1 systemd System and Service Manager
5 * When using systemd's default tmp.mount for /tmp, this will now be
6 mounted with the "nosuid" and "nodev" options. This avoids
7 privilege escalation attacks that put traps and exploits into /tmp.
8 However, this might cause some problems if you e. g. put container
9 images or overlays into /tmp; if you need this, override tmp.mount's
10 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
13 * systemd-resolved gained a new "Cache=" option in resolved.conf.
14 Local caching makes DNS poisoning attacks slightly easier and allows
15 a local user to detect whether any other user on the same machine has
16 recently visited a given DNS name (privacy). If that is a concern,
17 you can disable local caching with this option at the cost of slower
18 DNS resolution (which is particularly expensive with DNSSEC). The
19 default continues to be "yes" (i. e. caching is enabled).
21 Contributions from: ...
23 — Somewhere, 2016-XX-XX
27 * DNSSEC is now turned on by default in systemd-resolved (in
28 "allow-downgrade" mode), but may be turned off during compile time by
29 passing "--with-default-dnssec=no" to "configure" (and of course,
30 during runtime with DNSSEC= in resolved.conf). We recommend
31 downstreams to leave this on at least during development cycles and
32 report any issues with the DNSSEC logic upstream. We are very
33 interested in collecting feedback about the DNSSEC validator and its
34 limitations in the wild. Note however, that DNSSEC support is
35 probably nothing downstreams should turn on in stable distros just
36 yet, as it might create incompatibilities with a few DNS servers and
37 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
38 automatically whenever we detect such incompatible setups, but there
39 might be systems we do not cover yet. Hence: please help us testing
40 the DNSSEC code, leave this on where you can, report back, but then
41 again don't consider turning this on in your stable, LTS or
42 production release just yet. (Note that you have to enable
43 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
44 and its DNSSEC mode for host name resolution from local
47 * systemd-resolve conveniently resolves DANE records with the --tlsa
48 option and OPENPGPKEY records with the --openpgp option. It also
49 supports dumping raw DNS record data via the new --raw= switch.
51 * systemd-logind will now by default terminate user processes that are
52 part of the user session scope unit (session-XX.scope) when the user
53 logs out. This behavior is controlled by the KillUserProcesses=
54 setting in logind.conf, and the previous default of "no" is now
55 changed to "yes". This means that user sessions will be properly
56 cleaned up after, but additional steps are necessary to allow
57 intentionally long-running processes to survive logout.
59 While the user is logged in at least once, user@.service is running,
60 and any service that should survive the end of any individual login
61 session can be started at a user service or scope using systemd-run.
62 systemd-run(1) man page has been extended with an example which shows
63 how to run screen in a scope unit underneath user@.service. The same
64 command works for tmux.
66 After the user logs out of all sessions, user@.service will be
67 terminated too, by default, unless the user has "lingering" enabled.
68 To effectively allow users to run long-term tasks even if they are
69 logged out, lingering must be enabled for them. See loginctl(1) for
70 details. The default polkit policy was modified to allow users to
71 set lingering for themselves without authentication.
73 Previous defaults can be restored at compile time by the
74 --without-kill-user-processes option to "configure".
76 * systemd-logind gained new configuration settings SessionsMax= and
77 InhibitorsMax=, both with a default of 8192. It will not register new
78 user sessions or inhibitors above this limit.
80 * systemd-logind will now reload configuration on SIGHUP.
82 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
83 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
84 enable. Also, support for the "io" cgroup controller in the unified
85 hierarchy has been added, so that the "memory", "pids" and "io" are
86 now the controllers that are supported on the unified hierarchy.
88 WARNING: it is not possible to use previous systemd versions with
89 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
90 is necessary to also update systemd in the initramfs if using the
91 unified hierarchy. An updated SELinux policy is also required.
93 * LLDP support has been extended, and both passive (receive-only) and
94 active (sender) modes are supported. Passive mode ("routers-only") is
95 enabled by default in systemd-networkd. Active LLDP mode is enabled
96 by default for containers on the internal network. The "networkctl
97 lldp" command may be used to list information gathered. "networkctl
98 status" will also show basic LLDP information on connected peers now.
100 * The IAID and DUID unique identifier sent in DHCP requests may now be
101 configured for the system and each .network file managed by
102 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
104 * systemd-networkd gained support for configuring proxy ARP support for
105 each interface, via the ProxyArp= setting in .network files. It also
106 gained support for configuring the multicast querier feature of
107 bridge devices, via the new MulticastQuerier= setting in .netdev
108 files. Similarly, snooping on the IGMP traffic can be controlled
109 via the new setting MulticastSnooping=.
111 A new setting PreferredLifetime= has been added for addresses
112 configured in .network file to configure the lifetime intended for an
115 The systemd-networkd DHCP server gained the option EmitRouter=, which
116 defaults to yes, to configure whether the DHCP Option 3 (Router)
119 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
120 systemd-socket-activate and installed into /usr/bin. It is now fully
123 * systemd-journald now uses separate threads to flush changes to disk
124 when closing journal files, thus reducing impact of slow disk I/O on
127 * The sd-journal API gained two new calls
128 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
129 can be used to open journal files using file descriptors instead of
130 file or directory paths. sd_journal_open_container() has been
131 deprecated, sd_journal_open_directory_fd() should be used instead
132 with the flag SD_JOURNAL_OS_ROOT.
134 * journalctl learned a new output mode "-o short-unix" that outputs log
135 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
136 UTC). It also gained support for a new --no-hostname setting to
137 suppress the hostname column in the family of "short" output modes.
139 * systemd-ask-password now optionally skips printing of the password to
140 stdout with --no-output which can be useful in scripts.
142 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
143 (devices tagged with ID_MAKER_TOOL) are now tagged with
144 "uaccess" and are available to logged in users.
146 * The DeviceAllow= unit setting now supports specifiers (with "%").
148 * "systemctl show" gained a new --value switch, which allows print a
149 only the contents of a specific unit property, without also printing
150 the property's name. Similar support was added to "show*" verbs
151 of loginctl and machinectl that output "key=value" lists.
153 * A new unit type "generated" was added for files dynamically generated
154 by generator tools. Similarly, a new unit type "transient" is used
155 for unit files created using the runtime API. "systemctl enable" will
156 refuse to operate on such files.
158 * A new command "systemctl revert" has been added that may be used to
159 revert to the vendor version of a unit file, in case local changes
160 have been made by adding drop-ins or overriding the unit file.
162 * "machinectl clean" gained a new verb to automatically remove all or
163 just hidden container images.
165 * systemd-tmpfiles gained support for a new line type "e" for emptying
166 directories, if they exist, without creating them if they don't.
168 * systemd-nspawn gained support for automatically patching the UID/GIDs
169 of the owners and the ACLs of all files and directories in a
170 container tree to match the UID/GID user namespacing range selected
171 for the container invocation. This mode is enabled via the new
172 --private-user-chown switch. It also gained support for automatically
173 choosing a free, previously unused UID/GID range when starting a
174 container, via the new --private-users=pick setting (which implies
175 --private-user-chown). Together, these options for the first time
176 make user namespacing for nspawn containers fully automatic and thus
177 deployable. The systemd-nspawn@.service template unit file has been
178 changed to use this functionality by default.
180 * systemd-nspawn gained a new --network-zone= switch, that allows
181 creating ad-hoc virtual Ethernet links between multiple containers,
182 that only exist as long as at least one container referencing them is
183 running. This allows easy connecting of multiple containers with a
184 common link that implements an Ethernet broadcast domain. Each of
185 these network "zones" may be named relatively freely by the user, and
186 may be referenced by any number of containers, but each container may
187 only reference one of these "zones". On the lower level, this is
188 implemented by an automatically managed bridge network interface for
189 each zone, that is created when the first container referencing its
190 zone is created and removed when the last one referencing its zone
193 * The default start timeout may now be configured on the kernel command
194 line via systemd.default_timeout_start_sec=. It was already
195 configurable via the DefaultTimeoutStartSec= option in
196 /etc/systemd/system.conf.
198 * Socket units gained a new TriggerLimitIntervalSec= and
199 TriggerLimitBurst= setting to configure a limit on the activation
200 rate of the socket unit.
202 * The LimitNICE= setting now optionally takes normal UNIX nice values
203 in addition to the raw integer limit value. If the specified
204 parameter is prefixed with "+" or "-" and is in the range -20..19 the
205 value is understood as UNIX nice value. If not prefixed like this it
206 is understood as raw RLIMIT_NICE limit.
208 * Note that the effect of the PrivateDevices= unit file setting changed
209 slightly with this release: the per-device /dev file system will be
210 mounted read-only from this version on, and will have "noexec"
211 set. This (minor) change of behavior might cause some (exceptional)
212 legacy software to break, when PrivateDevices=yes is set for its
213 service. Please leave PrivateDevices= off if you run into problems
216 * systemd-bootchart has been split out to a separate repository:
217 https://github.com/systemd/systemd-bootchart
219 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
220 merged into the kernel in its current form.
222 * The compatibility libraries libsystemd-daemon.so,
223 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
224 which have been deprecated since systemd-209 have been removed along
225 with the corresponding pkg-config files. All symbols provided by
226 those libraries are provided by libsystemd.so.
228 * The Capabilities= unit file setting has been removed (it is ignored
229 for backwards compatibility). AmbientCapabilities= and
230 CapabilityBoundingSet= should be used instead.
232 * A new special target has been added, initrd-root-device.target,
233 which creates a synchronization point for dependencies of the root
234 device in early userspace. Initramfs builders must ensure that this
235 target is now included in early userspace.
237 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
238 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
239 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
240 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
241 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
242 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
243 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
244 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
245 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
246 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
247 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
248 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
249 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
250 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
251 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
252 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
253 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
254 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
255 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
256 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
257 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
258 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
259 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
260 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
261 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
264 — Fairfax, 2016-05-21
268 * The systemd-resolved DNS resolver service has gained a substantial
269 set of new features, most prominently it may now act as a DNSSEC
270 validating stub resolver. DNSSEC mode is currently turned off by
271 default, but is expected to be turned on by default in one of the
272 next releases. For now, we invite everybody to test the DNSSEC logic
273 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
274 service also gained a full set of D-Bus interfaces, including calls
275 to configure DNS and DNSSEC settings per link (for use by external
276 network management software). systemd-resolved and systemd-networkd
277 now distinguish between "search" and "routing" domains. The former
278 are used to qualify single-label names, the latter are used purely
279 for routing lookups within certain domains to specific links.
280 resolved now also synthesizes RRs for all entries from /etc/hosts.
282 * The systemd-resolve tool (which is a client utility for
283 systemd-resolved) has been improved considerably and is now fully
284 supported and documented. Hence it has moved from /usr/lib/systemd to
287 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
290 * The coredump collection logic has been reworked: when a coredump is
291 collected it is now written to disk, compressed and processed
292 (including stacktrace extraction) from a new instantiated service
293 systemd-coredump@.service, instead of directly from the
294 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
295 processing large coredumps can take up a substantial amount of
296 resources and time, and this previously happened entirely outside of
297 systemd's service supervision. With the new logic the core_pattern
298 hook only does minimal metadata collection before passing off control
299 to the new instantiated service, which is configured with a time
300 limit, a nice level and other settings to minimize negative impact on
301 the rest of the system. Also note that the new logic will honour the
302 RLIMIT_CORE setting of the crashed process, which now allows users
303 and processes to turn off coredumping for their processes by setting
306 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
307 and all forked processes by default. Previously, PID 1 would leave
308 the setting at "0" for all processes, as set by the kernel. Note that
309 the resource limit traditionally has no effect on the generated
310 coredumps on the system if the /proc/sys/kernel/core_pattern hook
311 logic is used. Since the limit is now honoured (see above) its
312 default has been changed so that the coredumping logic is enabled by
313 default for all processes, while allowing specific opt-out.
315 * When the stacktrace is extracted from processes of system users, this
316 is now done as "systemd-coredump" user, in order to sandbox this
317 potentially security sensitive parsing operation. (Note that when
318 processing coredumps of normal users this is done under the user ID
319 of process that crashed, as before.) Packagers should take notice
320 that it is now necessary to create the "systemd-coredump" system user
321 and group at package installation time.
323 * The systemd-activate socket activation testing tool gained support
324 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
325 and --seqpacket switches. It also has been extended to support both
326 new-style and inetd-style file descriptor passing. Use the new
327 --inetd switch to request inetd-style file descriptor passing.
329 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
330 variable, which takes a boolean value. If set to false, ANSI color
331 output is disabled in the tools even when run on a terminal that
334 * The VXLAN support in networkd now supports two new settings
335 DestinationPort= and PortRange=.
337 * A new systemd.machine_id= kernel command line switch has been added,
338 that may be used to set the machine ID in /etc/machine-id if it is
339 not initialized yet. This command line option has no effect if the
340 file is already initialized.
342 * systemd-nspawn gained a new --as-pid2 switch that invokes any
343 specified command line as PID 2 rather than PID 1 in the
344 container. In this mode PID 1 is a minimal stub init process that
345 implements the special POSIX and Linux semantics of PID 1 regarding
346 signal and child process management. Note that this stub init process
347 is implemented in nspawn itself and requires no support from the
348 container image. This new logic is useful to support running
349 arbitrary commands in the container, as normal processes are
350 generally not prepared to run as PID 1.
352 * systemd-nspawn gained a new --chdir= switch for setting the current
353 working directory for the process started in the container.
355 * "journalctl /dev/sda" will now output all kernel log messages for
356 specified device from the current boot, in addition to all devices
357 that are parents of it. This should make log output about devices
358 pretty useful, as long as kernel drivers attach enough metadata to
359 the log messages. (The usual SATA drivers do.)
361 * The sd-journal API gained two new calls
362 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
363 that report whether log data from /run or /var has been found.
365 * journalctl gained a new switch "--fields" that prints all journal
366 record field names currently in use in the journal. This is backed
367 by two new sd-journal API calls sd_journal_enumerate_fields() and
368 sd_journal_restart_fields().
370 * Most configurable timeouts in systemd now expect an argument of
371 "infinity" to turn them off, instead of "0" as before. The semantics
372 from now on is that a timeout of "0" means "now", and "infinity"
373 means "never". To maintain backwards compatibility, "0" continues to
374 turn off previously existing timeout settings.
376 * "systemctl reload-or-try-restart" has been renamed to "systemctl
377 try-reload-or-restart" to clarify what it actually does: the "try"
378 logic applies to both reloading and restarting, not just restarting.
379 The old name continues to be accepted for compatibility.
381 * On boot-up, when PID 1 detects that the system clock is behind the
382 release date of the systemd version in use, the clock is now set
383 to the latter. Previously, this was already done in timesyncd, in order
384 to avoid running with clocks set to the various clock epochs such as
385 1902, 1938 or 1970. With this change the logic is now done in PID 1
386 in addition to timesyncd during early boot-up, so that it is enforced
387 before the first process is spawned by systemd. Note that the logic
388 in timesyncd remains, as it is more comprehensive and ensures
389 clock monotonicity by maintaining a persistent timestamp file in
390 /var. Since /var is generally not available in earliest boot or the
391 initrd, this part of the logic remains in timesyncd, and is not done
394 * Support for tweaking details in net_cls.class_id through the
395 NetClass= configuration directive has been removed, as the kernel
396 people have decided to deprecate that controller in cgroup v2.
397 Userspace tools such as nftables are moving over to setting rules
398 that are specific to the full cgroup path of a task, which obsoletes
399 these controllers anyway. The NetClass= directive is kept around for
400 legacy compatibility reasons. For a more in-depth description of the
401 kernel change, please refer to the respective upstream commit:
403 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
405 * A new service setting RuntimeMaxSec= has been added that may be used
406 to specify a maximum runtime for a service. If the timeout is hit, the
407 service is terminated and put into a failure state.
409 * A new service setting AmbientCapabilities= has been added. It allows
410 configuration of additional Linux process capabilities that are
411 passed to the activated processes. This is only available on very
414 * The process resource limit settings in service units may now be used
415 to configure hard and soft limits individually.
417 * The various libsystemd APIs such as sd-bus or sd-event now publicly
418 expose support for gcc's __attribute__((cleanup())) C extension.
419 Specifically, for many object destructor functions alternative
420 versions have been added that have names suffixed with "p" and take a
421 pointer to a pointer to the object to destroy, instead of just a
422 pointer to the object itself. This is useful because these destructor
423 functions may be used directly as parameters to the cleanup
424 construct. Internally, systemd has been a heavy user of this GCC
425 extension for a long time, and with this change similar support is
426 now available to consumers of the library outside of systemd. Note
427 that by using this extension in your sources compatibility with old
428 and strictly ANSI compatible C compilers is lost. However, all gcc or
429 LLVM versions of recent years support this extension.
431 * Timer units gained support for a new setting RandomizedDelaySec= that
432 allows configuring some additional randomized delay to the configured
433 time. This is useful to spread out timer events to avoid load peaks in
434 clusters or larger setups.
436 * Calendar time specifications now support sub-second accuracy.
438 * Socket units now support listening on SCTP and UDP-lite protocol
441 * The sd-event API now comes with a full set of man pages.
443 * Older versions of systemd contained experimental support for
444 compressing journal files and coredumps with the LZ4 compressor that
445 was not compatible with the lz4 binary (due to API limitations of the
446 lz4 library). This support has been removed; only support for files
447 compatible with the lz4 binary remains. This LZ4 logic is now
448 officially supported and no longer considered experimental.
450 * The dkr image import logic has been removed again from importd. dkr's
451 micro-services focus doesn't fit into the machine image focus of
452 importd, and quickly got out of date with the upstream dkr API.
454 * Creation of the /run/lock/lockdev/ directory was dropped from
455 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
456 been available for many years. If you still need this, you need to
457 create your own tmpfiles.d config file with:
459 d /run/lock/lockdev 0775 root lock -
461 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
462 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
463 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
464 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
465 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
466 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
467 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
468 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
469 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
470 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
471 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
472 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
473 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
474 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
475 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
476 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
477 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
478 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
479 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
485 * A number of properties previously only settable in unit
486 files are now also available as properties to set when
487 creating transient units programmatically via the bus, as it
488 is exposed with systemd-run's --property=
489 setting. Specifically, these are: SyslogIdentifier=,
490 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
491 EnvironmentFile=, ReadWriteDirectories=,
492 ReadOnlyDirectories=, InaccessibleDirectories=,
493 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
495 * When creating transient services via the bus API it is now
496 possible to pass in a set of file descriptors to use as
497 STDIN/STDOUT/STDERR for the invoked process.
499 * Slice units may now be created transiently via the bus APIs,
500 similar to the way service and scope units may already be
503 * Wherever systemd expects a calendar timestamp specification
504 (like in journalctl's --since= and --until= switches) UTC
505 timestamps are now supported. Timestamps suffixed with "UTC"
506 are now considered to be in Universal Time Coordinated
507 instead of the local timezone. Also, timestamps may now
508 optionally be specified with sub-second accuracy. Both of
509 these additions also apply to recurring calendar event
510 specification, such as OnCalendar= in timer units.
512 * journalctl gained a new "--sync" switch that asks the
513 journal daemon to write all so far unwritten log messages to
514 disk and sync the files, before returning.
516 * systemd-tmpfiles learned two new line types "q" and "Q" that
517 operate like "v", but also set up a basic btrfs quota
518 hierarchy when used on a btrfs file system with quota
521 * tmpfiles' "v", "q" and "Q" will now create a plain directory
522 instead of a subvolume (even on a btrfs file system) if the
523 root directory is a plain directory, and not a
524 subvolume. This should simplify things with certain chroot()
525 environments which are not aware of the concept of btrfs
528 * systemd-detect-virt gained a new --chroot switch to detect
529 whether execution takes place in a chroot() environment.
531 * CPUAffinity= now takes CPU index ranges in addition to
534 * The various memory-related resource limit settings (such as
535 LimitAS=) now understand the usual K, M, G, ... suffixes to
536 the base of 1024 (IEC). Similar, the time-related resource
537 limit settings understand the usual min, h, day, ...
540 * There's a new system.conf setting DefaultTasksMax= to
541 control the default TasksMax= setting for services and
542 scopes running on the system. (TasksMax= is the primary
543 setting that exposes the "pids" cgroup controller on systemd
544 and was introduced in the previous systemd release.) The
545 setting now defaults to 512, which means services that are
546 not explicitly configured otherwise will only be able to
547 create 512 processes or threads at maximum, from this
548 version on. Note that this means that thread- or
549 process-heavy services might need to be reconfigured to set
550 TasksMax= to a higher value. It is sufficient to set
551 TasksMax= in these specific unit files to a higher value, or
552 even "infinity". Similar, there's now a logind.conf setting
553 UserTasksMax= that defaults to 4096 and limits the total
554 number of processes or tasks each user may own
555 concurrently. nspawn containers also have the TasksMax=
556 value set by default now, to 8192. Note that all of this
557 only has an effect if the "pids" cgroup controller is
558 enabled in the kernel. The general benefit of these changes
559 should be a more robust and safer system, that provides a
560 certain amount of per-service fork() bomb protection.
562 * systemd-nspawn gained the new --network-veth-extra= switch
563 to define additional and arbitrarily-named virtual Ethernet
564 links between the host and the container.
566 * A new service execution setting PassEnvironment= has been
567 added that allows importing select environment variables
568 from PID1's environment block into the environment block of
571 * Timer units gained support for a new RemainAfterElapse=
572 setting which takes a boolean argument. It defaults on,
573 exposing behaviour unchanged to previous releases. If set to
574 off, timer units are unloaded after they elapsed if they
575 cannot elapse again. This is particularly useful for
576 transient timer units, which shall not stay around longer
577 than until they first elapse.
579 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
580 default now (the kernel default is 16). This is beneficial
581 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
582 allows substantially larger numbers of queued
583 datagrams. This should increase the capability of systemd to
584 parallelize boot-up, as logging and sd_notify() are unlikely
585 to stall execution anymore. If you need to change the value
586 from the new defaults, use the usual sysctl.d/ snippets.
588 * The compression framing format used by the journal or
589 coredump processing has changed to be in line with what the
590 official LZ4 tools generate. LZ4 compression support in
591 systemd was considered unsupported previously, as the format
592 was not compatible with the normal tools. With this release
593 this has changed now, and it is hence safe for downstream
594 distributions to turn it on. While not compressing as well
595 as the XZ, LZ4 is substantially faster, which makes
596 it a good default choice for the compression logic in the
597 journal and in coredump handling.
599 * Any reference to /etc/mtab has been dropped from
600 systemd. The file has been obsolete since a while, but
601 systemd refused to work on systems where it was incorrectly
602 set up (it should be a symlink or non-existent). Please make
603 sure to update to util-linux 2.27.1 or newer in conjunction
604 with this systemd release, which also drops any reference to
605 /etc/mtab. If you maintain a distribution make sure that no
606 software you package still references it, as this is a
607 likely source of bugs. There's also a glibc bug pending,
608 asking for removal of any reference to this obsolete file:
610 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
612 Note that only util-linux versions built with
613 --enable-libmount-force-mountinfo are supported.
615 * Support for the ".snapshot" unit type has been removed. This
616 feature turned out to be little useful and little used, and
617 has now been removed from the core and from systemctl.
619 * The dependency types RequiresOverridable= and
620 RequisiteOverridable= have been removed from systemd. They
621 have been used only very sparingly to our knowledge and
622 other options that provide a similar effect (such as
623 systemctl --mode=ignore-dependencies) are much more useful
624 and commonly used. Moreover, they were only half-way
625 implemented as the option to control behaviour regarding
626 these dependencies was never added to systemctl. By removing
627 these dependency types the execution engine becomes a bit
628 simpler. Unit files that use these dependencies should be
629 changed to use the non-Overridable dependency types
630 instead. In fact, when parsing unit files with these
631 options, that's what systemd will automatically convert them
632 too, but it will also warn, asking users to fix the unit
633 files accordingly. Removal of these dependency types should
634 only affect a negligible number of unit files in the wild.
636 * Behaviour of networkd's IPForward= option changed
637 (again). It will no longer maintain a per-interface setting,
638 but propagate one way from interfaces where this is enabled
639 to the global kernel setting. The global setting will be
640 enabled when requested by a network that is set up, but
641 never be disabled again. This change was made to make sure
642 IPv4 and IPv6 behaviour regarding packet forwarding is
643 similar (as the Linux IPv6 stack does not support
644 per-interface control of this setting) and to minimize
647 * In unit files the behaviour of %u, %U, %h, %s has
648 changed. These specifiers will now unconditionally resolve
649 to the various user database fields of the user that the
650 systemd instance is running as, instead of the user
651 configured in the specific unit via User=. Note that this
652 effectively doesn't change much, as resolving of these
653 specifiers was already turned off in the --system instance
654 of systemd, as we cannot do NSS lookups from PID 1. In the
655 --user instance of systemd these specifiers where correctly
656 resolved, but hardly made any sense, since the user instance
657 lacks privileges to do user switches anyway, and User= is
658 hence useless. Morever, even in the --user instance of
659 systemd behaviour was awkward as it would only take settings
660 from User= assignment placed before the specifier into
661 account. In order to unify and simplify the logic around
662 this the specifiers will now always resolve to the
663 credentials of the user invoking the manager (which in case
664 of PID 1 is the root user).
666 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
667 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
668 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
669 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
670 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
671 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
672 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
673 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
674 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
675 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
676 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
677 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
678 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
679 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
686 * systemd now depends on util-linux v2.27. More specifically,
687 the newly added mount monitor feature in libmount now
688 replaces systemd's former own implementation.
690 * libmount mandates /etc/mtab not to be regular file, and
691 systemd now enforces this condition at early boot.
692 /etc/mtab has been deprecated and warned about for a very
693 long time, so systems running systemd should already have
694 stopped having this file around as anything else than a
695 symlink to /proc/self/mounts.
697 * Support for the "pids" cgroup controller has been added. It
698 allows accounting the number of tasks in a cgroup and
699 enforcing limits on it. This adds two new setting
700 TasksAccounting= and TasksMax= to each unit, as well as a
701 global option DefaultTasksAccounting=.
703 * Support for the "net_cls" cgroup controller has been added.
704 It allows assigning a net class ID to each task in the
705 cgroup, which can then be used in firewall rules and traffic
706 shaping configurations. Note that the kernel netfilter net
707 class code does not currently work reliably for ingress
708 packets on unestablished sockets.
710 This adds a new config directive called NetClass= to CGroup
711 enabled units. Allowed values are positive numbers for fixed
712 assignments and "auto" for picking a free value
715 * 'systemctl is-system-running' now returns 'offline' if the
716 system is not booted with systemd. This command can now be
717 used as a substitute for 'systemd-notify --booted'.
719 * Watchdog timeouts have been increased to 3 minutes for all
720 in-tree service files. Apparently, disk IO issues are more
721 frequent than we hoped, and user reported >1 minute waiting
724 * 'machine-id-commit' functionality has been merged into
725 'machine-id-setup --commit'. The separate binary has been
728 * The WorkingDirectory= directive in unit files may now be set
729 to the special value '~'. In this case, the working
730 directory is set to the home directory of the user
733 * "machinectl shell" will now open the shell in the home
734 directory of the selected user by default.
736 * The CrashChVT= configuration file setting is renamed to
737 CrashChangeVT=, following our usual logic of not
738 abbreviating unnecessarily. The old directive is still
739 supported for compat reasons. Also, this directive now takes
740 an integer value between 1 and 63, or a boolean value. The
741 formerly supported '-1' value for disabling stays around for
744 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
745 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
746 RootDirectory= properties can now be set for transient
749 * The systemd-analyze tool gained a new "set-log-target" verb
750 to change the logging target the system manager logs to
751 dynamically during runtime. This is similar to how
752 "systemd-analyze set-log-level" already changes the log
755 * In nspawn /sys is now mounted as tmpfs, with only a selected
756 set of subdirectories mounted in from the real sysfs. This
757 enhances security slightly, and is useful for ensuring user
758 namespaces work correctly.
760 * Support for USB FunctionFS activation has been added. This
761 allows implementation of USB gadget services that are
762 activated as soon as they are requested, so that they don't
763 have to run continously, similar to classic socket
766 * The "systemctl exit" command now optionally takes an
767 additional parameter that sets the exit code to return from
768 the systemd manager when exiting. This is only relevant when
769 running the systemd user instance, or when running the
770 system instance in a container.
772 * sd-bus gained the new API calls sd_bus_path_encode_many()
773 and sd_bus_path_decode_many() that allow easy encoding and
774 decoding of multiple identifier strings inside a D-Bus
775 object path. Another new call sd_bus_default_flush_close()
776 has been added to flush and close per-thread default
779 * systemd-cgtop gained support for a -M/--machine= switch to
780 show the control groups within a certain container only.
782 * "systemctl kill" gained support for an optional --fail
783 switch. If specified the requested operation will fail of no
784 processes have been killed, because the unit had no
785 processes attached, or similar.
787 * A new systemd.crash_reboot=1 kernel command line option has
788 been added that triggers a reboot after crashing. This can
789 also be set through CrashReboot= in systemd.conf.
791 * The RuntimeDirectory= setting now understands unit
792 specifiers like %i or %f.
794 * A new (still internal) libary API sd-ipv4acd has been added,
795 that implements address conflict detection for IPv4. It's
796 based on code from sd-ipv4ll, and will be useful for
797 detecting DHCP address conflicts.
799 * File descriptors passed during socket activation may now be
800 named. A new API sd_listen_fds_with_names() is added to
801 access the names. The default names may be overridden,
802 either in the .socket file using the FileDescriptorName=
803 parameter, or by passing FDNAME= when storing the file
804 descriptors using sd_notify().
806 * systemd-networkd gained support for:
808 - Setting the IPv6 Router Advertisement settings via
809 IPv6AcceptRouterAdvertisements= in .network files.
811 - Configuring the HelloTimeSec=, MaxAgeSec= and
812 ForwardDelaySec= bridge parameters in .netdev files.
814 - Configuring PreferredSource= for static routes in
817 * The "ask-password" framework used to query for LUKS harddisk
818 passwords or SSL passwords during boot gained support for
819 caching passwords in the kernel keyring, if it is
820 available. This makes sure that the user only has to type in
821 a passphrase once if there are multiple objects to unlock
822 with the same one. Previously, such password caching was
823 available only when Plymouth was used; this moves the
824 caching logic into the systemd codebase itself. The
825 "systemd-ask-password" utility gained a new --keyname=
826 switch to control which kernel keyring key to use for
827 caching a password in. This functionality is also useful for
828 enabling display managers such as gdm to automatically
829 unlock the user's GNOME keyring if its passphrase, the
830 user's password and the harddisk password are the same, if
831 gdm-autologin is used.
833 * When downloading tar or raw images using "machinectl
834 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
835 file is now also downloaded, if it is available and stored
836 next to the image file.
838 * Units of type ".socket" gained a new boolean setting
839 Writable= which is only useful in conjunction with
840 ListenSpecial=. If true, enables opening the specified
841 special file in O_RDWR mode rather than O_RDONLY mode.
843 * systemd-rfkill has been reworked to become a singleton
844 service that is activated through /dev/rfkill on each rfkill
845 state change and saves the settings to disk. This way,
846 systemd-rfkill is now compatible with devices that exist
847 only intermittendly, and even restores state if the previous
848 system shutdown was abrupt rather than clean.
850 * The journal daemon gained support for vacuuming old journal
851 files controlled by the number of files that shall remain,
852 in addition to the already existing control by size and by
853 date. This is useful as journal interleaving performance
854 degrades with too many separate journal files, and allows
855 putting an effective limit on them. The new setting defaults
856 to 100, but this may be changed by setting SystemMaxFiles=
857 and RuntimeMaxFiles= in journald.conf. Also, the
858 "journalctl" tool gained the new --vacuum-files= switch to
859 manually vacuum journal files to leave only the specified
860 number of files in place.
862 * udev will now create /dev/disk/by-path links for ATA devices
863 on kernels where that is supported.
865 * Galician, Serbian, Turkish and Korean translations were added.
867 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
868 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
869 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
870 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
871 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
872 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
873 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
874 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
875 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
876 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
877 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
878 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
879 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
880 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
881 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
882 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
883 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
884 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
890 * The DHCP implementation of systemd-networkd gained a set of
893 - The DHCP server now supports emitting DNS and NTP
894 information. It may be enabled and configured via
895 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
896 and NTP information is enabled, but no servers are
897 configured, the corresponding uplink information (if there
898 is any) is propagated.
900 - Server and client now support transmission and reception
901 of timezone information. It can be configured via the
902 newly introduced network options UseTimezone=,
903 EmitTimezone=, and Timezone=. Transmission of timezone
904 information is enabled between host and containers by
905 default now: the container will change its local timezone
906 to what the host has set.
908 - Lease timeouts can now be configured via
909 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
911 - The DHCP server improved on the stability of
912 leases. Clients are more likely to get the same lease
913 information back, even if the server loses state.
915 - The DHCP server supports two new configuration options to
916 control the lease address pool metrics, PoolOffset= and
919 * The encapsulation limit of tunnels in systemd-networkd may
920 now be configured via 'EncapsulationLimit='. It allows
921 modifying the maximum additional levels of encapsulation
922 that are permitted to be prepended to a packet.
924 * systemd now supports the concept of user buses replacing
925 session buses, if used with dbus-1.10 (and enabled via dbus
926 --enable-user-session). It previously only supported this on
927 kdbus-enabled systems, and this release expands this to
928 'dbus-daemon' systems.
930 * systemd-networkd now supports predictable interface names
933 * systemd now optionally supports the new Linux kernel
934 "unified" control group hierarchy. If enabled via the kernel
935 command-line option 'systemd.unified_cgroup_hierarchy=1',
936 systemd will try to mount the unified cgroup hierarchy
937 directly on /sys/fs/cgroup. If not enabled, or not
938 available, systemd will fall back to the legacy cgroup
939 hierarchy setup, as before. Host system and containers can
940 mix and match legacy and unified hierarchies as they
941 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
942 environment variable to individually select the hierarchy to
943 use for executed containers. By default, nspawn will use the
944 unified hierarchy for the containers if the host uses the
945 unified hierarchy, and the legacy hierarchy otherwise.
946 Please note that at this point the unified hierarchy is an
947 experimental kernel feature and is likely to change in one
948 of the next kernel releases. Therefore, it should not be
949 enabled by default in downstream distributions yet. The
950 minimum required kernel version for the unified hierarchy to
951 work is 4.2. Note that when the unified hierarchy is used
952 for the first time delegated access to controllers is
953 safe. Because of this systemd-nspawn containers will get
954 access to controllers now, as will systemd user
955 sessions. This means containers and user sessions may now
956 manage their own resources, partitioning up what the system
959 * A new special scope unit "init.scope" has been introduced
960 that encapsulates PID 1 of the system. It may be used to
961 determine resource usage and enforce resource limits on PID
962 1 itself. PID 1 hence moved out of the root of the control
965 * The cgtop tool gained support for filtering out kernel
966 threads when counting tasks in a control group. Also, the
967 count of processes is now recursively summed up by
968 default. Two options -k and --recursive= have been added to
969 revert to old behaviour. The tool has also been updated to
970 work correctly in containers now.
972 * systemd-nspawn's --bind= and --bind-ro= options have been
973 extended to allow creation of non-recursive bind mounts.
975 * libsystemd gained two new calls sd_pid_get_cgroup() and
976 sd_peer_get_cgroup() which return the control group path of
977 a process or peer of a connected AF_UNIX socket. This
978 function call is particularly useful when implementing
979 delegated subtrees support in the control group hierarchy.
981 * The "sd-event" event loop API of libsystemd now supports
982 correct dequeuing of real-time signals, without losing
985 * When systemd requests a PolicyKit decision when managing
986 units it will now add additional fields to the request,
987 including unit name and desired operation. This enables more
988 powerful PolicyKit policies, that make decisions depending
991 * nspawn learnt support for .nspawn settings files, that may
992 accompany the image files or directories of containers, and
993 may contain additional settings for the container. This is
994 an alternative to configuring container parameters via the
997 Contributions from: Cristian Rodríguez, Daniel Mack, David
998 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
999 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1000 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1001 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1002 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1003 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1004 Andersen, Tom Gundersen, Torstein Husebø
1006 — Berlin, 2015-09-08
1010 * machinectl gained a new verb 'shell' which opens a fresh
1011 shell on the target container or the host. It is similar to
1012 the existing 'login' command of machinectl, but spawns the
1013 shell directly without prompting for username or
1014 password. The pseudo machine '.host' now refers to the local
1015 host and is used by default. Hence, 'machinectl shell' can
1016 be used as replacement for 'su -' which spawns a session as
1017 a fresh systemd unit in a way that is fully isolated from
1018 the originating session.
1020 * systemd-networkd learned to cope with private-zone DHCP
1021 options and allows other programs to query the values.
1023 * SELinux access control when enabling/disabling units is no
1024 longer enforced with this release. The previous
1025 implementation was incorrect, and a new corrected
1026 implementation is not yet available. As unit file operations
1027 are still protected via PolicyKit and D-Bus policy this is
1028 not a security problem. Yet, distributions which care about
1029 optimal SELinux support should probably not stabilize on
1032 * sd-bus gained support for matches of type "arg0has=", that
1033 test for membership of strings in string arrays sent in bus
1036 * systemd-resolved now dumps the contents of its DNS and LLMNR
1037 caches to the logs on reception of the SIGUSR1 signal. This
1038 is useful to debug DNS behaviour.
1040 * The coredumpctl tool gained a new --directory= option to
1041 operate on journal files in a specific directory.
1043 * "systemctl reboot" and related commands gained a new
1044 "--message=" option which may be used to set a free-text
1045 wall message when shutting down or rebooting the
1046 system. This message is also logged, which is useful for
1047 figuring out the reason for a reboot or shutdown a
1050 * The "systemd-resolve-host" tool's -i switch now takes
1051 network interface numbers as alternative to interface names.
1053 * A new unit file setting for services has been introduced:
1054 UtmpMode= allows configuration of how precisely systemd
1055 handles utmp and wtmp entries for the service if this is
1056 enabled. This allows writing services that appear similar to
1057 user sessions in the output of the "w", "who", "last" and
1060 * systemd-resolved will now locally synthesize DNS resource
1061 records for the "localhost" and "gateway" domains as well as
1062 the local hostname. This should ensure that clients querying
1063 RRs via resolved will get similar results as those going via
1064 NSS, if nss-myhostname is enabled.
1066 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1067 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1068 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1069 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1070 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1071 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1072 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1073 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1074 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1075 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1076 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1077 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1079 — Berlin, 2015-08-27
1083 * The systemd-efi-boot-generator functionality was merged into
1084 systemd-gpt-auto-generator.
1086 * systemd-networkd now supports Group Policy for vxlan
1087 devices. It can be enabled via the new boolean configuration
1088 option called 'GroupPolicyExtension='.
1090 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1091 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1092 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1094 — Berlin, 2015-07-31
1098 * The python-systemd code has been removed from the systemd repository.
1099 A new repository has been created which accommodates the code from
1100 now on, and we kindly ask distributions to create a separate package
1101 for this: https://github.com/systemd/python-systemd
1103 * The systemd daemon will now reload its main configuration
1104 (/etc/systemd/system.conf) on daemon-reload.
1106 * sd-dhcp now exposes vendor specific extensions via
1107 sd_dhcp_lease_get_vendor_specific().
1109 * systemd-networkd gained a number of new configuration options.
1111 - A new boolean configuration option for TAP devices called
1112 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1113 device, thus allowing to send and receive GSO packets.
1115 - A new tunnel configuration option called 'CopyDSCP='.
1116 If enabled, the DSCP field of ip6 tunnels is copied into the
1117 decapsulated packet.
1119 - A set of boolean bridge configuration options were added.
1120 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1121 and 'UnicastFlood=' are now parsed by networkd and applied to the
1122 respective bridge link device via the respective IFLA_BRPORT_*
1125 - A new string configuration option to override the hostname sent
1126 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1127 is true, networkd will use the configured hostname instead of the
1128 system hostname when sending DHCP requests.
1130 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1131 networkd will configure the IPv6 flow-label of the tunnel device
1132 according to RFC2460.
1134 - The 'macvtap' virtual network devices are now supported, similar to
1135 the already supported 'macvlan' devices.
1137 * systemd-resolved now implements RFC5452 to improve resilience against
1138 cache poisoning. Additionally, source port randomization is enabled
1139 by default to further protect against DNS spoofing attacks.
1141 * nss-mymachines now supports translating UIDs and GIDs of running
1142 containers with user-namespaces enabled. If a container 'foo'
1143 translates a host uid 'UID' to the container uid 'TUID', then
1144 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1145 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1146 mapped as 'vg-foo-TGID'.
1148 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1149 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1150 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1151 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1152 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1153 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1154 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1155 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1156 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1157 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1159 — Berlin, 2015-07-29
1163 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1164 There are no known issues with current sysfs, and udev does not need
1165 or should be used to work around such bugs.
1167 * udev does no longer enable USB HID power management. Several reports
1168 indicate, that some devices cannot handle that setting.
1170 * The udev accelerometer helper was removed. The functionality
1171 is now fully included in iio-sensor-proxy. But this means,
1172 older iio-sensor-proxy versions will no longer provide
1173 accelerometer/orientation data with this systemd version.
1174 Please upgrade iio-sensor-proxy to version 1.0.
1176 * networkd gained a new configuration option IPv6PrivacyExtensions=
1177 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1178 for Stateless Address") on selected networks.
1180 * For the sake of fewer build-time dependencies and less code in the
1181 main repository, the python bindings are about to be removed in the
1182 next release. A new repository has been created which accommodates
1183 the code from now on, and we kindly ask distributions to create a
1184 separate package for this. The removal will take place in v223.
1186 https://github.com/systemd/python-systemd
1188 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1189 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1190 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1191 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1192 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1193 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1194 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1195 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1196 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1197 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1199 — Berlin, 2015-07-07
1203 * The sd-bus.h and sd-event.h APIs have now been declared
1204 stable and have been added to the official interface of
1205 libsystemd.so. sd-bus implements an alternative D-Bus client
1206 library, that is relatively easy to use, very efficient and
1207 supports both classic D-Bus as well as kdbus as transport
1208 backend. sd-event is a generic event loop abstraction that
1209 is built around Linux epoll, but adds features such as event
1210 prioritization or efficient timer handling. Both APIs are good
1211 choices for C programs looking for a bus and/or event loop
1212 implementation that is minimal and does not have to be
1213 portable to other kernels.
1215 * kdbus support is no longer compile-time optional. It is now
1216 always built-in. However, it can still be disabled at
1217 runtime using the kdbus=0 kernel command line setting, and
1218 that setting may be changed to default to off, by specifying
1219 --disable-kdbus at build-time. Note though that the kernel
1220 command line setting has no effect if the kdbus.ko kernel
1221 module is not installed, in which case kdbus is (obviously)
1222 also disabled. We encourage all downstream distributions to
1223 begin testing kdbus by adding it to the kernel images in the
1224 development distributions, and leaving kdbus support in
1227 * The minimal required util-linux version has been bumped to
1230 * Support for chkconfig (--enable-chkconfig) was removed in
1231 favor of calling an abstraction tool
1232 /lib/systemd/systemd-sysv-install. This needs to be
1233 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1234 in README for details.
1236 * If there's a systemd unit and a SysV init script for the
1237 same service name, and the user executes "systemctl enable"
1238 for it (or a related call), then this will now enable both
1239 (or execute the related operation on both), not just the
1242 * The libudev API documentation has been converted from gtkdoc
1245 * gudev has been removed from the systemd tree, it is now an
1248 * The systemd-cgtop tool learnt a new --raw switch to generate
1249 "raw" (machine parsable) output.
1251 * networkd's IPForwarding= .network file setting learnt the
1252 new setting "kernel", which ensures that networkd does not
1253 change the IP forwarding sysctl from the default kernel
1256 * The systemd-logind bus API now exposes a new boolean
1257 property "Docked" that reports whether logind considers the
1258 system "docked", i.e. connected to a docking station or not.
1260 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1261 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1262 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1263 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1264 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1265 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1266 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1267 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1268 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1269 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1270 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1271 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1272 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1273 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1274 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1275 Fink, Zbigniew Jędrzejewski-Szmek
1277 — Berlin, 2015-06-19
1281 * The gudev library has been extracted into a separate repository
1282 available at: https://git.gnome.org/browse/libgudev/
1283 It is now managed as part of the Gnome project. Distributions
1284 are recommended to pass --disable-gudev to systemd and use
1285 gudev from the Gnome project instead. gudev is still included
1286 in systemd, for now. It will be removed soon, though. Please
1287 also see the announcement-thread on systemd-devel:
1288 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1290 * systemd now exposes a CPUUsageNSec= property for each
1291 service unit on the bus, that contains the overall consumed
1292 CPU time of a service (the sum of what each process of the
1293 service consumed). This value is only available if
1294 CPUAccounting= is turned on for a service, and is then shown
1295 in the "systemctl status" output.
1297 * Support for configuring alternative mappings of the old SysV
1298 runlevels to systemd targets has been removed. They are now
1299 hardcoded in a way that runlevels 2, 3, 4 all map to
1300 multi-user.target and 5 to graphical.target (which
1301 previously was already the default behaviour).
1303 * The auto-mounter logic gained support for mount point
1304 expiry, using a new TimeoutIdleSec= setting in .automount
1305 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1307 * The EFI System Partition (ESP) as mounted to /boot by
1308 systemd-efi-boot-generator will now be unmounted
1309 automatically after 2 minutes of not being used. This should
1310 minimize the risk of ESP corruptions.
1312 * New /etc/fstab options x-systemd.requires= and
1313 x-systemd.requires-mounts-for= are now supported to express
1314 additional dependencies for mounts. This is useful for
1315 journalling file systems that support external journal
1316 devices or overlay file systems that require underlying file
1317 systems to be mounted.
1319 * systemd does not support direct live-upgrades (via systemctl
1320 daemon-reexec) from versions older than v44 anymore. As no
1321 distribution we are aware of shipped such old versions in a
1322 stable release this should not be problematic.
1324 * When systemd forks off a new per-connection service instance
1325 it will now set the $REMOTE_ADDR environment variable to the
1326 remote IP address, and $REMOTE_PORT environment variable to
1327 the remote IP port. This behaviour is similar to the
1328 corresponding environment variables defined by CGI.
1330 * systemd-networkd gained support for uplink failure
1331 detection. The BindCarrier= option allows binding interface
1332 configuration dynamically to the link sense of other
1333 interfaces. This is useful to achieve behaviour like in
1336 * systemd-networkd gained support for configuring the DHCP
1337 client identifier to use when requesting leases.
1339 * systemd-networkd now has a per-network UseNTP= option to
1340 configure whether NTP server information acquired via DHCP
1341 is passed on to services like systemd-timesyncd.
1343 * systemd-networkd gained support for vti6 tunnels.
1345 * Note that systemd-networkd manages the sysctl variable
1346 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1347 it is configured for since v219. The variable controls IP
1348 forwarding, and is a per-interface alternative to the global
1349 /proc/sys/net/ipv[46]/ip_forward. This setting is
1350 configurable in the IPForward= option, which defaults to
1351 "no". This means if networkd is used for an interface it is
1352 no longer sufficient to set the global sysctl option to turn
1353 on IP forwarding! Instead, the .network file option
1354 IPForward= needs to be turned on! Note that the
1355 implementation of this behaviour was broken in v219 and has
1358 * Many bonding and vxlan options are now configurable in
1361 * systemd-nspawn gained a new --property= setting to set unit
1362 properties for the container scope. This is useful for
1363 setting resource parameters (e.g "CPUShares=500") on
1364 containers started from the command line.
1366 * systemd-nspawn gained a new --private-users= switch to make
1367 use of user namespacing available on recent Linux kernels.
1369 * systemd-nspawn may now be called as part of a shell pipeline
1370 in which case the pipes used for stdin and stdout are passed
1371 directly to the process invoked in the container, without
1372 indirection via a pseudo tty.
1374 * systemd-nspawn gained a new switch to control the UNIX
1375 signal to use when killing the init process of the container
1378 * systemd-nspawn gained a new --overlay= switch for mounting
1379 overlay file systems into the container using the new kernel
1382 * When a container image is imported via systemd-importd and
1383 the host file system is not btrfs, a loopback block device
1384 file is created in /var/lib/machines.raw with a btrfs file
1385 system inside. It is then mounted to /var/lib/machines to
1386 enable btrfs features for container management. The loopback
1387 file and btrfs file system is grown as needed when container
1388 images are imported via systemd-importd.
1390 * systemd-machined/systemd-importd gained support for btrfs
1391 quota, to enforce container disk space limits on disk. This
1392 is exposed in "machinectl set-limit".
1394 * systemd-importd now can import containers from local .tar,
1395 .raw and .qcow2 images, and export them to .tar and .raw. It
1396 can also import dkr v2 images now from the network (on top
1399 * systemd-importd gained support for verifying downloaded
1400 images with gpg2 (previously only gpg1 was supported).
1402 * systemd-machined, systemd-logind, systemd: most bus calls
1403 are now accessible to unprivileged processes via
1404 PolicyKit. Also, systemd-logind will now allow users to kill
1405 their own sessions without further privileges or
1408 * systemd-shutdownd has been removed. This service was
1409 previously responsible for implementing scheduled shutdowns
1410 as exposed in /usr/bin/shutdown's time parameter. This
1411 functionality has now been moved into systemd-logind and is
1412 accessible via a bus interface.
1414 * "systemctl reboot" gained a new switch --firmware-setup that
1415 can be used to reboot into the EFI firmware setup, if that
1416 is available. systemd-logind now exposes an API on the bus
1417 to trigger such reboots, in case graphical desktop UIs want
1418 to cover this functionality.
1420 * "systemctl enable", "systemctl disable" and "systemctl mask"
1421 now support a new "--now" switch. If specified the units
1422 that are enabled will also be started, and the ones
1423 disabled/masked also stopped.
1425 * The Gummiboot EFI boot loader tool has been merged into
1426 systemd, and renamed to "systemd-boot". The bootctl tool has been
1427 updated to support systemd-boot.
1429 * An EFI kernel stub has been added that may be used to create
1430 kernel EFI binaries that contain not only the actual kernel,
1431 but also an initrd, boot splash, command line and OS release
1432 information. This combined binary can then be signed as a
1433 single image, so that the firmware can verify it all in one
1434 step. systemd-boot has special support for EFI binaries created
1435 like this and can extract OS release information from them
1436 and show them in the boot menu. This functionality is useful
1437 to implement cryptographically verified boot schemes.
1439 * Optional support has been added to systemd-fsck to pass
1440 fsck's progress report to an AF_UNIX socket in the file
1443 * udev will no longer create device symlinks for all block
1444 devices by default. A blacklist for excluding special block
1445 devices from this logic has been turned into a whitelist
1446 that requires picking block devices explicitly that require
1449 * A new (currently still internal) API sd-device.h has been
1450 added to libsystemd. This modernized API is supposed to
1451 replace libudev eventually. In fact, already much of libudev
1452 is now just a wrapper around sd-device.h.
1454 * A new hwdb database for storing metadata about pointing
1455 stick devices has been added.
1457 * systemd-tmpfiles gained support for setting file attributes
1458 similar to the "chattr" tool with new 'h' and 'H' lines.
1460 * systemd-journald will no longer unconditionally set the
1461 btrfs NOCOW flag on new journal files. This is instead done
1462 with tmpfiles snippet using the new 'h' line type. This
1463 allows easy disabling of this logic, by masking the
1464 journal-nocow.conf tmpfiles file.
1466 * systemd-journald will now translate audit message types to
1467 human readable identifiers when writing them to the
1468 journal. This should improve readability of audit messages.
1470 * The LUKS logic gained support for the offset= and skip=
1471 options in /etc/crypttab, as previously implemented by
1474 * /usr/lib/os-release gained a new optional field VARIANT= for
1475 distributions that support multiple variants (such as a
1476 desktop edition, a server edition, ...)
1478 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1479 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1480 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1481 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1482 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1483 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1484 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1485 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1486 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1487 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1488 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1489 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1490 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1491 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1492 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1493 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1494 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1495 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1496 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1497 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1498 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1499 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1500 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1501 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1502 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1503 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1504 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1506 — Berlin, 2015-05-22
1510 * Introduce a new API "sd-hwdb.h" for querying the hardware
1511 metadata database. With this minimal interface one can query
1512 and enumerate the udev hwdb, decoupled from the old libudev
1513 library. libudev's interface for this is now only a wrapper
1514 around sd-hwdb. A new tool systemd-hwdb has been added to
1515 interface with and update the database.
1517 * When any of systemd's tools copies files (for example due to
1518 tmpfiles' C lines) a btrfs reflink will attempted first,
1519 before bytewise copying is done.
1521 * systemd-nspawn gained a new --ephemeral switch. When
1522 specified a btrfs snapshot is taken of the container's root
1523 directory, and immediately removed when the container
1524 terminates again. Thus, a container can be started whose
1525 changes never alter the container's root directory, and are
1526 lost on container termination. This switch can also be used
1527 for starting a container off the root file system of the
1528 host without affecting the host OS. This switch is only
1529 available on btrfs file systems.
1531 * systemd-nspawn gained a new --template= switch. It takes the
1532 path to a container tree to use as template for the tree
1533 specified via --directory=, should that directory be
1534 missing. This allows instantiating containers dynamically,
1535 on first run. This switch is only available on btrfs file
1538 * When a .mount unit refers to a mount point on which multiple
1539 mounts are stacked, and the .mount unit is stopped all of
1540 the stacked mount points will now be unmounted until no
1541 mount point remains.
1543 * systemd now has an explicit notion of supported and
1544 unsupported unit types. Jobs enqueued for unsupported unit
1545 types will now fail with an "unsupported" error code. More
1546 specifically .swap, .automount and .device units are not
1547 supported in containers, .busname units are not supported on
1548 non-kdbus systems. .swap and .automount are also not
1549 supported if their respective kernel compile time options
1552 * machinectl gained support for two new "copy-from" and
1553 "copy-to" commands for copying files from a running
1554 container to the host or vice versa.
1556 * machinectl gained support for a new "bind" command to bind
1557 mount host directories into local containers. This is
1558 currently only supported for nspawn containers.
1560 * networkd gained support for configuring bridge forwarding
1561 database entries (fdb) from .network files.
1563 * A new tiny daemon "systemd-importd" has been added that can
1564 download container images in tar, raw, qcow2 or dkr formats,
1565 and make them available locally in /var/lib/machines, so
1566 that they can run as nspawn containers. The daemon can GPG
1567 verify the downloads (not supported for dkr, since it has no
1568 provisions for verifying downloads). It will transparently
1569 decompress bz2, xz, gzip compressed downloads if necessary,
1570 and restore sparse files on disk. The daemon uses privilege
1571 separation to ensure the actual download logic runs with
1572 fewer privileges than the daemon itself. machinectl has
1573 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1574 make the functionality of importd available to the
1575 user. With this in place the Fedora and Ubuntu "Cloud"
1576 images can be downloaded and booted as containers unmodified
1577 (the Fedora images lack the appropriate GPG signature files
1578 currently, so they cannot be verified, but this will change
1579 soon, hopefully). Note that downloading images is currently
1580 only fully supported on btrfs.
1582 * machinectl is now able to list container images found in
1583 /var/lib/machines, along with some metadata about sizes of
1584 disk and similar. If the directory is located on btrfs and
1585 quota is enabled, this includes quota display. A new command
1586 "image-status" has been added that shows additional
1587 information about images.
1589 * machinectl is now able to clone container images
1590 efficiently, if the underlying file system (btrfs) supports
1591 it, with the new "machinectl clone" command. It also
1592 gained commands for renaming and removing images, as well as
1593 marking them read-only or read-write (supported also on
1594 legacy file systems).
1596 * networkd gained support for collecting LLDP network
1597 announcements, from hardware that supports this. This is
1598 shown in networkctl output.
1600 * systemd-run gained support for a new -t (--pty) switch for
1601 invoking a binary on a pty whose input and output is
1602 connected to the invoking terminal. This allows executing
1603 processes as system services while interactively
1604 communicating with them via the terminal. Most interestingly
1605 this is supported across container boundaries. Invoking
1606 "systemd-run -t /bin/bash" is an alternative to running a
1607 full login session, the difference being that the former
1608 will not register a session, nor go through the PAM session
1611 * tmpfiles gained support for a new "v" line type for creating
1612 btrfs subvolumes. If the underlying file system is a legacy
1613 file system, this automatically degrades to creating a
1614 normal directory. Among others /var/lib/machines is now
1615 created like this at boot, should it be missing.
1617 * The directory /var/lib/containers/ has been deprecated and
1618 been replaced by /var/lib/machines. The term "machines" has
1619 been used in the systemd context as generic term for both
1620 VMs and containers, and hence appears more appropriate for
1621 this, as the directory can also contain raw images bootable
1624 * systemd-nspawn when invoked with -M but without --directory=
1625 or --image= is now capable of searching for the container
1626 root directory, subvolume or disk image automatically, in
1627 /var/lib/machines. systemd-nspawn@.service has been updated
1628 to make use of this, thus allowing it to be used for raw
1631 * A new machines.target unit has been introduced that is
1632 supposed to group all containers/VMs invoked as services on
1633 the system. systemd-nspawn@.service has been updated to
1634 integrate with that.
1636 * machinectl gained a new "start" command, for invoking a
1637 container as a service. "machinectl start foo" is mostly
1638 equivalent to "systemctl start systemd-nspawn@foo.service",
1639 but handles escaping in a nicer way.
1641 * systemd-nspawn will now mount most of the cgroupfs tree
1642 read-only into each container, with the exception of the
1643 container's own subtree in the name=systemd hierarchy.
1645 * journald now sets the special FS_NOCOW file flag for its
1646 journal files. This should improve performance on btrfs, by
1647 avoiding heavy fragmentation when journald's write-pattern
1648 is used on COW file systems. It degrades btrfs' data
1649 integrity guarantees for the files to the same levels as for
1650 ext3/ext4 however. This should be OK though as journald does
1651 its own data integrity checks and all its objects are
1652 checksummed on disk. Also, journald should handle btrfs disk
1653 full events a lot more gracefully now, by processing SIGBUS
1654 errors, and not relying on fallocate() anymore.
1656 * When journald detects that journal files it is writing to
1657 have been deleted it will immediately start new journal
1660 * systemd now provides a way to store file descriptors
1661 per-service in PID 1.This is useful for daemons to ensure
1662 that fds they require are not lost during a daemon
1663 restart. The fds are passed to the daemon on the next
1664 invocation in the same way socket activation fds are
1665 passed. This is now used by journald to ensure that the
1666 various sockets connected to all the system's stdout/stderr
1667 are not lost when journald is restarted. File descriptors
1668 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1669 an extension to sd_notify(). Note that a limit is enforced
1670 on the number of fds a service can store in PID 1, and it
1671 defaults to 0, so that no fds may be stored, unless this is
1672 explicitly turned on.
1674 * The default TERM variable to use for units connected to a
1675 terminal, when no other value is explicitly is set is now
1676 vt220 rather than vt102. This should be fairly safe still,
1677 but allows PgUp/PgDn work.
1679 * The /etc/crypttab option header= as known from Debian is now
1682 * "loginctl user-status" and "loginctl session-status" will
1683 now show the last 10 lines of log messages of the
1684 user/session following the status output. Similar,
1685 "machinectl status" will show the last 10 log lines
1686 associated with a virtual machine or container
1687 service. (Note that this is usually not the log messages
1688 done in the VM/container itself, but simply what the
1689 container manager logs. For nspawn this includes all console
1692 * "loginctl session-status" without further argument will now
1693 show the status of the session of the caller. Similar,
1694 "lock-session", "unlock-session", "activate",
1695 "enable-linger", "disable-linger" may now be called without
1696 session/user parameter in which case they apply to the
1697 caller's session/user.
1699 * An X11 session scriptlet is now shipped that uploads
1700 $DISPLAY and $XAUTHORITY into the environment of the systemd
1701 --user daemon if a session begins. This should improve
1702 compatibility with X11 enabled applications run as systemd
1705 * Generators are now subject to masking via /etc and /run, the
1706 same way as unit files.
1708 * networkd .network files gained support for configuring
1709 per-link IPv4/IPv6 packet forwarding as well as IPv4
1710 masquerading. This is by default turned on for veth links to
1711 containers, as registered by systemd-nspawn. This means that
1712 nspawn containers run with --network-veth will now get
1713 automatic routed access to the host's networks without any
1714 further configuration or setup, as long as networkd runs on
1717 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1718 or UDP posts of a container on the host. With this in place
1719 it is possible to run containers with private veth links
1720 (--network-veth), and have their functionality exposed on
1721 the host as if their services were running directly on the
1724 * systemd-nspawn's --network-veth switch now gained a short
1725 version "-n", since with the changes above it is now truly
1726 useful out-of-the-box. The systemd-nspawn@.service has been
1727 updated to make use of it too by default.
1729 * systemd-nspawn will now maintain a per-image R/W lock, to
1730 ensure that the same image is not started more than once
1731 writable. (It's OK to run an image multiple times
1732 simultaneously in read-only mode.)
1734 * systemd-nspawn's --image= option is now capable of
1735 dissecting and booting MBR and GPT disk images that contain
1736 only a single active Linux partition. Previously it
1737 supported only GPT disk images with proper GPT type
1738 IDs. This allows running cloud images from major
1739 distributions directly with systemd-nspawn, without
1742 * In addition to collecting mouse dpi data in the udev
1743 hardware database, there's now support for collecting angle
1744 information for mouse scroll wheels. The database is
1745 supposed to guarantee similar scrolling behavior on mice
1746 that it knows about. There's also support for collecting
1747 information about Touchpad types.
1749 * udev's input_id built-in will now also collect touch screen
1750 dimension data and attach it to probed devices.
1752 * /etc/os-release gained support for a Distribution Privacy
1755 * networkd gained support for creating "ipvlan", "gretap",
1756 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1758 * systemd-tmpfiles gained support for "a" lines for setting
1761 * systemd-nspawn will now mount /tmp in the container to
1762 tmpfs, automatically.
1764 * systemd now exposes the memory.usage_in_bytes cgroup
1765 attribute and shows it for each service in the "systemctl
1766 status" output, if available.
1768 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1769 immediate reboot is triggered. This useful if shutdown is
1770 hung and is unable to complete, to expedite the
1771 operation. Note that this kind of reboot will still unmount
1772 all file systems, and hence should not result in fsck being
1775 * A .device unit for an optical block device will now be
1776 considered active only when a medium is in the drive. Also,
1777 mount units are now bound to their backing devices thus
1778 triggering automatic unmounting when devices become
1779 unavailable. With this in place systemd will now
1780 automatically unmount left-over mounts when a CD-ROM is
1781 ejected or an USB stick is yanked from the system.
1783 * networkd-wait-online now has support for waiting for
1784 specific interfaces only (with globbing), and for giving up
1785 after a configurable timeout.
1787 * networkd now exits when idle. It will be automatically
1788 restarted as soon as interfaces show up, are removed or
1789 change state. networkd will stay around as long as there is
1790 at least one DHCP state machine or similar around, that keep
1793 * networkd may now configure IPv6 link-local addressing in
1794 addition to IPv4 link-local addressing.
1796 * The IPv6 "token" for use in SLAAC may now be configured for
1797 each .network interface in networkd.
1799 * Routes configured with networkd may now be assigned a scope
1802 * networkd's [Match] sections now support globbing and lists
1803 of multiple space-separated matches per item.
1805 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1806 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1807 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1808 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1809 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1810 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1811 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1812 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1813 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1814 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1815 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1816 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1817 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1818 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1819 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1820 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1821 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1822 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1823 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1824 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1825 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1826 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1827 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1828 Hoffmann, Zbigniew Jędrzejewski-Szmek
1830 — Berlin, 2015-02-16
1834 * When querying unit file enablement status (for example via
1835 "systemctl is-enabled"), a new state "indirect" is now known
1836 which indicates that a unit might not be enabled itself, but
1837 another unit listed in its Also= setting might be.
1839 * Similar to the various existing ConditionXYZ= settings for
1840 units, there are now matching AssertXYZ= settings. While
1841 failing conditions cause a unit to be skipped, but its job
1842 to succeed, failing assertions declared like this will cause
1843 a unit start operation and its job to fail.
1845 * hostnamed now knows a new chassis type "embedded".
1847 * systemctl gained a new "edit" command. When used on a unit
1848 file, this allows extending unit files with .d/ drop-in
1849 configuration snippets or editing the full file (after
1850 copying it from /usr/lib to /etc). This will invoke the
1851 user's editor (as configured with $EDITOR), and reload the
1852 modified configuration after editing.
1854 * "systemctl status" now shows the suggested enablement state
1855 for a unit, as declared in the (usually vendor-supplied)
1856 system preset files.
1858 * nss-myhostname will now resolve the single-label host name
1859 "gateway" to the locally configured default IP routing
1860 gateways, ordered by their metrics. This assigns a stable
1861 name to the used gateways, regardless which ones are
1862 currently configured. Note that the name will only be
1863 resolved after all other name sources (if nss-myhostname is
1864 configured properly) and should hence not negatively impact
1865 systems that use the single-label host name "gateway" in
1868 * systemd-inhibit now allows filtering by mode when listing
1871 * Scope and service units gained a new "Delegate" boolean
1872 property, which, when set, allows processes running inside the
1873 unit to further partition resources. This is primarily
1874 useful for systemd user instances as well as container
1877 * journald will now pick up audit messages directly from
1878 the kernel, and log them like any other log message. The
1879 audit fields are split up and fully indexed. This means that
1880 journalctl in many ways is now a (nicer!) alternative to
1881 ausearch, the traditional audit client. Note that this
1882 implements only a minimal audit client. If you want the
1883 special audit modes like reboot-on-log-overflow, please use
1884 the traditional auditd instead, which can be used in
1885 parallel to journald.
1887 * The ConditionSecurity= unit file option now understands the
1888 special string "audit" to check whether auditing is
1891 * journalctl gained two new commands --vacuum-size= and
1892 --vacuum-time= to delete old journal files until the
1893 remaining ones take up no more than the specified size on disk,
1894 or are not older than the specified time.
1896 * A new, native PPPoE library has been added to sd-network,
1897 systemd's library of light-weight networking protocols. This
1898 library will be used in a future version of networkd to
1899 enable PPPoE communication without an external pppd daemon.
1901 * The busctl tool now understands a new "capture" verb that
1902 works similar to "monitor", but writes a packet capture
1903 trace to STDOUT that can be redirected to a file which is
1904 compatible with libcap's capture file format. This can then
1905 be loaded in Wireshark and similar tools to inspect bus
1908 * The busctl tool now understands a new "tree" verb that shows
1909 the object trees of a specific service on the bus, or of all
1912 * The busctl tool now understands a new "introspect" verb that
1913 shows all interfaces and members of objects on the bus,
1914 including their signature and values. This is particularly
1915 useful to get more information about bus objects shown by
1916 the new "busctl tree" command.
1918 * The busctl tool now understands new verbs "call",
1919 "set-property" and "get-property" for invoking bus method
1920 calls, setting and getting bus object properties in a
1923 * busctl gained a new --augment-creds= argument that controls
1924 whether the tool shall augment credential information it
1925 gets from the bus with data from /proc, in a possibly
1928 * nspawn's --link-journal= switch gained two new values
1929 "try-guest" and "try-host" that work like "guest" and
1930 "host", but do not fail if the host has no persistent
1931 journalling enabled. -j is now equivalent to
1932 --link-journal=try-guest.
1934 * macvlan network devices created by nspawn will now have
1935 stable MAC addresses.
1937 * A new SmackProcessLabel= unit setting has been added, which
1938 controls the SMACK security label processes forked off by
1939 the respective unit shall use.
1941 * If compiled with --enable-xkbcommon, systemd-localed will
1942 verify x11 keymap settings by compiling the given keymap. It
1943 will spew out warnings if the compilation fails. This
1944 requires libxkbcommon to be installed.
1946 * When a coredump is collected, a larger number of metadata
1947 fields is now collected and included in the journal records
1948 created for it. More specifically, control group membership,
1949 environment variables, memory maps, working directory,
1950 chroot directory, /proc/$PID/status, and a list of open file
1951 descriptors is now stored in the log entry.
1953 * The udev hwdb now contains DPI information for mice. For
1956 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1958 * All systemd programs that read standalone configuration
1959 files in /etc now also support a corresponding series of
1960 .conf.d configuration directories in /etc/, /run/,
1961 /usr/local/lib/, /usr/lib/, and (if configured with
1962 --enable-split-usr) /lib/. In particular, the following
1963 configuration files now have corresponding configuration
1964 directories: system.conf user.conf, logind.conf,
1965 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1966 resolved.conf, timesyncd.conf, journal-remote.conf, and
1967 journal-upload.conf. Note that distributions should use the
1968 configuration directories in /usr/lib/; the directories in
1969 /etc/ are reserved for the system administrator.
1971 * systemd-rfkill will no longer take the rfkill device name
1972 into account when storing rfkill state on disk, as the name
1973 might be dynamically assigned and not stable. Instead, the
1974 ID_PATH udev variable combined with the rfkill type (wlan,
1975 bluetooth, ...) is used.
1977 * A new service systemd-machine-id-commit.service has been
1978 added. When used on systems where /etc is read-only during
1979 boot, and /etc/machine-id is not initialized (but an empty
1980 file), this service will copy the temporary machine ID
1981 created as replacement into /etc after the system is fully
1982 booted up. This is useful for systems that are freshly
1983 installed with a non-initialized machine ID, but should get
1984 a fixed machine ID for subsequent boots.
1986 * networkd's .netdev files now provide a large set of
1987 configuration parameters for VXLAN devices. Similarly, the
1988 bridge port cost parameter is now configurable in .network
1989 files. There's also new support for configuring IP source
1990 routing. networkd .link files gained support for a new
1991 OriginalName= match that is useful to match against the
1992 original interface name the kernel assigned. .network files
1993 may include MTU= and MACAddress= fields for altering the MTU
1994 and MAC address while being connected to a specific network
1997 * The LUKS logic gained supported for configuring
1998 UUID-specific key files. There's also new support for naming
1999 LUKS device from the kernel command line, using the new
2000 luks.name= argument.
2002 * Timer units may now be transiently created via the bus API
2003 (this was previously already available for scope and service
2004 units). In addition it is now possible to create multiple
2005 transient units at the same time with a single bus call. The
2006 "systemd-run" tool has been updated to make use of this for
2007 running commands on a specified time, in at(1)-style.
2009 * tmpfiles gained support for "t" lines, for assigning
2010 extended attributes to files. Among other uses this may be
2011 used to assign SMACK labels to files.
2013 Contributions from: Alin Rauta, Alison Chaiken, Andrej
2014 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
2015 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
2016 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
2017 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
2018 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
2019 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
2020 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
2021 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
2022 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
2023 Michael Biebl, Michael Chapman, Michael Marineau, Michal
2024 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
2025 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
2026 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
2027 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2028 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2029 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2030 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2032 — Berlin, 2014-12-10
2036 * journalctl gained the new options -t/--identifier= to match
2037 on the syslog identifier (aka "tag"), as well as --utc to
2038 show log timestamps in the UTC timezone. journalctl now also
2039 accepts -n/--lines=all to disable line capping in a pager.
2041 * journalctl gained a new switch, --flush, that synchronously
2042 flushes logs from /run/log/journal to /var/log/journal if
2043 persistent storage is enabled. systemd-journal-flush.service
2044 now waits until the operation is complete.
2046 * Services can notify the manager before they start a reload
2047 (by sending RELOADING=1) or shutdown (by sending
2048 STOPPING=1). This allows the manager to track and show the
2049 internal state of daemons and closes a race condition when
2050 the process is still running but has closed its D-Bus
2053 * Services with Type=oneshot do not have to have any ExecStart
2056 * User units are now loaded also from
2057 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2058 /run/systemd/user directory that was already previously
2059 supported, but is under the control of the user.
2061 * Job timeouts (i.e. time-outs on the time a job that is
2062 queued stays in the run queue) can now optionally result in
2063 immediate reboot or power-off actions (JobTimeoutAction= and
2064 JobTimeoutRebootArgument=). This is useful on ".target"
2065 units, to limit the maximum time a target remains
2066 undispatched in the run queue, and to trigger an emergency
2067 operation in such a case. This is now used by default to
2068 turn off the system if boot-up (as defined by everything in
2069 basic.target) hangs and does not complete for at least
2070 15min. Also, if power-off or reboot hang for at least 30min
2071 an immediate power-off/reboot operation is triggered. This
2072 functionality is particularly useful to increase reliability
2073 on embedded devices, but also on laptops which might
2074 accidentally get powered on when carried in a backpack and
2075 whose boot stays stuck in a hard disk encryption passphrase
2078 * systemd-logind can be configured to also handle lid switch
2079 events even when the machine is docked or multiple displays
2080 are attached (HandleLidSwitchDocked= option).
2082 * A helper binary and a service have been added which can be
2083 used to resume from hibernation in the initramfs. A
2084 generator will parse the resume= option on the kernel
2085 command line to trigger resume.
2087 * A user console daemon systemd-consoled has been
2088 added. Currently, it is a preview, and will so far open a
2089 single terminal on each session of the user marked as
2090 Desktop=systemd-console.
2092 * Route metrics can be specified for DHCP routes added by
2095 * The SELinux context of socket-activated services can be set
2096 from the information provided by the networking stack
2097 (SELinuxContextFromNet= option).
2099 * Userspace firmware loading support has been removed and
2100 the minimum supported kernel version is thus bumped to 3.7.
2102 * Timeout for udev workers has been increased from 1 to 3
2103 minutes, but a warning will be printed after 1 minute to
2104 help diagnose kernel modules that take a long time to load.
2106 * Udev rules can now remove tags on devices with TAG-="foobar".
2108 * systemd's readahead implementation has been removed. In many
2109 circumstances it didn't give expected benefits even for
2110 rotational disk drives and was becoming less relevant in the
2111 age of SSDs. As none of the developers has been using
2112 rotating media anymore, and nobody stepped up to actively
2113 maintain this component of systemd it has now been removed.
2115 * Swap units can use Options= to specify discard options.
2116 Discard options specified for swaps in /etc/fstab are now
2119 * Docker containers are now detected as a separate type of
2122 * The Password Agent protocol gained support for queries where
2123 the user input is shown, useful e.g. for user names.
2124 systemd-ask-password gained a new --echo option to turn that
2127 * The default sysctl.d/ snippets will now set:
2129 net.core.default_qdisc = fq_codel
2131 This selects Fair Queuing Controlled Delay as the default
2132 queuing discipline for network interfaces. fq_codel helps
2133 fight the network bufferbloat problem. It is believed to be
2134 a good default with no tuning required for most workloads.
2135 Downstream distributions may override this choice. On 10Gbit
2136 servers that do not do forwarding, "fq" may perform better.
2137 Systems without a good clocksource should use "pfifo_fast".
2139 * If kdbus is enabled during build a new option BusPolicy= is
2140 available for service units, that allows locking all service
2141 processes into a stricter bus policy, in order to limit
2142 access to various bus services, or even hide most of them
2143 from the service's view entirely.
2145 * networkctl will now show the .network and .link file
2146 networkd has applied to a specific interface.
2148 * sd-login gained a new API call sd_session_get_desktop() to
2149 query which desktop environment has been selected for a
2152 * UNIX utmp support is now compile-time optional to support
2153 legacy-free systems.
2155 * systemctl gained two new commands "add-wants" and
2156 "add-requires" for pulling in units from specific targets
2159 * If the word "rescue" is specified on the kernel command line
2160 the system will now boot into rescue mode (aka
2161 rescue.target), which was previously available only by
2162 specifying "1" or "systemd.unit=rescue.target" on the kernel
2163 command line. This new kernel command line option nicely
2164 mirrors the already existing "emergency" kernel command line
2167 * New kernel command line options mount.usr=, mount.usrflags=,
2168 mount.usrfstype= have been added that match root=, rootflags=,
2169 rootfstype= but allow mounting a specific file system to
2172 * The $NOTIFY_SOCKET is now also passed to control processes of
2173 services, not only the main process.
2175 * This version reenables support for fsck's -l switch. This
2176 means at least version v2.25 of util-linux is required for
2177 operation, otherwise dead-locks on device nodes may
2178 occur. Again: you need to update util-linux to at least
2179 v2.25 when updating systemd to v217.
2181 * The "multi-seat-x" tool has been removed from systemd, as
2182 its functionality has been integrated into X servers 1.16,
2183 and the tool is hence redundant. It is recommended to update
2184 display managers invoking this tool to simply invoke X
2185 directly from now on, again.
2187 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2188 message flag has been added for all of systemd's PolicyKit
2189 authenticated method calls has been added. In particular
2190 this now allows optional interactive authorization via
2191 PolicyKit for many of PID1's privileged operations such as
2192 unit file enabling and disabling.
2194 * "udevadm hwdb --update" learnt a new switch "--usr" for
2195 placing the rebuilt hardware database in /usr instead of
2196 /etc. When used only hardware database entries stored in
2197 /usr will be used, and any user database entries in /etc are
2198 ignored. This functionality is useful for vendors to ship a
2199 pre-built database on systems where local configuration is
2200 unnecessary or unlikely.
2202 * Calendar time specifications in .timer units now also
2203 understand the strings "semi-annually", "quarterly" and
2204 "minutely" as shortcuts (in addition to the preexisting
2205 "anually", "hourly", ...).
2207 * systemd-tmpfiles will now correctly create files in /dev
2208 at boot which are marked for creation only at boot. It is
2209 recommended to always create static device nodes with 'c!'
2210 and 'b!', so that they are created only at boot and not
2211 overwritten at runtime.
2213 * When the watchdog logic is used for a service (WatchdogSec=)
2214 and the watchdog timeout is hit the service will now be
2215 terminated with SIGABRT (instead of just SIGTERM), in order
2216 to make sure a proper coredump and backtrace is
2217 generated. This ensures that hanging services will result in
2218 similar coredump/backtrace behaviour as services that hit a
2221 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2222 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2223 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2224 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2225 Herrmann, David Sommerseth, David Strauss, Emil Renner
2226 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2227 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2228 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2229 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2230 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2231 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2232 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2233 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2234 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2235 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2236 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2237 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2238 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2239 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2240 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2241 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2244 — Berlin, 2014-10-28
2248 * timedated no longer reads NTP implementation unit names from
2249 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2250 implementations should add a
2252 Conflicts=systemd-timesyncd.service
2254 to their unit files to take over and replace systemd's NTP
2255 default functionality.
2257 * systemd-sysusers gained a new line type "r" for configuring
2258 which UID/GID ranges to allocate system users/groups
2259 from. Lines of type "u" may now add an additional column
2260 that specifies the home directory for the system user to be
2261 created. Also, systemd-sysusers may now optionally read user
2262 information from STDIN instead of a file. This is useful for
2263 invoking it from RPM preinst scriptlets that need to create
2264 users before the first RPM file is installed since these
2265 files might need to be owned by them. A new
2266 %sysusers_create_inline RPM macro has been introduced to do
2267 just that. systemd-sysusers now updates the shadow files as
2268 well as the user/group databases, which should enhance
2269 compatibility with certain tools like grpck.
2271 * A number of bus APIs of PID 1 now optionally consult
2272 PolicyKit to permit access for otherwise unprivileged
2273 clients under certain conditions. Note that this currently
2274 doesn't support interactive authentication yet, but this is
2275 expected to be added eventually, too.
2277 * /etc/machine-info now has new fields for configuring the
2278 deployment environment of the machine, as well as the
2279 location of the machine. hostnamectl has been updated with
2280 new command to update these fields.
2282 * systemd-timesyncd has been updated to automatically acquire
2283 NTP server information from systemd-networkd, which might
2284 have been discovered via DHCP.
2286 * systemd-resolved now includes a caching DNS stub resolver
2287 and a complete LLMNR name resolution implementation. A new
2288 NSS module "nss-resolve" has been added which can be used
2289 instead of glibc's own "nss-dns" to resolve hostnames via
2290 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2291 be resolved via systemd-resolved D-Bus APIs. In contrast to
2292 the glibc internal resolver systemd-resolved is aware of
2293 multi-homed system, and keeps DNS server and caches separate
2294 and per-interface. Queries are sent simultaneously on all
2295 interfaces that have DNS servers configured, in order to
2296 properly handle VPNs and local LANs which might resolve
2297 separate sets of domain names. systemd-resolved may acquire
2298 DNS server information from systemd-networkd automatically,
2299 which in turn might have discovered them via DHCP. A tool
2300 "systemd-resolve-host" has been added that may be used to
2301 query the DNS logic in resolved. systemd-resolved implements
2302 IDNA and automatically uses IDNA or UTF-8 encoding depending
2303 on whether classic DNS or LLMNR is used as transport. In the
2304 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2305 implementation to systemd-resolved.
2307 * A new NSS module nss-mymachines has been added, that
2308 automatically resolves the names of all local registered
2309 containers to their respective IP addresses.
2311 * A new client tool "networkctl" for systemd-networkd has been
2312 added. It currently is entirely passive and will query
2313 networking configuration from udev, rtnetlink and networkd,
2314 and present it to the user in a very friendly
2315 way. Eventually, we hope to extend it to become a full
2316 control utility for networkd.
2318 * .socket units gained a new DeferAcceptSec= setting that
2319 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2320 TCP. Similarly, support for controlling TCP keep-alive
2321 settings has been added (KeepAliveTimeSec=,
2322 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2323 turning off Nagle's algorithm on TCP has been added
2326 * logind learned a new session type "web", for use in projects
2327 like Cockpit which register web clients as PAM sessions.
2329 * timer units with at least one OnCalendar= setting will now
2330 be started only after timer-sync.target has been
2331 reached. This way they will not elapse before the system
2332 clock has been corrected by a local NTP client or
2333 similar. This is particular useful on RTC-less embedded
2334 machines, that come up with an invalid system clock.
2336 * systemd-nspawn's --network-veth= switch should now result in
2337 stable MAC addresses for both the outer and the inner side
2340 * systemd-nspawn gained a new --volatile= switch for running
2341 container instances with /etc or /var unpopulated.
2343 * The kdbus client code has been updated to use the new Linux
2344 3.17 memfd subsystem instead of the old kdbus-specific one.
2346 * systemd-networkd's DHCP client and server now support
2347 FORCERENEW. There are also new configuration options to
2348 configure the vendor client identifier and broadcast mode
2351 * systemd will no longer inform the kernel about the current
2352 timezone, as this is necessarily incorrect and racy as the
2353 kernel has no understanding of DST and similar
2354 concepts. This hence means FAT timestamps will be always
2355 considered UTC, similar to what Android is already
2356 doing. Also, when the RTC is configured to the local time
2357 (rather than UTC) systemd will never synchronize back to it,
2358 as this might confuse Windows at a later boot.
2360 * systemd-analyze gained a new command "verify" for offline
2361 validation of unit files.
2363 * systemd-networkd gained support for a couple of additional
2364 settings for bonding networking setups. Also, the metric for
2365 statically configured routes may now be configured. For
2366 network interfaces where this is appropriate the peer IP
2367 address may now be configured.
2369 * systemd-networkd's DHCP client will no longer request
2370 broadcasting by default, as this tripped up some networks.
2371 For hardware where broadcast is required the feature should
2372 be switched back on using RequestBroadcast=yes.
2374 * systemd-networkd will now set up IPv4LL addresses (when
2375 enabled) even if DHCP is configured successfully.
2377 * udev will now default to respect network device names given
2378 by the kernel when the kernel indicates that these are
2379 predictable. This behavior can be tweaked by changing
2380 NamePolicy= in the relevant .link file.
2382 * A new library systemd-terminal has been added that
2383 implements full TTY stream parsing and rendering. This
2384 library is supposed to be used later on for implementing a
2385 full userspace VT subsystem, replacing the current kernel
2388 * A new tool systemd-journal-upload has been added to push
2389 journal data to a remote system running
2390 systemd-journal-remote.
2392 * journald will no longer forward all local data to another
2393 running syslog daemon. This change has been made because
2394 rsyslog (which appears to be the most commonly used syslog
2395 implementation these days) no longer makes use of this, and
2396 instead pulls the data out of the journal on its own. Since
2397 forwarding the messages to a non-existent syslog server is
2398 more expensive than we assumed we have now turned this
2399 off. If you run a syslog server that is not a recent rsyslog
2400 version, you have to turn this option on again
2401 (ForwardToSyslog= in journald.conf).
2403 * journald now optionally supports the LZ4 compressor for
2404 larger journal fields. This compressor should perform much
2405 better than XZ which was the previous default.
2407 * machinectl now shows the IP addresses of local containers,
2408 if it knows them, plus the interface name of the container.
2410 * A new tool "systemd-escape" has been added that makes it
2411 easy to escape strings to build unit names and similar.
2413 * sd_notify() messages may now include a new ERRNO= field
2414 which is parsed and collected by systemd and shown among the
2415 "systemctl status" output for a service.
2417 * A new component "systemd-firstboot" has been added that
2418 queries the most basic systemd information (timezone,
2419 hostname, root password) interactively on first
2420 boot. Alternatively it may also be used to provision these
2421 things offline on OS images installed into directories.
2423 * The default sysctl.d/ snippets will now set
2425 net.ipv4.conf.default.promote_secondaries=1
2427 This has the benefit of no flushing secondary IP addresses
2428 when primary addresses are removed.
2430 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2431 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2432 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2433 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2434 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2435 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2436 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2437 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2438 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2439 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2440 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2441 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2442 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2443 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2444 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2446 — Berlin, 2014-08-19
2450 * A new tool systemd-sysusers has been added. This tool
2451 creates system users and groups in /etc/passwd and
2452 /etc/group, based on static declarative system user/group
2453 definitions in /usr/lib/sysusers.d/. This is useful to
2454 enable factory resets and volatile systems that boot up with
2455 an empty /etc directory, and thus need system users and
2456 groups created during early boot. systemd now also ships
2457 with two default sysusers.d/ files for the most basic
2458 users and groups systemd and the core operating system
2461 * A new tmpfiles snippet has been added that rebuilds the
2462 essential files in /etc on boot, should they be missing.
2464 * A directive for ensuring automatic clean-up of
2465 /var/cache/man/ has been removed from the default
2466 configuration. This line should now be shipped by the man
2467 implementation. The necessary change has been made to the
2468 man-db implementation. Note that you need to update your man
2469 implementation to one that ships this line, otherwise no
2470 automatic clean-up of /var/cache/man will take place.
2472 * A new condition ConditionNeedsUpdate= has been added that
2473 may conditionalize services to only run when /etc or /var
2474 are "older" than the vendor operating system resources in
2475 /usr. This is useful for reconstructing or updating /etc
2476 after an offline update of /usr or a factory reset, on the
2477 next reboot. Services that want to run once after such an
2478 update or reset should use this condition and order
2479 themselves before the new systemd-update-done.service, which
2480 will mark the two directories as fully updated. A number of
2481 service files have been added making use of this, to rebuild
2482 the udev hardware database, the journald message catalog and
2483 dynamic loader cache (ldconfig). The systemd-sysusers tool
2484 described above also makes use of this now. With this in
2485 place it is now possible to start up a minimal operating
2486 system with /etc empty cleanly. For more information on the
2487 concepts involved see this recent blog story:
2489 http://0pointer.de/blog/projects/stateless.html
2491 * A new system group "input" has been introduced, and all
2492 input device nodes get this group assigned. This is useful
2493 for system-level software to get access to input devices. It
2494 complements what is already done for "audio" and "video".
2496 * systemd-networkd learnt minimal DHCPv4 server support in
2497 addition to the existing DHCPv4 client support. It also
2498 learnt DHCPv6 client and IPv6 Router Solicitation client
2499 support. The DHCPv4 client gained support for static routes
2500 passed in from the server. Note that the [DHCPv4] section
2501 known in older systemd-networkd versions has been renamed to
2502 [DHCP] and is now also used by the DHCPv6 client. Existing
2503 .network files using settings of this section should be
2504 updated, though compatibility is maintained. Optionally, the
2505 client hostname may now be sent to the DHCP server.
2507 * networkd gained support for vxlan virtual networks as well
2508 as tun/tap and dummy devices.
2510 * networkd gained support for automatic allocation of address
2511 ranges for interfaces from a system-wide pool of
2512 addresses. This is useful for dynamically managing a large
2513 number of interfaces with a single network configuration
2514 file. In particular this is useful to easily assign
2515 appropriate IP addresses to the veth links of a large number
2516 of nspawn instances.
2518 * RPM macros for processing sysusers, sysctl and binfmt
2519 drop-in snippets at package installation time have been
2522 * The /etc/os-release file should now be placed in
2523 /usr/lib/os-release. The old location is automatically
2524 created as symlink. /usr/lib is the more appropriate
2525 location of this file, since it shall actually describe the
2526 vendor operating system shipped in /usr, and not the
2527 configuration stored in /etc.
2529 * .mount units gained a new boolean SloppyOptions= setting
2530 that maps to mount(8)'s -s option which enables permissive
2531 parsing of unknown mount options.
2533 * tmpfiles learnt a new "L+" directive which creates a symlink
2534 but (unlike "L") deletes a pre-existing file first, should
2535 it already exist and not already be the correct
2536 symlink. Similarly, "b+", "c+" and "p+" directives have been
2537 added as well, which create block and character devices, as
2538 well as fifos in the filesystem, possibly removing any
2539 pre-existing files of different types.
2541 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2542 'argument' field (which so far specified the source to
2543 symlink/copy the files from) is now optional. If omitted the
2544 same file os copied from /usr/share/factory/ suffixed by the
2545 full destination path. This is useful for populating /etc
2546 with essential files, by copying them from vendor defaults
2547 shipped in /usr/share/factory/etc.
2549 * A new command "systemctl preset-all" has been added that
2550 applies the service preset settings to all installed unit
2551 files. A new switch --preset-mode= has been added that
2552 controls whether only enable or only disable operations
2555 * A new command "systemctl is-system-running" has been added
2556 that allows checking the overall state of the system, for
2557 example whether it is fully up and running.
2559 * When the system boots up with an empty /etc, the equivalent
2560 to "systemctl preset-all" is executed during early boot, to
2561 make sure all default services are enabled after a factory
2564 * systemd now contains a minimal preset file that enables the
2565 most basic services systemd ships by default.
2567 * Unit files' [Install] section gained a new DefaultInstance=
2568 field for defining the default instance to create if a
2569 template unit is enabled with no instance specified.
2571 * A new passive target cryptsetup-pre.target has been added
2572 that may be used by services that need to make they run and
2573 finish before the first LUKS cryptographic device is set up.
2575 * The /dev/loop-control and /dev/btrfs-control device nodes
2576 are now owned by the "disk" group by default, opening up
2577 access to this group.
2579 * systemd-coredump will now automatically generate a
2580 stack trace of all core dumps taking place on the system,
2581 based on elfutils' libdw library. This stack trace is logged
2584 * systemd-coredump may now optionally store coredumps directly
2585 on disk (in /var/lib/systemd/coredump, possibly compressed),
2586 instead of storing them unconditionally in the journal. This
2587 mode is the new default. A new configuration file
2588 /etc/systemd/coredump.conf has been added to configure this
2589 and other parameters of systemd-coredump.
2591 * coredumpctl gained a new "info" verb to show details about a
2592 specific coredump. A new switch "-1" has also been added
2593 that makes sure to only show information about the most
2594 recent entry instead of all entries. Also, as the tool is
2595 generally useful now the "systemd-" prefix of the binary
2596 name has been removed. Distributions that want to maintain
2597 compatibility with the old name should add a symlink from
2598 the old name to the new name.
2600 * journald's SplitMode= now defaults to "uid". This makes sure
2601 that unprivileged users can access their own coredumps with
2602 coredumpctl without restrictions.
2604 * New kernel command line options "systemd.wants=" (for
2605 pulling an additional unit during boot), "systemd.mask="
2606 (for masking a specific unit for the boot), and
2607 "systemd.debug-shell" (for enabling the debug shell on tty9)
2608 have been added. This is implemented in the new generator
2609 "systemd-debug-generator".
2611 * systemd-nspawn will now by default filter a couple of
2612 syscalls for containers, among them those required for
2613 kernel module loading, direct x86 IO port access, swap
2614 management, and kexec. Most importantly though
2615 open_by_handle_at() is now prohibited for containers,
2616 closing a hole similar to a recently discussed vulnerability
2617 in docker regarding access to files on file hierarchies the
2618 container should normally not have access to. Note that, for
2619 nspawn, we generally make no security claims anyway (and
2620 this is explicitly documented in the man page), so this is
2621 just a fix for one of the most obvious problems.
2623 * A new man page file-hierarchy(7) has been added that
2624 contains a minimized, modernized version of the file system
2625 layout systemd expects, similar in style to the FHS
2626 specification or hier(5). A new tool systemd-path(1) has
2627 been added to query many of these paths for the local
2630 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2631 longer done. Since the directory now has a per-user size
2632 limit, and is cleaned on logout this appears unnecessary,
2633 in particular since this now brings the lifecycle of this
2634 directory closer in line with how IPC objects are handled.
2636 * systemd.pc now exports a number of additional directories,
2637 including $libdir (which is useful to identify the library
2638 path for the primary architecture of the system), and a
2639 couple of drop-in directories.
2641 * udev's predictable network interface names now use the dev_port
2642 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2643 distinguish between ports of the same PCI function. dev_id should
2644 only be used for ports using the same HW address, hence the need
2647 * machined has been updated to export the OS version of a
2648 container (read from /etc/os-release and
2649 /usr/lib/os-release) on the bus. This is now shown in
2650 "machinectl status" for a machine.
2652 * A new service setting RestartForceExitStatus= has been
2653 added. If configured to a set of exit signals or process
2654 return values, the service will be restarted when the main
2655 daemon process exits with any of them, regardless of the
2658 * systemctl's -H switch for connecting to remote systemd
2659 machines has been extended so that it may be used to
2660 directly connect to a specific container on the
2661 host. "systemctl -H root@foobar:waldi" will now connect as
2662 user "root" to host "foobar", and then proceed directly to
2663 the container named "waldi". Note that currently you have to
2664 authenticate as user "root" for this to work, as entering
2665 containers is a privileged operation.
2667 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2668 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2669 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2670 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2671 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2672 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2673 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2674 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2675 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2676 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2677 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2678 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2680 — Berlin, 2014-07-03
2684 * As an experimental feature, udev now tries to lock the
2685 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2686 executes events for the disk or any of its partitions.
2687 Applications like partitioning programs can lock the
2688 disk device node (flock(LOCK_EX)) and claim temporary
2689 device ownership that way; udev will entirely skip all event
2690 handling for this disk and its partitions. If the disk
2691 was opened for writing, the close will trigger a partition
2692 table rescan in udev's "watch" facility, and if needed
2693 synthesize "change" events for the disk and all its partitions.
2694 This is now unconditionally enabled, and if it turns out to
2695 cause major problems, we might turn it on only for specific
2696 devices, or might need to disable it entirely. Device Mapper
2697 devices are excluded from this logic.
2699 * We temporarily dropped the "-l" switch for fsck invocations,
2700 since they collide with the flock() logic above. util-linux
2701 upstream has been changed already to avoid this conflict,
2702 and we will readd "-l" as soon as util-linux with this
2703 change has been released.
2705 * The dependency on libattr has been removed. Since a long
2706 time, the extended attribute calls have moved to glibc, and
2707 libattr is thus unnecessary.
2709 * Virtualization detection works without priviliges now. This
2710 means the systemd-detect-virt binary no longer requires
2711 CAP_SYS_PTRACE file capabilities, and our daemons can run
2712 with fewer privileges.
2714 * systemd-networkd now runs under its own "systemd-network"
2715 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2716 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2717 loses the ability to write to files owned by root this way.
2719 * Similarly, systemd-resolved now runs under its own
2720 "systemd-resolve" user with no capabilities remaining.
2722 * Similarly, systemd-bus-proxyd now runs under its own
2723 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2725 * systemd-networkd gained support for setting up "veth"
2726 virtual Ethernet devices for container connectivity, as well
2727 as GRE and VTI tunnels.
2729 * systemd-networkd will no longer automatically attempt to
2730 manually load kernel modules necessary for certain tunnel
2731 transports. Instead, it is assumed the kernel loads them
2732 automatically when required. This only works correctly on
2733 very new kernels. On older kernels, please consider adding
2734 the kernel modules to /etc/modules-load.d/ as a work-around.
2736 * The resolv.conf file systemd-resolved generates has been
2737 moved to /run/systemd/resolve/. If you have a symlink from
2738 /etc/resolv.conf, it might be necessary to correct it.
2740 * Two new service settings, ProtectHome= and ProtectSystem=,
2741 have been added. When enabled, they will make the user data
2742 (such as /home) inaccessible or read-only and the system
2743 (such as /usr) read-only, for specific services. This allows
2744 very light-weight per-service sandboxing to avoid
2745 modifications of user data or system files from
2746 services. These two new switches have been enabled for all
2747 of systemd's long-running services, where appropriate.
2749 * Socket units gained new SocketUser= and SocketGroup=
2750 settings to set the owner user and group of AF_UNIX sockets
2751 and FIFOs in the file system.
2753 * Socket units gained a new RemoveOnStop= setting. If enabled,
2754 all FIFOS and sockets in the file system will be removed
2755 when the specific socket unit is stopped.
2757 * Socket units gained a new Symlinks= setting. It takes a list
2758 of symlinks to create to file system sockets or FIFOs
2759 created by the specific Unix sockets. This is useful to
2760 manage symlinks to socket nodes with the same life-cycle as
2763 * The /dev/log socket and /dev/initctl FIFO have been moved to
2764 /run, and have been replaced by symlinks. This allows
2765 connecting to these facilities even if PrivateDevices=yes is
2766 used for a service (which makes /dev/log itself unavailable,
2767 but /run is left). This also has the benefit of ensuring
2768 that /dev only contains device nodes, directories and
2769 symlinks, and nothing else.
2771 * sd-daemon gained two new calls sd_pid_notify() and
2772 sd_pid_notifyf(). They are similar to sd_notify() and
2773 sd_notifyf(), but allow overriding of the source PID of
2774 notification messages if permissions permit this. This is
2775 useful to send notify messages on behalf of a different
2776 process (for example, the parent process). The
2777 systemd-notify tool has been updated to make use of this
2778 when sending messages (so that notification messages now
2779 originate from the shell script invoking systemd-notify and
2780 not the systemd-notify process itself. This should minimize
2781 a race where systemd fails to associate notification
2782 messages to services when the originating process already
2785 * A new "on-abnormal" setting for Restart= has been added. If
2786 set, it will result in automatic restarts on all "abnormal"
2787 reasons for a process to exit, which includes unclean
2788 signals, core dumps, timeouts and watchdog timeouts, but
2789 does not include clean and unclean exit codes or clean
2790 signals. Restart=on-abnormal is an alternative for
2791 Restart=on-failure for services that shall be able to
2792 terminate and avoid restarts on certain errors, by
2793 indicating so with an unclean exit code. Restart=on-failure
2794 or Restart=on-abnormal is now the recommended setting for
2795 all long-running services.
2797 * If the InaccessibleDirectories= service setting points to a
2798 mount point (or if there are any submounts contained within
2799 it), it is now attempted to completely unmount it, to make
2800 the file systems truly unavailable for the respective
2803 * The ReadOnlyDirectories= service setting and
2804 systemd-nspawn's --read-only parameter are now recursively
2805 applied to all submounts, too.
2807 * Mount units may now be created transiently via the bus APIs.
2809 * The support for SysV and LSB init scripts has been removed
2810 from the systemd daemon itself. Instead, it is now
2811 implemented as a generator that creates native systemd units
2812 from these scripts when needed. This enables us to remove a
2813 substantial amount of legacy code from PID 1, following the
2814 fact that many distributions only ship a very small number
2815 of LSB/SysV init scripts nowadays.
2817 * Privileged Xen (dom0) domains are not considered
2818 virtualization anymore by the virtualization detection
2819 logic. After all, they generally have unrestricted access to
2820 the hardware and usually are used to manage the unprivileged
2823 * systemd-tmpfiles gained a new "C" line type, for copying
2824 files or entire directories.
2826 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2827 lines. So far, they have been non-globbing versions of the
2828 latter, and have thus been redundant. In future, it is
2829 recommended to only use "z". "m" has hence been removed
2830 from the documentation, even though it stays supported.
2832 * A tmpfiles snippet to recreate the most basic structure in
2833 /var has been added. This is enough to create the /var/run →
2834 /run symlink and create a couple of structural
2835 directories. This allows systems to boot up with an empty or
2836 volatile /var. Of course, while with this change, the core OS
2837 now is capable with dealing with a volatile /var, not all
2838 user services are ready for it. However, we hope that sooner
2839 or later, many service daemons will be changed upstream so
2840 that they are able to automatically create their necessary
2841 directories in /var at boot, should they be missing. This is
2842 the first step to allow state-less systems that only require
2843 the vendor image for /usr to boot.
2845 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2846 empty tmpfs instance to a specific directory. This is
2847 particularly useful for making use of the automatic
2848 reconstruction of /var (see above), by passing --tmpfs=/var.
2850 * Access modes specified in tmpfiles snippets may now be
2851 prefixed with "~", which indicates that they shall be masked
2852 by whether the existing file or directory is currently
2853 writable, readable or executable at all. Also, if specified,
2854 the sgid/suid/sticky bits will be masked for all
2857 * A new passive target unit "network-pre.target" has been
2858 added which is useful for services that shall run before any
2859 network is configured, for example firewall scripts.
2861 * The "floppy" group that previously owned the /dev/fd*
2862 devices is no longer used. The "disk" group is now used
2863 instead. Distributions should probably deprecate usage of
2866 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2867 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2868 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2869 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2870 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2871 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2874 — Berlin, 2014-06-11
2878 * A new "systemd-timesyncd" daemon has been added for
2879 synchronizing the system clock across the network. It
2880 implements an SNTP client. In contrast to NTP
2881 implementations such as chrony or the NTP reference server,
2882 this only implements a client side, and does not bother with
2883 the full NTP complexity, focusing only on querying time from
2884 one remote server and synchronizing the local clock to
2885 it. Unless you intend to serve NTP to networked clients or
2886 want to connect to local hardware clocks, this simple NTP
2887 client should be more than appropriate for most
2888 installations. The daemon runs with minimal privileges, and
2889 has been hooked up with networkd to only operate when
2890 network connectivity is available. The daemon saves the
2891 current clock to disk every time a new NTP sync has been
2892 acquired, and uses this to possibly correct the system clock
2893 early at bootup, in order to accommodate for systems that
2894 lack an RTC such as the Raspberry Pi and embedded devices,
2895 and to make sure that time monotonically progresses on these
2896 systems, even if it is not always correct. To make use of
2897 this daemon, a new system user and group "systemd-timesync"
2898 needs to be created on installation of systemd.
2900 * The queue "seqnum" interface of libudev has been disabled, as
2901 it was generally incompatible with device namespacing as
2902 sequence numbers of devices go "missing" if the devices are
2903 part of a different namespace.
2905 * "systemctl list-timers" and "systemctl list-sockets" gained
2906 a --recursive switch for showing units of these types also
2907 for all local containers, similar in style to the already
2908 supported --recursive switch for "systemctl list-units".
2910 * A new RebootArgument= setting has been added for service
2911 units, which may be used to specify a kernel reboot argument
2912 to use when triggering reboots with StartLimitAction=.
2914 * A new FailureAction= setting has been added for service
2915 units which may be used to specify an operation to trigger
2916 when a service fails. This works similarly to
2917 StartLimitAction=, but unlike it, controls what is done
2918 immediately rather than only after several attempts to
2919 restart the service in question.
2921 * hostnamed got updated to also expose the kernel name,
2922 release, and version on the bus. This is useful for
2923 executing commands like hostnamectl with the -H switch.
2924 systemd-analyze makes use of this to properly display
2925 details when running non-locally.
2927 * The bootchart tool can now show cgroup information in the
2928 graphs it generates.
2930 * The CFS CPU quota cgroup attribute is now exposed for
2931 services. The new CPUQuota= switch has been added for this
2932 which takes a percentage value. Setting this will have the
2933 result that a service may never get more CPU time than the
2934 specified percentage, even if the machine is otherwise idle.
2936 * systemd-networkd learned IPIP and SIT tunnel support.
2938 * LSB init scripts exposing a dependency on $network will now
2939 get a dependency on network-online.target rather than simply
2940 network.target. This should bring LSB handling closer to
2941 what it was on SysV systems.
2943 * A new fsck.repair= kernel option has been added to control
2944 how fsck shall deal with unclean file systems at boot.
2946 * The (.ini) configuration file parser will now silently
2947 ignore sections whose name begins with "X-". This may be
2948 used to maintain application-specific extension sections in unit
2951 * machined gained a new API to query the IP addresses of
2952 registered containers. "machinectl status" has been updated
2953 to show these addresses in its output.
2955 * A new call sd_uid_get_display() has been added to the
2956 sd-login APIs for querying the "primary" session of a
2957 user. The "primary" session of the user is elected from the
2958 user's sessions and generally a graphical session is
2959 preferred over a text one.
2961 * A minimal systemd-resolved daemon has been added. It
2962 currently simply acts as a companion to systemd-networkd and
2963 manages resolv.conf based on per-interface DNS
2964 configuration, possibly supplied via DHCP. In the long run
2965 we hope to extend this into a local DNSSEC enabled DNS and
2968 * The systemd-networkd-wait-online tool is now enabled by
2969 default. It will delay network-online.target until a network
2970 connection has been configured. The tool primarily integrates
2971 with networkd, but will also make a best effort to make sense
2972 of network configuration performed in some other way.
2974 * Two new service options StartupCPUShares= and
2975 StartupBlockIOWeight= have been added that work similarly to
2976 CPUShares= and BlockIOWeight= however only apply during
2977 system startup. This is useful to prioritize certain services
2978 differently during bootup than during normal runtime.
2980 * hostnamed has been changed to prefer the statically
2981 configured hostname in /etc/hostname (unless set to
2982 'localhost' or empty) over any dynamic one supplied by
2983 dhcp. With this change, the rules for picking the hostname
2984 match more closely the rules of other configuration settings
2985 where the local administrator's configuration in /etc always
2986 overrides any other settings.
2988 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2989 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2990 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2991 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2992 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2993 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2994 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2995 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2996 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2997 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2998 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2999 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3000 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3001 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3002 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3003 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3006 — Beijing, 2014-05-28
3010 * When restoring the screen brightness at boot, stay away from
3011 the darkest setting or from the lowest 5% of the available
3012 range, depending on which is the larger value of both. This
3013 should effectively protect the user from rebooting into a
3014 black screen, should the brightness have been set to minimum
3017 * sd-login gained a new sd_machine_get_class() call to
3018 determine the class ("vm" or "container") of a machine
3019 registered with machined.
3021 * sd-login gained new calls
3022 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
3023 to query the identity of the peer of a local AF_UNIX
3024 connection. They operate similarly to their sd_pid_get_xyz()
3027 * PID 1 will now maintain a system-wide system state engine
3028 with the states "starting", "running", "degraded",
3029 "maintenance", "stopping". These states are bound to system
3030 startup, normal runtime, runtime with at least one failed
3031 service, rescue/emergency mode and system shutdown. This
3032 state is shown in the "systemctl status" output when no unit
3033 name is passed. It is useful to determine system state, in
3034 particularly when doing so for many systems or containers at
3037 * A new command "list-machines" has been added to "systemctl"
3038 that lists all local OS containers and shows their system
3039 state (see above), if systemd runs inside of them.
3041 * systemctl gained a new "-r" switch to recursively enumerate
3042 units on all local containers, when used with the
3043 "list-unit" command (which is the default one that is
3044 executed when no parameters are specified).
3046 * The GPT automatic partition discovery logic will now honour
3047 two GPT partition flags: one may be set on a partition to
3048 cause it to be mounted read-only, and the other may be set
3049 on a partition to ignore it during automatic discovery.
3051 * Two new GPT type UUIDs have been added for automatic root
3052 partition discovery, for 32-bit and 64-bit ARM. This is not
3053 particularly useful for discovering the root directory on
3054 these architectures during bare-metal boots (since UEFI is
3055 not common there), but still very useful to allow booting of
3056 ARM disk images in nspawn with the -i option.
3058 * MAC addresses of interfaces created with nspawn's
3059 --network-interface= switch will now be generated from the
3060 machine name, and thus be stable between multiple invocations
3063 * logind will now automatically remove all IPC objects owned
3064 by a user if she or he fully logs out. This makes sure that
3065 users who are logged out cannot continue to consume IPC
3066 resources. This covers SysV memory, semaphores and message
3067 queues as well as POSIX shared memory and message
3068 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3069 limits. With this functionality, that is corrected. This may
3070 be turned off by using the RemoveIPC= switch of logind.conf.
3072 * The systemd-machine-id-setup and tmpfiles tools gained a
3073 --root= switch to operate on a specific root directory,
3076 * journald can now forward logged messages to the TTYs of all
3077 logged in users ("wall"). This is the default for all
3078 emergency messages now.
3080 * A new tool systemd-journal-remote has been added to stream
3081 journal log messages across the network.
3083 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3084 controller trees are mounted into it. Note that the
3085 directories mounted beneath it are not read-only. This is a
3086 security measure and is particularly useful because glibc
3087 actually includes a search logic to pick any tmpfs it can
3088 find to implement shm_open() if /dev/shm is not available
3089 (which it might very well be in namespaced setups).
3091 * machinectl gained a new "poweroff" command to cleanly power
3092 down a local OS container.
3094 * The PrivateDevices= unit file setting will now also drop the
3095 CAP_MKNOD capability from the capability bound set, and
3096 imply DevicePolicy=closed.
3098 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3099 comprehensively on all long-running systemd services where
3100 this is appropriate.
3102 * systemd-udevd will now run in a disassociated mount
3103 namespace. To mount directories from udev rules, make sure to
3104 pull in mount units via SYSTEMD_WANTS properties.
3106 * The kdbus support gained support for uploading policy into
3107 the kernel. sd-bus gained support for creating "monitoring"
3108 connections that can eavesdrop into all bus communication
3109 for debugging purposes.
3111 * Timestamps may now be specified in seconds since the UNIX
3112 epoch Jan 1st, 1970 by specifying "@" followed by the value
3115 * Native tcpwrap support in systemd has been removed. tcpwrap
3116 is old code, not really maintained anymore and has serious
3117 shortcomings, and better options such as firewalls
3118 exist. For setups that require tcpwrap usage, please
3119 consider invoking your socket-activated service via tcpd,
3120 like on traditional inetd.
3122 * A new system.conf configuration option
3123 DefaultTimerAccuracySec= has been added that controls the
3124 default AccuracySec= setting of .timer units.
3126 * Timer units gained a new WakeSystem= switch. If enabled,
3127 timers configured this way will cause the system to resume
3128 from system suspend (if the system supports that, which most
3131 * Timer units gained a new Persistent= switch. If enabled,
3132 timers configured this way will save to disk when they have
3133 been last triggered. This information is then used on next
3134 reboot to possible execute overdue timer events, that
3135 could not take place because the system was powered off.
3136 This enables simple anacron-like behaviour for timer units.
3138 * systemctl's "list-timers" will now also list the time a
3139 timer unit was last triggered in addition to the next time
3140 it will be triggered.
3142 * systemd-networkd will now assign predictable IPv4LL
3143 addresses to its local interfaces.
3145 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3146 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3147 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3148 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3149 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3150 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3151 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3152 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3155 — Berlin, 2014-03-25
3159 * A new unit file setting RestrictAddressFamilies= has been
3160 added to restrict which socket address families unit
3161 processes gain access to. This takes address family names
3162 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3163 attack surface of services via exotic protocol stacks. This
3164 is built on seccomp system call filters.
3166 * Two new unit file settings RuntimeDirectory= and
3167 RuntimeDirectoryMode= have been added that may be used to
3168 manage a per-daemon runtime directories below /run. This is
3169 an alternative for setting up directory permissions with
3170 tmpfiles snippets, and has the advantage that the runtime
3171 directory's lifetime is bound to the daemon runtime and that
3172 the daemon starts up with an empty directory each time. This
3173 is particularly useful when writing services that drop
3174 privileges using the User= or Group= setting.
3176 * The DeviceAllow= unit setting now supports globbing for
3177 matching against device group names.
3179 * The systemd configuration file system.conf gained new
3180 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3181 DefaultMemoryAccounting= to globally turn on/off accounting
3182 for specific resources (cgroups) for all units. These
3183 settings may still be overridden individually in each unit
3186 * systemd-gpt-auto-generator is now able to discover /srv and
3187 root partitions in addition to /home and swap partitions. It
3188 also supports LUKS-encrypted partitions now. With this in
3189 place, automatic discovery of partitions to mount following
3190 the Discoverable Partitions Specification
3191 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3192 is now a lot more complete. This allows booting without
3193 /etc/fstab and without root= on the kernel command line on
3194 systems prepared appropriately.
3196 * systemd-nspawn gained a new --image= switch which allows
3197 booting up disk images and Linux installations on any block
3198 device that follow the Discoverable Partitions Specification
3199 (see above). This means that installations made with
3200 appropriately updated installers may now be started and
3201 deployed using container managers, completely
3202 unmodified. (We hope that libvirt-lxc will add support for
3203 this feature soon, too.)
3205 * systemd-nspawn gained a new --network-macvlan= setting to
3206 set up a private macvlan interface for the
3207 container. Similarly, systemd-networkd gained a new
3208 Kind=macvlan setting in .netdev files.
3210 * systemd-networkd now supports configuring local addresses
3213 * A new tool systemd-network-wait-online has been added to
3214 synchronously wait for network connectivity using
3217 * The sd-bus.h bus API gained a new sd_bus_track object for
3218 tracking the life-cycle of bus peers. Note that sd-bus.h is
3219 still not a public API though (unless you specify
3220 --enable-kdbus on the configure command line, which however
3221 voids your warranty and you get no API stability guarantee).
3223 * The $XDG_RUNTIME_DIR runtime directories for each user are
3224 now individual tmpfs instances, which has the benefit of
3225 introducing separate pools for each user, with individual
3226 size limits, and thus making sure that unprivileged clients
3227 can no longer negatively impact the system or other users by
3228 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3229 RuntimeDirectorySize= has been introduced that allows
3230 controlling the default size limit for all users. It
3231 defaults to 10% of the available physical memory. This is no
3232 replacement for quotas on tmpfs though (which the kernel
3233 still does not support), as /dev/shm and /tmp are still
3234 shared resources used by both the system and unprivileged
3237 * logind will now automatically turn off automatic suspending
3238 on laptop lid close when more than one display is
3239 connected. This was previously expected to be implemented
3240 individually in desktop environments (such as GNOME),
3241 however has been added to logind now, in order to fix a
3242 boot-time race where a desktop environment might not have
3243 been started yet and thus not been able to take an inhibitor
3244 lock at the time where logind already suspends the system
3245 due to a closed lid.
3247 * logind will now wait at least 30s after each system
3248 suspend/resume cycle, and 3min after system boot before
3249 suspending the system due to a closed laptop lid. This
3250 should give USB docking stations and similar enough time to
3251 be probed and configured after system resume and boot in
3252 order to then act as suspend blocker.
3254 * systemd-run gained a new --property= setting which allows
3255 initialization of resource control properties (and others)
3256 for the created scope or service unit. Example: "systemd-run
3257 --property=BlockIOWeight=10 updatedb" may be used to run
3258 updatedb at a low block IO scheduling weight.
3260 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3261 now also work in --scope mode.
3263 * When systemd is compiled with kdbus support, basic support
3264 for enforced policies is now in place. (Note that enabling
3265 kdbus still voids your warranty and no API compatibility
3268 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3269 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3270 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3271 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3272 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3273 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3274 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3275 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3276 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3277 Zbigniew Jędrzejewski-Szmek
3279 — Berlin, 2014-03-12
3283 * systemd will now relabel /dev after loading the SMACK policy
3284 according to SMACK rules.
3286 * A new unit file option AppArmorProfile= has been added to
3287 set the AppArmor profile for the processes of a unit.
3289 * A new condition check ConditionArchitecture= has been added
3290 to conditionalize units based on the system architecture, as
3291 reported by uname()'s "machine" field.
3293 * systemd-networkd now supports matching on the system
3294 virtualization, architecture, kernel command line, host name
3297 * logind is now a lot more aggressive when suspending the
3298 machine due to a closed laptop lid. Instead of acting only
3299 on the lid close action, it will continuously watch the lid
3300 status and act on it. This is useful for laptops where the
3301 power button is on the outside of the chassis so that it can
3302 be reached without opening the lid (such as the Lenovo
3303 Yoga). On those machines, logind will now immediately
3304 re-suspend the machine if the power button has been
3305 accidentally pressed while the laptop was suspended and in a
3306 backpack or similar.
3308 * logind will now watch SW_DOCK switches and inhibit reaction
3309 to the lid switch if it is pressed. This means that logind
3310 will not suspend the machine anymore if the lid is closed
3311 and the system is docked, if the laptop supports SW_DOCK
3312 notifications via the input layer. Note that ACPI docking
3313 stations do not generate this currently. Also note that this
3314 logic is usually not fully sufficient and Desktop
3315 Environments should take a lid switch inhibitor lock when an
3316 external display is connected, as systemd will not watch
3319 * nspawn will now make use of the devices cgroup controller by
3320 default, and only permit creation of and access to the usual
3321 API device nodes like /dev/null or /dev/random, as well as
3322 access to (but not creation of) the pty devices.
3324 * We will now ship a default .network file for
3325 systemd-networkd that automatically configures DHCP for
3326 network interfaces created by nspawn's --network-veth or
3327 --network-bridge= switches.
3329 * systemd will now understand the usual M, K, G, T suffixes
3330 according to SI conventions (i.e. to the base 1000) when
3331 referring to throughput and hardware metrics. It will stay
3332 with IEC conventions (i.e. to the base 1024) for software
3333 metrics, according to what is customary according to
3334 Wikipedia. We explicitly document which base applies for
3335 each configuration option.
3337 * The DeviceAllow= setting in unit files now supports a syntax
3338 to whitelist an entire group of devices node majors at once,
3339 based on the /proc/devices listing. For example, with the
3340 string "char-pts", it is now possible to whitelist all
3341 current and future pseudo-TTYs at once.
3343 * sd-event learned a new "post" event source. Event sources of
3344 this type are triggered by the dispatching of any event
3345 source of a type that is not "post". This is useful for
3346 implementing clean-up and check event sources that are
3347 triggered by other work being done in the program.
3349 * systemd-networkd is no longer statically enabled, but uses
3350 the usual [Install] sections so that it can be
3351 enabled/disabled using systemctl. It still is enabled by
3354 * When creating a veth interface pair with systemd-nspawn, the
3355 host side will now be prefixed with "vb-" if
3356 --network-bridge= is used, and with "ve-" if --network-veth
3357 is used. This way, it is easy to distinguish these cases on
3358 the host, for example to apply different configuration to
3359 them with systemd-networkd.
3361 * The compatibility libraries for libsystemd-journal.so,
3362 libsystem-id128.so, libsystemd-login.so and
3363 libsystemd-daemon.so do not make use of IFUNC
3364 anymore. Instead, we now build libsystemd.so multiple times
3365 under these alternative names. This means that the footprint
3366 is drastically increased, but given that these are
3367 transitional compatibility libraries, this should not matter
3368 much. This change has been made necessary to support the ARM
3369 platform for these compatibility libraries, as the ARM
3370 toolchain is not really at the same level as the toolchain
3371 for other architectures like x86 and does not support
3372 IFUNC. Please make sure to use --enable-compat-libs only
3373 during a transitional period!
3375 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3376 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3377 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3378 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3379 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3380 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3381 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3382 Zbigniew Jędrzejewski-Szmek
3384 — Berlin, 2014-02-24
3388 * A new component "systemd-networkd" has been added that can
3389 be used to configure local network interfaces statically or
3390 via DHCP. It is capable of bringing up bridges, VLANs, and
3391 bonding. Currently, no hook-ups for interactive network
3392 configuration are provided. Use this for your initrd,
3393 container, embedded, or server setup if you need a simple,
3394 yet powerful, network configuration solution. This
3395 configuration subsystem is quite nifty, as it allows wildcard
3396 hotplug matching in interfaces. For example, with a single
3397 configuration snippet, you can configure that all Ethernet
3398 interfaces showing up are automatically added to a bridge,
3399 or similar. It supports link-sensing and more.
3401 * A new tool "systemd-socket-proxyd" has been added which can
3402 act as a bidirectional proxy for TCP sockets. This is
3403 useful for adding socket activation support to services that
3404 do not actually support socket activation, including virtual
3405 machines and the like.
3407 * Add a new tool to save/restore rfkill state on
3410 * Save/restore state of keyboard backlights in addition to
3411 display backlights on shutdown/boot.
3413 * udev learned a new SECLABEL{} construct to label device
3414 nodes with a specific security label when they appear. For
3415 now, only SECLABEL{selinux} is supported, but the syntax is
3416 prepared for additional security frameworks.
3418 * udev gained a new scheme to configure link-level attributes
3419 from files in /etc/systemd/network/*.link. These files can
3420 match against MAC address, device path, driver name and type,
3421 and will apply attributes like the naming policy, link speed,
3422 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3423 address assignment policy (randomized, ...).
3425 * The configuration of network interface naming rules for
3426 "permanent interface names" has changed: a new NamePolicy=
3427 setting in the [Link] section of .link files determines the
3428 priority of possible naming schemes (onboard, slot, MAC,
3429 path). The default value of this setting is determined by
3430 /usr/lib/net/links/99-default.link. Old
3431 80-net-name-slot.rules udev configuration file has been
3432 removed, so local configuration overriding this file should
3433 be adapated to override 99-default.link instead.
3435 * When the User= switch is used in a unit file, also
3436 initialize $SHELL= based on the user database entry.
3438 * systemd no longer depends on libdbus. All communication is
3439 now done with sd-bus, systemd's low-level bus library
3442 * kdbus support has been added to PID 1 itself. When kdbus is
3443 enabled, this causes PID 1 to set up the system bus and
3444 enable support for a new ".busname" unit type that
3445 encapsulates bus name activation on kdbus. It works a little
3446 bit like ".socket" units, except for bus names. A new
3447 generator has been added that converts classic dbus1 service
3448 activation files automatically into native systemd .busname
3451 * sd-bus: add a light-weight vtable implementation that allows
3452 defining objects on the bus with a simple static const
3453 vtable array of its methods, signals and properties.
3455 * systemd will not generate or install static dbus
3456 introspection data anymore to /usr/share/dbus-1/interfaces,
3457 as the precise format of these files is unclear, and
3458 nothing makes use of it.
3460 * A proxy daemon is now provided to proxy clients connecting
3461 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3462 compatibility with classic D-Bus.
3464 * A bus driver implementation has been added that supports the
3465 classic D-Bus bus driver calls on kdbus, also for
3466 compatibility purposes.
3468 * A new API "sd-event.h" has been added that implements a
3469 minimal event loop API built around epoll. It provides a
3470 couple of features that direct epoll usage is lacking:
3471 prioritization of events, scales to large numbers of timer
3472 events, per-event timer slack (accuracy), system-wide
3473 coalescing of timer events, exit handlers, watchdog
3474 supervision support using systemd's sd_notify() API, child
3477 * A new API "sd-rntl.h" has been added that provides an API
3478 around the route netlink interface of the kernel, similar in
3479 style to "sd-bus.h".
3481 * A new API "sd-dhcp-client.h" has been added that provides a
3482 small DHCPv4 client-side implementation. This is used by
3485 * There is a new kernel command line option
3486 "systemd.restore_state=0|1". When set to "0", none of the
3487 systemd tools will restore saved runtime state to hardware
3488 devices. More specifically, the rfkill and backlight states
3491 * The FsckPassNo= compatibility option in mount/service units
3492 has been removed. The fstab generator will now add the
3493 necessary dependencies automatically, and does not require
3494 PID1's support for that anymore.
3496 * journalctl gained a new switch, --list-boots, that lists
3497 recent boots with their times and boot IDs.
3499 * The various tools like systemctl, loginctl, timedatectl,
3500 busctl, systemd-run, ... have gained a new switch "-M" to
3501 connect to a specific, local OS container (as direct
3502 connection, without requiring SSH). This works on any
3503 container that is registered with machined, such as those
3504 created by libvirt-lxc or nspawn.
3506 * systemd-run and systemd-analyze also gained support for "-H"
3507 to connect to remote hosts via SSH. This is particularly
3508 useful for systemd-run because it enables queuing of jobs
3509 onto remote systems.
3511 * machinectl gained a new command "login" to open a getty
3512 login in any local container. This works with any container
3513 that is registered with machined (such as those created by
3514 libvirt-lxc or nspawn), and which runs systemd inside.
3516 * machinectl gained a new "reboot" command that may be used to
3517 trigger a reboot on a specific container that is registered
3518 with machined. This works on any container that runs an init
3519 system of some kind.
3521 * systemctl gained a new "list-timers" command to print a nice
3522 listing of installed timer units with the times they elapse
3525 * Alternative reboot() parameters may now be specified on the
3526 "systemctl reboot" command line and are passed to the
3527 reboot() system call.
3529 * systemctl gained a new --job-mode= switch to configure the
3530 mode to queue a job with. This is a more generic version of
3531 --fail, --irreversible, and --ignore-dependencies, which are
3532 still available but not advertised anymore.
3534 * /etc/systemd/system.conf gained new settings to configure
3535 various default timeouts of units, as well as the default
3536 start limit interval and burst. These may still be overridden
3539 * PID1 will now export on the bus profile data of the security
3540 policy upload process (such as the SELinux policy upload to
3543 * journald: when forwarding logs to the console, include
3544 timestamps (following the setting in
3545 /sys/module/printk/parameters/time).
3547 * OnCalendar= in timer units now understands the special
3548 strings "yearly" and "annually". (Both are equivalent)
3550 * The accuracy of timer units is now configurable with the new
3551 AccuracySec= setting. It defaults to 1min.
3553 * A new dependency type JoinsNamespaceOf= has been added that
3554 allows running two services within the same /tmp and network
3555 namespace, if PrivateNetwork= or PrivateTmp= are used.
3557 * A new command "cat" has been added to systemctl. It outputs
3558 the original unit file of a unit, and concatenates the
3559 contents of additional "drop-in" unit file snippets, so that
3560 the full configuration is shown.
3562 * systemctl now supports globbing on the various "list-xyz"
3563 commands, like "list-units" or "list-sockets", as well as on
3564 those commands which take multiple unit names.
3566 * journalctl's --unit= switch gained support for globbing.
3568 * All systemd daemons now make use of the watchdog logic so
3569 that systemd automatically notices when they hang.
3571 * If the $container_ttys environment variable is set,
3572 getty-generator will automatically spawn a getty for each
3573 listed tty. This is useful for container managers to request
3574 login gettys to be spawned on as many ttys as needed.
3576 * %h, %s, %U specifier support is not available anymore when
3577 used in unit files for PID 1. This is because NSS calls are
3578 not safe from PID 1. They stay available for --user
3579 instances of systemd, and as special case for the root user.
3581 * loginctl gained a new "--no-legend" switch to turn off output
3584 * The "sd-login.h" API gained three new calls:
3585 sd_session_is_remote(), sd_session_get_remote_user(),
3586 sd_session_get_remote_host() to query information about
3589 * The udev hardware database now also carries vendor/product
3590 information of SDIO devices.
3592 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3593 determine whether watchdog notifications are requested by
3596 * Socket-activated per-connection services now include a
3597 short description of the connection parameters in the
3600 * tmpfiles gained a new "--boot" option. When this is not used,
3601 only lines where the command character is not suffixed with
3602 "!" are executed. When this option is specified, those
3603 options are executed too. This partitions tmpfiles
3604 directives into those that can be safely executed at any
3605 time, and those which should be run only at boot (for
3606 example, a line that creates /run/nologin).
3608 * A new API "sd-resolve.h" has been added which provides a simple
3609 asynchronous wrapper around glibc NSS host name resolution
3610 calls, such as getaddrinfo(). In contrast to glibc's
3611 getaddrinfo_a(), it does not use signals. In contrast to most
3612 other asynchronous name resolution libraries, this one does
3613 not reimplement DNS, but reuses NSS, so that alternate
3614 host name resolution systems continue to work, such as mDNS,
3615 LDAP, etc. This API is based on libasyncns, but it has been
3616 cleaned up for inclusion in systemd.
3618 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3619 "sd-daemon.h" are no longer found in individual libraries
3620 libsystemd-journal.so, libsystemd-login.so,
3621 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3622 merged them into a single library, libsystemd.so, which
3623 provides all symbols. The reason for this is cyclic
3624 dependencies, as these libraries tend to use each other's
3625 symbols. So far, we have managed to workaround that by linking
3626 a copy of a good part of our code into each of these
3627 libraries again and again, which, however, makes certain
3628 things hard to do, like sharing static variables. Also, it
3629 substantially increases footprint. With this change, there
3630 is only one library for the basic APIs systemd
3631 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3632 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3633 library as well, however are subject to the --enable-kdbus
3634 switch (see below). Note that "sd-dhcp-client.h" is not part
3635 of this library (this is because it only consumes, never
3636 provides, services of/to other APIs). To make the transition
3637 easy from the separate libraries to the unified one, we
3638 provide the --enable-compat-libs compile-time switch which
3639 will generate stub libraries that are compatible with the
3640 old ones but redirect all calls to the new one.
3642 * All of the kdbus logic and the new APIs "sd-bus.h",
3643 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3644 and "sd-utf8.h" are compile-time optional via the
3645 "--enable-kdbus" switch, and they are not compiled in by
3646 default. To make use of kdbus, you have to explicitly enable
3647 the switch. Note however, that neither the kernel nor the
3648 userspace API for all of this is considered stable yet. We
3649 want to maintain the freedom to still change the APIs for
3650 now. By specifying this build-time switch, you acknowledge
3651 that you are aware of the instability of the current
3654 * Also, note that while kdbus is pretty much complete,
3655 it lacks one thing: proper policy support. This means you
3656 can build a fully working system with all features; however,
3657 it will be highly insecure. Policy support will be added in
3658 one of the next releases, at the same time that we will
3659 declare the APIs stable.
3661 * When the kernel command line argument "kdbus" is specified,
3662 systemd will automatically load the kdbus.ko kernel module. At
3663 this stage of development, it is only useful for testing kdbus
3664 and should not be used in production. Note: if "--enable-kdbus"
3665 is specified, and the kdbus.ko kernel module is available, and
3666 "kdbus" is added to the kernel command line, the entire system
3667 runs with kdbus instead of dbus-daemon, with the above mentioned
3668 problem of missing the system policy enforcement. Also a future
3669 version of kdbus.ko or a newer systemd will not be compatible with
3670 each other, and will unlikely be able to boot the machine if only
3671 one of them is updated.
3673 * systemctl gained a new "import-environment" command which
3674 uploads the caller's environment (or parts thereof) into the
3675 service manager so that it is inherited by services started
3676 by the manager. This is useful to upload variables like
3677 $DISPLAY into the user service manager.
3679 * A new PrivateDevices= switch has been added to service units
3680 which allows running a service with a namespaced /dev
3681 directory that does not contain any device nodes for
3682 physical devices. More specifically, it only includes devices
3683 such as /dev/null, /dev/urandom, and /dev/zero which are API
3686 * logind has been extended to support behaviour like VT
3687 switching on seats that do not support a VT. This makes
3688 multi-session available on seats that are not the first seat
3689 (seat0), and on systems where kernel support for VTs has
3690 been disabled at compile-time.
3692 * If a process holds a delay lock for system sleep or shutdown
3693 and fails to release it in time, we will now log its
3694 identity. This makes it easier to identify processes that
3695 cause slow suspends or power-offs.
3697 * When parsing /etc/crypttab, support for a new key-slot=
3698 option as supported by Debian is added. It allows indicating
3699 which LUKS slot to use on disk, speeding up key loading.
3701 * The sd_journald_sendv() API call has been checked and
3702 officially declared to be async-signal-safe so that it may
3703 be invoked from signal handlers for logging purposes.
3705 * Boot-time status output is now enabled automatically after a
3706 short timeout if boot does not progress, in order to give
3707 the user an indication what she or he is waiting for.
3709 * The boot-time output has been improved to show how much time
3710 remains until jobs expire.
3712 * The KillMode= switch in service units gained a new possible
3713 value "mixed". If set, and the unit is shut down, then the
3714 initial SIGTERM signal is sent only to the main daemon
3715 process, while the following SIGKILL signal is sent to
3716 all remaining processes of the service.
3718 * When a scope unit is registered, a new property "Controller"
3719 may be set. If set to a valid bus name, systemd will send a
3720 RequestStop() signal to this name when it would like to shut
3721 down the scope. This may be used to hook manager logic into
3722 the shutdown logic of scope units. Also, scope units may now
3723 be put in a special "abandoned" state, in which case the
3724 manager process which created them takes no further
3725 responsibilities for it.
3727 * When reading unit files, systemd will now verify
3728 the access mode of these files, and warn about certain
3729 suspicious combinations. This has been added to make it
3730 easier to track down packaging bugs where unit files are
3731 marked executable or world-writable.
3733 * systemd-nspawn gained a new "--setenv=" switch to set
3734 container-wide environment variables. The similar option in
3735 systemd-activate was renamed from "--environment=" to
3736 "--setenv=" for consistency.
3738 * systemd-nspawn has been updated to create a new kdbus domain
3739 for each container that is invoked, thus allowing each
3740 container to have its own set of system and user buses,
3741 independent of the host.
3743 * systemd-nspawn gained a new --drop-capability= switch to run
3744 the container with less capabilities than the default. Both
3745 --drop-capability= and --capability= now take the special
3746 string "all" for dropping or keeping all capabilities.
3748 * systemd-nspawn gained new switches for executing containers
3749 with specific SELinux labels set.
3751 * systemd-nspawn gained a new --quiet switch to not generate
3752 any additional output but the container's own console
3755 * systemd-nspawn gained a new --share-system switch to run a
3756 container without PID namespacing enabled.
3758 * systemd-nspawn gained a new --register= switch to control
3759 whether the container is registered with systemd-machined or
3760 not. This is useful for containers that do not run full
3761 OS images, but only specific apps.
3763 * systemd-nspawn gained a new --keep-unit which may be used
3764 when invoked as the only program from a service unit, and
3765 results in registration of the unit service itself in
3766 systemd-machined, instead of a newly opened scope unit.
3768 * systemd-nspawn gained a new --network-interface= switch for
3769 moving arbitrary interfaces to the container. The new
3770 --network-veth switch creates a virtual Ethernet connection
3771 between host and container. The new --network-bridge=
3772 switch then allows assigning the host side of this virtual
3773 Ethernet connection to a bridge device.
3775 * systemd-nspawn gained a new --personality= switch for
3776 setting the kernel personality for the container. This is
3777 useful when running a 32-bit container on a 64-bit host. A
3778 similar option Personality= is now also available for service
3781 * logind will now also track a "Desktop" identifier for each
3782 session which encodes the desktop environment of it. This is
3783 useful for desktop environments that want to identify
3784 multiple running sessions of itself easily.
3786 * A new SELinuxContext= setting for service units has been
3787 added that allows setting a specific SELinux execution
3788 context for a service.
3790 * Most systemd client tools will now honour $SYSTEMD_LESS for
3791 settings of the "less" pager. By default, these tools will
3792 override $LESS to allow certain operations to work, such as
3793 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3794 influence this logic.
3796 * systemd's "seccomp" hook-up has been changed to make use of
3797 the libseccomp library instead of using its own
3798 implementation. This has benefits for portability among
3801 * For usage together with SystemCallFilter=, a new
3802 SystemCallErrorNumber= setting has been introduced that
3803 allows configuration of a system error number to be returned
3804 on filtered system calls, instead of immediately killing the
3805 process. Also, SystemCallArchitectures= has been added to
3806 limit access to system calls of a particular architecture
3807 (in order to turn off support for unused secondary
3808 architectures). There is also a global
3809 SystemCallArchitectures= setting in system.conf now to turn
3810 off support for non-native system calls system-wide.
3812 * systemd requires a kernel with a working name_to_handle_at(),
3813 please see the kernel config requirements in the README file.
3815 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3816 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3817 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3818 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3819 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3820 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3821 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3822 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3823 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3824 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3825 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3826 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3827 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3828 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3829 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3830 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3831 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3832 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3833 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3834 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3835 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3836 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3837 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3838 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3840 — Berlin, 2014-02-20
3844 * logind has gained support for facilitating privileged input
3845 and drm device access for unprivileged clients. This work is
3846 useful to allow Wayland display servers (and similar
3847 programs, such as kmscon) to run under the user's ID and
3848 access input and drm devices which are normally
3849 protected. When this is used (and the kernel is new enough)
3850 logind will "mute" IO on the file descriptors passed to
3851 Wayland as long as it is in the background and "unmute" it
3852 if it returns into the foreground. This allows secure
3853 session switching without allowing background sessions to
3854 eavesdrop on input and display data. This also introduces
3855 session switching support if VT support is turned off in the
3856 kernel, and on seats that are not seat0.
3858 * A new kernel command line option luks.options= is understood
3859 now which allows specifying LUKS options for usage for LUKS
3860 encrypted partitions specified with luks.uuid=.
3862 * tmpfiles.d(5) snippets may now use specifier expansion in
3863 path names. More specifically %m, %b, %H, %v, are now
3864 replaced by the local machine id, boot id, hostname, and
3865 kernel version number.
3867 * A new tmpfiles.d(5) command "m" has been introduced which
3868 may be used to change the owner/group/access mode of a file
3869 or directory if it exists, but do nothing if it does not.
3871 * This release removes high-level support for the
3872 MemorySoftLimit= cgroup setting. The underlying kernel
3873 cgroup attribute memory.soft_limit= is currently badly
3874 designed and likely to be removed from the kernel API in its
3875 current form, hence we should not expose it for now.
3877 * The memory.use_hierarchy cgroup attribute is now enabled for
3878 all cgroups systemd creates in the memory cgroup
3879 hierarchy. This option is likely to be come the built-in
3880 default in the kernel anyway, and the non-hierarchical mode
3881 never made much sense in the intrinsically hierarchical
3884 * A new field _SYSTEMD_SLICE= is logged along with all journal
3885 messages containing the slice a message was generated
3886 from. This is useful to allow easy per-customer filtering of
3887 logs among other things.
3889 * systemd-journald will no longer adjust the group of journal
3890 files it creates to the "systemd-journal" group. Instead we
3891 rely on the journal directory to be owned by the
3892 "systemd-journal" group, and its setgid bit set, so that the
3893 kernel file system layer will automatically enforce that
3894 journal files inherit this group assignment. The reason for
3895 this change is that we cannot allow NSS look-ups from
3896 journald which would be necessary to resolve
3897 "systemd-journal" to a numeric GID, because this might
3898 create deadlocks if NSS involves synchronous queries to
3899 other daemons (such as nscd, or sssd) which in turn are
3900 logging clients of journald and might block on it, which
3901 would then dead lock. A tmpfiles.d(5) snippet included in
3902 systemd will make sure the setgid bit and group are
3903 properly set on the journal directory if it exists on every
3904 boot. However, we recommend adjusting it manually after
3905 upgrades too (or from RPM scriptlets), so that the change is
3906 not delayed until next reboot.
3908 * Backlight and random seed files in /var/lib/ have moved into
3909 the /var/lib/systemd/ directory, in order to centralize all
3910 systemd generated files in one directory.
3912 * Boot time performance measurements (as displayed by
3913 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3914 performance information if that's available to determine how
3915 much time BIOS and boot loader initialization required. With
3916 a sufficiently new BIOS you hence no longer need to boot
3917 with Gummiboot to get access to such information.
3919 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3920 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3921 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3922 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3923 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3924 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3925 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3927 — Berlin, 2013-10-02
3931 * The Restart= option for services now understands a new
3932 on-watchdog setting, which will restart the service
3933 automatically if the service stops sending out watchdog keep
3934 alive messages (as configured with WatchdogSec=).
3936 * The getty generator (which is responsible for bringing up a
3937 getty on configured serial consoles) will no longer only
3938 start a getty on the primary kernel console but on all
3939 others, too. This makes the order in which console= is
3940 specified on the kernel command line less important.
3942 * libsystemd-logind gained a new sd_session_get_vt() call to
3943 retrieve the VT number of a session.
3945 * If the option "tries=0" is set for an entry of /etc/crypttab
3946 its passphrase is queried indefinitely instead of any
3947 maximum number of tries.
3949 * If a service with a configure PID file terminates its PID
3950 file will now be removed automatically if it still exists
3951 afterwards. This should put an end to stale PID files.
3953 * systemd-run will now also take relative binary path names
3954 for execution and no longer insists on absolute paths.
3956 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3957 paths that are optionally prefixed with "-" to indicate that
3958 it should not be considered a failure if they do not exist.
3960 * journalctl -o (and similar commands) now understands a new
3961 output mode "short-precise", it is similar to "short" but
3962 shows timestamps with usec accuracy.
3964 * The option "discard" (as known from Debian) is now
3965 synonymous to "allow-discards" in /etc/crypttab. In fact,
3966 "discard" is preferred now (since it is easier to remember
3969 * Some licensing clean-ups were made, so that more code is now
3970 LGPL-2.1 licensed than before.
3972 * A minimal tool to save/restore the display backlight
3973 brightness across reboots has been added. It will store the
3974 backlight setting as late as possible at shutdown, and
3975 restore it as early as possible during reboot.
3977 * A logic to automatically discover and enable home and swap
3978 partitions on GPT disks has been added. With this in place
3979 /etc/fstab becomes optional for many setups as systemd can
3980 discover certain partitions located on the root disk
3981 automatically. Home partitions are recognized under their
3982 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3983 partitions are recognized under their GPT type ID
3984 0657fd6da4ab43c484e50933c84b4f4f.
3986 * systemd will no longer pass any environment from the kernel
3987 or initrd to system services. If you want to set an
3988 environment for all services, do so via the kernel command
3989 line systemd.setenv= assignment.
3991 * The systemd-sysctl tool no longer natively reads the file
3992 /etc/sysctl.conf. If desired, the file should be symlinked
3993 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3994 legacy support by a symlink rather than built-in code, it
3995 also makes the otherwise hidden order of application of the
3996 different files visible. (Note that this partly reverts to a
3997 pre-198 application order of sysctl knobs!)
3999 * The "systemctl set-log-level" and "systemctl dump" commands
4000 have been moved to systemd-analyze.
4002 * systemd-run learned the new --remain-after-exit switch,
4003 which causes the scope unit not to be cleaned up
4004 automatically after the process terminated.
4006 * tmpfiles learned a new --exclude-prefix= switch to exclude
4007 certain paths from operation.
4009 * journald will now automatically flush all messages to disk
4010 as soon as a message at the log level CRIT, ALERT or EMERG
4013 Contributions from: Andrew Cook, Brandon Philips, Christian
4014 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
4015 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
4016 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
4017 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
4018 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
4019 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
4020 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
4021 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
4022 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
4023 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
4024 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
4025 William Giokas, Zbigniew Jędrzejewski-Szmek
4027 — Berlin, 2013-09-13
4031 * The documentation has been updated to cover the various new
4032 concepts introduced with 205.
4034 * Unit files now understand the new %v specifier which
4035 resolves to the kernel version string as returned by "uname
4038 * systemctl now supports filtering the unit list output by
4039 load state, active state and sub state, using the new
4042 * "systemctl status" will now show the results of the
4043 condition checks (like ConditionPathExists= and similar) of
4044 the last start attempts of the unit. They are also logged to
4047 * "journalctl -b" may now be used to look for boot output of a
4048 specific boot. Try "journalctl -b -1" for the previous boot,
4049 but the syntax is substantially more powerful.
4051 * "journalctl --show-cursor" has been added which prints the
4052 cursor string the last shown log line. This may then be used
4053 with the new "journalctl --after-cursor=" switch to continue
4054 browsing logs from that point on.
4056 * "journalctl --force" may now be used to force regeneration
4059 * Creation of "dead" device nodes has been moved from udev
4060 into kmod and tmpfiles. Previously, udev would read the kmod
4061 databases to pre-generate dead device nodes based on meta
4062 information contained in kernel modules, so that these would
4063 be auto-loaded on access rather then at boot. As this
4064 does not really have much to do with the exposing actual
4065 kernel devices to userspace this has always been slightly
4066 alien in the udev codebase. Following the new scheme kmod
4067 will now generate a runtime snippet for tmpfiles from the
4068 module meta information and it now is tmpfiles' job to the
4069 create the nodes. This also allows overriding access and
4070 other parameters for the nodes using the usual tmpfiles
4071 facilities. As side effect this allows us to remove the
4072 CAP_SYS_MKNOD capability bit from udevd entirely.
4074 * logind's device ACLs may now be applied to these "dead"
4075 devices nodes too, thus finally allowing managed access to
4076 devices such as /dev/snd/sequencer whithout loading the
4077 backing module right-away.
4079 * A new RPM macro has been added that may be used to apply
4080 tmpfiles configuration during package installation.
4082 * systemd-detect-virt and ConditionVirtualization= now can
4083 detect User-Mode-Linux machines (UML).
4085 * journald will now implicitly log the effective capabilities
4086 set of processes in the message metadata.
4088 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4090 * The initrd interface has been simplified (more specifically,
4091 support for passing performance data via environment
4092 variables and fsck results via files in /run has been
4093 removed). These features were non-essential, and are
4094 nowadays available in a much nicer way by having systemd in
4095 the initrd serialize its state and have the hosts systemd
4096 deserialize it again.
4098 * The udev "keymap" data files and tools to apply keyboard
4099 specific mappings of scan to key codes, and force-release
4100 scan code lists have been entirely replaced by a udev
4101 "keyboard" builtin and a hwdb data file.
4103 * systemd will now honour the kernel's "quiet" command line
4104 argument also during late shutdown, resulting in a
4105 completely silent shutdown when used.
4107 * There's now an option to control the SO_REUSEPORT socket
4108 option in .socket units.
4110 * Instance units will now automatically get a per-template
4111 subslice of system.slice unless something else is explicitly
4112 configured. For example, instances of sshd@.service will now
4113 implicitly be placed in system-sshd.slice rather than
4114 system.slice as before.
4116 * Test coverage support may now be enabled at build time.
4118 Contributions from: Dave Reisner, Frederic Crozat, Harald
4119 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4120 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4121 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4122 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4123 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4124 Giokas, Zbigniew Jędrzejewski-Szmek
4126 — Berlin, 2013-07-23
4130 * Two new unit types have been introduced:
4132 Scope units are very similar to service units, however, are
4133 created out of pre-existing processes — instead of PID 1
4134 forking off the processes. By using scope units it is
4135 possible for system services and applications to group their
4136 own child processes (worker processes) in a powerful way
4137 which then maybe used to organize them, or kill them
4138 together, or apply resource limits on them.
4140 Slice units may be used to partition system resources in an
4141 hierarchical fashion and then assign other units to them. By
4142 default there are now three slices: system.slice (for all
4143 system services), user.slice (for all user sessions),
4144 machine.slice (for VMs and containers).
4146 Slices and scopes have been introduced primarily in
4147 context of the work to move cgroup handling to a
4148 single-writer scheme, where only PID 1
4149 creates/removes/manages cgroups.
4151 * There's a new concept of "transient" units. In contrast to
4152 normal units these units are created via an API at runtime,
4153 not from configuration from disk. More specifically this
4154 means it is now possible to run arbitrary programs as
4155 independent services, with all execution parameters passed
4156 in via bus APIs rather than read from disk. Transient units
4157 make systemd substantially more dynamic then it ever was,
4158 and useful as a general batch manager.
4160 * logind has been updated to make use of scope and slice units
4161 for managing user sessions. As a user logs in he will get
4162 his own private slice unit, to which all sessions are added
4163 as scope units. We also added support for automatically
4164 adding an instance of user@.service for the user into the
4165 slice. Effectively logind will no longer create cgroup
4166 hierarchies on its own now, it will defer entirely to PID 1
4167 for this by means of scope, service and slice units. Since
4168 user sessions this way become entities managed by PID 1
4169 the output of "systemctl" is now a lot more comprehensive.
4171 * A new mini-daemon "systemd-machined" has been added which
4172 may be used by virtualization managers to register local
4173 VMs/containers. nspawn has been updated accordingly, and
4174 libvirt will be updated shortly. machined will collect a bit
4175 of meta information about the VMs/containers, and assign
4176 them their own scope unit (see above). The collected
4177 meta-data is then made available via the "machinectl" tool,
4178 and exposed in "ps" and similar tools. machined/machinectl
4179 is compile-time optional.
4181 * As discussed earlier, the low-level cgroup configuration
4182 options ControlGroup=, ControlGroupModify=,
4183 ControlGroupPersistent=, ControlGroupAttribute= have been
4184 removed. Please use high-level attribute settings instead as
4185 well as slice units.
4187 * A new bus call SetUnitProperties() has been added to alter
4188 various runtime parameters of a unit. This is primarily
4189 useful to alter cgroup parameters dynamically in a nice way,
4190 but will be extended later on to make more properties
4191 modifiable at runtime. systemctl gained a new set-properties
4192 command that wraps this call.
4194 * A new tool "systemd-run" has been added which can be used to
4195 run arbitrary command lines as transient services or scopes,
4196 while configuring a number of settings via the command
4197 line. This tool is currently very basic, however already
4198 very useful. We plan to extend this tool to even allow
4199 queuing of execution jobs with time triggers from the
4200 command line, similar in fashion to "at".
4202 * nspawn will now inform the user explicitly that kernels with
4203 audit enabled break containers, and suggest the user to turn
4206 * Support for detecting the IMA and AppArmor security
4207 frameworks with ConditionSecurity= has been added.
4209 * journalctl gained a new "-k" switch for showing only kernel
4210 messages, mimicking dmesg output; in addition to "--user"
4211 and "--system" switches for showing only user's own logs
4214 * systemd-delta can now show information about drop-in
4215 snippets extending unit files.
4217 * libsystemd-bus has been substantially updated but is still
4218 not available as public API.
4220 * systemd will now look for the "debug" argument on the kernel
4221 command line and enable debug logging, similar to what
4222 "systemd.log_level=debug" already did before.
4224 * "systemctl set-default", "systemctl get-default" has been
4225 added to configure the default.target symlink, which
4226 controls what to boot into by default.
4228 * "systemctl set-log-level" has been added as a convenient
4229 way to raise and lower systemd logging threshold.
4231 * "systemd-analyze plot" will now show the time the various
4232 generators needed for execution, as well as information
4233 about the unit file loading.
4235 * libsystemd-journal gained a new sd_journal_open_files() call
4236 for opening specific journal files. journactl also gained a
4237 new switch to expose this new functionality. Previously we
4238 only supported opening all files from a directory, or all
4239 files from the system, as opening individual files only is
4240 racy due to journal file rotation.
4242 * systemd gained the new DefaultEnvironment= setting in
4243 /etc/systemd/system.conf to set environment variables for
4246 * If a privileged process logs a journal message with the
4247 OBJECT_PID= field set, then journald will automatically
4248 augment this with additional OBJECT_UID=, OBJECT_GID=,
4249 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4250 system services want to log events about specific client
4251 processes. journactl/systemctl has been updated to make use
4252 of this information if all log messages regarding a specific
4255 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4256 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4257 Reisner, David Coppa, David King, David Strauss, Eelco
4258 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4259 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4260 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4261 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4262 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4263 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4264 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4265 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4266 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4267 Łukasz Stelmach, 장동준
4271 * The Python bindings gained some minimal support for the APIs
4272 exposed by libsystemd-logind.
4274 * ConditionSecurity= gained support for detecting SMACK. Since
4275 this condition already supports SELinux and AppArmor we only
4276 miss IMA for this. Patches welcome!
4278 Contributions from: Karol Lewandowski, Lennart Poettering,
4279 Zbigniew Jędrzejewski-Szmek
4283 * systemd-nspawn will now create /etc/resolv.conf if
4284 necessary, before bind-mounting the host's file onto it.
4286 * systemd-nspawn will now store meta information about a
4287 container on the container's cgroup as extended attribute
4288 fields, including the root directory.
4290 * The cgroup hierarchy has been reworked in many ways. All
4291 objects any of the components systemd creates in the cgroup
4292 tree are now suffixed. More specifically, user sessions are
4293 now placed in cgroups suffixed with ".session", users in
4294 cgroups suffixed with ".user", and nspawn containers in
4295 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4296 names are now escaped in a simple scheme to avoid collision
4297 of userspace object names with kernel filenames. This work
4298 is preparation for making these objects relocatable in the
4299 cgroup tree, in order to allow easy resource partitioning of
4300 these objects without causing naming conflicts.
4302 * systemctl list-dependencies gained the new switches
4303 --plain, --reverse, --after and --before.
4305 * systemd-inhibit now shows the process name of processes that
4306 have taken an inhibitor lock.
4308 * nss-myhostname will now also resolve "localhost"
4309 implicitly. This makes /etc/hosts an optional file and
4310 nicely handles that on IPv6 ::1 maps to both "localhost" and
4313 * libsystemd-logind.so gained a new call
4314 sd_get_machine_names() to enumerate running containers and
4315 VMs (currently only supported by very new libvirt and
4316 nspawn). sd_login_monitor can now be used to watch
4317 VMs/containers coming and going.
4319 * .include is not allowed recursively anymore, and only in
4320 unit files. Usually it is better to use drop-in snippets in
4321 .d/*.conf anyway, as introduced with systemd 198.
4323 * systemd-analyze gained a new "critical-chain" command that
4324 determines the slowest chain of units run during system
4325 boot-up. It is very useful for tracking down where
4326 optimizing boot time is the most beneficial.
4328 * systemd will no longer allow manipulating service paths in
4329 the name=systemd:/system cgroup tree using ControlGroup= in
4330 units. (But is still fine with it in all other dirs.)
4332 * There's a new systemd-nspawn@.service service file that may
4333 be used to easily run nspawn containers as system
4334 services. With the container's root directory in
4335 /var/lib/container/foobar it is now sufficient to run
4336 "systemctl start systemd-nspawn@foobar.service" to boot it.
4338 * systemd-cgls gained a new parameter "--machine" to list only
4339 the processes within a certain container.
4341 * ConditionSecurity= now can check for "apparmor". We still
4342 are lacking checks for SMACK and IMA for this condition
4343 check though. Patches welcome!
4345 * A new configuration file /etc/systemd/sleep.conf has been
4346 added that may be used to configure which kernel operation
4347 systemd is supposed to execute when "suspend", "hibernate"
4348 or "hybrid-sleep" is requested. This makes the new kernel
4349 "freeze" state accessible to the user.
4351 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4352 the passed argument if applicable.
4354 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4355 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4356 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4357 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4358 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4359 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4360 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4365 * The output of 'systemctl list-jobs' got some polishing. The
4366 '--type=' argument may now be passed more than once. A new
4367 command 'systemctl list-sockets' has been added which shows
4368 a list of kernel sockets systemd is listening on with the
4369 socket units they belong to, plus the units these socket
4372 * The experimental libsystemd-bus library got substantial
4373 updates to work in conjunction with the (also experimental)
4374 kdbus kernel project. It works well enough to exchange
4375 messages with some sophistication. Note that kdbus is not
4376 ready yet, and the library is mostly an elaborate test case
4377 for now, and not installable.
4379 * systemd gained a new unit 'systemd-static-nodes.service'
4380 that generates static device nodes earlier during boot, and
4381 can run in conjunction with udev.
4383 * libsystemd-login gained a new call sd_pid_get_user_unit()
4384 to retrieve the user systemd unit a process is running
4385 in. This is useful for systems where systemd is used as
4388 * systemd-nspawn now places all containers in the new /machine
4389 top-level cgroup directory in the name=systemd
4390 hierarchy. libvirt will soon do the same, so that we get a
4391 uniform separation of /system, /user and /machine for system
4392 services, user processes and containers/virtual
4393 machines. This new cgroup hierarchy is also useful to stick
4394 stable names to specific container instances, which can be
4395 recognized later this way (this name may be controlled
4396 via systemd-nspawn's new -M switch). libsystemd-login also
4397 gained a new call sd_pid_get_machine_name() to retrieve the
4398 name of the container/VM a specific process belongs to.
4400 * bootchart can now store its data in the journal.
4402 * libsystemd-journal gained a new call
4403 sd_journal_add_conjunction() for AND expressions to the
4404 matching logic. This can be used to express more complex
4405 logical expressions.
4407 * journactl can now take multiple --unit= and --user-unit=
4410 * The cryptsetup logic now understands the "luks.key=" kernel
4411 command line switch for specifying a file to read the
4412 decryption key from. Also, if a configured key file is not
4413 found the tool will now automatically fall back to prompting
4416 * Python systemd.journal module was updated to wrap recently
4417 added functions from libsystemd-journal. The interface was
4418 changed to bring the low level interface in s.j._Reader
4419 closer to the C API, and the high level interface in
4420 s.j.Reader was updated to wrap and convert all data about
4423 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4424 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4425 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4426 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4427 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4428 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4432 * journalctl --update-catalog now understands a new --root=
4433 option to operate on catalogs found in a different root
4436 * During shutdown after systemd has terminated all running
4437 services a final killing loop kills all remaining left-over
4438 processes. We will now print the name of these processes
4439 when we send SIGKILL to them, since this usually indicates a
4442 * If /etc/crypttab refers to password files stored on
4443 configured mount points automatic dependencies will now be
4444 generated to ensure the specific mount is established first
4445 before the key file is attempted to be read.
4447 * 'systemctl status' will now show information about the
4448 network sockets a socket unit is listening on.
4450 * 'systemctl status' will also shown information about any
4451 drop-in configuration file for units. (Drop-In configuration
4452 files in this context are files such as
4453 /etc/systemd/systemd/foobar.service.d/*.conf)
4455 * systemd-cgtop now optionally shows summed up CPU times of
4456 cgroups. Press '%' while running cgtop to switch between
4457 percentage and absolute mode. This is useful to determine
4458 which cgroups use up the most CPU time over the entire
4459 runtime of the system. systemd-cgtop has also been updated
4460 to be 'pipeable' for processing with further shell tools.
4462 * 'hostnamectl set-hostname' will now allow setting of FQDN
4465 * The formatting and parsing of time span values has been
4466 changed. The parser now understands fractional expressions
4467 such as "5.5h". The formatter will now output fractional
4468 expressions for all time spans under 1min, i.e. "5.123456s"
4469 rather than "5s 123ms 456us". For time spans under 1s
4470 millisecond values are shown, for those under 1ms
4471 microsecond values are shown. This should greatly improve
4472 all time-related output of systemd.
4474 * libsystemd-login and libsystemd-journal gained new
4475 functions for querying the poll() events mask and poll()
4476 timeout value for integration into arbitrary event
4479 * localectl gained the ability to list available X11 keymaps
4480 (models, layouts, variants, options).
4482 * 'systemd-analyze dot' gained the ability to filter for
4483 specific units via shell-style globs, to create smaller,
4484 more useful graphs. I.e. it is now possible to create simple
4485 graphs of all the dependencies between only target units, or
4486 of all units that Avahi has dependencies with.
4488 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4489 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4490 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4491 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4492 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4493 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4494 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4498 * The boot-time readahead implementation for rotating media
4499 will now read the read-ahead data in multiple passes which
4500 consist of all read requests made in equidistant time
4501 intervals. This means instead of strictly reading read-ahead
4502 data in its physical order on disk we now try to find a
4503 middle ground between physical and access time order.
4505 * /etc/os-release files gained a new BUILD_ID= field for usage
4506 on operating systems that provide continuous builds of OS
4509 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4510 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4511 William Douglas, Zbigniew Jędrzejewski-Szmek
4515 * systemd-python gained an API exposing libsystemd-daemon.
4517 * The SMACK setup logic gained support for uploading CIPSO
4520 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4521 ReadOnlyDirectories= and InaccessibleDirectories= has
4522 changed. The private /tmp and /var/tmp directories are now
4523 shared by all processes of a service (which means
4524 ExecStartPre= may now leave data in /tmp that ExecStart= of
4525 the same service can still access). When a service is
4526 stopped its temporary directories are immediately deleted
4527 (normal clean-up with tmpfiles is still done in addition to
4530 * By default, systemd will now set a couple of sysctl
4531 variables in the kernel: the safe sysrq options are turned
4532 on, IP route verification is turned on, and source routing
4533 disabled. The recently added hardlink and softlink
4534 protection of the kernel is turned on. These settings should
4535 be reasonably safe, and good defaults for all new systems.
4537 * The predictable network naming logic may now be turned off
4538 with a new kernel command line switch: net.ifnames=0.
4540 * A new libsystemd-bus module has been added that implements a
4541 pretty complete D-Bus client library. For details see:
4543 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4545 * journald will now explicitly flush the journal files to disk
4546 at the latest 5min after each write. The file will then also
4547 be marked offline until the next write. This should increase
4548 reliability in case of a crash. The synchronization delay
4549 can be configured via SyncIntervalSec= in journald.conf.
4551 * There's a new remote-fs-setup.target unit that can be used
4552 to pull in specific services when at least one remote file
4553 system is to be mounted.
4555 * There are new targets timers.target and paths.target as
4556 canonical targets to pull user timer and path units in
4557 from. This complements sockets.target with a similar
4558 purpose for socket units.
4560 * libudev gained a new call udev_device_set_attribute_value()
4561 to set sysfs attributes of a device.
4563 * The udev daemon now sets the default number of worker
4564 processes executed in parallel based on the number of available
4565 CPUs instead of the amount of available RAM. This is supposed
4566 to provide a more reliable default and limit a too aggressive
4567 paralellism for setups with 1000s of devices connected.
4569 Contributions from: Auke Kok, Colin Walters, Cristian
4570 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4571 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4572 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4573 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4574 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4575 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4576 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4577 Zbigniew Jędrzejewski-Szmek
4581 * Configuration of unit files may now be extended via drop-in
4582 files without having to edit/override the unit files
4583 themselves. More specifically, if the administrator wants to
4584 change one value for a service file foobar.service he can
4585 now do so by dropping in a configuration snippet into
4586 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4587 will load all these snippets and apply them on top of the
4588 main unit configuration file, possibly extending or
4589 overriding its settings. Using these drop-in snippets is
4590 generally nicer than the two earlier options for changing
4591 unit files locally: copying the files from
4592 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4593 them there; or creating a new file in /etc/systemd/system/
4594 that incorporates the original one via ".include". Drop-in
4595 snippets into these .d/ directories can be placed in any
4596 directory systemd looks for units in, and the usual
4597 overriding semantics between /usr/lib, /etc and /run apply
4600 * Most unit file settings which take lists of items can now be
4601 reset by assigning the empty string to them. For example,
4602 normally, settings such as Environment=FOO=BAR append a new
4603 environment variable assignment to the environment block,
4604 each time they are used. By assigning Environment= the empty
4605 string the environment block can be reset to empty. This is
4606 particularly useful with the .d/*.conf drop-in snippets
4607 mentioned above, since this adds the ability to reset list
4608 settings from vendor unit files via these drop-ins.
4610 * systemctl gained a new "list-dependencies" command for
4611 listing the dependencies of a unit recursively.
4613 * Inhibitors are now honored and listed by "systemctl
4614 suspend", "systemctl poweroff" (and similar) too, not only
4615 GNOME. These commands will also list active sessions by
4618 * Resource limits (as exposed by the various control group
4619 controllers) can now be controlled dynamically at runtime
4620 for all units. More specifically, you can now use a command
4621 like "systemctl set-cgroup-attr foobar.service cpu.shares
4622 2000" to alter the CPU shares a specific service gets. These
4623 settings are stored persistently on disk, and thus allow the
4624 administrator to easily adjust the resource usage of
4625 services with a few simple commands. This dynamic resource
4626 management logic is also available to other programs via the
4627 bus. Almost any kernel cgroup attribute and controller is
4630 * systemd-vconsole-setup will now copy all font settings to
4631 all allocated VTs, where it previously applied them only to
4634 * libsystemd-login gained the new sd_session_get_tty() API
4637 * This release drops support for a few legacy or
4638 distribution-specific LSB facility names when parsing init
4639 scripts: $x-display-manager, $mail-transfer-agent,
4640 $mail-transport-agent, $mail-transfer-agent, $smtp,
4641 $null. Also, the mail-transfer-agent.target unit backing
4642 this has been removed. Distributions which want to retain
4643 compatibility with this should carry the burden for
4644 supporting this themselves and patch support for these back
4645 in, if they really need to. Also, the facilities $syslog and
4646 $local_fs are now ignored, since systemd does not support
4647 early-boot LSB init scripts anymore, and these facilities
4648 are implied anyway for normal services. syslog.target has
4651 * There are new bus calls on PID1's Manager object for
4652 cancelling jobs, and removing snapshot units. Previously,
4653 both calls were only available on the Job and Snapshot
4656 * systemd-journal-gatewayd gained SSL support.
4658 * The various "environment" files, such as /etc/locale.conf
4659 now support continuation lines with a backslash ("\") as
4660 last character in the line, similarly in style (but different)
4661 to how this is supported in shells.
4663 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4664 now implicitly appended to every log entry logged. systemctl
4665 has been updated to filter by this field when operating on a
4666 user systemd instance.
4668 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4669 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4670 the container. This makes it easier to boot unmodified
4671 Fedora systems in a container, which however still requires
4672 audit=0 to be passed on the kernel command line. Auditing in
4673 kernel and userspace is unfortunately still too broken in
4674 context of containers, hence we recommend compiling it out
4675 of the kernel or using audit=0. Hopefully this will be fixed
4676 one day for good in the kernel.
4678 * nspawn gained the new --bind= and --bind-ro= parameters to
4679 bind mount specific directories from the host into the
4682 * nspawn will now mount its own devpts file system instance
4683 into the container, in order not to leak pty devices from
4684 the host into the container.
4686 * systemd will now read the firmware boot time performance
4687 information from the EFI variables, if the used boot loader
4688 supports this, and takes it into account for boot performance
4689 analysis via "systemd-analyze". This is currently supported
4690 only in conjunction with Gummiboot, but could be supported
4691 by other boot loaders too. For details see:
4693 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4695 * A new generator has been added that automatically mounts the
4696 EFI System Partition (ESP) to /boot, if that directory
4697 exists, is empty, and no other file system has been
4698 configured to be mounted there.
4700 * logind will now send out PrepareForSleep(false) out
4701 unconditionally, after coming back from suspend. This may be
4702 used by applications as asynchronous notification for
4703 system resume events.
4705 * "systemctl unlock-sessions" has been added, that allows
4706 unlocking the screens of all user sessions at once, similar
4707 to how "systemctl lock-sessions" already locked all users
4708 sessions. This is backed by a new D-Bus call UnlockSessions().
4710 * "loginctl seat-status" will now show the master device of a
4711 seat. (i.e. the device of a seat that needs to be around for
4712 the seat to be considered available, usually the graphics
4715 * tmpfiles gained a new "X" line type, that allows
4716 configuration of files and directories (with wildcards) that
4717 shall be excluded from automatic cleanup ("aging").
4719 * udev default rules set the device node permissions now only
4720 at "add" events, and do not change them any longer with a
4721 later "change" event.
4723 * The log messages for lid events and power/sleep keypresses
4724 now carry a message ID.
4726 * We now have a substantially larger unit test suite, but this
4727 continues to be work in progress.
4729 * udevadm hwdb gained a new --root= parameter to change the
4730 root directory to operate relative to.
4732 * logind will now issue a background sync() request to the kernel
4733 early at shutdown, so that dirty buffers are flushed to disk early
4734 instead of at the last moment, in order to optimize shutdown
4737 * A new bootctl tool has been added that is an interface for
4738 certain boot loader operations. This is currently a preview
4739 and is likely to be extended into a small mechanism daemon
4740 like timedated, localed, hostnamed, and can be used by
4741 graphical UIs to enumerate available boot options, and
4742 request boot into firmware operations.
4744 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4745 the rest of the package. It also has been updated to work
4746 correctly in initrds.
4748 * Policykit previously has been runtime optional, and is now
4749 also compile time optional via a configure switch.
4751 * systemd-analyze has been reimplemented in C. Also "systemctl
4752 dot" has moved into systemd-analyze.
4754 * "systemctl status" with no further parameters will now print
4755 the status of all active or failed units.
4757 * Operations such as "systemctl start" can now be executed
4758 with a new mode "--irreversible" which may be used to queue
4759 operations that cannot accidentally be reversed by a later
4760 job queuing. This is by default used to make shutdown
4761 requests more robust.
4763 * The Python API of systemd now gained a new module for
4764 reading journal files.
4766 * A new tool kernel-install has been added that can install
4767 kernel images according to the Boot Loader Specification:
4769 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4771 * Boot time console output has been improved to provide
4772 animated boot time output for hanging jobs.
4774 * A new tool systemd-activate has been added which can be used
4775 to test socket activation with, directly from the command
4776 line. This should make it much easier to test and debug
4777 socket activation in daemons.
4779 * journalctl gained a new "--reverse" (or -r) option to show
4780 journal output in reverse order (i.e. newest line first).
4782 * journalctl gained a new "--pager-end" (or -e) option to jump
4783 to immediately jump to the end of the journal in the
4784 pager. This is only supported in conjunction with "less".
4786 * journalctl gained a new "--user-unit=" option, that works
4787 similarly to "--unit=" but filters for user units rather than
4790 * A number of unit files to ease adoption of systemd in
4791 initrds has been added. This moves some minimal logic from
4792 the various initrd implementations into systemd proper.
4794 * The journal files are now owned by a new group
4795 "systemd-journal", which exists specifically to allow access
4796 to the journal, and nothing else. Previously, we used the
4797 "adm" group for that, which however possibly covers more
4798 than just journal/log file access. This new group is now
4799 already used by systemd-journal-gatewayd to ensure this
4800 daemon gets access to the journal files and as little else
4801 as possible. Note that "make install" will also set FS ACLs
4802 up for /var/log/journal to give "adm" and "wheel" read
4803 access to it, in addition to "systemd-journal" which owns
4804 the journal files. We recommend that packaging scripts also
4805 add read access to "adm" + "wheel" to /var/log/journal, and
4806 all existing/future journal files. To normal users and
4807 administrators little changes, however packagers need to
4808 ensure to create the "systemd-journal" system group at
4809 package installation time.
4811 * The systemd-journal-gatewayd now runs as unprivileged user
4812 systemd-journal-gateway:systemd-journal-gateway. Packaging
4813 scripts need to create these system user/group at
4816 * timedated now exposes a new boolean property CanNTP that
4817 indicates whether a local NTP service is available or not.
4819 * systemd-detect-virt will now also detect xen PVs
4821 * The pstore file system is now mounted by default, if it is
4824 * In addition to the SELinux and IMA policies we will now also
4825 load SMACK policies at early boot.
4827 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4828 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4829 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4830 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4831 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4832 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4833 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4834 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4835 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4836 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4837 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4838 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4839 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4840 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4844 * Timer units now support calendar time events in addition to
4845 monotonic time events. That means you can now trigger a unit
4846 based on a calendar time specification such as "Thu,Fri
4847 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4848 or fifth day of any month of the year 2013, given that it is
4849 a thursday or friday. This brings timer event support
4850 considerably closer to cron's capabilities. For details on
4851 the supported calendar time specification language see
4854 * udev now supports a number of different naming policies for
4855 network interfaces for predictable names, and a combination
4856 of these policies is now the default. Please see this wiki
4857 document for details:
4859 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4861 * Auke Kok's bootchart implementation has been added to the
4862 systemd tree. It is an optional component that can graph the
4863 boot in quite some detail. It is one of the best bootchart
4864 implementations around and minimal in its code and
4867 * nss-myhostname has been integrated into the systemd source
4868 tree. nss-myhostname guarantees that the local hostname
4869 always stays resolvable via NSS. It has been a weak
4870 requirement of systemd-hostnamed since a long time, and
4871 since its code is actually trivial we decided to just
4872 include it in systemd's source tree. It can be turned off
4873 with a configure switch.
4875 * The read-ahead logic is now capable of properly detecting
4876 whether a btrfs file system is on SSD or rotating media, in
4877 order to optimize the read-ahead scheme. Previously, it was
4878 only capable of detecting this on traditional file systems
4881 * In udev, additional device properties are now read from the
4882 IAB in addition to the OUI database. Also, Bluetooth company
4883 identities are attached to the devices as well.
4885 * In service files %U may be used as specifier that is
4886 replaced by the configured user name of the service.
4888 * nspawn may now be invoked without a controlling TTY. This
4889 makes it suitable for invocation as its own service. This
4890 may be used to set up a simple containerized server system
4891 using only core OS tools.
4893 * systemd and nspawn can now accept socket file descriptors
4894 when they are started for socket activation. This enables
4895 implementation of socket activated nspawn
4896 containers. i.e. think about autospawning an entire OS image
4897 when the first SSH or HTTP connection is received. We expect
4898 that similar functionality will also be added to libvirt-lxc
4901 * journalctl will now suppress ANSI color codes when
4902 presenting log data.
4904 * systemctl will no longer show control group information for
4905 a unit if a the control group is empty anyway.
4907 * logind can now automatically suspend/hibernate/shutdown the
4910 * /etc/machine-info and hostnamed now also expose the chassis
4911 type of the system. This can be used to determine whether
4912 the local system is a laptop, desktop, handset or
4913 tablet. This information may either be configured by the
4914 user/vendor or is automatically determined from ACPI and DMI
4915 information if possible.
4917 * A number of PolicyKit actions are now bound together with
4918 "imply" rules. This should simplify creating UIs because
4919 many actions will now authenticate similar ones as well.
4921 * Unit files learnt a new condition ConditionACPower= which
4922 may be used to conditionalize a unit depending on whether an
4923 AC power source is connected or not, of whether the system
4924 is running on battery power.
4926 * systemctl gained a new "is-failed" verb that may be used in
4927 shell scripts and suchlike to check whether a specific unit
4928 is in the "failed" state.
4930 * The EnvironmentFile= setting in unit files now supports file
4931 globbing, and can hence be used to easily read a number of
4932 environment files at once.
4934 * systemd will no longer detect and recognize specific
4935 distributions. All distribution-specific #ifdeffery has been
4936 removed, systemd is now fully generic and
4937 distribution-agnostic. Effectively, not too much is lost as
4938 a lot of the code is still accessible via explicit configure
4939 switches. However, support for some distribution specific
4940 legacy configuration file formats has been dropped. We
4941 recommend distributions to simply adopt the configuration
4942 files everybody else uses now and convert the old
4943 configuration from packaging scripts. Most distributions
4944 already did that. If that's not possible or desirable,
4945 distributions are welcome to forward port the specific
4946 pieces of code locally from the git history.
4948 * When logging a message about a unit systemd will now always
4949 log the unit name in the message meta data.
4951 * localectl will now also discover system locale data that is
4952 not stored in locale archives, but directly unpacked.
4954 * logind will no longer unconditionally use framebuffer
4955 devices as seat masters, i.e. as devices that are required
4956 to be existing before a seat is considered preset. Instead,
4957 it will now look for all devices that are tagged as
4958 "seat-master" in udev. By default, framebuffer devices will
4959 be marked as such, but depending on local systems, other
4960 devices might be marked as well. This may be used to
4961 integrate graphics cards using closed source drivers (such
4962 as NVidia ones) more nicely into logind. Note however, that
4963 we recommend using the open source NVidia drivers instead,
4964 and no udev rules for the closed-source drivers will be
4965 shipped from us upstream.
4967 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4968 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4969 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4970 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4971 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4972 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4973 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4974 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4975 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4976 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4977 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4978 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4983 * udev gained support for loading additional device properties
4984 from an indexed database that is keyed by vendor/product IDs
4985 and similar device identifiers. For the beginning this
4986 "hwdb" is populated with data from the well-known PCI and
4987 USB database, but also includes PNP, ACPI and OID data. In
4988 the longer run this indexed database shall grow into
4989 becoming the one central database for non-essential
4990 userspace device metadata. Previously, data from the PCI/USB
4991 database was only attached to select devices, since the
4992 lookup was a relatively expensive operation due to O(n) time
4993 complexity (with n being the number of entries in the
4994 database). Since this is now O(1), we decided to add in this
4995 data for all devices where this is available, by
4996 default. Note that the indexed database needs to be rebuilt
4997 when new data files are installed. To achieve this you need
4998 to update your packaging scripts to invoke "udevadm hwdb
4999 --update" after installation of hwdb data files. For
5000 RPM-based distributions we introduced the new
5001 %udev_hwdb_update macro for this purpose.
5003 * The Journal gained support for the "Message Catalog", an
5004 indexed database to link up additional information with
5005 journal entries. For further details please check:
5007 http://www.freedesktop.org/wiki/Software/systemd/catalog
5009 The indexed message catalog database also needs to be
5010 rebuilt after installation of message catalog files. Use
5011 "journalctl --update-catalog" for this. For RPM-based
5012 distributions we introduced the %journal_catalog_update
5013 macro for this purpose.
5015 * The Python Journal bindings gained support for the standard
5016 Python logging framework.
5018 * The Journal API gained new functions for checking whether
5019 the underlying file system of a journal file is capable of
5020 properly reporting file change notifications, or whether
5021 applications that want to reflect journal changes "live"
5022 need to recheck journal files continuously in appropriate
5025 * It is now possible to set the "age" field for tmpfiles
5026 entries to 0, indicating that files matching this entry
5027 shall always be removed when the directories are cleaned up.
5029 * coredumpctl gained a new "gdb" verb which invokes gdb
5030 right-away on the selected coredump.
5032 * There's now support for "hybrid sleep" on kernels that
5033 support this, in addition to "suspend" and "hibernate". Use
5034 "systemctl hybrid-sleep" to make use of this.
5036 * logind's HandleSuspendKey= setting (and related settings)
5037 now gained support for a new "lock" setting to simply
5038 request the screen lock on all local sessions, instead of
5039 actually executing a suspend or hibernation.
5041 * systemd will now mount the EFI variables file system by
5044 * Socket units now gained support for configuration of the
5045 SMACK security label.
5047 * timedatectl will now output the time of the last and next
5048 daylight saving change.
5050 * We dropped support for various legacy and distro-specific
5051 concepts, such as insserv, early-boot SysV services
5052 (i.e. those for non-standard runlevels such as 'b' or 'S')
5053 or ArchLinux /etc/rc.conf support. We recommend the
5054 distributions who still need support this to either continue
5055 to maintain the necessary patches downstream, or find a
5056 different solution. (Talk to us if you have questions!)
5058 * Various systemd components will now bypass PolicyKit checks
5059 for root and otherwise handle properly if PolicyKit is not
5060 found to be around. This should fix most issues for
5061 PolicyKit-less systems. Quite frankly this should have been
5062 this way since day one. It is absolutely our intention to
5063 make systemd work fine on PolicyKit-less systems, and we
5064 consider it a bug if something does not work as it should if
5065 PolicyKit is not around.
5067 * For embedded systems it is now possible to build udev and
5068 systemd without blkid and/or kmod support.
5070 * "systemctl switch-root" is now capable of switching root
5071 more than once. I.e. in addition to transitions from the
5072 initrd to the host OS it is now possible to transition to
5073 further OS images from the host. This is useful to implement
5074 offline updating tools.
5076 * Various other additions have been made to the RPM macros
5077 shipped with systemd. Use %udev_rules_update() after
5078 installing new udev rules files. %_udevhwdbdir,
5079 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5080 %_sysctldir are now available which resolve to the right
5081 directories for packages to place various data files in.
5083 * journalctl gained the new --full switch (in addition to
5084 --all, to disable ellipsation for long messages.
5086 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5087 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5088 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5089 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5090 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5091 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5092 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5093 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5094 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5098 * journalctl gained new --since= and --until= switches to
5099 filter by time. It also now supports nice filtering for
5100 units via --unit=/-u.
5102 * Type=oneshot services may use ExecReload= and do the
5105 * The journal daemon now supports time-based rotation and
5106 vacuuming, in addition to the usual disk-space based
5109 * The journal will now index the available field values for
5110 each field name. This enables clients to show pretty drop
5111 downs of available match values when filtering. The bash
5112 completion of journalctl has been updated
5113 accordingly. journalctl gained a new switch -F to list all
5114 values a certain field takes in the journal database.
5116 * More service events are now written as structured messages
5117 to the journal, and made recognizable via message IDs.
5119 * The timedated, localed and hostnamed mini-services which
5120 previously only provided support for changing time, locale
5121 and hostname settings from graphical DEs such as GNOME now
5122 also have a minimal (but very useful) text-based client
5123 utility each. This is probably the nicest way to changing
5124 these settings from the command line now, especially since
5125 it lists available options and is fully integrated with bash
5128 * There's now a new tool "systemd-coredumpctl" to list and
5129 extract coredumps from the journal.
5131 * We now install a README each in /var/log/ and
5132 /etc/rc.d/init.d explaining where the system logs and init
5133 scripts went. This hopefully should help folks who go to
5134 that dirs and look into the otherwise now empty void and
5135 scratch their heads.
5137 * When user-services are invoked (by systemd --user) the
5138 $MANAGERPID env var is set to the PID of systemd.
5140 * SIGRTMIN+24 when sent to a --user instance will now result
5141 in immediate termination of systemd.
5143 * gatewayd received numerous feature additions such as a
5144 "follow" mode, for live syncing and filtering.
5146 * browse.html now allows filtering and showing detailed
5147 information on specific entries. Keyboard navigation and
5148 mouse screen support has been added.
5150 * gatewayd/journalctl now supports HTML5/JSON
5151 Server-Sent-Events as output.
5153 * The SysV init script compatibility logic will now
5154 heuristically determine whether a script supports the
5155 "reload" verb, and only then make this available as
5158 * "systemctl status --follow" has been removed, use "journalctl
5161 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5162 have been removed since they are hardly useful to be
5165 * And I'd like to take the opportunity to specifically mention
5166 Zbigniew for his great contributions. Zbigniew, you rock!
5168 Contributions from: Andrew Eikum, Christian Hesse, Colin
5169 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5170 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5171 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5172 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5173 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5174 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5178 * If /etc/vconsole.conf is non-existent or empty we will no
5179 longer load any console font or key map at boot by
5180 default. Instead the kernel defaults will be left
5181 intact. This is definitely the right thing to do, as no
5182 configuration should mean no configuration, and hard-coding
5183 font names that are different on all archs is probably a bad
5184 idea. Also, the kernel default key map and font should be
5185 good enough for most cases anyway, and mostly identical to
5186 the userspace fonts/key maps we previously overloaded them
5187 with. If distributions want to continue to default to a
5188 non-kernel font or key map they should ship a default
5189 /etc/vconsole.conf with the appropriate contents.
5191 Contributions from: Colin Walters, Daniel J Walsh, Dave
5192 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5193 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5197 * journalctl gained a new --cursor= switch to show entries
5198 starting from the specified location in the journal.
5200 * We now enforce a size limit on journal entry fields exported
5201 with "-o json" in journalctl. Fields larger than 4K will be
5202 assigned null. This can be turned off with --all.
5204 * An (optional) journal gateway daemon is now available as
5205 "systemd-journal-gatewayd.service". This service provides
5206 access to the journal via HTTP and JSON. This functionality
5207 will be used to implement live log synchronization in both
5208 pull and push modes, but has various other users too, such
5209 as easy log access for debugging of embedded devices. Right
5210 now it is already useful to retrieve the journal via HTTP:
5212 # systemctl start systemd-journal-gatewayd.service
5213 # wget http://localhost:19531/entries
5215 This will download the journal contents in a
5216 /var/log/messages compatible format. The same as JSON:
5218 # curl -H"Accept: application/json" http://localhost:19531/entries
5220 This service is also accessible via a web browser where a
5221 single static HTML5 app is served that uses the JSON logic
5222 to enable the user to do some basic browsing of the
5223 journal. This will be extended later on. Here's an example
5224 screenshot of this app in its current state:
5226 http://0pointer.de/public/journal-gatewayd
5228 Contributions from: Kay Sievers, Lennart Poettering, Robert
5229 Milasan, Tom Gundersen
5233 * The bash completion logic is now available for journalctl
5236 * We do not mount the "cpuset" controller anymore together with
5237 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5238 started if no parameters are assigned to it. "cpuset" hence
5239 broke code that assumed it could create "cpu" groups and
5242 * journalctl -f will now subscribe to terminal size changes,
5243 and line break accordingly.
5245 Contributions from: Dave Reisner, Kay Sievers, Lennart
5246 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5250 * nspawn will now create a symlink /etc/localtime in the
5251 container environment, copying the host's timezone
5252 setting. Previously this has been done via a bind mount, but
5253 since symlinks cannot be bind mounted this has now been
5254 changed to create/update the appropriate symlink.
5256 * journalctl -n's line number argument is now optional, and
5257 will default to 10 if omitted.
5259 * journald will now log the maximum size the journal files may
5260 take up on disk. This is particularly useful if the default
5261 built-in logic of determining this parameter from the file
5262 system size is used. Use "systemctl status
5263 systemd-journald.service" to see this information.
5265 * The multi-seat X wrapper tool has been stripped down. As X
5266 is now capable of enumerating graphics devices via udev in a
5267 seat-aware way the wrapper is not strictly necessary
5268 anymore. A stripped down temporary stop-gap is still shipped
5269 until the upstream display managers have been updated to
5270 fully support the new X logic. Expect this wrapper to be
5271 removed entirely in one of the next releases.
5273 * HandleSleepKey= in logind.conf has been split up into
5274 HandleSuspendKey= and HandleHibernateKey=. The old setting
5275 is not available anymore. X11 and the kernel are
5276 distinguishing between these keys and we should too. This
5277 also means the inhibition lock for these keys has been split
5280 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5281 Poettering, Lukas Nykryn, Václav Pavlín
5285 * Whenever a unit changes state we will now log this to the
5286 journal and show along the unit's own log output in
5289 * ConditionPathIsMountPoint= can now properly detect bind
5290 mount points too. (Previously, a bind mount of one file
5291 system to another place in the same file system could not be
5292 detected as mount, since they shared struct stat's st_dev
5295 * We will now mount the cgroup controllers cpu, cpuacct,
5296 cpuset and the controllers net_cls, net_prio together by
5299 * nspawn containers will now have a virtualized boot
5300 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5301 over with a randomized ID at container initialization). This
5302 has the effect of making "journalctl -b" do the right thing
5305 * The JSON output journal serialization has been updated not
5306 to generate "endless" list objects anymore, but rather one
5307 JSON object per line. This is more in line how most JSON
5308 parsers expect JSON objects. The new output mode
5309 "json-pretty" has been added to provide similar output, but
5310 neatly aligned for readability by humans.
5312 * We dropped all explicit sync() invocations in the shutdown
5313 code. The kernel does this implicitly anyway in the kernel
5314 reboot() syscall. halt(8)'s -n option is now a compatibility
5317 * We now support virtualized reboot() in containers, as
5318 supported by newer kernels. We will fall back to exit() if
5319 CAP_SYS_REBOOT is not available to the container. Also,
5320 nspawn makes use of this now and will actually reboot the
5321 container if the containerized OS asks for that.
5323 * journalctl will only show local log output by default
5324 now. Use --merge (-m) to show remote log output, too.
5326 * libsystemd-journal gained the new sd_journal_get_usage()
5327 call to determine the current disk usage of all journal
5328 files. This is exposed in the new "journalctl --disk-usage"
5331 * journald gained a new configuration setting SplitMode= in
5332 journald.conf which may be used to control how user journals
5333 are split off. See journald.conf(5) for details.
5335 * A new condition type ConditionFileNotEmpty= has been added.
5337 * tmpfiles' "w" lines now support file globbing, to write
5338 multiple files at once.
5340 * We added Python bindings for the journal submission
5341 APIs. More Python APIs for a number of selected APIs will
5342 likely follow. Note that we intend to add native bindings
5343 only for the Python language, as we consider it common
5344 enough to deserve bindings shipped within systemd. There are
5345 various projects outside of systemd that provide bindings
5346 for languages such as PHP or Lua.
5348 * Many conditions will now resolve specifiers such as %i. In
5349 addition, PathChanged= and related directives of .path units
5350 now support specifiers as well.
5352 * There's now a new RPM macro definition for the system preset
5355 * journald will now warn if it ca not forward a message to the
5356 syslog daemon because its socket is full.
5358 * timedated will no longer write or process /etc/timezone,
5359 except on Debian. As we do not support late mounted /usr
5360 anymore /etc/localtime always being a symlink is now safe,
5361 and hence the information in /etc/timezone is not necessary
5364 * logind will now always reserve one VT for a text getty (VT6
5365 by default). Previously if more than 6 X sessions where
5366 started they took up all the VTs with auto-spawned gettys,
5367 so that no text gettys were available anymore.
5369 * udev will now automatically inform the btrfs kernel logic
5370 about btrfs RAID components showing up. This should make
5371 simple hotplug based btrfs RAID assembly work.
5373 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5374 (but not for its children which will stay at the kernel
5375 default). This should allow setups with a lot more listening
5378 * systemd will now always pass the configured timezone to the
5379 kernel at boot. timedated will do the same when the timezone
5382 * logind's inhibition logic has been updated. By default,
5383 logind will now handle the lid switch, the power and sleep
5384 keys all the time, even in graphical sessions. If DEs want
5385 to handle these events on their own they should take the new
5386 handle-power-key, handle-sleep-key and handle-lid-switch
5387 inhibitors during their runtime. A simple way to achieve
5388 that is to invoke the DE wrapped in an invocation of:
5390 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5392 * Access to unit operations is now checked via SELinux taking
5393 the unit file label and client process label into account.
5395 * systemd will now notify the administrator in the journal
5396 when he over-mounts a non-empty directory.
5398 * There are new specifiers that are resolved in unit files,
5399 for the host name (%H), the machine ID (%m) and the boot ID
5402 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5403 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5404 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5405 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5406 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5407 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5408 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5412 * Support for reading structured kernel messages from
5413 /dev/kmsg has now been added and is enabled by default.
5415 * Support for reading kernel messages from /proc/kmsg has now
5416 been removed. If you want kernel messages in the journal
5417 make sure to run a recent kernel (>= 3.5) that supports
5418 reading structured messages from /dev/kmsg (see
5419 above). /proc/kmsg is now exclusive property of classic
5420 syslog daemons again.
5422 * The libudev API gained the new
5423 udev_device_new_from_device_id() call.
5425 * The logic for file system namespace (ReadOnlyDirectory=,
5426 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5427 require pivot_root() anymore. This means fewer temporary
5428 directories are created below /tmp for this feature.
5430 * nspawn containers will now see and receive all submounts
5431 made on the host OS below the root file system of the
5434 * Forward Secure Sealing is now supported for Journal files,
5435 which provide cryptographical sealing of journal files so
5436 that attackers cannot alter log history anymore without this
5437 being detectable. Lennart will soon post a blog story about
5438 this explaining it in more detail.
5440 * There are two new service settings RestartPreventExitStatus=
5441 and SuccessExitStatus= which allow configuration of exit
5442 status (exit code or signal) which will be excepted from the
5443 restart logic, resp. consider successful.
5445 * journalctl gained the new --verify switch that can be used
5446 to check the integrity of the structure of journal files and
5447 (if Forward Secure Sealing is enabled) the contents of
5450 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5451 and similar symlinks pre-created. This makes running shells
5452 as container init process a lot more fun.
5454 * The fstab support can now handle PARTUUID= and PARTLABEL=
5457 * A new ConditionHost= condition has been added to match
5458 against the hostname (with globs) and machine ID. This is
5459 useful for clusters where a single OS image is used to
5460 provision a large number of hosts which shall run slightly
5461 different sets of services.
5463 * Services which hit the restart limit will now be placed in a
5466 Contributions from: Bertram Poettering, Dave Reisner, Huang
5467 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5468 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5472 * When running in --user mode systemd will now become a
5473 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5474 tree a lot more organized.
5476 * A new PartOf= unit dependency type has been introduced that
5477 may be used to group services in a natural way.
5479 * "systemctl enable" may now be used to enable instances of
5482 * journalctl now prints error log levels in red, and
5483 warning/notice log levels in bright white. It also supports
5484 filtering by log level now.
5486 * cgtop gained a new -n switch (similar to top), to configure
5487 the maximum number of iterations to run for. It also gained
5488 -b, to run in batch mode (accepting no input).
5490 * The suffix ".service" may now be omitted on most systemctl
5491 command lines involving service unit names.
5493 * There's a new bus call in logind to lock all sessions, as
5494 well as a loginctl verb for it "lock-sessions".
5496 * libsystemd-logind.so gained a new call sd_journal_perror()
5497 that works similar to libc perror() but logs to the journal
5498 and encodes structured information about the error number.
5500 * /etc/crypttab entries now understand the new keyfile-size=
5503 * shutdown(8) now can send a (configurable) wall message when
5504 a shutdown is cancelled.
5506 * The mount propagation mode for the root file system will now
5507 default to "shared", which is useful to make containers work
5508 nicely out-of-the-box so that they receive new mounts from
5509 the host. This can be undone locally by running "mount
5510 --make-rprivate /" if needed.
5512 * The prefdm.service file has been removed. Distributions
5513 should maintain this unit downstream if they intend to keep
5514 it around. However, we recommend writing normal unit files
5515 for display managers instead.
5517 * Since systemd is a crucial part of the OS we will now
5518 default to a number of compiler switches that improve
5519 security (hardening) such as read-only relocations, stack
5520 protection, and suchlike.
5522 * The TimeoutSec= setting for services is now split into
5523 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5524 of individual time outs for the start and the stop phase of
5527 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5528 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5529 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5530 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5531 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5532 Gundersen, Zbigniew Jędrzejewski-Szmek
5536 * The journal and id128 C APIs are now fully documented as man
5539 * Extra safety checks have been added when transitioning from
5540 the initial RAM disk to the main system to avoid accidental
5543 * /etc/crypttab entries now understand the new keyfile-offset=
5546 * systemctl -t can now be used to filter by unit load state.
5548 * The journal C API gained the new sd_journal_wait() call to
5549 make writing synchronous journal clients easier.
5551 * journalctl gained the new -D switch to show journals from a
5554 * journalctl now displays a special marker between log
5555 messages of two different boots.
5557 * The journal is now explicitly flushed to /var via a service
5558 systemd-journal-flush.service, rather than implicitly simply
5559 by seeing /var/log/journal to be writable.
5561 * journalctl (and the journal C APIs) can now match for much
5562 more complex expressions, with alternatives and
5565 * When transitioning from the initial RAM disk to the main
5566 system we will now kill all processes in a killing spree to
5567 ensure no processes stay around by accident.
5569 * Three new specifiers may be used in unit files: %u, %h, %s
5570 resolve to the user name, user home directory resp. user
5571 shell. This is useful for running systemd user instances.
5573 * We now automatically rotate journal files if their data
5574 object hash table gets a fill level > 75%. We also size the
5575 hash table based on the configured maximum file size. This
5576 together should lower hash collisions drastically and thus
5577 speed things up a bit.
5579 * journalctl gained the new "--header" switch to introspect
5580 header data of journal files.
5582 * A new setting SystemCallFilters= has been added to services
5583 which may be used to apply blacklists or whitelists to
5584 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5586 * nspawn gained a new --link-journal= switch (and quicker: -j)
5587 to link the container journal with the host. This makes it
5588 very easy to centralize log viewing on the host for all
5589 guests while still keeping the journal files separated.
5591 * Many bugfixes and optimizations
5593 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5594 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5595 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5600 * Several tools now understand kernel command line arguments,
5601 which are only read when run in an initial RAM disk. They
5602 usually follow closely their normal counterparts, but are
5605 * There's a new tool to analyze the readahead files that are
5606 automatically generated at boot. Use:
5608 /usr/lib/systemd/systemd-readahead analyze /.readahead
5610 * We now provide an early debug shell on tty9 if this enabled. Use:
5612 systemctl enable debug-shell.service
5614 * All plymouth related units have been moved into the Plymouth
5615 package. Please make sure to upgrade your Plymouth version
5618 * systemd-tmpfiles now supports getting passed the basename of
5619 a configuration file only, in which case it will look for it
5620 in all appropriate directories automatically.
5622 * udevadm info now takes a /dev or /sys path as argument, and
5623 does the right thing. Example:
5625 udevadm info /dev/sda
5626 udevadm info /sys/class/block/sda
5628 * systemctl now prints a warning if a unit is stopped but a
5629 unit that might trigger it continues to run. Example: a
5630 service is stopped but the socket that activates it is left
5633 * "systemctl status" will now mention if the log output was
5634 shortened due to rotation since a service has been started.
5636 * The journal API now exposes functions to determine the
5637 "cutoff" times due to rotation.
5639 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5640 immediately flushing of runtime logs to /var if possible,
5641 resp. for triggering immediate rotation of the journal
5644 * It is now considered an error if a service is attempted to
5645 be stopped that is not loaded.
5647 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5649 * systemd-analyze now supports Python 3
5651 * tmpfiles now supports cleaning up directories via aging
5652 where the first level dirs are always kept around but
5653 directories beneath it automatically aged. This is enabled
5654 by prefixing the age field with '~'.
5656 * Seat objects now expose CanGraphical, CanTTY properties
5657 which is required to deal with very fast bootups where the
5658 display manager might be running before the graphics drivers
5659 completed initialization.
5661 * Seat objects now expose a State property.
5663 * We now include RPM macros for service enabling/disabling
5664 based on the preset logic. We recommend RPM based
5665 distributions to make use of these macros if possible. This
5666 makes it simpler to reuse RPM spec files across
5669 * We now make sure that the collected systemd unit name is
5670 always valid when services log to the journal via
5673 * There's a new man page kernel-command-line(7) detailing all
5674 command line options we understand.
5676 * The fstab generator may now be disabled at boot by passing
5677 fstab=0 on the kernel command line.
5679 * A new kernel command line option modules-load= is now understood
5680 to load a specific kernel module statically, early at boot.
5682 * Unit names specified on the systemctl command line are now
5683 automatically escaped as needed. Also, if file system or
5684 device paths are specified they are automatically turned
5685 into the appropriate mount or device unit names. Example:
5687 systemctl status /home
5688 systemctl status /dev/sda
5690 * The SysVConsole= configuration option has been removed from
5691 system.conf parsing.
5693 * The SysV search path is no longer exported on the D-Bus
5696 * The Names= option is been removed from unit file parsing.
5698 * There's a new man page bootup(7) detailing the boot process.
5700 * Every unit and every generator we ship with systemd now
5701 comes with full documentation. The self-explanatory boot is
5704 * A couple of services gained "systemd-" prefixes in their
5705 name if they wrap systemd code, rather than only external
5706 code. Among them fsck@.service which is now
5707 systemd-fsck@.service.
5709 * The HaveWatchdog property has been removed from the D-Bus
5712 * systemd.confirm_spawn= on the kernel command line should now
5715 * There's a new man page crypttab(5) which details all options
5716 we actually understand.
5718 * systemd-nspawn gained a new --capability= switch to pass
5719 additional capabilities to the container.
5721 * timedated will now read known NTP implementation unit names
5722 from /usr/lib/systemd/ntp-units.d/*.list,
5723 systemd-timedated-ntp.target has been removed.
5725 * journalctl gained a new switch "-b" that lists log data of
5726 the current boot only.
5728 * The notify socket is in the abstract namespace again, in
5729 order to support daemons which chroot() at start-up.
5731 * There is a new Storage= configuration option for journald
5732 which allows configuration of where log data should go. This
5733 also provides a way to disable journal logging entirely, so
5734 that data collected is only forwarded to the console, the
5735 kernel log buffer or another syslog implementation.
5737 * Many bugfixes and optimizations
5739 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5740 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5741 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5742 Shawn Landden, Tom Gundersen
5746 * "systemctl help <unit>" now shows the man page if one is
5749 * Several new man pages have been added.
5751 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5752 MaxLevelConsole= can now be specified in
5753 journald.conf. These options allow reducing the amount of
5754 data stored on disk or forwarded by the log level.
5756 * TimerSlackNSec= can now be specified in system.conf for
5757 PID1. This allows system-wide power savings.
5759 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5760 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5765 * logind is now capable of (optionally) handling power and
5766 sleep keys as well as the lid switch.
5768 * journalctl now understands the syntax "journalctl
5769 /usr/bin/avahi-daemon" to get all log output of a specific
5772 * CapabilityBoundingSet= in system.conf now also influences
5773 the capability bound set of usermode helpers of the kernel.
5775 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5776 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5777 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5778 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5782 * Note that we skipped 139 releases here in order to set the
5783 new version to something that is greater than both udev's
5784 and systemd's most recent version number.
5786 * udev: all udev sources are merged into the systemd source tree now.
5787 All future udev development will happen in the systemd tree. It
5788 is still fully supported to use the udev daemon and tools without
5789 systemd running, like in initramfs or other init systems. Building
5790 udev though, will require the *build* of the systemd tree, but
5791 udev can be properly *run* without systemd.
5793 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5794 should be used to create dead device nodes as workarounds for broken
5797 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5798 no longer supported. udev_monitor_new_from_netlink() needs to be
5799 used to subscribe to events.
5801 * udev: when udevd is started by systemd, processes which are left
5802 behind by forking them off of udev rules, are unconditionally cleaned
5803 up and killed now after the event handling has finished. Services or
5804 daemons must be started as systemd services. Services can be
5805 pulled-in by udev to get started, but they can no longer be directly
5806 forked by udev rules.
5808 * udev: the daemon binary is called systemd-udevd now and installed
5809 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5810 to adapt to that, create symlink, or rename the binary after building
5813 * libudev no longer provides these symbols:
5814 udev_monitor_from_socket()
5815 udev_queue_get_failed_list_entry()
5816 udev_get_{dev,sys,run}_path()
5817 The versions number was bumped and symbol versioning introduced.
5819 * systemd-loginctl and systemd-journalctl have been renamed
5820 to loginctl and journalctl to match systemctl.
5822 * The config files: /etc/systemd/systemd-logind.conf and
5823 /etc/systemd/systemd-journald.conf have been renamed to
5824 logind.conf and journald.conf. Package updates should rename
5825 the files to the new names on upgrade.
5827 * For almost all files the license is now LGPL2.1+, changed
5828 from the previous GPL2.0+. Exceptions are some minor stuff
5829 of udev (which will be changed to LGPL2.1 eventually, too),
5830 and the MIT licensed sd-daemon.[ch] library that is suitable
5831 to be used as drop-in files.
5833 * systemd and logind now handle system sleep states, in
5834 particular suspending and hibernating.
5836 * logind now implements a sleep/shutdown/idle inhibiting logic
5837 suitable for a variety of uses. Soonishly Lennart will blog
5838 about this in more detail.
5840 * var-run.mount and var-lock.mount are no longer provided
5841 (which prevously bind mounted these directories to their new
5842 places). Distributions which have not converted these
5843 directories to symlinks should consider stealing these files
5844 from git history and add them downstream.
5846 * We introduced the Documentation= field for units and added
5847 this to all our shipped units. This is useful to make it
5848 easier to explore the boot and the purpose of the various
5851 * All smaller setup units (such as
5852 systemd-vconsole-setup.service) now detect properly if they
5853 are run in a container and are skipped when
5854 appropriate. This guarantees an entirely noise-free boot in
5855 Linux container environments such as systemd-nspawn.
5857 * A framework for implementing offline system updates is now
5858 integrated, for details see:
5859 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5861 * A new service type Type=idle is available now which helps us
5862 avoiding ugly interleaving of getty output and boot status
5865 * There's now a system-wide CapabilityBoundingSet= option to
5866 globally reduce the set of capabilities for the
5867 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5868 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5869 even CAP_NET_ADMIN system-wide for secure systems.
5871 * There are now system-wide DefaultLimitXXX= options to
5872 globally change the defaults of the various resource limits
5873 for all units started by PID 1.
5875 * Harald Hoyer's systemd test suite has been integrated into
5876 systemd which allows easy testing of systemd builds in qemu
5877 and nspawn. (This is really awesome! Ask us for details!)
5879 * The fstab parser is now implemented as generator, not inside
5882 * systemctl will now warn you if .mount units generated from
5883 /etc/fstab are out of date due to changes in fstab that
5884 have not been read by systemd yet.
5886 * systemd is now suitable for usage in initrds. Dracut has
5887 already been updated to make use of this. With this in place
5888 initrds get a slight bit faster but primarily are much
5889 easier to introspect and debug since "systemctl status" in
5890 the host system can be used to introspect initrd services,
5891 and the journal from the initrd is kept around too.
5893 * systemd-delta has been added, a tool to explore differences
5894 between user/admin configuration and vendor defaults.
5896 * PrivateTmp= now affects both /tmp and /var/tmp.
5898 * Boot time status messages are now much prettier and feature
5899 proper english language. Booting up systemd has never been
5902 * Read-ahead pack files now include the inode number of all
5903 files to pre-cache. When the inode changes the pre-caching
5904 is not attempted. This should be nicer to deal with updated
5905 packages which might result in changes of read-ahead
5908 * We now temporaritly lower the kernel's read_ahead_kb variable
5909 when collecting read-ahead data to ensure the kernel's
5910 built-in read-ahead does not add noise to our measurements
5911 of necessary blocks to pre-cache.
5913 * There's now RequiresMountsFor= to add automatic dependencies
5914 for all mounts necessary for a specific file system path.
5916 * MountAuto= and SwapAuto= have been removed from
5917 system.conf. Mounting file systems at boot has to take place
5920 * nspawn now learned a new switch --uuid= to set the machine
5921 ID on the command line.
5923 * nspawn now learned the -b switch to automatically search
5926 * vt102 is now the default TERM for serial TTYs, upgraded from
5929 * systemd-logind now works on VT-less systems.
5931 * The build tree has been reorganized. The individual
5932 components now have directories of their own.
5934 * A new condition type ConditionPathIsReadWrite= is now available.
5936 * nspawn learned the new -C switch to create cgroups for the
5937 container in other hierarchies.
5939 * We now have support for hardware watchdogs, configurable in
5942 * The scheduled shutdown logic now has a public API.
5944 * We now mount /tmp as tmpfs by default, but this can be
5945 masked and /etc/fstab can override it.
5947 * Since udisks does not make use of /media anymore we are not
5948 mounting a tmpfs on it anymore.
5950 * journalctl gained a new --local switch to only interleave
5951 locally generated journal files.
5953 * We can now load the IMA policy at boot automatically.
5955 * The GTK tools have been split off into a systemd-ui.
5957 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5958 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5959 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5960 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5961 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5962 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5963 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5964 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5965 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5970 * This is mostly a bugfix release
5972 * Support optional initialization of the machine ID from the
5973 KVM or container configured UUID.
5975 * Support immediate reboots with "systemctl reboot -ff"
5977 * Show /etc/os-release data in systemd-analyze output
5979 * Many bugfixes for the journal, including endianness fixes and
5980 ensuring that disk space enforcement works
5982 * sd-login.h is C++ comptaible again
5984 * Extend the /etc/os-release format on request of the Debian
5987 * We now refuse non-UTF8 strings used in various configuration
5988 and unit files. This is done to ensure we do not pass invalid
5989 data over D-Bus or expose it elsewhere.
5991 * Register Mimo USB Screens as suitable for automatic seat
5994 * Read SELinux client context from journal clients in a race
5997 * Reorder configuration file lookup order. /etc now always
5998 overrides /run in order to allow the administrator to always
5999 and unconditionally override vendor-supplied or
6000 automatically generated data.
6002 * The various user visible bits of the journal now have man
6003 pages. We still lack man pages for the journal API calls
6006 * We now ship all man pages in HTML format again in the
6009 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6010 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6011 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
6016 * This is mostly a bugfix release
6018 * systems lacking /etc/os-release are no longer supported.
6020 * Various functionality updates to libsystemd-login.so
6022 * Track class of PAM logins to distinguish greeters from
6025 Contributions from: Kay Sievers, Lennart Poettering, Michael
6030 * This is an important bugfix release for v41.
6032 * Building man pages is now optional which should be useful
6033 for those building systemd from git but unwilling to install
6036 * Watchdog support for supervising services is now usable. In
6037 a future release support for hardware watchdogs
6038 (i.e. /dev/watchdog) will be added building on this.
6040 * Service start rate limiting is now configurable and can be
6041 turned off per service. When a start rate limit is hit a
6042 reboot can automatically be triggered.
6044 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6046 Contributions from: Benjamin Franzke, Bill Nottingham,
6047 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6048 Schmidt, Michał Górny, Piotr Drąg
6052 * The systemd binary is installed /usr/lib/systemd/systemd now;
6053 An existing /sbin/init symlink needs to be adapted with the
6056 * The code that loads kernel modules has been ported to invoke
6057 libkmod directly, instead of modprobe. This means we do not
6058 support systems with module-init-tools anymore.
6060 * Watchdog support is now already useful, but still not
6063 * A new kernel command line option systemd.setenv= is
6064 understood to set system wide environment variables
6065 dynamically at boot.
6067 * We now limit the set of capabilities of systemd-journald.
6069 * We now set SIGPIPE to ignore by default, since it only is
6070 useful in shell pipelines, and has little use in general
6071 code. This can be disabled with IgnoreSIPIPE=no in unit
6074 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6075 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6080 * This is mostly a bugfix release
6082 * We now expose the reason why a service failed in the
6083 "Result" D-Bus property.
6085 * Rudimentary service watchdog support (will be completed over
6086 the next few releases.)
6088 * When systemd forks off in order execute some service we will
6089 now immediately changes its argv[0] to reflect which process
6090 it will execute. This is useful to minimize the time window
6091 with a generic argv[0], which makes bootcharts more useful
6093 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6094 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6095 Mike Kazantsev, Ray Strode
6099 * This is mostly a test release, but incorporates many
6102 * New systemd-cgtop tool to show control groups by their
6105 * Linking against libacl for ACLs is optional again. If
6106 disabled, support tracking device access for active logins
6107 goes becomes unavailable, and so does access to the user
6108 journals by the respective users.
6110 * If a group "adm" exists, journal files are automatically
6111 owned by them, thus allow members of this group full access
6112 to the system journal as well as all user journals.
6114 * The journal now stores the SELinux context of the logging
6115 client for all entries.
6117 * Add C++ inclusion guards to all public headers
6119 * New output mode "cat" in the journal to print only text
6120 messages, without any meta data like date or time.
6122 * Include tiny X server wrapper as a temporary stop-gap to
6123 teach XOrg udev display enumeration. This is used by display
6124 managers such as gdm, and will go away as soon as XOrg
6125 learned native udev hotplugging for display devices.
6127 * Add new systemd-cat tool for executing arbitrary programs
6128 with STDERR/STDOUT connected to the journal. Can also act as
6129 BSD logger replacement, and does so by default.
6131 * Optionally store all locally generated coredumps in the
6132 journal along with meta data.
6134 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6135 writing short strings to files (for usage for /sys), and for
6136 creating symlinks, character and block device nodes.
6138 * New unit file option ControlGroupPersistent= to make cgroups
6139 persistent, following the mechanisms outlined in
6140 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6142 * Support multiple local RTCs in a sane way
6144 * No longer monopolize IO when replaying readahead data on
6145 rotating disks, since we might starve non-file-system IO to
6146 death, since fanotify() will not see accesses done by blkid,
6149 * Do not show kernel threads in systemd-cgls anymore, unless
6150 requested with new -k switch.
6152 Contributions from: Dan Horák, Kay Sievers, Lennart
6153 Poettering, Michal Schmidt
6157 * This is mostly a test release, but incorporates many
6160 * The git repository moved to:
6161 git://anongit.freedesktop.org/systemd/systemd
6162 ssh://git.freedesktop.org/git/systemd/systemd
6164 * First release with the journal
6165 http://0pointer.de/blog/projects/the-journal.html
6167 * The journal replaces both systemd-kmsg-syslogd and
6168 systemd-stdout-bridge.
6170 * New sd_pid_get_unit() API call in libsystemd-logind
6172 * Many systemadm clean-ups
6174 * Introduce remote-fs-pre.target which is ordered before all
6175 remote mounts and may be used to start services before all
6178 * Added Mageia support
6180 * Add bash completion for systemd-loginctl
6182 * Actively monitor PID file creation for daemons which exit in
6183 the parent process before having finished writing the PID
6184 file in the daemon process. Daemons which do this need to be
6185 fixed (i.e. PID file creation must have finished before the
6186 parent exits), but we now react a bit more gracefully to them.
6188 * Add colourful boot output, mimicking the well-known output
6189 of existing distributions.
6191 * New option PassCredentials= for socket units, for
6192 compatibility with a recent kernel ABI breakage.
6194 * /etc/rc.local is now hooked in via a generator binary, and
6195 thus will no longer act as synchronization point during
6198 * systemctl list-unit-files now supports --root=.
6200 * systemd-tmpfiles now understands two new commands: z, Z for
6201 relabelling files according to the SELinux database. This is
6202 useful to apply SELinux labels to specific files in /sys,
6205 * Output of SysV services is now forwarded to both the console
6206 and the journal by default, not only just the console.
6208 * New man pages for all APIs from libsystemd-login.
6210 * The build tree got reorganized and a the build system is a
6211 lot more modular allowing embedded setups to specifically
6212 select the components of systemd they are interested in.
6214 * Support for Linux systems lacking the kernel VT subsystem is
6217 * configure's --with-rootdir= got renamed to
6218 --with-rootprefix= to follow the naming used by udev and
6221 * Unless specified otherwise we will now install to /usr instead
6222 of /usr/local by default.
6224 * Processes with '@' in argv[0][0] are now excluded from the
6225 final shut-down killing spree, following the logic explained
6227 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6229 * All processes remaining in a service cgroup when we enter
6230 the START or START_PRE states are now killed with
6231 SIGKILL. That means it is no longer possible to spawn
6232 background processes from ExecStart= lines (which was never
6233 supported anyway, and bad style).
6235 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6236 reloading of units together.
6238 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6239 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6240 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6241 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6242 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob