var/cache/man will take place.
2419 * A new condition ConditionNeedsUpdate= has been added that
2420 may conditionalize services to only run when /etc or /var
2421 are "older" than the vendor operating system resources in
2422 /usr. This is useful for reconstructing or updating /etc
2423 after an offline update of /usr or a factory reset, on the
2424 next reboot. Services that want to run once after such an
2425 update or reset should use this condition and order
2426 themselves before the new systemd-update-done.service, which
2427 will mark the two directories as fully updated. A number of
2428 service files have been added making use of this, to rebuild
2429 the udev hardware database, the journald message catalog and
2430 dynamic loader cache (ldconfig). The systemd-sysusers tool
2431 described above also makes use of this now. With this in
2432 place it is now possible to start up a minimal operating
2433 system with /etc empty cleanly. For more information on the
2434 concepts involved see this recent blog story:
2436 http://0pointer.de/blog/projects/stateless.html
2438 * A new system group "input" has been introduced, and all
2439 input device nodes get this group assigned. This is useful
2440 for system-level software to get access to input devices. It
2441 complements what is already done for "audio" and "video".
2443 * systemd-networkd learnt minimal DHCPv4 server support in
2444 addition to the existing DHCPv4 client support. It also
2445 learnt DHCPv6 client and IPv6 Router Solicitation client
2446 support. The DHCPv4 client gained support for static routes
2447 passed in from the server. Note that the [DHCPv4] section
2448 known in older systemd-networkd versions has been renamed to
2449 [DHCP] and is now also used by the DHCPv6 client. Existing
2450 .network files using settings of this section should be
2451 updated, though compatibility is maintained. Optionally, the
2452 client hostname may now be sent to the DHCP server.
2454 * networkd gained support for vxlan virtual networks as well
2455 as tun/tap and dummy devices.
2457 * networkd gained support for automatic allocation of address
2458 ranges for interfaces from a system-wide pool of
2459 addresses. This is useful for dynamically managing a large
2460 number of interfaces with a single network configuration
2461 file. In particular this is useful to easily assign
2462 appropriate IP addresses to the veth links of a large number
2463 of nspawn instances.
2465 * RPM macros for processing sysusers, sysctl and binfmt
2466 drop-in snippets at package installation time have been
2469 * The /etc/os-release file should now be placed in
2470 /usr/lib/os-release. The old location is automatically
2471 created as symlink. /usr/lib is the more appropriate
2472 location of this file, since it shall actually describe the
2473 vendor operating system shipped in /usr, and not the
2474 configuration stored in /etc.
2476 * .mount units gained a new boolean SloppyOptions= setting
2477 that maps to mount(8)'s -s option which enables permissive
2478 parsing of unknown mount options.
2480 * tmpfiles learnt a new "L+" directive which creates a symlink
2481 but (unlike "L") deletes a pre-existing file first, should
2482 it already exist and not already be the correct
2483 symlink. Similarly, "b+", "c+" and "p+" directives have been
2484 added as well, which create block and character devices, as
2485 well as fifos in the filesystem, possibly removing any
2486 pre-existing files of different types.
2488 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2489 'argument' field (which so far specified the source to
2490 symlink/copy the files from) is now optional. If omitted the
2491 same file os copied from /usr/share/factory/ suffixed by the
2492 full destination path. This is useful for populating /etc
2493 with essential files, by copying them from vendor defaults
2494 shipped in /usr/share/factory/etc.
2496 * A new command "systemctl preset-all" has been added that
2497 applies the service preset settings to all installed unit
2498 files. A new switch --preset-mode= has been added that
2499 controls whether only enable or only disable operations
2502 * A new command "systemctl is-system-running" has been added
2503 that allows checking the overall state of the system, for
2504 example whether it is fully up and running.
2506 * When the system boots up with an empty /etc, the equivalent
2507 to "systemctl preset-all" is executed during early boot, to
2508 make sure all default services are enabled after a factory
2511 * systemd now contains a minimal preset file that enables the
2512 most basic services systemd ships by default.
2514 * Unit files' [Install] section gained a new DefaultInstance=
2515 field for defining the default instance to create if a
2516 template unit is enabled with no instance specified.
2518 * A new passive target cryptsetup-pre.target has been added
2519 that may be used by services that need to make they run and
2520 finish before the first LUKS cryptographic device is set up.
2522 * The /dev/loop-control and /dev/btrfs-control device nodes
2523 are now owned by the "disk" group by default, opening up
2524 access to this group.
2526 * systemd-coredump will now automatically generate a
2527 stack trace of all core dumps taking place on the system,
2528 based on elfutils' libdw library. This stack trace is logged
2531 * systemd-coredump may now optionally store coredumps directly
2532 on disk (in /var/lib/systemd/coredump, possibly compressed),
2533 instead of storing them unconditionally in the journal. This
2534 mode is the new default. A new configuration file
2535 /etc/systemd/coredump.conf has been added to configure this
2536 and other parameters of systemd-coredump.
2538 * coredumpctl gained a new "info" verb to show details about a
2539 specific coredump. A new switch "-1" has also been added
2540 that makes sure to only show information about the most
2541 recent entry instead of all entries. Also, as the tool is
2542 generally useful now the "systemd-" prefix of the binary
2543 name has been removed. Distributions that want to maintain
2544 compatibility with the old name should add a symlink from
2545 the old name to the new name.
2547 * journald's SplitMode= now defaults to "uid". This makes sure
2548 that unprivileged users can access their own coredumps with
2549 coredumpctl without restrictions.
2551 * New kernel command line options "systemd.wants=" (for
2552 pulling an additional unit during boot), "systemd.mask="
2553 (for masking a specific unit for the boot), and
2554 "systemd.debug-shell" (for enabling the debug shell on tty9)
2555 have been added. This is implemented in the new generator
2556 "systemd-debug-generator".
2558 * systemd-nspawn will now by default filter a couple of
2559 syscalls for containers, among them those required for
2560 kernel module loading, direct x86 IO port access, swap
2561 management, and kexec. Most importantly though
2562 open_by_handle_at() is now prohibited for containers,
2563 closing a hole similar to a recently discussed vulnerability
2564 in docker regarding access to files on file hierarchies the
2565 container should normally not have access to. Note that, for
2566 nspawn, we generally make no security claims anyway (and
2567 this is explicitly documented in the man page), so this is
2568 just a fix for one of the most obvious problems.
2570 * A new man page file-hierarchy(7) has been added that
2571 contains a minimized, modernized version of the file system
2572 layout systemd expects, similar in style to the FHS
2573 specification or hier(5). A new tool systemd-path(1) has
2574 been added to query many of these paths for the local
2577 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2578 longer done. Since the directory now has a per-user size
2579 limit, and is cleaned on logout this appears unnecessary,
2580 in particular since this now brings the lifecycle of this
2581 directory closer in line with how IPC objects are handled.
2583 * systemd.pc now exports a number of additional directories,
2584 including $libdir (which is useful to identify the library
2585 path for the primary architecture of the system), and a
2586 couple of drop-in directories.
2588 * udev's predictable network interface names now use the dev_port
2589 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2590 distinguish between ports of the same PCI function. dev_id should
2591 only be used for ports using the same HW address, hence the need
2594 * machined has been updated to export the OS version of a
2595 container (read from /etc/os-release and
2596 /usr/lib/os-release) on the bus. This is now shown in
2597 "machinectl status" for a machine.
2599 * A new service setting RestartForceExitStatus= has been
2600 added. If configured to a set of exit signals or process
2601 return values, the service will be restarted when the main
2602 daemon process exits with any of them, regardless of the
2605 * systemctl's -H switch for connecting to remote systemd
2606 machines has been extended so that it may be used to
2607 directly connect to a specific container on the
2608 host. "systemctl -H root@foobar:waldi" will now connect as
2609 user "root" to host "foobar", and then proceed directly to
2610 the container named "waldi". Note that currently you have to
2611 authenticate as user "root" for this to work, as entering
2612 containers is a privileged operation.
2614 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2615 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2616 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2617 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2618 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2619 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2620 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2621 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2622 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2623 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2624 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2625 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2627 — Berlin, 2014-07-03
2631 * As an experimental feature, udev now tries to lock the
2632 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2633 executes events for the disk or any of its partitions.
2634 Applications like partitioning programs can lock the
2635 disk device node (flock(LOCK_EX)) and claim temporary
2636 device ownership that way; udev will entirely skip all event
2637 handling for this disk and its partitions. If the disk
2638 was opened for writing, the close will trigger a partition
2639 table rescan in udev's "watch" facility, and if needed
2640 synthesize "change" events for the disk and all its partitions.
2641 This is now unconditionally enabled, and if it turns out to
2642 cause major problems, we might turn it on only for specific
2643 devices, or might need to disable it entirely. Device Mapper
2644 devices are excluded from this logic.
2646 * We temporarily dropped the "-l" switch for fsck invocations,
2647 since they collide with the flock() logic above. util-linux
2648 upstream has been changed already to avoid this conflict,
2649 and we will readd "-l" as soon as util-linux with this
2650 change has been released.
2652 * The dependency on libattr has been removed. Since a long
2653 time, the extended attribute calls have moved to glibc, and
2654 libattr is thus unnecessary.
2656 * Virtualization detection works without priviliges now. This
2657 means the systemd-detect-virt binary no longer requires
2658 CAP_SYS_PTRACE file capabilities, and our daemons can run
2659 with fewer privileges.
2661 * systemd-networkd now runs under its own "systemd-network"
2662 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2663 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2664 loses the ability to write to files owned by root this way.
2666 * Similarly, systemd-resolved now runs under its own
2667 "systemd-resolve" user with no capabilities remaining.
2669 * Similarly, systemd-bus-proxyd now runs under its own
2670 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2672 * systemd-networkd gained support for setting up "veth"
2673 virtual Ethernet devices for container connectivity, as well
2674 as GRE and VTI tunnels.
2676 * systemd-networkd will no longer automatically attempt to
2677 manually load kernel modules necessary for certain tunnel
2678 transports. Instead, it is assumed the kernel loads them
2679 automatically when required. This only works correctly on
2680 very new kernels. On older kernels, please consider adding
2681 the kernel modules to /etc/modules-load.d/ as a work-around.
2683 * The resolv.conf file systemd-resolved generates has been
2684 moved to /run/systemd/resolve/. If you have a symlink from
2685 /etc/resolv.conf, it might be necessary to correct it.
2687 * Two new service settings, ProtectHome= and ProtectSystem=,
2688 have been added. When enabled, they will make the user data
2689 (such as /home) inaccessible or read-only and the system
2690 (such as /usr) read-only, for specific services. This allows
2691 very light-weight per-service sandboxing to avoid
2692 modifications of user data or system files from
2693 services. These two new switches have been enabled for all
2694 of systemd's long-running services, where appropriate.
2696 * Socket units gained new SocketUser= and SocketGroup=
2697 settings to set the owner user and group of AF_UNIX sockets
2698 and FIFOs in the file system.
2700 * Socket units gained a new RemoveOnStop= setting. If enabled,
2701 all FIFOS and sockets in the file system will be removed
2702 when the specific socket unit is stopped.
2704 * Socket units gained a new Symlinks= setting. It takes a list
2705 of symlinks to create to file system sockets or FIFOs
2706 created by the specific Unix sockets. This is useful to
2707 manage symlinks to socket nodes with the same life-cycle as
2710 * The /dev/log socket and /dev/initctl FIFO have been moved to
2711 /run, and have been replaced by symlinks. This allows
2712 connecting to these facilities even if PrivateDevices=yes is
2713 used for a service (which makes /dev/log itself unavailable,
2714 but /run is left). This also has the benefit of ensuring
2715 that /dev only contains device nodes, directories and
2716 symlinks, and nothing else.
2718 * sd-daemon gained two new calls sd_pid_notify() and
2719 sd_pid_notifyf(). They are similar to sd_notify() and
2720 sd_notifyf(), but allow overriding of the source PID of
2721 notification messages if permissions permit this. This is
2722 useful to send notify messages on behalf of a different
2723 process (for example, the parent process). The
2724 systemd-notify tool has been updated to make use of this
2725 when sending messages (so that notification messages now
2726 originate from the shell script invoking systemd-notify and
2727 not the systemd-notify process itself. This should minimize
2728 a race where systemd fails to associate notification
2729 messages to services when the originating process already
2732 * A new "on-abnormal" setting for Restart= has been added. If
2733 set, it will result in automatic restarts on all "abnormal"
2734 reasons for a process to exit, which includes unclean
2735 signals, core dumps, timeouts and watchdog timeouts, but
2736 does not include clean and unclean exit codes or clean
2737 signals. Restart=on-abnormal is an alternative for
2738 Restart=on-failure for services that shall be able to
2739 terminate and avoid restarts on certain errors, by
2740 indicating so with an unclean exit code. Restart=on-failure
2741 or Restart=on-abnormal is now the recommended setting for
2742 all long-running services.
2744 * If the InaccessibleDirectories= service setting points to a
2745 mount point (or if there are any submounts contained within
2746 it), it is now attempted to completely unmount it, to make
2747 the file systems truly unavailable for the respective
2750 * The ReadOnlyDirectories= service setting and
2751 systemd-nspawn's --read-only parameter are now recursively
2752 applied to all submounts, too.
2754 * Mount units may now be created transiently via the bus APIs.
2756 * The support for SysV and LSB init scripts has been removed
2757 from the systemd daemon itself. Instead, it is now
2758 implemented as a generator that creates native systemd units
2759 from these scripts when needed. This enables us to remove a
2760 substantial amount of legacy code from PID 1, following the
2761 fact that many distributions only ship a very small number
2762 of LSB/SysV init scripts nowadays.
2764 * Privileged Xen (dom0) domains are not considered
2765 virtualization anymore by the virtualization detection
2766 logic. After all, they generally have unrestricted access to
2767 the hardware and usually are used to manage the unprivileged
2770 * systemd-tmpfiles gained a new "C" line type, for copying
2771 files or entire directories.
2773 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2774 lines. So far, they have been non-globbing versions of the
2775 latter, and have thus been redundant. In future, it is
2776 recommended to only use "z". "m" has hence been removed
2777 from the documentation, even though it stays supported.
2779 * A tmpfiles snippet to recreate the most basic structure in
2780 /var has been added. This is enough to create the /var/run →
2781 /run symlink and create a couple of structural
2782 directories. This allows systems to boot up with an empty or
2783 volatile /var. Of course, while with this change, the core OS
2784 now is capable with dealing with a volatile /var, not all
2785 user services are ready for it. However, we hope that sooner
2786 or later, many service daemons will be changed upstream so
2787 that they are able to automatically create their necessary
2788 directories in /var at boot, should they be missing. This is
2789 the first step to allow state-less systems that only require
2790 the vendor image for /usr to boot.
2792 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2793 empty tmpfs instance to a specific directory. This is
2794 particularly useful for making use of the automatic
2795 reconstruction of /var (see above), by passing --tmpfs=/var.
2797 * Access modes specified in tmpfiles snippets may now be
2798 prefixed with "~", which indicates that they shall be masked
2799 by whether the existing file or directory is currently
2800 writable, readable or executable at all. Also, if specified,
2801 the sgid/suid/sticky bits will be masked for all
2804 * A new passive target unit "network-pre.target" has been
2805 added which is useful for services that shall run before any
2806 network is configured, for example firewall scripts.
2808 * The "floppy" group that previously owned the /dev/fd*
2809 devices is no longer used. The "disk" group is now used
2810 instead. Distributions should probably deprecate usage of
2813 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2814 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2815 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2816 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2817 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2818 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2821 — Berlin, 2014-06-11
2825 * A new "systemd-timesyncd" daemon has been added for
2826 synchronizing the system clock across the network. It
2827 implements an SNTP client. In contrast to NTP
2828 implementations such as chrony or the NTP reference server,
2829 this only implements a client side, and does not bother with
2830 the full NTP complexity, focusing only on querying time from
2831 one remote server and synchronizing the local clock to
2832 it. Unless you intend to serve NTP to networked clients or
2833 want to connect to local hardware clocks, this simple NTP
2834 client should be more than appropriate for most
2835 installations. The daemon runs with minimal privileges, and
2836 has been hooked up with networkd to only operate when
2837 network connectivity is available. The daemon saves the
2838 current clock to disk every time a new NTP sync has been
2839 acquired, and uses this to possibly correct the system clock
2840 early at bootup, in order to accommodate for systems that
2841 lack an RTC such as the Raspberry Pi and embedded devices,
2842 and to make sure that time monotonically progresses on these
2843 systems, even if it is not always correct. To make use of
2844 this daemon, a new system user and group "systemd-timesync"
2845 needs to be created on installation of systemd.
2847 * The queue "seqnum" interface of libudev has been disabled, as
2848 it was generally incompatible with device namespacing as
2849 sequence numbers of devices go "missing" if the devices are
2850 part of a different namespace.
2852 * "systemctl list-timers" and "systemctl list-sockets" gained
2853 a --recursive switch for showing units of these types also
2854 for all local containers, similar in style to the already
2855 supported --recursive switch for "systemctl list-units".
2857 * A new RebootArgument= setting has been added for service
2858 units, which may be used to specify a kernel reboot argument
2859 to use when triggering reboots with StartLimitAction=.
2861 * A new FailureAction= setting has been added for service
2862 units which may be used to specify an operation to trigger
2863 when a service fails. This works similarly to
2864 StartLimitAction=, but unlike it, controls what is done
2865 immediately rather than only after several attempts to
2866 restart the service in question.
2868 * hostnamed got updated to also expose the kernel name,
2869 release, and version on the bus. This is useful for
2870 executing commands like hostnamectl with the -H switch.
2871 systemd-analyze makes use of this to properly display
2872 details when running non-locally.
2874 * The bootchart tool can now show cgroup information in the
2875 graphs it generates.
2877 * The CFS CPU quota cgroup attribute is now exposed for
2878 services. The new CPUQuota= switch has been added for this
2879 which takes a percentage value. Setting this will have the
2880 result that a service may never get more CPU time than the
2881 specified percentage, even if the machine is otherwise idle.
2883 * systemd-networkd learned IPIP and SIT tunnel support.
2885 * LSB init scripts exposing a dependency on $network will now
2886 get a dependency on network-online.target rather than simply
2887 network.target. This should bring LSB handling closer to
2888 what it was on SysV systems.
2890 * A new fsck.repair= kernel option has been added to control
2891 how fsck shall deal with unclean file systems at boot.
2893 * The (.ini) configuration file parser will now silently
2894 ignore sections whose name begins with "X-". This may be
2895 used to maintain application-specific extension sections in unit
2898 * machined gained a new API to query the IP addresses of
2899 registered containers. "machinectl status" has been updated
2900 to show these addresses in its output.
2902 * A new call sd_uid_get_display() has been added to the
2903 sd-login APIs for querying the "primary" session of a
2904 user. The "primary" session of the user is elected from the
2905 user's sessions and generally a graphical session is
2906 preferred over a text one.
2908 * A minimal systemd-resolved daemon has been added. It
2909 currently simply acts as a companion to systemd-networkd and
2910 manages resolv.conf based on per-interface DNS
2911 configuration, possibly supplied via DHCP. In the long run
2912 we hope to extend this into a local DNSSEC enabled DNS and
2915 * The systemd-networkd-wait-online tool is now enabled by
2916 default. It will delay network-online.target until a network
2917 connection has been configured. The tool primarily integrates
2918 with networkd, but will also make a best effort to make sense
2919 of network configuration performed in some other way.
2921 * Two new service options StartupCPUShares= and
2922 StartupBlockIOWeight= have been added that work similarly to
2923 CPUShares= and BlockIOWeight= however only apply during
2924 system startup. This is useful to prioritize certain services
2925 differently during bootup than during normal runtime.
2927 * hostnamed has been changed to prefer the statically
2928 configured hostname in /etc/hostname (unless set to
2929 'localhost' or empty) over any dynamic one supplied by
2930 dhcp. With this change, the rules for picking the hostname
2931 match more closely the rules of other configuration settings
2932 where the local administrator's configuration in /etc always
2933 overrides any other settings.
2935 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2936 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2937 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2938 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2939 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2940 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2941 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2942 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2943 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2944 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2945 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2946 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2947 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2948 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2949 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2950 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2953 — Beijing, 2014-05-28
2957 * When restoring the screen brightness at boot, stay away from
2958 the darkest setting or from the lowest 5% of the available
2959 range, depending on which is the larger value of both. This
2960 should effectively protect the user from rebooting into a
2961 black screen, should the brightness have been set to minimum
2964 * sd-login gained a new sd_machine_get_class() call to
2965 determine the class ("vm" or "container") of a machine
2966 registered with machined.
2968 * sd-login gained new calls
2969 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2970 to query the identity of the peer of a local AF_UNIX
2971 connection. They operate similarly to their sd_pid_get_xyz()
2974 * PID 1 will now maintain a system-wide system state engine
2975 with the states "starting", "running", "degraded",
2976 "maintenance", "stopping". These states are bound to system
2977 startup, normal runtime, runtime with at least one failed
2978 service, rescue/emergency mode and system shutdown. This
2979 state is shown in the "systemctl status" output when no unit
2980 name is passed. It is useful to determine system state, in
2981 particularly when doing so for many systems or containers at
2984 * A new command "list-machines" has been added to "systemctl"
2985 that lists all local OS containers and shows their system
2986 state (see above), if systemd runs inside of them.
2988 * systemctl gained a new "-r" switch to recursively enumerate
2989 units on all local containers, when used with the
2990 "list-unit" command (which is the default one that is
2991 executed when no parameters are specified).
2993 * The GPT automatic partition discovery logic will now honour
2994 two GPT partition flags: one may be set on a partition to
2995 cause it to be mounted read-only, and the other may be set
2996 on a partition to ignore it during automatic discovery.
2998 * Two new GPT type UUIDs have been added for automatic root
2999 partition discovery, for 32-bit and 64-bit ARM. This is not
3000 particularly useful for discovering the root directory on
3001 these architectures during bare-metal boots (since UEFI is
3002 not common there), but still very useful to allow booting of
3003 ARM disk images in nspawn with the -i option.
3005 * MAC addresses of interfaces created with nspawn's
3006 --network-interface= switch will now be generated from the
3007 machine name, and thus be stable between multiple invocations
3010 * logind will now automatically remove all IPC objects owned
3011 by a user if she or he fully logs out. This makes sure that
3012 users who are logged out cannot continue to consume IPC
3013 resources. This covers SysV memory, semaphores and message
3014 queues as well as POSIX shared memory and message
3015 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3016 limits. With this functionality, that is corrected. This may
3017 be turned off by using the RemoveIPC= switch of logind.conf.
3019 * The systemd-machine-id-setup and tmpfiles tools gained a
3020 --root= switch to operate on a specific root directory,
3023 * journald can now forward logged messages to the TTYs of all
3024 logged in users ("wall"). This is the default for all
3025 emergency messages now.
3027 * A new tool systemd-journal-remote has been added to stream
3028 journal log messages across the network.
3030 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3031 controller trees are mounted into it. Note that the
3032 directories mounted beneath it are not read-only. This is a
3033 security measure and is particularly useful because glibc
3034 actually includes a search logic to pick any tmpfs it can
3035 find to implement shm_open() if /dev/shm is not available
3036 (which it might very well be in namespaced setups).
3038 * machinectl gained a new "poweroff" command to cleanly power
3039 down a local OS container.
3041 * The PrivateDevices= unit file setting will now also drop the
3042 CAP_MKNOD capability from the capability bound set, and
3043 imply DevicePolicy=closed.
3045 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3046 comprehensively on all long-running systemd services where
3047 this is appropriate.
3049 * systemd-udevd will now run in a disassociated mount
3050 namespace. To mount directories from udev rules, make sure to
3051 pull in mount units via SYSTEMD_WANTS properties.
3053 * The kdbus support gained support for uploading policy into
3054 the kernel. sd-bus gained support for creating "monitoring"
3055 connections that can eavesdrop into all bus communication
3056 for debugging purposes.
3058 * Timestamps may now be specified in seconds since the UNIX
3059 epoch Jan 1st, 1970 by specifying "@" followed by the value
3062 * Native tcpwrap support in systemd has been removed. tcpwrap
3063 is old code, not really maintained anymore and has serious
3064 shortcomings, and better options such as firewalls
3065 exist. For setups that require tcpwrap usage, please
3066 consider invoking your socket-activated service via tcpd,
3067 like on traditional inetd.
3069 * A new system.conf configuration option
3070 DefaultTimerAccuracySec= has been added that controls the
3071 default AccuracySec= setting of .timer units.
3073 * Timer units gained a new WakeSystem= switch. If enabled,
3074 timers configured this way will cause the system to resume
3075 from system suspend (if the system supports that, which most
3078 * Timer units gained a new Persistent= switch. If enabled,
3079 timers configured this way will save to disk when they have
3080 been last triggered. This information is then used on next
3081 reboot to possible execute overdue timer events, that
3082 could not take place because the system was powered off.
3083 This enables simple anacron-like behaviour for timer units.
3085 * systemctl's "list-timers" will now also list the time a
3086 timer unit was last triggered in addition to the next time
3087 it will be triggered.
3089 * systemd-networkd will now assign predictable IPv4LL
3090 addresses to its local interfaces.
3092 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3093 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3094 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3095 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3096 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3097 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3098 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3099 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3102 — Berlin, 2014-03-25
3106 * A new unit file setting RestrictAddressFamilies= has been
3107 added to restrict which socket address families unit
3108 processes gain access to. This takes address family names
3109 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3110 attack surface of services via exotic protocol stacks. This
3111 is built on seccomp system call filters.
3113 * Two new unit file settings RuntimeDirectory= and
3114 RuntimeDirectoryMode= have been added that may be used to
3115 manage a per-daemon runtime directories below /run. This is
3116 an alternative for setting up directory permissions with
3117 tmpfiles snippets, and has the advantage that the runtime
3118 directory's lifetime is bound to the daemon runtime and that
3119 the daemon starts up with an empty directory each time. This
3120 is particularly useful when writing services that drop
3121 privileges using the User= or Group= setting.
3123 * The DeviceAllow= unit setting now supports globbing for
3124 matching against device group names.
3126 * The systemd configuration file system.conf gained new
3127 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3128 DefaultMemoryAccounting= to globally turn on/off accounting
3129 for specific resources (cgroups) for all units. These
3130 settings may still be overridden individually in each unit
3133 * systemd-gpt-auto-generator is now able to discover /srv and
3134 root partitions in addition to /home and swap partitions. It
3135 also supports LUKS-encrypted partitions now. With this in
3136 place, automatic discovery of partitions to mount following
3137 the Discoverable Partitions Specification
3138 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3139 is now a lot more complete. This allows booting without
3140 /etc/fstab and without root= on the kernel command line on
3141 systems prepared appropriately.
3143 * systemd-nspawn gained a new --image= switch which allows
3144 booting up disk images and Linux installations on any block
3145 device that follow the Discoverable Partitions Specification
3146 (see above). This means that installations made with
3147 appropriately updated installers may now be started and
3148 deployed using container managers, completely
3149 unmodified. (We hope that libvirt-lxc will add support for
3150 this feature soon, too.)
3152 * systemd-nspawn gained a new --network-macvlan= setting to
3153 set up a private macvlan interface for the
3154 container. Similarly, systemd-networkd gained a new
3155 Kind=macvlan setting in .netdev files.
3157 * systemd-networkd now supports configuring local addresses
3160 * A new tool systemd-network-wait-online has been added to
3161 synchronously wait for network connectivity using
3164 * The sd-bus.h bus API gained a new sd_bus_track object for
3165 tracking the life-cycle of bus peers. Note that sd-bus.h is
3166 still not a public API though (unless you specify
3167 --enable-kdbus on the configure command line, which however
3168 voids your warranty and you get no API stability guarantee).
3170 * The $XDG_RUNTIME_DIR runtime directories for each user are
3171 now individual tmpfs instances, which has the benefit of
3172 introducing separate pools for each user, with individual
3173 size limits, and thus making sure that unprivileged clients
3174 can no longer negatively impact the system or other users by
3175 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3176 RuntimeDirectorySize= has been introduced that allows
3177 controlling the default size limit for all users. It
3178 defaults to 10% of the available physical memory. This is no
3179 replacement for quotas on tmpfs though (which the kernel
3180 still does not support), as /dev/shm and /tmp are still
3181 shared resources used by both the system and unprivileged
3184 * logind will now automatically turn off automatic suspending
3185 on laptop lid close when more than one display is
3186 connected. This was previously expected to be implemented
3187 individually in desktop environments (such as GNOME),
3188 however has been added to logind now, in order to fix a
3189 boot-time race where a desktop environment might not have
3190 been started yet and thus not been able to take an inhibitor
3191 lock at the time where logind already suspends the system
3192 due to a closed lid.
3194 * logind will now wait at least 30s after each system
3195 suspend/resume cycle, and 3min after system boot before
3196 suspending the system due to a closed laptop lid. This
3197 should give USB docking stations and similar enough time to
3198 be probed and configured after system resume and boot in
3199 order to then act as suspend blocker.
3201 * systemd-run gained a new --property= setting which allows
3202 initialization of resource control properties (and others)
3203 for the created scope or service unit. Example: "systemd-run
3204 --property=BlockIOWeight=10 updatedb" may be used to run
3205 updatedb at a low block IO scheduling weight.
3207 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3208 now also work in --scope mode.
3210 * When systemd is compiled with kdbus support, basic support
3211 for enforced policies is now in place. (Note that enabling
3212 kdbus still voids your warranty and no API compatibility
3215 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3216 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3217 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3218 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3219 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3220 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3221 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3222 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3223 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3224 Zbigniew Jędrzejewski-Szmek
3226 — Berlin, 2014-03-12
3230 * systemd will now relabel /dev after loading the SMACK policy
3231 according to SMACK rules.
3233 * A new unit file option AppArmorProfile= has been added to
3234 set the AppArmor profile for the processes of a unit.
3236 * A new condition check ConditionArchitecture= has been added
3237 to conditionalize units based on the system architecture, as
3238 reported by uname()'s "machine" field.
3240 * systemd-networkd now supports matching on the system
3241 virtualization, architecture, kernel command line, host name
3244 * logind is now a lot more aggressive when suspending the
3245 machine due to a closed laptop lid. Instead of acting only
3246 on the lid close action, it will continuously watch the lid
3247 status and act on it. This is useful for laptops where the
3248 power button is on the outside of the chassis so that it can
3249 be reached without opening the lid (such as the Lenovo
3250 Yoga). On those machines, logind will now immediately
3251 re-suspend the machine if the power button has been
3252 accidentally pressed while the laptop was suspended and in a
3253 backpack or similar.
3255 * logind will now watch SW_DOCK switches and inhibit reaction
3256 to the lid switch if it is pressed. This means that logind
3257 will not suspend the machine anymore if the lid is closed
3258 and the system is docked, if the laptop supports SW_DOCK
3259 notifications via the input layer. Note that ACPI docking
3260 stations do not generate this currently. Also note that this
3261 logic is usually not fully sufficient and Desktop
3262 Environments should take a lid switch inhibitor lock when an
3263 external display is connected, as systemd will not watch
3266 * nspawn will now make use of the devices cgroup controller by
3267 default, and only permit creation of and access to the usual
3268 API device nodes like /dev/null or /dev/random, as well as
3269 access to (but not creation of) the pty devices.
3271 * We will now ship a default .network file for
3272 systemd-networkd that automatically configures DHCP for
3273 network interfaces created by nspawn's --network-veth or
3274 --network-bridge= switches.
3276 * systemd will now understand the usual M, K, G, T suffixes
3277 according to SI conventions (i.e. to the base 1000) when
3278 referring to throughput and hardware metrics. It will stay
3279 with IEC conventions (i.e. to the base 1024) for software
3280 metrics, according to what is customary according to
3281 Wikipedia. We explicitly document which base applies for
3282 each configuration option.
3284 * The DeviceAllow= setting in unit files now supports a syntax
3285 to whitelist an entire group of devices node majors at once,
3286 based on the /proc/devices listing. For example, with the
3287 string "char-pts", it is now possible to whitelist all
3288 current and future pseudo-TTYs at once.
3290 * sd-event learned a new "post" event source. Event sources of
3291 this type are triggered by the dispatching of any event
3292 source of a type that is not "post". This is useful for
3293 implementing clean-up and check event sources that are
3294 triggered by other work being done in the program.
3296 * systemd-networkd is no longer statically enabled, but uses
3297 the usual [Install] sections so that it can be
3298 enabled/disabled using systemctl. It still is enabled by
3301 * When creating a veth interface pair with systemd-nspawn, the
3302 host side will now be prefixed with "vb-" if
3303 --network-bridge= is used, and with "ve-" if --network-veth
3304 is used. This way, it is easy to distinguish these cases on
3305 the host, for example to apply different configuration to
3306 them with systemd-networkd.
3308 * The compatibility libraries for libsystemd-journal.so,
3309 libsystem-id128.so, libsystemd-login.so and
3310 libsystemd-daemon.so do not make use of IFUNC
3311 anymore. Instead, we now build libsystemd.so multiple times
3312 under these alternative names. This means that the footprint
3313 is drastically increased, but given that these are
3314 transitional compatibility libraries, this should not matter
3315 much. This change has been made necessary to support the ARM
3316 platform for these compatibility libraries, as the ARM
3317 toolchain is not really at the same level as the toolchain
3318 for other architectures like x86 and does not support
3319 IFUNC. Please make sure to use --enable-compat-libs only
3320 during a transitional period!
3322 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3323 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3324 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3325 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3326 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3327 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3328 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3329 Zbigniew Jędrzejewski-Szmek
3331 — Berlin, 2014-02-24
3335 * A new component "systemd-networkd" has been added that can
3336 be used to configure local network interfaces statically or
3337 via DHCP. It is capable of bringing up bridges, VLANs, and
3338 bonding. Currently, no hook-ups for interactive network
3339 configuration are provided. Use this for your initrd,
3340 container, embedded, or server setup if you need a simple,
3341 yet powerful, network configuration solution. This
3342 configuration subsystem is quite nifty, as it allows wildcard
3343 hotplug matching in interfaces. For example, with a single
3344 configuration snippet, you can configure that all Ethernet
3345 interfaces showing up are automatically added to a bridge,
3346 or similar. It supports link-sensing and more.
3348 * A new tool "systemd-socket-proxyd" has been added which can
3349 act as a bidirectional proxy for TCP sockets. This is
3350 useful for adding socket activation support to services that
3351 do not actually support socket activation, including virtual
3352 machines and the like.
3354 * Add a new tool to save/restore rfkill state on
3357 * Save/restore state of keyboard backlights in addition to
3358 display backlights on shutdown/boot.
3360 * udev learned a new SECLABEL{} construct to label device
3361 nodes with a specific security label when they appear. For
3362 now, only SECLABEL{selinux} is supported, but the syntax is
3363 prepared for additional security frameworks.
3365 * udev gained a new scheme to configure link-level attributes
3366 from files in /etc/systemd/network/*.link. These files can
3367 match against MAC address, device path, driver name and type,
3368 and will apply attributes like the naming policy, link speed,
3369 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3370 address assignment policy (randomized, ...).
3372 * The configuration of network interface naming rules for
3373 "permanent interface names" has changed: a new NamePolicy=
3374 setting in the [Link] section of .link files determines the
3375 priority of possible naming schemes (onboard, slot, MAC,
3376 path). The default value of this setting is determined by
3377 /usr/lib/net/links/99-default.link. Old
3378 80-net-name-slot.rules udev configuration file has been
3379 removed, so local configuration overriding this file should
3380 be adapated to override 99-default.link instead.
3382 * When the User= switch is used in a unit file, also
3383 initialize $SHELL= based on the user database entry.
3385 * systemd no longer depends on libdbus. All communication is
3386 now done with sd-bus, systemd's low-level bus library
3389 * kdbus support has been added to PID 1 itself. When kdbus is
3390 enabled, this causes PID 1 to set up the system bus and
3391 enable support for a new ".busname" unit type that
3392 encapsulates bus name activation on kdbus. It works a little
3393 bit like ".socket" units, except for bus names. A new
3394 generator has been added that converts classic dbus1 service
3395 activation files automatically into native systemd .busname
3398 * sd-bus: add a light-weight vtable implementation that allows
3399 defining objects on the bus with a simple static const
3400 vtable array of its methods, signals and properties.
3402 * systemd will not generate or install static dbus
3403 introspection data anymore to /usr/share/dbus-1/interfaces,
3404 as the precise format of these files is unclear, and
3405 nothing makes use of it.
3407 * A proxy daemon is now provided to proxy clients connecting
3408 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3409 compatibility with classic D-Bus.
3411 * A bus driver implementation has been added that supports the
3412 classic D-Bus bus driver calls on kdbus, also for
3413 compatibility purposes.
3415 * A new API "sd-event.h" has been added that implements a
3416 minimal event loop API built around epoll. It provides a
3417 couple of features that direct epoll usage is lacking:
3418 prioritization of events, scales to large numbers of timer
3419 events, per-event timer slack (accuracy), system-wide
3420 coalescing of timer events, exit handlers, watchdog
3421 supervision support using systemd's sd_notify() API, child
3424 * A new API "sd-rntl.h" has been added that provides an API
3425 around the route netlink interface of the kernel, similar in
3426 style to "sd-bus.h".
3428 * A new API "sd-dhcp-client.h" has been added that provides a
3429 small DHCPv4 client-side implementation. This is used by
3432 * There is a new kernel command line option
3433 "systemd.restore_state=0|1". When set to "0", none of the
3434 systemd tools will restore saved runtime state to hardware
3435 devices. More specifically, the rfkill and backlight states
3438 * The FsckPassNo= compatibility option in mount/service units
3439 has been removed. The fstab generator will now add the
3440 necessary dependencies automatically, and does not require
3441 PID1's support for that anymore.
3443 * journalctl gained a new switch, --list-boots, that lists
3444 recent boots with their times and boot IDs.
3446 * The various tools like systemctl, loginctl, timedatectl,
3447 busctl, systemd-run, ... have gained a new switch "-M" to
3448 connect to a specific, local OS container (as direct
3449 connection, without requiring SSH). This works on any
3450 container that is registered with machined, such as those
3451 created by libvirt-lxc or nspawn.
3453 * systemd-run and systemd-analyze also gained support for "-H"
3454 to connect to remote hosts via SSH. This is particularly
3455 useful for systemd-run because it enables queuing of jobs
3456 onto remote systems.
3458 * machinectl gained a new command "login" to open a getty
3459 login in any local container. This works with any container
3460 that is registered with machined (such as those created by
3461 libvirt-lxc or nspawn), and which runs systemd inside.
3463 * machinectl gained a new "reboot" command that may be used to
3464 trigger a reboot on a specific container that is registered
3465 with machined. This works on any container that runs an init
3466 system of some kind.
3468 * systemctl gained a new "list-timers" command to print a nice
3469 listing of installed timer units with the times they elapse
3472 * Alternative reboot() parameters may now be specified on the
3473 "systemctl reboot" command line and are passed to the
3474 reboot() system call.
3476 * systemctl gained a new --job-mode= switch to configure the
3477 mode to queue a job with. This is a more generic version of
3478 --fail, --irreversible, and --ignore-dependencies, which are
3479 still available but not advertised anymore.
3481 * /etc/systemd/system.conf gained new settings to configure
3482 various default timeouts of units, as well as the default
3483 start limit interval and burst. These may still be overridden
3486 * PID1 will now export on the bus profile data of the security
3487 policy upload process (such as the SELinux policy upload to
3490 * journald: when forwarding logs to the console, include
3491 timestamps (following the setting in
3492 /sys/module/printk/parameters/time).
3494 * OnCalendar= in timer units now understands the special
3495 strings "yearly" and "annually". (Both are equivalent)
3497 * The accuracy of timer units is now configurable with the new
3498 AccuracySec= setting. It defaults to 1min.
3500 * A new dependency type JoinsNamespaceOf= has been added that
3501 allows running two services within the same /tmp and network
3502 namespace, if PrivateNetwork= or PrivateTmp= are used.
3504 * A new command "cat" has been added to systemctl. It outputs
3505 the original unit file of a unit, and concatenates the
3506 contents of additional "drop-in" unit file snippets, so that
3507 the full configuration is shown.
3509 * systemctl now supports globbing on the various "list-xyz"
3510 commands, like "list-units" or "list-sockets", as well as on
3511 those commands which take multiple unit names.
3513 * journalctl's --unit= switch gained support for globbing.
3515 * All systemd daemons now make use of the watchdog logic so
3516 that systemd automatically notices when they hang.
3518 * If the $container_ttys environment variable is set,
3519 getty-generator will automatically spawn a getty for each
3520 listed tty. This is useful for container managers to request
3521 login gettys to be spawned on as many ttys as needed.
3523 * %h, %s, %U specifier support is not available anymore when
3524 used in unit files for PID 1. This is because NSS calls are
3525 not safe from PID 1. They stay available for --user
3526 instances of systemd, and as special case for the root user.
3528 * loginctl gained a new "--no-legend" switch to turn off output
3531 * The "sd-login.h" API gained three new calls:
3532 sd_session_is_remote(), sd_session_get_remote_user(),
3533 sd_session_get_remote_host() to query information about
3536 * The udev hardware database now also carries vendor/product
3537 information of SDIO devices.
3539 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3540 determine whether watchdog notifications are requested by
3543 * Socket-activated per-connection services now include a
3544 short description of the connection parameters in the
3547 * tmpfiles gained a new "--boot" option. When this is not used,
3548 only lines where the command character is not suffixed with
3549 "!" are executed. When this option is specified, those
3550 options are executed too. This partitions tmpfiles
3551 directives into those that can be safely executed at any
3552 time, and those which should be run only at boot (for
3553 example, a line that creates /run/nologin).
3555 * A new API "sd-resolve.h" has been added which provides a simple
3556 asynchronous wrapper around glibc NSS host name resolution
3557 calls, such as getaddrinfo(). In contrast to glibc's
3558 getaddrinfo_a(), it does not use signals. In contrast to most
3559 other asynchronous name resolution libraries, this one does
3560 not reimplement DNS, but reuses NSS, so that alternate
3561 host name resolution systems continue to work, such as mDNS,
3562 LDAP, etc. This API is based on libasyncns, but it has been
3563 cleaned up for inclusion in systemd.
3565 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3566 "sd-daemon.h" are no longer found in individual libraries
3567 libsystemd-journal.so, libsystemd-login.so,
3568 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3569 merged them into a single library, libsystemd.so, which
3570 provides all symbols. The reason for this is cyclic
3571 dependencies, as these libraries tend to use each other's
3572 symbols. So far, we have managed to workaround that by linking
3573 a copy of a good part of our code into each of these
3574 libraries again and again, which, however, makes certain
3575 things hard to do, like sharing static variables. Also, it
3576 substantially increases footprint. With this change, there
3577 is only one library for the basic APIs systemd
3578 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3579 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3580 library as well, however are subject to the --enable-kdbus
3581 switch (see below). Note that "sd-dhcp-client.h" is not part
3582 of this library (this is because it only consumes, never
3583 provides, services of/to other APIs). To make the transition
3584 easy from the separate libraries to the unified one, we
3585 provide the --enable-compat-libs compile-time switch which
3586 will generate stub libraries that are compatible with the
3587 old ones but redirect all calls to the new one.
3589 * All of the kdbus logic and the new APIs "sd-bus.h",
3590 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3591 and "sd-utf8.h" are compile-time optional via the
3592 "--enable-kdbus" switch, and they are not compiled in by
3593 default. To make use of kdbus, you have to explicitly enable
3594 the switch. Note however, that neither the kernel nor the
3595 userspace API for all of this is considered stable yet. We
3596 want to maintain the freedom to still change the APIs for
3597 now. By specifying this build-time switch, you acknowledge
3598 that you are aware of the instability of the current
3601 * Also, note that while kdbus is pretty much complete,
3602 it lacks one thing: proper policy support. This means you
3603 can build a fully working system with all features; however,
3604 it will be highly insecure. Policy support will be added in
3605 one of the next releases, at the same time that we will
3606 declare the APIs stable.
3608 * When the kernel command line argument "kdbus" is specified,
3609 systemd will automatically load the kdbus.ko kernel module. At
3610 this stage of development, it is only useful for testing kdbus
3611 and should not be used in production. Note: if "--enable-kdbus"
3612 is specified, and the kdbus.ko kernel module is available, and
3613 "kdbus" is added to the kernel command line, the entire system
3614 runs with kdbus instead of dbus-daemon, with the above mentioned
3615 problem of missing the system policy enforcement. Also a future
3616 version of kdbus.ko or a newer systemd will not be compatible with
3617 each other, and will unlikely be able to boot the machine if only
3618 one of them is updated.
3620 * systemctl gained a new "import-environment" command which
3621 uploads the caller's environment (or parts thereof) into the
3622 service manager so that it is inherited by services started
3623 by the manager. This is useful to upload variables like
3624 $DISPLAY into the user service manager.
3626 * A new PrivateDevices= switch has been added to service units
3627 which allows running a service with a namespaced /dev
3628 directory that does not contain any device nodes for
3629 physical devices. More specifically, it only includes devices
3630 such as /dev/null, /dev/urandom, and /dev/zero which are API
3633 * logind has been extended to support behaviour like VT
3634 switching on seats that do not support a VT. This makes
3635 multi-session available on seats that are not the first seat
3636 (seat0), and on systems where kernel support for VTs has
3637 been disabled at compile-time.
3639 * If a process holds a delay lock for system sleep or shutdown
3640 and fails to release it in time, we will now log its
3641 identity. This makes it easier to identify processes that
3642 cause slow suspends or power-offs.
3644 * When parsing /etc/crypttab, support for a new key-slot=
3645 option as supported by Debian is added. It allows indicating
3646 which LUKS slot to use on disk, speeding up key loading.
3648 * The sd_journald_sendv() API call has been checked and
3649 officially declared to be async-signal-safe so that it may
3650 be invoked from signal handlers for logging purposes.
3652 * Boot-time status output is now enabled automatically after a
3653 short timeout if boot does not progress, in order to give
3654 the user an indication what she or he is waiting for.
3656 * The boot-time output has been improved to show how much time
3657 remains until jobs expire.
3659 * The KillMode= switch in service units gained a new possible
3660 value "mixed". If set, and the unit is shut down, then the
3661 initial SIGTERM signal is sent only to the main daemon
3662 process, while the following SIGKILL signal is sent to
3663 all remaining processes of the service.
3665 * When a scope unit is registered, a new property "Controller"
3666 may be set. If set to a valid bus name, systemd will send a
3667 RequestStop() signal to this name when it would like to shut
3668 down the scope. This may be used to hook manager logic into
3669 the shutdown logic of scope units. Also, scope units may now
3670 be put in a special "abandoned" state, in which case the
3671 manager process which created them takes no further
3672 responsibilities for it.
3674 * When reading unit files, systemd will now verify
3675 the access mode of these files, and warn about certain
3676 suspicious combinations. This has been added to make it
3677 easier to track down packaging bugs where unit files are
3678 marked executable or world-writable.
3680 * systemd-nspawn gained a new "--setenv=" switch to set
3681 container-wide environment variables. The similar option in
3682 systemd-activate was renamed from "--environment=" to
3683 "--setenv=" for consistency.
3685 * systemd-nspawn has been updated to create a new kdbus domain
3686 for each container that is invoked, thus allowing each
3687 container to have its own set of system and user buses,
3688 independent of the host.
3690 * systemd-nspawn gained a new --drop-capability= switch to run
3691 the container with less capabilities than the default. Both
3692 --drop-capability= and --capability= now take the special
3693 string "all" for dropping or keeping all capabilities.
3695 * systemd-nspawn gained new switches for executing containers
3696 with specific SELinux labels set.
3698 * systemd-nspawn gained a new --quiet switch to not generate
3699 any additional output but the container's own console
3702 * systemd-nspawn gained a new --share-system switch to run a
3703 container without PID namespacing enabled.
3705 * systemd-nspawn gained a new --register= switch to control
3706 whether the container is registered with systemd-machined or
3707 not. This is useful for containers that do not run full
3708 OS images, but only specific apps.
3710 * systemd-nspawn gained a new --keep-unit which may be used
3711 when invoked as the only program from a service unit, and
3712 results in registration of the unit service itself in
3713 systemd-machined, instead of a newly opened scope unit.
3715 * systemd-nspawn gained a new --network-interface= switch for
3716 moving arbitrary interfaces to the container. The new
3717 --network-veth switch creates a virtual Ethernet connection
3718 between host and container. The new --network-bridge=
3719 switch then allows assigning the host side of this virtual
3720 Ethernet connection to a bridge device.
3722 * systemd-nspawn gained a new --personality= switch for
3723 setting the kernel personality for the container. This is
3724 useful when running a 32-bit container on a 64-bit host. A
3725 similar option Personality= is now also available for service
3728 * logind will now also track a "Desktop" identifier for each
3729 session which encodes the desktop environment of it. This is
3730 useful for desktop environments that want to identify
3731 multiple running sessions of itself easily.
3733 * A new SELinuxContext= setting for service units has been
3734 added that allows setting a specific SELinux execution
3735 context for a service.
3737 * Most systemd client tools will now honour $SYSTEMD_LESS for
3738 settings of the "less" pager. By default, these tools will
3739 override $LESS to allow certain operations to work, such as
3740 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3741 influence this logic.
3743 * systemd's "seccomp" hook-up has been changed to make use of
3744 the libseccomp library instead of using its own
3745 implementation. This has benefits for portability among
3748 * For usage together with SystemCallFilter=, a new
3749 SystemCallErrorNumber= setting has been introduced that
3750 allows configuration of a system error number to be returned
3751 on filtered system calls, instead of immediately killing the
3752 process. Also, SystemCallArchitectures= has been added to
3753 limit access to system calls of a particular architecture
3754 (in order to turn off support for unused secondary
3755 architectures). There is also a global
3756 SystemCallArchitectures= setting in system.conf now to turn
3757 off support for non-native system calls system-wide.
3759 * systemd requires a kernel with a working name_to_handle_at(),
3760 please see the kernel config requirements in the README file.
3762 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3763 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3764 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3765 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3766 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3767 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3768 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3769 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3770 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3771 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3772 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3773 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3774 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3775 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3776 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3777 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3778 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3779 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3780 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3781 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3782 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3783 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3784 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3785 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3787 — Berlin, 2014-02-20
3791 * logind has gained support for facilitating privileged input
3792 and drm device access for unprivileged clients. This work is
3793 useful to allow Wayland display servers (and similar
3794 programs, such as kmscon) to run under the user's ID and
3795 access input and drm devices which are normally
3796 protected. When this is used (and the kernel is new enough)
3797 logind will "mute" IO on the file descriptors passed to
3798 Wayland as long as it is in the background and "unmute" it
3799 if it returns into the foreground. This allows secure
3800 session switching without allowing background sessions to
3801 eavesdrop on input and display data. This also introduces
3802 session switching support if VT support is turned off in the
3803 kernel, and on seats that are not seat0.
3805 * A new kernel command line option luks.options= is understood
3806 now which allows specifying LUKS options for usage for LUKS
3807 encrypted partitions specified with luks.uuid=.
3809 * tmpfiles.d(5) snippets may now use specifier expansion in
3810 path names. More specifically %m, %b, %H, %v, are now
3811 replaced by the local machine id, boot id, hostname, and
3812 kernel version number.
3814 * A new tmpfiles.d(5) command "m" has been introduced which
3815 may be used to change the owner/group/access mode of a file
3816 or directory if it exists, but do nothing if it does not.
3818 * This release removes high-level support for the
3819 MemorySoftLimit= cgroup setting. The underlying kernel
3820 cgroup attribute memory.soft_limit= is currently badly
3821 designed and likely to be removed from the kernel API in its
3822 current form, hence we should not expose it for now.
3824 * The memory.use_hierarchy cgroup attribute is now enabled for
3825 all cgroups systemd creates in the memory cgroup
3826 hierarchy. This option is likely to be come the built-in
3827 default in the kernel anyway, and the non-hierarchical mode
3828 never made much sense in the intrinsically hierarchical
3831 * A new field _SYSTEMD_SLICE= is logged along with all journal
3832 messages containing the slice a message was generated
3833 from. This is useful to allow easy per-customer filtering of
3834 logs among other things.
3836 * systemd-journald will no longer adjust the group of journal
3837 files it creates to the "systemd-journal" group. Instead we
3838 rely on the journal directory to be owned by the
3839 "systemd-journal" group, and its setgid bit set, so that the
3840 kernel file system layer will automatically enforce that
3841 journal files inherit this group assignment. The reason for
3842 this change is that we cannot allow NSS look-ups from
3843 journald which would be necessary to resolve
3844 "systemd-journal" to a numeric GID, because this might
3845 create deadlocks if NSS involves synchronous queries to
3846 other daemons (such as nscd, or sssd) which in turn are
3847 logging clients of journald and might block on it, which
3848 would then dead lock. A tmpfiles.d(5) snippet included in
3849 systemd will make sure the setgid bit and group are
3850 properly set on the journal directory if it exists on every
3851 boot. However, we recommend adjusting it manually after
3852 upgrades too (or from RPM scriptlets), so that the change is
3853 not delayed until next reboot.
3855 * Backlight and random seed files in /var/lib/ have moved into
3856 the /var/lib/systemd/ directory, in order to centralize all
3857 systemd generated files in one directory.
3859 * Boot time performance measurements (as displayed by
3860 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3861 performance information if that's available to determine how
3862 much time BIOS and boot loader initialization required. With
3863 a sufficiently new BIOS you hence no longer need to boot
3864 with Gummiboot to get access to such information.
3866 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3867 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3868 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3869 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3870 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3871 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3872 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3874 — Berlin, 2013-10-02
3878 * The Restart= option for services now understands a new
3879 on-watchdog setting, which will restart the service
3880 automatically if the service stops sending out watchdog keep
3881 alive messages (as configured with WatchdogSec=).
3883 * The getty generator (which is responsible for bringing up a
3884 getty on configured serial consoles) will no longer only
3885 start a getty on the primary kernel console but on all
3886 others, too. This makes the order in which console= is
3887 specified on the kernel command line less important.
3889 * libsystemd-logind gained a new sd_session_get_vt() call to
3890 retrieve the VT number of a session.
3892 * If the option "tries=0" is set for an entry of /etc/crypttab
3893 its passphrase is queried indefinitely instead of any
3894 maximum number of tries.
3896 * If a service with a configure PID file terminates its PID
3897 file will now be removed automatically if it still exists
3898 afterwards. This should put an end to stale PID files.
3900 * systemd-run will now also take relative binary path names
3901 for execution and no longer insists on absolute paths.
3903 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3904 paths that are optionally prefixed with "-" to indicate that
3905 it should not be considered a failure if they do not exist.
3907 * journalctl -o (and similar commands) now understands a new
3908 output mode "short-precise", it is similar to "short" but
3909 shows timestamps with usec accuracy.
3911 * The option "discard" (as known from Debian) is now
3912 synonymous to "allow-discards" in /etc/crypttab. In fact,
3913 "discard" is preferred now (since it is easier to remember
3916 * Some licensing clean-ups were made, so that more code is now
3917 LGPL-2.1 licensed than before.
3919 * A minimal tool to save/restore the display backlight
3920 brightness across reboots has been added. It will store the
3921 backlight setting as late as possible at shutdown, and
3922 restore it as early as possible during reboot.
3924 * A logic to automatically discover and enable home and swap
3925 partitions on GPT disks has been added. With this in place
3926 /etc/fstab becomes optional for many setups as systemd can
3927 discover certain partitions located on the root disk
3928 automatically. Home partitions are recognized under their
3929 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3930 partitions are recognized under their GPT type ID
3931 0657fd6da4ab43c484e50933c84b4f4f.
3933 * systemd will no longer pass any environment from the kernel
3934 or initrd to system services. If you want to set an
3935 environment for all services, do so via the kernel command
3936 line systemd.setenv= assignment.
3938 * The systemd-sysctl tool no longer natively reads the file
3939 /etc/sysctl.conf. If desired, the file should be symlinked
3940 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3941 legacy support by a symlink rather than built-in code, it
3942 also makes the otherwise hidden order of application of the
3943 different files visible. (Note that this partly reverts to a
3944 pre-198 application order of sysctl knobs!)
3946 * The "systemctl set-log-level" and "systemctl dump" commands
3947 have been moved to systemd-analyze.
3949 * systemd-run learned the new --remain-after-exit switch,
3950 which causes the scope unit not to be cleaned up
3951 automatically after the process terminated.
3953 * tmpfiles learned a new --exclude-prefix= switch to exclude
3954 certain paths from operation.
3956 * journald will now automatically flush all messages to disk
3957 as soon as a message at the log level CRIT, ALERT or EMERG
3960 Contributions from: Andrew Cook, Brandon Philips, Christian
3961 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3962 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3963 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3964 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3965 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3966 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3967 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3968 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3969 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3970 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3971 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3972 William Giokas, Zbigniew Jędrzejewski-Szmek
3974 — Berlin, 2013-09-13
3978 * The documentation has been updated to cover the various new
3979 concepts introduced with 205.
3981 * Unit files now understand the new %v specifier which
3982 resolves to the kernel version string as returned by "uname
3985 * systemctl now supports filtering the unit list output by
3986 load state, active state and sub state, using the new
3989 * "systemctl status" will now show the results of the
3990 condition checks (like ConditionPathExists= and similar) of
3991 the last start attempts of the unit. They are also logged to
3994 * "journalctl -b" may now be used to look for boot output of a
3995 specific boot. Try "journalctl -b -1" for the previous boot,
3996 but the syntax is substantially more powerful.
3998 * "journalctl --show-cursor" has been added which prints the
3999 cursor string the last shown log line. This may then be used
4000 with the new "journalctl --after-cursor=" switch to continue
4001 browsing logs from that point on.
4003 * "journalctl --force" may now be used to force regeneration
4006 * Creation of "dead" device nodes has been moved from udev
4007 into kmod and tmpfiles. Previously, udev would read the kmod
4008 databases to pre-generate dead device nodes based on meta
4009 information contained in kernel modules, so that these would
4010 be auto-loaded on access rather then at boot. As this
4011 does not really have much to do with the exposing actual
4012 kernel devices to userspace this has always been slightly
4013 alien in the udev codebase. Following the new scheme kmod
4014 will now generate a runtime snippet for tmpfiles from the
4015 module meta information and it now is tmpfiles' job to the
4016 create the nodes. This also allows overriding access and
4017 other parameters for the nodes using the usual tmpfiles
4018 facilities. As side effect this allows us to remove the
4019 CAP_SYS_MKNOD capability bit from udevd entirely.
4021 * logind's device ACLs may now be applied to these "dead"
4022 devices nodes too, thus finally allowing managed access to
4023 devices such as /dev/snd/sequencer whithout loading the
4024 backing module right-away.
4026 * A new RPM macro has been added that may be used to apply
4027 tmpfiles configuration during package installation.
4029 * systemd-detect-virt and ConditionVirtualization= now can
4030 detect User-Mode-Linux machines (UML).
4032 * journald will now implicitly log the effective capabilities
4033 set of processes in the message metadata.
4035 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4037 * The initrd interface has been simplified (more specifically,
4038 support for passing performance data via environment
4039 variables and fsck results via files in /run has been
4040 removed). These features were non-essential, and are
4041 nowadays available in a much nicer way by having systemd in
4042 the initrd serialize its state and have the hosts systemd
4043 deserialize it again.
4045 * The udev "keymap" data files and tools to apply keyboard
4046 specific mappings of scan to key codes, and force-release
4047 scan code lists have been entirely replaced by a udev
4048 "keyboard" builtin and a hwdb data file.
4050 * systemd will now honour the kernel's "quiet" command line
4051 argument also during late shutdown, resulting in a
4052 completely silent shutdown when used.
4054 * There's now an option to control the SO_REUSEPORT socket
4055 option in .socket units.
4057 * Instance units will now automatically get a per-template
4058 subslice of system.slice unless something else is explicitly
4059 configured. For example, instances of sshd@.service will now
4060 implicitly be placed in system-sshd.slice rather than
4061 system.slice as before.
4063 * Test coverage support may now be enabled at build time.
4065 Contributions from: Dave Reisner, Frederic Crozat, Harald
4066 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4067 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4068 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4069 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4070 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4071 Giokas, Zbigniew Jędrzejewski-Szmek
4073 — Berlin, 2013-07-23
4077 * Two new unit types have been introduced:
4079 Scope units are very similar to service units, however, are
4080 created out of pre-existing processes — instead of PID 1
4081 forking off the processes. By using scope units it is
4082 possible for system services and applications to group their
4083 own child processes (worker processes) in a powerful way
4084 which then maybe used to organize them, or kill them
4085 together, or apply resource limits on them.
4087 Slice units may be used to partition system resources in an
4088 hierarchical fashion and then assign other units to them. By
4089 default there are now three slices: system.slice (for all
4090 system services), user.slice (for all user sessions),
4091 machine.slice (for VMs and containers).
4093 Slices and scopes have been introduced primarily in
4094 context of the work to move cgroup handling to a
4095 single-writer scheme, where only PID 1
4096 creates/removes/manages cgroups.
4098 * There's a new concept of "transient" units. In contrast to
4099 normal units these units are created via an API at runtime,
4100 not from configuration from disk. More specifically this
4101 means it is now possible to run arbitrary programs as
4102 independent services, with all execution parameters passed
4103 in via bus APIs rather than read from disk. Transient units
4104 make systemd substantially more dynamic then it ever was,
4105 and useful as a general batch manager.
4107 * logind has been updated to make use of scope and slice units
4108 for managing user sessions. As a user logs in he will get
4109 his own private slice unit, to which all sessions are added
4110 as scope units. We also added support for automatically
4111 adding an instance of user@.service for the user into the
4112 slice. Effectively logind will no longer create cgroup
4113 hierarchies on its own now, it will defer entirely to PID 1
4114 for this by means of scope, service and slice units. Since
4115 user sessions this way become entities managed by PID 1
4116 the output of "systemctl" is now a lot more comprehensive.
4118 * A new mini-daemon "systemd-machined" has been added which
4119 may be used by virtualization managers to register local
4120 VMs/containers. nspawn has been updated accordingly, and
4121 libvirt will be updated shortly. machined will collect a bit
4122 of meta information about the VMs/containers, and assign
4123 them their own scope unit (see above). The collected
4124 meta-data is then made available via the "machinectl" tool,
4125 and exposed in "ps" and similar tools. machined/machinectl
4126 is compile-time optional.
4128 * As discussed earlier, the low-level cgroup configuration
4129 options ControlGroup=, ControlGroupModify=,
4130 ControlGroupPersistent=, ControlGroupAttribute= have been
4131 removed. Please use high-level attribute settings instead as
4132 well as slice units.
4134 * A new bus call SetUnitProperties() has been added to alter
4135 various runtime parameters of a unit. This is primarily
4136 useful to alter cgroup parameters dynamically in a nice way,
4137 but will be extended later on to make more properties
4138 modifiable at runtime. systemctl gained a new set-properties
4139 command that wraps this call.
4141 * A new tool "systemd-run" has been added which can be used to
4142 run arbitrary command lines as transient services or scopes,
4143 while configuring a number of settings via the command
4144 line. This tool is currently very basic, however already
4145 very useful. We plan to extend this tool to even allow
4146 queuing of execution jobs with time triggers from the
4147 command line, similar in fashion to "at".
4149 * nspawn will now inform the user explicitly that kernels with
4150 audit enabled break containers, and suggest the user to turn
4153 * Support for detecting the IMA and AppArmor security
4154 frameworks with ConditionSecurity= has been added.
4156 * journalctl gained a new "-k" switch for showing only kernel
4157 messages, mimicking dmesg output; in addition to "--user"
4158 and "--system" switches for showing only user's own logs
4161 * systemd-delta can now show information about drop-in
4162 snippets extending unit files.
4164 * libsystemd-bus has been substantially updated but is still
4165 not available as public API.
4167 * systemd will now look for the "debug" argument on the kernel
4168 command line and enable debug logging, similar to what
4169 "systemd.log_level=debug" already did before.
4171 * "systemctl set-default", "systemctl get-default" has been
4172 added to configure the default.target symlink, which
4173 controls what to boot into by default.
4175 * "systemctl set-log-level" has been added as a convenient
4176 way to raise and lower systemd logging threshold.
4178 * "systemd-analyze plot" will now show the time the various
4179 generators needed for execution, as well as information
4180 about the unit file loading.
4182 * libsystemd-journal gained a new sd_journal_open_files() call
4183 for opening specific journal files. journactl also gained a
4184 new switch to expose this new functionality. Previously we
4185 only supported opening all files from a directory, or all
4186 files from the system, as opening individual files only is
4187 racy due to journal file rotation.
4189 * systemd gained the new DefaultEnvironment= setting in
4190 /etc/systemd/system.conf to set environment variables for
4193 * If a privileged process logs a journal message with the
4194 OBJECT_PID= field set, then journald will automatically
4195 augment this with additional OBJECT_UID=, OBJECT_GID=,
4196 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4197 system services want to log events about specific client
4198 processes. journactl/systemctl has been updated to make use
4199 of this information if all log messages regarding a specific
4202 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4203 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4204 Reisner, David Coppa, David King, David Strauss, Eelco
4205 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4206 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4207 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4208 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4209 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4210 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4211 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4212 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4213 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4214 Łukasz Stelmach, 장동준
4218 * The Python bindings gained some minimal support for the APIs
4219 exposed by libsystemd-logind.
4221 * ConditionSecurity= gained support for detecting SMACK. Since
4222 this condition already supports SELinux and AppArmor we only
4223 miss IMA for this. Patches welcome!
4225 Contributions from: Karol Lewandowski, Lennart Poettering,
4226 Zbigniew Jędrzejewski-Szmek
4230 * systemd-nspawn will now create /etc/resolv.conf if
4231 necessary, before bind-mounting the host's file onto it.
4233 * systemd-nspawn will now store meta information about a
4234 container on the container's cgroup as extended attribute
4235 fields, including the root directory.
4237 * The cgroup hierarchy has been reworked in many ways. All
4238 objects any of the components systemd creates in the cgroup
4239 tree are now suffixed. More specifically, user sessions are
4240 now placed in cgroups suffixed with ".session", users in
4241 cgroups suffixed with ".user", and nspawn containers in
4242 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4243 names are now escaped in a simple scheme to avoid collision
4244 of userspace object names with kernel filenames. This work
4245 is preparation for making these objects relocatable in the
4246 cgroup tree, in order to allow easy resource partitioning of
4247 these objects without causing naming conflicts.
4249 * systemctl list-dependencies gained the new switches
4250 --plain, --reverse, --after and --before.
4252 * systemd-inhibit now shows the process name of processes that
4253 have taken an inhibitor lock.
4255 * nss-myhostname will now also resolve "localhost"
4256 implicitly. This makes /etc/hosts an optional file and
4257 nicely handles that on IPv6 ::1 maps to both "localhost" and
4260 * libsystemd-logind.so gained a new call
4261 sd_get_machine_names() to enumerate running containers and
4262 VMs (currently only supported by very new libvirt and
4263 nspawn). sd_login_monitor can now be used to watch
4264 VMs/containers coming and going.
4266 * .include is not allowed recursively anymore, and only in
4267 unit files. Usually it is better to use drop-in snippets in
4268 .d/*.conf anyway, as introduced with systemd 198.
4270 * systemd-analyze gained a new "critical-chain" command that
4271 determines the slowest chain of units run during system
4272 boot-up. It is very useful for tracking down where
4273 optimizing boot time is the most beneficial.
4275 * systemd will no longer allow manipulating service paths in
4276 the name=systemd:/system cgroup tree using ControlGroup= in
4277 units. (But is still fine with it in all other dirs.)
4279 * There's a new systemd-nspawn@.service service file that may
4280 be used to easily run nspawn containers as system
4281 services. With the container's root directory in
4282 /var/lib/container/foobar it is now sufficient to run
4283 "systemctl start systemd-nspawn@foobar.service" to boot it.
4285 * systemd-cgls gained a new parameter "--machine" to list only
4286 the processes within a certain container.
4288 * ConditionSecurity= now can check for "apparmor". We still
4289 are lacking checks for SMACK and IMA for this condition
4290 check though. Patches welcome!
4292 * A new configuration file /etc/systemd/sleep.conf has been
4293 added that may be used to configure which kernel operation
4294 systemd is supposed to execute when "suspend", "hibernate"
4295 or "hybrid-sleep" is requested. This makes the new kernel
4296 "freeze" state accessible to the user.
4298 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4299 the passed argument if applicable.
4301 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4302 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4303 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4304 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4305 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4306 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4307 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4312 * The output of 'systemctl list-jobs' got some polishing. The
4313 '--type=' argument may now be passed more than once. A new
4314 command 'systemctl list-sockets' has been added which shows
4315 a list of kernel sockets systemd is listening on with the
4316 socket units they belong to, plus the units these socket
4319 * The experimental libsystemd-bus library got substantial
4320 updates to work in conjunction with the (also experimental)
4321 kdbus kernel project. It works well enough to exchange
4322 messages with some sophistication. Note that kdbus is not
4323 ready yet, and the library is mostly an elaborate test case
4324 for now, and not installable.
4326 * systemd gained a new unit 'systemd-static-nodes.service'
4327 that generates static device nodes earlier during boot, and
4328 can run in conjunction with udev.
4330 * libsystemd-login gained a new call sd_pid_get_user_unit()
4331 to retrieve the user systemd unit a process is running
4332 in. This is useful for systems where systemd is used as
4335 * systemd-nspawn now places all containers in the new /machine
4336 top-level cgroup directory in the name=systemd
4337 hierarchy. libvirt will soon do the same, so that we get a
4338 uniform separation of /system, /user and /machine for system
4339 services, user processes and containers/virtual
4340 machines. This new cgroup hierarchy is also useful to stick
4341 stable names to specific container instances, which can be
4342 recognized later this way (this name may be controlled
4343 via systemd-nspawn's new -M switch). libsystemd-login also
4344 gained a new call sd_pid_get_machine_name() to retrieve the
4345 name of the container/VM a specific process belongs to.
4347 * bootchart can now store its data in the journal.
4349 * libsystemd-journal gained a new call
4350 sd_journal_add_conjunction() for AND expressions to the
4351 matching logic. This can be used to express more complex
4352 logical expressions.
4354 * journactl can now take multiple --unit= and --user-unit=
4357 * The cryptsetup logic now understands the "luks.key=" kernel
4358 command line switch for specifying a file to read the
4359 decryption key from. Also, if a configured key file is not
4360 found the tool will now automatically fall back to prompting
4363 * Python systemd.journal module was updated to wrap recently
4364 added functions from libsystemd-journal. The interface was
4365 changed to bring the low level interface in s.j._Reader
4366 closer to the C API, and the high level interface in
4367 s.j.Reader was updated to wrap and convert all data about
4370 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4371 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4372 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4373 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4374 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4375 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4379 * journalctl --update-catalog now understands a new --root=
4380 option to operate on catalogs found in a different root
4383 * During shutdown after systemd has terminated all running
4384 services a final killing loop kills all remaining left-over
4385 processes. We will now print the name of these processes
4386 when we send SIGKILL to them, since this usually indicates a
4389 * If /etc/crypttab refers to password files stored on
4390 configured mount points automatic dependencies will now be
4391 generated to ensure the specific mount is established first
4392 before the key file is attempted to be read.
4394 * 'systemctl status' will now show information about the
4395 network sockets a socket unit is listening on.
4397 * 'systemctl status' will also shown information about any
4398 drop-in configuration file for units. (Drop-In configuration
4399 files in this context are files such as
4400 /etc/systemd/systemd/foobar.service.d/*.conf)
4402 * systemd-cgtop now optionally shows summed up CPU times of
4403 cgroups. Press '%' while running cgtop to switch between
4404 percentage and absolute mode. This is useful to determine
4405 which cgroups use up the most CPU time over the entire
4406 runtime of the system. systemd-cgtop has also been updated
4407 to be 'pipeable' for processing with further shell tools.
4409 * 'hostnamectl set-hostname' will now allow setting of FQDN
4412 * The formatting and parsing of time span values has been
4413 changed. The parser now understands fractional expressions
4414 such as "5.5h". The formatter will now output fractional
4415 expressions for all time spans under 1min, i.e. "5.123456s"
4416 rather than "5s 123ms 456us". For time spans under 1s
4417 millisecond values are shown, for those under 1ms
4418 microsecond values are shown. This should greatly improve
4419 all time-related output of systemd.
4421 * libsystemd-login and libsystemd-journal gained new
4422 functions for querying the poll() events mask and poll()
4423 timeout value for integration into arbitrary event
4426 * localectl gained the ability to list available X11 keymaps
4427 (models, layouts, variants, options).
4429 * 'systemd-analyze dot' gained the ability to filter for
4430 specific units via shell-style globs, to create smaller,
4431 more useful graphs. I.e. it is now possible to create simple
4432 graphs of all the dependencies between only target units, or
4433 of all units that Avahi has dependencies with.
4435 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4436 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4437 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4438 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4439 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4440 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4441 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4445 * The boot-time readahead implementation for rotating media
4446 will now read the read-ahead data in multiple passes which
4447 consist of all read requests made in equidistant time
4448 intervals. This means instead of strictly reading read-ahead
4449 data in its physical order on disk we now try to find a
4450 middle ground between physical and access time order.
4452 * /etc/os-release files gained a new BUILD_ID= field for usage
4453 on operating systems that provide continuous builds of OS
4456 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4457 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4458 William Douglas, Zbigniew Jędrzejewski-Szmek
4462 * systemd-python gained an API exposing libsystemd-daemon.
4464 * The SMACK setup logic gained support for uploading CIPSO
4467 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4468 ReadOnlyDirectories= and InaccessibleDirectories= has
4469 changed. The private /tmp and /var/tmp directories are now
4470 shared by all processes of a service (which means
4471 ExecStartPre= may now leave data in /tmp that ExecStart= of
4472 the same service can still access). When a service is
4473 stopped its temporary directories are immediately deleted
4474 (normal clean-up with tmpfiles is still done in addition to
4477 * By default, systemd will now set a couple of sysctl
4478 variables in the kernel: the safe sysrq options are turned
4479 on, IP route verification is turned on, and source routing
4480 disabled. The recently added hardlink and softlink
4481 protection of the kernel is turned on. These settings should
4482 be reasonably safe, and good defaults for all new systems.
4484 * The predictable network naming logic may now be turned off
4485 with a new kernel command line switch: net.ifnames=0.
4487 * A new libsystemd-bus module has been added that implements a
4488 pretty complete D-Bus client library. For details see:
4490 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4492 * journald will now explicitly flush the journal files to disk
4493 at the latest 5min after each write. The file will then also
4494 be marked offline until the next write. This should increase
4495 reliability in case of a crash. The synchronization delay
4496 can be configured via SyncIntervalSec= in journald.conf.
4498 * There's a new remote-fs-setup.target unit that can be used
4499 to pull in specific services when at least one remote file
4500 system is to be mounted.
4502 * There are new targets timers.target and paths.target as
4503 canonical targets to pull user timer and path units in
4504 from. This complements sockets.target with a similar
4505 purpose for socket units.
4507 * libudev gained a new call udev_device_set_attribute_value()
4508 to set sysfs attributes of a device.
4510 * The udev daemon now sets the default number of worker
4511 processes executed in parallel based on the number of available
4512 CPUs instead of the amount of available RAM. This is supposed
4513 to provide a more reliable default and limit a too aggressive
4514 paralellism for setups with 1000s of devices connected.
4516 Contributions from: Auke Kok, Colin Walters, Cristian
4517 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4518 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4519 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4520 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4521 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4522 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4523 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4524 Zbigniew Jędrzejewski-Szmek
4528 * Configuration of unit files may now be extended via drop-in
4529 files without having to edit/override the unit files
4530 themselves. More specifically, if the administrator wants to
4531 change one value for a service file foobar.service he can
4532 now do so by dropping in a configuration snippet into
4533 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4534 will load all these snippets and apply them on top of the
4535 main unit configuration file, possibly extending or
4536 overriding its settings. Using these drop-in snippets is
4537 generally nicer than the two earlier options for changing
4538 unit files locally: copying the files from
4539 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4540 them there; or creating a new file in /etc/systemd/system/
4541 that incorporates the original one via ".include". Drop-in
4542 snippets into these .d/ directories can be placed in any
4543 directory systemd looks for units in, and the usual
4544 overriding semantics between /usr/lib, /etc and /run apply
4547 * Most unit file settings which take lists of items can now be
4548 reset by assigning the empty string to them. For example,
4549 normally, settings such as Environment=FOO=BAR append a new
4550 environment variable assignment to the environment block,
4551 each time they are used. By assigning Environment= the empty
4552 string the environment block can be reset to empty. This is
4553 particularly useful with the .d/*.conf drop-in snippets
4554 mentioned above, since this adds the ability to reset list
4555 settings from vendor unit files via these drop-ins.
4557 * systemctl gained a new "list-dependencies" command for
4558 listing the dependencies of a unit recursively.
4560 * Inhibitors are now honored and listed by "systemctl
4561 suspend", "systemctl poweroff" (and similar) too, not only
4562 GNOME. These commands will also list active sessions by
4565 * Resource limits (as exposed by the various control group
4566 controllers) can now be controlled dynamically at runtime
4567 for all units. More specifically, you can now use a command
4568 like "systemctl set-cgroup-attr foobar.service cpu.shares
4569 2000" to alter the CPU shares a specific service gets. These
4570 settings are stored persistently on disk, and thus allow the
4571 administrator to easily adjust the resource usage of
4572 services with a few simple commands. This dynamic resource
4573 management logic is also available to other programs via the
4574 bus. Almost any kernel cgroup attribute and controller is
4577 * systemd-vconsole-setup will now copy all font settings to
4578 all allocated VTs, where it previously applied them only to
4581 * libsystemd-login gained the new sd_session_get_tty() API
4584 * This release drops support for a few legacy or
4585 distribution-specific LSB facility names when parsing init
4586 scripts: $x-display-manager, $mail-transfer-agent,
4587 $mail-transport-agent, $mail-transfer-agent, $smtp,
4588 $null. Also, the mail-transfer-agent.target unit backing
4589 this has been removed. Distributions which want to retain
4590 compatibility with this should carry the burden for
4591 supporting this themselves and patch support for these back
4592 in, if they really need to. Also, the facilities $syslog and
4593 $local_fs are now ignored, since systemd does not support
4594 early-boot LSB init scripts anymore, and these facilities
4595 are implied anyway for normal services. syslog.target has
4598 * There are new bus calls on PID1's Manager object for
4599 cancelling jobs, and removing snapshot units. Previously,
4600 both calls were only available on the Job and Snapshot
4603 * systemd-journal-gatewayd gained SSL support.
4605 * The various "environment" files, such as /etc/locale.conf
4606 now support continuation lines with a backslash ("\") as
4607 last character in the line, similarly in style (but different)
4608 to how this is supported in shells.
4610 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4611 now implicitly appended to every log entry logged. systemctl
4612 has been updated to filter by this field when operating on a
4613 user systemd instance.
4615 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4616 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4617 the container. This makes it easier to boot unmodified
4618 Fedora systems in a container, which however still requires
4619 audit=0 to be passed on the kernel command line. Auditing in
4620 kernel and userspace is unfortunately still too broken in
4621 context of containers, hence we recommend compiling it out
4622 of the kernel or using audit=0. Hopefully this will be fixed
4623 one day for good in the kernel.
4625 * nspawn gained the new --bind= and --bind-ro= parameters to
4626 bind mount specific directories from the host into the
4629 * nspawn will now mount its own devpts file system instance
4630 into the container, in order not to leak pty devices from
4631 the host into the container.
4633 * systemd will now read the firmware boot time performance
4634 information from the EFI variables, if the used boot loader
4635 supports this, and takes it into account for boot performance
4636 analysis via "systemd-analyze". This is currently supported
4637 only in conjunction with Gummiboot, but could be supported
4638 by other boot loaders too. For details see:
4640 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4642 * A new generator has been added that automatically mounts the
4643 EFI System Partition (ESP) to /boot, if that directory
4644 exists, is empty, and no other file system has been
4645 configured to be mounted there.
4647 * logind will now send out PrepareForSleep(false) out
4648 unconditionally, after coming back from suspend. This may be
4649 used by applications as asynchronous notification for
4650 system resume events.
4652 * "systemctl unlock-sessions" has been added, that allows
4653 unlocking the screens of all user sessions at once, similar
4654 to how "systemctl lock-sessions" already locked all users
4655 sessions. This is backed by a new D-Bus call UnlockSessions().
4657 * "loginctl seat-status" will now show the master device of a
4658 seat. (i.e. the device of a seat that needs to be around for
4659 the seat to be considered available, usually the graphics
4662 * tmpfiles gained a new "X" line type, that allows
4663 configuration of files and directories (with wildcards) that
4664 shall be excluded from automatic cleanup ("aging").
4666 * udev default rules set the device node permissions now only
4667 at "add" events, and do not change them any longer with a
4668 later "change" event.
4670 * The log messages for lid events and power/sleep keypresses
4671 now carry a message ID.
4673 * We now have a substantially larger unit test suite, but this
4674 continues to be work in progress.
4676 * udevadm hwdb gained a new --root= parameter to change the
4677 root directory to operate relative to.
4679 * logind will now issue a background sync() request to the kernel
4680 early at shutdown, so that dirty buffers are flushed to disk early
4681 instead of at the last moment, in order to optimize shutdown
4684 * A new bootctl tool has been added that is an interface for
4685 certain boot loader operations. This is currently a preview
4686 and is likely to be extended into a small mechanism daemon
4687 like timedated, localed, hostnamed, and can be used by
4688 graphical UIs to enumerate available boot options, and
4689 request boot into firmware operations.
4691 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4692 the rest of the package. It also has been updated to work
4693 correctly in initrds.
4695 * Policykit previously has been runtime optional, and is now
4696 also compile time optional via a configure switch.
4698 * systemd-analyze has been reimplemented in C. Also "systemctl
4699 dot" has moved into systemd-analyze.
4701 * "systemctl status" with no further parameters will now print
4702 the status of all active or failed units.
4704 * Operations such as "systemctl start" can now be executed
4705 with a new mode "--irreversible" which may be used to queue
4706 operations that cannot accidentally be reversed by a later
4707 job queuing. This is by default used to make shutdown
4708 requests more robust.
4710 * The Python API of systemd now gained a new module for
4711 reading journal files.
4713 * A new tool kernel-install has been added that can install
4714 kernel images according to the Boot Loader Specification:
4716 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4718 * Boot time console output has been improved to provide
4719 animated boot time output for hanging jobs.
4721 * A new tool systemd-activate has been added which can be used
4722 to test socket activation with, directly from the command
4723 line. This should make it much easier to test and debug
4724 socket activation in daemons.
4726 * journalctl gained a new "--reverse" (or -r) option to show
4727 journal output in reverse order (i.e. newest line first).
4729 * journalctl gained a new "--pager-end" (or -e) option to jump
4730 to immediately jump to the end of the journal in the
4731 pager. This is only supported in conjunction with "less".
4733 * journalctl gained a new "--user-unit=" option, that works
4734 similarly to "--unit=" but filters for user units rather than
4737 * A number of unit files to ease adoption of systemd in
4738 initrds has been added. This moves some minimal logic from
4739 the various initrd implementations into systemd proper.
4741 * The journal files are now owned by a new group
4742 "systemd-journal", which exists specifically to allow access
4743 to the journal, and nothing else. Previously, we used the
4744 "adm" group for that, which however possibly covers more
4745 than just journal/log file access. This new group is now
4746 already used by systemd-journal-gatewayd to ensure this
4747 daemon gets access to the journal files and as little else
4748 as possible. Note that "make install" will also set FS ACLs
4749 up for /var/log/journal to give "adm" and "wheel" read
4750 access to it, in addition to "systemd-journal" which owns
4751 the journal files. We recommend that packaging scripts also
4752 add read access to "adm" + "wheel" to /var/log/journal, and
4753 all existing/future journal files. To normal users and
4754 administrators little changes, however packagers need to
4755 ensure to create the "systemd-journal" system group at
4756 package installation time.
4758 * The systemd-journal-gatewayd now runs as unprivileged user
4759 systemd-journal-gateway:systemd-journal-gateway. Packaging
4760 scripts need to create these system user/group at
4763 * timedated now exposes a new boolean property CanNTP that
4764 indicates whether a local NTP service is available or not.
4766 * systemd-detect-virt will now also detect xen PVs
4768 * The pstore file system is now mounted by default, if it is
4771 * In addition to the SELinux and IMA policies we will now also
4772 load SMACK policies at early boot.
4774 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4775 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4776 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4777 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4778 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4779 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4780 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4781 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4782 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4783 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4784 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4785 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4786 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4787 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4791 * Timer units now support calendar time events in addition to
4792 monotonic time events. That means you can now trigger a unit
4793 based on a calendar time specification such as "Thu,Fri
4794 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4795 or fifth day of any month of the year 2013, given that it is
4796 a thursday or friday. This brings timer event support
4797 considerably closer to cron's capabilities. For details on
4798 the supported calendar time specification language see
4801 * udev now supports a number of different naming policies for
4802 network interfaces for predictable names, and a combination
4803 of these policies is now the default. Please see this wiki
4804 document for details:
4806 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4808 * Auke Kok's bootchart implementation has been added to the
4809 systemd tree. It is an optional component that can graph the
4810 boot in quite some detail. It is one of the best bootchart
4811 implementations around and minimal in its code and
4814 * nss-myhostname has been integrated into the systemd source
4815 tree. nss-myhostname guarantees that the local hostname
4816 always stays resolvable via NSS. It has been a weak
4817 requirement of systemd-hostnamed since a long time, and
4818 since its code is actually trivial we decided to just
4819 include it in systemd's source tree. It can be turned off
4820 with a configure switch.
4822 * The read-ahead logic is now capable of properly detecting
4823 whether a btrfs file system is on SSD or rotating media, in
4824 order to optimize the read-ahead scheme. Previously, it was
4825 only capable of detecting this on traditional file systems
4828 * In udev, additional device properties are now read from the
4829 IAB in addition to the OUI database. Also, Bluetooth company
4830 identities are attached to the devices as well.
4832 * In service files %U may be used as specifier that is
4833 replaced by the configured user name of the service.
4835 * nspawn may now be invoked without a controlling TTY. This
4836 makes it suitable for invocation as its own service. This
4837 may be used to set up a simple containerized server system
4838 using only core OS tools.
4840 * systemd and nspawn can now accept socket file descriptors
4841 when they are started for socket activation. This enables
4842 implementation of socket activated nspawn
4843 containers. i.e. think about autospawning an entire OS image
4844 when the first SSH or HTTP connection is received. We expect
4845 that similar functionality will also be added to libvirt-lxc
4848 * journalctl will now suppress ANSI color codes when
4849 presenting log data.
4851 * systemctl will no longer show control group information for
4852 a unit if a the control group is empty anyway.
4854 * logind can now automatically suspend/hibernate/shutdown the
4857 * /etc/machine-info and hostnamed now also expose the chassis
4858 type of the system. This can be used to determine whether
4859 the local system is a laptop, desktop, handset or
4860 tablet. This information may either be configured by the
4861 user/vendor or is automatically determined from ACPI and DMI
4862 information if possible.
4864 * A number of PolicyKit actions are now bound together with
4865 "imply" rules. This should simplify creating UIs because
4866 many actions will now authenticate similar ones as well.
4868 * Unit files learnt a new condition ConditionACPower= which
4869 may be used to conditionalize a unit depending on whether an
4870 AC power source is connected or not, of whether the system
4871 is running on battery power.
4873 * systemctl gained a new "is-failed" verb that may be used in
4874 shell scripts and suchlike to check whether a specific unit
4875 is in the "failed" state.
4877 * The EnvironmentFile= setting in unit files now supports file
4878 globbing, and can hence be used to easily read a number of
4879 environment files at once.
4881 * systemd will no longer detect and recognize specific
4882 distributions. All distribution-specific #ifdeffery has been
4883 removed, systemd is now fully generic and
4884 distribution-agnostic. Effectively, not too much is lost as
4885 a lot of the code is still accessible via explicit configure
4886 switches. However, support for some distribution specific
4887 legacy configuration file formats has been dropped. We
4888 recommend distributions to simply adopt the configuration
4889 files everybody else uses now and convert the old
4890 configuration from packaging scripts. Most distributions
4891 already did that. If that's not possible or desirable,
4892 distributions are welcome to forward port the specific
4893 pieces of code locally from the git history.
4895 * When logging a message about a unit systemd will now always
4896 log the unit name in the message meta data.
4898 * localectl will now also discover system locale data that is
4899 not stored in locale archives, but directly unpacked.
4901 * logind will no longer unconditionally use framebuffer
4902 devices as seat masters, i.e. as devices that are required
4903 to be existing before a seat is considered preset. Instead,
4904 it will now look for all devices that are tagged as
4905 "seat-master" in udev. By default, framebuffer devices will
4906 be marked as such, but depending on local systems, other
4907 devices might be marked as well. This may be used to
4908 integrate graphics cards using closed source drivers (such
4909 as NVidia ones) more nicely into logind. Note however, that
4910 we recommend using the open source NVidia drivers instead,
4911 and no udev rules for the closed-source drivers will be
4912 shipped from us upstream.
4914 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4915 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4916 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4917 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4918 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4919 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4920 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4921 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4922 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4923 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4924 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4925 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4930 * udev gained support for loading additional device properties
4931 from an indexed database that is keyed by vendor/product IDs
4932 and similar device identifiers. For the beginning this
4933 "hwdb" is populated with data from the well-known PCI and
4934 USB database, but also includes PNP, ACPI and OID data. In
4935 the longer run this indexed database shall grow into
4936 becoming the one central database for non-essential
4937 userspace device metadata. Previously, data from the PCI/USB
4938 database was only attached to select devices, since the
4939 lookup was a relatively expensive operation due to O(n) time
4940 complexity (with n being the number of entries in the
4941 database). Since this is now O(1), we decided to add in this
4942 data for all devices where this is available, by
4943 default. Note that the indexed database needs to be rebuilt
4944 when new data files are installed. To achieve this you need
4945 to update your packaging scripts to invoke "udevadm hwdb
4946 --update" after installation of hwdb data files. For
4947 RPM-based distributions we introduced the new
4948 %udev_hwdb_update macro for this purpose.
4950 * The Journal gained support for the "Message Catalog", an
4951 indexed database to link up additional information with
4952 journal entries. For further details please check:
4954 http://www.freedesktop.org/wiki/Software/systemd/catalog
4956 The indexed message catalog database also needs to be
4957 rebuilt after installation of message catalog files. Use
4958 "journalctl --update-catalog" for this. For RPM-based
4959 distributions we introduced the %journal_catalog_update
4960 macro for this purpose.
4962 * The Python Journal bindings gained support for the standard
4963 Python logging framework.
4965 * The Journal API gained new functions for checking whether
4966 the underlying file system of a journal file is capable of
4967 properly reporting file change notifications, or whether
4968 applications that want to reflect journal changes "live"
4969 need to recheck journal files continuously in appropriate
4972 * It is now possible to set the "age" field for tmpfiles
4973 entries to 0, indicating that files matching this entry
4974 shall always be removed when the directories are cleaned up.
4976 * coredumpctl gained a new "gdb" verb which invokes gdb
4977 right-away on the selected coredump.
4979 * There's now support for "hybrid sleep" on kernels that
4980 support this, in addition to "suspend" and "hibernate". Use
4981 "systemctl hybrid-sleep" to make use of this.
4983 * logind's HandleSuspendKey= setting (and related settings)
4984 now gained support for a new "lock" setting to simply
4985 request the screen lock on all local sessions, instead of
4986 actually executing a suspend or hibernation.
4988 * systemd will now mount the EFI variables file system by
4991 * Socket units now gained support for configuration of the
4992 SMACK security label.
4994 * timedatectl will now output the time of the last and next
4995 daylight saving change.
4997 * We dropped support for various legacy and distro-specific
4998 concepts, such as insserv, early-boot SysV services
4999 (i.e. those for non-standard runlevels such as 'b' or 'S')
5000 or ArchLinux /etc/rc.conf support. We recommend the
5001 distributions who still need support this to either continue
5002 to maintain the necessary patches downstream, or find a
5003 different solution. (Talk to us if you have questions!)
5005 * Various systemd components will now bypass PolicyKit checks
5006 for root and otherwise handle properly if PolicyKit is not
5007 found to be around. This should fix most issues for
5008 PolicyKit-less systems. Quite frankly this should have been
5009 this way since day one. It is absolutely our intention to
5010 make systemd work fine on PolicyKit-less systems, and we
5011 consider it a bug if something does not work as it should if
5012 PolicyKit is not around.
5014 * For embedded systems it is now possible to build udev and
5015 systemd without blkid and/or kmod support.
5017 * "systemctl switch-root" is now capable of switching root
5018 more than once. I.e. in addition to transitions from the
5019 initrd to the host OS it is now possible to transition to
5020 further OS images from the host. This is useful to implement
5021 offline updating tools.
5023 * Various other additions have been made to the RPM macros
5024 shipped with systemd. Use %udev_rules_update() after
5025 installing new udev rules files. %_udevhwdbdir,
5026 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5027 %_sysctldir are now available which resolve to the right
5028 directories for packages to place various data files in.
5030 * journalctl gained the new --full switch (in addition to
5031 --all, to disable ellipsation for long messages.
5033 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5034 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5035 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5036 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5037 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5038 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5039 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5040 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5041 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5045 * journalctl gained new --since= and --until= switches to
5046 filter by time. It also now supports nice filtering for
5047 units via --unit=/-u.
5049 * Type=oneshot services may use ExecReload= and do the
5052 * The journal daemon now supports time-based rotation and
5053 vacuuming, in addition to the usual disk-space based
5056 * The journal will now index the available field values for
5057 each field name. This enables clients to show pretty drop
5058 downs of available match values when filtering. The bash
5059 completion of journalctl has been updated
5060 accordingly. journalctl gained a new switch -F to list all
5061 values a certain field takes in the journal database.
5063 * More service events are now written as structured messages
5064 to the journal, and made recognizable via message IDs.
5066 * The timedated, localed and hostnamed mini-services which
5067 previously only provided support for changing time, locale
5068 and hostname settings from graphical DEs such as GNOME now
5069 also have a minimal (but very useful) text-based client
5070 utility each. This is probably the nicest way to changing
5071 these settings from the command line now, especially since
5072 it lists available options and is fully integrated with bash
5075 * There's now a new tool "systemd-coredumpctl" to list and
5076 extract coredumps from the journal.
5078 * We now install a README each in /var/log/ and
5079 /etc/rc.d/init.d explaining where the system logs and init
5080 scripts went. This hopefully should help folks who go to
5081 that dirs and look into the otherwise now empty void and
5082 scratch their heads.
5084 * When user-services are invoked (by systemd --user) the
5085 $MANAGERPID env var is set to the PID of systemd.
5087 * SIGRTMIN+24 when sent to a --user instance will now result
5088 in immediate termination of systemd.
5090 * gatewayd received numerous feature additions such as a
5091 "follow" mode, for live syncing and filtering.
5093 * browse.html now allows filtering and showing detailed
5094 information on specific entries. Keyboard navigation and
5095 mouse screen support has been added.
5097 * gatewayd/journalctl now supports HTML5/JSON
5098 Server-Sent-Events as output.
5100 * The SysV init script compatibility logic will now
5101 heuristically determine whether a script supports the
5102 "reload" verb, and only then make this available as
5105 * "systemctl status --follow" has been removed, use "journalctl
5108 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5109 have been removed since they are hardly useful to be
5112 * And I'd like to take the opportunity to specifically mention
5113 Zbigniew for his great contributions. Zbigniew, you rock!
5115 Contributions from: Andrew Eikum, Christian Hesse, Colin
5116 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5117 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5118 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5119 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5120 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5121 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5125 * If /etc/vconsole.conf is non-existent or empty we will no
5126 longer load any console font or key map at boot by
5127 default. Instead the kernel defaults will be left
5128 intact. This is definitely the right thing to do, as no
5129 configuration should mean no configuration, and hard-coding
5130 font names that are different on all archs is probably a bad
5131 idea. Also, the kernel default key map and font should be
5132 good enough for most cases anyway, and mostly identical to
5133 the userspace fonts/key maps we previously overloaded them
5134 with. If distributions want to continue to default to a
5135 non-kernel font or key map they should ship a default
5136 /etc/vconsole.conf with the appropriate contents.
5138 Contributions from: Colin Walters, Daniel J Walsh, Dave
5139 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5140 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5144 * journalctl gained a new --cursor= switch to show entries
5145 starting from the specified location in the journal.
5147 * We now enforce a size limit on journal entry fields exported
5148 with "-o json" in journalctl. Fields larger than 4K will be
5149 assigned null. This can be turned off with --all.
5151 * An (optional) journal gateway daemon is now available as
5152 "systemd-journal-gatewayd.service". This service provides
5153 access to the journal via HTTP and JSON. This functionality
5154 will be used to implement live log synchronization in both
5155 pull and push modes, but has various other users too, such
5156 as easy log access for debugging of embedded devices. Right
5157 now it is already useful to retrieve the journal via HTTP:
5159 # systemctl start systemd-journal-gatewayd.service
5160 # wget http://localhost:19531/entries
5162 This will download the journal contents in a
5163 /var/log/messages compatible format. The same as JSON:
5165 # curl -H"Accept: application/json" http://localhost:19531/entries
5167 This service is also accessible via a web browser where a
5168 single static HTML5 app is served that uses the JSON logic
5169 to enable the user to do some basic browsing of the
5170 journal. This will be extended later on. Here's an example
5171 screenshot of this app in its current state:
5173 http://0pointer.de/public/journal-gatewayd
5175 Contributions from: Kay Sievers, Lennart Poettering, Robert
5176 Milasan, Tom Gundersen
5180 * The bash completion logic is now available for journalctl
5183 * We do not mount the "cpuset" controller anymore together with
5184 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5185 started if no parameters are assigned to it. "cpuset" hence
5186 broke code that assumed it it could create "cpu" groups and
5189 * journalctl -f will now subscribe to terminal size changes,
5190 and line break accordingly.
5192 Contributions from: Dave Reisner, Kay Sievers, Lennart
5193 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5197 * nspawn will now create a symlink /etc/localtime in the
5198 container environment, copying the host's timezone
5199 setting. Previously this has been done via a bind mount, but
5200 since symlinks cannot be bind mounted this has now been
5201 changed to create/update the appropriate symlink.
5203 * journalctl -n's line number argument is now optional, and
5204 will default to 10 if omitted.
5206 * journald will now log the maximum size the journal files may
5207 take up on disk. This is particularly useful if the default
5208 built-in logic of determining this parameter from the file
5209 system size is used. Use "systemctl status
5210 systemd-journald.service" to see this information.
5212 * The multi-seat X wrapper tool has been stripped down. As X
5213 is now capable of enumerating graphics devices via udev in a
5214 seat-aware way the wrapper is not strictly necessary
5215 anymore. A stripped down temporary stop-gap is still shipped
5216 until the upstream display managers have been updated to
5217 fully support the new X logic. Expect this wrapper to be
5218 removed entirely in one of the next releases.
5220 * HandleSleepKey= in logind.conf has been split up into
5221 HandleSuspendKey= and HandleHibernateKey=. The old setting
5222 is not available anymore. X11 and the kernel are
5223 distinguishing between these keys and we should too. This
5224 also means the inhibition lock for these keys has been split
5227 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5228 Poettering, Lukas Nykryn, Václav Pavlín
5232 * Whenever a unit changes state we will now log this to the
5233 journal and show along the unit's own log output in
5236 * ConditionPathIsMountPoint= can now properly detect bind
5237 mount points too. (Previously, a bind mount of one file
5238 system to another place in the same file system could not be
5239 detected as mount, since they shared struct stat's st_dev
5242 * We will now mount the cgroup controllers cpu, cpuacct,
5243 cpuset and the controllers net_cls, net_prio together by
5246 * nspawn containers will now have a virtualized boot
5247 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5248 over with a randomized ID at container initialization). This
5249 has the effect of making "journalctl -b" do the right thing
5252 * The JSON output journal serialization has been updated not
5253 to generate "endless" list objects anymore, but rather one
5254 JSON object per line. This is more in line how most JSON
5255 parsers expect JSON objects. The new output mode
5256 "json-pretty" has been added to provide similar output, but
5257 neatly aligned for readability by humans.
5259 * We dropped all explicit sync() invocations in the shutdown
5260 code. The kernel does this implicitly anyway in the kernel
5261 reboot() syscall. halt(8)'s -n option is now a compatibility
5264 * We now support virtualized reboot() in containers, as
5265 supported by newer kernels. We will fall back to exit() if
5266 CAP_SYS_REBOOT is not available to the container. Also,
5267 nspawn makes use of this now and will actually reboot the
5268 container if the containerized OS asks for that.
5270 * journalctl will only show local log output by default
5271 now. Use --merge (-m) to show remote log output, too.
5273 * libsystemd-journal gained the new sd_journal_get_usage()
5274 call to determine the current disk usage of all journal
5275 files. This is exposed in the new "journalctl --disk-usage"
5278 * journald gained a new configuration setting SplitMode= in
5279 journald.conf which may be used to control how user journals
5280 are split off. See journald.conf(5) for details.
5282 * A new condition type ConditionFileNotEmpty= has been added.
5284 * tmpfiles' "w" lines now support file globbing, to write
5285 multiple files at once.
5287 * We added Python bindings for the journal submission
5288 APIs. More Python APIs for a number of selected APIs will
5289 likely follow. Note that we intend to add native bindings
5290 only for the Python language, as we consider it common
5291 enough to deserve bindings shipped within systemd. There are
5292 various projects outside of systemd that provide bindings
5293 for languages such as PHP or Lua.
5295 * Many conditions will now resolve specifiers such as %i. In
5296 addition, PathChanged= and related directives of .path units
5297 now support specifiers as well.
5299 * There's now a new RPM macro definition for the system preset
5302 * journald will now warn if it ca not forward a message to the
5303 syslog daemon because its socket is full.
5305 * timedated will no longer write or process /etc/timezone,
5306 except on Debian. As we do not support late mounted /usr
5307 anymore /etc/localtime always being a symlink is now safe,
5308 and hence the information in /etc/timezone is not necessary
5311 * logind will now always reserve one VT for a text getty (VT6
5312 by default). Previously if more than 6 X sessions where
5313 started they took up all the VTs with auto-spawned gettys,
5314 so that no text gettys were available anymore.
5316 * udev will now automatically inform the btrfs kernel logic
5317 about btrfs RAID components showing up. This should make
5318 simple hotplug based btrfs RAID assembly work.
5320 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5321 (but not for its children which will stay at the kernel
5322 default). This should allow setups with a lot more listening
5325 * systemd will now always pass the configured timezone to the
5326 kernel at boot. timedated will do the same when the timezone
5329 * logind's inhibition logic has been updated. By default,
5330 logind will now handle the lid switch, the power and sleep
5331 keys all the time, even in graphical sessions. If DEs want
5332 to handle these events on their own they should take the new
5333 handle-power-key, handle-sleep-key and handle-lid-switch
5334 inhibitors during their runtime. A simple way to achieve
5335 that is to invoke the DE wrapped in an invocation of:
5337 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5339 * Access to unit operations is now checked via SELinux taking
5340 the unit file label and client process label into account.
5342 * systemd will now notify the administrator in the journal
5343 when he over-mounts a non-empty directory.
5345 * There are new specifiers that are resolved in unit files,
5346 for the host name (%H), the machine ID (%m) and the boot ID
5349 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5350 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5351 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5352 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5353 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5354 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5355 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5359 * Support for reading structured kernel messages from
5360 /dev/kmsg has now been added and is enabled by default.
5362 * Support for reading kernel messages from /proc/kmsg has now
5363 been removed. If you want kernel messages in the journal
5364 make sure to run a recent kernel (>= 3.5) that supports
5365 reading structured messages from /dev/kmsg (see
5366 above). /proc/kmsg is now exclusive property of classic
5367 syslog daemons again.
5369 * The libudev API gained the new
5370 udev_device_new_from_device_id() call.
5372 * The logic for file system namespace (ReadOnlyDirectory=,
5373 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5374 require pivot_root() anymore. This means fewer temporary
5375 directories are created below /tmp for this feature.
5377 * nspawn containers will now see and receive all submounts
5378 made on the host OS below the root file system of the
5381 * Forward Secure Sealing is now supported for Journal files,
5382 which provide cryptographical sealing of journal files so
5383 that attackers cannot alter log history anymore without this
5384 being detectable. Lennart will soon post a blog story about
5385 this explaining it in more detail.
5387 * There are two new service settings RestartPreventExitStatus=
5388 and SuccessExitStatus= which allow configuration of exit
5389 status (exit code or signal) which will be excepted from the
5390 restart logic, resp. consider successful.
5392 * journalctl gained the new --verify switch that can be used
5393 to check the integrity of the structure of journal files and
5394 (if Forward Secure Sealing is enabled) the contents of
5397 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5398 and similar symlinks pre-created. This makes running shells
5399 as container init process a lot more fun.
5401 * The fstab support can now handle PARTUUID= and PARTLABEL=
5404 * A new ConditionHost= condition has been added to match
5405 against the hostname (with globs) and machine ID. This is
5406 useful for clusters where a single OS image is used to
5407 provision a large number of hosts which shall run slightly
5408 different sets of services.
5410 * Services which hit the restart limit will now be placed in a
5413 Contributions from: Bertram Poettering, Dave Reisner, Huang
5414 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5415 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5419 * When running in --user mode systemd will now become a
5420 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5421 tree a lot more organized.
5423 * A new PartOf= unit dependency type has been introduced that
5424 may be used to group services in a natural way.
5426 * "systemctl enable" may now be used to enable instances of
5429 * journalctl now prints error log levels in red, and
5430 warning/notice log levels in bright white. It also supports
5431 filtering by log level now.
5433 * cgtop gained a new -n switch (similar to top), to configure
5434 the maximum number of iterations to run for. It also gained
5435 -b, to run in batch mode (accepting no input).
5437 * The suffix ".service" may now be omitted on most systemctl
5438 command lines involving service unit names.
5440 * There's a new bus call in logind to lock all sessions, as
5441 well as a loginctl verb for it "lock-sessions".
5443 * libsystemd-logind.so gained a new call sd_journal_perror()
5444 that works similar to libc perror() but logs to the journal
5445 and encodes structured information about the error number.
5447 * /etc/crypttab entries now understand the new keyfile-size=
5450 * shutdown(8) now can send a (configurable) wall message when
5451 a shutdown is cancelled.
5453 * The mount propagation mode for the root file system will now
5454 default to "shared", which is useful to make containers work
5455 nicely out-of-the-box so that they receive new mounts from
5456 the host. This can be undone locally by running "mount
5457 --make-rprivate /" if needed.
5459 * The prefdm.service file has been removed. Distributions
5460 should maintain this unit downstream if they intend to keep
5461 it around. However, we recommend writing normal unit files
5462 for display managers instead.
5464 * Since systemd is a crucial part of the OS we will now
5465 default to a number of compiler switches that improve
5466 security (hardening) such as read-only relocations, stack
5467 protection, and suchlike.
5469 * The TimeoutSec= setting for services is now split into
5470 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5471 of individual time outs for the start and the stop phase of
5474 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5475 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5476 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5477 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5478 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5479 Gundersen, Zbigniew Jędrzejewski-Szmek
5483 * The journal and id128 C APIs are now fully documented as man
5486 * Extra safety checks have been added when transitioning from
5487 the initial RAM disk to the main system to avoid accidental
5490 * /etc/crypttab entries now understand the new keyfile-offset=
5493 * systemctl -t can now be used to filter by unit load state.
5495 * The journal C API gained the new sd_journal_wait() call to
5496 make writing synchronous journal clients easier.
5498 * journalctl gained the new -D switch to show journals from a
5501 * journalctl now displays a special marker between log
5502 messages of two different boots.
5504 * The journal is now explicitly flushed to /var via a service
5505 systemd-journal-flush.service, rather than implicitly simply
5506 by seeing /var/log/journal to be writable.
5508 * journalctl (and the journal C APIs) can now match for much
5509 more complex expressions, with alternatives and
5512 * When transitioning from the initial RAM disk to the main
5513 system we will now kill all processes in a killing spree to
5514 ensure no processes stay around by accident.
5516 * Three new specifiers may be used in unit files: %u, %h, %s
5517 resolve to the user name, user home directory resp. user
5518 shell. This is useful for running systemd user instances.
5520 * We now automatically rotate journal files if their data
5521 object hash table gets a fill level > 75%. We also size the
5522 hash table based on the configured maximum file size. This
5523 together should lower hash collisions drastically and thus
5524 speed things up a bit.
5526 * journalctl gained the new "--header" switch to introspect
5527 header data of journal files.
5529 * A new setting SystemCallFilters= has been added to services
5530 which may be used to apply blacklists or whitelists to
5531 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5533 * nspawn gained a new --link-journal= switch (and quicker: -j)
5534 to link the container journal with the host. This makes it
5535 very easy to centralize log viewing on the host for all
5536 guests while still keeping the journal files separated.
5538 * Many bugfixes and optimizations
5540 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5541 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5542 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5547 * Several tools now understand kernel command line arguments,
5548 which are only read when run in an initial RAM disk. They
5549 usually follow closely their normal counterparts, but are
5552 * There's a new tool to analyze the readahead files that are
5553 automatically generated at boot. Use:
5555 /usr/lib/systemd/systemd-readahead analyze /.readahead
5557 * We now provide an early debug shell on tty9 if this enabled. Use:
5559 systemctl enable debug-shell.service
5561 * All plymouth related units have been moved into the Plymouth
5562 package. Please make sure to upgrade your Plymouth version
5565 * systemd-tmpfiles now supports getting passed the basename of
5566 a configuration file only, in which case it will look for it
5567 in all appropriate directories automatically.
5569 * udevadm info now takes a /dev or /sys path as argument, and
5570 does the right thing. Example:
5572 udevadm info /dev/sda
5573 udevadm info /sys/class/block/sda
5575 * systemctl now prints a warning if a unit is stopped but a
5576 unit that might trigger it continues to run. Example: a
5577 service is stopped but the socket that activates it is left
5580 * "systemctl status" will now mention if the log output was
5581 shortened due to rotation since a service has been started.
5583 * The journal API now exposes functions to determine the
5584 "cutoff" times due to rotation.
5586 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5587 immediately flushing of runtime logs to /var if possible,
5588 resp. for triggering immediate rotation of the journal
5591 * It is now considered an error if a service is attempted to
5592 be stopped that is not loaded.
5594 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5596 * systemd-analyze now supports Python 3
5598 * tmpfiles now supports cleaning up directories via aging
5599 where the first level dirs are always kept around but
5600 directories beneath it automatically aged. This is enabled
5601 by prefixing the age field with '~'.
5603 * Seat objects now expose CanGraphical, CanTTY properties
5604 which is required to deal with very fast bootups where the
5605 display manager might be running before the graphics drivers
5606 completed initialization.
5608 * Seat objects now expose a State property.
5610 * We now include RPM macros for service enabling/disabling
5611 based on the preset logic. We recommend RPM based
5612 distributions to make use of these macros if possible. This
5613 makes it simpler to reuse RPM spec files across
5616 * We now make sure that the collected systemd unit name is
5617 always valid when services log to the journal via
5620 * There's a new man page kernel-command-line(7) detailing all
5621 command line options we understand.
5623 * The fstab generator may now be disabled at boot by passing
5624 fstab=0 on the kernel command line.
5626 * A new kernel command line option modules-load= is now understood
5627 to load a specific kernel module statically, early at boot.
5629 * Unit names specified on the systemctl command line are now
5630 automatically escaped as needed. Also, if file system or
5631 device paths are specified they are automatically turned
5632 into the appropriate mount or device unit names. Example:
5634 systemctl status /home
5635 systemctl status /dev/sda
5637 * The SysVConsole= configuration option has been removed from
5638 system.conf parsing.
5640 * The SysV search path is no longer exported on the D-Bus
5643 * The Names= option is been removed from unit file parsing.
5645 * There's a new man page bootup(7) detailing the boot process.
5647 * Every unit and every generator we ship with systemd now
5648 comes with full documentation. The self-explanatory boot is
5651 * A couple of services gained "systemd-" prefixes in their
5652 name if they wrap systemd code, rather than only external
5653 code. Among them fsck@.service which is now
5654 systemd-fsck@.service.
5656 * The HaveWatchdog property has been removed from the D-Bus
5659 * systemd.confirm_spawn= on the kernel command line should now
5662 * There's a new man page crypttab(5) which details all options
5663 we actually understand.
5665 * systemd-nspawn gained a new --capability= switch to pass
5666 additional capabilities to the container.
5668 * timedated will now read known NTP implementation unit names
5669 from /usr/lib/systemd/ntp-units.d/*.list,
5670 systemd-timedated-ntp.target has been removed.
5672 * journalctl gained a new switch "-b" that lists log data of
5673 the current boot only.
5675 * The notify socket is in the abstract namespace again, in
5676 order to support daemons which chroot() at start-up.
5678 * There is a new Storage= configuration option for journald
5679 which allows configuration of where log data should go. This
5680 also provides a way to disable journal logging entirely, so
5681 that data collected is only forwarded to the console, the
5682 kernel log buffer or another syslog implementation.
5684 * Many bugfixes and optimizations
5686 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5687 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5688 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5689 Shawn Landden, Tom Gundersen
5693 * "systemctl help <unit>" now shows the man page if one is
5696 * Several new man pages have been added.
5698 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5699 MaxLevelConsole= can now be specified in
5700 journald.conf. These options allow reducing the amount of
5701 data stored on disk or forwarded by the log level.
5703 * TimerSlackNSec= can now be specified in system.conf for
5704 PID1. This allows system-wide power savings.
5706 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5707 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5712 * logind is now capable of (optionally) handling power and
5713 sleep keys as well as the lid switch.
5715 * journalctl now understands the syntax "journalctl
5716 /usr/bin/avahi-daemon" to get all log output of a specific
5719 * CapabilityBoundingSet= in system.conf now also influences
5720 the capability bound set of usermode helpers of the kernel.
5722 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5723 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5724 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5725 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5729 * Note that we skipped 139 releases here in order to set the
5730 new version to something that is greater than both udev's
5731 and systemd's most recent version number.
5733 * udev: all udev sources are merged into the systemd source tree now.
5734 All future udev development will happen in the systemd tree. It
5735 is still fully supported to use the udev daemon and tools without
5736 systemd running, like in initramfs or other init systems. Building
5737 udev though, will require the *build* of the systemd tree, but
5738 udev can be properly *run* without systemd.
5740 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5741 should be used to create dead device nodes as workarounds for broken
5744 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5745 no longer supported. udev_monitor_new_from_netlink() needs to be
5746 used to subscribe to events.
5748 * udev: when udevd is started by systemd, processes which are left
5749 behind by forking them off of udev rules, are unconditionally cleaned
5750 up and killed now after the event handling has finished. Services or
5751 daemons must be started as systemd services. Services can be
5752 pulled-in by udev to get started, but they can no longer be directly
5753 forked by udev rules.
5755 * udev: the daemon binary is called systemd-udevd now and installed
5756 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5757 to adapt to that, create symlink, or rename the binary after building
5760 * libudev no longer provides these symbols:
5761 udev_monitor_from_socket()
5762 udev_queue_get_failed_list_entry()
5763 udev_get_{dev,sys,run}_path()
5764 The versions number was bumped and symbol versioning introduced.
5766 * systemd-loginctl and systemd-journalctl have been renamed
5767 to loginctl and journalctl to match systemctl.
5769 * The config files: /etc/systemd/systemd-logind.conf and
5770 /etc/systemd/systemd-journald.conf have been renamed to
5771 logind.conf and journald.conf. Package updates should rename
5772 the files to the new names on upgrade.
5774 * For almost all files the license is now LGPL2.1+, changed
5775 from the previous GPL2.0+. Exceptions are some minor stuff
5776 of udev (which will be changed to LGPL2.1 eventually, too),
5777 and the MIT licensed sd-daemon.[ch] library that is suitable
5778 to be used as drop-in files.
5780 * systemd and logind now handle system sleep states, in
5781 particular suspending and hibernating.
5783 * logind now implements a sleep/shutdown/idle inhibiting logic
5784 suitable for a variety of uses. Soonishly Lennart will blog
5785 about this in more detail.
5787 * var-run.mount and var-lock.mount are no longer provided
5788 (which prevously bind mounted these directories to their new
5789 places). Distributions which have not converted these
5790 directories to symlinks should consider stealing these files
5791 from git history and add them downstream.
5793 * We introduced the Documentation= field for units and added
5794 this to all our shipped units. This is useful to make it
5795 easier to explore the boot and the purpose of the various
5798 * All smaller setup units (such as
5799 systemd-vconsole-setup.service) now detect properly if they
5800 are run in a container and are skipped when
5801 appropriate. This guarantees an entirely noise-free boot in
5802 Linux container environments such as systemd-nspawn.
5804 * A framework for implementing offline system updates is now
5805 integrated, for details see:
5806 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5808 * A new service type Type=idle is available now which helps us
5809 avoiding ugly interleaving of getty output and boot status
5812 * There's now a system-wide CapabilityBoundingSet= option to
5813 globally reduce the set of capabilities for the
5814 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5815 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5816 even CAP_NET_ADMIN system-wide for secure systems.
5818 * There are now system-wide DefaultLimitXXX= options to
5819 globally change the defaults of the various resource limits
5820 for all units started by PID 1.
5822 * Harald Hoyer's systemd test suite has been integrated into
5823 systemd which allows easy testing of systemd builds in qemu
5824 and nspawn. (This is really awesome! Ask us for details!)
5826 * The fstab parser is now implemented as generator, not inside
5829 * systemctl will now warn you if .mount units generated from
5830 /etc/fstab are out of date due to changes in fstab that
5831 have not been read by systemd yet.
5833 * systemd is now suitable for usage in initrds. Dracut has
5834 already been updated to make use of this. With this in place
5835 initrds get a slight bit faster but primarily are much
5836 easier to introspect and debug since "systemctl status" in
5837 the host system can be used to introspect initrd services,
5838 and the journal from the initrd is kept around too.
5840 * systemd-delta has been added, a tool to explore differences
5841 between user/admin configuration and vendor defaults.
5843 * PrivateTmp= now affects both /tmp and /var/tmp.
5845 * Boot time status messages are now much prettier and feature
5846 proper english language. Booting up systemd has never been
5849 * Read-ahead pack files now include the inode number of all
5850 files to pre-cache. When the inode changes the pre-caching
5851 is not attempted. This should be nicer to deal with updated
5852 packages which might result in changes of read-ahead
5855 * We now temporaritly lower the kernel's read_ahead_kb variable
5856 when collecting read-ahead data to ensure the kernel's
5857 built-in read-ahead does not add noise to our measurements
5858 of necessary blocks to pre-cache.
5860 * There's now RequiresMountsFor= to add automatic dependencies
5861 for all mounts necessary for a specific file system path.
5863 * MountAuto= and SwapAuto= have been removed from
5864 system.conf. Mounting file systems at boot has to take place
5867 * nspawn now learned a new switch --uuid= to set the machine
5868 ID on the command line.
5870 * nspawn now learned the -b switch to automatically search
5873 * vt102 is now the default TERM for serial TTYs, upgraded from
5876 * systemd-logind now works on VT-less systems.
5878 * The build tree has been reorganized. The individual
5879 components now have directories of their own.
5881 * A new condition type ConditionPathIsReadWrite= is now available.
5883 * nspawn learned the new -C switch to create cgroups for the
5884 container in other hierarchies.
5886 * We now have support for hardware watchdogs, configurable in
5889 * The scheduled shutdown logic now has a public API.
5891 * We now mount /tmp as tmpfs by default, but this can be
5892 masked and /etc/fstab can override it.
5894 * Since udisks does not make use of /media anymore we are not
5895 mounting a tmpfs on it anymore.
5897 * journalctl gained a new --local switch to only interleave
5898 locally generated journal files.
5900 * We can now load the IMA policy at boot automatically.
5902 * The GTK tools have been split off into a systemd-ui.
5904 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5905 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5906 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5907 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5908 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5909 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5910 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5911 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5912 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5917 * This is mostly a bugfix release
5919 * Support optional initialization of the machine ID from the
5920 KVM or container configured UUID.
5922 * Support immediate reboots with "systemctl reboot -ff"
5924 * Show /etc/os-release data in systemd-analyze output
5926 * Many bugfixes for the journal, including endianness fixes and
5927 ensuring that disk space enforcement works
5929 * sd-login.h is C++ comptaible again
5931 * Extend the /etc/os-release format on request of the Debian
5934 * We now refuse non-UTF8 strings used in various configuration
5935 and unit files. This is done to ensure we do not pass invalid
5936 data over D-Bus or expose it elsewhere.
5938 * Register Mimo USB Screens as suitable for automatic seat
5941 * Read SELinux client context from journal clients in a race
5944 * Reorder configuration file lookup order. /etc now always
5945 overrides /run in order to allow the administrator to always
5946 and unconditionally override vendor-supplied or
5947 automatically generated data.
5949 * The various user visible bits of the journal now have man
5950 pages. We still lack man pages for the journal API calls
5953 * We now ship all man pages in HTML format again in the
5956 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5957 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5958 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5963 * This is mostly a bugfix release
5965 * systems lacking /etc/os-release are no longer supported.
5967 * Various functionality updates to libsystemd-login.so
5969 * Track class of PAM logins to distinguish greeters from
5972 Contributions from: Kay Sievers, Lennart Poettering, Michael
5977 * This is an important bugfix release for v41.
5979 * Building man pages is now optional which should be useful
5980 for those building systemd from git but unwilling to install
5983 * Watchdog support for supervising services is now usable. In
5984 a future release support for hardware watchdogs
5985 (i.e. /dev/watchdog) will be added building on this.
5987 * Service start rate limiting is now configurable and can be
5988 turned off per service. When a start rate limit is hit a
5989 reboot can automatically be triggered.
5991 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5993 Contributions from: Benjamin Franzke, Bill Nottingham,
5994 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5995 Schmidt, Michał Górny, Piotr Drąg
5999 * The systemd binary is installed /usr/lib/systemd/systemd now;
6000 An existing /sbin/init symlink needs to be adapted with the
6003 * The code that loads kernel modules has been ported to invoke
6004 libkmod directly, instead of modprobe. This means we do not
6005 support systems with module-init-tools anymore.
6007 * Watchdog support is now already useful, but still not
6010 * A new kernel command line option systemd.setenv= is
6011 understood to set system wide environment variables
6012 dynamically at boot.
6014 * We now limit the set of capabilities of systemd-journald.
6016 * We now set SIGPIPE to ignore by default, since it only is
6017 useful in shell pipelines, and has little use in general
6018 code. This can be disabled with IgnoreSIPIPE=no in unit
6021 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6022 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6027 * This is mostly a bugfix release
6029 * We now expose the reason why a service failed in the
6030 "Result" D-Bus property.
6032 * Rudimentary service watchdog support (will be completed over
6033 the next few releases.)
6035 * When systemd forks off in order execute some service we will
6036 now immediately changes its argv[0] to reflect which process
6037 it will execute. This is useful to minimize the time window
6038 with a generic argv[0], which makes bootcharts more useful
6040 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6041 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6042 Mike Kazantsev, Ray Strode
6046 * This is mostly a test release, but incorporates many
6049 * New systemd-cgtop tool to show control groups by their
6052 * Linking against libacl for ACLs is optional again. If
6053 disabled, support tracking device access for active logins
6054 goes becomes unavailable, and so does access to the user
6055 journals by the respective users.
6057 * If a group "adm" exists, journal files are automatically
6058 owned by them, thus allow members of this group full access
6059 to the system journal as well as all user journals.
6061 * The journal now stores the SELinux context of the logging
6062 client for all entries.
6064 * Add C++ inclusion guards to all public headers
6066 * New output mode "cat" in the journal to print only text
6067 messages, without any meta data like date or time.
6069 * Include tiny X server wrapper as a temporary stop-gap to
6070 teach XOrg udev display enumeration. This is used by display
6071 managers such as gdm, and will go away as soon as XOrg
6072 learned native udev hotplugging for display devices.
6074 * Add new systemd-cat tool for executing arbitrary programs
6075 with STDERR/STDOUT connected to the journal. Can also act as
6076 BSD logger replacement, and does so by default.
6078 * Optionally store all locally generated coredumps in the
6079 journal along with meta data.
6081 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6082 writing short strings to files (for usage for /sys), and for
6083 creating symlinks, character and block device nodes.
6085 * New unit file option ControlGroupPersistent= to make cgroups
6086 persistent, following the mechanisms outlined in
6087 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6089 * Support multiple local RTCs in a sane way
6091 * No longer monopolize IO when replaying readahead data on
6092 rotating disks, since we might starve non-file-system IO to
6093 death, since fanotify() will not see accesses done by blkid,
6096 * Do not show kernel threads in systemd-cgls anymore, unless
6097 requested with new -k switch.
6099 Contributions from: Dan Horák, Kay Sievers, Lennart
6100 Poettering, Michal Schmidt
6104 * This is mostly a test release, but incorporates many
6107 * The git repository moved to:
6108 git://anongit.freedesktop.org/systemd/systemd
6109 ssh://git.freedesktop.org/git/systemd/systemd
6111 * First release with the journal
6112 http://0pointer.de/blog/projects/the-journal.html
6114 * The journal replaces both systemd-kmsg-syslogd and
6115 systemd-stdout-bridge.
6117 * New sd_pid_get_unit() API call in libsystemd-logind
6119 * Many systemadm clean-ups
6121 * Introduce remote-fs-pre.target which is ordered before all
6122 remote mounts and may be used to start services before all
6125 * Added Mageia support
6127 * Add bash completion for systemd-loginctl
6129 * Actively monitor PID file creation for daemons which exit in
6130 the parent process before having finished writing the PID
6131 file in the daemon process. Daemons which do this need to be
6132 fixed (i.e. PID file creation must have finished before the
6133 parent exits), but we now react a bit more gracefully to them.
6135 * Add colourful boot output, mimicking the well-known output
6136 of existing distributions.
6138 * New option PassCredentials= for socket units, for
6139 compatibility with a recent kernel ABI breakage.
6141 * /etc/rc.local is now hooked in via a generator binary, and
6142 thus will no longer act as synchronization point during
6145 * systemctl list-unit-files now supports --root=.
6147 * systemd-tmpfiles now understands two new commands: z, Z for
6148 relabelling files according to the SELinux database. This is
6149 useful to apply SELinux labels to specific files in /sys,
6152 * Output of SysV services is now forwarded to both the console
6153 and the journal by default, not only just the console.
6155 * New man pages for all APIs from libsystemd-login.
6157 * The build tree got reorganized and a the build system is a
6158 lot more modular allowing embedded setups to specifically
6159 select the components of systemd they are interested in.
6161 * Support for Linux systems lacking the kernel VT subsystem is
6164 * configure's --with-rootdir= got renamed to
6165 --with-rootprefix= to follow the naming used by udev and
6168 * Unless specified otherwise we will now install to /usr instead
6169 of /usr/local by default.
6171 * Processes with '@' in argv[0][0] are now excluded from the
6172 final shut-down killing spree, following the logic explained
6174 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6176 * All processes remaining in a service cgroup when we enter
6177 the START or START_PRE states are now killed with
6178 SIGKILL. That means it is no longer possible to spawn
6179 background processes from ExecStart= lines (which was never
6180 supported anyway, and bad style).
6182 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6183 reloading of units together.
6185 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6186 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6187 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6188 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6189 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek