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 Contributions from: Alban Crequy, Alexander Kuleshov, Alex Crawford,
176 Andrew Eikum, Beniamino Galvani, Benjamin Robin, Benjamin ROBIN, Biao
177 Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Colin Guthrie, Daniel
178 J Walsh, Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
179 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
180 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
181 Bui, frankheckenbach, Georgia Brikis, Harald Hoyer, Hendrik Brueckner,
182 Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo Puustinen, Jakub
183 Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth, kayrus, Klearchos
184 Chaloulos, Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukáš
185 Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt, Michael Biebl,
186 michaelolbrich, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletar,
187 Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin, mulkieran,
188 muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween, Nicolas
189 Braud-Santoni, Patrik Flykt, Peter Hutterer, Petr Lautrbach, Petros
190 Angelatos, Piotr Drąg, Rabin Vincent, Robert Węcławski, Ronny
191 Chevalier, Samuel Tardieu, Stefan Schallenberg, Steven Siloti, Susant
192 Sahani, Sylvain Plantefève, Taylor Smock, tblume, Tejun Heo, Thomas
193 Blume, Thomas Haller, Thomas Hindoe Paaboel Andersen, Thomas
194 H. P. Andersen, Tobias Klauser, Tom Gundersen, Torstein Husebø, Umut
195 Tezduyar Lindskog, Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam),
196 Vladimir Panteleev, Wieland Hoffmann, Wouter Verhelst, Yu Watanabe,
197 Zbigniew Jędrzejewski-Szmek
203 * The systemd-resolved DNS resolver service has gained a substantial
204 set of new features, most prominently it may now act as a DNSSEC
205 validating stub resolver. DNSSEC mode is currently turned off by
206 default, but is expected to be turned on by default in one of the
207 next releases. For now, we invite everybody to test the DNSSEC logic
208 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
209 service also gained a full set of D-Bus interfaces, including calls
210 to configure DNS and DNSSEC settings per link (for use by external
211 network management software). systemd-resolved and systemd-networkd
212 now distinguish between "search" and "routing" domains. The former
213 are used to qualify single-label names, the latter are used purely
214 for routing lookups within certain domains to specific links.
215 resolved now also synthesizes RRs for all entries from /etc/hosts.
217 * The systemd-resolve tool (which is a client utility for
218 systemd-resolved) has been improved considerably and is now fully
219 supported and documented. Hence it has moved from /usr/lib/systemd to
222 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
225 * The coredump collection logic has been reworked: when a coredump is
226 collected it is now written to disk, compressed and processed
227 (including stacktrace extraction) from a new instantiated service
228 systemd-coredump@.service, instead of directly from the
229 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
230 processing large coredumps can take up a substantial amount of
231 resources and time, and this previously happened entirely outside of
232 systemd's service supervision. With the new logic the core_pattern
233 hook only does minimal metadata collection before passing off control
234 to the new instantiated service, which is configured with a time
235 limit, a nice level and other settings to minimize negative impact on
236 the rest of the system. Also note that the new logic will honour the
237 RLIMIT_CORE setting of the crashed process, which now allows users
238 and processes to turn off coredumping for their processes by setting
241 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
242 and all forked processes by default. Previously, PID 1 would leave
243 the setting at "0" for all processes, as set by the kernel. Note that
244 the resource limit traditionally has no effect on the generated
245 coredumps on the system if the /proc/sys/kernel/core_pattern hook
246 logic is used. Since the limit is now honoured (see above) its
247 default has been changed so that the coredumping logic is enabled by
248 default for all processes, while allowing specific opt-out.
250 * When the stacktrace is extracted from processes of system users, this
251 is now done as "systemd-coredump" user, in order to sandbox this
252 potentially security sensitive parsing operation. (Note that when
253 processing coredumps of normal users this is done under the user ID
254 of process that crashed, as before.) Packagers should take notice
255 that it is now necessary to create the "systemd-coredump" system user
256 and group at package installation time.
258 * The systemd-activate socket activation testing tool gained support
259 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
260 and --seqpacket switches. It also has been extended to support both
261 new-style and inetd-style file descriptor passing. Use the new
262 --inetd switch to request inetd-style file descriptor passing.
264 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
265 variable, which takes a boolean value. If set to false, ANSI color
266 output is disabled in the tools even when run on a terminal that
269 * The VXLAN support in networkd now supports two new settings
270 DestinationPort= and PortRange=.
272 * A new systemd.machine_id= kernel command line switch has been added,
273 that may be used to set the machine ID in /etc/machine-id if it is
274 not initialized yet. This command line option has no effect if the
275 file is already initialized.
277 * systemd-nspawn gained a new --as-pid2 switch that invokes any
278 specified command line as PID 2 rather than PID 1 in the
279 container. In this mode PID 1 is a minimal stub init process that
280 implements the special POSIX and Linux semantics of PID 1 regarding
281 signal and child process management. Note that this stub init process
282 is implemented in nspawn itself and requires no support from the
283 container image. This new logic is useful to support running
284 arbitrary commands in the container, as normal processes are
285 generally not prepared to run as PID 1.
287 * systemd-nspawn gained a new --chdir= switch for setting the current
288 working directory for the process started in the container.
290 * "journalctl /dev/sda" will now output all kernel log messages for
291 specified device from the current boot, in addition to all devices
292 that are parents of it. This should make log output about devices
293 pretty useful, as long as kernel drivers attach enough metadata to
294 the log messages. (The usual SATA drivers do.)
296 * The sd-journal API gained two new calls
297 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
298 that report whether log data from /run or /var has been found.
300 * journalctl gained a new switch "--fields" that prints all journal
301 record field names currently in use in the journal. This is backed
302 by two new sd-journal API calls sd_journal_enumerate_fields() and
303 sd_journal_restart_fields().
305 * Most configurable timeouts in systemd now expect an argument of
306 "infinity" to turn them off, instead of "0" as before. The semantics
307 from now on is that a timeout of "0" means "now", and "infinity"
308 means "never". To maintain backwards compatibility, "0" continues to
309 turn off previously existing timeout settings.
311 * "systemctl reload-or-try-restart" has been renamed to "systemctl
312 try-reload-or-restart" to clarify what it actually does: the "try"
313 logic applies to both reloading and restarting, not just restarting.
314 The old name continues to be accepted for compatibility.
316 * On boot-up, when PID 1 detects that the system clock is behind the
317 release date of the systemd version in use, the clock is now set
318 to the latter. Previously, this was already done in timesyncd, in order
319 to avoid running with clocks set to the various clock epochs such as
320 1902, 1938 or 1970. With this change the logic is now done in PID 1
321 in addition to timesyncd during early boot-up, so that it is enforced
322 before the first process is spawned by systemd. Note that the logic
323 in timesyncd remains, as it is more comprehensive and ensures
324 clock monotonicity by maintaining a persistent timestamp file in
325 /var. Since /var is generally not available in earliest boot or the
326 initrd, this part of the logic remains in timesyncd, and is not done
329 * Support for tweaking details in net_cls.class_id through the
330 NetClass= configuration directive has been removed, as the kernel
331 people have decided to deprecate that controller in cgroup v2.
332 Userspace tools such as nftables are moving over to setting rules
333 that are specific to the full cgroup path of a task, which obsoletes
334 these controllers anyway. The NetClass= directive is kept around for
335 legacy compatibility reasons. For a more in-depth description of the
336 kernel change, please refer to the respective upstream commit:
338 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
340 * A new service setting RuntimeMaxSec= has been added that may be used
341 to specify a maximum runtime for a service. If the timeout is hit, the
342 service is terminated and put into a failure state.
344 * A new service setting AmbientCapabilities= has been added. It allows
345 configuration of additional Linux process capabilities that are
346 passed to the activated processes. This is only available on very
349 * The process resource limit settings in service units may now be used
350 to configure hard and soft limits individually.
352 * The various libsystemd APIs such as sd-bus or sd-event now publicly
353 expose support for gcc's __attribute__((cleanup())) C extension.
354 Specifically, for many object destructor functions alternative
355 versions have been added that have names suffixed with "p" and take a
356 pointer to a pointer to the object to destroy, instead of just a
357 pointer to the object itself. This is useful because these destructor
358 functions may be used directly as parameters to the cleanup
359 construct. Internally, systemd has been a heavy user of this GCC
360 extension for a long time, and with this change similar support is
361 now available to consumers of the library outside of systemd. Note
362 that by using this extension in your sources compatibility with old
363 and strictly ANSI compatible C compilers is lost. However, all gcc or
364 LLVM versions of recent years support this extension.
366 * Timer units gained support for a new setting RandomizedDelaySec= that
367 allows configuring some additional randomized delay to the configured
368 time. This is useful to spread out timer events to avoid load peaks in
369 clusters or larger setups.
371 * Calendar time specifications now support sub-second accuracy.
373 * Socket units now support listening on SCTP and UDP-lite protocol
376 * The sd-event API now comes with a full set of man pages.
378 * Older versions of systemd contained experimental support for
379 compressing journal files and coredumps with the LZ4 compressor that
380 was not compatible with the lz4 binary (due to API limitations of the
381 lz4 library). This support has been removed; only support for files
382 compatible with the lz4 binary remains. This LZ4 logic is now
383 officially supported and no longer considered experimental.
385 * The dkr image import logic has been removed again from importd. dkr's
386 micro-services focus doesn't fit into the machine image focus of
387 importd, and quickly got out of date with the upstream dkr API.
389 * Creation of the /run/lock/lockdev/ directory was dropped from
390 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
391 been available for many years. If you still need this, you need to
392 create your own tmpfiles.d config file with:
394 d /run/lock/lockdev 0775 root lock -
396 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
397 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
398 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
399 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
400 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
401 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
402 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
403 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
404 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
405 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
406 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
407 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
408 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
409 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
410 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
411 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
412 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
413 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
414 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
420 * A number of properties previously only settable in unit
421 files are now also available as properties to set when
422 creating transient units programmatically via the bus, as it
423 is exposed with systemd-run's --property=
424 setting. Specifically, these are: SyslogIdentifier=,
425 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
426 EnvironmentFile=, ReadWriteDirectories=,
427 ReadOnlyDirectories=, InaccessibleDirectories=,
428 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
430 * When creating transient services via the bus API it is now
431 possible to pass in a set of file descriptors to use as
432 STDIN/STDOUT/STDERR for the invoked process.
434 * Slice units may now be created transiently via the bus APIs,
435 similar to the way service and scope units may already be
438 * Wherever systemd expects a calendar timestamp specification
439 (like in journalctl's --since= and --until= switches) UTC
440 timestamps are now supported. Timestamps suffixed with "UTC"
441 are now considered to be in Universal Time Coordinated
442 instead of the local timezone. Also, timestamps may now
443 optionally be specified with sub-second accuracy. Both of
444 these additions also apply to recurring calendar event
445 specification, such as OnCalendar= in timer units.
447 * journalctl gained a new "--sync" switch that asks the
448 journal daemon to write all so far unwritten log messages to
449 disk and sync the files, before returning.
451 * systemd-tmpfiles learned two new line types "q" and "Q" that
452 operate like "v", but also set up a basic btrfs quota
453 hierarchy when used on a btrfs file system with quota
456 * tmpfiles' "v", "q" and "Q" will now create a plain directory
457 instead of a subvolume (even on a btrfs file system) if the
458 root directory is a plain directory, and not a
459 subvolume. This should simplify things with certain chroot()
460 environments which are not aware of the concept of btrfs
463 * systemd-detect-virt gained a new --chroot switch to detect
464 whether execution takes place in a chroot() environment.
466 * CPUAffinity= now takes CPU index ranges in addition to
469 * The various memory-related resource limit settings (such as
470 LimitAS=) now understand the usual K, M, G, ... suffixes to
471 the base of 1024 (IEC). Similar, the time-related resource
472 limit settings understand the usual min, h, day, ...
475 * There's a new system.conf setting DefaultTasksMax= to
476 control the default TasksMax= setting for services and
477 scopes running on the system. (TasksMax= is the primary
478 setting that exposes the "pids" cgroup controller on systemd
479 and was introduced in the previous systemd release.) The
480 setting now defaults to 512, which means services that are
481 not explicitly configured otherwise will only be able to
482 create 512 processes or threads at maximum, from this
483 version on. Note that this means that thread- or
484 process-heavy services might need to be reconfigured to set
485 TasksMax= to a higher value. It is sufficient to set
486 TasksMax= in these specific unit files to a higher value, or
487 even "infinity". Similar, there's now a logind.conf setting
488 UserTasksMax= that defaults to 4096 and limits the total
489 number of processes or tasks each user may own
490 concurrently. nspawn containers also have the TasksMax=
491 value set by default now, to 8192. Note that all of this
492 only has an effect if the "pids" cgroup controller is
493 enabled in the kernel. The general benefit of these changes
494 should be a more robust and safer system, that provides a
495 certain amount of per-service fork() bomb protection.
497 * systemd-nspawn gained the new --network-veth-extra= switch
498 to define additional and arbitrarily-named virtual Ethernet
499 links between the host and the container.
501 * A new service execution setting PassEnvironment= has been
502 added that allows importing select environment variables
503 from PID1's environment block into the environment block of
506 * Timer units gained support for a new RemainAfterElapse=
507 setting which takes a boolean argument. It defaults on on,
508 exposing behaviour unchanged to previous releases. If set to
509 off, timer units are unloaded after they elapsed if they
510 cannot elapse again. This is particularly useful for
511 transient timer units, which shall not stay around longer
512 than until they first elapse.
514 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
515 default now (the kernel default is 16). This is beneficial
516 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
517 allows substantially larger numbers of queued
518 datagrams. This should increase the capability of systemd to
519 parallelize boot-up, as logging and sd_notify() are unlikely
520 to stall execution anymore. If you need to change the value
521 from the new defaults, use the usual sysctl.d/ snippets.
523 * The compression framing format used by the journal or
524 coredump processing has changed to be in line with what the
525 official LZ4 tools generate. LZ4 compression support in
526 systemd was considered unsupported previously, as the format
527 was not compatible with the normal tools. With this release
528 this has changed now, and it is hence safe for downstream
529 distributions to turn it on. While not compressing as well
530 as the XZ, LZ4 is substantially faster, which makes
531 it a good default choice for the compression logic in the
532 journal and in coredump handling.
534 * Any reference to /etc/mtab has been dropped from
535 systemd. The file has been obsolete since a while, but
536 systemd refused to work on systems where it was incorrectly
537 set up (it should be a symlink or non-existent). Please make
538 sure to update to util-linux 2.27.1 or newer in conjunction
539 with this systemd release, which also drops any reference to
540 /etc/mtab. If you maintain a distribution make sure that no
541 software you package still references it, as this is a
542 likely source of bugs. There's also a glibc bug pending,
543 asking for removal of any reference to this obsolete file:
545 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
547 Note that only util-linux versions built with
548 --enable-libmount-force-mountinfo are supported.
550 * Support for the ".snapshot" unit type has been removed. This
551 feature turned out to be little useful and little used, and
552 has now been removed from the core and from systemctl.
554 * The dependency types RequiresOverridable= and
555 RequisiteOverridable= have been removed from systemd. They
556 have been used only very sparingly to our knowledge and
557 other options that provide a similar effect (such as
558 systemctl --mode=ignore-dependencies) are much more useful
559 and commonly used. Moreover, they were only half-way
560 implemented as the option to control behaviour regarding
561 these dependencies was never added to systemctl. By removing
562 these dependency types the execution engine becomes a bit
563 simpler. Unit files that use these dependencies should be
564 changed to use the non-Overridable dependency types
565 instead. In fact, when parsing unit files with these
566 options, that's what systemd will automatically convert them
567 too, but it will also warn, asking users to fix the unit
568 files accordingly. Removal of these dependency types should
569 only affect a negligible number of unit files in the wild.
571 * Behaviour of networkd's IPForward= option changed
572 (again). It will no longer maintain a per-interface setting,
573 but propagate one way from interfaces where this is enabled
574 to the global kernel setting. The global setting will be
575 enabled when requested by a network that is set up, but
576 never be disabled again. This change was made to make sure
577 IPv4 and IPv6 behaviour regarding packet forwarding is
578 similar (as the Linux IPv6 stack does not support
579 per-interface control of this setting) and to minimize
582 * In unit files the behaviour of %u, %U, %h, %s has
583 changed. These specifiers will now unconditionally resolve
584 to the various user database fields of the user that the
585 systemd instance is running as, instead of the user
586 configured in the specific unit via User=. Note that this
587 effectively doesn't change much, as resolving of these
588 specifiers was already turned off in the --system instance
589 of systemd, as we cannot do NSS lookups from PID 1. In the
590 --user instance of systemd these specifiers where correctly
591 resolved, but hardly made any sense, since the user instance
592 lacks privileges to do user switches anyway, and User= is
593 hence useless. Morever, even in the --user instance of
594 systemd behaviour was awkward as it would only take settings
595 from User= assignment placed before the specifier into
596 account. In order to unify and simplify the logic around
597 this the specifiers will now always resolve to the
598 credentials of the user invoking the manager (which in case
599 of PID 1 is the root user).
601 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
602 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
603 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
604 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
605 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
606 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
607 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
608 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
609 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
610 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
611 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
612 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
613 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
614 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
621 * systemd now depends on util-linux v2.27. More specifically,
622 the newly added mount monitor feature in libmount now
623 replaces systemd's former own implementation.
625 * libmount mandates /etc/mtab not to be regular file, and
626 systemd now enforces this condition at early boot.
627 /etc/mtab has been deprecated and warned about for a very
628 long time, so systems running systemd should already have
629 stopped having this file around as anything else than a
630 symlink to /proc/self/mounts.
632 * Support for the "pids" cgroup controller has been added. It
633 allows accounting the number of tasks in a cgroup and
634 enforcing limits on it. This adds two new setting
635 TasksAccounting= and TasksMax= to each unit, as well as a
636 global option DefaultTasksAccounting=.
638 * Support for the "net_cls" cgroup controller has been added.
639 It allows assigning a net class ID to each task in the
640 cgroup, which can then be used in firewall rules and traffic
641 shaping configurations. Note that the kernel netfilter net
642 class code does not currently work reliably for ingress
643 packets on unestablished sockets.
645 This adds a new config directive called NetClass= to CGroup
646 enabled units. Allowed values are positive numbers for fixed
647 assignments and "auto" for picking a free value
650 * 'systemctl is-system-running' now returns 'offline' if the
651 system is not booted with systemd. This command can now be
652 used as a substitute for 'systemd-notify --booted'.
654 * Watchdog timeouts have been increased to 3 minutes for all
655 in-tree service files. Apparently, disk IO issues are more
656 frequent than we hoped, and user reported >1 minute waiting
659 * 'machine-id-commit' functionality has been merged into
660 'machine-id-setup --commit'. The separate binary has been
663 * The WorkingDirectory= directive in unit files may now be set
664 to the special value '~'. In this case, the working
665 directory is set to the home directory of the user
668 * "machinectl shell" will now open the shell in the home
669 directory of the selected user by default.
671 * The CrashChVT= configuration file setting is renamed to
672 CrashChangeVT=, following our usual logic of not
673 abbreviating unnecessarily. The old directive is still
674 supported for compat reasons. Also, this directive now takes
675 an integer value between 1 and 63, or a boolean value. The
676 formerly supported '-1' value for disabling stays around for
679 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
680 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
681 RootDirectory= properties can now be set for transient
684 * The systemd-analyze tool gained a new "set-log-target" verb
685 to change the logging target the system manager logs to
686 dynamically during runtime. This is similar to how
687 "systemd-analyze set-log-level" already changes the log
690 * In nspawn /sys is now mounted as tmpfs, with only a selected
691 set of subdirectories mounted in from the real sysfs. This
692 enhances security slightly, and is useful for ensuring user
693 namespaces work correctly.
695 * Support for USB FunctionFS activation has been added. This
696 allows implementation of USB gadget services that are
697 activated as soon as they are requested, so that they don't
698 have to run continously, similar to classic socket
701 * The "systemctl exit" command now optionally takes an
702 additional parameter that sets the exit code to return from
703 the systemd manager when exiting. This is only relevant when
704 running the systemd user instance, or when running the
705 system instance in a container.
707 * sd-bus gained the new API calls sd_bus_path_encode_many()
708 and sd_bus_path_decode_many() that allow easy encoding and
709 decoding of multiple identifier strings inside a D-Bus
710 object path. Another new call sd_bus_default_flush_close()
711 has been added to flush and close per-thread default
714 * systemd-cgtop gained support for a -M/--machine= switch to
715 show the control groups within a certain container only.
717 * "systemctl kill" gained support for an optional --fail
718 switch. If specified the requested operation will fail of no
719 processes have been killed, because the unit had no
720 processes attached, or similar.
722 * A new systemd.crash_reboot=1 kernel command line option has
723 been added that triggers a reboot after crashing. This can
724 also be set through CrashReboot= in systemd.conf.
726 * The RuntimeDirectory= setting now understands unit
727 specifiers like %i or %f.
729 * A new (still internal) libary API sd-ipv4acd has been added,
730 that implements address conflict detection for IPv4. It's
731 based on code from sd-ipv4ll, and will be useful for
732 detecting DHCP address conflicts.
734 * File descriptors passed during socket activation may now be
735 named. A new API sd_listen_fds_with_names() is added to
736 access the names. The default names may be overridden,
737 either in the .socket file using the FileDescriptorName=
738 parameter, or by passing FDNAME= when storing the file
739 descriptors using sd_notify().
741 * systemd-networkd gained support for:
743 - Setting the IPv6 Router Advertisment settings via
744 IPv6AcceptRouterAdvertisements= in .network files.
746 - Configuring the HelloTimeSec=, MaxAgeSec= and
747 ForwardDelaySec= bridge parameters in .netdev files.
749 - Configuring PreferredSource= for static routes in
752 * The "ask-password" framework used to query for LUKS harddisk
753 passwords or SSL passwords during boot gained support for
754 caching passwords in the kernel keyring, if it is
755 available. This makes sure that the user only has to type in
756 a passphrase once if there are multiple objects to unlock
757 with the same one. Previously, such password caching was
758 available only when Plymouth was used; this moves the
759 caching logic into the systemd codebase itself. The
760 "systemd-ask-password" utility gained a new --keyname=
761 switch to control which kernel keyring key to use for
762 caching a password in. This functionality is also useful for
763 enabling display managers such as gdm to automatically
764 unlock the user's GNOME keyring if its passphrase, the
765 user's password and the harddisk password are the same, if
766 gdm-autologin is used.
768 * When downloading tar or raw images using "machinectl
769 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
770 file is now also downloaded, if it is available and stored
771 next to the image file.
773 * Units of type ".socket" gained a new boolean setting
774 Writable= which is only useful in conjunction with
775 ListenSpecial=. If true, enables opening the specified
776 special file in O_RDWR mode rather than O_RDONLY mode.
778 * systemd-rfkill has been reworked to become a singleton
779 service that is activated through /dev/rfkill on each rfkill
780 state change and saves the settings to disk. This way,
781 systemd-rfkill is now compatible with devices that exist
782 only intermittendly, and even restores state if the previous
783 system shutdown was abrupt rather than clean.
785 * The journal daemon gained support for vacuuming old journal
786 files controlled by the number of files that shall remain,
787 in addition to the already existing control by size and by
788 date. This is useful as journal interleaving performance
789 degrades with too many seperate journal files, and allows
790 putting an effective limit on them. The new setting defaults
791 to 100, but this may be changed by setting SystemMaxFiles=
792 and RuntimeMaxFiles= in journald.conf. Also, the
793 "journalctl" tool gained the new --vacuum-files= switch to
794 manually vacuum journal files to leave only the specified
795 number of files in place.
797 * udev will now create /dev/disk/by-path links for ATA devices
798 on kernels where that is supported.
800 * Galician, Serbian, Turkish and Korean translations were added.
802 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
803 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
804 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
805 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
806 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
807 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
808 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
809 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
810 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
811 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
812 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
813 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
814 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
815 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
816 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
817 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
818 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
819 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
825 * The DHCP implementation of systemd-networkd gained a set of
828 - The DHCP server now supports emitting DNS and NTP
829 information. It may be enabled and configured via
830 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
831 and NTP information is enabled, but no servers are
832 configured, the corresponding uplink information (if there
833 is any) is propagated.
835 - Server and client now support transmission and reception
836 of timezone information. It can be configured via the
837 newly introduced network options UseTimezone=,
838 EmitTimezone=, and Timezone=. Transmission of timezone
839 information is enabled between host and containers by
840 default now: the container will change its local timezone
841 to what the host has set.
843 - Lease timeouts can now be configured via
844 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
846 - The DHCP server improved on the stability of
847 leases. Clients are more likely to get the same lease
848 information back, even if the server loses state.
850 - The DHCP server supports two new configuration options to
851 control the lease address pool metrics, PoolOffset= and
854 * The encapsulation limit of tunnels in systemd-networkd may
855 now be configured via 'EncapsulationLimit='. It allows
856 modifying the maximum additional levels of encapsulation
857 that are permitted to be prepended to a packet.
859 * systemd now supports the concept of user buses replacing
860 session buses, if used with dbus-1.10 (and enabled via dbus
861 --enable-user-session). It previously only supported this on
862 kdbus-enabled systems, and this release expands this to
863 'dbus-daemon' systems.
865 * systemd-networkd now supports predictable interface names
868 * systemd now optionally supports the new Linux kernel
869 "unified" control group hierarchy. If enabled via the kernel
870 command-line option 'systemd.unified_cgroup_hierarchy=1',
871 systemd will try to mount the unified cgroup hierarchy
872 directly on /sys/fs/cgroup. If not enabled, or not
873 available, systemd will fall back to the legacy cgroup
874 hierarchy setup, as before. Host system and containers can
875 mix and match legacy and unified hierarchies as they
876 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
877 environment variable to individually select the hierarchy to
878 use for executed containers. By default, nspawn will use the
879 unified hierarchy for the containers if the host uses the
880 unified hierarchy, and the legacy hierarchy otherwise.
881 Please note that at this point the unified hierarchy is an
882 experimental kernel feature and is likely to change in one
883 of the next kernel releases. Therefore, it should not be
884 enabled by default in downstream distributions yet. The
885 minimum required kernel version for the unified hierarchy to
886 work is 4.2. Note that when the unified hierarchy is used
887 for the first time delegated access to controllers is
888 safe. Because of this systemd-nspawn containers will get
889 access to controllers now, as will systemd user
890 sessions. This means containers and user sessions may now
891 manage their own resources, partitioning up what the system
894 * A new special scope unit "init.scope" has been introduced
895 that encapsulates PID 1 of the system. It may be used to
896 determine resource usage and enforce resource limits on PID
897 1 itself. PID 1 hence moved out of the root of the control
900 * The cgtop tool gained support for filtering out kernel
901 threads when counting tasks in a control group. Also, the
902 count of processes is now recursively summed up by
903 default. Two options -k and --recursive= have been added to
904 revert to old behaviour. The tool has also been updated to
905 work correctly in containers now.
907 * systemd-nspawn's --bind= and --bind-ro= options have been
908 extended to allow creation of non-recursive bind mounts.
910 * libsystemd gained two new calls sd_pid_get_cgroup() and
911 sd_peer_get_cgroup() which return the control group path of
912 a process or peer of a connected AF_UNIX socket. This
913 function call is particularly useful when implementing
914 delegated subtrees support in the control group hierarchy.
916 * The "sd-event" event loop API of libsystemd now supports
917 correct dequeuing of real-time signals, without losing
920 * When systemd requests a PolicyKit decision when managing
921 units it will now add additional fields to the request,
922 including unit name and desired operation. This enables more
923 powerful PolicyKit policies, that make decisions depending
926 * nspawn learnt support for .nspawn settings files, that may
927 accompany the image files or directories of containers, and
928 may contain additional settings for the container. This is
929 an alternative to configuring container parameters via the
932 Contributions from: Cristian Rodríguez, Daniel Mack, David
933 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
934 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
935 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
936 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
937 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
938 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
939 Andersen, Tom Gundersen, Torstein Husebø
945 * machinectl gained a new verb 'shell' which opens a fresh
946 shell on the target container or the host. It is similar to
947 the existing 'login' command of machinectl, but spawns the
948 shell directly without prompting for username or
949 password. The pseudo machine '.host' now refers to the local
950 host and is used by default. Hence, 'machinectl shell' can
951 be used as replacement for 'su -' which spawns a session as
952 a fresh systemd unit in a way that is fully isolated from
953 the originating session.
955 * systemd-networkd learned to cope with private-zone DHCP
956 options and allows other programs to query the values.
958 * SELinux access control when enabling/disabling units is no
959 longer enforced with this release. The previous
960 implementation was incorrect, and a new corrected
961 implementation is not yet available. As unit file operations
962 are still protected via PolicyKit and D-Bus policy this is
963 not a security problem. Yet, distributions which care about
964 optimal SELinux support should probably not stabilize on
967 * sd-bus gained support for matches of type "arg0has=", that
968 test for membership of strings in string arrays sent in bus
971 * systemd-resolved now dumps the contents of its DNS and LLMNR
972 caches to the logs on reception of the SIGUSR1 signal. This
973 is useful to debug DNS behaviour.
975 * The coredumpctl tool gained a new --directory= option to
976 operate on journal files in a specific directory.
978 * "systemctl reboot" and related commands gained a new
979 "--message=" option which may be used to set a free-text
980 wall message when shutting down or rebooting the
981 system. This message is also logged, which is useful for
982 figuring out the reason for a reboot or shutdown a
985 * The "systemd-resolve-host" tool's -i switch now takes
986 network interface numbers as alternative to interface names.
988 * A new unit file setting for services has been introduced:
989 UtmpMode= allows configuration of how precisely systemd
990 handles utmp and wtmp entries for the service if this is
991 enabled. This allows writing services that appear similar to
992 user sessions in the output of the "w", "who", "last" and
995 * systemd-resolved will now locally synthesize DNS resource
996 records for the "localhost" and "gateway" domains as well as
997 the local hostname. This should ensure that clients querying
998 RRs via resolved will get similar results as those going via
999 NSS, if nss-myhostname is enabled.
1001 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1002 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1003 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1004 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1005 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1006 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1007 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1008 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1009 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1010 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1011 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1012 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1014 — Berlin, 2015-08-27
1018 * The systemd-efi-boot-generator functionality was merged into
1019 systemd-gpt-auto-generator.
1021 * systemd-networkd now supports Group Policy for vxlan
1022 devices. It can be enabled via the new boolean configuration
1023 option called 'GroupPolicyExtension='.
1025 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1026 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1027 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1029 — Berlin, 2015-07-31
1033 * The python-systemd code has been removed from the systemd repository.
1034 A new repository has been created which accommodates the code from
1035 now on, and we kindly ask distributions to create a separate package
1036 for this: https://github.com/systemd/python-systemd
1038 * The systemd daemon will now reload its main configuration
1039 (/etc/systemd/system.conf) on daemon-reload.
1041 * sd-dhcp now exposes vendor specific extensions via
1042 sd_dhcp_lease_get_vendor_specific().
1044 * systemd-networkd gained a number of new configuration options.
1046 - A new boolean configuration option for TAP devices called
1047 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1048 device, thus allowing to send and receive GSO packets.
1050 - A new tunnel configuration option called 'CopyDSCP='.
1051 If enabled, the DSCP field of ip6 tunnels is copied into the
1052 decapsulated packet.
1054 - A set of boolean bridge configuration options were added.
1055 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1056 and 'UnicastFlood=' are now parsed by networkd and applied to the
1057 respective bridge link device via the respective IFLA_BRPORT_*
1060 - A new string configuration option to override the hostname sent
1061 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1062 is true, networkd will use the configured hostname instead of the
1063 system hostname when sending DHCP requests.
1065 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1066 networkd will configure the IPv6 flow-label of the tunnel device
1067 according to RFC2460.
1069 - The 'macvtap' virtual network devices are now supported, similar to
1070 the already supported 'macvlan' devices.
1072 * systemd-resolved now implements RFC5452 to improve resilience against
1073 cache poisoning. Additionally, source port randomization is enabled
1074 by default to further protect against DNS spoofing attacks.
1076 * nss-mymachines now supports translating UIDs and GIDs of running
1077 containers with user-namespaces enabled. If a container 'foo'
1078 translates a host uid 'UID' to the container uid 'TUID', then
1079 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1080 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1081 mapped as 'vg-foo-TGID'.
1083 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1084 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1085 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1086 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1087 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1088 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1089 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1090 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1091 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1092 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1094 — Berlin, 2015-07-29
1098 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1099 There are no known issues with current sysfs, and udev does not need
1100 or should be used to work around such bugs.
1102 * udev does no longer enable USB HID power management. Several reports
1103 indicate, that some devices cannot handle that setting.
1105 * The udev accelerometer helper was removed. The functionality
1106 is now fully included in iio-sensor-proxy. But this means,
1107 older iio-sensor-proxy versions will no longer provide
1108 accelerometer/orientation data with this systemd version.
1109 Please upgrade iio-sensor-proxy to version 1.0.
1111 * networkd gained a new configuration option IPv6PrivacyExtensions=
1112 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1113 for Stateless Address") on selected networks.
1115 * For the sake of fewer build-time dependencies and less code in the
1116 main repository, the python bindings are about to be removed in the
1117 next release. A new repository has been created which accommodates
1118 the code from now on, and we kindly ask distributions to create a
1119 separate package for this. The removal will take place in v223.
1121 https://github.com/systemd/python-systemd
1123 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1124 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1125 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1126 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1127 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1128 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1129 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1130 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1131 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1132 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1134 — Berlin, 2015-07-07
1138 * The sd-bus.h and sd-event.h APIs have now been declared
1139 stable and have been added to the official interface of
1140 libsystemd.so. sd-bus implements an alternative D-Bus client
1141 library, that is relatively easy to use, very efficient and
1142 supports both classic D-Bus as well as kdbus as transport
1143 backend. sd-event is a generic event loop abstraction that
1144 is built around Linux epoll, but adds features such as event
1145 prioritization or efficient timer handling. Both APIs are good
1146 choices for C programs looking for a bus and/or event loop
1147 implementation that is minimal and does not have to be
1148 portable to other kernels.
1150 * kdbus support is no longer compile-time optional. It is now
1151 always built-in. However, it can still be disabled at
1152 runtime using the kdbus=0 kernel command line setting, and
1153 that setting may be changed to default to off, by specifying
1154 --disable-kdbus at build-time. Note though that the kernel
1155 command line setting has no effect if the kdbus.ko kernel
1156 module is not installed, in which case kdbus is (obviously)
1157 also disabled. We encourage all downstream distributions to
1158 begin testing kdbus by adding it to the kernel images in the
1159 development distributions, and leaving kdbus support in
1162 * The minimal required util-linux version has been bumped to
1165 * Support for chkconfig (--enable-chkconfig) was removed in
1166 favor of calling an abstraction tool
1167 /lib/systemd/systemd-sysv-install. This needs to be
1168 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1169 in README for details.
1171 * If there's a systemd unit and a SysV init script for the
1172 same service name, and the user executes "systemctl enable"
1173 for it (or a related call), then this will now enable both
1174 (or execute the related operation on both), not just the
1177 * The libudev API documentation has been converted from gtkdoc
1180 * gudev has been removed from the systemd tree, it is now an
1183 * The systemd-cgtop tool learnt a new --raw switch to generate
1184 "raw" (machine parsable) output.
1186 * networkd's IPForwarding= .network file setting learnt the
1187 new setting "kernel", which ensures that networkd does not
1188 change the IP forwarding sysctl from the default kernel
1191 * The systemd-logind bus API now exposes a new boolean
1192 property "Docked" that reports whether logind considers the
1193 system "docked", i.e. connected to a docking station or not.
1195 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1196 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1197 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1198 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1199 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1200 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1201 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1202 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1203 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1204 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1205 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1206 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1207 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1208 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1209 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1210 Fink, Zbigniew Jędrzejewski-Szmek
1212 — Berlin, 2015-06-19
1216 * The gudev library has been extracted into a separate repository
1217 available at: https://git.gnome.org/browse/libgudev/
1218 It is now managed as part of the Gnome project. Distributions
1219 are recommended to pass --disable-gudev to systemd and use
1220 gudev from the Gnome project instead. gudev is still included
1221 in systemd, for now. It will be removed soon, though. Please
1222 also see the announcement-thread on systemd-devel:
1223 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1225 * systemd now exposes a CPUUsageNSec= property for each
1226 service unit on the bus, that contains the overall consumed
1227 CPU time of a service (the sum of what each process of the
1228 service consumed). This value is only available if
1229 CPUAccounting= is turned on for a service, and is then shown
1230 in the "systemctl status" output.
1232 * Support for configuring alternative mappings of the old SysV
1233 runlevels to systemd targets has been removed. They are now
1234 hardcoded in a way that runlevels 2, 3, 4 all map to
1235 multi-user.target and 5 to graphical.target (which
1236 previously was already the default behaviour).
1238 * The auto-mounter logic gained support for mount point
1239 expiry, using a new TimeoutIdleSec= setting in .automount
1240 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1242 * The EFI System Partition (ESP) as mounted to /boot by
1243 systemd-efi-boot-generator will now be unmounted
1244 automatically after 2 minutes of not being used. This should
1245 minimize the risk of ESP corruptions.
1247 * New /etc/fstab options x-systemd.requires= and
1248 x-systemd.requires-mounts-for= are now supported to express
1249 additional dependencies for mounts. This is useful for
1250 journalling file systems that support external journal
1251 devices or overlay file systems that require underlying file
1252 systems to be mounted.
1254 * systemd does not support direct live-upgrades (via systemctl
1255 daemon-reexec) from versions older than v44 anymore. As no
1256 distribution we are aware of shipped such old versions in a
1257 stable release this should not be problematic.
1259 * When systemd forks off a new per-connection service instance
1260 it will now set the $REMOTE_ADDR environment variable to the
1261 remote IP address, and $REMOTE_PORT environment variable to
1262 the remote IP port. This behaviour is similar to the
1263 corresponding environment variables defined by CGI.
1265 * systemd-networkd gained support for uplink failure
1266 detection. The BindCarrier= option allows binding interface
1267 configuration dynamically to the link sense of other
1268 interfaces. This is useful to achieve behaviour like in
1271 * systemd-networkd gained support for configuring the DHCP
1272 client identifier to use when requesting leases.
1274 * systemd-networkd now has a per-network UseNTP= option to
1275 configure whether NTP server information acquired via DHCP
1276 is passed on to services like systemd-timesyncd.
1278 * systemd-networkd gained support for vti6 tunnels.
1280 * Note that systemd-networkd manages the sysctl variable
1281 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1282 it is configured for since v219. The variable controls IP
1283 forwarding, and is a per-interface alternative to the global
1284 /proc/sys/net/ipv[46]/ip_forward. This setting is
1285 configurable in the IPForward= option, which defaults to
1286 "no". This means if networkd is used for an interface it is
1287 no longer sufficient to set the global sysctl option to turn
1288 on IP forwarding! Instead, the .network file option
1289 IPForward= needs to be turned on! Note that the
1290 implementation of this behaviour was broken in v219 and has
1293 * Many bonding and vxlan options are now configurable in
1296 * systemd-nspawn gained a new --property= setting to set unit
1297 properties for the container scope. This is useful for
1298 setting resource parameters (e.g "CPUShares=500") on
1299 containers started from the command line.
1301 * systemd-nspawn gained a new --private-users= switch to make
1302 use of user namespacing available on recent Linux kernels.
1304 * systemd-nspawn may now be called as part of a shell pipeline
1305 in which case the pipes used for stdin and stdout are passed
1306 directly to the process invoked in the container, without
1307 indirection via a pseudo tty.
1309 * systemd-nspawn gained a new switch to control the UNIX
1310 signal to use when killing the init process of the container
1313 * systemd-nspawn gained a new --overlay= switch for mounting
1314 overlay file systems into the container using the new kernel
1317 * When a container image is imported via systemd-importd and
1318 the host file system is not btrfs, a loopback block device
1319 file is created in /var/lib/machines.raw with a btrfs file
1320 system inside. It is then mounted to /var/lib/machines to
1321 enable btrfs features for container management. The loopback
1322 file and btrfs file system is grown as needed when container
1323 images are imported via systemd-importd.
1325 * systemd-machined/systemd-importd gained support for btrfs
1326 quota, to enforce container disk space limits on disk. This
1327 is exposed in "machinectl set-limit".
1329 * systemd-importd now can import containers from local .tar,
1330 .raw and .qcow2 images, and export them to .tar and .raw. It
1331 can also import dkr v2 images now from the network (on top
1334 * systemd-importd gained support for verifying downloaded
1335 images with gpg2 (previously only gpg1 was supported).
1337 * systemd-machined, systemd-logind, systemd: most bus calls
1338 are now accessible to unprivileged processes via
1339 PolicyKit. Also, systemd-logind will now allow users to kill
1340 their own sessions without further privileges or
1343 * systemd-shutdownd has been removed. This service was
1344 previously responsible for implementing scheduled shutdowns
1345 as exposed in /usr/bin/shutdown's time parameter. This
1346 functionality has now been moved into systemd-logind and is
1347 accessible via a bus interface.
1349 * "systemctl reboot" gained a new switch --firmware-setup that
1350 can be used to reboot into the EFI firmware setup, if that
1351 is available. systemd-logind now exposes an API on the bus
1352 to trigger such reboots, in case graphical desktop UIs want
1353 to cover this functionality.
1355 * "systemctl enable", "systemctl disable" and "systemctl mask"
1356 now support a new "--now" switch. If specified the units
1357 that are enabled will also be started, and the ones
1358 disabled/masked also stopped.
1360 * The Gummiboot EFI boot loader tool has been merged into
1361 systemd, and renamed to "systemd-boot". The bootctl tool has been
1362 updated to support systemd-boot.
1364 * An EFI kernel stub has been added that may be used to create
1365 kernel EFI binaries that contain not only the actual kernel,
1366 but also an initrd, boot splash, command line and OS release
1367 information. This combined binary can then be signed as a
1368 single image, so that the firmware can verify it all in one
1369 step. systemd-boot has special support for EFI binaries created
1370 like this and can extract OS release information from them
1371 and show them in the boot menu. This functionality is useful
1372 to implement cryptographically verified boot schemes.
1374 * Optional support has been added to systemd-fsck to pass
1375 fsck's progress report to an AF_UNIX socket in the file
1378 * udev will no longer create device symlinks for all block
1379 devices by default. A blacklist for excluding special block
1380 devices from this logic has been turned into a whitelist
1381 that requires picking block devices explicitly that require
1384 * A new (currently still internal) API sd-device.h has been
1385 added to libsystemd. This modernized API is supposed to
1386 replace libudev eventually. In fact, already much of libudev
1387 is now just a wrapper around sd-device.h.
1389 * A new hwdb database for storing metadata about pointing
1390 stick devices has been added.
1392 * systemd-tmpfiles gained support for setting file attributes
1393 similar to the "chattr" tool with new 'h' and 'H' lines.
1395 * systemd-journald will no longer unconditionally set the
1396 btrfs NOCOW flag on new journal files. This is instead done
1397 with tmpfiles snippet using the new 'h' line type. This
1398 allows easy disabling of this logic, by masking the
1399 journal-nocow.conf tmpfiles file.
1401 * systemd-journald will now translate audit message types to
1402 human readable identifiers when writing them to the
1403 journal. This should improve readability of audit messages.
1405 * The LUKS logic gained support for the offset= and skip=
1406 options in /etc/crypttab, as previously implemented by
1409 * /usr/lib/os-release gained a new optional field VARIANT= for
1410 distributions that support multiple variants (such as a
1411 desktop edition, a server edition, ...)
1413 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1414 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1415 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1416 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1417 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1418 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1419 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1420 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1421 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1422 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1423 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1424 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1425 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1426 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1427 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1428 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1429 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1430 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1431 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1432 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1433 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1434 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1435 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1436 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1437 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1438 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1439 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1441 — Berlin, 2015-05-22
1445 * Introduce a new API "sd-hwdb.h" for querying the hardware
1446 metadata database. With this minimal interface one can query
1447 and enumerate the udev hwdb, decoupled from the old libudev
1448 library. libudev's interface for this is now only a wrapper
1449 around sd-hwdb. A new tool systemd-hwdb has been added to
1450 interface with and update the database.
1452 * When any of systemd's tools copies files (for example due to
1453 tmpfiles' C lines) a btrfs reflink will attempted first,
1454 before bytewise copying is done.
1456 * systemd-nspawn gained a new --ephemeral switch. When
1457 specified a btrfs snapshot is taken of the container's root
1458 directory, and immediately removed when the container
1459 terminates again. Thus, a container can be started whose
1460 changes never alter the container's root directory, and are
1461 lost on container termination. This switch can also be used
1462 for starting a container off the root file system of the
1463 host without affecting the host OS. This switch is only
1464 available on btrfs file systems.
1466 * systemd-nspawn gained a new --template= switch. It takes the
1467 path to a container tree to use as template for the tree
1468 specified via --directory=, should that directory be
1469 missing. This allows instantiating containers dynamically,
1470 on first run. This switch is only available on btrfs file
1473 * When a .mount unit refers to a mount point on which multiple
1474 mounts are stacked, and the .mount unit is stopped all of
1475 the stacked mount points will now be unmounted until no
1476 mount point remains.
1478 * systemd now has an explicit notion of supported and
1479 unsupported unit types. Jobs enqueued for unsupported unit
1480 types will now fail with an "unsupported" error code. More
1481 specifically .swap, .automount and .device units are not
1482 supported in containers, .busname units are not supported on
1483 non-kdbus systems. .swap and .automount are also not
1484 supported if their respective kernel compile time options
1487 * machinectl gained support for two new "copy-from" and
1488 "copy-to" commands for copying files from a running
1489 container to the host or vice versa.
1491 * machinectl gained support for a new "bind" command to bind
1492 mount host directories into local containers. This is
1493 currently only supported for nspawn containers.
1495 * networkd gained support for configuring bridge forwarding
1496 database entries (fdb) from .network files.
1498 * A new tiny daemon "systemd-importd" has been added that can
1499 download container images in tar, raw, qcow2 or dkr formats,
1500 and make them available locally in /var/lib/machines, so
1501 that they can run as nspawn containers. The daemon can GPG
1502 verify the downloads (not supported for dkr, since it has no
1503 provisions for verifying downloads). It will transparently
1504 decompress bz2, xz, gzip compressed downloads if necessary,
1505 and restore sparse files on disk. The daemon uses privilege
1506 separation to ensure the actual download logic runs with
1507 fewer privileges than the daemon itself. machinectl has
1508 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1509 make the functionality of importd available to the
1510 user. With this in place the Fedora and Ubuntu "Cloud"
1511 images can be downloaded and booted as containers unmodified
1512 (the Fedora images lack the appropriate GPG signature files
1513 currently, so they cannot be verified, but this will change
1514 soon, hopefully). Note that downloading images is currently
1515 only fully supported on btrfs.
1517 * machinectl is now able to list container images found in
1518 /var/lib/machines, along with some metadata about sizes of
1519 disk and similar. If the directory is located on btrfs and
1520 quota is enabled, this includes quota display. A new command
1521 "image-status" has been added that shows additional
1522 information about images.
1524 * machinectl is now able to clone container images
1525 efficiently, if the underlying file system (btrfs) supports
1526 it, with the new "machinectl clone" command. It also
1527 gained commands for renaming and removing images, as well as
1528 marking them read-only or read-write (supported also on
1529 legacy file systems).
1531 * networkd gained support for collecting LLDP network
1532 announcements, from hardware that supports this. This is
1533 shown in networkctl output.
1535 * systemd-run gained support for a new -t (--pty) switch for
1536 invoking a binary on a pty whose input and output is
1537 connected to the invoking terminal. This allows executing
1538 processes as system services while interactively
1539 communicating with them via the terminal. Most interestingly
1540 this is supported across container boundaries. Invoking
1541 "systemd-run -t /bin/bash" is an alternative to running a
1542 full login session, the difference being that the former
1543 will not register a session, nor go through the PAM session
1546 * tmpfiles gained support for a new "v" line type for creating
1547 btrfs subvolumes. If the underlying file system is a legacy
1548 file system, this automatically degrades to creating a
1549 normal directory. Among others /var/lib/machines is now
1550 created like this at boot, should it be missing.
1552 * The directory /var/lib/containers/ has been deprecated and
1553 been replaced by /var/lib/machines. The term "machines" has
1554 been used in the systemd context as generic term for both
1555 VMs and containers, and hence appears more appropriate for
1556 this, as the directory can also contain raw images bootable
1559 * systemd-nspawn when invoked with -M but without --directory=
1560 or --image= is now capable of searching for the container
1561 root directory, subvolume or disk image automatically, in
1562 /var/lib/machines. systemd-nspawn@.service has been updated
1563 to make use of this, thus allowing it to be used for raw
1566 * A new machines.target unit has been introduced that is
1567 supposed to group all containers/VMs invoked as services on
1568 the system. systemd-nspawn@.service has been updated to
1569 integrate with that.
1571 * machinectl gained a new "start" command, for invoking a
1572 container as a service. "machinectl start foo" is mostly
1573 equivalent to "systemctl start systemd-nspawn@foo.service",
1574 but handles escaping in a nicer way.
1576 * systemd-nspawn will now mount most of the cgroupfs tree
1577 read-only into each container, with the exception of the
1578 container's own subtree in the name=systemd hierarchy.
1580 * journald now sets the special FS_NOCOW file flag for its
1581 journal files. This should improve performance on btrfs, by
1582 avoiding heavy fragmentation when journald's write-pattern
1583 is used on COW file systems. It degrades btrfs' data
1584 integrity guarantees for the files to the same levels as for
1585 ext3/ext4 however. This should be OK though as journald does
1586 its own data integrity checks and all its objects are
1587 checksummed on disk. Also, journald should handle btrfs disk
1588 full events a lot more gracefully now, by processing SIGBUS
1589 errors, and not relying on fallocate() anymore.
1591 * When journald detects that journal files it is writing to
1592 have been deleted it will immediately start new journal
1595 * systemd now provides a way to store file descriptors
1596 per-service in PID 1.This is useful for daemons to ensure
1597 that fds they require are not lost during a daemon
1598 restart. The fds are passed to the daemon on the next
1599 invocation in the same way socket activation fds are
1600 passed. This is now used by journald to ensure that the
1601 various sockets connected to all the system's stdout/stderr
1602 are not lost when journald is restarted. File descriptors
1603 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1604 an extension to sd_notify(). Note that a limit is enforced
1605 on the number of fds a service can store in PID 1, and it
1606 defaults to 0, so that no fds may be stored, unless this is
1607 explicitly turned on.
1609 * The default TERM variable to use for units connected to a
1610 terminal, when no other value is explicitly is set is now
1611 vt220 rather than vt102. This should be fairly safe still,
1612 but allows PgUp/PgDn work.
1614 * The /etc/crypttab option header= as known from Debian is now
1617 * "loginctl user-status" and "loginctl session-status" will
1618 now show the last 10 lines of log messages of the
1619 user/session following the status output. Similar,
1620 "machinectl status" will show the last 10 log lines
1621 associated with a virtual machine or container
1622 service. (Note that this is usually not the log messages
1623 done in the VM/container itself, but simply what the
1624 container manager logs. For nspawn this includes all console
1627 * "loginctl session-status" without further argument will now
1628 show the status of the session of the caller. Similar,
1629 "lock-session", "unlock-session", "activate",
1630 "enable-linger", "disable-linger" may now be called without
1631 session/user parameter in which case they apply to the
1632 caller's session/user.
1634 * An X11 session scriptlet is now shipped that uploads
1635 $DISPLAY and $XAUTHORITY into the environment of the systemd
1636 --user daemon if a session begins. This should improve
1637 compatibility with X11 enabled applications run as systemd
1640 * Generators are now subject to masking via /etc and /run, the
1641 same way as unit files.
1643 * networkd .network files gained support for configuring
1644 per-link IPv4/IPv6 packet forwarding as well as IPv4
1645 masquerading. This is by default turned on for veth links to
1646 containers, as registered by systemd-nspawn. This means that
1647 nspawn containers run with --network-veth will now get
1648 automatic routed access to the host's networks without any
1649 further configuration or setup, as long as networkd runs on
1652 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1653 or UDP posts of a container on the host. With this in place
1654 it is possible to run containers with private veth links
1655 (--network-veth), and have their functionality exposed on
1656 the host as if their services were running directly on the
1659 * systemd-nspawn's --network-veth switch now gained a short
1660 version "-n", since with the changes above it is now truly
1661 useful out-of-the-box. The systemd-nspawn@.service has been
1662 updated to make use of it too by default.
1664 * systemd-nspawn will now maintain a per-image R/W lock, to
1665 ensure that the same image is not started more than once
1666 writable. (It's OK to run an image multiple times
1667 simultaneously in read-only mode.)
1669 * systemd-nspawn's --image= option is now capable of
1670 dissecting and booting MBR and GPT disk images that contain
1671 only a single active Linux partition. Previously it
1672 supported only GPT disk images with proper GPT type
1673 IDs. This allows running cloud images from major
1674 distributions directly with systemd-nspawn, without
1677 * In addition to collecting mouse dpi data in the udev
1678 hardware database, there's now support for collecting angle
1679 information for mouse scroll wheels. The database is
1680 supposed to guarantee similar scrolling behavior on mice
1681 that it knows about. There's also support for collecting
1682 information about Touchpad types.
1684 * udev's input_id built-in will now also collect touch screen
1685 dimension data and attach it to probed devices.
1687 * /etc/os-release gained support for a Distribution Privacy
1690 * networkd gained support for creating "ipvlan", "gretap",
1691 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1693 * systemd-tmpfiles gained support for "a" lines for setting
1696 * systemd-nspawn will now mount /tmp in the container to
1697 tmpfs, automatically.
1699 * systemd now exposes the memory.usage_in_bytes cgroup
1700 attribute and shows it for each service in the "systemctl
1701 status" output, if available.
1703 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1704 immediate reboot is triggered. This useful if shutdown is
1705 hung and is unable to complete, to expedite the
1706 operation. Note that this kind of reboot will still unmount
1707 all file systems, and hence should not result in fsck being
1710 * A .device unit for an optical block device will now be
1711 considered active only when a medium is in the drive. Also,
1712 mount units are now bound to their backing devices thus
1713 triggering automatic unmounting when devices become
1714 unavailable. With this in place systemd will now
1715 automatically unmount left-over mounts when a CD-ROM is
1716 ejected or an USB stick is yanked from the system.
1718 * networkd-wait-online now has support for waiting for
1719 specific interfaces only (with globbing), and for giving up
1720 after a configurable timeout.
1722 * networkd now exits when idle. It will be automatically
1723 restarted as soon as interfaces show up, are removed or
1724 change state. networkd will stay around as long as there is
1725 at least one DHCP state machine or similar around, that keep
1728 * networkd may now configure IPv6 link-local addressing in
1729 addition to IPv4 link-local addressing.
1731 * The IPv6 "token" for use in SLAAC may now be configured for
1732 each .network interface in networkd.
1734 * Routes configured with networkd may now be assigned a scope
1737 * networkd's [Match] sections now support globbing and lists
1738 of multiple space-separated matches per item.
1740 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1741 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1742 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1743 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1744 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1745 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1746 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1747 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1748 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1749 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1750 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1751 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1752 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1753 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1754 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1755 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1756 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1757 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1758 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1759 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1760 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1761 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1762 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1763 Hoffmann, Zbigniew Jędrzejewski-Szmek
1765 — Berlin, 2015-02-16
1769 * When querying unit file enablement status (for example via
1770 "systemctl is-enabled"), a new state "indirect" is now known
1771 which indicates that a unit might not be enabled itself, but
1772 another unit listed in its Also= setting might be.
1774 * Similar to the various existing ConditionXYZ= settings for
1775 units, there are now matching AssertXYZ= settings. While
1776 failing conditions cause a unit to be skipped, but its job
1777 to succeed, failing assertions declared like this will cause
1778 a unit start operation and its job to fail.
1780 * hostnamed now knows a new chassis type "embedded".
1782 * systemctl gained a new "edit" command. When used on a unit
1783 file, this allows extending unit files with .d/ drop-in
1784 configuration snippets or editing the full file (after
1785 copying it from /usr/lib to /etc). This will invoke the
1786 user's editor (as configured with $EDITOR), and reload the
1787 modified configuration after editing.
1789 * "systemctl status" now shows the suggested enablement state
1790 for a unit, as declared in the (usually vendor-supplied)
1791 system preset files.
1793 * nss-myhostname will now resolve the single-label host name
1794 "gateway" to the locally configured default IP routing
1795 gateways, ordered by their metrics. This assigns a stable
1796 name to the used gateways, regardless which ones are
1797 currently configured. Note that the name will only be
1798 resolved after all other name sources (if nss-myhostname is
1799 configured properly) and should hence not negatively impact
1800 systems that use the single-label host name "gateway" in
1803 * systemd-inhibit now allows filtering by mode when listing
1806 * Scope and service units gained a new "Delegate" boolean
1807 property, which, when set, allows processes running inside the
1808 unit to further partition resources. This is primarily
1809 useful for systemd user instances as well as container
1812 * journald will now pick up audit messages directly from
1813 the kernel, and log them like any other log message. The
1814 audit fields are split up and fully indexed. This means that
1815 journalctl in many ways is now a (nicer!) alternative to
1816 ausearch, the traditional audit client. Note that this
1817 implements only a minimal audit client. If you want the
1818 special audit modes like reboot-on-log-overflow, please use
1819 the traditional auditd instead, which can be used in
1820 parallel to journald.
1822 * The ConditionSecurity= unit file option now understands the
1823 special string "audit" to check whether auditing is
1826 * journalctl gained two new commands --vacuum-size= and
1827 --vacuum-time= to delete old journal files until the
1828 remaining ones take up no more than the specified size on disk,
1829 or are not older than the specified time.
1831 * A new, native PPPoE library has been added to sd-network,
1832 systemd's library of light-weight networking protocols. This
1833 library will be used in a future version of networkd to
1834 enable PPPoE communication without an external pppd daemon.
1836 * The busctl tool now understands a new "capture" verb that
1837 works similar to "monitor", but writes a packet capture
1838 trace to STDOUT that can be redirected to a file which is
1839 compatible with libcap's capture file format. This can then
1840 be loaded in Wireshark and similar tools to inspect bus
1843 * The busctl tool now understands a new "tree" verb that shows
1844 the object trees of a specific service on the bus, or of all
1847 * The busctl tool now understands a new "introspect" verb that
1848 shows all interfaces and members of objects on the bus,
1849 including their signature and values. This is particularly
1850 useful to get more information about bus objects shown by
1851 the new "busctl tree" command.
1853 * The busctl tool now understands new verbs "call",
1854 "set-property" and "get-property" for invoking bus method
1855 calls, setting and getting bus object properties in a
1858 * busctl gained a new --augment-creds= argument that controls
1859 whether the tool shall augment credential information it
1860 gets from the bus with data from /proc, in a possibly
1863 * nspawn's --link-journal= switch gained two new values
1864 "try-guest" and "try-host" that work like "guest" and
1865 "host", but do not fail if the host has no persistent
1866 journalling enabled. -j is now equivalent to
1867 --link-journal=try-guest.
1869 * macvlan network devices created by nspawn will now have
1870 stable MAC addresses.
1872 * A new SmackProcessLabel= unit setting has been added, which
1873 controls the SMACK security label processes forked off by
1874 the respective unit shall use.
1876 * If compiled with --enable-xkbcommon, systemd-localed will
1877 verify x11 keymap settings by compiling the given keymap. It
1878 will spew out warnings if the compilation fails. This
1879 requires libxkbcommon to be installed.
1881 * When a coredump is collected, a larger number of metadata
1882 fields is now collected and included in the journal records
1883 created for it. More specifically, control group membership,
1884 environment variables, memory maps, working directory,
1885 chroot directory, /proc/$PID/status, and a list of open file
1886 descriptors is now stored in the log entry.
1888 * The udev hwdb now contains DPI information for mice. For
1891 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1893 * All systemd programs that read standalone configuration
1894 files in /etc now also support a corresponding series of
1895 .conf.d configuration directories in /etc/, /run/,
1896 /usr/local/lib/, /usr/lib/, and (if configured with
1897 --enable-split-usr) /lib/. In particular, the following
1898 configuration files now have corresponding configuration
1899 directories: system.conf user.conf, logind.conf,
1900 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1901 resolved.conf, timesyncd.conf, journal-remote.conf, and
1902 journal-upload.conf. Note that distributions should use the
1903 configuration directories in /usr/lib/; the directories in
1904 /etc/ are reserved for the system administrator.
1906 * systemd-rfkill will no longer take the rfkill device name
1907 into account when storing rfkill state on disk, as the name
1908 might be dynamically assigned and not stable. Instead, the
1909 ID_PATH udev variable combined with the rfkill type (wlan,
1910 bluetooth, ...) is used.
1912 * A new service systemd-machine-id-commit.service has been
1913 added. When used on systems where /etc is read-only during
1914 boot, and /etc/machine-id is not initialized (but an empty
1915 file), this service will copy the temporary machine ID
1916 created as replacement into /etc after the system is fully
1917 booted up. This is useful for systems that are freshly
1918 installed with a non-initialized machine ID, but should get
1919 a fixed machine ID for subsequent boots.
1921 * networkd's .netdev files now provide a large set of
1922 configuration parameters for VXLAN devices. Similarly, the
1923 bridge port cost parameter is now configurable in .network
1924 files. There's also new support for configuring IP source
1925 routing. networkd .link files gained support for a new
1926 OriginalName= match that is useful to match against the
1927 original interface name the kernel assigned. .network files
1928 may include MTU= and MACAddress= fields for altering the MTU
1929 and MAC address while being connected to a specific network
1932 * The LUKS logic gained supported for configuring
1933 UUID-specific key files. There's also new support for naming
1934 LUKS device from the kernel command line, using the new
1935 luks.name= argument.
1937 * Timer units may now be transiently created via the bus API
1938 (this was previously already available for scope and service
1939 units). In addition it is now possible to create multiple
1940 transient units at the same time with a single bus call. The
1941 "systemd-run" tool has been updated to make use of this for
1942 running commands on a specified time, in at(1)-style.
1944 * tmpfiles gained support for "t" lines, for assigning
1945 extended attributes to files. Among other uses this may be
1946 used to assign SMACK labels to files.
1948 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1949 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1950 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1951 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1952 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1953 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1954 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1955 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1956 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1957 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1958 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1959 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1960 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1961 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1962 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1963 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1964 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1965 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1967 — Berlin, 2014-12-10
1971 * journalctl gained the new options -t/--identifier= to match
1972 on the syslog identifier (aka "tag"), as well as --utc to
1973 show log timestamps in the UTC timezone. journalctl now also
1974 accepts -n/--lines=all to disable line capping in a pager.
1976 * journalctl gained a new switch, --flush, that synchronously
1977 flushes logs from /run/log/journal to /var/log/journal if
1978 persistent storage is enabled. systemd-journal-flush.service
1979 now waits until the operation is complete.
1981 * Services can notify the manager before they start a reload
1982 (by sending RELOADING=1) or shutdown (by sending
1983 STOPPING=1). This allows the manager to track and show the
1984 internal state of daemons and closes a race condition when
1985 the process is still running but has closed its D-Bus
1988 * Services with Type=oneshot do not have to have any ExecStart
1991 * User units are now loaded also from
1992 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1993 /run/systemd/user directory that was already previously
1994 supported, but is under the control of the user.
1996 * Job timeouts (i.e. time-outs on the time a job that is
1997 queued stays in the run queue) can now optionally result in
1998 immediate reboot or power-off actions (JobTimeoutAction= and
1999 JobTimeoutRebootArgument=). This is useful on ".target"
2000 units, to limit the maximum time a target remains
2001 undispatched in the run queue, and to trigger an emergency
2002 operation in such a case. This is now used by default to
2003 turn off the system if boot-up (as defined by everything in
2004 basic.target) hangs and does not complete for at least
2005 15min. Also, if power-off or reboot hang for at least 30min
2006 an immediate power-off/reboot operation is triggered. This
2007 functionality is particularly useful to increase reliability
2008 on embedded devices, but also on laptops which might
2009 accidentally get powered on when carried in a backpack and
2010 whose boot stays stuck in a hard disk encryption passphrase
2013 * systemd-logind can be configured to also handle lid switch
2014 events even when the machine is docked or multiple displays
2015 are attached (HandleLidSwitchDocked= option).
2017 * A helper binary and a service have been added which can be
2018 used to resume from hibernation in the initramfs. A
2019 generator will parse the resume= option on the kernel
2020 command line to trigger resume.
2022 * A user console daemon systemd-consoled has been
2023 added. Currently, it is a preview, and will so far open a
2024 single terminal on each session of the user marked as
2025 Desktop=systemd-console.
2027 * Route metrics can be specified for DHCP routes added by
2030 * The SELinux context of socket-activated services can be set
2031 from the information provided by the networking stack
2032 (SELinuxContextFromNet= option).
2034 * Userspace firmware loading support has been removed and
2035 the minimum supported kernel version is thus bumped to 3.7.
2037 * Timeout for udev workers has been increased from 1 to 3
2038 minutes, but a warning will be printed after 1 minute to
2039 help diagnose kernel modules that take a long time to load.
2041 * Udev rules can now remove tags on devices with TAG-="foobar".
2043 * systemd's readahead implementation has been removed. In many
2044 circumstances it didn't give expected benefits even for
2045 rotational disk drives and was becoming less relevant in the
2046 age of SSDs. As none of the developers has been using
2047 rotating media anymore, and nobody stepped up to actively
2048 maintain this component of systemd it has now been removed.
2050 * Swap units can use Options= to specify discard options.
2051 Discard options specified for swaps in /etc/fstab are now
2054 * Docker containers are now detected as a separate type of
2057 * The Password Agent protocol gained support for queries where
2058 the user input is shown, useful e.g. for user names.
2059 systemd-ask-password gained a new --echo option to turn that
2062 * The default sysctl.d/ snippets will now set:
2064 net.core.default_qdisc = fq_codel
2066 This selects Fair Queuing Controlled Delay as the default
2067 queuing discipline for network interfaces. fq_codel helps
2068 fight the network bufferbloat problem. It is believed to be
2069 a good default with no tuning required for most workloads.
2070 Downstream distributions may override this choice. On 10Gbit
2071 servers that do not do forwarding, "fq" may perform better.
2072 Systems without a good clocksource should use "pfifo_fast".
2074 * If kdbus is enabled during build a new option BusPolicy= is
2075 available for service units, that allows locking all service
2076 processes into a stricter bus policy, in order to limit
2077 access to various bus services, or even hide most of them
2078 from the service's view entirely.
2080 * networkctl will now show the .network and .link file
2081 networkd has applied to a specific interface.
2083 * sd-login gained a new API call sd_session_get_desktop() to
2084 query which desktop environment has been selected for a
2087 * UNIX utmp support is now compile-time optional to support
2088 legacy-free systems.
2090 * systemctl gained two new commands "add-wants" and
2091 "add-requires" for pulling in units from specific targets
2094 * If the word "rescue" is specified on the kernel command line
2095 the system will now boot into rescue mode (aka
2096 rescue.target), which was previously available only by
2097 specifying "1" or "systemd.unit=rescue.target" on the kernel
2098 command line. This new kernel command line option nicely
2099 mirrors the already existing "emergency" kernel command line
2102 * New kernel command line options mount.usr=, mount.usrflags=,
2103 mount.usrfstype= have been added that match root=, rootflags=,
2104 rootfstype= but allow mounting a specific file system to
2107 * The $NOTIFY_SOCKET is now also passed to control processes of
2108 services, not only the main process.
2110 * This version reenables support for fsck's -l switch. This
2111 means at least version v2.25 of util-linux is required for
2112 operation, otherwise dead-locks on device nodes may
2113 occur. Again: you need to update util-linux to at least
2114 v2.25 when updating systemd to v217.
2116 * The "multi-seat-x" tool has been removed from systemd, as
2117 its functionality has been integrated into X servers 1.16,
2118 and the tool is hence redundant. It is recommended to update
2119 display managers invoking this tool to simply invoke X
2120 directly from now on, again.
2122 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2123 message flag has been added for all of systemd's PolicyKit
2124 authenticated method calls has been added. In particular
2125 this now allows optional interactive authorization via
2126 PolicyKit for many of PID1's privileged operations such as
2127 unit file enabling and disabling.
2129 * "udevadm hwdb --update" learnt a new switch "--usr" for
2130 placing the rebuilt hardware database in /usr instead of
2131 /etc. When used only hardware database entries stored in
2132 /usr will be used, and any user database entries in /etc are
2133 ignored. This functionality is useful for vendors to ship a
2134 pre-built database on systems where local configuration is
2135 unnecessary or unlikely.
2137 * Calendar time specifications in .timer units now also
2138 understand the strings "semi-annually", "quarterly" and
2139 "minutely" as shortcuts (in addition to the preexisting
2140 "anually", "hourly", ...).
2142 * systemd-tmpfiles will now correctly create files in /dev
2143 at boot which are marked for creation only at boot. It is
2144 recommended to always create static device nodes with 'c!'
2145 and 'b!', so that they are created only at boot and not
2146 overwritten at runtime.
2148 * When the watchdog logic is used for a service (WatchdogSec=)
2149 and the watchdog timeout is hit the service will now be
2150 terminated with SIGABRT (instead of just SIGTERM), in order
2151 to make sure a proper coredump and backtrace is
2152 generated. This ensures that hanging services will result in
2153 similar coredump/backtrace behaviour as services that hit a
2156 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2157 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2158 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2159 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2160 Herrmann, David Sommerseth, David Strauss, Emil Renner
2161 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2162 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2163 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2164 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2165 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2166 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2167 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2168 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2169 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2170 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2171 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2172 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2173 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2174 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2175 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2176 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2179 — Berlin, 2014-10-28
2183 * timedated no longer reads NTP implementation unit names from
2184 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2185 implementations should add a
2187 Conflicts=systemd-timesyncd.service
2189 to their unit files to take over and replace systemd's NTP
2190 default functionality.
2192 * systemd-sysusers gained a new line type "r" for configuring
2193 which UID/GID ranges to allocate system users/groups
2194 from. Lines of type "u" may now add an additional column
2195 that specifies the home directory for the system user to be
2196 created. Also, systemd-sysusers may now optionally read user
2197 information from STDIN instead of a file. This is useful for
2198 invoking it from RPM preinst scriptlets that need to create
2199 users before the first RPM file is installed since these
2200 files might need to be owned by them. A new
2201 %sysusers_create_inline RPM macro has been introduced to do
2202 just that. systemd-sysusers now updates the shadow files as
2203 well as the user/group databases, which should enhance
2204 compatibility with certain tools like grpck.
2206 * A number of bus APIs of PID 1 now optionally consult
2207 PolicyKit to permit access for otherwise unprivileged
2208 clients under certain conditions. Note that this currently
2209 doesn't support interactive authentication yet, but this is
2210 expected to be added eventually, too.
2212 * /etc/machine-info now has new fields for configuring the
2213 deployment environment of the machine, as well as the
2214 location of the machine. hostnamectl has been updated with
2215 new command to update these fields.
2217 * systemd-timesyncd has been updated to automatically acquire
2218 NTP server information from systemd-networkd, which might
2219 have been discovered via DHCP.
2221 * systemd-resolved now includes a caching DNS stub resolver
2222 and a complete LLMNR name resolution implementation. A new
2223 NSS module "nss-resolve" has been added which can be used
2224 instead of glibc's own "nss-dns" to resolve hostnames via
2225 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2226 be resolved via systemd-resolved D-Bus APIs. In contrast to
2227 the glibc internal resolver systemd-resolved is aware of
2228 multi-homed system, and keeps DNS server and caches separate
2229 and per-interface. Queries are sent simultaneously on all
2230 interfaces that have DNS servers configured, in order to
2231 properly handle VPNs and local LANs which might resolve
2232 separate sets of domain names. systemd-resolved may acquire
2233 DNS server information from systemd-networkd automatically,
2234 which in turn might have discovered them via DHCP. A tool
2235 "systemd-resolve-host" has been added that may be used to
2236 query the DNS logic in resolved. systemd-resolved implements
2237 IDNA and automatically uses IDNA or UTF-8 encoding depending
2238 on whether classic DNS or LLMNR is used as transport. In the
2239 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2240 implementation to systemd-resolved.
2242 * A new NSS module nss-mymachines has been added, that
2243 automatically resolves the names of all local registered
2244 containers to their respective IP addresses.
2246 * A new client tool "networkctl" for systemd-networkd has been
2247 added. It currently is entirely passive and will query
2248 networking configuration from udev, rtnetlink and networkd,
2249 and present it to the user in a very friendly
2250 way. Eventually, we hope to extend it to become a full
2251 control utility for networkd.
2253 * .socket units gained a new DeferAcceptSec= setting that
2254 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2255 TCP. Similarly, support for controlling TCP keep-alive
2256 settings has been added (KeepAliveTimeSec=,
2257 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2258 turning off Nagle's algorithm on TCP has been added
2261 * logind learned a new session type "web", for use in projects
2262 like Cockpit which register web clients as PAM sessions.
2264 * timer units with at least one OnCalendar= setting will now
2265 be started only after timer-sync.target has been
2266 reached. This way they will not elapse before the system
2267 clock has been corrected by a local NTP client or
2268 similar. This is particular useful on RTC-less embedded
2269 machines, that come up with an invalid system clock.
2271 * systemd-nspawn's --network-veth= switch should now result in
2272 stable MAC addresses for both the outer and the inner side
2275 * systemd-nspawn gained a new --volatile= switch for running
2276 container instances with /etc or /var unpopulated.
2278 * The kdbus client code has been updated to use the new Linux
2279 3.17 memfd subsystem instead of the old kdbus-specific one.
2281 * systemd-networkd's DHCP client and server now support
2282 FORCERENEW. There are also new configuration options to
2283 configure the vendor client identifier and broadcast mode
2286 * systemd will no longer inform the kernel about the current
2287 timezone, as this is necessarily incorrect and racy as the
2288 kernel has no understanding of DST and similar
2289 concepts. This hence means FAT timestamps will be always
2290 considered UTC, similar to what Android is already
2291 doing. Also, when the RTC is configured to the local time
2292 (rather than UTC) systemd will never synchronize back to it,
2293 as this might confuse Windows at a later boot.
2295 * systemd-analyze gained a new command "verify" for offline
2296 validation of unit files.
2298 * systemd-networkd gained support for a couple of additional
2299 settings for bonding networking setups. Also, the metric for
2300 statically configured routes may now be configured. For
2301 network interfaces where this is appropriate the peer IP
2302 address may now be configured.
2304 * systemd-networkd's DHCP client will no longer request
2305 broadcasting by default, as this tripped up some networks.
2306 For hardware where broadcast is required the feature should
2307 be switched back on using RequestBroadcast=yes.
2309 * systemd-networkd will now set up IPv4LL addresses (when
2310 enabled) even if DHCP is configured successfully.
2312 * udev will now default to respect network device names given
2313 by the kernel when the kernel indicates that these are
2314 predictable. This behavior can be tweaked by changing
2315 NamePolicy= in the relevant .link file.
2317 * A new library systemd-terminal has been added that
2318 implements full TTY stream parsing and rendering. This
2319 library is supposed to be used later on for implementing a
2320 full userspace VT subsystem, replacing the current kernel
2323 * A new tool systemd-journal-upload has been added to push
2324 journal data to a remote system running
2325 systemd-journal-remote.
2327 * journald will no longer forward all local data to another
2328 running syslog daemon. This change has been made because
2329 rsyslog (which appears to be the most commonly used syslog
2330 implementation these days) no longer makes use of this, and
2331 instead pulls the data out of the journal on its own. Since
2332 forwarding the messages to a non-existent syslog server is
2333 more expensive than we assumed we have now turned this
2334 off. If you run a syslog server that is not a recent rsyslog
2335 version, you have to turn this option on again
2336 (ForwardToSyslog= in journald.conf).
2338 * journald now optionally supports the LZ4 compressor for
2339 larger journal fields. This compressor should perform much
2340 better than XZ which was the previous default.
2342 * machinectl now shows the IP addresses of local containers,
2343 if it knows them, plus the interface name of the container.
2345 * A new tool "systemd-escape" has been added that makes it
2346 easy to escape strings to build unit names and similar.
2348 * sd_notify() messages may now include a new ERRNO= field
2349 which is parsed and collected by systemd and shown among the
2350 "systemctl status" output for a service.
2352 * A new component "systemd-firstboot" has been added that
2353 queries the most basic systemd information (timezone,
2354 hostname, root password) interactively on first
2355 boot. Alternatively it may also be used to provision these
2356 things offline on OS images installed into directories.
2358 * The default sysctl.d/ snippets will now set
2360 net.ipv4.conf.default.promote_secondaries=1
2362 This has the benefit of no flushing secondary IP addresses
2363 when primary addresses are removed.
2365 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2366 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2367 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2368 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2369 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2370 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2371 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2372 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2373 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2374 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2375 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2376 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2377 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2378 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2379 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2381 — Berlin, 2014-08-19
2385 * A new tool systemd-sysusers has been added. This tool
2386 creates system users and groups in /etc/passwd and
2387 /etc/group, based on static declarative system user/group
2388 definitions in /usr/lib/sysusers.d/. This is useful to
2389 enable factory resets and volatile systems that boot up with
2390 an empty /etc directory, and thus need system users and
2391 groups created during early boot. systemd now also ships
2392 with two default sysusers.d/ files for the most basic
2393 users and groups systemd and the core operating system
2396 * A new tmpfiles snippet has been added that rebuilds the
2397 essential files in /etc on boot, should they be missing.
2399 * A directive for ensuring automatic clean-up of
2400 /var/cache/man/ has been removed from the default
2401 configuration. This line should now be shipped by the man
2402 implementation. The necessary change has been made to the
2403 man-db implementation. Note that you need to update your man
2404 implementation to one that ships this line, otherwise no
2405 automatic clean-up of /var/cache/man will take place.
2407 * A new condition ConditionNeedsUpdate= has been added that
2408 may conditionalize services to only run when /etc or /var
2409 are "older" than the vendor operating system resources in
2410 /usr. This is useful for reconstructing or updating /etc
2411 after an offline update of /usr or a factory reset, on the
2412 next reboot. Services that want to run once after such an
2413 update or reset should use this condition and order
2414 themselves before the new systemd-update-done.service, which
2415 will mark the two directories as fully updated. A number of
2416 service files have been added making use of this, to rebuild
2417 the udev hardware database, the journald message catalog and
2418 dynamic loader cache (ldconfig). The systemd-sysusers tool
2419 described above also makes use of this now. With this in
2420 place it is now possible to start up a minimal operating
2421 system with /etc empty cleanly. For more information on the
2422 concepts involved see this recent blog story:
2424 http://0pointer.de/blog/projects/stateless.html
2426 * A new system group "input" has been introduced, and all
2427 input device nodes get this group assigned. This is useful
2428 for system-level software to get access to input devices. It
2429 complements what is already done for "audio" and "video".
2431 * systemd-networkd learnt minimal DHCPv4 server support in
2432 addition to the existing DHCPv4 client support. It also
2433 learnt DHCPv6 client and IPv6 Router Solicitation client
2434 support. The DHCPv4 client gained support for static routes
2435 passed in from the server. Note that the [DHCPv4] section
2436 known in older systemd-networkd versions has been renamed to
2437 [DHCP] and is now also used by the DHCPv6 client. Existing
2438 .network files using settings of this section should be
2439 updated, though compatibility is maintained. Optionally, the
2440 client hostname may now be sent to the DHCP server.
2442 * networkd gained support for vxlan virtual networks as well
2443 as tun/tap and dummy devices.
2445 * networkd gained support for automatic allocation of address
2446 ranges for interfaces from a system-wide pool of
2447 addresses. This is useful for dynamically managing a large
2448 number of interfaces with a single network configuration
2449 file. In particular this is useful to easily assign
2450 appropriate IP addresses to the veth links of a large number
2451 of nspawn instances.
2453 * RPM macros for processing sysusers, sysctl and binfmt
2454 drop-in snippets at package installation time have been
2457 * The /etc/os-release file should now be placed in
2458 /usr/lib/os-release. The old location is automatically
2459 created as symlink. /usr/lib is the more appropriate
2460 location of this file, since it shall actually describe the
2461 vendor operating system shipped in /usr, and not the
2462 configuration stored in /etc.
2464 * .mount units gained a new boolean SloppyOptions= setting
2465 that maps to mount(8)'s -s option which enables permissive
2466 parsing of unknown mount options.
2468 * tmpfiles learnt a new "L+" directive which creates a symlink
2469 but (unlike "L") deletes a pre-existing file first, should
2470 it already exist and not already be the correct
2471 symlink. Similarly, "b+", "c+" and "p+" directives have been
2472 added as well, which create block and character devices, as
2473 well as fifos in the filesystem, possibly removing any
2474 pre-existing files of different types.
2476 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2477 'argument' field (which so far specified the source to
2478 symlink/copy the files from) is now optional. If omitted the
2479 same file os copied from /usr/share/factory/ suffixed by the
2480 full destination path. This is useful for populating /etc
2481 with essential files, by copying them from vendor defaults
2482 shipped in /usr/share/factory/etc.
2484 * A new command "systemctl preset-all" has been added that
2485 applies the service preset settings to all installed unit
2486 files. A new switch --preset-mode= has been added that
2487 controls whether only enable or only disable operations
2490 * A new command "systemctl is-system-running" has been added
2491 that allows checking the overall state of the system, for
2492 example whether it is fully up and running.
2494 * When the system boots up with an empty /etc, the equivalent
2495 to "systemctl preset-all" is executed during early boot, to
2496 make sure all default services are enabled after a factory
2499 * systemd now contains a minimal preset file that enables the
2500 most basic services systemd ships by default.
2502 * Unit files' [Install] section gained a new DefaultInstance=
2503 field for defining the default instance to create if a
2504 template unit is enabled with no instance specified.
2506 * A new passive target cryptsetup-pre.target has been added
2507 that may be used by services that need to make they run and
2508 finish before the first LUKS cryptographic device is set up.
2510 * The /dev/loop-control and /dev/btrfs-control device nodes
2511 are now owned by the "disk" group by default, opening up
2512 access to this group.
2514 * systemd-coredump will now automatically generate a
2515 stack trace of all core dumps taking place on the system,
2516 based on elfutils' libdw library. This stack trace is logged
2519 * systemd-coredump may now optionally store coredumps directly
2520 on disk (in /var/lib/systemd/coredump, possibly compressed),
2521 instead of storing them unconditionally in the journal. This
2522 mode is the new default. A new configuration file
2523 /etc/systemd/coredump.conf has been added to configure this
2524 and other parameters of systemd-coredump.
2526 * coredumpctl gained a new "info" verb to show details about a
2527 specific coredump. A new switch "-1" has also been added
2528 that makes sure to only show information about the most
2529 recent entry instead of all entries. Also, as the tool is
2530 generally useful now the "systemd-" prefix of the binary
2531 name has been removed. Distributions that want to maintain
2532 compatibility with the old name should add a symlink from
2533 the old name to the new name.
2535 * journald's SplitMode= now defaults to "uid". This makes sure
2536 that unprivileged users can access their own coredumps with
2537 coredumpctl without restrictions.
2539 * New kernel command line options "systemd.wants=" (for
2540 pulling an additional unit during boot), "systemd.mask="
2541 (for masking a specific unit for the boot), and
2542 "systemd.debug-shell" (for enabling the debug shell on tty9)
2543 have been added. This is implemented in the new generator
2544 "systemd-debug-generator".
2546 * systemd-nspawn will now by default filter a couple of
2547 syscalls for containers, among them those required for
2548 kernel module loading, direct x86 IO port access, swap
2549 management, and kexec. Most importantly though
2550 open_by_handle_at() is now prohibited for containers,
2551 closing a hole similar to a recently discussed vulnerability
2552 in docker regarding access to files on file hierarchies the
2553 container should normally not have access to. Note that, for
2554 nspawn, we generally make no security claims anyway (and
2555 this is explicitly documented in the man page), so this is
2556 just a fix for one of the most obvious problems.
2558 * A new man page file-hierarchy(7) has been added that
2559 contains a minimized, modernized version of the file system
2560 layout systemd expects, similar in style to the FHS
2561 specification or hier(5). A new tool systemd-path(1) has
2562 been added to query many of these paths for the local
2565 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2566 longer done. Since the directory now has a per-user size
2567 limit, and is cleaned on logout this appears unnecessary,
2568 in particular since this now brings the lifecycle of this
2569 directory closer in line with how IPC objects are handled.
2571 * systemd.pc now exports a number of additional directories,
2572 including $libdir (which is useful to identify the library
2573 path for the primary architecture of the system), and a
2574 couple of drop-in directories.
2576 * udev's predictable network interface names now use the dev_port
2577 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2578 distinguish between ports of the same PCI function. dev_id should
2579 only be used for ports using the same HW address, hence the need
2582 * machined has been updated to export the OS version of a
2583 container (read from /etc/os-release and
2584 /usr/lib/os-release) on the bus. This is now shown in
2585 "machinectl status" for a machine.
2587 * A new service setting RestartForceExitStatus= has been
2588 added. If configured to a set of exit signals or process
2589 return values, the service will be restarted when the main
2590 daemon process exits with any of them, regardless of the
2593 * systemctl's -H switch for connecting to remote systemd
2594 machines has been extended so that it may be used to
2595 directly connect to a specific container on the
2596 host. "systemctl -H root@foobar:waldi" will now connect as
2597 user "root" to host "foobar", and then proceed directly to
2598 the container named "waldi". Note that currently you have to
2599 authenticate as user "root" for this to work, as entering
2600 containers is a privileged operation.
2602 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2603 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2604 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2605 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2606 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2607 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2608 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2609 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2610 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2611 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2612 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2613 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2615 — Berlin, 2014-07-03
2619 * As an experimental feature, udev now tries to lock the
2620 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2621 executes events for the disk or any of its partitions.
2622 Applications like partitioning programs can lock the
2623 disk device node (flock(LOCK_EX)) and claim temporary
2624 device ownership that way; udev will entirely skip all event
2625 handling for this disk and its partitions. If the disk
2626 was opened for writing, the close will trigger a partition
2627 table rescan in udev's "watch" facility, and if needed
2628 synthesize "change" events for the disk and all its partitions.
2629 This is now unconditionally enabled, and if it turns out to
2630 cause major problems, we might turn it on only for specific
2631 devices, or might need to disable it entirely. Device Mapper
2632 devices are excluded from this logic.
2634 * We temporarily dropped the "-l" switch for fsck invocations,
2635 since they collide with the flock() logic above. util-linux
2636 upstream has been changed already to avoid this conflict,
2637 and we will readd "-l" as soon as util-linux with this
2638 change has been released.
2640 * The dependency on libattr has been removed. Since a long
2641 time, the extended attribute calls have moved to glibc, and
2642 libattr is thus unnecessary.
2644 * Virtualization detection works without priviliges now. This
2645 means the systemd-detect-virt binary no longer requires
2646 CAP_SYS_PTRACE file capabilities, and our daemons can run
2647 with fewer privileges.
2649 * systemd-networkd now runs under its own "systemd-network"
2650 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2651 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2652 loses the ability to write to files owned by root this way.
2654 * Similarly, systemd-resolved now runs under its own
2655 "systemd-resolve" user with no capabilities remaining.
2657 * Similarly, systemd-bus-proxyd now runs under its own
2658 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2660 * systemd-networkd gained support for setting up "veth"
2661 virtual Ethernet devices for container connectivity, as well
2662 as GRE and VTI tunnels.
2664 * systemd-networkd will no longer automatically attempt to
2665 manually load kernel modules necessary for certain tunnel
2666 transports. Instead, it is assumed the kernel loads them
2667 automatically when required. This only works correctly on
2668 very new kernels. On older kernels, please consider adding
2669 the kernel modules to /etc/modules-load.d/ as a work-around.
2671 * The resolv.conf file systemd-resolved generates has been
2672 moved to /run/systemd/resolve/. If you have a symlink from
2673 /etc/resolv.conf, it might be necessary to correct it.
2675 * Two new service settings, ProtectHome= and ProtectSystem=,
2676 have been added. When enabled, they will make the user data
2677 (such as /home) inaccessible or read-only and the system
2678 (such as /usr) read-only, for specific services. This allows
2679 very light-weight per-service sandboxing to avoid
2680 modifications of user data or system files from
2681 services. These two new switches have been enabled for all
2682 of systemd's long-running services, where appropriate.
2684 * Socket units gained new SocketUser= and SocketGroup=
2685 settings to set the owner user and group of AF_UNIX sockets
2686 and FIFOs in the file system.
2688 * Socket units gained a new RemoveOnStop= setting. If enabled,
2689 all FIFOS and sockets in the file system will be removed
2690 when the specific socket unit is stopped.
2692 * Socket units gained a new Symlinks= setting. It takes a list
2693 of symlinks to create to file system sockets or FIFOs
2694 created by the specific Unix sockets. This is useful to
2695 manage symlinks to socket nodes with the same life-cycle as
2698 * The /dev/log socket and /dev/initctl FIFO have been moved to
2699 /run, and have been replaced by symlinks. This allows
2700 connecting to these facilities even if PrivateDevices=yes is
2701 used for a service (which makes /dev/log itself unavailable,
2702 but /run is left). This also has the benefit of ensuring
2703 that /dev only contains device nodes, directories and
2704 symlinks, and nothing else.
2706 * sd-daemon gained two new calls sd_pid_notify() and
2707 sd_pid_notifyf(). They are similar to sd_notify() and
2708 sd_notifyf(), but allow overriding of the source PID of
2709 notification messages if permissions permit this. This is
2710 useful to send notify messages on behalf of a different
2711 process (for example, the parent process). The
2712 systemd-notify tool has been updated to make use of this
2713 when sending messages (so that notification messages now
2714 originate from the shell script invoking systemd-notify and
2715 not the systemd-notify process itself. This should minimize
2716 a race where systemd fails to associate notification
2717 messages to services when the originating process already
2720 * A new "on-abnormal" setting for Restart= has been added. If
2721 set, it will result in automatic restarts on all "abnormal"
2722 reasons for a process to exit, which includes unclean
2723 signals, core dumps, timeouts and watchdog timeouts, but
2724 does not include clean and unclean exit codes or clean
2725 signals. Restart=on-abnormal is an alternative for
2726 Restart=on-failure for services that shall be able to
2727 terminate and avoid restarts on certain errors, by
2728 indicating so with an unclean exit code. Restart=on-failure
2729 or Restart=on-abnormal is now the recommended setting for
2730 all long-running services.
2732 * If the InaccessibleDirectories= service setting points to a
2733 mount point (or if there are any submounts contained within
2734 it), it is now attempted to completely unmount it, to make
2735 the file systems truly unavailable for the respective
2738 * The ReadOnlyDirectories= service setting and
2739 systemd-nspawn's --read-only parameter are now recursively
2740 applied to all submounts, too.
2742 * Mount units may now be created transiently via the bus APIs.
2744 * The support for SysV and LSB init scripts has been removed
2745 from the systemd daemon itself. Instead, it is now
2746 implemented as a generator that creates native systemd units
2747 from these scripts when needed. This enables us to remove a
2748 substantial amount of legacy code from PID 1, following the
2749 fact that many distributions only ship a very small number
2750 of LSB/SysV init scripts nowadays.
2752 * Privileged Xen (dom0) domains are not considered
2753 virtualization anymore by the virtualization detection
2754 logic. After all, they generally have unrestricted access to
2755 the hardware and usually are used to manage the unprivileged
2758 * systemd-tmpfiles gained a new "C" line type, for copying
2759 files or entire directories.
2761 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2762 lines. So far, they have been non-globbing versions of the
2763 latter, and have thus been redundant. In future, it is
2764 recommended to only use "z". "m" has hence been removed
2765 from the documentation, even though it stays supported.
2767 * A tmpfiles snippet to recreate the most basic structure in
2768 /var has been added. This is enough to create the /var/run →
2769 /run symlink and create a couple of structural
2770 directories. This allows systems to boot up with an empty or
2771 volatile /var. Of course, while with this change, the core OS
2772 now is capable with dealing with a volatile /var, not all
2773 user services are ready for it. However, we hope that sooner
2774 or later, many service daemons will be changed upstream so
2775 that they are able to automatically create their necessary
2776 directories in /var at boot, should they be missing. This is
2777 the first step to allow state-less systems that only require
2778 the vendor image for /usr to boot.
2780 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2781 empty tmpfs instance to a specific directory. This is
2782 particularly useful for making use of the automatic
2783 reconstruction of /var (see above), by passing --tmpfs=/var.
2785 * Access modes specified in tmpfiles snippets may now be
2786 prefixed with "~", which indicates that they shall be masked
2787 by whether the existing file or directory is currently
2788 writable, readable or executable at all. Also, if specified,
2789 the sgid/suid/sticky bits will be masked for all
2792 * A new passive target unit "network-pre.target" has been
2793 added which is useful for services that shall run before any
2794 network is configured, for example firewall scripts.
2796 * The "floppy" group that previously owned the /dev/fd*
2797 devices is no longer used. The "disk" group is now used
2798 instead. Distributions should probably deprecate usage of
2801 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2802 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2803 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2804 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2805 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2806 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2809 — Berlin, 2014-06-11
2813 * A new "systemd-timesyncd" daemon has been added for
2814 synchronizing the system clock across the network. It
2815 implements an SNTP client. In contrast to NTP
2816 implementations such as chrony or the NTP reference server,
2817 this only implements a client side, and does not bother with
2818 the full NTP complexity, focusing only on querying time from
2819 one remote server and synchronizing the local clock to
2820 it. Unless you intend to serve NTP to networked clients or
2821 want to connect to local hardware clocks, this simple NTP
2822 client should be more than appropriate for most
2823 installations. The daemon runs with minimal privileges, and
2824 has been hooked up with networkd to only operate when
2825 network connectivity is available. The daemon saves the
2826 current clock to disk every time a new NTP sync has been
2827 acquired, and uses this to possibly correct the system clock
2828 early at bootup, in order to accommodate for systems that
2829 lack an RTC such as the Raspberry Pi and embedded devices,
2830 and to make sure that time monotonically progresses on these
2831 systems, even if it is not always correct. To make use of
2832 this daemon, a new system user and group "systemd-timesync"
2833 needs to be created on installation of systemd.
2835 * The queue "seqnum" interface of libudev has been disabled, as
2836 it was generally incompatible with device namespacing as
2837 sequence numbers of devices go "missing" if the devices are
2838 part of a different namespace.
2840 * "systemctl list-timers" and "systemctl list-sockets" gained
2841 a --recursive switch for showing units of these types also
2842 for all local containers, similar in style to the already
2843 supported --recursive switch for "systemctl list-units".
2845 * A new RebootArgument= setting has been added for service
2846 units, which may be used to specify a kernel reboot argument
2847 to use when triggering reboots with StartLimitAction=.
2849 * A new FailureAction= setting has been added for service
2850 units which may be used to specify an operation to trigger
2851 when a service fails. This works similarly to
2852 StartLimitAction=, but unlike it, controls what is done
2853 immediately rather than only after several attempts to
2854 restart the service in question.
2856 * hostnamed got updated to also expose the kernel name,
2857 release, and version on the bus. This is useful for
2858 executing commands like hostnamectl with the -H switch.
2859 systemd-analyze makes use of this to properly display
2860 details when running non-locally.
2862 * The bootchart tool can now show cgroup information in the
2863 graphs it generates.
2865 * The CFS CPU quota cgroup attribute is now exposed for
2866 services. The new CPUQuota= switch has been added for this
2867 which takes a percentage value. Setting this will have the
2868 result that a service may never get more CPU time than the
2869 specified percentage, even if the machine is otherwise idle.
2871 * systemd-networkd learned IPIP and SIT tunnel support.
2873 * LSB init scripts exposing a dependency on $network will now
2874 get a dependency on network-online.target rather than simply
2875 network.target. This should bring LSB handling closer to
2876 what it was on SysV systems.
2878 * A new fsck.repair= kernel option has been added to control
2879 how fsck shall deal with unclean file systems at boot.
2881 * The (.ini) configuration file parser will now silently
2882 ignore sections whose name begins with "X-". This may be
2883 used to maintain application-specific extension sections in unit
2886 * machined gained a new API to query the IP addresses of
2887 registered containers. "machinectl status" has been updated
2888 to show these addresses in its output.
2890 * A new call sd_uid_get_display() has been added to the
2891 sd-login APIs for querying the "primary" session of a
2892 user. The "primary" session of the user is elected from the
2893 user's sessions and generally a graphical session is
2894 preferred over a text one.
2896 * A minimal systemd-resolved daemon has been added. It
2897 currently simply acts as a companion to systemd-networkd and
2898 manages resolv.conf based on per-interface DNS
2899 configuration, possibly supplied via DHCP. In the long run
2900 we hope to extend this into a local DNSSEC enabled DNS and
2903 * The systemd-networkd-wait-online tool is now enabled by
2904 default. It will delay network-online.target until a network
2905 connection has been configured. The tool primarily integrates
2906 with networkd, but will also make a best effort to make sense
2907 of network configuration performed in some other way.
2909 * Two new service options StartupCPUShares= and
2910 StartupBlockIOWeight= have been added that work similarly to
2911 CPUShares= and BlockIOWeight= however only apply during
2912 system startup. This is useful to prioritize certain services
2913 differently during bootup than during normal runtime.
2915 * hostnamed has been changed to prefer the statically
2916 configured hostname in /etc/hostname (unless set to
2917 'localhost' or empty) over any dynamic one supplied by
2918 dhcp. With this change, the rules for picking the hostname
2919 match more closely the rules of other configuration settings
2920 where the local administrator's configuration in /etc always
2921 overrides any other settings.
2923 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2924 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2925 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2926 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2927 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2928 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2929 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2930 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2931 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2932 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2933 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2934 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2935 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2936 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2937 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2938 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2941 — Beijing, 2014-05-28
2945 * When restoring the screen brightness at boot, stay away from
2946 the darkest setting or from the lowest 5% of the available
2947 range, depending on which is the larger value of both. This
2948 should effectively protect the user from rebooting into a
2949 black screen, should the brightness have been set to minimum
2952 * sd-login gained a new sd_machine_get_class() call to
2953 determine the class ("vm" or "container") of a machine
2954 registered with machined.
2956 * sd-login gained new calls
2957 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2958 to query the identity of the peer of a local AF_UNIX
2959 connection. They operate similarly to their sd_pid_get_xyz()
2962 * PID 1 will now maintain a system-wide system state engine
2963 with the states "starting", "running", "degraded",
2964 "maintenance", "stopping". These states are bound to system
2965 startup, normal runtime, runtime with at least one failed
2966 service, rescue/emergency mode and system shutdown. This
2967 state is shown in the "systemctl status" output when no unit
2968 name is passed. It is useful to determine system state, in
2969 particularly when doing so for many systems or containers at
2972 * A new command "list-machines" has been added to "systemctl"
2973 that lists all local OS containers and shows their system
2974 state (see above), if systemd runs inside of them.
2976 * systemctl gained a new "-r" switch to recursively enumerate
2977 units on all local containers, when used with the
2978 "list-unit" command (which is the default one that is
2979 executed when no parameters are specified).
2981 * The GPT automatic partition discovery logic will now honour
2982 two GPT partition flags: one may be set on a partition to
2983 cause it to be mounted read-only, and the other may be set
2984 on a partition to ignore it during automatic discovery.
2986 * Two new GPT type UUIDs have been added for automatic root
2987 partition discovery, for 32-bit and 64-bit ARM. This is not
2988 particularly useful for discovering the root directory on
2989 these architectures during bare-metal boots (since UEFI is
2990 not common there), but still very useful to allow booting of
2991 ARM disk images in nspawn with the -i option.
2993 * MAC addresses of interfaces created with nspawn's
2994 --network-interface= switch will now be generated from the
2995 machine name, and thus be stable between multiple invocations
2998 * logind will now automatically remove all IPC objects owned
2999 by a user if she or he fully logs out. This makes sure that
3000 users who are logged out cannot continue to consume IPC
3001 resources. This covers SysV memory, semaphores and message
3002 queues as well as POSIX shared memory and message
3003 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3004 limits. With this functionality, that is corrected. This may
3005 be turned off by using the RemoveIPC= switch of logind.conf.
3007 * The systemd-machine-id-setup and tmpfiles tools gained a
3008 --root= switch to operate on a specific root directory,
3011 * journald can now forward logged messages to the TTYs of all
3012 logged in users ("wall"). This is the default for all
3013 emergency messages now.
3015 * A new tool systemd-journal-remote has been added to stream
3016 journal log messages across the network.
3018 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3019 controller trees are mounted into it. Note that the
3020 directories mounted beneath it are not read-only. This is a
3021 security measure and is particularly useful because glibc
3022 actually includes a search logic to pick any tmpfs it can
3023 find to implement shm_open() if /dev/shm is not available
3024 (which it might very well be in namespaced setups).
3026 * machinectl gained a new "poweroff" command to cleanly power
3027 down a local OS container.
3029 * The PrivateDevices= unit file setting will now also drop the
3030 CAP_MKNOD capability from the capability bound set, and
3031 imply DevicePolicy=closed.
3033 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3034 comprehensively on all long-running systemd services where
3035 this is appropriate.
3037 * systemd-udevd will now run in a disassociated mount
3038 namespace. To mount directories from udev rules, make sure to
3039 pull in mount units via SYSTEMD_WANTS properties.
3041 * The kdbus support gained support for uploading policy into
3042 the kernel. sd-bus gained support for creating "monitoring"
3043 connections that can eavesdrop into all bus communication
3044 for debugging purposes.
3046 * Timestamps may now be specified in seconds since the UNIX
3047 epoch Jan 1st, 1970 by specifying "@" followed by the value
3050 * Native tcpwrap support in systemd has been removed. tcpwrap
3051 is old code, not really maintained anymore and has serious
3052 shortcomings, and better options such as firewalls
3053 exist. For setups that require tcpwrap usage, please
3054 consider invoking your socket-activated service via tcpd,
3055 like on traditional inetd.
3057 * A new system.conf configuration option
3058 DefaultTimerAccuracySec= has been added that controls the
3059 default AccuracySec= setting of .timer units.
3061 * Timer units gained a new WakeSystem= switch. If enabled,
3062 timers configured this way will cause the system to resume
3063 from system suspend (if the system supports that, which most
3066 * Timer units gained a new Persistent= switch. If enabled,
3067 timers configured this way will save to disk when they have
3068 been last triggered. This information is then used on next
3069 reboot to possible execute overdue timer events, that
3070 could not take place because the system was powered off.
3071 This enables simple anacron-like behaviour for timer units.
3073 * systemctl's "list-timers" will now also list the time a
3074 timer unit was last triggered in addition to the next time
3075 it will be triggered.
3077 * systemd-networkd will now assign predictable IPv4LL
3078 addresses to its local interfaces.
3080 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3081 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3082 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3083 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3084 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3085 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3086 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3087 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3090 — Berlin, 2014-03-25
3094 * A new unit file setting RestrictAddressFamilies= has been
3095 added to restrict which socket address families unit
3096 processes gain access to. This takes address family names
3097 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3098 attack surface of services via exotic protocol stacks. This
3099 is built on seccomp system call filters.
3101 * Two new unit file settings RuntimeDirectory= and
3102 RuntimeDirectoryMode= have been added that may be used to
3103 manage a per-daemon runtime directories below /run. This is
3104 an alternative for setting up directory permissions with
3105 tmpfiles snippets, and has the advantage that the runtime
3106 directory's lifetime is bound to the daemon runtime and that
3107 the daemon starts up with an empty directory each time. This
3108 is particularly useful when writing services that drop
3109 privileges using the User= or Group= setting.
3111 * The DeviceAllow= unit setting now supports globbing for
3112 matching against device group names.
3114 * The systemd configuration file system.conf gained new
3115 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3116 DefaultMemoryAccounting= to globally turn on/off accounting
3117 for specific resources (cgroups) for all units. These
3118 settings may still be overridden individually in each unit
3121 * systemd-gpt-auto-generator is now able to discover /srv and
3122 root partitions in addition to /home and swap partitions. It
3123 also supports LUKS-encrypted partitions now. With this in
3124 place, automatic discovery of partitions to mount following
3125 the Discoverable Partitions Specification
3126 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3127 is now a lot more complete. This allows booting without
3128 /etc/fstab and without root= on the kernel command line on
3129 systems prepared appropriately.
3131 * systemd-nspawn gained a new --image= switch which allows
3132 booting up disk images and Linux installations on any block
3133 device that follow the Discoverable Partitions Specification
3134 (see above). This means that installations made with
3135 appropriately updated installers may now be started and
3136 deployed using container managers, completely
3137 unmodified. (We hope that libvirt-lxc will add support for
3138 this feature soon, too.)
3140 * systemd-nspawn gained a new --network-macvlan= setting to
3141 set up a private macvlan interface for the
3142 container. Similarly, systemd-networkd gained a new
3143 Kind=macvlan setting in .netdev files.
3145 * systemd-networkd now supports configuring local addresses
3148 * A new tool systemd-network-wait-online has been added to
3149 synchronously wait for network connectivity using
3152 * The sd-bus.h bus API gained a new sd_bus_track object for
3153 tracking the life-cycle of bus peers. Note that sd-bus.h is
3154 still not a public API though (unless you specify
3155 --enable-kdbus on the configure command line, which however
3156 voids your warranty and you get no API stability guarantee).
3158 * The $XDG_RUNTIME_DIR runtime directories for each user are
3159 now individual tmpfs instances, which has the benefit of
3160 introducing separate pools for each user, with individual
3161 size limits, and thus making sure that unprivileged clients
3162 can no longer negatively impact the system or other users by
3163 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3164 RuntimeDirectorySize= has been introduced that allows
3165 controlling the default size limit for all users. It
3166 defaults to 10% of the available physical memory. This is no
3167 replacement for quotas on tmpfs though (which the kernel
3168 still does not support), as /dev/shm and /tmp are still
3169 shared resources used by both the system and unprivileged
3172 * logind will now automatically turn off automatic suspending
3173 on laptop lid close when more than one display is
3174 connected. This was previously expected to be implemented
3175 individually in desktop environments (such as GNOME),
3176 however has been added to logind now, in order to fix a
3177 boot-time race where a desktop environment might not have
3178 been started yet and thus not been able to take an inhibitor
3179 lock at the time where logind already suspends the system
3180 due to a closed lid.
3182 * logind will now wait at least 30s after each system
3183 suspend/resume cycle, and 3min after system boot before
3184 suspending the system due to a closed laptop lid. This
3185 should give USB docking stations and similar enough time to
3186 be probed and configured after system resume and boot in
3187 order to then act as suspend blocker.
3189 * systemd-run gained a new --property= setting which allows
3190 initialization of resource control properties (and others)
3191 for the created scope or service unit. Example: "systemd-run
3192 --property=BlockIOWeight=10 updatedb" may be used to run
3193 updatedb at a low block IO scheduling weight.
3195 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3196 now also work in --scope mode.
3198 * When systemd is compiled with kdbus support, basic support
3199 for enforced policies is now in place. (Note that enabling
3200 kdbus still voids your warranty and no API compatibility
3203 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3204 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3205 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3206 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3207 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3208 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3209 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3210 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3211 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3212 Zbigniew Jędrzejewski-Szmek
3214 — Berlin, 2014-03-12
3218 * systemd will now relabel /dev after loading the SMACK policy
3219 according to SMACK rules.
3221 * A new unit file option AppArmorProfile= has been added to
3222 set the AppArmor profile for the processes of a unit.
3224 * A new condition check ConditionArchitecture= has been added
3225 to conditionalize units based on the system architecture, as
3226 reported by uname()'s "machine" field.
3228 * systemd-networkd now supports matching on the system
3229 virtualization, architecture, kernel command line, host name
3232 * logind is now a lot more aggressive when suspending the
3233 machine due to a closed laptop lid. Instead of acting only
3234 on the lid close action, it will continuously watch the lid
3235 status and act on it. This is useful for laptops where the
3236 power button is on the outside of the chassis so that it can
3237 be reached without opening the lid (such as the Lenovo
3238 Yoga). On those machines, logind will now immediately
3239 re-suspend the machine if the power button has been
3240 accidentally pressed while the laptop was suspended and in a
3241 backpack or similar.
3243 * logind will now watch SW_DOCK switches and inhibit reaction
3244 to the lid switch if it is pressed. This means that logind
3245 will not suspend the machine anymore if the lid is closed
3246 and the system is docked, if the laptop supports SW_DOCK
3247 notifications via the input layer. Note that ACPI docking
3248 stations do not generate this currently. Also note that this
3249 logic is usually not fully sufficient and Desktop
3250 Environments should take a lid switch inhibitor lock when an
3251 external display is connected, as systemd will not watch
3254 * nspawn will now make use of the devices cgroup controller by
3255 default, and only permit creation of and access to the usual
3256 API device nodes like /dev/null or /dev/random, as well as
3257 access to (but not creation of) the pty devices.
3259 * We will now ship a default .network file for
3260 systemd-networkd that automatically configures DHCP for
3261 network interfaces created by nspawn's --network-veth or
3262 --network-bridge= switches.
3264 * systemd will now understand the usual M, K, G, T suffixes
3265 according to SI conventions (i.e. to the base 1000) when
3266 referring to throughput and hardware metrics. It will stay
3267 with IEC conventions (i.e. to the base 1024) for software
3268 metrics, according to what is customary according to
3269 Wikipedia. We explicitly document which base applies for
3270 each configuration option.
3272 * The DeviceAllow= setting in unit files now supports a syntax
3273 to whitelist an entire group of devices node majors at once,
3274 based on the /proc/devices listing. For example, with the
3275 string "char-pts", it is now possible to whitelist all
3276 current and future pseudo-TTYs at once.
3278 * sd-event learned a new "post" event source. Event sources of
3279 this type are triggered by the dispatching of any event
3280 source of a type that is not "post". This is useful for
3281 implementing clean-up and check event sources that are
3282 triggered by other work being done in the program.
3284 * systemd-networkd is no longer statically enabled, but uses
3285 the usual [Install] sections so that it can be
3286 enabled/disabled using systemctl. It still is enabled by
3289 * When creating a veth interface pair with systemd-nspawn, the
3290 host side will now be prefixed with "vb-" if
3291 --network-bridge= is used, and with "ve-" if --network-veth
3292 is used. This way, it is easy to distinguish these cases on
3293 the host, for example to apply different configuration to
3294 them with systemd-networkd.
3296 * The compatibility libraries for libsystemd-journal.so,
3297 libsystem-id128.so, libsystemd-login.so and
3298 libsystemd-daemon.so do not make use of IFUNC
3299 anymore. Instead, we now build libsystemd.so multiple times
3300 under these alternative names. This means that the footprint
3301 is drastically increased, but given that these are
3302 transitional compatibility libraries, this should not matter
3303 much. This change has been made necessary to support the ARM
3304 platform for these compatibility libraries, as the ARM
3305 toolchain is not really at the same level as the toolchain
3306 for other architectures like x86 and does not support
3307 IFUNC. Please make sure to use --enable-compat-libs only
3308 during a transitional period!
3310 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3311 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3312 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3313 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3314 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3315 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3316 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3317 Zbigniew Jędrzejewski-Szmek
3319 — Berlin, 2014-02-24
3323 * A new component "systemd-networkd" has been added that can
3324 be used to configure local network interfaces statically or
3325 via DHCP. It is capable of bringing up bridges, VLANs, and
3326 bonding. Currently, no hook-ups for interactive network
3327 configuration are provided. Use this for your initrd,
3328 container, embedded, or server setup if you need a simple,
3329 yet powerful, network configuration solution. This
3330 configuration subsystem is quite nifty, as it allows wildcard
3331 hotplug matching in interfaces. For example, with a single
3332 configuration snippet, you can configure that all Ethernet
3333 interfaces showing up are automatically added to a bridge,
3334 or similar. It supports link-sensing and more.
3336 * A new tool "systemd-socket-proxyd" has been added which can
3337 act as a bidirectional proxy for TCP sockets. This is
3338 useful for adding socket activation support to services that
3339 do not actually support socket activation, including virtual
3340 machines and the like.
3342 * Add a new tool to save/restore rfkill state on
3345 * Save/restore state of keyboard backlights in addition to
3346 display backlights on shutdown/boot.
3348 * udev learned a new SECLABEL{} construct to label device
3349 nodes with a specific security label when they appear. For
3350 now, only SECLABEL{selinux} is supported, but the syntax is
3351 prepared for additional security frameworks.
3353 * udev gained a new scheme to configure link-level attributes
3354 from files in /etc/systemd/network/*.link. These files can
3355 match against MAC address, device path, driver name and type,
3356 and will apply attributes like the naming policy, link speed,
3357 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3358 address assignment policy (randomized, ...).
3360 * The configuration of network interface naming rules for
3361 "permanent interface names" has changed: a new NamePolicy=
3362 setting in the [Link] section of .link files determines the
3363 priority of possible naming schemes (onboard, slot, MAC,
3364 path). The default value of this setting is determined by
3365 /usr/lib/net/links/99-default.link. Old
3366 80-net-name-slot.rules udev configuration file has been
3367 removed, so local configuration overriding this file should
3368 be adapated to override 99-default.link instead.
3370 * When the User= switch is used in a unit file, also
3371 initialize $SHELL= based on the user database entry.
3373 * systemd no longer depends on libdbus. All communication is
3374 now done with sd-bus, systemd's low-level bus library
3377 * kdbus support has been added to PID 1 itself. When kdbus is
3378 enabled, this causes PID 1 to set up the system bus and
3379 enable support for a new ".busname" unit type that
3380 encapsulates bus name activation on kdbus. It works a little
3381 bit like ".socket" units, except for bus names. A new
3382 generator has been added that converts classic dbus1 service
3383 activation files automatically into native systemd .busname
3386 * sd-bus: add a light-weight vtable implementation that allows
3387 defining objects on the bus with a simple static const
3388 vtable array of its methods, signals and properties.
3390 * systemd will not generate or install static dbus
3391 introspection data anymore to /usr/share/dbus-1/interfaces,
3392 as the precise format of these files is unclear, and
3393 nothing makes use of it.
3395 * A proxy daemon is now provided to proxy clients connecting
3396 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3397 compatibility with classic D-Bus.
3399 * A bus driver implementation has been added that supports the
3400 classic D-Bus bus driver calls on kdbus, also for
3401 compatibility purposes.
3403 * A new API "sd-event.h" has been added that implements a
3404 minimal event loop API built around epoll. It provides a
3405 couple of features that direct epoll usage is lacking:
3406 prioritization of events, scales to large numbers of timer
3407 events, per-event timer slack (accuracy), system-wide
3408 coalescing of timer events, exit handlers, watchdog
3409 supervision support using systemd's sd_notify() API, child
3412 * A new API "sd-rntl.h" has been added that provides an API
3413 around the route netlink interface of the kernel, similar in
3414 style to "sd-bus.h".
3416 * A new API "sd-dhcp-client.h" has been added that provides a
3417 small DHCPv4 client-side implementation. This is used by
3420 * There is a new kernel command line option
3421 "systemd.restore_state=0|1". When set to "0", none of the
3422 systemd tools will restore saved runtime state to hardware
3423 devices. More specifically, the rfkill and backlight states
3426 * The FsckPassNo= compatibility option in mount/service units
3427 has been removed. The fstab generator will now add the
3428 necessary dependencies automatically, and does not require
3429 PID1's support for that anymore.
3431 * journalctl gained a new switch, --list-boots, that lists
3432 recent boots with their times and boot IDs.
3434 * The various tools like systemctl, loginctl, timedatectl,
3435 busctl, systemd-run, ... have gained a new switch "-M" to
3436 connect to a specific, local OS container (as direct
3437 connection, without requiring SSH). This works on any
3438 container that is registered with machined, such as those
3439 created by libvirt-lxc or nspawn.
3441 * systemd-run and systemd-analyze also gained support for "-H"
3442 to connect to remote hosts via SSH. This is particularly
3443 useful for systemd-run because it enables queuing of jobs
3444 onto remote systems.
3446 * machinectl gained a new command "login" to open a getty
3447 login in any local container. This works with any container
3448 that is registered with machined (such as those created by
3449 libvirt-lxc or nspawn), and which runs systemd inside.
3451 * machinectl gained a new "reboot" command that may be used to
3452 trigger a reboot on a specific container that is registered
3453 with machined. This works on any container that runs an init
3454 system of some kind.
3456 * systemctl gained a new "list-timers" command to print a nice
3457 listing of installed timer units with the times they elapse
3460 * Alternative reboot() parameters may now be specified on the
3461 "systemctl reboot" command line and are passed to the
3462 reboot() system call.
3464 * systemctl gained a new --job-mode= switch to configure the
3465 mode to queue a job with. This is a more generic version of
3466 --fail, --irreversible, and --ignore-dependencies, which are
3467 still available but not advertised anymore.
3469 * /etc/systemd/system.conf gained new settings to configure
3470 various default timeouts of units, as well as the default
3471 start limit interval and burst. These may still be overridden
3474 * PID1 will now export on the bus profile data of the security
3475 policy upload process (such as the SELinux policy upload to
3478 * journald: when forwarding logs to the console, include
3479 timestamps (following the setting in
3480 /sys/module/printk/parameters/time).
3482 * OnCalendar= in timer units now understands the special
3483 strings "yearly" and "annually". (Both are equivalent)
3485 * The accuracy of timer units is now configurable with the new
3486 AccuracySec= setting. It defaults to 1min.
3488 * A new dependency type JoinsNamespaceOf= has been added that
3489 allows running two services within the same /tmp and network
3490 namespace, if PrivateNetwork= or PrivateTmp= are used.
3492 * A new command "cat" has been added to systemctl. It outputs
3493 the original unit file of a unit, and concatenates the
3494 contents of additional "drop-in" unit file snippets, so that
3495 the full configuration is shown.
3497 * systemctl now supports globbing on the various "list-xyz"
3498 commands, like "list-units" or "list-sockets", as well as on
3499 those commands which take multiple unit names.
3501 * journalctl's --unit= switch gained support for globbing.
3503 * All systemd daemons now make use of the watchdog logic so
3504 that systemd automatically notices when they hang.
3506 * If the $container_ttys environment variable is set,
3507 getty-generator will automatically spawn a getty for each
3508 listed tty. This is useful for container managers to request
3509 login gettys to be spawned on as many ttys as needed.
3511 * %h, %s, %U specifier support is not available anymore when
3512 used in unit files for PID 1. This is because NSS calls are
3513 not safe from PID 1. They stay available for --user
3514 instances of systemd, and as special case for the root user.
3516 * loginctl gained a new "--no-legend" switch to turn off output
3519 * The "sd-login.h" API gained three new calls:
3520 sd_session_is_remote(), sd_session_get_remote_user(),
3521 sd_session_get_remote_host() to query information about
3524 * The udev hardware database now also carries vendor/product
3525 information of SDIO devices.
3527 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3528 determine whether watchdog notifications are requested by
3531 * Socket-activated per-connection services now include a
3532 short description of the connection parameters in the
3535 * tmpfiles gained a new "--boot" option. When this is not used,
3536 only lines where the command character is not suffixed with
3537 "!" are executed. When this option is specified, those
3538 options are executed too. This partitions tmpfiles
3539 directives into those that can be safely executed at any
3540 time, and those which should be run only at boot (for
3541 example, a line that creates /run/nologin).
3543 * A new API "sd-resolve.h" has been added which provides a simple
3544 asynchronous wrapper around glibc NSS host name resolution
3545 calls, such as getaddrinfo(). In contrast to glibc's
3546 getaddrinfo_a(), it does not use signals. In contrast to most
3547 other asynchronous name resolution libraries, this one does
3548 not reimplement DNS, but reuses NSS, so that alternate
3549 host name resolution systems continue to work, such as mDNS,
3550 LDAP, etc. This API is based on libasyncns, but it has been
3551 cleaned up for inclusion in systemd.
3553 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3554 "sd-daemon.h" are no longer found in individual libraries
3555 libsystemd-journal.so, libsystemd-login.so,
3556 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3557 merged them into a single library, libsystemd.so, which
3558 provides all symbols. The reason for this is cyclic
3559 dependencies, as these libraries tend to use each other's
3560 symbols. So far, we have managed to workaround that by linking
3561 a copy of a good part of our code into each of these
3562 libraries again and again, which, however, makes certain
3563 things hard to do, like sharing static variables. Also, it
3564 substantially increases footprint. With this change, there
3565 is only one library for the basic APIs systemd
3566 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3567 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3568 library as well, however are subject to the --enable-kdbus
3569 switch (see below). Note that "sd-dhcp-client.h" is not part
3570 of this library (this is because it only consumes, never
3571 provides, services of/to other APIs). To make the transition
3572 easy from the separate libraries to the unified one, we
3573 provide the --enable-compat-libs compile-time switch which
3574 will generate stub libraries that are compatible with the
3575 old ones but redirect all calls to the new one.
3577 * All of the kdbus logic and the new APIs "sd-bus.h",
3578 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3579 and "sd-utf8.h" are compile-time optional via the
3580 "--enable-kdbus" switch, and they are not compiled in by
3581 default. To make use of kdbus, you have to explicitly enable
3582 the switch. Note however, that neither the kernel nor the
3583 userspace API for all of this is considered stable yet. We
3584 want to maintain the freedom to still change the APIs for
3585 now. By specifying this build-time switch, you acknowledge
3586 that you are aware of the instability of the current
3589 * Also, note that while kdbus is pretty much complete,
3590 it lacks one thing: proper policy support. This means you
3591 can build a fully working system with all features; however,
3592 it will be highly insecure. Policy support will be added in
3593 one of the next releases, at the same time that we will
3594 declare the APIs stable.
3596 * When the kernel command line argument "kdbus" is specified,
3597 systemd will automatically load the kdbus.ko kernel module. At
3598 this stage of development, it is only useful for testing kdbus
3599 and should not be used in production. Note: if "--enable-kdbus"
3600 is specified, and the kdbus.ko kernel module is available, and
3601 "kdbus" is added to the kernel command line, the entire system
3602 runs with kdbus instead of dbus-daemon, with the above mentioned
3603 problem of missing the system policy enforcement. Also a future
3604 version of kdbus.ko or a newer systemd will not be compatible with
3605 each other, and will unlikely be able to boot the machine if only
3606 one of them is updated.
3608 * systemctl gained a new "import-environment" command which
3609 uploads the caller's environment (or parts thereof) into the
3610 service manager so that it is inherited by services started
3611 by the manager. This is useful to upload variables like
3612 $DISPLAY into the user service manager.
3614 * A new PrivateDevices= switch has been added to service units
3615 which allows running a service with a namespaced /dev
3616 directory that does not contain any device nodes for
3617 physical devices. More specifically, it only includes devices
3618 such as /dev/null, /dev/urandom, and /dev/zero which are API
3621 * logind has been extended to support behaviour like VT
3622 switching on seats that do not support a VT. This makes
3623 multi-session available on seats that are not the first seat
3624 (seat0), and on systems where kernel support for VTs has
3625 been disabled at compile-time.
3627 * If a process holds a delay lock for system sleep or shutdown
3628 and fails to release it in time, we will now log its
3629 identity. This makes it easier to identify processes that
3630 cause slow suspends or power-offs.
3632 * When parsing /etc/crypttab, support for a new key-slot=
3633 option as supported by Debian is added. It allows indicating
3634 which LUKS slot to use on disk, speeding up key loading.
3636 * The sd_journald_sendv() API call has been checked and
3637 officially declared to be async-signal-safe so that it may
3638 be invoked from signal handlers for logging purposes.
3640 * Boot-time status output is now enabled automatically after a
3641 short timeout if boot does not progress, in order to give
3642 the user an indication what she or he is waiting for.
3644 * The boot-time output has been improved to show how much time
3645 remains until jobs expire.
3647 * The KillMode= switch in service units gained a new possible
3648 value "mixed". If set, and the unit is shut down, then the
3649 initial SIGTERM signal is sent only to the main daemon
3650 process, while the following SIGKILL signal is sent to
3651 all remaining processes of the service.
3653 * When a scope unit is registered, a new property "Controller"
3654 may be set. If set to a valid bus name, systemd will send a
3655 RequestStop() signal to this name when it would like to shut
3656 down the scope. This may be used to hook manager logic into
3657 the shutdown logic of scope units. Also, scope units may now
3658 be put in a special "abandoned" state, in which case the
3659 manager process which created them takes no further
3660 responsibilities for it.
3662 * When reading unit files, systemd will now verify
3663 the access mode of these files, and warn about certain
3664 suspicious combinations. This has been added to make it
3665 easier to track down packaging bugs where unit files are
3666 marked executable or world-writable.
3668 * systemd-nspawn gained a new "--setenv=" switch to set
3669 container-wide environment variables. The similar option in
3670 systemd-activate was renamed from "--environment=" to
3671 "--setenv=" for consistency.
3673 * systemd-nspawn has been updated to create a new kdbus domain
3674 for each container that is invoked, thus allowing each
3675 container to have its own set of system and user buses,
3676 independent of the host.
3678 * systemd-nspawn gained a new --drop-capability= switch to run
3679 the container with less capabilities than the default. Both
3680 --drop-capability= and --capability= now take the special
3681 string "all" for dropping or keeping all capabilities.
3683 * systemd-nspawn gained new switches for executing containers
3684 with specific SELinux labels set.
3686 * systemd-nspawn gained a new --quiet switch to not generate
3687 any additional output but the container's own console
3690 * systemd-nspawn gained a new --share-system switch to run a
3691 container without PID namespacing enabled.
3693 * systemd-nspawn gained a new --register= switch to control
3694 whether the container is registered with systemd-machined or
3695 not. This is useful for containers that do not run full
3696 OS images, but only specific apps.
3698 * systemd-nspawn gained a new --keep-unit which may be used
3699 when invoked as the only program from a service unit, and
3700 results in registration of the unit service itself in
3701 systemd-machined, instead of a newly opened scope unit.
3703 * systemd-nspawn gained a new --network-interface= switch for
3704 moving arbitrary interfaces to the container. The new
3705 --network-veth switch creates a virtual Ethernet connection
3706 between host and container. The new --network-bridge=
3707 switch then allows assigning the host side of this virtual
3708 Ethernet connection to a bridge device.
3710 * systemd-nspawn gained a new --personality= switch for
3711 setting the kernel personality for the container. This is
3712 useful when running a 32-bit container on a 64-bit host. A
3713 similar option Personality= is now also available for service
3716 * logind will now also track a "Desktop" identifier for each
3717 session which encodes the desktop environment of it. This is
3718 useful for desktop environments that want to identify
3719 multiple running sessions of itself easily.
3721 * A new SELinuxContext= setting for service units has been
3722 added that allows setting a specific SELinux execution
3723 context for a service.
3725 * Most systemd client tools will now honour $SYSTEMD_LESS for
3726 settings of the "less" pager. By default, these tools will
3727 override $LESS to allow certain operations to work, such as
3728 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3729 influence this logic.
3731 * systemd's "seccomp" hook-up has been changed to make use of
3732 the libseccomp library instead of using its own
3733 implementation. This has benefits for portability among
3736 * For usage together with SystemCallFilter=, a new
3737 SystemCallErrorNumber= setting has been introduced that
3738 allows configuration of a system error number to be returned
3739 on filtered system calls, instead of immediately killing the
3740 process. Also, SystemCallArchitectures= has been added to
3741 limit access to system calls of a particular architecture
3742 (in order to turn off support for unused secondary
3743 architectures). There is also a global
3744 SystemCallArchitectures= setting in system.conf now to turn
3745 off support for non-native system calls system-wide.
3747 * systemd requires a kernel with a working name_to_handle_at(),
3748 please see the kernel config requirements in the README file.
3750 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3751 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3752 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3753 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3754 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3755 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3756 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3757 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3758 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3759 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3760 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3761 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3762 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3763 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3764 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3765 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3766 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3767 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3768 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3769 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3770 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3771 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3772 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3773 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3775 — Berlin, 2014-02-20
3779 * logind has gained support for facilitating privileged input
3780 and drm device access for unprivileged clients. This work is
3781 useful to allow Wayland display servers (and similar
3782 programs, such as kmscon) to run under the user's ID and
3783 access input and drm devices which are normally
3784 protected. When this is used (and the kernel is new enough)
3785 logind will "mute" IO on the file descriptors passed to
3786 Wayland as long as it is in the background and "unmute" it
3787 if it returns into the foreground. This allows secure
3788 session switching without allowing background sessions to
3789 eavesdrop on input and display data. This also introduces
3790 session switching support if VT support is turned off in the
3791 kernel, and on seats that are not seat0.
3793 * A new kernel command line option luks.options= is understood
3794 now which allows specifying LUKS options for usage for LUKS
3795 encrypted partitions specified with luks.uuid=.
3797 * tmpfiles.d(5) snippets may now use specifier expansion in
3798 path names. More specifically %m, %b, %H, %v, are now
3799 replaced by the local machine id, boot id, hostname, and
3800 kernel version number.
3802 * A new tmpfiles.d(5) command "m" has been introduced which
3803 may be used to change the owner/group/access mode of a file
3804 or directory if it exists, but do nothing if it does not.
3806 * This release removes high-level support for the
3807 MemorySoftLimit= cgroup setting. The underlying kernel
3808 cgroup attribute memory.soft_limit= is currently badly
3809 designed and likely to be removed from the kernel API in its
3810 current form, hence we should not expose it for now.
3812 * The memory.use_hierarchy cgroup attribute is now enabled for
3813 all cgroups systemd creates in the memory cgroup
3814 hierarchy. This option is likely to be come the built-in
3815 default in the kernel anyway, and the non-hierarchical mode
3816 never made much sense in the intrinsically hierarchical
3819 * A new field _SYSTEMD_SLICE= is logged along with all journal
3820 messages containing the slice a message was generated
3821 from. This is useful to allow easy per-customer filtering of
3822 logs among other things.
3824 * systemd-journald will no longer adjust the group of journal
3825 files it creates to the "systemd-journal" group. Instead we
3826 rely on the journal directory to be owned by the
3827 "systemd-journal" group, and its setgid bit set, so that the
3828 kernel file system layer will automatically enforce that
3829 journal files inherit this group assignment. The reason for
3830 this change is that we cannot allow NSS look-ups from
3831 journald which would be necessary to resolve
3832 "systemd-journal" to a numeric GID, because this might
3833 create deadlocks if NSS involves synchronous queries to
3834 other daemons (such as nscd, or sssd) which in turn are
3835 logging clients of journald and might block on it, which
3836 would then dead lock. A tmpfiles.d(5) snippet included in
3837 systemd will make sure the setgid bit and group are
3838 properly set on the journal directory if it exists on every
3839 boot. However, we recommend adjusting it manually after
3840 upgrades too (or from RPM scriptlets), so that the change is
3841 not delayed until next reboot.
3843 * Backlight and random seed files in /var/lib/ have moved into
3844 the /var/lib/systemd/ directory, in order to centralize all
3845 systemd generated files in one directory.
3847 * Boot time performance measurements (as displayed by
3848 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3849 performance information if that's available to determine how
3850 much time BIOS and boot loader initialization required. With
3851 a sufficiently new BIOS you hence no longer need to boot
3852 with Gummiboot to get access to such information.
3854 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3855 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3856 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3857 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3858 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3859 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3860 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3862 — Berlin, 2013-10-02
3866 * The Restart= option for services now understands a new
3867 on-watchdog setting, which will restart the service
3868 automatically if the service stops sending out watchdog keep
3869 alive messages (as configured with WatchdogSec=).
3871 * The getty generator (which is responsible for bringing up a
3872 getty on configured serial consoles) will no longer only
3873 start a getty on the primary kernel console but on all
3874 others, too. This makes the order in which console= is
3875 specified on the kernel command line less important.
3877 * libsystemd-logind gained a new sd_session_get_vt() call to
3878 retrieve the VT number of a session.
3880 * If the option "tries=0" is set for an entry of /etc/crypttab
3881 its passphrase is queried indefinitely instead of any
3882 maximum number of tries.
3884 * If a service with a configure PID file terminates its PID
3885 file will now be removed automatically if it still exists
3886 afterwards. This should put an end to stale PID files.
3888 * systemd-run will now also take relative binary path names
3889 for execution and no longer insists on absolute paths.
3891 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3892 paths that are optionally prefixed with "-" to indicate that
3893 it should not be considered a failure if they do not exist.
3895 * journalctl -o (and similar commands) now understands a new
3896 output mode "short-precise", it is similar to "short" but
3897 shows timestamps with usec accuracy.
3899 * The option "discard" (as known from Debian) is now
3900 synonymous to "allow-discards" in /etc/crypttab. In fact,
3901 "discard" is preferred now (since it is easier to remember
3904 * Some licensing clean-ups were made, so that more code is now
3905 LGPL-2.1 licensed than before.
3907 * A minimal tool to save/restore the display backlight
3908 brightness across reboots has been added. It will store the
3909 backlight setting as late as possible at shutdown, and
3910 restore it as early as possible during reboot.
3912 * A logic to automatically discover and enable home and swap
3913 partitions on GPT disks has been added. With this in place
3914 /etc/fstab becomes optional for many setups as systemd can
3915 discover certain partitions located on the root disk
3916 automatically. Home partitions are recognized under their
3917 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3918 partitions are recognized under their GPT type ID
3919 0657fd6da4ab43c484e50933c84b4f4f.
3921 * systemd will no longer pass any environment from the kernel
3922 or initrd to system services. If you want to set an
3923 environment for all services, do so via the kernel command
3924 line systemd.setenv= assignment.
3926 * The systemd-sysctl tool no longer natively reads the file
3927 /etc/sysctl.conf. If desired, the file should be symlinked
3928 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3929 legacy support by a symlink rather than built-in code, it
3930 also makes the otherwise hidden order of application of the
3931 different files visible. (Note that this partly reverts to a
3932 pre-198 application order of sysctl knobs!)
3934 * The "systemctl set-log-level" and "systemctl dump" commands
3935 have been moved to systemd-analyze.
3937 * systemd-run learned the new --remain-after-exit switch,
3938 which causes the scope unit not to be cleaned up
3939 automatically after the process terminated.
3941 * tmpfiles learned a new --exclude-prefix= switch to exclude
3942 certain paths from operation.
3944 * journald will now automatically flush all messages to disk
3945 as soon as a message at the log level CRIT, ALERT or EMERG
3948 Contributions from: Andrew Cook, Brandon Philips, Christian
3949 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3950 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3951 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3952 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3953 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3954 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3955 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3956 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3957 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3958 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3959 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3960 William Giokas, Zbigniew Jędrzejewski-Szmek
3962 — Berlin, 2013-09-13
3966 * The documentation has been updated to cover the various new
3967 concepts introduced with 205.
3969 * Unit files now understand the new %v specifier which
3970 resolves to the kernel version string as returned by "uname
3973 * systemctl now supports filtering the unit list output by
3974 load state, active state and sub state, using the new
3977 * "systemctl status" will now show the results of the
3978 condition checks (like ConditionPathExists= and similar) of
3979 the last start attempts of the unit. They are also logged to
3982 * "journalctl -b" may now be used to look for boot output of a
3983 specific boot. Try "journalctl -b -1" for the previous boot,
3984 but the syntax is substantially more powerful.
3986 * "journalctl --show-cursor" has been added which prints the
3987 cursor string the last shown log line. This may then be used
3988 with the new "journalctl --after-cursor=" switch to continue
3989 browsing logs from that point on.
3991 * "journalctl --force" may now be used to force regeneration
3994 * Creation of "dead" device nodes has been moved from udev
3995 into kmod and tmpfiles. Previously, udev would read the kmod
3996 databases to pre-generate dead device nodes based on meta
3997 information contained in kernel modules, so that these would
3998 be auto-loaded on access rather then at boot. As this
3999 does not really have much to do with the exposing actual
4000 kernel devices to userspace this has always been slightly
4001 alien in the udev codebase. Following the new scheme kmod
4002 will now generate a runtime snippet for tmpfiles from the
4003 module meta information and it now is tmpfiles' job to the
4004 create the nodes. This also allows overriding access and
4005 other parameters for the nodes using the usual tmpfiles
4006 facilities. As side effect this allows us to remove the
4007 CAP_SYS_MKNOD capability bit from udevd entirely.
4009 * logind's device ACLs may now be applied to these "dead"
4010 devices nodes too, thus finally allowing managed access to
4011 devices such as /dev/snd/sequencer whithout loading the
4012 backing module right-away.
4014 * A new RPM macro has been added that may be used to apply
4015 tmpfiles configuration during package installation.
4017 * systemd-detect-virt and ConditionVirtualization= now can
4018 detect User-Mode-Linux machines (UML).
4020 * journald will now implicitly log the effective capabilities
4021 set of processes in the message metadata.
4023 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4025 * The initrd interface has been simplified (more specifically,
4026 support for passing performance data via environment
4027 variables and fsck results via files in /run has been
4028 removed). These features were non-essential, and are
4029 nowadays available in a much nicer way by having systemd in
4030 the initrd serialize its state and have the hosts systemd
4031 deserialize it again.
4033 * The udev "keymap" data files and tools to apply keyboard
4034 specific mappings of scan to key codes, and force-release
4035 scan code lists have been entirely replaced by a udev
4036 "keyboard" builtin and a hwdb data file.
4038 * systemd will now honour the kernel's "quiet" command line
4039 argument also during late shutdown, resulting in a
4040 completely silent shutdown when used.
4042 * There's now an option to control the SO_REUSEPORT socket
4043 option in .socket units.
4045 * Instance units will now automatically get a per-template
4046 subslice of system.slice unless something else is explicitly
4047 configured. For example, instances of sshd@.service will now
4048 implicitly be placed in system-sshd.slice rather than
4049 system.slice as before.
4051 * Test coverage support may now be enabled at build time.
4053 Contributions from: Dave Reisner, Frederic Crozat, Harald
4054 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4055 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4056 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4057 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4058 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4059 Giokas, Zbigniew Jędrzejewski-Szmek
4061 — Berlin, 2013-07-23
4065 * Two new unit types have been introduced:
4067 Scope units are very similar to service units, however, are
4068 created out of pre-existing processes — instead of PID 1
4069 forking off the processes. By using scope units it is
4070 possible for system services and applications to group their
4071 own child processes (worker processes) in a powerful way
4072 which then maybe used to organize them, or kill them
4073 together, or apply resource limits on them.
4075 Slice units may be used to partition system resources in an
4076 hierarchical fashion and then assign other units to them. By
4077 default there are now three slices: system.slice (for all
4078 system services), user.slice (for all user sessions),
4079 machine.slice (for VMs and containers).
4081 Slices and scopes have been introduced primarily in
4082 context of the work to move cgroup handling to a
4083 single-writer scheme, where only PID 1
4084 creates/removes/manages cgroups.
4086 * There's a new concept of "transient" units. In contrast to
4087 normal units these units are created via an API at runtime,
4088 not from configuration from disk. More specifically this
4089 means it is now possible to run arbitrary programs as
4090 independent services, with all execution parameters passed
4091 in via bus APIs rather than read from disk. Transient units
4092 make systemd substantially more dynamic then it ever was,
4093 and useful as a general batch manager.
4095 * logind has been updated to make use of scope and slice units
4096 for managing user sessions. As a user logs in he will get
4097 his own private slice unit, to which all sessions are added
4098 as scope units. We also added support for automatically
4099 adding an instance of user@.service for the user into the
4100 slice. Effectively logind will no longer create cgroup
4101 hierarchies on its own now, it will defer entirely to PID 1
4102 for this by means of scope, service and slice units. Since
4103 user sessions this way become entities managed by PID 1
4104 the output of "systemctl" is now a lot more comprehensive.
4106 * A new mini-daemon "systemd-machined" has been added which
4107 may be used by virtualization managers to register local
4108 VMs/containers. nspawn has been updated accordingly, and
4109 libvirt will be updated shortly. machined will collect a bit
4110 of meta information about the VMs/containers, and assign
4111 them their own scope unit (see above). The collected
4112 meta-data is then made available via the "machinectl" tool,
4113 and exposed in "ps" and similar tools. machined/machinectl
4114 is compile-time optional.
4116 * As discussed earlier, the low-level cgroup configuration
4117 options ControlGroup=, ControlGroupModify=,
4118 ControlGroupPersistent=, ControlGroupAttribute= have been
4119 removed. Please use high-level attribute settings instead as
4120 well as slice units.
4122 * A new bus call SetUnitProperties() has been added to alter
4123 various runtime parameters of a unit. This is primarily
4124 useful to alter cgroup parameters dynamically in a nice way,
4125 but will be extended later on to make more properties
4126 modifiable at runtime. systemctl gained a new set-properties
4127 command that wraps this call.
4129 * A new tool "systemd-run" has been added which can be used to
4130 run arbitrary command lines as transient services or scopes,
4131 while configuring a number of settings via the command
4132 line. This tool is currently very basic, however already
4133 very useful. We plan to extend this tool to even allow
4134 queuing of execution jobs with time triggers from the
4135 command line, similar in fashion to "at".
4137 * nspawn will now inform the user explicitly that kernels with
4138 audit enabled break containers, and suggest the user to turn
4141 * Support for detecting the IMA and AppArmor security
4142 frameworks with ConditionSecurity= has been added.
4144 * journalctl gained a new "-k" switch for showing only kernel
4145 messages, mimicking dmesg output; in addition to "--user"
4146 and "--system" switches for showing only user's own logs
4149 * systemd-delta can now show information about drop-in
4150 snippets extending unit files.
4152 * libsystemd-bus has been substantially updated but is still
4153 not available as public API.
4155 * systemd will now look for the "debug" argument on the kernel
4156 command line and enable debug logging, similar to what
4157 "systemd.log_level=debug" already did before.
4159 * "systemctl set-default", "systemctl get-default" has been
4160 added to configure the default.target symlink, which
4161 controls what to boot into by default.
4163 * "systemctl set-log-level" has been added as a convenient
4164 way to raise and lower systemd logging threshold.
4166 * "systemd-analyze plot" will now show the time the various
4167 generators needed for execution, as well as information
4168 about the unit file loading.
4170 * libsystemd-journal gained a new sd_journal_open_files() call
4171 for opening specific journal files. journactl also gained a
4172 new switch to expose this new functionality. Previously we
4173 only supported opening all files from a directory, or all
4174 files from the system, as opening individual files only is
4175 racy due to journal file rotation.
4177 * systemd gained the new DefaultEnvironment= setting in
4178 /etc/systemd/system.conf to set environment variables for
4181 * If a privileged process logs a journal message with the
4182 OBJECT_PID= field set, then journald will automatically
4183 augment this with additional OBJECT_UID=, OBJECT_GID=,
4184 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4185 system services want to log events about specific client
4186 processes. journactl/systemctl has been updated to make use
4187 of this information if all log messages regarding a specific
4190 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4191 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4192 Reisner, David Coppa, David King, David Strauss, Eelco
4193 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4194 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4195 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4196 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4197 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4198 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4199 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4200 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4201 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4202 Łukasz Stelmach, 장동준
4206 * The Python bindings gained some minimal support for the APIs
4207 exposed by libsystemd-logind.
4209 * ConditionSecurity= gained support for detecting SMACK. Since
4210 this condition already supports SELinux and AppArmor we only
4211 miss IMA for this. Patches welcome!
4213 Contributions from: Karol Lewandowski, Lennart Poettering,
4214 Zbigniew Jędrzejewski-Szmek
4218 * systemd-nspawn will now create /etc/resolv.conf if
4219 necessary, before bind-mounting the host's file onto it.
4221 * systemd-nspawn will now store meta information about a
4222 container on the container's cgroup as extended attribute
4223 fields, including the root directory.
4225 * The cgroup hierarchy has been reworked in many ways. All
4226 objects any of the components systemd creates in the cgroup
4227 tree are now suffixed. More specifically, user sessions are
4228 now placed in cgroups suffixed with ".session", users in
4229 cgroups suffixed with ".user", and nspawn containers in
4230 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4231 names are now escaped in a simple scheme to avoid collision
4232 of userspace object names with kernel filenames. This work
4233 is preparation for making these objects relocatable in the
4234 cgroup tree, in order to allow easy resource partitioning of
4235 these objects without causing naming conflicts.
4237 * systemctl list-dependencies gained the new switches
4238 --plain, --reverse, --after and --before.
4240 * systemd-inhibit now shows the process name of processes that
4241 have taken an inhibitor lock.
4243 * nss-myhostname will now also resolve "localhost"
4244 implicitly. This makes /etc/hosts an optional file and
4245 nicely handles that on IPv6 ::1 maps to both "localhost" and
4248 * libsystemd-logind.so gained a new call
4249 sd_get_machine_names() to enumerate running containers and
4250 VMs (currently only supported by very new libvirt and
4251 nspawn). sd_login_monitor can now be used to watch
4252 VMs/containers coming and going.
4254 * .include is not allowed recursively anymore, and only in
4255 unit files. Usually it is better to use drop-in snippets in
4256 .d/*.conf anyway, as introduced with systemd 198.
4258 * systemd-analyze gained a new "critical-chain" command that
4259 determines the slowest chain of units run during system
4260 boot-up. It is very useful for tracking down where
4261 optimizing boot time is the most beneficial.
4263 * systemd will no longer allow manipulating service paths in
4264 the name=systemd:/system cgroup tree using ControlGroup= in
4265 units. (But is still fine with it in all other dirs.)
4267 * There's a new systemd-nspawn@.service service file that may
4268 be used to easily run nspawn containers as system
4269 services. With the container's root directory in
4270 /var/lib/container/foobar it is now sufficient to run
4271 "systemctl start systemd-nspawn@foobar.service" to boot it.
4273 * systemd-cgls gained a new parameter "--machine" to list only
4274 the processes within a certain container.
4276 * ConditionSecurity= now can check for "apparmor". We still
4277 are lacking checks for SMACK and IMA for this condition
4278 check though. Patches welcome!
4280 * A new configuration file /etc/systemd/sleep.conf has been
4281 added that may be used to configure which kernel operation
4282 systemd is supposed to execute when "suspend", "hibernate"
4283 or "hybrid-sleep" is requested. This makes the new kernel
4284 "freeze" state accessible to the user.
4286 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4287 the passed argument if applicable.
4289 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4290 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4291 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4292 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4293 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4294 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4295 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4300 * The output of 'systemctl list-jobs' got some polishing. The
4301 '--type=' argument may now be passed more than once. A new
4302 command 'systemctl list-sockets' has been added which shows
4303 a list of kernel sockets systemd is listening on with the
4304 socket units they belong to, plus the units these socket
4307 * The experimental libsystemd-bus library got substantial
4308 updates to work in conjunction with the (also experimental)
4309 kdbus kernel project. It works well enough to exchange
4310 messages with some sophistication. Note that kdbus is not
4311 ready yet, and the library is mostly an elaborate test case
4312 for now, and not installable.
4314 * systemd gained a new unit 'systemd-static-nodes.service'
4315 that generates static device nodes earlier during boot, and
4316 can run in conjunction with udev.
4318 * libsystemd-login gained a new call sd_pid_get_user_unit()
4319 to retrieve the user systemd unit a process is running
4320 in. This is useful for systems where systemd is used as
4323 * systemd-nspawn now places all containers in the new /machine
4324 top-level cgroup directory in the name=systemd
4325 hierarchy. libvirt will soon do the same, so that we get a
4326 uniform separation of /system, /user and /machine for system
4327 services, user processes and containers/virtual
4328 machines. This new cgroup hierarchy is also useful to stick
4329 stable names to specific container instances, which can be
4330 recognized later this way (this name may be controlled
4331 via systemd-nspawn's new -M switch). libsystemd-login also
4332 gained a new call sd_pid_get_machine_name() to retrieve the
4333 name of the container/VM a specific process belongs to.
4335 * bootchart can now store its data in the journal.
4337 * libsystemd-journal gained a new call
4338 sd_journal_add_conjunction() for AND expressions to the
4339 matching logic. This can be used to express more complex
4340 logical expressions.
4342 * journactl can now take multiple --unit= and --user-unit=
4345 * The cryptsetup logic now understands the "luks.key=" kernel
4346 command line switch for specifying a file to read the
4347 decryption key from. Also, if a configured key file is not
4348 found the tool will now automatically fall back to prompting
4351 * Python systemd.journal module was updated to wrap recently
4352 added functions from libsystemd-journal. The interface was
4353 changed to bring the low level interface in s.j._Reader
4354 closer to the C API, and the high level interface in
4355 s.j.Reader was updated to wrap and convert all data about
4358 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4359 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4360 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4361 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4362 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4363 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4367 * journalctl --update-catalog now understands a new --root=
4368 option to operate on catalogs found in a different root
4371 * During shutdown after systemd has terminated all running
4372 services a final killing loop kills all remaining left-over
4373 processes. We will now print the name of these processes
4374 when we send SIGKILL to them, since this usually indicates a
4377 * If /etc/crypttab refers to password files stored on
4378 configured mount points automatic dependencies will now be
4379 generated to ensure the specific mount is established first
4380 before the key file is attempted to be read.
4382 * 'systemctl status' will now show information about the
4383 network sockets a socket unit is listening on.
4385 * 'systemctl status' will also shown information about any
4386 drop-in configuration file for units. (Drop-In configuration
4387 files in this context are files such as
4388 /etc/systemd/systemd/foobar.service.d/*.conf)
4390 * systemd-cgtop now optionally shows summed up CPU times of
4391 cgroups. Press '%' while running cgtop to switch between
4392 percentage and absolute mode. This is useful to determine
4393 which cgroups use up the most CPU time over the entire
4394 runtime of the system. systemd-cgtop has also been updated
4395 to be 'pipeable' for processing with further shell tools.
4397 * 'hostnamectl set-hostname' will now allow setting of FQDN
4400 * The formatting and parsing of time span values has been
4401 changed. The parser now understands fractional expressions
4402 such as "5.5h". The formatter will now output fractional
4403 expressions for all time spans under 1min, i.e. "5.123456s"
4404 rather than "5s 123ms 456us". For time spans under 1s
4405 millisecond values are shown, for those under 1ms
4406 microsecond values are shown. This should greatly improve
4407 all time-related output of systemd.
4409 * libsystemd-login and libsystemd-journal gained new
4410 functions for querying the poll() events mask and poll()
4411 timeout value for integration into arbitrary event
4414 * localectl gained the ability to list available X11 keymaps
4415 (models, layouts, variants, options).
4417 * 'systemd-analyze dot' gained the ability to filter for
4418 specific units via shell-style globs, to create smaller,
4419 more useful graphs. I.e. it is now possible to create simple
4420 graphs of all the dependencies between only target units, or
4421 of all units that Avahi has dependencies with.
4423 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4424 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4425 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4426 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4427 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4428 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4429 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4433 * The boot-time readahead implementation for rotating media
4434 will now read the read-ahead data in multiple passes which
4435 consist of all read requests made in equidistant time
4436 intervals. This means instead of strictly reading read-ahead
4437 data in its physical order on disk we now try to find a
4438 middle ground between physical and access time order.
4440 * /etc/os-release files gained a new BUILD_ID= field for usage
4441 on operating systems that provide continuous builds of OS
4444 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4445 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4446 William Douglas, Zbigniew Jędrzejewski-Szmek
4450 * systemd-python gained an API exposing libsystemd-daemon.
4452 * The SMACK setup logic gained support for uploading CIPSO
4455 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4456 ReadOnlyDirectories= and InaccessibleDirectories= has
4457 changed. The private /tmp and /var/tmp directories are now
4458 shared by all processes of a service (which means
4459 ExecStartPre= may now leave data in /tmp that ExecStart= of
4460 the same service can still access). When a service is
4461 stopped its temporary directories are immediately deleted
4462 (normal clean-up with tmpfiles is still done in addition to
4465 * By default, systemd will now set a couple of sysctl
4466 variables in the kernel: the safe sysrq options are turned
4467 on, IP route verification is turned on, and source routing
4468 disabled. The recently added hardlink and softlink
4469 protection of the kernel is turned on. These settings should
4470 be reasonably safe, and good defaults for all new systems.
4472 * The predictable network naming logic may now be turned off
4473 with a new kernel command line switch: net.ifnames=0.
4475 * A new libsystemd-bus module has been added that implements a
4476 pretty complete D-Bus client library. For details see:
4478 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4480 * journald will now explicitly flush the journal files to disk
4481 at the latest 5min after each write. The file will then also
4482 be marked offline until the next write. This should increase
4483 reliability in case of a crash. The synchronization delay
4484 can be configured via SyncIntervalSec= in journald.conf.
4486 * There's a new remote-fs-setup.target unit that can be used
4487 to pull in specific services when at least one remote file
4488 system is to be mounted.
4490 * There are new targets timers.target and paths.target as
4491 canonical targets to pull user timer and path units in
4492 from. This complements sockets.target with a similar
4493 purpose for socket units.
4495 * libudev gained a new call udev_device_set_attribute_value()
4496 to set sysfs attributes of a device.
4498 * The udev daemon now sets the default number of worker
4499 processes executed in parallel based on the number of available
4500 CPUs instead of the amount of available RAM. This is supposed
4501 to provide a more reliable default and limit a too aggressive
4502 paralellism for setups with 1000s of devices connected.
4504 Contributions from: Auke Kok, Colin Walters, Cristian
4505 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4506 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4507 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4508 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4509 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4510 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4511 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4512 Zbigniew Jędrzejewski-Szmek
4516 * Configuration of unit files may now be extended via drop-in
4517 files without having to edit/override the unit files
4518 themselves. More specifically, if the administrator wants to
4519 change one value for a service file foobar.service he can
4520 now do so by dropping in a configuration snippet into
4521 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4522 will load all these snippets and apply them on top of the
4523 main unit configuration file, possibly extending or
4524 overriding its settings. Using these drop-in snippets is
4525 generally nicer than the two earlier options for changing
4526 unit files locally: copying the files from
4527 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4528 them there; or creating a new file in /etc/systemd/system/
4529 that incorporates the original one via ".include". Drop-in
4530 snippets into these .d/ directories can be placed in any
4531 directory systemd looks for units in, and the usual
4532 overriding semantics between /usr/lib, /etc and /run apply
4535 * Most unit file settings which take lists of items can now be
4536 reset by assigning the empty string to them. For example,
4537 normally, settings such as Environment=FOO=BAR append a new
4538 environment variable assignment to the environment block,
4539 each time they are used. By assigning Environment= the empty
4540 string the environment block can be reset to empty. This is
4541 particularly useful with the .d/*.conf drop-in snippets
4542 mentioned above, since this adds the ability to reset list
4543 settings from vendor unit files via these drop-ins.
4545 * systemctl gained a new "list-dependencies" command for
4546 listing the dependencies of a unit recursively.
4548 * Inhibitors are now honored and listed by "systemctl
4549 suspend", "systemctl poweroff" (and similar) too, not only
4550 GNOME. These commands will also list active sessions by
4553 * Resource limits (as exposed by the various control group
4554 controllers) can now be controlled dynamically at runtime
4555 for all units. More specifically, you can now use a command
4556 like "systemctl set-cgroup-attr foobar.service cpu.shares
4557 2000" to alter the CPU shares a specific service gets. These
4558 settings are stored persistently on disk, and thus allow the
4559 administrator to easily adjust the resource usage of
4560 services with a few simple commands. This dynamic resource
4561 management logic is also available to other programs via the
4562 bus. Almost any kernel cgroup attribute and controller is
4565 * systemd-vconsole-setup will now copy all font settings to
4566 all allocated VTs, where it previously applied them only to
4569 * libsystemd-login gained the new sd_session_get_tty() API
4572 * This release drops support for a few legacy or
4573 distribution-specific LSB facility names when parsing init
4574 scripts: $x-display-manager, $mail-transfer-agent,
4575 $mail-transport-agent, $mail-transfer-agent, $smtp,
4576 $null. Also, the mail-transfer-agent.target unit backing
4577 this has been removed. Distributions which want to retain
4578 compatibility with this should carry the burden for
4579 supporting this themselves and patch support for these back
4580 in, if they really need to. Also, the facilities $syslog and
4581 $local_fs are now ignored, since systemd does not support
4582 early-boot LSB init scripts anymore, and these facilities
4583 are implied anyway for normal services. syslog.target has
4586 * There are new bus calls on PID1's Manager object for
4587 cancelling jobs, and removing snapshot units. Previously,
4588 both calls were only available on the Job and Snapshot
4591 * systemd-journal-gatewayd gained SSL support.
4593 * The various "environment" files, such as /etc/locale.conf
4594 now support continuation lines with a backslash ("\") as
4595 last character in the line, similarly in style (but different)
4596 to how this is supported in shells.
4598 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4599 now implicitly appended to every log entry logged. systemctl
4600 has been updated to filter by this field when operating on a
4601 user systemd instance.
4603 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4604 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4605 the container. This makes it easier to boot unmodified
4606 Fedora systems in a container, which however still requires
4607 audit=0 to be passed on the kernel command line. Auditing in
4608 kernel and userspace is unfortunately still too broken in
4609 context of containers, hence we recommend compiling it out
4610 of the kernel or using audit=0. Hopefully this will be fixed
4611 one day for good in the kernel.
4613 * nspawn gained the new --bind= and --bind-ro= parameters to
4614 bind mount specific directories from the host into the
4617 * nspawn will now mount its own devpts file system instance
4618 into the container, in order not to leak pty devices from
4619 the host into the container.
4621 * systemd will now read the firmware boot time performance
4622 information from the EFI variables, if the used boot loader
4623 supports this, and takes it into account for boot performance
4624 analysis via "systemd-analyze". This is currently supported
4625 only in conjunction with Gummiboot, but could be supported
4626 by other boot loaders too. For details see:
4628 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4630 * A new generator has been added that automatically mounts the
4631 EFI System Partition (ESP) to /boot, if that directory
4632 exists, is empty, and no other file system has been
4633 configured to be mounted there.
4635 * logind will now send out PrepareForSleep(false) out
4636 unconditionally, after coming back from suspend. This may be
4637 used by applications as asynchronous notification for
4638 system resume events.
4640 * "systemctl unlock-sessions" has been added, that allows
4641 unlocking the screens of all user sessions at once, similar
4642 to how "systemctl lock-sessions" already locked all users
4643 sessions. This is backed by a new D-Bus call UnlockSessions().
4645 * "loginctl seat-status" will now show the master device of a
4646 seat. (i.e. the device of a seat that needs to be around for
4647 the seat to be considered available, usually the graphics
4650 * tmpfiles gained a new "X" line type, that allows
4651 configuration of files and directories (with wildcards) that
4652 shall be excluded from automatic cleanup ("aging").
4654 * udev default rules set the device node permissions now only
4655 at "add" events, and do not change them any longer with a
4656 later "change" event.
4658 * The log messages for lid events and power/sleep keypresses
4659 now carry a message ID.
4661 * We now have a substantially larger unit test suite, but this
4662 continues to be work in progress.
4664 * udevadm hwdb gained a new --root= parameter to change the
4665 root directory to operate relative to.
4667 * logind will now issue a background sync() request to the kernel
4668 early at shutdown, so that dirty buffers are flushed to disk early
4669 instead of at the last moment, in order to optimize shutdown
4672 * A new bootctl tool has been added that is an interface for
4673 certain boot loader operations. This is currently a preview
4674 and is likely to be extended into a small mechanism daemon
4675 like timedated, localed, hostnamed, and can be used by
4676 graphical UIs to enumerate available boot options, and
4677 request boot into firmware operations.
4679 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4680 the rest of the package. It also has been updated to work
4681 correctly in initrds.
4683 * Policykit previously has been runtime optional, and is now
4684 also compile time optional via a configure switch.
4686 * systemd-analyze has been reimplemented in C. Also "systemctl
4687 dot" has moved into systemd-analyze.
4689 * "systemctl status" with no further parameters will now print
4690 the status of all active or failed units.
4692 * Operations such as "systemctl start" can now be executed
4693 with a new mode "--irreversible" which may be used to queue
4694 operations that cannot accidentally be reversed by a later
4695 job queuing. This is by default used to make shutdown
4696 requests more robust.
4698 * The Python API of systemd now gained a new module for
4699 reading journal files.
4701 * A new tool kernel-install has been added that can install
4702 kernel images according to the Boot Loader Specification:
4704 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4706 * Boot time console output has been improved to provide
4707 animated boot time output for hanging jobs.
4709 * A new tool systemd-activate has been added which can be used
4710 to test socket activation with, directly from the command
4711 line. This should make it much easier to test and debug
4712 socket activation in daemons.
4714 * journalctl gained a new "--reverse" (or -r) option to show
4715 journal output in reverse order (i.e. newest line first).
4717 * journalctl gained a new "--pager-end" (or -e) option to jump
4718 to immediately jump to the end of the journal in the
4719 pager. This is only supported in conjunction with "less".
4721 * journalctl gained a new "--user-unit=" option, that works
4722 similarly to "--unit=" but filters for user units rather than
4725 * A number of unit files to ease adoption of systemd in
4726 initrds has been added. This moves some minimal logic from
4727 the various initrd implementations into systemd proper.
4729 * The journal files are now owned by a new group
4730 "systemd-journal", which exists specifically to allow access
4731 to the journal, and nothing else. Previously, we used the
4732 "adm" group for that, which however possibly covers more
4733 than just journal/log file access. This new group is now
4734 already used by systemd-journal-gatewayd to ensure this
4735 daemon gets access to the journal files and as little else
4736 as possible. Note that "make install" will also set FS ACLs
4737 up for /var/log/journal to give "adm" and "wheel" read
4738 access to it, in addition to "systemd-journal" which owns
4739 the journal files. We recommend that packaging scripts also
4740 add read access to "adm" + "wheel" to /var/log/journal, and
4741 all existing/future journal files. To normal users and
4742 administrators little changes, however packagers need to
4743 ensure to create the "systemd-journal" system group at
4744 package installation time.
4746 * The systemd-journal-gatewayd now runs as unprivileged user
4747 systemd-journal-gateway:systemd-journal-gateway. Packaging
4748 scripts need to create these system user/group at
4751 * timedated now exposes a new boolean property CanNTP that
4752 indicates whether a local NTP service is available or not.
4754 * systemd-detect-virt will now also detect xen PVs
4756 * The pstore file system is now mounted by default, if it is
4759 * In addition to the SELinux and IMA policies we will now also
4760 load SMACK policies at early boot.
4762 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4763 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4764 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4765 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4766 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4767 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4768 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4769 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4770 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4771 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4772 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4773 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4774 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4775 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4779 * Timer units now support calendar time events in addition to
4780 monotonic time events. That means you can now trigger a unit
4781 based on a calendar time specification such as "Thu,Fri
4782 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4783 or fifth day of any month of the year 2013, given that it is
4784 a thursday or friday. This brings timer event support
4785 considerably closer to cron's capabilities. For details on
4786 the supported calendar time specification language see
4789 * udev now supports a number of different naming policies for
4790 network interfaces for predictable names, and a combination
4791 of these policies is now the default. Please see this wiki
4792 document for details:
4794 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4796 * Auke Kok's bootchart implementation has been added to the
4797 systemd tree. It is an optional component that can graph the
4798 boot in quite some detail. It is one of the best bootchart
4799 implementations around and minimal in its code and
4802 * nss-myhostname has been integrated into the systemd source
4803 tree. nss-myhostname guarantees that the local hostname
4804 always stays resolvable via NSS. It has been a weak
4805 requirement of systemd-hostnamed since a long time, and
4806 since its code is actually trivial we decided to just
4807 include it in systemd's source tree. It can be turned off
4808 with a configure switch.
4810 * The read-ahead logic is now capable of properly detecting
4811 whether a btrfs file system is on SSD or rotating media, in
4812 order to optimize the read-ahead scheme. Previously, it was
4813 only capable of detecting this on traditional file systems
4816 * In udev, additional device properties are now read from the
4817 IAB in addition to the OUI database. Also, Bluetooth company
4818 identities are attached to the devices as well.
4820 * In service files %U may be used as specifier that is
4821 replaced by the configured user name of the service.
4823 * nspawn may now be invoked without a controlling TTY. This
4824 makes it suitable for invocation as its own service. This
4825 may be used to set up a simple containerized server system
4826 using only core OS tools.
4828 * systemd and nspawn can now accept socket file descriptors
4829 when they are started for socket activation. This enables
4830 implementation of socket activated nspawn
4831 containers. i.e. think about autospawning an entire OS image
4832 when the first SSH or HTTP connection is received. We expect
4833 that similar functionality will also be added to libvirt-lxc
4836 * journalctl will now suppress ANSI color codes when
4837 presenting log data.
4839 * systemctl will no longer show control group information for
4840 a unit if a the control group is empty anyway.
4842 * logind can now automatically suspend/hibernate/shutdown the
4845 * /etc/machine-info and hostnamed now also expose the chassis
4846 type of the system. This can be used to determine whether
4847 the local system is a laptop, desktop, handset or
4848 tablet. This information may either be configured by the
4849 user/vendor or is automatically determined from ACPI and DMI
4850 information if possible.
4852 * A number of PolicyKit actions are now bound together with
4853 "imply" rules. This should simplify creating UIs because
4854 many actions will now authenticate similar ones as well.
4856 * Unit files learnt a new condition ConditionACPower= which
4857 may be used to conditionalize a unit depending on whether an
4858 AC power source is connected or not, of whether the system
4859 is running on battery power.
4861 * systemctl gained a new "is-failed" verb that may be used in
4862 shell scripts and suchlike to check whether a specific unit
4863 is in the "failed" state.
4865 * The EnvironmentFile= setting in unit files now supports file
4866 globbing, and can hence be used to easily read a number of
4867 environment files at once.
4869 * systemd will no longer detect and recognize specific
4870 distributions. All distribution-specific #ifdeffery has been
4871 removed, systemd is now fully generic and
4872 distribution-agnostic. Effectively, not too much is lost as
4873 a lot of the code is still accessible via explicit configure
4874 switches. However, support for some distribution specific
4875 legacy configuration file formats has been dropped. We
4876 recommend distributions to simply adopt the configuration
4877 files everybody else uses now and convert the old
4878 configuration from packaging scripts. Most distributions
4879 already did that. If that's not possible or desirable,
4880 distributions are welcome to forward port the specific
4881 pieces of code locally from the git history.
4883 * When logging a message about a unit systemd will now always
4884 log the unit name in the message meta data.
4886 * localectl will now also discover system locale data that is
4887 not stored in locale archives, but directly unpacked.
4889 * logind will no longer unconditionally use framebuffer
4890 devices as seat masters, i.e. as devices that are required
4891 to be existing before a seat is considered preset. Instead,
4892 it will now look for all devices that are tagged as
4893 "seat-master" in udev. By default, framebuffer devices will
4894 be marked as such, but depending on local systems, other
4895 devices might be marked as well. This may be used to
4896 integrate graphics cards using closed source drivers (such
4897 as NVidia ones) more nicely into logind. Note however, that
4898 we recommend using the open source NVidia drivers instead,
4899 and no udev rules for the closed-source drivers will be
4900 shipped from us upstream.
4902 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4903 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4904 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4905 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4906 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4907 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4908 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4909 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4910 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4911 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4912 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4913 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4918 * udev gained support for loading additional device properties
4919 from an indexed database that is keyed by vendor/product IDs
4920 and similar device identifiers. For the beginning this
4921 "hwdb" is populated with data from the well-known PCI and
4922 USB database, but also includes PNP, ACPI and OID data. In
4923 the longer run this indexed database shall grow into
4924 becoming the one central database for non-essential
4925 userspace device metadata. Previously, data from the PCI/USB
4926 database was only attached to select devices, since the
4927 lookup was a relatively expensive operation due to O(n) time
4928 complexity (with n being the number of entries in the
4929 database). Since this is now O(1), we decided to add in this
4930 data for all devices where this is available, by
4931 default. Note that the indexed database needs to be rebuilt
4932 when new data files are installed. To achieve this you need
4933 to update your packaging scripts to invoke "udevadm hwdb
4934 --update" after installation of hwdb data files. For
4935 RPM-based distributions we introduced the new
4936 %udev_hwdb_update macro for this purpose.
4938 * The Journal gained support for the "Message Catalog", an
4939 indexed database to link up additional information with
4940 journal entries. For further details please check:
4942 http://www.freedesktop.org/wiki/Software/systemd/catalog
4944 The indexed message catalog database also needs to be
4945 rebuilt after installation of message catalog files. Use
4946 "journalctl --update-catalog" for this. For RPM-based
4947 distributions we introduced the %journal_catalog_update
4948 macro for this purpose.
4950 * The Python Journal bindings gained support for the standard
4951 Python logging framework.
4953 * The Journal API gained new functions for checking whether
4954 the underlying file system of a journal file is capable of
4955 properly reporting file change notifications, or whether
4956 applications that want to reflect journal changes "live"
4957 need to recheck journal files continuously in appropriate
4960 * It is now possible to set the "age" field for tmpfiles
4961 entries to 0, indicating that files matching this entry
4962 shall always be removed when the directories are cleaned up.
4964 * coredumpctl gained a new "gdb" verb which invokes gdb
4965 right-away on the selected coredump.
4967 * There's now support for "hybrid sleep" on kernels that
4968 support this, in addition to "suspend" and "hibernate". Use
4969 "systemctl hybrid-sleep" to make use of this.
4971 * logind's HandleSuspendKey= setting (and related settings)
4972 now gained support for a new "lock" setting to simply
4973 request the screen lock on all local sessions, instead of
4974 actually executing a suspend or hibernation.
4976 * systemd will now mount the EFI variables file system by
4979 * Socket units now gained support for configuration of the
4980 SMACK security label.
4982 * timedatectl will now output the time of the last and next
4983 daylight saving change.
4985 * We dropped support for various legacy and distro-specific
4986 concepts, such as insserv, early-boot SysV services
4987 (i.e. those for non-standard runlevels such as 'b' or 'S')
4988 or ArchLinux /etc/rc.conf support. We recommend the
4989 distributions who still need support this to either continue
4990 to maintain the necessary patches downstream, or find a
4991 different solution. (Talk to us if you have questions!)
4993 * Various systemd components will now bypass PolicyKit checks
4994 for root and otherwise handle properly if PolicyKit is not
4995 found to be around. This should fix most issues for
4996 PolicyKit-less systems. Quite frankly this should have been
4997 this way since day one. It is absolutely our intention to
4998 make systemd work fine on PolicyKit-less systems, and we
4999 consider it a bug if something does not work as it should if
5000 PolicyKit is not around.
5002 * For embedded systems it is now possible to build udev and
5003 systemd without blkid and/or kmod support.
5005 * "systemctl switch-root" is now capable of switching root
5006 more than once. I.e. in addition to transitions from the
5007 initrd to the host OS it is now possible to transition to
5008 further OS images from the host. This is useful to implement
5009 offline updating tools.
5011 * Various other additions have been made to the RPM macros
5012 shipped with systemd. Use %udev_rules_update() after
5013 installing new udev rules files. %_udevhwdbdir,
5014 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5015 %_sysctldir are now available which resolve to the right
5016 directories for packages to place various data files in.
5018 * journalctl gained the new --full switch (in addition to
5019 --all, to disable ellipsation for long messages.
5021 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5022 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5023 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5024 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5025 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5026 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5027 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5028 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5029 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5033 * journalctl gained new --since= and --until= switches to
5034 filter by time. It also now supports nice filtering for
5035 units via --unit=/-u.
5037 * Type=oneshot services may use ExecReload= and do the
5040 * The journal daemon now supports time-based rotation and
5041 vacuuming, in addition to the usual disk-space based
5044 * The journal will now index the available field values for
5045 each field name. This enables clients to show pretty drop
5046 downs of available match values when filtering. The bash
5047 completion of journalctl has been updated
5048 accordingly. journalctl gained a new switch -F to list all
5049 values a certain field takes in the journal database.
5051 * More service events are now written as structured messages
5052 to the journal, and made recognizable via message IDs.
5054 * The timedated, localed and hostnamed mini-services which
5055 previously only provided support for changing time, locale
5056 and hostname settings from graphical DEs such as GNOME now
5057 also have a minimal (but very useful) text-based client
5058 utility each. This is probably the nicest way to changing
5059 these settings from the command line now, especially since
5060 it lists available options and is fully integrated with bash
5063 * There's now a new tool "systemd-coredumpctl" to list and
5064 extract coredumps from the journal.
5066 * We now install a README each in /var/log/ and
5067 /etc/rc.d/init.d explaining where the system logs and init
5068 scripts went. This hopefully should help folks who go to
5069 that dirs and look into the otherwise now empty void and
5070 scratch their heads.
5072 * When user-services are invoked (by systemd --user) the
5073 $MANAGERPID env var is set to the PID of systemd.
5075 * SIGRTMIN+24 when sent to a --user instance will now result
5076 in immediate termination of systemd.
5078 * gatewayd received numerous feature additions such as a
5079 "follow" mode, for live syncing and filtering.
5081 * browse.html now allows filtering and showing detailed
5082 information on specific entries. Keyboard navigation and
5083 mouse screen support has been added.
5085 * gatewayd/journalctl now supports HTML5/JSON
5086 Server-Sent-Events as output.
5088 * The SysV init script compatibility logic will now
5089 heuristically determine whether a script supports the
5090 "reload" verb, and only then make this available as
5093 * "systemctl status --follow" has been removed, use "journalctl
5096 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5097 have been removed since they are hardly useful to be
5100 * And I'd like to take the opportunity to specifically mention
5101 Zbigniew for his great contributions. Zbigniew, you rock!
5103 Contributions from: Andrew Eikum, Christian Hesse, Colin
5104 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5105 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5106 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5107 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5108 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5109 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5113 * If /etc/vconsole.conf is non-existent or empty we will no
5114 longer load any console font or key map at boot by
5115 default. Instead the kernel defaults will be left
5116 intact. This is definitely the right thing to do, as no
5117 configuration should mean no configuration, and hard-coding
5118 font names that are different on all archs is probably a bad
5119 idea. Also, the kernel default key map and font should be
5120 good enough for most cases anyway, and mostly identical to
5121 the userspace fonts/key maps we previously overloaded them
5122 with. If distributions want to continue to default to a
5123 non-kernel font or key map they should ship a default
5124 /etc/vconsole.conf with the appropriate contents.
5126 Contributions from: Colin Walters, Daniel J Walsh, Dave
5127 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5128 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5132 * journalctl gained a new --cursor= switch to show entries
5133 starting from the specified location in the journal.
5135 * We now enforce a size limit on journal entry fields exported
5136 with "-o json" in journalctl. Fields larger than 4K will be
5137 assigned null. This can be turned off with --all.
5139 * An (optional) journal gateway daemon is now available as
5140 "systemd-journal-gatewayd.service". This service provides
5141 access to the journal via HTTP and JSON. This functionality
5142 will be used to implement live log synchronization in both
5143 pull and push modes, but has various other users too, such
5144 as easy log access for debugging of embedded devices. Right
5145 now it is already useful to retrieve the journal via HTTP:
5147 # systemctl start systemd-journal-gatewayd.service
5148 # wget http://localhost:19531/entries
5150 This will download the journal contents in a
5151 /var/log/messages compatible format. The same as JSON:
5153 # curl -H"Accept: application/json" http://localhost:19531/entries
5155 This service is also accessible via a web browser where a
5156 single static HTML5 app is served that uses the JSON logic
5157 to enable the user to do some basic browsing of the
5158 journal. This will be extended later on. Here's an example
5159 screenshot of this app in its current state:
5161 http://0pointer.de/public/journal-gatewayd
5163 Contributions from: Kay Sievers, Lennart Poettering, Robert
5164 Milasan, Tom Gundersen
5168 * The bash completion logic is now available for journalctl
5171 * We do not mount the "cpuset" controller anymore together with
5172 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5173 started if no parameters are assigned to it. "cpuset" hence
5174 broke code that assumed it it could create "cpu" groups and
5177 * journalctl -f will now subscribe to terminal size changes,
5178 and line break accordingly.
5180 Contributions from: Dave Reisner, Kay Sievers, Lennart
5181 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5185 * nspawn will now create a symlink /etc/localtime in the
5186 container environment, copying the host's timezone
5187 setting. Previously this has been done via a bind mount, but
5188 since symlinks cannot be bind mounted this has now been
5189 changed to create/update the appropriate symlink.
5191 * journalctl -n's line number argument is now optional, and
5192 will default to 10 if omitted.
5194 * journald will now log the maximum size the journal files may
5195 take up on disk. This is particularly useful if the default
5196 built-in logic of determining this parameter from the file
5197 system size is used. Use "systemctl status
5198 systemd-journald.service" to see this information.
5200 * The multi-seat X wrapper tool has been stripped down. As X
5201 is now capable of enumerating graphics devices via udev in a
5202 seat-aware way the wrapper is not strictly necessary
5203 anymore. A stripped down temporary stop-gap is still shipped
5204 until the upstream display managers have been updated to
5205 fully support the new X logic. Expect this wrapper to be
5206 removed entirely in one of the next releases.
5208 * HandleSleepKey= in logind.conf has been split up into
5209 HandleSuspendKey= and HandleHibernateKey=. The old setting
5210 is not available anymore. X11 and the kernel are
5211 distinguishing between these keys and we should too. This
5212 also means the inhibition lock for these keys has been split
5215 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5216 Poettering, Lukas Nykryn, Václav Pavlín
5220 * Whenever a unit changes state we will now log this to the
5221 journal and show along the unit's own log output in
5224 * ConditionPathIsMountPoint= can now properly detect bind
5225 mount points too. (Previously, a bind mount of one file
5226 system to another place in the same file system could not be
5227 detected as mount, since they shared struct stat's st_dev
5230 * We will now mount the cgroup controllers cpu, cpuacct,
5231 cpuset and the controllers net_cls, net_prio together by
5234 * nspawn containers will now have a virtualized boot
5235 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5236 over with a randomized ID at container initialization). This
5237 has the effect of making "journalctl -b" do the right thing
5240 * The JSON output journal serialization has been updated not
5241 to generate "endless" list objects anymore, but rather one
5242 JSON object per line. This is more in line how most JSON
5243 parsers expect JSON objects. The new output mode
5244 "json-pretty" has been added to provide similar output, but
5245 neatly aligned for readability by humans.
5247 * We dropped all explicit sync() invocations in the shutdown
5248 code. The kernel does this implicitly anyway in the kernel
5249 reboot() syscall. halt(8)'s -n option is now a compatibility
5252 * We now support virtualized reboot() in containers, as
5253 supported by newer kernels. We will fall back to exit() if
5254 CAP_SYS_REBOOT is not available to the container. Also,
5255 nspawn makes use of this now and will actually reboot the
5256 container if the containerized OS asks for that.
5258 * journalctl will only show local log output by default
5259 now. Use --merge (-m) to show remote log output, too.
5261 * libsystemd-journal gained the new sd_journal_get_usage()
5262 call to determine the current disk usage of all journal
5263 files. This is exposed in the new "journalctl --disk-usage"
5266 * journald gained a new configuration setting SplitMode= in
5267 journald.conf which may be used to control how user journals
5268 are split off. See journald.conf(5) for details.
5270 * A new condition type ConditionFileNotEmpty= has been added.
5272 * tmpfiles' "w" lines now support file globbing, to write
5273 multiple files at once.
5275 * We added Python bindings for the journal submission
5276 APIs. More Python APIs for a number of selected APIs will
5277 likely follow. Note that we intend to add native bindings
5278 only for the Python language, as we consider it common
5279 enough to deserve bindings shipped within systemd. There are
5280 various projects outside of systemd that provide bindings
5281 for languages such as PHP or Lua.
5283 * Many conditions will now resolve specifiers such as %i. In
5284 addition, PathChanged= and related directives of .path units
5285 now support specifiers as well.
5287 * There's now a new RPM macro definition for the system preset
5290 * journald will now warn if it ca not forward a message to the
5291 syslog daemon because its socket is full.
5293 * timedated will no longer write or process /etc/timezone,
5294 except on Debian. As we do not support late mounted /usr
5295 anymore /etc/localtime always being a symlink is now safe,
5296 and hence the information in /etc/timezone is not necessary
5299 * logind will now always reserve one VT for a text getty (VT6
5300 by default). Previously if more than 6 X sessions where
5301 started they took up all the VTs with auto-spawned gettys,
5302 so that no text gettys were available anymore.
5304 * udev will now automatically inform the btrfs kernel logic
5305 about btrfs RAID components showing up. This should make
5306 simple hotplug based btrfs RAID assembly work.
5308 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5309 (but not for its children which will stay at the kernel
5310 default). This should allow setups with a lot more listening
5313 * systemd will now always pass the configured timezone to the
5314 kernel at boot. timedated will do the same when the timezone
5317 * logind's inhibition logic has been updated. By default,
5318 logind will now handle the lid switch, the power and sleep
5319 keys all the time, even in graphical sessions. If DEs want
5320 to handle these events on their own they should take the new
5321 handle-power-key, handle-sleep-key and handle-lid-switch
5322 inhibitors during their runtime. A simple way to achieve
5323 that is to invoke the DE wrapped in an invocation of:
5325 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5327 * Access to unit operations is now checked via SELinux taking
5328 the unit file label and client process label into account.
5330 * systemd will now notify the administrator in the journal
5331 when he over-mounts a non-empty directory.
5333 * There are new specifiers that are resolved in unit files,
5334 for the host name (%H), the machine ID (%m) and the boot ID
5337 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5338 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5339 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5340 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5341 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5342 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5343 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5347 * Support for reading structured kernel messages from
5348 /dev/kmsg has now been added and is enabled by default.
5350 * Support for reading kernel messages from /proc/kmsg has now
5351 been removed. If you want kernel messages in the journal
5352 make sure to run a recent kernel (>= 3.5) that supports
5353 reading structured messages from /dev/kmsg (see
5354 above). /proc/kmsg is now exclusive property of classic
5355 syslog daemons again.
5357 * The libudev API gained the new
5358 udev_device_new_from_device_id() call.
5360 * The logic for file system namespace (ReadOnlyDirectory=,
5361 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5362 require pivot_root() anymore. This means fewer temporary
5363 directories are created below /tmp for this feature.
5365 * nspawn containers will now see and receive all submounts
5366 made on the host OS below the root file system of the
5369 * Forward Secure Sealing is now supported for Journal files,
5370 which provide cryptographical sealing of journal files so
5371 that attackers cannot alter log history anymore without this
5372 being detectable. Lennart will soon post a blog story about
5373 this explaining it in more detail.
5375 * There are two new service settings RestartPreventExitStatus=
5376 and SuccessExitStatus= which allow configuration of exit
5377 status (exit code or signal) which will be excepted from the
5378 restart logic, resp. consider successful.
5380 * journalctl gained the new --verify switch that can be used
5381 to check the integrity of the structure of journal files and
5382 (if Forward Secure Sealing is enabled) the contents of
5385 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5386 and similar symlinks pre-created. This makes running shells
5387 as container init process a lot more fun.
5389 * The fstab support can now handle PARTUUID= and PARTLABEL=
5392 * A new ConditionHost= condition has been added to match
5393 against the hostname (with globs) and machine ID. This is
5394 useful for clusters where a single OS image is used to
5395 provision a large number of hosts which shall run slightly
5396 different sets of services.
5398 * Services which hit the restart limit will now be placed in a
5401 Contributions from: Bertram Poettering, Dave Reisner, Huang
5402 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5403 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5407 * When running in --user mode systemd will now become a
5408 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5409 tree a lot more organized.
5411 * A new PartOf= unit dependency type has been introduced that
5412 may be used to group services in a natural way.
5414 * "systemctl enable" may now be used to enable instances of
5417 * journalctl now prints error log levels in red, and
5418 warning/notice log levels in bright white. It also supports
5419 filtering by log level now.
5421 * cgtop gained a new -n switch (similar to top), to configure
5422 the maximum number of iterations to run for. It also gained
5423 -b, to run in batch mode (accepting no input).
5425 * The suffix ".service" may now be omitted on most systemctl
5426 command lines involving service unit names.
5428 * There's a new bus call in logind to lock all sessions, as
5429 well as a loginctl verb for it "lock-sessions".
5431 * libsystemd-logind.so gained a new call sd_journal_perror()
5432 that works similar to libc perror() but logs to the journal
5433 and encodes structured information about the error number.
5435 * /etc/crypttab entries now understand the new keyfile-size=
5438 * shutdown(8) now can send a (configurable) wall message when
5439 a shutdown is cancelled.
5441 * The mount propagation mode for the root file system will now
5442 default to "shared", which is useful to make containers work
5443 nicely out-of-the-box so that they receive new mounts from
5444 the host. This can be undone locally by running "mount
5445 --make-rprivate /" if needed.
5447 * The prefdm.service file has been removed. Distributions
5448 should maintain this unit downstream if they intend to keep
5449 it around. However, we recommend writing normal unit files
5450 for display managers instead.
5452 * Since systemd is a crucial part of the OS we will now
5453 default to a number of compiler switches that improve
5454 security (hardening) such as read-only relocations, stack
5455 protection, and suchlike.
5457 * The TimeoutSec= setting for services is now split into
5458 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5459 of individual time outs for the start and the stop phase of
5462 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5463 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5464 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5465 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5466 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5467 Gundersen, Zbigniew Jędrzejewski-Szmek
5471 * The journal and id128 C APIs are now fully documented as man
5474 * Extra safety checks have been added when transitioning from
5475 the initial RAM disk to the main system to avoid accidental
5478 * /etc/crypttab entries now understand the new keyfile-offset=
5481 * systemctl -t can now be used to filter by unit load state.
5483 * The journal C API gained the new sd_journal_wait() call to
5484 make writing synchronous journal clients easier.
5486 * journalctl gained the new -D switch to show journals from a
5489 * journalctl now displays a special marker between log
5490 messages of two different boots.
5492 * The journal is now explicitly flushed to /var via a service
5493 systemd-journal-flush.service, rather than implicitly simply
5494 by seeing /var/log/journal to be writable.
5496 * journalctl (and the journal C APIs) can now match for much
5497 more complex expressions, with alternatives and
5500 * When transitioning from the initial RAM disk to the main
5501 system we will now kill all processes in a killing spree to
5502 ensure no processes stay around by accident.
5504 * Three new specifiers may be used in unit files: %u, %h, %s
5505 resolve to the user name, user home directory resp. user
5506 shell. This is useful for running systemd user instances.
5508 * We now automatically rotate journal files if their data
5509 object hash table gets a fill level > 75%. We also size the
5510 hash table based on the configured maximum file size. This
5511 together should lower hash collisions drastically and thus
5512 speed things up a bit.
5514 * journalctl gained the new "--header" switch to introspect
5515 header data of journal files.
5517 * A new setting SystemCallFilters= has been added to services
5518 which may be used to apply blacklists or whitelists to
5519 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5521 * nspawn gained a new --link-journal= switch (and quicker: -j)
5522 to link the container journal with the host. This makes it
5523 very easy to centralize log viewing on the host for all
5524 guests while still keeping the journal files separated.
5526 * Many bugfixes and optimizations
5528 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5529 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5530 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5535 * Several tools now understand kernel command line arguments,
5536 which are only read when run in an initial RAM disk. They
5537 usually follow closely their normal counterparts, but are
5540 * There's a new tool to analyze the readahead files that are
5541 automatically generated at boot. Use:
5543 /usr/lib/systemd/systemd-readahead analyze /.readahead
5545 * We now provide an early debug shell on tty9 if this enabled. Use:
5547 systemctl enable debug-shell.service
5549 * All plymouth related units have been moved into the Plymouth
5550 package. Please make sure to upgrade your Plymouth version
5553 * systemd-tmpfiles now supports getting passed the basename of
5554 a configuration file only, in which case it will look for it
5555 in all appropriate directories automatically.
5557 * udevadm info now takes a /dev or /sys path as argument, and
5558 does the right thing. Example:
5560 udevadm info /dev/sda
5561 udevadm info /sys/class/block/sda
5563 * systemctl now prints a warning if a unit is stopped but a
5564 unit that might trigger it continues to run. Example: a
5565 service is stopped but the socket that activates it is left
5568 * "systemctl status" will now mention if the log output was
5569 shortened due to rotation since a service has been started.
5571 * The journal API now exposes functions to determine the
5572 "cutoff" times due to rotation.
5574 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5575 immediately flushing of runtime logs to /var if possible,
5576 resp. for triggering immediate rotation of the journal
5579 * It is now considered an error if a service is attempted to
5580 be stopped that is not loaded.
5582 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5584 * systemd-analyze now supports Python 3
5586 * tmpfiles now supports cleaning up directories via aging
5587 where the first level dirs are always kept around but
5588 directories beneath it automatically aged. This is enabled
5589 by prefixing the age field with '~'.
5591 * Seat objects now expose CanGraphical, CanTTY properties
5592 which is required to deal with very fast bootups where the
5593 display manager might be running before the graphics drivers
5594 completed initialization.
5596 * Seat objects now expose a State property.
5598 * We now include RPM macros for service enabling/disabling
5599 based on the preset logic. We recommend RPM based
5600 distributions to make use of these macros if possible. This
5601 makes it simpler to reuse RPM spec files across
5604 * We now make sure that the collected systemd unit name is
5605 always valid when services log to the journal via
5608 * There's a new man page kernel-command-line(7) detailing all
5609 command line options we understand.
5611 * The fstab generator may now be disabled at boot by passing
5612 fstab=0 on the kernel command line.
5614 * A new kernel command line option modules-load= is now understood
5615 to load a specific kernel module statically, early at boot.
5617 * Unit names specified on the systemctl command line are now
5618 automatically escaped as needed. Also, if file system or
5619 device paths are specified they are automatically turned
5620 into the appropriate mount or device unit names. Example:
5622 systemctl status /home
5623 systemctl status /dev/sda
5625 * The SysVConsole= configuration option has been removed from
5626 system.conf parsing.
5628 * The SysV search path is no longer exported on the D-Bus
5631 * The Names= option is been removed from unit file parsing.
5633 * There's a new man page bootup(7) detailing the boot process.
5635 * Every unit and every generator we ship with systemd now
5636 comes with full documentation. The self-explanatory boot is
5639 * A couple of services gained "systemd-" prefixes in their
5640 name if they wrap systemd code, rather than only external
5641 code. Among them fsck@.service which is now
5642 systemd-fsck@.service.
5644 * The HaveWatchdog property has been removed from the D-Bus
5647 * systemd.confirm_spawn= on the kernel command line should now
5650 * There's a new man page crypttab(5) which details all options
5651 we actually understand.
5653 * systemd-nspawn gained a new --capability= switch to pass
5654 additional capabilities to the container.
5656 * timedated will now read known NTP implementation unit names
5657 from /usr/lib/systemd/ntp-units.d/*.list,
5658 systemd-timedated-ntp.target has been removed.
5660 * journalctl gained a new switch "-b" that lists log data of
5661 the current boot only.
5663 * The notify socket is in the abstract namespace again, in
5664 order to support daemons which chroot() at start-up.
5666 * There is a new Storage= configuration option for journald
5667 which allows configuration of where log data should go. This
5668 also provides a way to disable journal logging entirely, so
5669 that data collected is only forwarded to the console, the
5670 kernel log buffer or another syslog implementation.
5672 * Many bugfixes and optimizations
5674 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5675 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5676 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5677 Shawn Landden, Tom Gundersen
5681 * "systemctl help <unit>" now shows the man page if one is
5684 * Several new man pages have been added.
5686 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5687 MaxLevelConsole= can now be specified in
5688 journald.conf. These options allow reducing the amount of
5689 data stored on disk or forwarded by the log level.
5691 * TimerSlackNSec= can now be specified in system.conf for
5692 PID1. This allows system-wide power savings.
5694 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5695 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5700 * logind is now capable of (optionally) handling power and
5701 sleep keys as well as the lid switch.
5703 * journalctl now understands the syntax "journalctl
5704 /usr/bin/avahi-daemon" to get all log output of a specific
5707 * CapabilityBoundingSet= in system.conf now also influences
5708 the capability bound set of usermode helpers of the kernel.
5710 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5711 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5712 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5713 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5717 * Note that we skipped 139 releases here in order to set the
5718 new version to something that is greater than both udev's
5719 and systemd's most recent version number.
5721 * udev: all udev sources are merged into the systemd source tree now.
5722 All future udev development will happen in the systemd tree. It
5723 is still fully supported to use the udev daemon and tools without
5724 systemd running, like in initramfs or other init systems. Building
5725 udev though, will require the *build* of the systemd tree, but
5726 udev can be properly *run* without systemd.
5728 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5729 should be used to create dead device nodes as workarounds for broken
5732 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5733 no longer supported. udev_monitor_new_from_netlink() needs to be
5734 used to subscribe to events.
5736 * udev: when udevd is started by systemd, processes which are left
5737 behind by forking them off of udev rules, are unconditionally cleaned
5738 up and killed now after the event handling has finished. Services or
5739 daemons must be started as systemd services. Services can be
5740 pulled-in by udev to get started, but they can no longer be directly
5741 forked by udev rules.
5743 * udev: the daemon binary is called systemd-udevd now and installed
5744 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5745 to adapt to that, create symlink, or rename the binary after building
5748 * libudev no longer provides these symbols:
5749 udev_monitor_from_socket()
5750 udev_queue_get_failed_list_entry()
5751 udev_get_{dev,sys,run}_path()
5752 The versions number was bumped and symbol versioning introduced.
5754 * systemd-loginctl and systemd-journalctl have been renamed
5755 to loginctl and journalctl to match systemctl.
5757 * The config files: /etc/systemd/systemd-logind.conf and
5758 /etc/systemd/systemd-journald.conf have been renamed to
5759 logind.conf and journald.conf. Package updates should rename
5760 the files to the new names on upgrade.
5762 * For almost all files the license is now LGPL2.1+, changed
5763 from the previous GPL2.0+. Exceptions are some minor stuff
5764 of udev (which will be changed to LGPL2.1 eventually, too),
5765 and the MIT licensed sd-daemon.[ch] library that is suitable
5766 to be used as drop-in files.
5768 * systemd and logind now handle system sleep states, in
5769 particular suspending and hibernating.
5771 * logind now implements a sleep/shutdown/idle inhibiting logic
5772 suitable for a variety of uses. Soonishly Lennart will blog
5773 about this in more detail.
5775 * var-run.mount and var-lock.mount are no longer provided
5776 (which prevously bind mounted these directories to their new
5777 places). Distributions which have not converted these
5778 directories to symlinks should consider stealing these files
5779 from git history and add them downstream.
5781 * We introduced the Documentation= field for units and added
5782 this to all our shipped units. This is useful to make it
5783 easier to explore the boot and the purpose of the various
5786 * All smaller setup units (such as
5787 systemd-vconsole-setup.service) now detect properly if they
5788 are run in a container and are skipped when
5789 appropriate. This guarantees an entirely noise-free boot in
5790 Linux container environments such as systemd-nspawn.
5792 * A framework for implementing offline system updates is now
5793 integrated, for details see:
5794 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5796 * A new service type Type=idle is available now which helps us
5797 avoiding ugly interleaving of getty output and boot status
5800 * There's now a system-wide CapabilityBoundingSet= option to
5801 globally reduce the set of capabilities for the
5802 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5803 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5804 even CAP_NET_ADMIN system-wide for secure systems.
5806 * There are now system-wide DefaultLimitXXX= options to
5807 globally change the defaults of the various resource limits
5808 for all units started by PID 1.
5810 * Harald Hoyer's systemd test suite has been integrated into
5811 systemd which allows easy testing of systemd builds in qemu
5812 and nspawn. (This is really awesome! Ask us for details!)
5814 * The fstab parser is now implemented as generator, not inside
5817 * systemctl will now warn you if .mount units generated from
5818 /etc/fstab are out of date due to changes in fstab that
5819 have not been read by systemd yet.
5821 * systemd is now suitable for usage in initrds. Dracut has
5822 already been updated to make use of this. With this in place
5823 initrds get a slight bit faster but primarily are much
5824 easier to introspect and debug since "systemctl status" in
5825 the host system can be used to introspect initrd services,
5826 and the journal from the initrd is kept around too.
5828 * systemd-delta has been added, a tool to explore differences
5829 between user/admin configuration and vendor defaults.
5831 * PrivateTmp= now affects both /tmp and /var/tmp.
5833 * Boot time status messages are now much prettier and feature
5834 proper english language. Booting up systemd has never been
5837 * Read-ahead pack files now include the inode number of all
5838 files to pre-cache. When the inode changes the pre-caching
5839 is not attempted. This should be nicer to deal with updated
5840 packages which might result in changes of read-ahead
5843 * We now temporaritly lower the kernel's read_ahead_kb variable
5844 when collecting read-ahead data to ensure the kernel's
5845 built-in read-ahead does not add noise to our measurements
5846 of necessary blocks to pre-cache.
5848 * There's now RequiresMountsFor= to add automatic dependencies
5849 for all mounts necessary for a specific file system path.
5851 * MountAuto= and SwapAuto= have been removed from
5852 system.conf. Mounting file systems at boot has to take place
5855 * nspawn now learned a new switch --uuid= to set the machine
5856 ID on the command line.
5858 * nspawn now learned the -b switch to automatically search
5861 * vt102 is now the default TERM for serial TTYs, upgraded from
5864 * systemd-logind now works on VT-less systems.
5866 * The build tree has been reorganized. The individual
5867 components now have directories of their own.
5869 * A new condition type ConditionPathIsReadWrite= is now available.
5871 * nspawn learned the new -C switch to create cgroups for the
5872 container in other hierarchies.
5874 * We now have support for hardware watchdogs, configurable in
5877 * The scheduled shutdown logic now has a public API.
5879 * We now mount /tmp as tmpfs by default, but this can be
5880 masked and /etc/fstab can override it.
5882 * Since udisks does not make use of /media anymore we are not
5883 mounting a tmpfs on it anymore.
5885 * journalctl gained a new --local switch to only interleave
5886 locally generated journal files.
5888 * We can now load the IMA policy at boot automatically.
5890 * The GTK tools have been split off into a systemd-ui.
5892 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5893 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5894 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5895 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5896 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5897 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5898 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5899 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5900 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5905 * This is mostly a bugfix release
5907 * Support optional initialization of the machine ID from the
5908 KVM or container configured UUID.
5910 * Support immediate reboots with "systemctl reboot -ff"
5912 * Show /etc/os-release data in systemd-analyze output
5914 * Many bugfixes for the journal, including endianness fixes and
5915 ensuring that disk space enforcement works
5917 * sd-login.h is C++ comptaible again
5919 * Extend the /etc/os-release format on request of the Debian
5922 * We now refuse non-UTF8 strings used in various configuration
5923 and unit files. This is done to ensure we do not pass invalid
5924 data over D-Bus or expose it elsewhere.
5926 * Register Mimo USB Screens as suitable for automatic seat
5929 * Read SELinux client context from journal clients in a race
5932 * Reorder configuration file lookup order. /etc now always
5933 overrides /run in order to allow the administrator to always
5934 and unconditionally override vendor-supplied or
5935 automatically generated data.
5937 * The various user visible bits of the journal now have man
5938 pages. We still lack man pages for the journal API calls
5941 * We now ship all man pages in HTML format again in the
5944 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5945 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5946 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5951 * This is mostly a bugfix release
5953 * systems lacking /etc/os-release are no longer supported.
5955 * Various functionality updates to libsystemd-login.so
5957 * Track class of PAM logins to distinguish greeters from
5960 Contributions from: Kay Sievers, Lennart Poettering, Michael
5965 * This is an important bugfix release for v41.
5967 * Building man pages is now optional which should be useful
5968 for those building systemd from git but unwilling to install
5971 * Watchdog support for supervising services is now usable. In
5972 a future release support for hardware watchdogs
5973 (i.e. /dev/watchdog) will be added building on this.
5975 * Service start rate limiting is now configurable and can be
5976 turned off per service. When a start rate limit is hit a
5977 reboot can automatically be triggered.
5979 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5981 Contributions from: Benjamin Franzke, Bill Nottingham,
5982 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5983 Schmidt, Michał Górny, Piotr Drąg
5987 * The systemd binary is installed /usr/lib/systemd/systemd now;
5988 An existing /sbin/init symlink needs to be adapted with the
5991 * The code that loads kernel modules has been ported to invoke
5992 libkmod directly, instead of modprobe. This means we do not
5993 support systems with module-init-tools anymore.
5995 * Watchdog support is now already useful, but still not
5998 * A new kernel command line option systemd.setenv= is
5999 understood to set system wide environment variables
6000 dynamically at boot.
6002 * We now limit the set of capabilities of systemd-journald.
6004 * We now set SIGPIPE to ignore by default, since it only is
6005 useful in shell pipelines, and has little use in general
6006 code. This can be disabled with IgnoreSIPIPE=no in unit
6009 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6010 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6015 * This is mostly a bugfix release
6017 * We now expose the reason why a service failed in the
6018 "Result" D-Bus property.
6020 * Rudimentary service watchdog support (will be completed over
6021 the next few releases.)
6023 * When systemd forks off in order execute some service we will
6024 now immediately changes its argv[0] to reflect which process
6025 it will execute. This is useful to minimize the time window
6026 with a generic argv[0], which makes bootcharts more useful
6028 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6029 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6030 Mike Kazantsev, Ray Strode
6034 * This is mostly a test release, but incorporates many
6037 * New systemd-cgtop tool to show control groups by their
6040 * Linking against libacl for ACLs is optional again. If
6041 disabled, support tracking device access for active logins
6042 goes becomes unavailable, and so does access to the user
6043 journals by the respective users.
6045 * If a group "adm" exists, journal files are automatically
6046 owned by them, thus allow members of this group full access
6047 to the system journal as well as all user journals.
6049 * The journal now stores the SELinux context of the logging
6050 client for all entries.
6052 * Add C++ inclusion guards to all public headers
6054 * New output mode "cat" in the journal to print only text
6055 messages, without any meta data like date or time.
6057 * Include tiny X server wrapper as a temporary stop-gap to
6058 teach XOrg udev display enumeration. This is used by display
6059 managers such as gdm, and will go away as soon as XOrg
6060 learned native udev hotplugging for display devices.
6062 * Add new systemd-cat tool for executing arbitrary programs
6063 with STDERR/STDOUT connected to the journal. Can also act as
6064 BSD logger replacement, and does so by default.
6066 * Optionally store all locally generated coredumps in the
6067 journal along with meta data.
6069 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6070 writing short strings to files (for usage for /sys), and for
6071 creating symlinks, character and block device nodes.
6073 * New unit file option ControlGroupPersistent= to make cgroups
6074 persistent, following the mechanisms outlined in
6075 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6077 * Support multiple local RTCs in a sane way
6079 * No longer monopolize IO when replaying readahead data on
6080 rotating disks, since we might starve non-file-system IO to
6081 death, since fanotify() will not see accesses done by blkid,
6084 * Do not show kernel threads in systemd-cgls anymore, unless
6085 requested with new -k switch.
6087 Contributions from: Dan Horák, Kay Sievers, Lennart
6088 Poettering, Michal Schmidt
6092 * This is mostly a test release, but incorporates many
6095 * The git repository moved to:
6096 git://anongit.freedesktop.org/systemd/systemd
6097 ssh://git.freedesktop.org/git/systemd/systemd
6099 * First release with the journal
6100 http://0pointer.de/blog/projects/the-journal.html
6102 * The journal replaces both systemd-kmsg-syslogd and
6103 systemd-stdout-bridge.
6105 * New sd_pid_get_unit() API call in libsystemd-logind
6107 * Many systemadm clean-ups
6109 * Introduce remote-fs-pre.target which is ordered before all
6110 remote mounts and may be used to start services before all
6113 * Added Mageia support
6115 * Add bash completion for systemd-loginctl
6117 * Actively monitor PID file creation for daemons which exit in
6118 the parent process before having finished writing the PID
6119 file in the daemon process. Daemons which do this need to be
6120 fixed (i.e. PID file creation must have finished before the
6121 parent exits), but we now react a bit more gracefully to them.
6123 * Add colourful boot output, mimicking the well-known output
6124 of existing distributions.
6126 * New option PassCredentials= for socket units, for
6127 compatibility with a recent kernel ABI breakage.
6129 * /etc/rc.local is now hooked in via a generator binary, and
6130 thus will no longer act as synchronization point during
6133 * systemctl list-unit-files now supports --root=.
6135 * systemd-tmpfiles now understands two new commands: z, Z for
6136 relabelling files according to the SELinux database. This is
6137 useful to apply SELinux labels to specific files in /sys,
6140 * Output of SysV services is now forwarded to both the console
6141 and the journal by default, not only just the console.
6143 * New man pages for all APIs from libsystemd-login.
6145 * The build tree got reorganized and a the build system is a
6146 lot more modular allowing embedded setups to specifically
6147 select the components of systemd they are interested in.
6149 * Support for Linux systems lacking the kernel VT subsystem is
6152 * configure's --with-rootdir= got renamed to
6153 --with-rootprefix= to follow the naming used by udev and
6156 * Unless specified otherwise we will now install to /usr instead
6157 of /usr/local by default.
6159 * Processes with '@' in argv[0][0] are now excluded from the
6160 final shut-down killing spree, following the logic explained
6162 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6164 * All processes remaining in a service cgroup when we enter
6165 the START or START_PRE states are now killed with
6166 SIGKILL. That means it is no longer possible to spawn
6167 background processes from ExecStart= lines (which was never
6168 supported anyway, and bad style).
6170 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6171 reloading of units together.
6173 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6174 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6175 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6176 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6177 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek