1 systemd System and Service Manager
3 CHANGES WITH 237 in spe:
5 * Some keyboards come with a zoom see-saw or rocker which until now got
6 mapped to the Linux "zoomin/out" keys in hwdb. However, these
7 keycodes are not recognized by any major desktop. They now produce
8 Up/Down key events so that they can be used for scrolling.
10 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
11 slightly: previously, if an argument was specified for lines of this
12 type (i.e. the right-most column was set) this string was appended to
13 existing files each time elogind-tmpfiles was run. This behaviour was
14 different from what the documentation said, and not particularly
15 useful, as repeated elogind-tmpfiles invocations would not be
16 idempotent and grow such files without bounds. With this release
17 behaviour has been altered slightly, to match what the documentation
18 says: lines of this type only have an effect if the indicated files
19 don't exist yet, and only then the argument string is written to the
22 * A new environment variable $SYSTEMD_OFFLINE is now understood by
23 systemctl. It takes a boolean argument. If on, systemctl assumes it
24 operates on an "offline" OS tree, and will not attempt to talk to the
25 service manager. Previously, this mode was implicitly enabled if a
26 chroot() environment was detected, and this new environment variable
27 now provides explicit control.
29 * .path and .socket units may now be created transiently,
30 too. Previously only service, mount, automount and timer units were
31 supported as transient units. The elogind-run tool has been updated
32 to expose this new functionality, you may hence use it now to bind
33 arbitrary commands to path or socket activation on-the-fly from the
34 command line. Moreover, for the unit types already supporting
35 transient operation most properties are now hooked up.
37 * The elogind-mount command gained support for a new --owner= parameter
38 which takes a user name, which is then resolved and included in uid=
39 and gid= mount options string of the file system to mount.
41 * A new unit condition ConditionControlGroupController= has been added
42 that checks whether a specific cgroup controller is available.
44 * Unit files, udev's .link files, and elogind-networkd's .netdev and
45 .network files all gained support for a new condition
46 ConditionKernelVersion= for checking against specific kernel
49 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
50 support for configuring device flags in the Flags= setting. In the
51 same files, the [Tunnel] section gained support for configuring
52 AllowLocalRemote=. The [Route] section in .network files gained
53 support for configuring InitialCongestionWindow=,
54 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
55 understands RapidCommit=.
57 * elogind-networkd's DHCPv6 support gained support for Prefix
60 * sd-bus gained support for a new "watch-bind" feature. When this
61 feature is enabled, an sd_bus connection may be set up to connect to
62 an AF_UNIX socket in the file system as soon as it is created. This
63 functionality is useful for writing early-boot services that
64 automatically connect to the system bus as soon as it is started,
65 without ugly time-based polling. elogind-networkd and
66 elogind-resolved have been updated to make use of this
67 functionality. busctl exposes this functionality in a new
68 --watch-bind= command line switch.
70 * sd-bus will now optionally synthesize a local "Connected" signal as
71 soon as a D-Bus connection is set up fully. This message mirrors the
72 already existing "Disconnected" signal which is synthesized when the
73 connection is terminated. This signal is generally useful but
74 particularly handy in combination with the "watch-bind" feature
75 described above. Synthesizing of this message has to be requested
76 explicitly through the new API call sd_bus_set_connected_signal(). In
77 addition a new call sd_bus_is_ready() has been added that checks
78 whether a connection is fully set up (i.e. between the "Connected" and
79 "Disconnected" signals).
81 * sd-bus gained two new calls sd_bus_request_name_async() and
82 sd_bus_release_name_async() for asynchronously registering bus
83 names. Similar, there is now sd_bus_add_match_async() for installing
84 a signal match asynchronously. All of elogind's own services have
85 been updated to make use of these calls. Doing these operations
86 asynchronously has two benefits: it reduces the risk of deadlocks in
87 case of cyclic dependencies between bus services, and it speeds up
88 service initialization since synchronization points for bus
89 round-trips are removed.
91 * sd-bus gained two new calls sd_bus_match_signal() and
92 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
93 and sd_bus_add_match_async() but instead of taking a D-Bus match
94 string take match fields as normal function parameters.
96 * sd-bus gained two new calls sd_bus_set_sender() and
97 sd_bus_message_set_sender() for setting the sender name of outgoing
98 messages (either for all outgoing messages or for just one specific
99 one). These calls are only useful in direct connections as on
100 brokered connections the broker fills in the sender anyway,
101 overwriting whatever the client filled in.
103 * sd-event gained a new pseudo-handle that may be specified on all API
104 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
105 used this refers to the default event loop object of the calling
106 thread. Note however that this does not implicitly allocate one —
107 which has to be done prior by using sd_event_default(). Similarly
108 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
109 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
110 to the default bus of the specified type of the calling thread. Here
111 too this does not implicitly allocate bus connection objects, this
112 has to be done prior with sd_bus_default() and friends.
114 * sd-event gained a new call pair
115 sd_event_source_{get|set}_io_fd_own(). This may be used to request
116 automatic closure of the file descriptor an IO event source watches
117 when the event source is destroyed.
119 * elogind-networkd gained support for natively configuring WireGuard
122 * In previous versions elogind synthesized user records both for the
123 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
124 internally. In order to simplify distribution-wide renames of the
125 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
126 new transitional flag file has been added: if
127 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
128 user and group record within the elogind codebase is disabled.
130 * elogind-notify gained a new --uid= option for selecting the source
131 user/UID to use for notification messages sent to the service
134 * journalctl gained a new --grep option to list only entries in which
135 the message matches a certain pattern. By default matching is case
136 insensitive if the pattern is lowercase, and case sensitive
137 otherwise. Option --case-sensitive=yes|no can be used to override
138 this an specify case sensitivity or case insensitivity.
140 * There's now a "elogind-analyze service-watchdogs" command for printing
141 the current state of the service runtime watchdog, and optionally
142 enabling or disabling the per-service watchdogs system-wide if given a
143 boolean argument (i.e. the concept you configure in WatchdogSec=), for
144 debugging purposes. There's also a kernel command line option
145 elogind.service_watchdogs= for controlling the same.
147 * Two new "log-level" and "log-target" options for elogind-analyze were
148 addded that merge the now deprecated get-log-level, set-log-level and
149 get-log-target, set-log-target pairs. The deprecated options are still
150 understood for backwards compatibility. The two new options print the
151 current value when no arguments are given, and set them when a
152 level/target is given as an argument.
154 * sysusers.d's "u" lines now optionally accept both a UID and a GID
155 specification, separated by a ":" character, in order to create users
156 where UID and GID do not match.
158 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
159 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
160 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
161 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
162 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
163 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
164 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
165 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
166 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
167 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
168 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
169 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
170 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
171 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
172 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
173 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
174 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
175 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
176 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
177 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
184 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
185 in v235 has been extended to also set the dummy.ko module option
186 numdummies=0, preventing the kernel from automatically creating
187 dummy0. All dummy interfaces must now be explicitly created.
189 * Unknown '%' specifiers in configuration files are now rejected. This
190 applies to units and tmpfiles.d configuration. Any percent characters
191 that are followed by a letter or digit that are not supposed to be
192 interpreted as the beginning of a specifier should be escaped by
193 doubling ("%%"). (So "size=5%" is still accepted, as well as
194 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
195 valid specifiers today.)
197 * systemd-resolved now maintains a new dynamic
198 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
199 recommended to make /etc/resolv.conf a symlink to it. This file
200 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
201 includes dynamically acquired search domains, achieving more correct
202 DNS resolution by software that bypasses local DNS APIs such as NSS.
204 * The "uaccess" udev tag has been dropped from /dev/kvm and
205 /dev/dri/renderD*. These devices now have the 0666 permissions by
206 default (but this may be changed at build-time). /dev/dri/renderD*
207 will now be owned by the "render" group along with /dev/kfd.
209 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
210 systemd-journal-gatewayd.service and
211 systemd-journal-upload.service. This means "nss-systemd" must be
212 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
213 services are resolved properly.
215 * In /etc/fstab two new mount options are now understood:
216 x-systemd.makefs and x-systemd.growfs. The former has the effect that
217 the configured file system is formatted before it is mounted, the
218 latter that the file system is resized to the full block device size
219 after it is mounted (i.e. if the file system is smaller than the
220 partition it resides on, it's grown). This is similar to the fsck
221 logic in /etc/fstab, and pulls in systemd-makefs@.service and
222 systemd-growfs@.service as necessary, similar to
223 systemd-fsck@.service. Resizing is currently only supported on ext4
226 * In systemd-networkd, the IPv6 RA logic now optionally may announce
227 DNS server and domain information.
229 * Support for the LUKS2 on-disk format for encrypted partitions has
230 been added. This requires libcryptsetup2 during compilation and
233 * The systemd --user instance will now signal "readiness" when its
234 basic.target unit has been reached, instead of when the run queue ran
235 empty for the first time.
237 * Tmpfiles.d with user configuration are now also supported.
238 systemd-tmpfiles gained a new --user switch, and snippets placed in
239 ~/.config/user-tmpfiles.d/ and corresponding directories will be
240 executed by systemd-tmpfiles --user running in the new
241 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
242 running in the user session.
244 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
245 %S resolves to the top-level state directory (/var/lib for the system
246 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
247 top-level cache directory (/var/cache for the system instance,
248 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
249 logs directory (/var/log for the system instance,
250 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
251 existing %t specifier, that resolves to the top-level runtime
252 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
255 * journalctl learnt a new parameter --output-fields= for limiting the
256 set of journal fields to output in verbose and JSON output modes.
258 * systemd-timesyncd's configuration file gained a new option
259 RootDistanceMaxSec= for setting the maximum root distance of servers
260 it'll use, as well as the new options PollIntervalMinSec= and
261 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
263 * bootctl gained a new command "list" for listing all available boot
264 menu items on systems that follow the boot loader specification.
266 * systemctl gained a new --dry-run switch that shows what would be done
267 instead of doing it, and is currently supported by the shutdown and
270 * ConditionSecurity= can now detect the TOMOYO security module.
272 * Unit file [Install] sections are now also respected in unit drop-in
273 files. This is intended to be used by drop-ins under /usr/lib/.
275 * systemd-firstboot may now also set the initial keyboard mapping.
277 * Udev "changed" events for devices which are exposed as systemd
278 .device units are now propagated to units specified in
279 ReloadPropagatedFrom= as reload requests.
281 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
282 unit template name (i.e. a name in the form of 'foobar@.service',
283 without the instance component between the '@' and - the '.'), then
284 the escaped sysfs path of the device is automatically used as the
287 * SystemCallFilter= in unit files has been extended so that an "errno"
288 can be specified individually for each system call. Example:
289 SystemCallFilter=~uname:EILSEQ.
291 * The cgroup delegation logic has been substantially updated. Delegate=
292 now optionally takes a list of controllers (instead of a boolean, as
293 before), which lists the controllers to delegate at least.
295 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
297 * A new LogLevelMax= setting configures the maximum log level any
298 process of the service may log at (i.e. anything with a lesser
299 priority than what is specified is automatically dropped). A new
300 LogExtraFields= setting allows configuration of additional journal
301 fields to attach to all log records generated by any of the unit's
304 * New StandardInputData= and StandardInputText= settings along with the
305 new option StandardInput=data may be used to configure textual or
306 binary data that shall be passed to the executed service process via
307 standard input, encoded in-line in the unit file.
309 * StandardInput=, StandardOutput= and StandardError= may now be used to
310 connect stdin/stdout/stderr of executed processes directly with a
311 file or AF_UNIX socket in the file system, using the new "file:" option.
313 * A new unit file option CollectMode= has been added, that allows
314 tweaking the garbage collection logic for units. It may be used to
315 tell systemd to garbage collect units that have failed automatically
316 (normally it only GCs units that exited successfully). systemd-run
317 and systemd-mount expose this new functionality with a new -G option.
319 * "machinectl bind" may now be used to bind mount non-directories
320 (i.e. regularfiles, devices, fifos, sockets).
322 * systemd-analyze gained a new verb "calendar" for validating and
323 testing calendar time specifications to use for OnCalendar= in timer
324 units. Besides validating the expression it will calculate the next
325 time the specified expression would elapse.
327 * In addition to the pre-existing FailureAction= unit file setting
328 there's now SuccessAction=, for configuring a shutdown action to
329 execute when a unit completes successfully. This is useful in
330 particular inside containers that shall terminate after some workload
331 has been completed. Also, both options are now supported for all unit
332 types, not just services.
334 * networkds's IP rule support gained two new options
335 IncomingInterface= and OutgoingInterface= for configuring the incoming
336 and outgoing interfaces of configured rules. systemd-networkd also
337 gained support for "vxcan" network devices.
339 * networkd gained a new setting RequiredForOnline=, taking a
340 boolean. If set, systemd-wait-online will take it into consideration
341 when determining that the system is up, otherwise it will ignore the
342 interface for this purpose.
344 * The sd_notify() protocol gained support for a new operation: with
345 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
346 store again, ahead of POLLHUP or POLLERR when they are removed
349 * A new document UIDS-GIDS.md has been added to the source tree, that
350 documents the UID/GID range and assignment assumptions and
351 requirements of systemd.
353 * The watchdog device PID 1 will ping may now be configured through the
354 WatchdogDevice= configuration file setting, or by setting the
355 systemd.watchdog_service= kernel commandline option.
357 * systemd-resolved's gained support for registering DNS-SD services on
358 the local network using MulticastDNS. Services may either be
359 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
360 the same dir below /run, /usr/lib), or through its D-Bus API.
362 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
363 extend the effective start, runtime, and stop time. The service must
364 continue to send EXTEND_TIMEOUT_USEC within the period specified to
365 prevent the service manager from making the service as timedout.
367 * elogind-resolved's DNSSEC support gained support for RFC 8080
368 (Ed25519 keys and signatures).
370 * The elogind-resolve command line tool gained a new set of options
371 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
372 --set-nta= and --revert to configure per-interface DNS configuration
373 dynamically during runtime. It's useful for pushing DNS information
374 into elogind-resolved from DNS hook scripts that various interface
375 managing software supports (such as pppd).
377 * elogind-nspawn gained a new --network-namespace-path= command line
378 option, which may be used to make a container join an existing
379 network namespace, by specifying a path to a "netns" file.
381 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
382 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
383 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
384 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
385 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
386 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
387 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
388 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
389 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
390 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
391 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
392 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
393 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
394 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
395 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
396 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
397 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
398 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
399 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
400 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
401 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
402 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
403 Jędrzejewski-Szmek, Zeal Jagannatha
409 * INCOMPATIBILITY: systemd-logind.service and other long-running
410 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
411 communication with the outside. This generally improves security of
412 the system, and is in almost all cases a safe and good choice, as
413 these services do not and should not provide any network-facing
414 functionality. However, systemd-logind uses the glibc NSS API to
415 query the user database. This creates problems on systems where NSS
416 is set up to directly consult network services for user database
417 lookups. In particular, this creates incompatibilities with the
418 "nss-nis" module, which attempts to directly contact the NIS/YP
419 network servers it is configured for, and will now consistently
420 fail. In such cases, it is possible to turn off IP sandboxing for
421 systemd-logind.service (set IPAddressDeny= in its [Service] section
422 to the empty string, via a .d/ unit file drop-in). Downstream
423 distributions might want to update their nss-nis packaging to include
424 such a drop-in snippet, accordingly, to hide this incompatibility
425 from the user. Another option is to make use of glibc's nscd service
426 to proxy such network requests through a privilege-separated, minimal
427 local caching daemon, or to switch to more modern technologies such
428 sssd, whose NSS hook-ups generally do not involve direct network
429 access. In general, we think it's definitely time to question the
430 implementation choices of nss-nis, i.e. whether it's a good idea
431 today to embed a network-facing loadable module into all local
432 processes that need to query the user database, including the most
433 trivial and benign ones, such as "ls". For more details about
434 IPAddressDeny= see below.
436 * A new modprobe.d drop-in is now shipped by default that sets the
437 bonding module option max_bonds=0. This overrides the kernel default,
438 to avoid conflicts and ambiguity as to whether or not bond0 should be
439 managed by systemd-networkd or not. This resolves multiple issues
440 with bond0 properties not being applied, when bond0 is configured
441 with systemd-networkd. Distributors may choose to not package this,
442 however in that case users will be prevented from correctly managing
443 bond0 interface using systemd-networkd.
445 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
446 which print the logging level and target of the system manager. They
447 complement the existing "set-log-level" and "set-log-target" verbs
448 used to change those values.
450 * journald.conf gained a new boolean setting ReadKMsg= which defaults
451 to on. If turned off kernel log messages will not be read by
452 systemd-journald or included in the logs. It also gained a new
453 setting LineMax= for configuring the maximum line length in
454 STDOUT/STDERR log streams. The new default for this value is 48K, up
455 from the previous hardcoded 2048.
457 * A new unit setting RuntimeDirectoryPreserve= has been added, which
458 allows more detailed control of what to do with a runtime directory
459 configured with RuntimeDirectory= (i.e. a directory below /run or
460 $XDG_RUNTIME_DIR) after a unit is stopped.
462 * The RuntimeDirectory= setting for units gained support for creating
463 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
464 one top-level directory.
466 * Units gained new options StateDirectory=, CacheDirectory=,
467 LogsDirectory= and ConfigurationDirectory= which are closely related
468 to RuntimeDirectory= but manage per-service directories below
469 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
470 possible to write unit files which when activated automatically gain
471 properly owned service specific directories in these locations, thus
472 making unit files self-contained and increasing compatibility with
473 stateless systems and factory reset where /etc or /var are
474 unpopulated at boot. Matching these new settings there's also
475 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
476 ConfigurationDirectoryMode= for configuring the access mode of these
477 directories. These settings are particularly useful in combination
478 with DynamicUser=yes as they provide secure, properly-owned,
479 writable, and stateful locations for storage, excluded from the
480 sandbox that such services live in otherwise.
482 * Automake support has been removed from this release. systemd is now
485 * systemd-journald will now aggressively cache client metadata during
486 runtime, speeding up log write performance under pressure. This comes
487 at a small price though: as much of the metadata is read
488 asynchronously from /proc/ (and isn't implicitly attached to log
489 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
490 metadata stored alongside a log entry might be slightly
491 out-of-date. Previously it could only be slightly newer than the log
492 message. The time window is small however, and given that the kernel
493 is unlikely to be improved anytime soon in this regard, this appears
496 * nss-myhostname/systemd-resolved will now by default synthesize an
497 A/AAAA resource record for the "_gateway" hostname, pointing to the
498 current default IP gateway. Previously it did that for the "gateway"
499 name, hampering adoption, as some distributions wanted to leave that
500 host name open for local use. The old behaviour may still be
501 requested at build time.
503 * systemd-networkd's [Address] section in .network files gained a new
504 Scope= setting for configuring the IP address scope. The [Network]
505 section gained a new boolean setting ConfigureWithoutCarrier= that
506 tells systemd-networkd to ignore link sensing when configuring the
507 device. The [DHCP] section gained a new Anonymize= boolean option for
508 turning on a number of options suggested in RFC 7844. A new
509 [RoutingPolicyRule] section has been added for configuring the IP
510 routing policy. The [Route] section has gained support for a new
511 Type= setting which permits configuring
512 blackhole/unreachable/prohibit routes.
514 * The [VRF] section in .netdev files gained a new Table= setting for
515 configuring the routing table to use. The [Tunnel] section gained a
516 new Independent= boolean field for configuring tunnels independent of
517 an underlying network interface. The [Bridge] section gained a new
518 GroupForwardMask= option for configuration of propagation of link
519 local frames between bridge ports.
521 * The WakeOnLan= setting in .link files gained support for a number of
522 new modes. A new TCP6SegmentationOffload= setting has been added for
523 configuring TCP/IPv6 hardware segmentation offload.
525 * The IPv6 RA sender implementation may now optionally send out RDNSS
526 and RDNSSL records to supply DNS configuration to peers.
528 * systemd-nspawn gained support for a new --system-call-filter= command
529 line option for adding and removing entries in the default system
530 call filter it applies. Moreover systemd-nspawn has been changed to
531 implement a system call whitelist instead of a blacklist.
533 * systemd-run gained support for a new --pipe command line option. If
534 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
535 are directly passed on to the activated transient service
536 executable. This allows invoking arbitrary processes as systemd
537 services (for example to take benefit of dependency management,
538 accounting management, resource management or log management that is
539 done automatically for services) — while still allowing them to be
540 integrated in a classic UNIX shell pipeline.
542 * When a service sends RELOAD=1 via sd_notify() and reload propagation
543 using ReloadPropagationTo= is configured, a reload is now propagated
544 to configured units. (Previously this was only done on explicitly
545 requested reloads, using "systemctl reload" or an equivalent
548 * For each service unit a restart counter is now kept: it is increased
549 each time the service is restarted due to Restart=, and may be
550 queried using "systemctl show -p NRestarts …".
552 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
553 @signal and @timer have been added, for usage with SystemCallFilter=
554 in unit files and the new --system-call-filter= command line option
555 of systemd-nspawn (see above).
557 * ExecStart= lines in unit files gained two new modifiers: when a
558 command line is prefixed with "!" the command will be executed as
559 configured, except for the credentials applied by
560 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
561 "+", but does still apply namespacing options unlike "+". There's
562 also "!!" now, which is mostly identical, but becomes a NOP on
563 systems that support ambient capabilities. This is useful to write
564 unit files that work with ambient capabilities where possible but
565 automatically fall back to traditional privilege dropping mechanisms
566 on systems where this is not supported.
568 * ListenNetlink= settings in socket units now support RDMA netlink
571 * A new unit file setting LockPersonality= has been added which permits
572 locking down the chosen execution domain ("personality") of a service
575 * A new special target "getty-pre.target" has been added, which is
576 ordered before all text logins, and may be used to order services
577 before textual logins acquire access to the console.
579 * systemd will now attempt to load the virtio-rng.ko kernel module very
580 early on if a VM environment supporting this is detected. This should
581 improve entropy during early boot in virtualized environments.
583 * A _netdev option is now supported in /etc/crypttab that operates in a
584 similar way as the same option in /etc/fstab: it permits configuring
585 encrypted devices that need to be ordered after the network is up.
586 Following this logic, two new special targets
587 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
588 added that are to cryptsetup.target what remote-fs.target and
589 remote-fs-pre.target are to local-fs.target.
591 * Service units gained a new UnsetEnvironment= setting which permits
592 unsetting specific environment variables for services that are
593 normally passed to it (for example in order to mask out locale
594 settings for specific services that can't deal with it).
596 * Units acquired a new boolean option IPAccounting=. When turned on, IP
597 traffic accounting (packet count as well as byte count) is done for
598 the service, and shown as part of "systemctl status" or "systemd-run
601 * Service units acquired two new options IPAddressAllow= and
602 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
603 for configuring a simple IP access control list for all sockets of
604 the unit. These options are available also on .slice and .socket
605 units, permitting flexible access list configuration for individual
606 services as well as groups of services (as defined by a slice unit),
607 including system-wide. Note that IP ACLs configured this way are
608 enforced on every single IPv4 and IPv6 socket created by any process
609 of the service unit, and apply to ingress as well as egress traffic.
611 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
612 structured log message is generated each time the unit is stopped,
613 containing information about the consumed resources of this
616 * A new setting KeyringMode= has been added to unit files, which may be
617 used to control how the kernel keyring is set up for executed
620 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
621 "systemctl kexec" and "systemctl exit" are now always asynchronous in
622 behaviour (that is: these commands return immediately after the
623 operation was enqueued instead of waiting for the operation to
624 complete). Previously, "systemctl poweroff" and "systemctl reboot"
625 were asynchronous on systems using systemd-logind (i.e. almost
626 always, and like they were on sysvinit), and the other three commands
627 were unconditionally synchronous. With this release this is cleaned
628 up, and callers will see the same asynchronous behaviour on all
629 systems for all five operations.
631 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
634 * .timer units now accept calendar specifications in other timezones
635 than UTC or the local timezone.
637 * The tmpfiles snippet var.conf has been changed to create
638 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
639 the "utmp" group already, and it appears to be generally understood
640 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
641 databases. Previously this was implemented correctly for all these
642 databases excepts btmp, which has been opened up like this now
643 too. Note that while the other databases are world-readable
644 (i.e. 0644), btmp is not and remains more restrictive.
646 * The systemd-resolve tool gained a new --reset-server-features
647 switch. When invoked like this systemd-resolved will forget
648 everything it learnt about the features supported by the configured
649 upstream DNS servers, and restarts the feature probing logic on the
650 next resolver look-up for them at the highest feature level
653 * The status dump systemd-resolved sends to the logs upon receiving
654 SIGUSR1 now also includes information about all DNS servers it is
655 configured to use, and the features levels it probed for them.
657 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
658 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
659 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
660 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
661 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
662 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
663 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
664 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
665 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
666 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
667 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
668 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
669 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
670 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
671 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
672 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
673 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
674 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
675 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
676 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
682 * Meson is now supported as build system in addition to Automake. It is
683 our plan to remove Automake in one of our next releases, so that
684 Meson becomes our exclusive build system. Hence, please start using
685 the Meson build system in your downstream packaging. There's plenty
686 of documentation around how to use Meson, the extremely brief
689 ./autogen.sh && ./configure && make && sudo make install
693 meson build && ninja -C build && sudo ninja -C build install
695 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
696 which permits configuring a timeout on the time a job is
697 running. This is particularly useful for setting timeouts on jobs for
700 * Unit files gained two new options ConditionUser= and ConditionGroup=
701 for conditionalizing units based on the identity of the user/group
702 running a systemd user instance.
704 * systemd-networkd now understands a new FlowLabel= setting in the
705 [VXLAN] section of .network files, as well as a Priority= in
706 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
707 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
708 gained support for configuration of GENEVE links, and IPv6 address
709 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
711 * .link files now understand a new Port= setting.
713 * systemd-networkd's DHCP support gained support for DHCP option 119
714 (domain search list).
716 * systemd-networkd gained support for serving IPv6 address ranges using
717 the Router Advertisment protocol. The new .network configuration
718 section [IPv6Prefix] may be used to configure the ranges to
719 serve. This is implemented based on a new, minimal, native server
720 implementation of RA.
722 * journalctl's --output= switch gained support for a new parameter
723 "short-iso-precise" for a mode where timestamps are shown as precise
726 * systemd-udevd's "net_id" builtin may now generate stable network
727 interface names from IBM PowerVM VIO devices as well as ACPI platform
730 * MulticastDNS support in systemd-resolved may now be explicitly
731 enabled/disabled using the new MulticastDNS= configuration file
734 * systemd-resolved may now optionally use libidn2 instead of the libidn
735 for processing internationalized domain names. Support for libidn2
736 should be considered experimental and should not be enabled by
739 * "machinectl pull-tar" and related call may now do verification of
740 downloaded images using SUSE-style .sha256 checksum files in addition
741 to the already existing support for validating using Ubuntu-style
744 * sd-bus gained support for a new sd_bus_message_appendv() call which
745 is va_list equivalent of sd_bus_message_append().
747 * sd-boot gained support for validating images using SHIM/MOK.
749 * The SMACK code learnt support for "onlycap".
751 * systemd-mount --umount is now much smarter in figuring out how to
752 properly unmount a device given its mount or device path.
754 * The code to call libnss_dns as a fallback from libnss_resolve when
755 the communication with systemd-resolved fails was removed. This
756 fallback was redundant and interfered with the [!UNAVAIL=return]
757 suffix. See nss-resolve(8) for the recommended configuration.
759 * systemd-logind may now be restarted without losing state. It stores
760 the file descriptors for devices it manages in the system manager
761 using the FDSTORE= mechanism. Please note that further changes in
762 other components may be required to make use of this (for example
763 Xorg has code to listen for stops of systemd-logind and terminate
764 itself when logind is stopped or restarted, in order to avoid using
765 stale file descriptors for graphical devices, which is now
766 counterproductive and must be reverted in order for restarts of
767 systemd-logind to be safe. See
768 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
770 * All kernel install plugins are called with the environment variable
771 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
772 /etc/machine-id. If the file is missing or empty, the variable is
773 empty and BOOT_DIR_ABS is the path of a temporary directory which is
774 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
775 is empty, all plugins should not put anything in BOOT_DIR_ABS.
777 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
778 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
779 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
780 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
781 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
782 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
783 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
784 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
785 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
786 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
787 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
788 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
789 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
790 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
791 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
792 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
793 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
794 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
795 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
796 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
797 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
798 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
799 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
800 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
801 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
802 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
803 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
804 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
805 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
812 * The "hybrid" control group mode has been modified to improve
813 compatibility with "legacy" cgroups-v1 setups. Specifically, the
814 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
815 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
816 cgroups-v1 hierarchy), the only externally visible change being that
817 the cgroups-v2 hierarchy is also mounted, to
818 /sys/fs/cgroup/unified. This should provide a large degree of
819 compatibility with "legacy" cgroups-v1, while taking benefit of the
820 better management capabilities of cgroups-v2.
822 * The default control group setup mode may be selected both a boot-time
823 via a set of kernel command line parameters (specifically:
824 systemd.unified_cgroup_hierarchy= and
825 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
826 default selected on the configure command line
827 (--with-default-hierarchy=). The upstream default is "hybrid"
828 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
829 this will change in a future systemd version to be "unified" (pure
830 cgroups-v2 mode). The third option for the compile time option is
831 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
832 distributions to default to "hybrid" mode for release distributions,
833 starting with v233. We recommend "unified" for development
834 distributions (specifically: distributions such as Fedora's rawhide)
835 as that's where things are headed in the long run. Use "legacy" for
836 greatest stability and compatibility only.
838 * Note one current limitation of "unified" and "hybrid" control group
839 setup modes: the kernel currently does not permit the systemd --user
840 instance (i.e. unprivileged code) to migrate processes between two
841 disconnected cgroup subtrees, even if both are managed and owned by
842 the user. This effectively means "systemd-run --user --scope" doesn't
843 work when invoked from outside of any "systemd --user" service or
844 scope. Specifically, it is not supported from session scopes. We are
845 working on fixing this in a future systemd version. (See #3388 for
846 further details about this.)
848 * DBus policy files are now installed into /usr rather than /etc. Make
849 sure your system has dbus >= 1.9.18 running before upgrading to this
850 version, or override the install path with --with-dbuspolicydir= .
852 * All python scripts shipped with systemd (specifically: the various
853 tests written in Python) now require Python 3.
855 * systemd unit tests can now run standalone (without the source or
856 build directories), and can be installed into /usr/lib/systemd/tests/
857 with 'make install-tests'.
859 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
860 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
863 * Support for the %c, %r, %R specifiers in unit files has been
864 removed. Specifiers are not supposed to be dependent on configuration
865 in the unit file itself (so that they resolve the same regardless
866 where used in the unit files), but these specifiers were influenced
867 by the Slice= option.
869 * The shell invoked by debug-shell.service now defaults to /bin/sh in
870 all cases. If distributions want to use a different shell for this
871 purpose (for example Fedora's /sbin/sushell) they need to specify
872 this explicitly at configure time using --with-debug-shell=.
874 * The confirmation spawn prompt has been reworked to offer the
877 (c)ontinue, proceed without asking anymore
878 (D)ump, show the state of the unit
879 (f)ail, don't execute the command and pretend it failed
881 (i)nfo, show a short summary of the unit
882 (j)obs, show jobs that are in progress
883 (s)kip, don't execute the command and pretend it succeeded
884 (y)es, execute the command
886 The 'n' choice for the confirmation spawn prompt has been removed,
887 because its meaning was confusing.
889 The prompt may now also be redirected to an alternative console by
890 specifying the console as parameter to systemd.confirm_spawn=.
892 * Services of Type=notify require a READY=1 notification to be sent
893 during startup. If no such message is sent, the service now fails,
894 even if the main process exited with a successful exit code.
896 * Services that fail to start up correctly now always have their
897 ExecStopPost= commands executed. Previously, they'd enter "failed"
898 state directly, without executing these commands.
900 * The option MulticastDNS= of network configuration files has acquired
901 an actual implementation. With MulticastDNS=yes a host can resolve
902 names of remote hosts and reply to mDNS A and AAAA requests.
904 * When units are about to be started an additional check is now done to
905 ensure that all dependencies of type BindsTo= (when used in
906 combination with After=) have been started.
908 * systemd-analyze gained a new verb "syscall-filter" which shows which
909 system call groups are defined for the SystemCallFilter= unit file
910 setting, and which system calls they contain.
912 * A new system call filter group "@filesystem" has been added,
913 consisting of various file system related system calls. Group
914 "@reboot" has been added, covering reboot, kexec and shutdown related
915 calls. Finally, group "@swap" has been added covering swap
916 configuration related calls.
918 * A new unit file option RestrictNamespaces= has been added that may be
919 used to restrict access to the various process namespace types the
920 Linux kernel provides. Specifically, it may be used to take away the
921 right for a service unit to create additional file system, network,
922 user, and other namespaces. This sandboxing option is particularly
923 relevant due to the high amount of recently discovered namespacing
924 related vulnerabilities in the kernel.
926 * systemd-udev's .link files gained support for a new AutoNegotiation=
927 setting for configuring Ethernet auto-negotiation.
929 * systemd-networkd's .network files gained support for a new
930 ListenPort= setting in the [DHCP] section to explicitly configure the
931 UDP client port the DHCP client shall listen on.
933 * .network files gained a new Unmanaged= boolean setting for explicitly
934 excluding one or more interfaces from management by systemd-networkd.
936 * The systemd-networkd ProxyARP= option has been renamed to
937 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
938 renamed to ReduceARPProxy=. The old names continue to be available
941 * systemd-networkd gained support for configuring IPv6 Proxy NDP
942 addresses via the new IPv6ProxyNDPAddress= .network file setting.
944 * systemd-networkd's bonding device support gained support for two new
945 configuration options ActiveSlave= and PrimarySlave=.
947 * The various options in the [Match] section of .network files gained
948 support for negative matching.
950 * New systemd-specific mount options are now understood in /etc/fstab:
952 x-systemd.mount-timeout= may be used to configure the maximum
953 permitted runtime of the mount command.
955 x-systemd.device-bound may be set to bind a mount point to its
956 backing device unit, in order to automatically remove a mount point
957 if its backing device is unplugged. This option may also be
958 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
959 on the block device, which is now automatically set for all CDROM
960 drives, so that mounted CDs are automatically unmounted when they are
961 removed from the drive.
963 x-systemd.after= and x-systemd.before= may be used to explicitly
964 order a mount after or before another unit or mount point.
966 * Enqueued start jobs for device units are now automatically garbage
967 collected if there are no jobs waiting for them anymore.
969 * systemctl list-jobs gained two new switches: with --after, for every
970 queued job the jobs it's waiting for are shown; with --before the
971 jobs which it's blocking are shown.
973 * systemd-nspawn gained support for ephemeral boots from disk images
974 (or in other words: --ephemeral and --image= may now be
975 combined). Moreover, ephemeral boots are now supported for normal
976 directories, even if the backing file system is not btrfs. Of course,
977 if the file system does not support file system snapshots or
978 reflinks, the initial copy operation will be relatively expensive, but
979 this should still be suitable for many use cases.
981 * Calendar time specifications in .timer units now support
982 specifications relative to the end of a month by using "~" instead of
983 "-" as separator between month and day. For example, "*-02~03" means
984 "the third last day in February". In addition a new syntax for
985 repeated events has been added using the "/" character. For example,
986 "9..17/2:00" means "every two hours from 9am to 5pm".
988 * systemd-socket-proxyd gained a new parameter --connections-max= for
989 configuring the maximum number of concurrent connections.
991 * sd-id128 gained a new API for generating unique IDs for the host in a
992 way that does not leak the machine ID. Specifically,
993 sd_id128_get_machine_app_specific() derives an ID based on the
994 machine ID a in well-defined, non-reversible, stable way. This is
995 useful whenever an identifier for the host is needed but where the
996 identifier shall not be useful to identify the system beyond the
997 scope of the application itself. (Internally this uses HMAC-SHA256 as
998 keyed hash function using the machine ID as input.)
1000 * NotifyAccess= gained a new supported value "exec". When set
1001 notifications are accepted from all processes systemd itself invoked,
1002 including all control processes.
1004 * .nspawn files gained support for defining overlay mounts using the
1005 Overlay= and OverlayReadOnly= options. Previously this functionality
1006 was only available on the systemd-nspawn command line.
1008 * systemd-nspawn's --bind= and --overlay= options gained support for
1009 bind/overlay mounts whose source lies within the container tree by
1010 prefixing the source path with "+".
1012 * systemd-nspawn's --bind= and --overlay= options gained support for
1013 automatically allocating a temporary source directory in /var/tmp
1014 that is removed when the container dies. Specifically, if the source
1015 directory is specified as empty string this mechanism is selected. An
1016 example usage is --overlay=+/var::/var, which creates an overlay
1017 mount based on the original /var contained in the image, overlayed
1018 with a temporary directory in the host's /var/tmp. This way changes
1019 to /var are automatically flushed when the container shuts down.
1021 * systemd-nspawn --image= option does now permit raw file system block
1022 devices (in addition to images containing partition tables, as
1025 * The disk image dissection logic in systemd-nspawn gained support for
1026 automatically setting up LUKS encrypted as well as Verity protected
1027 partitions. When a container is booted from an encrypted image the
1028 passphrase is queried at start-up time. When a container with Verity
1029 data is started, the root hash is search in a ".roothash" file
1030 accompanying the disk image (alternatively, pass the root hash via
1031 the new --root-hash= command line option).
1033 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1034 be used to dissect disk images the same way as systemd-nspawn does
1035 it, following the Bootable Partition Specification. It may even be
1036 used to mount disk images with complex partition setups (including
1037 LUKS and Verity partitions) to a local host directory, in order to
1038 inspect them. This tool is not considered public API (yet), and is
1039 thus not installed into /usr/bin. Please do not rely on its
1040 existence, since it might go away or be changed in later systemd
1043 * A new generator "systemd-verity-generator" has been added, similar in
1044 style to "systemd-cryptsetup-generator", permitting automatic setup of
1045 Verity root partitions when systemd boots up. In order to make use of
1046 this your partition setup should follow the Discoverable Partitions
1047 Specification, and the GPT partition ID of the root file system
1048 partition should be identical to the upper 128bit of the Verity root
1049 hash. The GPT partition ID of the Verity partition protecting it
1050 should be the lower 128bit of the Verity root hash. If the partition
1051 image follows this model it is sufficient to specify a single
1052 "roothash=" kernel command line argument to both configure which root
1053 image and verity partition to use as well as the root hash for
1054 it. Note that systemd-nspawn's Verity support follows the same
1055 semantics, meaning that disk images with proper Verity data in place
1056 may be booted in containers with systemd-nspawn as well as on
1057 physical systems via the verity generator. Also note that the "mkosi"
1058 tool available at https://github.com/systemd/mkosi has been updated
1059 to generate Verity protected disk images following this scheme. In
1060 fact, it has been updated to generate disk images that optionally
1061 implement a complete UEFI SecureBoot trust chain, involving a signed
1062 kernel and initrd image that incorporates such a root hash as well as
1063 a Verity-enabled root partition.
1065 * The hardware database (hwdb) udev supports has been updated to carry
1066 accelerometer quirks.
1068 * All system services are now run with a fresh kernel keyring set up
1069 for them. The invocation ID is stored by default in it, thus
1070 providing a safe, non-overridable way to determine the invocation
1073 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1074 options for bind mounting arbitrary paths in a service-specific
1075 way. When these options are used, arbitrary host or service files and
1076 directories may be mounted to arbitrary locations in the service's
1079 * Documentation has been added that lists all of systemd's low-level
1080 environment variables:
1082 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
1084 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1085 whether a specific socket file descriptor matches a specified socket
1088 * systemd-firstboot has been updated to check for the
1089 systemd.firstboot= kernel command line option. It accepts a boolean
1090 and when set to false the first boot questions are skipped.
1092 * systemd-fstab-generator has been updated to check for the
1093 systemd.volatile= kernel command line option, which either takes an
1094 optional boolean parameter or the special value "state". If used the
1095 system may be booted in a "volatile" boot mode. Specifically,
1096 "systemd.volatile" is used, the root directory will be mounted as
1097 tmpfs, and only /usr is mounted from the actual root file system. If
1098 "systemd.volatile=state" is used, the root directory will be mounted
1099 as usual, but /var is mounted as tmpfs. This concept provides similar
1100 functionality as systemd-nspawn's --volatile= option, but provides it
1101 on physical boots. Use this option for implementing stateless
1102 systems, or testing systems with all state and/or configuration reset
1103 to the defaults. (Note though that many distributions are not
1104 prepared to boot up without a populated /etc or /var, though.)
1106 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1107 partitions. Previously it only supported LUKS encrypted partitions
1108 for all other uses, except for the root partition itself.
1110 * Socket units gained support for listening on AF_VSOCK sockets for
1111 communication in virtualized QEMU environments.
1113 * The "configure" script gained a new option --with-fallback-hostname=
1114 for specifying the fallback hostname to use if none is configured in
1115 /etc/hostname. For example, by specifying
1116 --with-fallback-hostname=fedora it is possible to default to a
1117 hostname of "fedora" on pristine installations.
1119 * systemd-cgls gained support for a new --unit= switch for listing only
1120 the control groups of a specific unit. Similar --user-unit= has been
1121 added for listing only the control groups of a specific user unit.
1123 * systemd-mount gained a new --umount switch for unmounting a mount or
1124 automount point (and all mount/automount points below it).
1126 * systemd will now refuse full configuration reloads (via systemctl
1127 daemon-reload and related calls) unless at least 16MiB of free space
1128 are available in /run. This is a safety precaution in order to ensure
1129 that generators can safely operate after the reload completed.
1131 * A new unit file option RootImage= has been added, which has a similar
1132 effect as RootDirectory= but mounts the service's root directory from
1133 a disk image instead of plain directory. This logic reuses the same
1134 image dissection and mount logic that systemd-nspawn already uses,
1135 and hence supports any disk images systemd-nspawn supports, including
1136 those following the Discoverable Partition Specification, as well as
1137 Verity enabled images. This option enables systemd to run system
1138 services directly off disk images acting as resource bundles,
1139 possibly even including full integrity data.
1141 * A new MountAPIVFS= unit file option has been added, taking a boolean
1142 argument. If enabled /proc, /sys and /dev (collectively called the
1143 "API VFS") will be mounted for the service. This is only relevant if
1144 RootDirectory= or RootImage= is used for the service, as these mounts
1145 are of course in place in the host mount namespace anyway.
1147 * systemd-nspawn gained support for a new --pivot-root= switch. If
1148 specified the root directory within the container image is pivoted to
1149 the specified mount point, while the original root disk is moved to a
1150 different place. This option enables booting of ostree images
1151 directly with systemd-nspawn.
1153 * The systemd build scripts will no longer complain if the NTP server
1154 addresses are not changed from the defaults. Google now supports
1155 these NTP servers officially. We still recommend downstreams to
1156 properly register an NTP pool with the NTP pool project though.
1158 * coredumpctl gained a new "--reverse" option for printing the list
1159 of coredumps in reverse order.
1161 * coredumpctl will now show additional information about truncated and
1162 inaccessible coredumps, as well as coredumps that are still being
1163 processed. It also gained a new --quiet switch for suppressing
1164 additional informational message in its output.
1166 * coredumpctl gained support for only showing coredumps newer and/or
1167 older than specific timestamps, using the new --since= and --until=
1168 options, reminiscent of journalctl's options by the same name.
1170 * The systemd-coredump logic has been improved so that it may be reused
1171 to collect backtraces in non-compiled languages, for example in
1172 scripting languages such as Python.
1174 * machinectl will now show the UID shift of local containers, if user
1175 namespacing is enabled for them.
1177 * systemd will now optionally run "environment generator" binaries at
1178 configuration load time. They may be used to add environment
1179 variables to the environment block passed to services invoked. One
1180 user environment generator is shipped by default that sets up
1181 environment variables based on files dropped into /etc/environment.d
1182 and ~/.config/environment.d/.
1184 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1187 * hostnamed has been updated to report a new chassis type of
1188 "convertible" to cover "foldable" laptops that can both act as a
1189 tablet and as a laptop, such as various Lenovo Yoga devices.
1191 Contributions from: Adrián López, Alexander Galanin, Alexander
1192 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1193 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1194 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1195 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1196 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1197 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1198 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1199 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1200 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1201 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1202 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1203 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1204 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1205 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1206 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1207 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1208 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1209 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1210 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1211 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1212 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1213 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1214 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1215 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1216 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1217 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1218 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1221 — Berlin, 2017-03-01
1225 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1226 RestrictAddressFamilies= enabled. These sandboxing options should
1227 generally be compatible with the various external udev call-out
1228 binaries we are aware of, however there may be exceptions, in
1229 particular when exotic languages for these call-outs are used. In
1230 this case, consider turning off these settings locally.
1232 * The new RemoveIPC= option can be used to remove IPC objects owned by
1233 the user or group of a service when that service exits.
1235 * The new ProtectKernelModules= option can be used to disable explicit
1236 load and unload operations of kernel modules by a service. In
1237 addition access to /usr/lib/modules is removed if this option is set.
1239 * ProtectSystem= option gained a new value "strict", which causes the
1240 whole file system tree with the exception of /dev, /proc, and /sys,
1241 to be remounted read-only for a service.
1243 * The new ProtectKernelTunables= option can be used to disable
1244 modification of configuration files in /sys and /proc by a service.
1245 Various directories and files are remounted read-only, so access is
1246 restricted even if the file permissions would allow it.
1248 * The new ProtectControlGroups= option can be used to disable write
1249 access by a service to /sys/fs/cgroup.
1251 * Various systemd services have been hardened with
1252 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1253 RestrictAddressFamilies=.
1255 * Support for dynamically creating users for the lifetime of a service
1256 has been added. If DynamicUser=yes is specified, user and group IDs
1257 will be allocated from the range 61184..65519 for the lifetime of the
1258 service. They can be resolved using the new nss-systemd.so NSS
1259 module. The module must be enabled in /etc/nsswitch.conf. Services
1260 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1261 any resources allocated by the service will be cleaned up when the
1262 service exits. They also have ProtectHome=read-only and
1263 ProtectSystem=strict enabled, so they are not able to make any
1264 permanent modifications to the system.
1266 * The nss-systemd module also always resolves root and nobody, making
1267 it possible to have no /etc/passwd or /etc/group files in minimal
1268 container or chroot environments.
1270 * Services may be started with their own user namespace using the new
1271 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1272 under which the service is running are mapped. All other users are
1275 * Support for the cgroup namespace has been added to systemd-nspawn. If
1276 supported by kernel, the container system started by systemd-nspawn
1277 will have its own view of the cgroup hierarchy. This new behaviour
1278 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1280 * The new MemorySwapMax= option can be used to limit the maximum swap
1281 usage under the unified cgroup hierarchy.
1283 * Support for the CPU controller in the unified cgroup hierarchy has
1284 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1285 options. This controller requires out-of-tree patches for the kernel
1286 and the support is provisional.
1288 * Mount and automount units may now be created transiently
1289 (i.e. dynamically at runtime via the bus API, instead of requiring
1290 unit files in the file system).
1292 * systemd-mount is a new tool which may mount file systems – much like
1293 mount(8), optionally pulling in additional dependencies through
1294 transient .mount and .automount units. For example, this tool
1295 automatically runs fsck on a backing block device before mounting,
1296 and allows the automount logic to be used dynamically from the
1297 command line for establishing mount points. This tool is particularly
1298 useful when dealing with removable media, as it will ensure fsck is
1299 run – if necessary – before the first access and that the file system
1300 is quickly unmounted after each access by utilizing the automount
1301 logic. This maximizes the chance that the file system on the
1302 removable media stays in a clean state, and if it isn't in a clean
1303 state is fixed automatically.
1305 * LazyUnmount=yes option for mount units has been added to expose the
1306 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1309 * /efi will be used as the mount point of the EFI boot partition, if
1310 the directory is present, and the mount point was not configured
1311 through other means (e.g. fstab). If /efi directory does not exist,
1312 /boot will be used as before. This makes it easier to automatically
1313 mount the EFI partition on systems where /boot is used for something
1316 * When operating on GPT disk images for containers, systemd-nspawn will
1317 now mount the ESP to /boot or /efi according to the same rules as PID
1318 1 running on a host. This allows tools like "bootctl" to operate
1319 correctly within such containers, in order to make container images
1320 bootable on physical systems.
1322 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1324 * Two new user session targets have been added to support running
1325 graphical sessions under the systemd --user instance:
1326 graphical-session.target and graphical-session-pre.target. See
1327 systemd.special(7) for a description of how those targets should be
1330 * The vconsole initialization code has been significantly reworked to
1331 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1332 support unicode keymaps. Font and keymap configuration will now be
1333 copied to all allocated virtual consoles.
1335 * FreeBSD's bhyve virtualization is now detected.
1337 * Information recorded in the journal for core dumps now includes the
1338 contents of /proc/mountinfo and the command line of the process at
1339 the top of the process hierarchy (which is usually the init process
1342 * systemd-journal-gatewayd learned the --directory= option to serve
1343 files from the specified location.
1345 * journalctl --root=… can be used to peruse the journal in the
1346 /var/log/ directories inside of a container tree. This is similar to
1347 the existing --machine= option, but does not require the container to
1350 * The hardware database has been extended to support
1351 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1354 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1355 specify the click rate for mice which include a horizontal wheel with
1356 a click rate that is different than the one for the vertical wheel.
1358 * systemd-run gained a new --wait option that makes service execution
1359 synchronous. (Specifically, the command will not return until the
1360 specified service binary exited.)
1362 * systemctl gained a new --wait option that causes the start command to
1363 wait until the units being started have terminated again.
1365 * A new journal output mode "short-full" has been added which displays
1366 timestamps with abbreviated English day names and adds a timezone
1367 suffix. Those timestamps include more information than the default
1368 "short" output mode, and can be passed directly to journalctl's
1369 --since= and --until= options.
1371 * /etc/resolv.conf will be bind-mounted into containers started by
1372 systemd-nspawn, if possible, so any changes to resolv.conf contents
1373 are automatically propagated to the container.
1375 * The number of instances for socket-activated services originating
1376 from a single IP address can be limited with
1377 MaxConnectionsPerSource=, extending the existing setting of
1380 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1383 * .netdev and .network configuration can now be extended through
1386 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1387 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1388 can be enabled and disabled using the new UDPSegmentationOffload=,
1389 TCPSegmentationOffload=, GenericSegmentationOffload=,
1390 GenericReceiveOffload=, LargeReceiveOffload= options in the
1391 [Link] section of .link files.
1393 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1394 Port VLAN ID can be configured for bridge devices using the new STP=,
1395 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1396 section of .netdev files.
1398 * The route table to which routes received over DHCP or RA should be
1399 added can be configured with the new RouteTable= option in the [DHCP]
1400 and [IPv6AcceptRA] sections of .network files.
1402 * The Address Resolution Protocol can be disabled on links managed by
1403 systemd-networkd using the ARP=no setting in the [Link] section of
1406 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1407 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1408 encode information about the result and exit codes of the current
1409 service runtime cycle.
1411 * systemd-sysctl will now configure kernel parameters in the order
1412 they occur in the configuration files. This matches what sysctl
1413 has been traditionally doing.
1415 * kernel-install "plugins" that are executed to perform various
1416 tasks after a new kernel is added and before an old one is removed
1417 can now return a special value to terminate the procedure and
1418 prevent any later plugins from running.
1420 * Journald's SplitMode=login setting has been deprecated. It has been
1421 removed from documentation, and its use is discouraged. In a future
1422 release it will be completely removed, and made equivalent to current
1423 default of SplitMode=uid.
1425 * Storage=both option setting in /etc/systemd/coredump.conf has been
1426 removed. With fast LZ4 compression storing the core dump twice is not
1429 * The --share-system systemd-nspawn option has been replaced with an
1430 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1431 this functionality is discouraged. In addition the variables
1432 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1433 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1434 individual namespaces.
1436 * "machinectl list" now shows the IP address of running containers in
1437 the output, as well as OS release information.
1439 * "loginctl list" now shows the TTY of each session in the output.
1441 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1442 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1443 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1444 tracking objects in a "recursive" mode, where a single client can be
1445 counted multiple times, if it takes multiple references.
1447 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1448 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1449 process using sd-bus automatically exit if the bus connection is
1452 * Bus clients of the service manager may now "pin" loaded units into
1453 memory, by taking an explicit reference on them. This is useful to
1454 ensure the client can retrieve runtime data about the service even
1455 after the service completed execution. Taking such a reference is
1456 available only for privileged clients and should be helpful to watch
1457 running services in a race-free manner, and in particular collect
1458 information about exit statuses and results.
1460 * The nss-resolve module has been changed to strictly return UNAVAIL
1461 when communication via D-Bus with resolved failed, and NOTFOUND when
1462 a lookup completed but was negative. This means it is now possible to
1463 neatly configure fallbacks using nsswitch.conf result checking
1464 expressions. Taking benefit of this, the new recommended
1465 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1467 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1469 * A new setting CtrlAltDelBurstAction= has been added to
1470 /etc/systemd/system.conf which may be used to configure the precise
1471 behaviour if the user on the console presses Ctrl-Alt-Del more often
1472 than 7 times in 2s. Previously this would unconditionally result in
1473 an expedited, immediate reboot. With this new setting the precise
1474 operation may be configured in more detail, and also turned off
1477 * In .netdev files two new settings RemoteChecksumTx= and
1478 RemoteChecksumRx= are now understood that permit configuring the
1479 remote checksumming logic for VXLAN networks.
1481 * The service manager learnt a new "invocation ID" concept for invoked
1482 services. Each runtime cycle of a service will get a new invocation
1483 ID (a 128bit random UUID) assigned that identifies the current
1484 run of the service uniquely and globally. A new invocation ID
1485 is generated each time a service starts up. The journal will store
1486 the invocation ID of a service along with any logged messages, thus
1487 making the invocation ID useful for matching the online runtime of a
1488 service with the offline log data it generated in a safe way without
1489 relying on synchronized timestamps. In many ways this new service
1490 invocation ID concept is similar to the kernel's boot ID concept that
1491 uniquely and globally identifies the runtime of each boot. The
1492 invocation ID of a service is passed to the service itself via an
1493 environment variable ($INVOCATION_ID). A new bus call
1494 GetUnitByInvocationID() has been added that is similar to GetUnit()
1495 but instead of retrieving the bus path for a unit by its name
1496 retrieves it by its invocation ID. The returned path is valid only as
1497 long as the passed invocation ID is current.
1499 * systemd-resolved gained a new "DNSStubListener" setting in
1500 resolved.conf. It either takes a boolean value or the special values
1501 "udp" and "tcp", and configures whether to enable the stub DNS
1502 listener on 127.0.0.53:53.
1504 * IP addresses configured via networkd may now carry additional
1505 configuration settings supported by the kernel. New options include:
1506 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1507 PrefixRoute=, AutoJoin=.
1509 * The PAM configuration fragment file for "user@.service" shipped with
1510 systemd (i.e. the --user instance of systemd) has been stripped to
1511 the minimum necessary to make the system boot. Previously, it
1512 contained Fedora-specific stanzas that did not apply to other
1513 distributions. It is expected that downstream distributions add
1514 additional configuration lines, matching their needs to this file,
1515 using it only as rough template of what systemd itself needs. Note
1516 that this reduced fragment does not even include an invocation of
1517 pam_limits which most distributions probably want to add, even though
1518 systemd itself does not need it. (There's also the new build time
1519 option --with-pamconfdir=no to disable installation of the PAM
1522 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1523 capability is now also dropped from its set (in addition to
1524 CAP_SYS_MKNOD as before).
1526 * In service unit files it is now possible to connect a specific named
1527 file descriptor with stdin/stdout/stdout of an executed service. The
1528 name may be specified in matching .socket units using the
1529 FileDescriptorName= setting.
1531 * A number of journal settings may now be configured on the kernel
1532 command line. Specifically, the following options are now understood:
1533 systemd.journald.max_level_console=,
1534 systemd.journald.max_level_store=,
1535 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1536 systemd.journald.max_level_wall=.
1538 * "systemctl is-enabled --full" will now show by which symlinks a unit
1539 file is enabled in the unit dependency tree.
1541 * Support for VeraCrypt encrypted partitions has been added to the
1542 "cryptsetup" logic and /etc/crypttab.
1544 * systemd-detect-virt gained support for a new --private-users switch
1545 that checks whether the invoking processes are running inside a user
1546 namespace. Similar, a new special value "private-users" for the
1547 existing ConditionVirtualization= setting has been added, permitting
1548 skipping of specific units in user namespace environments.
1550 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1551 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1552 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1553 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1554 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1555 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1556 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1557 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1558 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1559 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1560 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1561 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1562 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1563 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1564 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1565 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1566 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1567 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1568 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1569 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1570 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1571 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1572 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1573 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1574 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1575 Jędrzejewski-Szmek, Zeal Jagannatha
1577 — Santa Fe, 2016-11-03
1581 * In service units the various ExecXYZ= settings have been extended
1582 with an additional special character as first argument of the
1583 assigned value: if the character '+' is used the specified command
1584 line it will be run with full privileges, regardless of User=,
1585 Group=, CapabilityBoundingSet= and similar options. The effect is
1586 similar to the existing PermissionsStartOnly= option, but allows
1587 configuration of this concept for each executed command line
1590 * Services may now alter the service watchdog timeout at runtime by
1591 sending a WATCHDOG_USEC= message via sd_notify().
1593 * MemoryLimit= and related unit settings now optionally take percentage
1594 specifications. The percentage is taken relative to the amount of
1595 physical memory in the system (or in case of containers, the assigned
1596 amount of memory). This allows scaling service resources neatly with
1597 the amount of RAM available on the system. Similarly, systemd-logind's
1598 RuntimeDirectorySize= option now also optionally takes percentage
1601 * In similar fashion TasksMax= takes percentage values now, too. The
1602 value is taken relative to the configured maximum number of processes
1603 on the system. The per-service task maximum has been changed to 15%
1604 using this functionality. (Effectively this is an increase of 512 →
1605 4915 for service units, given the kernel's default pid_max setting.)
1607 * Calendar time specifications in .timer units now understand a ".."
1608 syntax for time ranges. Example: "4..7:10" may now be used for
1609 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1612 * The InaccessableDirectories=, ReadOnlyDirectories= and
1613 ReadWriteDirectories= unit file settings have been renamed to
1614 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1615 applied to all kinds of file nodes, and not just directories, with
1616 the exception of symlinks. Specifically these settings may now be
1617 used on block and character device nodes, UNIX sockets and FIFOS as
1618 well as regular files. The old names of these settings remain
1619 available for compatibility.
1621 * systemd will now log about all service processes it kills forcibly
1622 (using SIGKILL) because they remained after the clean shutdown phase
1623 of the service completed. This should help identifying services that
1624 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1625 systemd-logind's configuration a similar log message is generated for
1626 processes killed at the end of each session due to this setting.
1628 * systemd will now set the $JOURNAL_STREAM environment variable for all
1629 services whose stdout/stderr are connected to the Journal (which
1630 effectively means by default: all services). The variable contains
1631 the device and inode number of the file descriptor used for
1632 stdout/stderr. This may be used by invoked programs to detect whether
1633 their stdout/stderr is connected to the Journal, in which case they
1634 can switch over to direct Journal communication, thus being able to
1635 pass extended, structured metadata along with their log messages. As
1636 one example, this is now used by glib's logging primitives.
1638 * When using systemd's default tmp.mount unit for /tmp, the mount point
1639 will now be established with the "nosuid" and "nodev" options. This
1640 avoids privilege escalation attacks that put traps and exploits into
1641 /tmp. However, this might cause problems if you e. g. put container
1642 images or overlays into /tmp; if you need this, override tmp.mount's
1643 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1646 * systemd now supports the "memory" cgroup controller also on
1649 * The systemd-cgtop tool now optionally takes a control group path as
1650 command line argument. If specified, the control group list shown is
1651 limited to subgroups of that group.
1653 * The SystemCallFilter= unit file setting gained support for
1654 pre-defined, named system call filter sets. For example
1655 SystemCallFilter=@clock is now an effective way to make all clock
1656 changing-related system calls unavailable to a service. A number of
1657 similar pre-defined groups are defined. Writing system call filters
1658 for system services is simplified substantially with this new
1659 concept. Accordingly, all of systemd's own, long-running services now
1660 enable system call filtering based on this, by default.
1662 * A new service setting MemoryDenyWriteExecute= has been added, taking
1663 a boolean value. If turned on, a service may no longer create memory
1664 mappings that are writable and executable at the same time. This
1665 enhances security for services where this is enabled as it becomes
1666 harder to dynamically write and then execute memory in exploited
1667 service processes. This option has been enabled for all of systemd's
1668 own long-running services.
1670 * A new RestrictRealtime= service setting has been added, taking a
1671 boolean argument. If set the service's processes may no longer
1672 acquire realtime scheduling. This improves security as realtime
1673 scheduling may otherwise be used to easily freeze the system.
1675 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1676 value. This may be used for requesting that the system manager inside
1677 of the container reports start-up completion to nspawn which then
1678 propagates this notification further to the service manager
1679 supervising nspawn itself. A related option NotifyReady= in .nspawn
1680 files has been added too. This functionality allows ordering of the
1681 start-up of multiple containers using the usual systemd ordering
1684 * machinectl gained a new command "stop" that is an alias for
1687 * systemd-resolved gained support for contacting DNS servers on
1688 link-local IPv6 addresses.
1690 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1691 its caches. A method call for requesting the same operation has been
1692 added to the bus API too, and is made available via "systemd-resolve
1695 * systemd-resolve gained a new --status switch. If passed a brief
1696 summary of the used DNS configuration with per-interface information
1699 * resolved.conf gained a new Cache= boolean option, defaulting to
1700 on. If turned off local DNS caching is disabled. This comes with a
1701 performance penalty in particular when DNSSEC is enabled. Note that
1702 resolved disables its internal caching implicitly anyway, when the
1703 configured DNS server is on a host-local IP address such as ::1 or
1704 127.0.0.1, thus automatically avoiding double local caching.
1706 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1707 for DNS requests. This improves compatibility with local programs
1708 that do not use the libc NSS or systemd-resolved's bus APIs for name
1709 resolution. This minimal DNS service is only available to local
1710 programs and does not implement the full DNS protocol, but enough to
1711 cover local DNS clients. A new, static resolv.conf file, listing just
1712 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1713 now recommended to make /etc/resolv.conf a symlink to this file in
1714 order to route all DNS lookups to systemd-resolved, regardless if
1715 done via NSS, the bus API or raw DNS packets. Note that this local
1716 DNS service is not as fully featured as the libc NSS or
1717 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1718 used to deliver link-local address information (as this implies
1719 sending a local interface index along), LLMNR/mDNS support via this
1720 interface is severely restricted. It is thus strongly recommended for
1721 all applications to use the libc NSS API or native systemd-resolved
1724 * systemd-networkd's bridge support learned a new setting
1725 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1726 in .network files has been added for configuring VLAN bridging in
1727 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1729 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1730 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1731 now be acquired without relying on DHCPv6. Two new options
1732 UseDomains= and UseDNS= have been added to configure this behaviour.
1734 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1735 renamed IPv6AcceptRA=, without altering its behaviour. The old
1736 setting name remains available for compatibility reasons.
1738 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1739 Key=, InputKey= and OutputKey=.
1741 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1742 interface configuration.
1744 * "systemctl edit" may now be used to create new unit files by
1745 specifying the --force switch.
1747 * sd-event gained a new function sd_event_get_iteration() for
1748 requesting the current iteration counter of the event loop. It starts
1749 at zero and is increased by one with each event loop iteration.
1751 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1752 file. It can be used in lieu of %systemd_requires in packages which
1753 don't use any systemd functionality and are intended to be installed
1754 in minimal containers without systemd present. This macro provides
1755 ordering dependencies to ensure that if the package is installed in
1756 the same rpm transaction as systemd, systemd will be installed before
1757 the scriptlets for the package are executed, allowing unit presets
1760 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1761 been added to simplify packaging of generators.
1763 * The os-release file gained VERSION_CODENAME field for the
1764 distribution nickname (e.g. VERSION_CODENAME=woody).
1766 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1767 can be set to disable parsing of metadata and the creation
1768 of persistent symlinks for that device.
1770 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1771 to make them available to logged-in users has been reverted.
1773 * Much of the common code of the various systemd components is now
1774 built into an internal shared library libsystemd-shared-231.so
1775 (incorporating the systemd version number in the name, to be updated
1776 with future releases) that the components link to. This should
1777 decrease systemd footprint both in memory during runtime and on
1778 disk. Note that the shared library is not for public use, and is
1779 neither API not ABI stable, but is likely to change with every new
1780 released update. Packagers need to make sure that binaries
1781 linking to libsystemd-shared.so are updated in step with the
1784 * Configuration for "mkosi" is now part of the systemd
1785 repository. mkosi is a tool to easily build legacy-free OS images,
1786 and is available on github: https://github.com/systemd/mkosi. If
1787 "mkosi" is invoked in the build tree a new raw OS image is generated
1788 incorporating the systemd sources currently being worked on and a
1789 clean, fresh distribution installation. The generated OS image may be
1790 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1791 UEFI PC. This functionality is particularly useful to easily test
1792 local changes made to systemd in a pristine, defined environment. See
1793 HACKING for details.
1795 * configure learned the --with-support-url= option to specify the
1796 distribution's bugtracker.
1798 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1799 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1800 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1801 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1802 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1803 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1804 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1805 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1806 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1807 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1808 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1809 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1810 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1811 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1812 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1813 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1814 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1815 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1816 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1818 — Berlin, 2016-07-25
1822 * DNSSEC is now turned on by default in systemd-resolved (in
1823 "allow-downgrade" mode), but may be turned off during compile time by
1824 passing "--with-default-dnssec=no" to "configure" (and of course,
1825 during runtime with DNSSEC= in resolved.conf). We recommend
1826 downstreams to leave this on at least during development cycles and
1827 report any issues with the DNSSEC logic upstream. We are very
1828 interested in collecting feedback about the DNSSEC validator and its
1829 limitations in the wild. Note however, that DNSSEC support is
1830 probably nothing downstreams should turn on in stable distros just
1831 yet, as it might create incompatibilities with a few DNS servers and
1832 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1833 automatically whenever we detect such incompatible setups, but there
1834 might be systems we do not cover yet. Hence: please help us testing
1835 the DNSSEC code, leave this on where you can, report back, but then
1836 again don't consider turning this on in your stable, LTS or
1837 production release just yet. (Note that you have to enable
1838 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1839 and its DNSSEC mode for host name resolution from local
1842 * systemd-resolve conveniently resolves DANE records with the --tlsa
1843 option and OPENPGPKEY records with the --openpgp option. It also
1844 supports dumping raw DNS record data via the new --raw= switch.
1846 * systemd-logind will now by default terminate user processes that are
1847 part of the user session scope unit (session-XX.scope) when the user
1848 logs out. This behavior is controlled by the KillUserProcesses=
1849 setting in logind.conf, and the previous default of "no" is now
1850 changed to "yes". This means that user sessions will be properly
1851 cleaned up after, but additional steps are necessary to allow
1852 intentionally long-running processes to survive logout.
1854 While the user is logged in at least once, user@.service is running,
1855 and any service that should survive the end of any individual login
1856 session can be started at a user service or scope using systemd-run.
1857 systemd-run(1) man page has been extended with an example which shows
1858 how to run screen in a scope unit underneath user@.service. The same
1859 command works for tmux.
1861 After the user logs out of all sessions, user@.service will be
1862 terminated too, by default, unless the user has "lingering" enabled.
1863 To effectively allow users to run long-term tasks even if they are
1864 logged out, lingering must be enabled for them. See loginctl(1) for
1865 details. The default polkit policy was modified to allow users to
1866 set lingering for themselves without authentication.
1868 Previous defaults can be restored at compile time by the
1869 --without-kill-user-processes option to "configure".
1871 * systemd-logind gained new configuration settings SessionsMax= and
1872 InhibitorsMax=, both with a default of 8192. It will not register new
1873 user sessions or inhibitors above this limit.
1875 * systemd-logind will now reload configuration on SIGHUP.
1877 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1878 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1879 enable. Also, support for the "io" cgroup controller in the unified
1880 hierarchy has been added, so that the "memory", "pids" and "io" are
1881 now the controllers that are supported on the unified hierarchy.
1883 WARNING: it is not possible to use previous systemd versions with
1884 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1885 is necessary to also update systemd in the initramfs if using the
1886 unified hierarchy. An updated SELinux policy is also required.
1888 * LLDP support has been extended, and both passive (receive-only) and
1889 active (sender) modes are supported. Passive mode ("routers-only") is
1890 enabled by default in systemd-networkd. Active LLDP mode is enabled
1891 by default for containers on the internal network. The "networkctl
1892 lldp" command may be used to list information gathered. "networkctl
1893 status" will also show basic LLDP information on connected peers now.
1895 * The IAID and DUID unique identifier sent in DHCP requests may now be
1896 configured for the system and each .network file managed by
1897 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1899 * systemd-networkd gained support for configuring proxy ARP support for
1900 each interface, via the ProxyArp= setting in .network files. It also
1901 gained support for configuring the multicast querier feature of
1902 bridge devices, via the new MulticastQuerier= setting in .netdev
1903 files. Similarly, snooping on the IGMP traffic can be controlled
1904 via the new setting MulticastSnooping=.
1906 A new setting PreferredLifetime= has been added for addresses
1907 configured in .network file to configure the lifetime intended for an
1910 The systemd-networkd DHCP server gained the option EmitRouter=, which
1911 defaults to yes, to configure whether the DHCP Option 3 (Router)
1914 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1915 systemd-socket-activate and installed into /usr/bin. It is now fully
1918 * systemd-journald now uses separate threads to flush changes to disk
1919 when closing journal files, thus reducing impact of slow disk I/O on
1920 logging performance.
1922 * The sd-journal API gained two new calls
1923 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1924 can be used to open journal files using file descriptors instead of
1925 file or directory paths. sd_journal_open_container() has been
1926 deprecated, sd_journal_open_directory_fd() should be used instead
1927 with the flag SD_JOURNAL_OS_ROOT.
1929 * journalctl learned a new output mode "-o short-unix" that outputs log
1930 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1931 UTC). It also gained support for a new --no-hostname setting to
1932 suppress the hostname column in the family of "short" output modes.
1934 * systemd-ask-password now optionally skips printing of the password to
1935 stdout with --no-output which can be useful in scripts.
1937 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1938 (devices tagged with ID_MAKER_TOOL) are now tagged with
1939 "uaccess" and are available to logged in users.
1941 * The DeviceAllow= unit setting now supports specifiers (with "%").
1943 * "systemctl show" gained a new --value switch, which allows print a
1944 only the contents of a specific unit property, without also printing
1945 the property's name. Similar support was added to "show*" verbs
1946 of loginctl and machinectl that output "key=value" lists.
1948 * A new unit type "generated" was added for files dynamically generated
1949 by generator tools. Similarly, a new unit type "transient" is used
1950 for unit files created using the runtime API. "systemctl enable" will
1951 refuse to operate on such files.
1953 * A new command "systemctl revert" has been added that may be used to
1954 revert to the vendor version of a unit file, in case local changes
1955 have been made by adding drop-ins or overriding the unit file.
1957 * "machinectl clean" gained a new verb to automatically remove all or
1958 just hidden container images.
1960 * systemd-tmpfiles gained support for a new line type "e" for emptying
1961 directories, if they exist, without creating them if they don't.
1963 * systemd-nspawn gained support for automatically patching the UID/GIDs
1964 of the owners and the ACLs of all files and directories in a
1965 container tree to match the UID/GID user namespacing range selected
1966 for the container invocation. This mode is enabled via the new
1967 --private-users-chown switch. It also gained support for
1968 automatically choosing a free, previously unused UID/GID range when
1969 starting a container, via the new --private-users=pick setting (which
1970 implies --private-users-chown). Together, these options for the first
1971 time make user namespacing for nspawn containers fully automatic and
1972 thus deployable. The systemd-nspawn@.service template unit file has
1973 been changed to use this functionality by default.
1975 * systemd-nspawn gained a new --network-zone= switch, that allows
1976 creating ad-hoc virtual Ethernet links between multiple containers,
1977 that only exist as long as at least one container referencing them is
1978 running. This allows easy connecting of multiple containers with a
1979 common link that implements an Ethernet broadcast domain. Each of
1980 these network "zones" may be named relatively freely by the user, and
1981 may be referenced by any number of containers, but each container may
1982 only reference one of these "zones". On the lower level, this is
1983 implemented by an automatically managed bridge network interface for
1984 each zone, that is created when the first container referencing its
1985 zone is created and removed when the last one referencing its zone
1988 * The default start timeout may now be configured on the kernel command
1989 line via systemd.default_timeout_start_sec=. It was already
1990 configurable via the DefaultTimeoutStartSec= option in
1991 /etc/systemd/system.conf.
1993 * Socket units gained a new TriggerLimitIntervalSec= and
1994 TriggerLimitBurst= setting to configure a limit on the activation
1995 rate of the socket unit.
1997 * The LimitNICE= setting now optionally takes normal UNIX nice values
1998 in addition to the raw integer limit value. If the specified
1999 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2000 value is understood as UNIX nice value. If not prefixed like this it
2001 is understood as raw RLIMIT_NICE limit.
2003 * Note that the effect of the PrivateDevices= unit file setting changed
2004 slightly with this release: the per-device /dev file system will be
2005 mounted read-only from this version on, and will have "noexec"
2006 set. This (minor) change of behavior might cause some (exceptional)
2007 legacy software to break, when PrivateDevices=yes is set for its
2008 service. Please leave PrivateDevices= off if you run into problems
2011 * systemd-bootchart has been split out to a separate repository:
2012 https://github.com/systemd/systemd-bootchart
2014 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2015 merged into the kernel in its current form.
2017 * The compatibility libraries libsystemd-daemon.so,
2018 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2019 which have been deprecated since systemd-209 have been removed along
2020 with the corresponding pkg-config files. All symbols provided by
2021 those libraries are provided by libsystemd.so.
2023 * The Capabilities= unit file setting has been removed (it is ignored
2024 for backwards compatibility). AmbientCapabilities= and
2025 CapabilityBoundingSet= should be used instead.
2027 * A new special target has been added, initrd-root-device.target,
2028 which creates a synchronization point for dependencies of the root
2029 device in early userspace. Initramfs builders must ensure that this
2030 target is now included in early userspace.
2032 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2033 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2034 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2035 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2036 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2037 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2038 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2039 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2040 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2041 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2042 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2043 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2044 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2045 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2046 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2047 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2048 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2049 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2050 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2051 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2052 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2053 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2054 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2055 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2056 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2059 — Fairfax, 2016-05-21
2063 * The systemd-resolved DNS resolver service has gained a substantial
2064 set of new features, most prominently it may now act as a DNSSEC
2065 validating stub resolver. DNSSEC mode is currently turned off by
2066 default, but is expected to be turned on by default in one of the
2067 next releases. For now, we invite everybody to test the DNSSEC logic
2068 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2069 service also gained a full set of D-Bus interfaces, including calls
2070 to configure DNS and DNSSEC settings per link (for use by external
2071 network management software). systemd-resolved and systemd-networkd
2072 now distinguish between "search" and "routing" domains. The former
2073 are used to qualify single-label names, the latter are used purely
2074 for routing lookups within certain domains to specific links.
2075 resolved now also synthesizes RRs for all entries from /etc/hosts.
2077 * The systemd-resolve tool (which is a client utility for
2078 systemd-resolved) has been improved considerably and is now fully
2079 supported and documented. Hence it has moved from /usr/lib/systemd to
2082 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2085 * The coredump collection logic has been reworked: when a coredump is
2086 collected it is now written to disk, compressed and processed
2087 (including stacktrace extraction) from a new instantiated service
2088 systemd-coredump@.service, instead of directly from the
2089 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2090 processing large coredumps can take up a substantial amount of
2091 resources and time, and this previously happened entirely outside of
2092 systemd's service supervision. With the new logic the core_pattern
2093 hook only does minimal metadata collection before passing off control
2094 to the new instantiated service, which is configured with a time
2095 limit, a nice level and other settings to minimize negative impact on
2096 the rest of the system. Also note that the new logic will honour the
2097 RLIMIT_CORE setting of the crashed process, which now allows users
2098 and processes to turn off coredumping for their processes by setting
2101 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2102 and all forked processes by default. Previously, PID 1 would leave
2103 the setting at "0" for all processes, as set by the kernel. Note that
2104 the resource limit traditionally has no effect on the generated
2105 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2106 logic is used. Since the limit is now honoured (see above) its
2107 default has been changed so that the coredumping logic is enabled by
2108 default for all processes, while allowing specific opt-out.
2110 * When the stacktrace is extracted from processes of system users, this
2111 is now done as "systemd-coredump" user, in order to sandbox this
2112 potentially security sensitive parsing operation. (Note that when
2113 processing coredumps of normal users this is done under the user ID
2114 of process that crashed, as before.) Packagers should take notice
2115 that it is now necessary to create the "systemd-coredump" system user
2116 and group at package installation time.
2118 * The systemd-activate socket activation testing tool gained support
2119 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2120 and --seqpacket switches. It also has been extended to support both
2121 new-style and inetd-style file descriptor passing. Use the new
2122 --inetd switch to request inetd-style file descriptor passing.
2124 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2125 variable, which takes a boolean value. If set to false, ANSI color
2126 output is disabled in the tools even when run on a terminal that
2129 * The VXLAN support in networkd now supports two new settings
2130 DestinationPort= and PortRange=.
2132 * A new systemd.machine_id= kernel command line switch has been added,
2133 that may be used to set the machine ID in /etc/machine-id if it is
2134 not initialized yet. This command line option has no effect if the
2135 file is already initialized.
2137 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2138 specified command line as PID 2 rather than PID 1 in the
2139 container. In this mode PID 1 is a minimal stub init process that
2140 implements the special POSIX and Linux semantics of PID 1 regarding
2141 signal and child process management. Note that this stub init process
2142 is implemented in nspawn itself and requires no support from the
2143 container image. This new logic is useful to support running
2144 arbitrary commands in the container, as normal processes are
2145 generally not prepared to run as PID 1.
2147 * systemd-nspawn gained a new --chdir= switch for setting the current
2148 working directory for the process started in the container.
2150 * "journalctl /dev/sda" will now output all kernel log messages for
2151 specified device from the current boot, in addition to all devices
2152 that are parents of it. This should make log output about devices
2153 pretty useful, as long as kernel drivers attach enough metadata to
2154 the log messages. (The usual SATA drivers do.)
2156 * The sd-journal API gained two new calls
2157 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2158 that report whether log data from /run or /var has been found.
2160 * journalctl gained a new switch "--fields" that prints all journal
2161 record field names currently in use in the journal. This is backed
2162 by two new sd-journal API calls sd_journal_enumerate_fields() and
2163 sd_journal_restart_fields().
2165 * Most configurable timeouts in systemd now expect an argument of
2166 "infinity" to turn them off, instead of "0" as before. The semantics
2167 from now on is that a timeout of "0" means "now", and "infinity"
2168 means "never". To maintain backwards compatibility, "0" continues to
2169 turn off previously existing timeout settings.
2171 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2172 try-reload-or-restart" to clarify what it actually does: the "try"
2173 logic applies to both reloading and restarting, not just restarting.
2174 The old name continues to be accepted for compatibility.
2176 * On boot-up, when PID 1 detects that the system clock is behind the
2177 release date of the systemd version in use, the clock is now set
2178 to the latter. Previously, this was already done in timesyncd, in order
2179 to avoid running with clocks set to the various clock epochs such as
2180 1902, 1938 or 1970. With this change the logic is now done in PID 1
2181 in addition to timesyncd during early boot-up, so that it is enforced
2182 before the first process is spawned by systemd. Note that the logic
2183 in timesyncd remains, as it is more comprehensive and ensures
2184 clock monotonicity by maintaining a persistent timestamp file in
2185 /var. Since /var is generally not available in earliest boot or the
2186 initrd, this part of the logic remains in timesyncd, and is not done
2189 * Support for tweaking details in net_cls.class_id through the
2190 NetClass= configuration directive has been removed, as the kernel
2191 people have decided to deprecate that controller in cgroup v2.
2192 Userspace tools such as nftables are moving over to setting rules
2193 that are specific to the full cgroup path of a task, which obsoletes
2194 these controllers anyway. The NetClass= directive is kept around for
2195 legacy compatibility reasons. For a more in-depth description of the
2196 kernel change, please refer to the respective upstream commit:
2198 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2200 * A new service setting RuntimeMaxSec= has been added that may be used
2201 to specify a maximum runtime for a service. If the timeout is hit, the
2202 service is terminated and put into a failure state.
2204 * A new service setting AmbientCapabilities= has been added. It allows
2205 configuration of additional Linux process capabilities that are
2206 passed to the activated processes. This is only available on very
2209 * The process resource limit settings in service units may now be used
2210 to configure hard and soft limits individually.
2212 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2213 expose support for gcc's __attribute__((cleanup())) C extension.
2214 Specifically, for many object destructor functions alternative
2215 versions have been added that have names suffixed with "p" and take a
2216 pointer to a pointer to the object to destroy, instead of just a
2217 pointer to the object itself. This is useful because these destructor
2218 functions may be used directly as parameters to the cleanup
2219 construct. Internally, systemd has been a heavy user of this GCC
2220 extension for a long time, and with this change similar support is
2221 now available to consumers of the library outside of systemd. Note
2222 that by using this extension in your sources compatibility with old
2223 and strictly ANSI compatible C compilers is lost. However, all gcc or
2224 LLVM versions of recent years support this extension.
2226 * Timer units gained support for a new setting RandomizedDelaySec= that
2227 allows configuring some additional randomized delay to the configured
2228 time. This is useful to spread out timer events to avoid load peaks in
2229 clusters or larger setups.
2231 * Calendar time specifications now support sub-second accuracy.
2233 * Socket units now support listening on SCTP and UDP-lite protocol
2236 * The sd-event API now comes with a full set of man pages.
2238 * Older versions of systemd contained experimental support for
2239 compressing journal files and coredumps with the LZ4 compressor that
2240 was not compatible with the lz4 binary (due to API limitations of the
2241 lz4 library). This support has been removed; only support for files
2242 compatible with the lz4 binary remains. This LZ4 logic is now
2243 officially supported and no longer considered experimental.
2245 * The dkr image import logic has been removed again from importd. dkr's
2246 micro-services focus doesn't fit into the machine image focus of
2247 importd, and quickly got out of date with the upstream dkr API.
2249 * Creation of the /run/lock/lockdev/ directory was dropped from
2250 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2251 been available for many years. If you still need this, you need to
2252 create your own tmpfiles.d config file with:
2254 d /run/lock/lockdev 0775 root lock -
2256 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2257 and RebootArgument= have been moved from the [Service] section of
2258 unit files to [Unit], and they are now supported on all unit types,
2259 not just service units. Of course, elogind will continue to
2260 understand these settings also at the old location, in order to
2261 maintain compatibility.
2263 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2264 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2265 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2266 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2267 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2268 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2269 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2270 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2271 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2272 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2273 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2274 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2275 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2276 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2277 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2278 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2279 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2280 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2281 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2283 — Berlin, 2016-02-11
2287 * A number of properties previously only settable in unit
2288 files are now also available as properties to set when
2289 creating transient units programmatically via the bus, as it
2290 is exposed with systemd-run's --property=
2291 setting. Specifically, these are: SyslogIdentifier=,
2292 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2293 EnvironmentFile=, ReadWriteDirectories=,
2294 ReadOnlyDirectories=, InaccessibleDirectories=,
2295 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2297 * When creating transient services via the bus API it is now
2298 possible to pass in a set of file descriptors to use as
2299 STDIN/STDOUT/STDERR for the invoked process.
2301 * Slice units may now be created transiently via the bus APIs,
2302 similar to the way service and scope units may already be
2303 created transiently.
2305 * Wherever systemd expects a calendar timestamp specification
2306 (like in journalctl's --since= and --until= switches) UTC
2307 timestamps are now supported. Timestamps suffixed with "UTC"
2308 are now considered to be in Universal Time Coordinated
2309 instead of the local timezone. Also, timestamps may now
2310 optionally be specified with sub-second accuracy. Both of
2311 these additions also apply to recurring calendar event
2312 specification, such as OnCalendar= in timer units.
2314 * journalctl gained a new "--sync" switch that asks the
2315 journal daemon to write all so far unwritten log messages to
2316 disk and sync the files, before returning.
2318 * systemd-tmpfiles learned two new line types "q" and "Q" that
2319 operate like "v", but also set up a basic btrfs quota
2320 hierarchy when used on a btrfs file system with quota
2323 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2324 instead of a subvolume (even on a btrfs file system) if the
2325 root directory is a plain directory, and not a
2326 subvolume. This should simplify things with certain chroot()
2327 environments which are not aware of the concept of btrfs
2330 * systemd-detect-virt gained a new --chroot switch to detect
2331 whether execution takes place in a chroot() environment.
2333 * CPUAffinity= now takes CPU index ranges in addition to
2336 * The various memory-related resource limit settings (such as
2337 LimitAS=) now understand the usual K, M, G, ... suffixes to
2338 the base of 1024 (IEC). Similar, the time-related resource
2339 limit settings understand the usual min, h, day, ...
2342 * There's a new system.conf setting DefaultTasksMax= to
2343 control the default TasksMax= setting for services and
2344 scopes running on the system. (TasksMax= is the primary
2345 setting that exposes the "pids" cgroup controller on systemd
2346 and was introduced in the previous systemd release.) The
2347 setting now defaults to 512, which means services that are
2348 not explicitly configured otherwise will only be able to
2349 create 512 processes or threads at maximum, from this
2350 version on. Note that this means that thread- or
2351 process-heavy services might need to be reconfigured to set
2352 TasksMax= to a higher value. It is sufficient to set
2353 TasksMax= in these specific unit files to a higher value, or
2354 even "infinity". Similar, there's now a logind.conf setting
2355 UserTasksMax= that defaults to 4096 and limits the total
2356 number of processes or tasks each user may own
2357 concurrently. nspawn containers also have the TasksMax=
2358 value set by default now, to 8192. Note that all of this
2359 only has an effect if the "pids" cgroup controller is
2360 enabled in the kernel. The general benefit of these changes
2361 should be a more robust and safer system, that provides a
2362 certain amount of per-service fork() bomb protection.
2364 * systemd-nspawn gained the new --network-veth-extra= switch
2365 to define additional and arbitrarily-named virtual Ethernet
2366 links between the host and the container.
2368 * A new service execution setting PassEnvironment= has been
2369 added that allows importing select environment variables
2370 from PID1's environment block into the environment block of
2373 * Timer units gained support for a new RemainAfterElapse=
2374 setting which takes a boolean argument. It defaults to on,
2375 exposing behaviour unchanged to previous releases. If set to
2376 off, timer units are unloaded after they elapsed if they
2377 cannot elapse again. This is particularly useful for
2378 transient timer units, which shall not stay around longer
2379 than until they first elapse.
2381 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2382 default now (the kernel default is 16). This is beneficial
2383 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2384 allows substantially larger numbers of queued
2385 datagrams. This should increase the capability of systemd to
2386 parallelize boot-up, as logging and sd_notify() are unlikely
2387 to stall execution anymore. If you need to change the value
2388 from the new defaults, use the usual sysctl.d/ snippets.
2390 * The compression framing format used by the journal or
2391 coredump processing has changed to be in line with what the
2392 official LZ4 tools generate. LZ4 compression support in
2393 systemd was considered unsupported previously, as the format
2394 was not compatible with the normal tools. With this release
2395 this has changed now, and it is hence safe for downstream
2396 distributions to turn it on. While not compressing as well
2397 as the XZ, LZ4 is substantially faster, which makes
2398 it a good default choice for the compression logic in the
2399 journal and in coredump handling.
2401 * Any reference to /etc/mtab has been dropped from
2402 systemd. The file has been obsolete since a while, but
2403 systemd refused to work on systems where it was incorrectly
2404 set up (it should be a symlink or non-existent). Please make
2405 sure to update to util-linux 2.27.1 or newer in conjunction
2406 with this systemd release, which also drops any reference to
2407 /etc/mtab. If you maintain a distribution make sure that no
2408 software you package still references it, as this is a
2409 likely source of bugs. There's also a glibc bug pending,
2410 asking for removal of any reference to this obsolete file:
2412 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2414 Note that only util-linux versions built with
2415 --enable-libmount-force-mountinfo are supported.
2417 * Support for the ".snapshot" unit type has been removed. This
2418 feature turned out to be little useful and little used, and
2419 has now been removed from the core and from systemctl.
2421 * The dependency types RequiresOverridable= and
2422 RequisiteOverridable= have been removed from systemd. They
2423 have been used only very sparingly to our knowledge and
2424 other options that provide a similar effect (such as
2425 systemctl --mode=ignore-dependencies) are much more useful
2426 and commonly used. Moreover, they were only half-way
2427 implemented as the option to control behaviour regarding
2428 these dependencies was never added to systemctl. By removing
2429 these dependency types the execution engine becomes a bit
2430 simpler. Unit files that use these dependencies should be
2431 changed to use the non-Overridable dependency types
2432 instead. In fact, when parsing unit files with these
2433 options, that's what systemd will automatically convert them
2434 too, but it will also warn, asking users to fix the unit
2435 files accordingly. Removal of these dependency types should
2436 only affect a negligible number of unit files in the wild.
2438 * Behaviour of networkd's IPForward= option changed
2439 (again). It will no longer maintain a per-interface setting,
2440 but propagate one way from interfaces where this is enabled
2441 to the global kernel setting. The global setting will be
2442 enabled when requested by a network that is set up, but
2443 never be disabled again. This change was made to make sure
2444 IPv4 and IPv6 behaviour regarding packet forwarding is
2445 similar (as the Linux IPv6 stack does not support
2446 per-interface control of this setting) and to minimize
2449 * In unit files the behaviour of %u, %U, %h, %s has
2450 changed. These specifiers will now unconditionally resolve
2451 to the various user database fields of the user that the
2452 systemd instance is running as, instead of the user
2453 configured in the specific unit via User=. Note that this
2454 effectively doesn't change much, as resolving of these
2455 specifiers was already turned off in the --system instance
2456 of systemd, as we cannot do NSS lookups from PID 1. In the
2457 --user instance of systemd these specifiers where correctly
2458 resolved, but hardly made any sense, since the user instance
2459 lacks privileges to do user switches anyway, and User= is
2460 hence useless. Moreover, even in the --user instance of
2461 systemd behaviour was awkward as it would only take settings
2462 from User= assignment placed before the specifier into
2463 account. In order to unify and simplify the logic around
2464 this the specifiers will now always resolve to the
2465 credentials of the user invoking the manager (which in case
2466 of PID 1 is the root user).
2468 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2469 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2470 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2471 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2472 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2473 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2474 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2475 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2476 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2477 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2478 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2479 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2480 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2481 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2484 — Berlin, 2015-11-18
2488 * systemd now depends on util-linux v2.27. More specifically,
2489 the newly added mount monitor feature in libmount now
2490 replaces systemd's former own implementation.
2492 * libmount mandates /etc/mtab not to be regular file, and
2493 systemd now enforces this condition at early boot.
2494 /etc/mtab has been deprecated and warned about for a very
2495 long time, so systems running systemd should already have
2496 stopped having this file around as anything else than a
2497 symlink to /proc/self/mounts.
2499 * Support for the "pids" cgroup controller has been added. It
2500 allows accounting the number of tasks in a cgroup and
2501 enforcing limits on it. This adds two new setting
2502 TasksAccounting= and TasksMax= to each unit, as well as a
2503 global option DefaultTasksAccounting=.
2505 * Support for the "net_cls" cgroup controller has been added.
2506 It allows assigning a net class ID to each task in the
2507 cgroup, which can then be used in firewall rules and traffic
2508 shaping configurations. Note that the kernel netfilter net
2509 class code does not currently work reliably for ingress
2510 packets on unestablished sockets.
2512 This adds a new config directive called NetClass= to CGroup
2513 enabled units. Allowed values are positive numbers for fixed
2514 assignments and "auto" for picking a free value
2517 * 'systemctl is-system-running' now returns 'offline' if the
2518 system is not booted with systemd. This command can now be
2519 used as a substitute for 'systemd-notify --booted'.
2521 * Watchdog timeouts have been increased to 3 minutes for all
2522 in-tree service files. Apparently, disk IO issues are more
2523 frequent than we hoped, and user reported >1 minute waiting
2526 * 'machine-id-commit' functionality has been merged into
2527 'machine-id-setup --commit'. The separate binary has been
2530 * The WorkingDirectory= directive in unit files may now be set
2531 to the special value '~'. In this case, the working
2532 directory is set to the home directory of the user
2533 configured in User=.
2535 * "machinectl shell" will now open the shell in the home
2536 directory of the selected user by default.
2538 * The CrashChVT= configuration file setting is renamed to
2539 CrashChangeVT=, following our usual logic of not
2540 abbreviating unnecessarily. The old directive is still
2541 supported for compat reasons. Also, this directive now takes
2542 an integer value between 1 and 63, or a boolean value. The
2543 formerly supported '-1' value for disabling stays around for
2546 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2547 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2548 RootDirectory= properties can now be set for transient
2551 * The systemd-analyze tool gained a new "set-log-target" verb
2552 to change the logging target the system manager logs to
2553 dynamically during runtime. This is similar to how
2554 "systemd-analyze set-log-level" already changes the log
2557 * In nspawn /sys is now mounted as tmpfs, with only a selected
2558 set of subdirectories mounted in from the real sysfs. This
2559 enhances security slightly, and is useful for ensuring user
2560 namespaces work correctly.
2562 * Support for USB FunctionFS activation has been added. This
2563 allows implementation of USB gadget services that are
2564 activated as soon as they are requested, so that they don't
2565 have to run continuously, similar to classic socket
2568 * The "systemctl exit" command now optionally takes an
2569 additional parameter that sets the exit code to return from
2570 the systemd manager when exiting. This is only relevant when
2571 running the systemd user instance, or when running the
2572 system instance in a container.
2574 * sd-bus gained the new API calls sd_bus_path_encode_many()
2575 and sd_bus_path_decode_many() that allow easy encoding and
2576 decoding of multiple identifier strings inside a D-Bus
2577 object path. Another new call sd_bus_default_flush_close()
2578 has been added to flush and close per-thread default
2581 * systemd-cgtop gained support for a -M/--machine= switch to
2582 show the control groups within a certain container only.
2584 * "systemctl kill" gained support for an optional --fail
2585 switch. If specified the requested operation will fail of no
2586 processes have been killed, because the unit had no
2587 processes attached, or similar.
2589 * A new systemd.crash_reboot=1 kernel command line option has
2590 been added that triggers a reboot after crashing. This can
2591 also be set through CrashReboot= in systemd.conf.
2593 * The RuntimeDirectory= setting now understands unit
2594 specifiers like %i or %f.
2596 * A new (still internal) library API sd-ipv4acd has been added,
2597 that implements address conflict detection for IPv4. It's
2598 based on code from sd-ipv4ll, and will be useful for
2599 detecting DHCP address conflicts.
2601 * File descriptors passed during socket activation may now be
2602 named. A new API sd_listen_fds_with_names() is added to
2603 access the names. The default names may be overridden,
2604 either in the .socket file using the FileDescriptorName=
2605 parameter, or by passing FDNAME= when storing the file
2606 descriptors using sd_notify().
2608 * systemd-networkd gained support for:
2610 - Setting the IPv6 Router Advertisement settings via
2611 IPv6AcceptRouterAdvertisements= in .network files.
2613 - Configuring the HelloTimeSec=, MaxAgeSec= and
2614 ForwardDelaySec= bridge parameters in .netdev files.
2616 - Configuring PreferredSource= for static routes in
2619 * The "ask-password" framework used to query for LUKS harddisk
2620 passwords or SSL passwords during boot gained support for
2621 caching passwords in the kernel keyring, if it is
2622 available. This makes sure that the user only has to type in
2623 a passphrase once if there are multiple objects to unlock
2624 with the same one. Previously, such password caching was
2625 available only when Plymouth was used; this moves the
2626 caching logic into the systemd codebase itself. The
2627 "systemd-ask-password" utility gained a new --keyname=
2628 switch to control which kernel keyring key to use for
2629 caching a password in. This functionality is also useful for
2630 enabling display managers such as gdm to automatically
2631 unlock the user's GNOME keyring if its passphrase, the
2632 user's password and the harddisk password are the same, if
2633 gdm-autologin is used.
2635 * When downloading tar or raw images using "machinectl
2636 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2637 file is now also downloaded, if it is available and stored
2638 next to the image file.
2640 * Units of type ".socket" gained a new boolean setting
2641 Writable= which is only useful in conjunction with
2642 ListenSpecial=. If true, enables opening the specified
2643 special file in O_RDWR mode rather than O_RDONLY mode.
2645 * systemd-rfkill has been reworked to become a singleton
2646 service that is activated through /dev/rfkill on each rfkill
2647 state change and saves the settings to disk. This way,
2648 systemd-rfkill is now compatible with devices that exist
2649 only intermittendly, and even restores state if the previous
2650 system shutdown was abrupt rather than clean.
2652 * The journal daemon gained support for vacuuming old journal
2653 files controlled by the number of files that shall remain,
2654 in addition to the already existing control by size and by
2655 date. This is useful as journal interleaving performance
2656 degrades with too many separate journal files, and allows
2657 putting an effective limit on them. The new setting defaults
2658 to 100, but this may be changed by setting SystemMaxFiles=
2659 and RuntimeMaxFiles= in journald.conf. Also, the
2660 "journalctl" tool gained the new --vacuum-files= switch to
2661 manually vacuum journal files to leave only the specified
2662 number of files in place.
2664 * udev will now create /dev/disk/by-path links for ATA devices
2665 on kernels where that is supported.
2667 * Galician, Serbian, Turkish and Korean translations were added.
2669 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2670 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2671 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2672 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2673 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2674 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2675 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2676 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2677 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2678 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2679 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2680 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2681 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2682 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2683 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2684 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2685 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2686 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2688 — Berlin, 2015-10-07
2692 * The DHCP implementation of systemd-networkd gained a set of
2695 - The DHCP server now supports emitting DNS and NTP
2696 information. It may be enabled and configured via
2697 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2698 and NTP information is enabled, but no servers are
2699 configured, the corresponding uplink information (if there
2700 is any) is propagated.
2702 - Server and client now support transmission and reception
2703 of timezone information. It can be configured via the
2704 newly introduced network options UseTimezone=,
2705 EmitTimezone=, and Timezone=. Transmission of timezone
2706 information is enabled between host and containers by
2707 default now: the container will change its local timezone
2708 to what the host has set.
2710 - Lease timeouts can now be configured via
2711 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2713 - The DHCP server improved on the stability of
2714 leases. Clients are more likely to get the same lease
2715 information back, even if the server loses state.
2717 - The DHCP server supports two new configuration options to
2718 control the lease address pool metrics, PoolOffset= and
2721 * The encapsulation limit of tunnels in systemd-networkd may
2722 now be configured via 'EncapsulationLimit='. It allows
2723 modifying the maximum additional levels of encapsulation
2724 that are permitted to be prepended to a packet.
2726 * systemd now supports the concept of user buses replacing
2727 session buses, if used with dbus-1.10 (and enabled via dbus
2728 --enable-user-session). It previously only supported this on
2729 kdbus-enabled systems, and this release expands this to
2730 'dbus-daemon' systems.
2732 * systemd-networkd now supports predictable interface names
2735 * systemd now optionally supports the new Linux kernel
2736 "unified" control group hierarchy. If enabled via the kernel
2737 command-line option 'systemd.unified_cgroup_hierarchy=1',
2738 systemd will try to mount the unified cgroup hierarchy
2739 directly on /sys/fs/cgroup. If not enabled, or not
2740 available, systemd will fall back to the legacy cgroup
2741 hierarchy setup, as before. Host system and containers can
2742 mix and match legacy and unified hierarchies as they
2743 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2744 environment variable to individually select the hierarchy to
2745 use for executed containers. By default, nspawn will use the
2746 unified hierarchy for the containers if the host uses the
2747 unified hierarchy, and the legacy hierarchy otherwise.
2748 Please note that at this point the unified hierarchy is an
2749 experimental kernel feature and is likely to change in one
2750 of the next kernel releases. Therefore, it should not be
2751 enabled by default in downstream distributions yet. The
2752 minimum required kernel version for the unified hierarchy to
2753 work is 4.2. Note that when the unified hierarchy is used
2754 for the first time delegated access to controllers is
2755 safe. Because of this systemd-nspawn containers will get
2756 access to controllers now, as will systemd user
2757 sessions. This means containers and user sessions may now
2758 manage their own resources, partitioning up what the system
2761 * A new special scope unit "init.scope" has been introduced
2762 that encapsulates PID 1 of the system. It may be used to
2763 determine resource usage and enforce resource limits on PID
2764 1 itself. PID 1 hence moved out of the root of the control
2767 * The cgtop tool gained support for filtering out kernel
2768 threads when counting tasks in a control group. Also, the
2769 count of processes is now recursively summed up by
2770 default. Two options -k and --recursive= have been added to
2771 revert to old behaviour. The tool has also been updated to
2772 work correctly in containers now.
2774 * systemd-nspawn's --bind= and --bind-ro= options have been
2775 extended to allow creation of non-recursive bind mounts.
2777 * libsystemd gained two new calls sd_pid_get_cgroup() and
2778 sd_peer_get_cgroup() which return the control group path of
2779 a process or peer of a connected AF_UNIX socket. This
2780 function call is particularly useful when implementing
2781 delegated subtrees support in the control group hierarchy.
2783 * The "sd-event" event loop API of libsystemd now supports
2784 correct dequeuing of real-time signals, without losing
2787 * When systemd requests a PolicyKit decision when managing
2788 units it will now add additional fields to the request,
2789 including unit name and desired operation. This enables more
2790 powerful PolicyKit policies, that make decisions depending
2791 on these parameters.
2793 * nspawn learnt support for .nspawn settings files, that may
2794 accompany the image files or directories of containers, and
2795 may contain additional settings for the container. This is
2796 an alternative to configuring container parameters via the
2797 nspawn command line.
2799 Contributions from: Cristian Rodríguez, Daniel Mack, David
2800 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2801 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2802 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2803 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2804 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2805 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2806 Andersen, Tom Gundersen, Torstein Husebø
2808 — Berlin, 2015-09-08
2812 * machinectl gained a new verb 'shell' which opens a fresh
2813 shell on the target container or the host. It is similar to
2814 the existing 'login' command of machinectl, but spawns the
2815 shell directly without prompting for username or
2816 password. The pseudo machine '.host' now refers to the local
2817 host and is used by default. Hence, 'machinectl shell' can
2818 be used as replacement for 'su -' which spawns a session as
2819 a fresh systemd unit in a way that is fully isolated from
2820 the originating session.
2822 * systemd-networkd learned to cope with private-zone DHCP
2823 options and allows other programs to query the values.
2825 * SELinux access control when enabling/disabling units is no
2826 longer enforced with this release. The previous
2827 implementation was incorrect, and a new corrected
2828 implementation is not yet available. As unit file operations
2829 are still protected via PolicyKit and D-Bus policy this is
2830 not a security problem. Yet, distributions which care about
2831 optimal SELinux support should probably not stabilize on
2834 * sd-bus gained support for matches of type "arg0has=", that
2835 test for membership of strings in string arrays sent in bus
2838 * systemd-resolved now dumps the contents of its DNS and LLMNR
2839 caches to the logs on reception of the SIGUSR1 signal. This
2840 is useful to debug DNS behaviour.
2842 * The coredumpctl tool gained a new --directory= option to
2843 operate on journal files in a specific directory.
2845 * "systemctl reboot" and related commands gained a new
2846 "--message=" option which may be used to set a free-text
2847 wall message when shutting down or rebooting the
2848 system. This message is also logged, which is useful for
2849 figuring out the reason for a reboot or shutdown a
2852 * The "systemd-resolve-host" tool's -i switch now takes
2853 network interface numbers as alternative to interface names.
2855 * A new unit file setting for services has been introduced:
2856 UtmpMode= allows configuration of how precisely systemd
2857 handles utmp and wtmp entries for the service if this is
2858 enabled. This allows writing services that appear similar to
2859 user sessions in the output of the "w", "who", "last" and
2862 * systemd-resolved will now locally synthesize DNS resource
2863 records for the "localhost" and "gateway" domains as well as
2864 the local hostname. This should ensure that clients querying
2865 RRs via resolved will get similar results as those going via
2866 NSS, if nss-myhostname is enabled.
2868 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2869 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2870 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2871 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2872 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2873 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2874 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2875 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2876 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2877 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2878 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2879 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2881 — Berlin, 2015-08-27
2885 * The systemd-efi-boot-generator functionality was merged into
2886 systemd-gpt-auto-generator.
2888 * systemd-networkd now supports Group Policy for vxlan
2889 devices. It can be enabled via the new boolean configuration
2890 option called 'GroupPolicyExtension='.
2892 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2893 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2894 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2896 — Berlin, 2015-07-31
2900 * The python-systemd code has been removed from the systemd repository.
2901 A new repository has been created which accommodates the code from
2902 now on, and we kindly ask distributions to create a separate package
2903 for this: https://github.com/systemd/python-systemd
2905 * The systemd daemon will now reload its main configuration
2906 (/etc/systemd/system.conf) on daemon-reload.
2908 * sd-dhcp now exposes vendor specific extensions via
2909 sd_dhcp_lease_get_vendor_specific().
2911 * systemd-networkd gained a number of new configuration options.
2913 - A new boolean configuration option for TAP devices called
2914 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2915 device, thus allowing to send and receive GSO packets.
2917 - A new tunnel configuration option called 'CopyDSCP='.
2918 If enabled, the DSCP field of ip6 tunnels is copied into the
2919 decapsulated packet.
2921 - A set of boolean bridge configuration options were added.
2922 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2923 and 'UnicastFlood=' are now parsed by networkd and applied to the
2924 respective bridge link device via the respective IFLA_BRPORT_*
2927 - A new string configuration option to override the hostname sent
2928 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2929 is true, networkd will use the configured hostname instead of the
2930 system hostname when sending DHCP requests.
2932 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2933 networkd will configure the IPv6 flow-label of the tunnel device
2934 according to RFC2460.
2936 - The 'macvtap' virtual network devices are now supported, similar to
2937 the already supported 'macvlan' devices.
2939 * systemd-resolved now implements RFC5452 to improve resilience against
2940 cache poisoning. Additionally, source port randomization is enabled
2941 by default to further protect against DNS spoofing attacks.
2943 * nss-mymachines now supports translating UIDs and GIDs of running
2944 containers with user-namespaces enabled. If a container 'foo'
2945 translates a host uid 'UID' to the container uid 'TUID', then
2946 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2947 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2948 mapped as 'vg-foo-TGID'.
2950 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2951 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2952 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2953 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2954 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2955 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2956 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2957 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2958 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2959 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2961 — Berlin, 2015-07-29
2965 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2966 There are no known issues with current sysfs, and udev does not need
2967 or should be used to work around such bugs.
2969 * udev does no longer enable USB HID power management. Several reports
2970 indicate, that some devices cannot handle that setting.
2972 * The udev accelerometer helper was removed. The functionality
2973 is now fully included in iio-sensor-proxy. But this means,
2974 older iio-sensor-proxy versions will no longer provide
2975 accelerometer/orientation data with this systemd version.
2976 Please upgrade iio-sensor-proxy to version 1.0.
2978 * networkd gained a new configuration option IPv6PrivacyExtensions=
2979 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2980 for Stateless Address") on selected networks.
2982 * For the sake of fewer build-time dependencies and less code in the
2983 main repository, the python bindings are about to be removed in the
2984 next release. A new repository has been created which accommodates
2985 the code from now on, and we kindly ask distributions to create a
2986 separate package for this. The removal will take place in v223.
2988 https://github.com/systemd/python-systemd
2990 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2991 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2992 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2993 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2994 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2995 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2996 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2997 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2998 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2999 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3001 — Berlin, 2015-07-07
3005 * The sd-bus.h and sd-event.h APIs have now been declared
3006 stable and have been added to the official interface of
3007 libsystemd.so. sd-bus implements an alternative D-Bus client
3008 library, that is relatively easy to use, very efficient and
3009 supports both classic D-Bus as well as kdbus as transport
3010 backend. sd-event is a generic event loop abstraction that
3011 is built around Linux epoll, but adds features such as event
3012 prioritization or efficient timer handling. Both APIs are good
3013 choices for C programs looking for a bus and/or event loop
3014 implementation that is minimal and does not have to be
3015 portable to other kernels.
3017 * kdbus support is no longer compile-time optional. It is now
3018 always built-in. However, it can still be disabled at
3019 runtime using the kdbus=0 kernel command line setting, and
3020 that setting may be changed to default to off, by specifying
3021 --disable-kdbus at build-time. Note though that the kernel
3022 command line setting has no effect if the kdbus.ko kernel
3023 module is not installed, in which case kdbus is (obviously)
3024 also disabled. We encourage all downstream distributions to
3025 begin testing kdbus by adding it to the kernel images in the
3026 development distributions, and leaving kdbus support in
3029 * The minimal required util-linux version has been bumped to
3032 * Support for chkconfig (--enable-chkconfig) was removed in
3033 favor of calling an abstraction tool
3034 /lib/systemd/systemd-sysv-install. This needs to be
3035 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3036 in README for details.
3038 * If there's a systemd unit and a SysV init script for the
3039 same service name, and the user executes "systemctl enable"
3040 for it (or a related call), then this will now enable both
3041 (or execute the related operation on both), not just the
3044 * The libudev API documentation has been converted from gtkdoc
3047 * gudev has been removed from the systemd tree, it is now an
3050 * The systemd-cgtop tool learnt a new --raw switch to generate
3051 "raw" (machine parsable) output.
3053 * networkd's IPForwarding= .network file setting learnt the
3054 new setting "kernel", which ensures that networkd does not
3055 change the IP forwarding sysctl from the default kernel
3058 * The systemd-logind bus API now exposes a new boolean
3059 property "Docked" that reports whether logind considers the
3060 system "docked", i.e. connected to a docking station or not.
3062 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3063 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3064 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3065 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3066 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3067 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3068 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3069 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3070 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3071 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3072 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3073 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3074 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3075 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3076 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3077 Fink, Zbigniew Jędrzejewski-Szmek
3079 — Berlin, 2015-06-19
3083 * The gudev library has been extracted into a separate repository
3084 available at: https://git.gnome.org/browse/libgudev/
3085 It is now managed as part of the Gnome project. Distributions
3086 are recommended to pass --disable-gudev to systemd and use
3087 gudev from the Gnome project instead. gudev is still included
3088 in systemd, for now. It will be removed soon, though. Please
3089 also see the announcement-thread on systemd-devel:
3090 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3092 * systemd now exposes a CPUUsageNSec= property for each
3093 service unit on the bus, that contains the overall consumed
3094 CPU time of a service (the sum of what each process of the
3095 service consumed). This value is only available if
3096 CPUAccounting= is turned on for a service, and is then shown
3097 in the "systemctl status" output.
3099 * Support for configuring alternative mappings of the old SysV
3100 runlevels to systemd targets has been removed. They are now
3101 hardcoded in a way that runlevels 2, 3, 4 all map to
3102 multi-user.target and 5 to graphical.target (which
3103 previously was already the default behaviour).
3105 * The auto-mounter logic gained support for mount point
3106 expiry, using a new TimeoutIdleSec= setting in .automount
3107 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3109 * The EFI System Partition (ESP) as mounted to /boot by
3110 systemd-efi-boot-generator will now be unmounted
3111 automatically after 2 minutes of not being used. This should
3112 minimize the risk of ESP corruptions.
3114 * New /etc/fstab options x-systemd.requires= and
3115 x-systemd.requires-mounts-for= are now supported to express
3116 additional dependencies for mounts. This is useful for
3117 journalling file systems that support external journal
3118 devices or overlay file systems that require underlying file
3119 systems to be mounted.
3121 * systemd does not support direct live-upgrades (via systemctl
3122 daemon-reexec) from versions older than v44 anymore. As no
3123 distribution we are aware of shipped such old versions in a
3124 stable release this should not be problematic.
3126 * When systemd forks off a new per-connection service instance
3127 it will now set the $REMOTE_ADDR environment variable to the
3128 remote IP address, and $REMOTE_PORT environment variable to
3129 the remote IP port. This behaviour is similar to the
3130 corresponding environment variables defined by CGI.
3132 * systemd-networkd gained support for uplink failure
3133 detection. The BindCarrier= option allows binding interface
3134 configuration dynamically to the link sense of other
3135 interfaces. This is useful to achieve behaviour like in
3138 * systemd-networkd gained support for configuring the DHCP
3139 client identifier to use when requesting leases.
3141 * systemd-networkd now has a per-network UseNTP= option to
3142 configure whether NTP server information acquired via DHCP
3143 is passed on to services like systemd-timesyncd.
3145 * systemd-networkd gained support for vti6 tunnels.
3147 * Note that systemd-networkd manages the sysctl variable
3148 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3149 it is configured for since v219. The variable controls IP
3150 forwarding, and is a per-interface alternative to the global
3151 /proc/sys/net/ipv[46]/ip_forward. This setting is
3152 configurable in the IPForward= option, which defaults to
3153 "no". This means if networkd is used for an interface it is
3154 no longer sufficient to set the global sysctl option to turn
3155 on IP forwarding! Instead, the .network file option
3156 IPForward= needs to be turned on! Note that the
3157 implementation of this behaviour was broken in v219 and has
3160 * Many bonding and vxlan options are now configurable in
3163 * systemd-nspawn gained a new --property= setting to set unit
3164 properties for the container scope. This is useful for
3165 setting resource parameters (e.g. "CPUShares=500") on
3166 containers started from the command line.
3168 * systemd-nspawn gained a new --private-users= switch to make
3169 use of user namespacing available on recent Linux kernels.
3171 * systemd-nspawn may now be called as part of a shell pipeline
3172 in which case the pipes used for stdin and stdout are passed
3173 directly to the process invoked in the container, without
3174 indirection via a pseudo tty.
3176 * systemd-nspawn gained a new switch to control the UNIX
3177 signal to use when killing the init process of the container
3180 * systemd-nspawn gained a new --overlay= switch for mounting
3181 overlay file systems into the container using the new kernel
3184 * When a container image is imported via systemd-importd and
3185 the host file system is not btrfs, a loopback block device
3186 file is created in /var/lib/machines.raw with a btrfs file
3187 system inside. It is then mounted to /var/lib/machines to
3188 enable btrfs features for container management. The loopback
3189 file and btrfs file system is grown as needed when container
3190 images are imported via systemd-importd.
3192 * systemd-machined/systemd-importd gained support for btrfs
3193 quota, to enforce container disk space limits on disk. This
3194 is exposed in "machinectl set-limit".
3196 * systemd-importd now can import containers from local .tar,
3197 .raw and .qcow2 images, and export them to .tar and .raw. It
3198 can also import dkr v2 images now from the network (on top
3201 * systemd-importd gained support for verifying downloaded
3202 images with gpg2 (previously only gpg1 was supported).
3204 * systemd-machined, systemd-logind, systemd: most bus calls
3205 are now accessible to unprivileged processes via
3206 PolicyKit. Also, systemd-logind will now allow users to kill
3207 their own sessions without further privileges or
3210 * systemd-shutdownd has been removed. This service was
3211 previously responsible for implementing scheduled shutdowns
3212 as exposed in /usr/bin/shutdown's time parameter. This
3213 functionality has now been moved into systemd-logind and is
3214 accessible via a bus interface.
3216 * "systemctl reboot" gained a new switch --firmware-setup that
3217 can be used to reboot into the EFI firmware setup, if that
3218 is available. systemd-logind now exposes an API on the bus
3219 to trigger such reboots, in case graphical desktop UIs want
3220 to cover this functionality.
3222 * "systemctl enable", "systemctl disable" and "systemctl mask"
3223 now support a new "--now" switch. If specified the units
3224 that are enabled will also be started, and the ones
3225 disabled/masked also stopped.
3227 * The Gummiboot EFI boot loader tool has been merged into
3228 systemd, and renamed to "systemd-boot". The bootctl tool has been
3229 updated to support systemd-boot.
3231 * An EFI kernel stub has been added that may be used to create
3232 kernel EFI binaries that contain not only the actual kernel,
3233 but also an initrd, boot splash, command line and OS release
3234 information. This combined binary can then be signed as a
3235 single image, so that the firmware can verify it all in one
3236 step. systemd-boot has special support for EFI binaries created
3237 like this and can extract OS release information from them
3238 and show them in the boot menu. This functionality is useful
3239 to implement cryptographically verified boot schemes.
3241 * Optional support has been added to systemd-fsck to pass
3242 fsck's progress report to an AF_UNIX socket in the file
3245 * udev will no longer create device symlinks for all block
3246 devices by default. A blacklist for excluding special block
3247 devices from this logic has been turned into a whitelist
3248 that requires picking block devices explicitly that require
3251 * A new (currently still internal) API sd-device.h has been
3252 added to libsystemd. This modernized API is supposed to
3253 replace libudev eventually. In fact, already much of libudev
3254 is now just a wrapper around sd-device.h.
3256 * A new hwdb database for storing metadata about pointing
3257 stick devices has been added.
3259 * systemd-tmpfiles gained support for setting file attributes
3260 similar to the "chattr" tool with new 'h' and 'H' lines.
3262 * systemd-journald will no longer unconditionally set the
3263 btrfs NOCOW flag on new journal files. This is instead done
3264 with tmpfiles snippet using the new 'h' line type. This
3265 allows easy disabling of this logic, by masking the
3266 journal-nocow.conf tmpfiles file.
3268 * systemd-journald will now translate audit message types to
3269 human readable identifiers when writing them to the
3270 journal. This should improve readability of audit messages.
3272 * The LUKS logic gained support for the offset= and skip=
3273 options in /etc/crypttab, as previously implemented by
3276 * /usr/lib/os-release gained a new optional field VARIANT= for
3277 distributions that support multiple variants (such as a
3278 desktop edition, a server edition, ...)
3280 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3281 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3282 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3283 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3284 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3285 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3286 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3287 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3288 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3289 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3290 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3291 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3292 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3293 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3294 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3295 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3296 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3297 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3298 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3299 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3300 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3301 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3302 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3303 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3304 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3305 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3306 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3308 — Berlin, 2015-05-22
3312 * Introduce a new API "sd-hwdb.h" for querying the hardware
3313 metadata database. With this minimal interface one can query
3314 and enumerate the udev hwdb, decoupled from the old libudev
3315 library. libudev's interface for this is now only a wrapper
3316 around sd-hwdb. A new tool systemd-hwdb has been added to
3317 interface with and update the database.
3319 * When any of systemd's tools copies files (for example due to
3320 tmpfiles' C lines) a btrfs reflink will attempted first,
3321 before bytewise copying is done.
3323 * systemd-nspawn gained a new --ephemeral switch. When
3324 specified a btrfs snapshot is taken of the container's root
3325 directory, and immediately removed when the container
3326 terminates again. Thus, a container can be started whose
3327 changes never alter the container's root directory, and are
3328 lost on container termination. This switch can also be used
3329 for starting a container off the root file system of the
3330 host without affecting the host OS. This switch is only
3331 available on btrfs file systems.
3333 * systemd-nspawn gained a new --template= switch. It takes the
3334 path to a container tree to use as template for the tree
3335 specified via --directory=, should that directory be
3336 missing. This allows instantiating containers dynamically,
3337 on first run. This switch is only available on btrfs file
3340 * When a .mount unit refers to a mount point on which multiple
3341 mounts are stacked, and the .mount unit is stopped all of
3342 the stacked mount points will now be unmounted until no
3343 mount point remains.
3345 * systemd now has an explicit notion of supported and
3346 unsupported unit types. Jobs enqueued for unsupported unit
3347 types will now fail with an "unsupported" error code. More
3348 specifically .swap, .automount and .device units are not
3349 supported in containers, .busname units are not supported on
3350 non-kdbus systems. .swap and .automount are also not
3351 supported if their respective kernel compile time options
3354 * machinectl gained support for two new "copy-from" and
3355 "copy-to" commands for copying files from a running
3356 container to the host or vice versa.
3358 * machinectl gained support for a new "bind" command to bind
3359 mount host directories into local containers. This is
3360 currently only supported for nspawn containers.
3362 * networkd gained support for configuring bridge forwarding
3363 database entries (fdb) from .network files.
3365 * A new tiny daemon "systemd-importd" has been added that can
3366 download container images in tar, raw, qcow2 or dkr formats,
3367 and make them available locally in /var/lib/machines, so
3368 that they can run as nspawn containers. The daemon can GPG
3369 verify the downloads (not supported for dkr, since it has no
3370 provisions for verifying downloads). It will transparently
3371 decompress bz2, xz, gzip compressed downloads if necessary,
3372 and restore sparse files on disk. The daemon uses privilege
3373 separation to ensure the actual download logic runs with
3374 fewer privileges than the daemon itself. machinectl has
3375 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3376 make the functionality of importd available to the
3377 user. With this in place the Fedora and Ubuntu "Cloud"
3378 images can be downloaded and booted as containers unmodified
3379 (the Fedora images lack the appropriate GPG signature files
3380 currently, so they cannot be verified, but this will change
3381 soon, hopefully). Note that downloading images is currently
3382 only fully supported on btrfs.
3384 * machinectl is now able to list container images found in
3385 /var/lib/machines, along with some metadata about sizes of
3386 disk and similar. If the directory is located on btrfs and
3387 quota is enabled, this includes quota display. A new command
3388 "image-status" has been added that shows additional
3389 information about images.
3391 * machinectl is now able to clone container images
3392 efficiently, if the underlying file system (btrfs) supports
3393 it, with the new "machinectl clone" command. It also
3394 gained commands for renaming and removing images, as well as
3395 marking them read-only or read-write (supported also on
3396 legacy file systems).
3398 * networkd gained support for collecting LLDP network
3399 announcements, from hardware that supports this. This is
3400 shown in networkctl output.
3402 * systemd-run gained support for a new -t (--pty) switch for
3403 invoking a binary on a pty whose input and output is
3404 connected to the invoking terminal. This allows executing
3405 processes as system services while interactively
3406 communicating with them via the terminal. Most interestingly
3407 this is supported across container boundaries. Invoking
3408 "systemd-run -t /bin/bash" is an alternative to running a
3409 full login session, the difference being that the former
3410 will not register a session, nor go through the PAM session
3413 * tmpfiles gained support for a new "v" line type for creating
3414 btrfs subvolumes. If the underlying file system is a legacy
3415 file system, this automatically degrades to creating a
3416 normal directory. Among others /var/lib/machines is now
3417 created like this at boot, should it be missing.
3419 * The directory /var/lib/containers/ has been deprecated and
3420 been replaced by /var/lib/machines. The term "machines" has
3421 been used in the systemd context as generic term for both
3422 VMs and containers, and hence appears more appropriate for
3423 this, as the directory can also contain raw images bootable
3426 * systemd-nspawn when invoked with -M but without --directory=
3427 or --image= is now capable of searching for the container
3428 root directory, subvolume or disk image automatically, in
3429 /var/lib/machines. systemd-nspawn@.service has been updated
3430 to make use of this, thus allowing it to be used for raw
3433 * A new machines.target unit has been introduced that is
3434 supposed to group all containers/VMs invoked as services on
3435 the system. systemd-nspawn@.service has been updated to
3436 integrate with that.
3438 * machinectl gained a new "start" command, for invoking a
3439 container as a service. "machinectl start foo" is mostly
3440 equivalent to "systemctl start systemd-nspawn@foo.service",
3441 but handles escaping in a nicer way.
3443 * systemd-nspawn will now mount most of the cgroupfs tree
3444 read-only into each container, with the exception of the
3445 container's own subtree in the name=systemd hierarchy.
3447 * journald now sets the special FS_NOCOW file flag for its
3448 journal files. This should improve performance on btrfs, by
3449 avoiding heavy fragmentation when journald's write-pattern
3450 is used on COW file systems. It degrades btrfs' data
3451 integrity guarantees for the files to the same levels as for
3452 ext3/ext4 however. This should be OK though as journald does
3453 its own data integrity checks and all its objects are
3454 checksummed on disk. Also, journald should handle btrfs disk
3455 full events a lot more gracefully now, by processing SIGBUS
3456 errors, and not relying on fallocate() anymore.
3458 * When journald detects that journal files it is writing to
3459 have been deleted it will immediately start new journal
3462 * systemd now provides a way to store file descriptors
3463 per-service in PID 1. This is useful for daemons to ensure
3464 that fds they require are not lost during a daemon
3465 restart. The fds are passed to the daemon on the next
3466 invocation in the same way socket activation fds are
3467 passed. This is now used by journald to ensure that the
3468 various sockets connected to all the system's stdout/stderr
3469 are not lost when journald is restarted. File descriptors
3470 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3471 an extension to sd_notify(). Note that a limit is enforced
3472 on the number of fds a service can store in PID 1, and it
3473 defaults to 0, so that no fds may be stored, unless this is
3474 explicitly turned on.
3476 * The default TERM variable to use for units connected to a
3477 terminal, when no other value is explicitly is set is now
3478 vt220 rather than vt102. This should be fairly safe still,
3479 but allows PgUp/PgDn work.
3481 * The /etc/crypttab option header= as known from Debian is now
3484 * "loginctl user-status" and "loginctl session-status" will
3485 now show the last 10 lines of log messages of the
3486 user/session following the status output. Similar,
3487 "machinectl status" will show the last 10 log lines
3488 associated with a virtual machine or container
3489 service. (Note that this is usually not the log messages
3490 done in the VM/container itself, but simply what the
3491 container manager logs. For nspawn this includes all console
3494 * "loginctl session-status" without further argument will now
3495 show the status of the session of the caller. Similar,
3496 "lock-session", "unlock-session", "activate",
3497 "enable-linger", "disable-linger" may now be called without
3498 session/user parameter in which case they apply to the
3499 caller's session/user.
3501 * An X11 session scriptlet is now shipped that uploads
3502 $DISPLAY and $XAUTHORITY into the environment of the systemd
3503 --user daemon if a session begins. This should improve
3504 compatibility with X11 enabled applications run as systemd
3507 * Generators are now subject to masking via /etc and /run, the
3508 same way as unit files.
3510 * networkd .network files gained support for configuring
3511 per-link IPv4/IPv6 packet forwarding as well as IPv4
3512 masquerading. This is by default turned on for veth links to
3513 containers, as registered by systemd-nspawn. This means that
3514 nspawn containers run with --network-veth will now get
3515 automatic routed access to the host's networks without any
3516 further configuration or setup, as long as networkd runs on
3519 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3520 or UDP posts of a container on the host. With this in place
3521 it is possible to run containers with private veth links
3522 (--network-veth), and have their functionality exposed on
3523 the host as if their services were running directly on the
3526 * systemd-nspawn's --network-veth switch now gained a short
3527 version "-n", since with the changes above it is now truly
3528 useful out-of-the-box. The systemd-nspawn@.service has been
3529 updated to make use of it too by default.
3531 * systemd-nspawn will now maintain a per-image R/W lock, to
3532 ensure that the same image is not started more than once
3533 writable. (It's OK to run an image multiple times
3534 simultaneously in read-only mode.)
3536 * systemd-nspawn's --image= option is now capable of
3537 dissecting and booting MBR and GPT disk images that contain
3538 only a single active Linux partition. Previously it
3539 supported only GPT disk images with proper GPT type
3540 IDs. This allows running cloud images from major
3541 distributions directly with systemd-nspawn, without
3544 * In addition to collecting mouse dpi data in the udev
3545 hardware database, there's now support for collecting angle
3546 information for mouse scroll wheels. The database is
3547 supposed to guarantee similar scrolling behavior on mice
3548 that it knows about. There's also support for collecting
3549 information about Touchpad types.
3551 * udev's input_id built-in will now also collect touch screen
3552 dimension data and attach it to probed devices.
3554 * /etc/os-release gained support for a Distribution Privacy
3557 * networkd gained support for creating "ipvlan", "gretap",
3558 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3560 * systemd-tmpfiles gained support for "a" lines for setting
3563 * systemd-nspawn will now mount /tmp in the container to
3564 tmpfs, automatically.
3566 * systemd now exposes the memory.usage_in_bytes cgroup
3567 attribute and shows it for each service in the "systemctl
3568 status" output, if available.
3570 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3571 immediate reboot is triggered. This useful if shutdown is
3572 hung and is unable to complete, to expedite the
3573 operation. Note that this kind of reboot will still unmount
3574 all file systems, and hence should not result in fsck being
3577 * A .device unit for an optical block device will now be
3578 considered active only when a medium is in the drive. Also,
3579 mount units are now bound to their backing devices thus
3580 triggering automatic unmounting when devices become
3581 unavailable. With this in place systemd will now
3582 automatically unmount left-over mounts when a CD-ROM is
3583 ejected or an USB stick is yanked from the system.
3585 * networkd-wait-online now has support for waiting for
3586 specific interfaces only (with globbing), and for giving up
3587 after a configurable timeout.
3589 * networkd now exits when idle. It will be automatically
3590 restarted as soon as interfaces show up, are removed or
3591 change state. networkd will stay around as long as there is
3592 at least one DHCP state machine or similar around, that keep
3595 * networkd may now configure IPv6 link-local addressing in
3596 addition to IPv4 link-local addressing.
3598 * The IPv6 "token" for use in SLAAC may now be configured for
3599 each .network interface in networkd.
3601 * Routes configured with networkd may now be assigned a scope
3604 * networkd's [Match] sections now support globbing and lists
3605 of multiple space-separated matches per item.
3607 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3608 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3609 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3610 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3611 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3612 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3613 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3614 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3615 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3616 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3617 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3618 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3619 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3620 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3621 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3622 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3623 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3624 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3625 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3626 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3627 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3628 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3629 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3630 Hoffmann, Zbigniew Jędrzejewski-Szmek
3632 — Berlin, 2015-02-16
3636 * When querying unit file enablement status (for example via
3637 "systemctl is-enabled"), a new state "indirect" is now known
3638 which indicates that a unit might not be enabled itself, but
3639 another unit listed in its Also= setting might be.
3641 * Similar to the various existing ConditionXYZ= settings for
3642 units, there are now matching AssertXYZ= settings. While
3643 failing conditions cause a unit to be skipped, but its job
3644 to succeed, failing assertions declared like this will cause
3645 a unit start operation and its job to fail.
3647 * hostnamed now knows a new chassis type "embedded".
3649 * systemctl gained a new "edit" command. When used on a unit
3650 file, this allows extending unit files with .d/ drop-in
3651 configuration snippets or editing the full file (after
3652 copying it from /usr/lib to /etc). This will invoke the
3653 user's editor (as configured with $EDITOR), and reload the
3654 modified configuration after editing.
3656 * "systemctl status" now shows the suggested enablement state
3657 for a unit, as declared in the (usually vendor-supplied)
3658 system preset files.
3660 * nss-myhostname will now resolve the single-label host name
3661 "gateway" to the locally configured default IP routing
3662 gateways, ordered by their metrics. This assigns a stable
3663 name to the used gateways, regardless which ones are
3664 currently configured. Note that the name will only be
3665 resolved after all other name sources (if nss-myhostname is
3666 configured properly) and should hence not negatively impact
3667 systems that use the single-label host name "gateway" in
3670 * systemd-inhibit now allows filtering by mode when listing
3673 * Scope and service units gained a new "Delegate" boolean
3674 property, which, when set, allows processes running inside the
3675 unit to further partition resources. This is primarily
3676 useful for systemd user instances as well as container
3679 * journald will now pick up audit messages directly from
3680 the kernel, and log them like any other log message. The
3681 audit fields are split up and fully indexed. This means that
3682 journalctl in many ways is now a (nicer!) alternative to
3683 ausearch, the traditional audit client. Note that this
3684 implements only a minimal audit client. If you want the
3685 special audit modes like reboot-on-log-overflow, please use
3686 the traditional auditd instead, which can be used in
3687 parallel to journald.
3689 * The ConditionSecurity= unit file option now understands the
3690 special string "audit" to check whether auditing is
3693 * journalctl gained two new commands --vacuum-size= and
3694 --vacuum-time= to delete old journal files until the
3695 remaining ones take up no more than the specified size on disk,
3696 or are not older than the specified time.
3698 * A new, native PPPoE library has been added to sd-network,
3699 systemd's library of light-weight networking protocols. This
3700 library will be used in a future version of networkd to
3701 enable PPPoE communication without an external pppd daemon.
3703 * The busctl tool now understands a new "capture" verb that
3704 works similar to "monitor", but writes a packet capture
3705 trace to STDOUT that can be redirected to a file which is
3706 compatible with libcap's capture file format. This can then
3707 be loaded in Wireshark and similar tools to inspect bus
3710 * The busctl tool now understands a new "tree" verb that shows
3711 the object trees of a specific service on the bus, or of all
3714 * The busctl tool now understands a new "introspect" verb that
3715 shows all interfaces and members of objects on the bus,
3716 including their signature and values. This is particularly
3717 useful to get more information about bus objects shown by
3718 the new "busctl tree" command.
3720 * The busctl tool now understands new verbs "call",
3721 "set-property" and "get-property" for invoking bus method
3722 calls, setting and getting bus object properties in a
3725 * busctl gained a new --augment-creds= argument that controls
3726 whether the tool shall augment credential information it
3727 gets from the bus with data from /proc, in a possibly
3730 * nspawn's --link-journal= switch gained two new values
3731 "try-guest" and "try-host" that work like "guest" and
3732 "host", but do not fail if the host has no persistent
3733 journalling enabled. -j is now equivalent to
3734 --link-journal=try-guest.
3736 * macvlan network devices created by nspawn will now have
3737 stable MAC addresses.
3739 * A new SmackProcessLabel= unit setting has been added, which
3740 controls the SMACK security label processes forked off by
3741 the respective unit shall use.
3743 * If compiled with --enable-xkbcommon, systemd-localed will
3744 verify x11 keymap settings by compiling the given keymap. It
3745 will spew out warnings if the compilation fails. This
3746 requires libxkbcommon to be installed.
3748 * When a coredump is collected, a larger number of metadata
3749 fields is now collected and included in the journal records
3750 created for it. More specifically, control group membership,
3751 environment variables, memory maps, working directory,
3752 chroot directory, /proc/$PID/status, and a list of open file
3753 descriptors is now stored in the log entry.
3755 * The udev hwdb now contains DPI information for mice. For
3758 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3760 * All systemd programs that read standalone configuration
3761 files in /etc now also support a corresponding series of
3762 .conf.d configuration directories in /etc/, /run/,
3763 /usr/local/lib/, /usr/lib/, and (if configured with
3764 --enable-split-usr) /lib/. In particular, the following
3765 configuration files now have corresponding configuration
3766 directories: system.conf user.conf, logind.conf,
3767 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3768 resolved.conf, timesyncd.conf, journal-remote.conf, and
3769 journal-upload.conf. Note that distributions should use the
3770 configuration directories in /usr/lib/; the directories in
3771 /etc/ are reserved for the system administrator.
3773 * systemd-rfkill will no longer take the rfkill device name
3774 into account when storing rfkill state on disk, as the name
3775 might be dynamically assigned and not stable. Instead, the
3776 ID_PATH udev variable combined with the rfkill type (wlan,
3777 bluetooth, ...) is used.
3779 * A new service systemd-machine-id-commit.service has been
3780 added. When used on systems where /etc is read-only during
3781 boot, and /etc/machine-id is not initialized (but an empty
3782 file), this service will copy the temporary machine ID
3783 created as replacement into /etc after the system is fully
3784 booted up. This is useful for systems that are freshly
3785 installed with a non-initialized machine ID, but should get
3786 a fixed machine ID for subsequent boots.
3788 * networkd's .netdev files now provide a large set of
3789 configuration parameters for VXLAN devices. Similarly, the
3790 bridge port cost parameter is now configurable in .network
3791 files. There's also new support for configuring IP source
3792 routing. networkd .link files gained support for a new
3793 OriginalName= match that is useful to match against the
3794 original interface name the kernel assigned. .network files
3795 may include MTU= and MACAddress= fields for altering the MTU
3796 and MAC address while being connected to a specific network
3799 * The LUKS logic gained supported for configuring
3800 UUID-specific key files. There's also new support for naming
3801 LUKS device from the kernel command line, using the new
3802 luks.name= argument.
3804 * Timer units may now be transiently created via the bus API
3805 (this was previously already available for scope and service
3806 units). In addition it is now possible to create multiple
3807 transient units at the same time with a single bus call. The
3808 "systemd-run" tool has been updated to make use of this for
3809 running commands on a specified time, in at(1)-style.
3811 * tmpfiles gained support for "t" lines, for assigning
3812 extended attributes to files. Among other uses this may be
3813 used to assign SMACK labels to files.
3815 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3816 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3817 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3818 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3819 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3820 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3821 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3822 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3823 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3824 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3825 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3826 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3827 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3828 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3829 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3830 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3831 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3832 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3834 — Berlin, 2014-12-10
3838 * journalctl gained the new options -t/--identifier= to match
3839 on the syslog identifier (aka "tag"), as well as --utc to
3840 show log timestamps in the UTC timezone. journalctl now also
3841 accepts -n/--lines=all to disable line capping in a pager.
3843 * journalctl gained a new switch, --flush, that synchronously
3844 flushes logs from /run/log/journal to /var/log/journal if
3845 persistent storage is enabled. systemd-journal-flush.service
3846 now waits until the operation is complete.
3848 * Services can notify the manager before they start a reload
3849 (by sending RELOADING=1) or shutdown (by sending
3850 STOPPING=1). This allows the manager to track and show the
3851 internal state of daemons and closes a race condition when
3852 the process is still running but has closed its D-Bus
3855 * Services with Type=oneshot do not have to have any ExecStart
3858 * User units are now loaded also from
3859 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3860 /run/systemd/user directory that was already previously
3861 supported, but is under the control of the user.
3863 * Job timeouts (i.e. time-outs on the time a job that is
3864 queued stays in the run queue) can now optionally result in
3865 immediate reboot or power-off actions (JobTimeoutAction= and
3866 JobTimeoutRebootArgument=). This is useful on ".target"
3867 units, to limit the maximum time a target remains
3868 undispatched in the run queue, and to trigger an emergency
3869 operation in such a case. This is now used by default to
3870 turn off the system if boot-up (as defined by everything in
3871 basic.target) hangs and does not complete for at least
3872 15min. Also, if power-off or reboot hang for at least 30min
3873 an immediate power-off/reboot operation is triggered. This
3874 functionality is particularly useful to increase reliability
3875 on embedded devices, but also on laptops which might
3876 accidentally get powered on when carried in a backpack and
3877 whose boot stays stuck in a hard disk encryption passphrase
3880 * systemd-logind can be configured to also handle lid switch
3881 events even when the machine is docked or multiple displays
3882 are attached (HandleLidSwitchDocked= option).
3884 * A helper binary and a service have been added which can be
3885 used to resume from hibernation in the initramfs. A
3886 generator will parse the resume= option on the kernel
3887 command line to trigger resume.
3889 * A user console daemon systemd-consoled has been
3890 added. Currently, it is a preview, and will so far open a
3891 single terminal on each session of the user marked as
3892 Desktop=systemd-console.
3894 * Route metrics can be specified for DHCP routes added by
3897 * The SELinux context of socket-activated services can be set
3898 from the information provided by the networking stack
3899 (SELinuxContextFromNet= option).
3901 * Userspace firmware loading support has been removed and
3902 the minimum supported kernel version is thus bumped to 3.7.
3904 * Timeout for udev workers has been increased from 1 to 3
3905 minutes, but a warning will be printed after 1 minute to
3906 help diagnose kernel modules that take a long time to load.
3908 * Udev rules can now remove tags on devices with TAG-="foobar".
3910 * systemd's readahead implementation has been removed. In many
3911 circumstances it didn't give expected benefits even for
3912 rotational disk drives and was becoming less relevant in the
3913 age of SSDs. As none of the developers has been using
3914 rotating media anymore, and nobody stepped up to actively
3915 maintain this component of systemd it has now been removed.
3917 * Swap units can use Options= to specify discard options.
3918 Discard options specified for swaps in /etc/fstab are now
3921 * Docker containers are now detected as a separate type of
3924 * The Password Agent protocol gained support for queries where
3925 the user input is shown, useful e.g. for user names.
3926 systemd-ask-password gained a new --echo option to turn that
3929 * The default sysctl.d/ snippets will now set:
3931 net.core.default_qdisc = fq_codel
3933 This selects Fair Queuing Controlled Delay as the default
3934 queuing discipline for network interfaces. fq_codel helps
3935 fight the network bufferbloat problem. It is believed to be
3936 a good default with no tuning required for most workloads.
3937 Downstream distributions may override this choice. On 10Gbit
3938 servers that do not do forwarding, "fq" may perform better.
3939 Systems without a good clocksource should use "pfifo_fast".
3941 * If kdbus is enabled during build a new option BusPolicy= is
3942 available for service units, that allows locking all service
3943 processes into a stricter bus policy, in order to limit
3944 access to various bus services, or even hide most of them
3945 from the service's view entirely.
3947 * networkctl will now show the .network and .link file
3948 networkd has applied to a specific interface.
3950 * sd-login gained a new API call sd_session_get_desktop() to
3951 query which desktop environment has been selected for a
3954 * UNIX utmp support is now compile-time optional to support
3955 legacy-free systems.
3957 * systemctl gained two new commands "add-wants" and
3958 "add-requires" for pulling in units from specific targets
3961 * If the word "rescue" is specified on the kernel command line
3962 the system will now boot into rescue mode (aka
3963 rescue.target), which was previously available only by
3964 specifying "1" or "systemd.unit=rescue.target" on the kernel
3965 command line. This new kernel command line option nicely
3966 mirrors the already existing "emergency" kernel command line
3969 * New kernel command line options mount.usr=, mount.usrflags=,
3970 mount.usrfstype= have been added that match root=, rootflags=,
3971 rootfstype= but allow mounting a specific file system to
3974 * The $NOTIFY_SOCKET is now also passed to control processes of
3975 services, not only the main process.
3977 * This version reenables support for fsck's -l switch. This
3978 means at least version v2.25 of util-linux is required for
3979 operation, otherwise dead-locks on device nodes may
3980 occur. Again: you need to update util-linux to at least
3981 v2.25 when updating systemd to v217.
3983 * The "multi-seat-x" tool has been removed from systemd, as
3984 its functionality has been integrated into X servers 1.16,
3985 and the tool is hence redundant. It is recommended to update
3986 display managers invoking this tool to simply invoke X
3987 directly from now on, again.
3989 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3990 message flag has been added for all of systemd's PolicyKit
3991 authenticated method calls has been added. In particular
3992 this now allows optional interactive authorization via
3993 PolicyKit for many of PID1's privileged operations such as
3994 unit file enabling and disabling.
3996 * "udevadm hwdb --update" learnt a new switch "--usr" for
3997 placing the rebuilt hardware database in /usr instead of
3998 /etc. When used only hardware database entries stored in
3999 /usr will be used, and any user database entries in /etc are
4000 ignored. This functionality is useful for vendors to ship a
4001 pre-built database on systems where local configuration is
4002 unnecessary or unlikely.
4004 * Calendar time specifications in .timer units now also
4005 understand the strings "semi-annually", "quarterly" and
4006 "minutely" as shortcuts (in addition to the preexisting
4007 "anually", "hourly", ...).
4009 * systemd-tmpfiles will now correctly create files in /dev
4010 at boot which are marked for creation only at boot. It is
4011 recommended to always create static device nodes with 'c!'
4012 and 'b!', so that they are created only at boot and not
4013 overwritten at runtime.
4015 * When the watchdog logic is used for a service (WatchdogSec=)
4016 and the watchdog timeout is hit the service will now be
4017 terminated with SIGABRT (instead of just SIGTERM), in order
4018 to make sure a proper coredump and backtrace is
4019 generated. This ensures that hanging services will result in
4020 similar coredump/backtrace behaviour as services that hit a
4023 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4024 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4025 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4026 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4027 Herrmann, David Sommerseth, David Strauss, Emil Renner
4028 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4029 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4030 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4031 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4032 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4033 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4034 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4035 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4036 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4037 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4038 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4039 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4040 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4041 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4042 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4043 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4046 — Berlin, 2014-10-28
4050 * timedated no longer reads NTP implementation unit names from
4051 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4052 implementations should add a
4054 Conflicts=systemd-timesyncd.service
4056 to their unit files to take over and replace systemd's NTP
4057 default functionality.
4059 * systemd-sysusers gained a new line type "r" for configuring
4060 which UID/GID ranges to allocate system users/groups
4061 from. Lines of type "u" may now add an additional column
4062 that specifies the home directory for the system user to be
4063 created. Also, systemd-sysusers may now optionally read user
4064 information from STDIN instead of a file. This is useful for
4065 invoking it from RPM preinst scriptlets that need to create
4066 users before the first RPM file is installed since these
4067 files might need to be owned by them. A new
4068 %sysusers_create_inline RPM macro has been introduced to do
4069 just that. systemd-sysusers now updates the shadow files as
4070 well as the user/group databases, which should enhance
4071 compatibility with certain tools like grpck.
4073 * A number of bus APIs of PID 1 now optionally consult
4074 PolicyKit to permit access for otherwise unprivileged
4075 clients under certain conditions. Note that this currently
4076 doesn't support interactive authentication yet, but this is
4077 expected to be added eventually, too.
4079 * /etc/machine-info now has new fields for configuring the
4080 deployment environment of the machine, as well as the
4081 location of the machine. hostnamectl has been updated with
4082 new command to update these fields.
4084 * systemd-timesyncd has been updated to automatically acquire
4085 NTP server information from systemd-networkd, which might
4086 have been discovered via DHCP.
4088 * systemd-resolved now includes a caching DNS stub resolver
4089 and a complete LLMNR name resolution implementation. A new
4090 NSS module "nss-resolve" has been added which can be used
4091 instead of glibc's own "nss-dns" to resolve hostnames via
4092 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4093 be resolved via systemd-resolved D-Bus APIs. In contrast to
4094 the glibc internal resolver systemd-resolved is aware of
4095 multi-homed system, and keeps DNS server and caches separate
4096 and per-interface. Queries are sent simultaneously on all
4097 interfaces that have DNS servers configured, in order to
4098 properly handle VPNs and local LANs which might resolve
4099 separate sets of domain names. systemd-resolved may acquire
4100 DNS server information from systemd-networkd automatically,
4101 which in turn might have discovered them via DHCP. A tool
4102 "systemd-resolve-host" has been added that may be used to
4103 query the DNS logic in resolved. systemd-resolved implements
4104 IDNA and automatically uses IDNA or UTF-8 encoding depending
4105 on whether classic DNS or LLMNR is used as transport. In the
4106 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4107 implementation to systemd-resolved.
4109 * A new NSS module nss-mymachines has been added, that
4110 automatically resolves the names of all local registered
4111 containers to their respective IP addresses.
4113 * A new client tool "networkctl" for systemd-networkd has been
4114 added. It currently is entirely passive and will query
4115 networking configuration from udev, rtnetlink and networkd,
4116 and present it to the user in a very friendly
4117 way. Eventually, we hope to extend it to become a full
4118 control utility for networkd.
4120 * .socket units gained a new DeferAcceptSec= setting that
4121 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4122 TCP. Similarly, support for controlling TCP keep-alive
4123 settings has been added (KeepAliveTimeSec=,
4124 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4125 turning off Nagle's algorithm on TCP has been added
4128 * logind learned a new session type "web", for use in projects
4129 like Cockpit which register web clients as PAM sessions.
4131 * timer units with at least one OnCalendar= setting will now
4132 be started only after time-sync.target has been
4133 reached. This way they will not elapse before the system
4134 clock has been corrected by a local NTP client or
4135 similar. This is particular useful on RTC-less embedded
4136 machines, that come up with an invalid system clock.
4138 * systemd-nspawn's --network-veth= switch should now result in
4139 stable MAC addresses for both the outer and the inner side
4142 * systemd-nspawn gained a new --volatile= switch for running
4143 container instances with /etc or /var unpopulated.
4145 * The kdbus client code has been updated to use the new Linux
4146 3.17 memfd subsystem instead of the old kdbus-specific one.
4148 * systemd-networkd's DHCP client and server now support
4149 FORCERENEW. There are also new configuration options to
4150 configure the vendor client identifier and broadcast mode
4153 * systemd will no longer inform the kernel about the current
4154 timezone, as this is necessarily incorrect and racy as the
4155 kernel has no understanding of DST and similar
4156 concepts. This hence means FAT timestamps will be always
4157 considered UTC, similar to what Android is already
4158 doing. Also, when the RTC is configured to the local time
4159 (rather than UTC) systemd will never synchronize back to it,
4160 as this might confuse Windows at a later boot.
4162 * systemd-analyze gained a new command "verify" for offline
4163 validation of unit files.
4165 * systemd-networkd gained support for a couple of additional
4166 settings for bonding networking setups. Also, the metric for
4167 statically configured routes may now be configured. For
4168 network interfaces where this is appropriate the peer IP
4169 address may now be configured.
4171 * systemd-networkd's DHCP client will no longer request
4172 broadcasting by default, as this tripped up some networks.
4173 For hardware where broadcast is required the feature should
4174 be switched back on using RequestBroadcast=yes.
4176 * systemd-networkd will now set up IPv4LL addresses (when
4177 enabled) even if DHCP is configured successfully.
4179 * udev will now default to respect network device names given
4180 by the kernel when the kernel indicates that these are
4181 predictable. This behavior can be tweaked by changing
4182 NamePolicy= in the relevant .link file.
4184 * A new library systemd-terminal has been added that
4185 implements full TTY stream parsing and rendering. This
4186 library is supposed to be used later on for implementing a
4187 full userspace VT subsystem, replacing the current kernel
4190 * A new tool systemd-journal-upload has been added to push
4191 journal data to a remote system running
4192 systemd-journal-remote.
4194 * journald will no longer forward all local data to another
4195 running syslog daemon. This change has been made because
4196 rsyslog (which appears to be the most commonly used syslog
4197 implementation these days) no longer makes use of this, and
4198 instead pulls the data out of the journal on its own. Since
4199 forwarding the messages to a non-existent syslog server is
4200 more expensive than we assumed we have now turned this
4201 off. If you run a syslog server that is not a recent rsyslog
4202 version, you have to turn this option on again
4203 (ForwardToSyslog= in journald.conf).
4205 * journald now optionally supports the LZ4 compressor for
4206 larger journal fields. This compressor should perform much
4207 better than XZ which was the previous default.
4209 * machinectl now shows the IP addresses of local containers,
4210 if it knows them, plus the interface name of the container.
4212 * A new tool "systemd-escape" has been added that makes it
4213 easy to escape strings to build unit names and similar.
4215 * sd_notify() messages may now include a new ERRNO= field
4216 which is parsed and collected by systemd and shown among the
4217 "systemctl status" output for a service.
4219 * A new component "systemd-firstboot" has been added that
4220 queries the most basic systemd information (timezone,
4221 hostname, root password) interactively on first
4222 boot. Alternatively it may also be used to provision these
4223 things offline on OS images installed into directories.
4225 * The default sysctl.d/ snippets will now set
4227 net.ipv4.conf.default.promote_secondaries=1
4229 This has the benefit of no flushing secondary IP addresses
4230 when primary addresses are removed.
4232 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4233 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4234 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4235 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4236 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4237 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4238 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4239 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4240 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4241 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4242 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4243 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4244 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4245 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4246 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4248 — Berlin, 2014-08-19
4252 * A new tool systemd-sysusers has been added. This tool
4253 creates system users and groups in /etc/passwd and
4254 /etc/group, based on static declarative system user/group
4255 definitions in /usr/lib/sysusers.d/. This is useful to
4256 enable factory resets and volatile systems that boot up with
4257 an empty /etc directory, and thus need system users and
4258 groups created during early boot. systemd now also ships
4259 with two default sysusers.d/ files for the most basic
4260 users and groups systemd and the core operating system
4263 * A new tmpfiles snippet has been added that rebuilds the
4264 essential files in /etc on boot, should they be missing.
4266 * A directive for ensuring automatic clean-up of
4267 /var/cache/man/ has been removed from the default
4268 configuration. This line should now be shipped by the man
4269 implementation. The necessary change has been made to the
4270 man-db implementation. Note that you need to update your man
4271 implementation to one that ships this line, otherwise no
4272 automatic clean-up of /var/cache/man will take place.
4274 * A new condition ConditionNeedsUpdate= has been added that
4275 may conditionalize services to only run when /etc or /var
4276 are "older" than the vendor operating system resources in
4277 /usr. This is useful for reconstructing or updating /etc
4278 after an offline update of /usr or a factory reset, on the
4279 next reboot. Services that want to run once after such an
4280 update or reset should use this condition and order
4281 themselves before the new systemd-update-done.service, which
4282 will mark the two directories as fully updated. A number of
4283 service files have been added making use of this, to rebuild
4284 the udev hardware database, the journald message catalog and
4285 dynamic loader cache (ldconfig). The systemd-sysusers tool
4286 described above also makes use of this now. With this in
4287 place it is now possible to start up a minimal operating
4288 system with /etc empty cleanly. For more information on the
4289 concepts involved see this recent blog story:
4291 http://0pointer.de/blog/projects/stateless.html
4293 * A new system group "input" has been introduced, and all
4294 input device nodes get this group assigned. This is useful
4295 for system-level software to get access to input devices. It
4296 complements what is already done for "audio" and "video".
4298 * systemd-networkd learnt minimal DHCPv4 server support in
4299 addition to the existing DHCPv4 client support. It also
4300 learnt DHCPv6 client and IPv6 Router Solicitation client
4301 support. The DHCPv4 client gained support for static routes
4302 passed in from the server. Note that the [DHCPv4] section
4303 known in older systemd-networkd versions has been renamed to
4304 [DHCP] and is now also used by the DHCPv6 client. Existing
4305 .network files using settings of this section should be
4306 updated, though compatibility is maintained. Optionally, the
4307 client hostname may now be sent to the DHCP server.
4309 * networkd gained support for vxlan virtual networks as well
4310 as tun/tap and dummy devices.
4312 * networkd gained support for automatic allocation of address
4313 ranges for interfaces from a system-wide pool of
4314 addresses. This is useful for dynamically managing a large
4315 number of interfaces with a single network configuration
4316 file. In particular this is useful to easily assign
4317 appropriate IP addresses to the veth links of a large number
4318 of nspawn instances.
4320 * RPM macros for processing sysusers, sysctl and binfmt
4321 drop-in snippets at package installation time have been
4324 * The /etc/os-release file should now be placed in
4325 /usr/lib/os-release. The old location is automatically
4326 created as symlink. /usr/lib is the more appropriate
4327 location of this file, since it shall actually describe the
4328 vendor operating system shipped in /usr, and not the
4329 configuration stored in /etc.
4331 * .mount units gained a new boolean SloppyOptions= setting
4332 that maps to mount(8)'s -s option which enables permissive
4333 parsing of unknown mount options.
4335 * tmpfiles learnt a new "L+" directive which creates a symlink
4336 but (unlike "L") deletes a pre-existing file first, should
4337 it already exist and not already be the correct
4338 symlink. Similarly, "b+", "c+" and "p+" directives have been
4339 added as well, which create block and character devices, as
4340 well as fifos in the filesystem, possibly removing any
4341 pre-existing files of different types.
4343 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4344 'argument' field (which so far specified the source to
4345 symlink/copy the files from) is now optional. If omitted the
4346 same file os copied from /usr/share/factory/ suffixed by the
4347 full destination path. This is useful for populating /etc
4348 with essential files, by copying them from vendor defaults
4349 shipped in /usr/share/factory/etc.
4351 * A new command "systemctl preset-all" has been added that
4352 applies the service preset settings to all installed unit
4353 files. A new switch --preset-mode= has been added that
4354 controls whether only enable or only disable operations
4357 * A new command "systemctl is-system-running" has been added
4358 that allows checking the overall state of the system, for
4359 example whether it is fully up and running.
4361 * When the system boots up with an empty /etc, the equivalent
4362 to "systemctl preset-all" is executed during early boot, to
4363 make sure all default services are enabled after a factory
4366 * systemd now contains a minimal preset file that enables the
4367 most basic services systemd ships by default.
4369 * Unit files' [Install] section gained a new DefaultInstance=
4370 field for defining the default instance to create if a
4371 template unit is enabled with no instance specified.
4373 * A new passive target cryptsetup-pre.target has been added
4374 that may be used by services that need to make they run and
4375 finish before the first LUKS cryptographic device is set up.
4377 * The /dev/loop-control and /dev/btrfs-control device nodes
4378 are now owned by the "disk" group by default, opening up
4379 access to this group.
4381 * systemd-coredump will now automatically generate a
4382 stack trace of all core dumps taking place on the system,
4383 based on elfutils' libdw library. This stack trace is logged
4386 * systemd-coredump may now optionally store coredumps directly
4387 on disk (in /var/lib/systemd/coredump, possibly compressed),
4388 instead of storing them unconditionally in the journal. This
4389 mode is the new default. A new configuration file
4390 /etc/systemd/coredump.conf has been added to configure this
4391 and other parameters of systemd-coredump.
4393 * coredumpctl gained a new "info" verb to show details about a
4394 specific coredump. A new switch "-1" has also been added
4395 that makes sure to only show information about the most
4396 recent entry instead of all entries. Also, as the tool is
4397 generally useful now the "systemd-" prefix of the binary
4398 name has been removed. Distributions that want to maintain
4399 compatibility with the old name should add a symlink from
4400 the old name to the new name.
4402 * journald's SplitMode= now defaults to "uid". This makes sure
4403 that unprivileged users can access their own coredumps with
4404 coredumpctl without restrictions.
4406 * New kernel command line options "systemd.wants=" (for
4407 pulling an additional unit during boot), "systemd.mask="
4408 (for masking a specific unit for the boot), and
4409 "systemd.debug-shell" (for enabling the debug shell on tty9)
4410 have been added. This is implemented in the new generator
4411 "systemd-debug-generator".
4413 * systemd-nspawn will now by default filter a couple of
4414 syscalls for containers, among them those required for
4415 kernel module loading, direct x86 IO port access, swap
4416 management, and kexec. Most importantly though
4417 open_by_handle_at() is now prohibited for containers,
4418 closing a hole similar to a recently discussed vulnerability
4419 in docker regarding access to files on file hierarchies the
4420 container should normally not have access to. Note that, for
4421 nspawn, we generally make no security claims anyway (and
4422 this is explicitly documented in the man page), so this is
4423 just a fix for one of the most obvious problems.
4425 * A new man page file-hierarchy(7) has been added that
4426 contains a minimized, modernized version of the file system
4427 layout systemd expects, similar in style to the FHS
4428 specification or hier(5). A new tool systemd-path(1) has
4429 been added to query many of these paths for the local
4432 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4433 longer done. Since the directory now has a per-user size
4434 limit, and is cleaned on logout this appears unnecessary,
4435 in particular since this now brings the lifecycle of this
4436 directory closer in line with how IPC objects are handled.
4438 * systemd.pc now exports a number of additional directories,
4439 including $libdir (which is useful to identify the library
4440 path for the primary architecture of the system), and a
4441 couple of drop-in directories.
4443 * udev's predictable network interface names now use the dev_port
4444 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4445 distinguish between ports of the same PCI function. dev_id should
4446 only be used for ports using the same HW address, hence the need
4449 * machined has been updated to export the OS version of a
4450 container (read from /etc/os-release and
4451 /usr/lib/os-release) on the bus. This is now shown in
4452 "machinectl status" for a machine.
4454 * A new service setting RestartForceExitStatus= has been
4455 added. If configured to a set of exit signals or process
4456 return values, the service will be restarted when the main
4457 daemon process exits with any of them, regardless of the
4460 * systemctl's -H switch for connecting to remote systemd
4461 machines has been extended so that it may be used to
4462 directly connect to a specific container on the
4463 host. "systemctl -H root@foobar:waldi" will now connect as
4464 user "root" to host "foobar", and then proceed directly to
4465 the container named "waldi". Note that currently you have to
4466 authenticate as user "root" for this to work, as entering
4467 containers is a privileged operation.
4469 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4470 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4471 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4472 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4473 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4474 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4475 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4476 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4477 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4478 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4479 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4480 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4482 — Berlin, 2014-07-03
4486 * As an experimental feature, udev now tries to lock the
4487 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4488 executes events for the disk or any of its partitions.
4489 Applications like partitioning programs can lock the
4490 disk device node (flock(LOCK_EX)) and claim temporary
4491 device ownership that way; udev will entirely skip all event
4492 handling for this disk and its partitions. If the disk
4493 was opened for writing, the close will trigger a partition
4494 table rescan in udev's "watch" facility, and if needed
4495 synthesize "change" events for the disk and all its partitions.
4496 This is now unconditionally enabled, and if it turns out to
4497 cause major problems, we might turn it on only for specific
4498 devices, or might need to disable it entirely. Device Mapper
4499 devices are excluded from this logic.
4501 * We temporarily dropped the "-l" switch for fsck invocations,
4502 since they collide with the flock() logic above. util-linux
4503 upstream has been changed already to avoid this conflict,
4504 and we will readd "-l" as soon as util-linux with this
4505 change has been released.
4507 * The dependency on libattr has been removed. Since a long
4508 time, the extended attribute calls have moved to glibc, and
4509 libattr is thus unnecessary.
4511 * Virtualization detection works without privileges now. This
4512 means the systemd-detect-virt binary no longer requires
4513 CAP_SYS_PTRACE file capabilities, and our daemons can run
4514 with fewer privileges.
4516 * systemd-networkd now runs under its own "systemd-network"
4517 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4518 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4519 loses the ability to write to files owned by root this way.
4521 * Similarly, systemd-resolved now runs under its own
4522 "systemd-resolve" user with no capabilities remaining.
4524 * Similarly, systemd-bus-proxyd now runs under its own
4525 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4527 * systemd-networkd gained support for setting up "veth"
4528 virtual Ethernet devices for container connectivity, as well
4529 as GRE and VTI tunnels.
4531 * systemd-networkd will no longer automatically attempt to
4532 manually load kernel modules necessary for certain tunnel
4533 transports. Instead, it is assumed the kernel loads them
4534 automatically when required. This only works correctly on
4535 very new kernels. On older kernels, please consider adding
4536 the kernel modules to /etc/modules-load.d/ as a work-around.
4538 * The resolv.conf file systemd-resolved generates has been
4539 moved to /run/systemd/resolve/. If you have a symlink from
4540 /etc/resolv.conf, it might be necessary to correct it.
4542 * Two new service settings, ProtectHome= and ProtectSystem=,
4543 have been added. When enabled, they will make the user data
4544 (such as /home) inaccessible or read-only and the system
4545 (such as /usr) read-only, for specific services. This allows
4546 very light-weight per-service sandboxing to avoid
4547 modifications of user data or system files from
4548 services. These two new switches have been enabled for all
4549 of systemd's long-running services, where appropriate.
4551 * Socket units gained new SocketUser= and SocketGroup=
4552 settings to set the owner user and group of AF_UNIX sockets
4553 and FIFOs in the file system.
4555 * Socket units gained a new RemoveOnStop= setting. If enabled,
4556 all FIFOS and sockets in the file system will be removed
4557 when the specific socket unit is stopped.
4559 * Socket units gained a new Symlinks= setting. It takes a list
4560 of symlinks to create to file system sockets or FIFOs
4561 created by the specific Unix sockets. This is useful to
4562 manage symlinks to socket nodes with the same life-cycle as
4565 * The /dev/log socket and /dev/initctl FIFO have been moved to
4566 /run, and have been replaced by symlinks. This allows
4567 connecting to these facilities even if PrivateDevices=yes is
4568 used for a service (which makes /dev/log itself unavailable,
4569 but /run is left). This also has the benefit of ensuring
4570 that /dev only contains device nodes, directories and
4571 symlinks, and nothing else.
4573 * sd-daemon gained two new calls sd_pid_notify() and
4574 sd_pid_notifyf(). They are similar to sd_notify() and
4575 sd_notifyf(), but allow overriding of the source PID of
4576 notification messages if permissions permit this. This is
4577 useful to send notify messages on behalf of a different
4578 process (for example, the parent process). The
4579 systemd-notify tool has been updated to make use of this
4580 when sending messages (so that notification messages now
4581 originate from the shell script invoking systemd-notify and
4582 not the systemd-notify process itself. This should minimize
4583 a race where systemd fails to associate notification
4584 messages to services when the originating process already
4587 * A new "on-abnormal" setting for Restart= has been added. If
4588 set, it will result in automatic restarts on all "abnormal"
4589 reasons for a process to exit, which includes unclean
4590 signals, core dumps, timeouts and watchdog timeouts, but
4591 does not include clean and unclean exit codes or clean
4592 signals. Restart=on-abnormal is an alternative for
4593 Restart=on-failure for services that shall be able to
4594 terminate and avoid restarts on certain errors, by
4595 indicating so with an unclean exit code. Restart=on-failure
4596 or Restart=on-abnormal is now the recommended setting for
4597 all long-running services.
4599 * If the InaccessibleDirectories= service setting points to a
4600 mount point (or if there are any submounts contained within
4601 it), it is now attempted to completely unmount it, to make
4602 the file systems truly unavailable for the respective
4605 * The ReadOnlyDirectories= service setting and
4606 systemd-nspawn's --read-only parameter are now recursively
4607 applied to all submounts, too.
4609 * Mount units may now be created transiently via the bus APIs.
4611 * The support for SysV and LSB init scripts has been removed
4612 from the systemd daemon itself. Instead, it is now
4613 implemented as a generator that creates native systemd units
4614 from these scripts when needed. This enables us to remove a
4615 substantial amount of legacy code from PID 1, following the
4616 fact that many distributions only ship a very small number
4617 of LSB/SysV init scripts nowadays.
4619 * Privileged Xen (dom0) domains are not considered
4620 virtualization anymore by the virtualization detection
4621 logic. After all, they generally have unrestricted access to
4622 the hardware and usually are used to manage the unprivileged
4625 * systemd-tmpfiles gained a new "C" line type, for copying
4626 files or entire directories.
4628 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4629 lines. So far, they have been non-globbing versions of the
4630 latter, and have thus been redundant. In future, it is
4631 recommended to only use "z". "m" has hence been removed
4632 from the documentation, even though it stays supported.
4634 * A tmpfiles snippet to recreate the most basic structure in
4635 /var has been added. This is enough to create the /var/run →
4636 /run symlink and create a couple of structural
4637 directories. This allows systems to boot up with an empty or
4638 volatile /var. Of course, while with this change, the core OS
4639 now is capable with dealing with a volatile /var, not all
4640 user services are ready for it. However, we hope that sooner
4641 or later, many service daemons will be changed upstream so
4642 that they are able to automatically create their necessary
4643 directories in /var at boot, should they be missing. This is
4644 the first step to allow state-less systems that only require
4645 the vendor image for /usr to boot.
4647 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4648 empty tmpfs instance to a specific directory. This is
4649 particularly useful for making use of the automatic
4650 reconstruction of /var (see above), by passing --tmpfs=/var.
4652 * Access modes specified in tmpfiles snippets may now be
4653 prefixed with "~", which indicates that they shall be masked
4654 by whether the existing file or directory is currently
4655 writable, readable or executable at all. Also, if specified,
4656 the sgid/suid/sticky bits will be masked for all
4659 * A new passive target unit "network-pre.target" has been
4660 added which is useful for services that shall run before any
4661 network is configured, for example firewall scripts.
4663 * The "floppy" group that previously owned the /dev/fd*
4664 devices is no longer used. The "disk" group is now used
4665 instead. Distributions should probably deprecate usage of
4668 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4669 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4670 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4671 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4672 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4673 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4676 — Berlin, 2014-06-11
4680 * A new "systemd-timesyncd" daemon has been added for
4681 synchronizing the system clock across the network. It
4682 implements an SNTP client. In contrast to NTP
4683 implementations such as chrony or the NTP reference server,
4684 this only implements a client side, and does not bother with
4685 the full NTP complexity, focusing only on querying time from
4686 one remote server and synchronizing the local clock to
4687 it. Unless you intend to serve NTP to networked clients or
4688 want to connect to local hardware clocks, this simple NTP
4689 client should be more than appropriate for most
4690 installations. The daemon runs with minimal privileges, and
4691 has been hooked up with networkd to only operate when
4692 network connectivity is available. The daemon saves the
4693 current clock to disk every time a new NTP sync has been
4694 acquired, and uses this to possibly correct the system clock
4695 early at bootup, in order to accommodate for systems that
4696 lack an RTC such as the Raspberry Pi and embedded devices,
4697 and to make sure that time monotonically progresses on these
4698 systems, even if it is not always correct. To make use of
4699 this daemon, a new system user and group "systemd-timesync"
4700 needs to be created on installation of systemd.
4702 * The queue "seqnum" interface of libudev has been disabled, as
4703 it was generally incompatible with device namespacing as
4704 sequence numbers of devices go "missing" if the devices are
4705 part of a different namespace.
4707 * "systemctl list-timers" and "systemctl list-sockets" gained
4708 a --recursive switch for showing units of these types also
4709 for all local containers, similar in style to the already
4710 supported --recursive switch for "systemctl list-units".
4712 * A new RebootArgument= setting has been added for service
4713 units, which may be used to specify a kernel reboot argument
4714 to use when triggering reboots with StartLimitAction=.
4716 * A new FailureAction= setting has been added for service
4717 units which may be used to specify an operation to trigger
4718 when a service fails. This works similarly to
4719 StartLimitAction=, but unlike it, controls what is done
4720 immediately rather than only after several attempts to
4721 restart the service in question.
4723 * hostnamed got updated to also expose the kernel name,
4724 release, and version on the bus. This is useful for
4725 executing commands like hostnamectl with the -H switch.
4726 systemd-analyze makes use of this to properly display
4727 details when running non-locally.
4729 * The bootchart tool can now show cgroup information in the
4730 graphs it generates.
4732 * The CFS CPU quota cgroup attribute is now exposed for
4733 services. The new CPUQuota= switch has been added for this
4734 which takes a percentage value. Setting this will have the
4735 result that a service may never get more CPU time than the
4736 specified percentage, even if the machine is otherwise idle.
4738 * systemd-networkd learned IPIP and SIT tunnel support.
4740 * LSB init scripts exposing a dependency on $network will now
4741 get a dependency on network-online.target rather than simply
4742 network.target. This should bring LSB handling closer to
4743 what it was on SysV systems.
4745 * A new fsck.repair= kernel option has been added to control
4746 how fsck shall deal with unclean file systems at boot.
4748 * The (.ini) configuration file parser will now silently
4749 ignore sections whose name begins with "X-". This may be
4750 used to maintain application-specific extension sections in unit
4753 * machined gained a new API to query the IP addresses of
4754 registered containers. "machinectl status" has been updated
4755 to show these addresses in its output.
4757 * A new call sd_uid_get_display() has been added to the
4758 sd-login APIs for querying the "primary" session of a
4759 user. The "primary" session of the user is elected from the
4760 user's sessions and generally a graphical session is
4761 preferred over a text one.
4763 * A minimal systemd-resolved daemon has been added. It
4764 currently simply acts as a companion to systemd-networkd and
4765 manages resolv.conf based on per-interface DNS
4766 configuration, possibly supplied via DHCP. In the long run
4767 we hope to extend this into a local DNSSEC enabled DNS and
4770 * The systemd-networkd-wait-online tool is now enabled by
4771 default. It will delay network-online.target until a network
4772 connection has been configured. The tool primarily integrates
4773 with networkd, but will also make a best effort to make sense
4774 of network configuration performed in some other way.
4776 * Two new service options StartupCPUShares= and
4777 StartupBlockIOWeight= have been added that work similarly to
4778 CPUShares= and BlockIOWeight= however only apply during
4779 system startup. This is useful to prioritize certain services
4780 differently during bootup than during normal runtime.
4782 * hostnamed has been changed to prefer the statically
4783 configured hostname in /etc/hostname (unless set to
4784 'localhost' or empty) over any dynamic one supplied by
4785 dhcp. With this change, the rules for picking the hostname
4786 match more closely the rules of other configuration settings
4787 where the local administrator's configuration in /etc always
4788 overrides any other settings.
4790 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4791 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4792 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4793 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4794 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4795 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4796 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4797 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4798 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4799 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4800 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4801 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4802 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4803 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4804 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4805 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4808 — Beijing, 2014-05-28
4812 * When restoring the screen brightness at boot, stay away from
4813 the darkest setting or from the lowest 5% of the available
4814 range, depending on which is the larger value of both. This
4815 should effectively protect the user from rebooting into a
4816 black screen, should the brightness have been set to minimum
4819 * sd-login gained a new sd_machine_get_class() call to
4820 determine the class ("vm" or "container") of a machine
4821 registered with machined.
4823 * sd-login gained new calls
4824 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4825 to query the identity of the peer of a local AF_UNIX
4826 connection. They operate similarly to their sd_pid_get_xyz()
4829 * PID 1 will now maintain a system-wide system state engine
4830 with the states "starting", "running", "degraded",
4831 "maintenance", "stopping". These states are bound to system
4832 startup, normal runtime, runtime with at least one failed
4833 service, rescue/emergency mode and system shutdown. This
4834 state is shown in the "systemctl status" output when no unit
4835 name is passed. It is useful to determine system state, in
4836 particularly when doing so for many systems or containers at
4839 * A new command "list-machines" has been added to "systemctl"
4840 that lists all local OS containers and shows their system
4841 state (see above), if systemd runs inside of them.
4843 * systemctl gained a new "-r" switch to recursively enumerate
4844 units on all local containers, when used with the
4845 "list-unit" command (which is the default one that is
4846 executed when no parameters are specified).
4848 * The GPT automatic partition discovery logic will now honour
4849 two GPT partition flags: one may be set on a partition to
4850 cause it to be mounted read-only, and the other may be set
4851 on a partition to ignore it during automatic discovery.
4853 * Two new GPT type UUIDs have been added for automatic root
4854 partition discovery, for 32-bit and 64-bit ARM. This is not
4855 particularly useful for discovering the root directory on
4856 these architectures during bare-metal boots (since UEFI is
4857 not common there), but still very useful to allow booting of
4858 ARM disk images in nspawn with the -i option.
4860 * MAC addresses of interfaces created with nspawn's
4861 --network-interface= switch will now be generated from the
4862 machine name, and thus be stable between multiple invocations
4865 * logind will now automatically remove all IPC objects owned
4866 by a user if she or he fully logs out. This makes sure that
4867 users who are logged out cannot continue to consume IPC
4868 resources. This covers SysV memory, semaphores and message
4869 queues as well as POSIX shared memory and message
4870 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4871 limits. With this functionality, that is corrected. This may
4872 be turned off by using the RemoveIPC= switch of logind.conf.
4874 * The systemd-machine-id-setup and tmpfiles tools gained a
4875 --root= switch to operate on a specific root directory,
4878 * journald can now forward logged messages to the TTYs of all
4879 logged in users ("wall"). This is the default for all
4880 emergency messages now.
4882 * A new tool systemd-journal-remote has been added to stream
4883 journal log messages across the network.
4885 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4886 controller trees are mounted into it. Note that the
4887 directories mounted beneath it are not read-only. This is a
4888 security measure and is particularly useful because glibc
4889 actually includes a search logic to pick any tmpfs it can
4890 find to implement shm_open() if /dev/shm is not available
4891 (which it might very well be in namespaced setups).
4893 * machinectl gained a new "poweroff" command to cleanly power
4894 down a local OS container.
4896 * The PrivateDevices= unit file setting will now also drop the
4897 CAP_MKNOD capability from the capability bound set, and
4898 imply DevicePolicy=closed.
4900 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4901 comprehensively on all long-running systemd services where
4902 this is appropriate.
4904 * systemd-udevd will now run in a disassociated mount
4905 namespace. To mount directories from udev rules, make sure to
4906 pull in mount units via SYSTEMD_WANTS properties.
4908 * The kdbus support gained support for uploading policy into
4909 the kernel. sd-bus gained support for creating "monitoring"
4910 connections that can eavesdrop into all bus communication
4911 for debugging purposes.
4913 * Timestamps may now be specified in seconds since the UNIX
4914 epoch Jan 1st, 1970 by specifying "@" followed by the value
4917 * Native tcpwrap support in systemd has been removed. tcpwrap
4918 is old code, not really maintained anymore and has serious
4919 shortcomings, and better options such as firewalls
4920 exist. For setups that require tcpwrap usage, please
4921 consider invoking your socket-activated service via tcpd,
4922 like on traditional inetd.
4924 * A new system.conf configuration option
4925 DefaultTimerAccuracySec= has been added that controls the
4926 default AccuracySec= setting of .timer units.
4928 * Timer units gained a new WakeSystem= switch. If enabled,
4929 timers configured this way will cause the system to resume
4930 from system suspend (if the system supports that, which most
4933 * Timer units gained a new Persistent= switch. If enabled,
4934 timers configured this way will save to disk when they have
4935 been last triggered. This information is then used on next
4936 reboot to possible execute overdue timer events, that
4937 could not take place because the system was powered off.
4938 This enables simple anacron-like behaviour for timer units.
4940 * systemctl's "list-timers" will now also list the time a
4941 timer unit was last triggered in addition to the next time
4942 it will be triggered.
4944 * systemd-networkd will now assign predictable IPv4LL
4945 addresses to its local interfaces.
4947 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4948 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4949 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4950 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4951 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4952 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4953 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4954 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4957 — Berlin, 2014-03-25
4961 * A new unit file setting RestrictAddressFamilies= has been
4962 added to restrict which socket address families unit
4963 processes gain access to. This takes address family names
4964 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4965 attack surface of services via exotic protocol stacks. This
4966 is built on seccomp system call filters.
4968 * Two new unit file settings RuntimeDirectory= and
4969 RuntimeDirectoryMode= have been added that may be used to
4970 manage a per-daemon runtime directories below /run. This is
4971 an alternative for setting up directory permissions with
4972 tmpfiles snippets, and has the advantage that the runtime
4973 directory's lifetime is bound to the daemon runtime and that
4974 the daemon starts up with an empty directory each time. This
4975 is particularly useful when writing services that drop
4976 privileges using the User= or Group= setting.
4978 * The DeviceAllow= unit setting now supports globbing for
4979 matching against device group names.
4981 * The systemd configuration file system.conf gained new
4982 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4983 DefaultMemoryAccounting= to globally turn on/off accounting
4984 for specific resources (cgroups) for all units. These
4985 settings may still be overridden individually in each unit
4988 * systemd-gpt-auto-generator is now able to discover /srv and
4989 root partitions in addition to /home and swap partitions. It
4990 also supports LUKS-encrypted partitions now. With this in
4991 place, automatic discovery of partitions to mount following
4992 the Discoverable Partitions Specification
4993 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4994 is now a lot more complete. This allows booting without
4995 /etc/fstab and without root= on the kernel command line on
4996 systems prepared appropriately.
4998 * systemd-nspawn gained a new --image= switch which allows
4999 booting up disk images and Linux installations on any block
5000 device that follow the Discoverable Partitions Specification
5001 (see above). This means that installations made with
5002 appropriately updated installers may now be started and
5003 deployed using container managers, completely
5004 unmodified. (We hope that libvirt-lxc will add support for
5005 this feature soon, too.)
5007 * systemd-nspawn gained a new --network-macvlan= setting to
5008 set up a private macvlan interface for the
5009 container. Similarly, systemd-networkd gained a new
5010 Kind=macvlan setting in .netdev files.
5012 * systemd-networkd now supports configuring local addresses
5015 * A new tool systemd-network-wait-online has been added to
5016 synchronously wait for network connectivity using
5019 * The sd-bus.h bus API gained a new sd_bus_track object for
5020 tracking the life-cycle of bus peers. Note that sd-bus.h is
5021 still not a public API though (unless you specify
5022 --enable-kdbus on the configure command line, which however
5023 voids your warranty and you get no API stability guarantee).
5025 * The $XDG_RUNTIME_DIR runtime directories for each user are
5026 now individual tmpfs instances, which has the benefit of
5027 introducing separate pools for each user, with individual
5028 size limits, and thus making sure that unprivileged clients
5029 can no longer negatively impact the system or other users by
5030 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5031 RuntimeDirectorySize= has been introduced that allows
5032 controlling the default size limit for all users. It
5033 defaults to 10% of the available physical memory. This is no
5034 replacement for quotas on tmpfs though (which the kernel
5035 still does not support), as /dev/shm and /tmp are still
5036 shared resources used by both the system and unprivileged
5039 * logind will now automatically turn off automatic suspending
5040 on laptop lid close when more than one display is
5041 connected. This was previously expected to be implemented
5042 individually in desktop environments (such as GNOME),
5043 however has been added to logind now, in order to fix a
5044 boot-time race where a desktop environment might not have
5045 been started yet and thus not been able to take an inhibitor
5046 lock at the time where logind already suspends the system
5047 due to a closed lid.
5049 * logind will now wait at least 30s after each system
5050 suspend/resume cycle, and 3min after system boot before
5051 suspending the system due to a closed laptop lid. This
5052 should give USB docking stations and similar enough time to
5053 be probed and configured after system resume and boot in
5054 order to then act as suspend blocker.
5056 * systemd-run gained a new --property= setting which allows
5057 initialization of resource control properties (and others)
5058 for the created scope or service unit. Example: "systemd-run
5059 --property=BlockIOWeight=10 updatedb" may be used to run
5060 updatedb at a low block IO scheduling weight.
5062 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5063 now also work in --scope mode.
5065 * When systemd is compiled with kdbus support, basic support
5066 for enforced policies is now in place. (Note that enabling
5067 kdbus still voids your warranty and no API compatibility
5070 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5071 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5072 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5073 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5074 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5075 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5076 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5077 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5078 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5079 Zbigniew Jędrzejewski-Szmek
5081 — Berlin, 2014-03-12
5085 * systemd will now relabel /dev after loading the SMACK policy
5086 according to SMACK rules.
5088 * A new unit file option AppArmorProfile= has been added to
5089 set the AppArmor profile for the processes of a unit.
5091 * A new condition check ConditionArchitecture= has been added
5092 to conditionalize units based on the system architecture, as
5093 reported by uname()'s "machine" field.
5095 * systemd-networkd now supports matching on the system
5096 virtualization, architecture, kernel command line, host name
5099 * logind is now a lot more aggressive when suspending the
5100 machine due to a closed laptop lid. Instead of acting only
5101 on the lid close action, it will continuously watch the lid
5102 status and act on it. This is useful for laptops where the
5103 power button is on the outside of the chassis so that it can
5104 be reached without opening the lid (such as the Lenovo
5105 Yoga). On those machines, logind will now immediately
5106 re-suspend the machine if the power button has been
5107 accidentally pressed while the laptop was suspended and in a
5108 backpack or similar.
5110 * logind will now watch SW_DOCK switches and inhibit reaction
5111 to the lid switch if it is pressed. This means that logind
5112 will not suspend the machine anymore if the lid is closed
5113 and the system is docked, if the laptop supports SW_DOCK
5114 notifications via the input layer. Note that ACPI docking
5115 stations do not generate this currently. Also note that this
5116 logic is usually not fully sufficient and Desktop
5117 Environments should take a lid switch inhibitor lock when an
5118 external display is connected, as systemd will not watch
5121 * nspawn will now make use of the devices cgroup controller by
5122 default, and only permit creation of and access to the usual
5123 API device nodes like /dev/null or /dev/random, as well as
5124 access to (but not creation of) the pty devices.
5126 * We will now ship a default .network file for
5127 systemd-networkd that automatically configures DHCP for
5128 network interfaces created by nspawn's --network-veth or
5129 --network-bridge= switches.
5131 * systemd will now understand the usual M, K, G, T suffixes
5132 according to SI conventions (i.e. to the base 1000) when
5133 referring to throughput and hardware metrics. It will stay
5134 with IEC conventions (i.e. to the base 1024) for software
5135 metrics, according to what is customary according to
5136 Wikipedia. We explicitly document which base applies for
5137 each configuration option.
5139 * The DeviceAllow= setting in unit files now supports a syntax
5140 to whitelist an entire group of devices node majors at once,
5141 based on the /proc/devices listing. For example, with the
5142 string "char-pts", it is now possible to whitelist all
5143 current and future pseudo-TTYs at once.
5145 * sd-event learned a new "post" event source. Event sources of
5146 this type are triggered by the dispatching of any event
5147 source of a type that is not "post". This is useful for
5148 implementing clean-up and check event sources that are
5149 triggered by other work being done in the program.
5151 * systemd-networkd is no longer statically enabled, but uses
5152 the usual [Install] sections so that it can be
5153 enabled/disabled using systemctl. It still is enabled by
5156 * When creating a veth interface pair with systemd-nspawn, the
5157 host side will now be prefixed with "vb-" if
5158 --network-bridge= is used, and with "ve-" if --network-veth
5159 is used. This way, it is easy to distinguish these cases on
5160 the host, for example to apply different configuration to
5161 them with systemd-networkd.
5163 * The compatibility libraries for libsystemd-journal.so,
5164 libsystem-id128.so, libsystemd-login.so and
5165 libsystemd-daemon.so do not make use of IFUNC
5166 anymore. Instead, we now build libsystemd.so multiple times
5167 under these alternative names. This means that the footprint
5168 is drastically increased, but given that these are
5169 transitional compatibility libraries, this should not matter
5170 much. This change has been made necessary to support the ARM
5171 platform for these compatibility libraries, as the ARM
5172 toolchain is not really at the same level as the toolchain
5173 for other architectures like x86 and does not support
5174 IFUNC. Please make sure to use --enable-compat-libs only
5175 during a transitional period!
5177 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5178 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5179 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5180 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5181 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5182 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5183 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5184 Zbigniew Jędrzejewski-Szmek
5186 — Berlin, 2014-02-24
5190 * A new component "systemd-networkd" has been added that can
5191 be used to configure local network interfaces statically or
5192 via DHCP. It is capable of bringing up bridges, VLANs, and
5193 bonding. Currently, no hook-ups for interactive network
5194 configuration are provided. Use this for your initrd,
5195 container, embedded, or server setup if you need a simple,
5196 yet powerful, network configuration solution. This
5197 configuration subsystem is quite nifty, as it allows wildcard
5198 hotplug matching in interfaces. For example, with a single
5199 configuration snippet, you can configure that all Ethernet
5200 interfaces showing up are automatically added to a bridge,
5201 or similar. It supports link-sensing and more.
5203 * A new tool "systemd-socket-proxyd" has been added which can
5204 act as a bidirectional proxy for TCP sockets. This is
5205 useful for adding socket activation support to services that
5206 do not actually support socket activation, including virtual
5207 machines and the like.
5209 * Add a new tool to save/restore rfkill state on
5212 * Save/restore state of keyboard backlights in addition to
5213 display backlights on shutdown/boot.
5215 * udev learned a new SECLABEL{} construct to label device
5216 nodes with a specific security label when they appear. For
5217 now, only SECLABEL{selinux} is supported, but the syntax is
5218 prepared for additional security frameworks.
5220 * udev gained a new scheme to configure link-level attributes
5221 from files in /etc/systemd/network/*.link. These files can
5222 match against MAC address, device path, driver name and type,
5223 and will apply attributes like the naming policy, link speed,
5224 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5225 address assignment policy (randomized, ...).
5227 * The configuration of network interface naming rules for
5228 "permanent interface names" has changed: a new NamePolicy=
5229 setting in the [Link] section of .link files determines the
5230 priority of possible naming schemes (onboard, slot, MAC,
5231 path). The default value of this setting is determined by
5232 /usr/lib/net/links/99-default.link. Old
5233 80-net-name-slot.rules udev configuration file has been
5234 removed, so local configuration overriding this file should
5235 be adapted to override 99-default.link instead.
5237 * When the User= switch is used in a unit file, also
5238 initialize $SHELL= based on the user database entry.
5240 * systemd no longer depends on libdbus. All communication is
5241 now done with sd-bus, systemd's low-level bus library
5244 * kdbus support has been added to PID 1 itself. When kdbus is
5245 enabled, this causes PID 1 to set up the system bus and
5246 enable support for a new ".busname" unit type that
5247 encapsulates bus name activation on kdbus. It works a little
5248 bit like ".socket" units, except for bus names. A new
5249 generator has been added that converts classic dbus1 service
5250 activation files automatically into native systemd .busname
5253 * sd-bus: add a light-weight vtable implementation that allows
5254 defining objects on the bus with a simple static const
5255 vtable array of its methods, signals and properties.
5257 * systemd will not generate or install static dbus
5258 introspection data anymore to /usr/share/dbus-1/interfaces,
5259 as the precise format of these files is unclear, and
5260 nothing makes use of it.
5262 * A proxy daemon is now provided to proxy clients connecting
5263 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5264 compatibility with classic D-Bus.
5266 * A bus driver implementation has been added that supports the
5267 classic D-Bus bus driver calls on kdbus, also for
5268 compatibility purposes.
5270 * A new API "sd-event.h" has been added that implements a
5271 minimal event loop API built around epoll. It provides a
5272 couple of features that direct epoll usage is lacking:
5273 prioritization of events, scales to large numbers of timer
5274 events, per-event timer slack (accuracy), system-wide
5275 coalescing of timer events, exit handlers, watchdog
5276 supervision support using systemd's sd_notify() API, child
5279 * A new API "sd-rntl.h" has been added that provides an API
5280 around the route netlink interface of the kernel, similar in
5281 style to "sd-bus.h".
5283 * A new API "sd-dhcp-client.h" has been added that provides a
5284 small DHCPv4 client-side implementation. This is used by
5287 * There is a new kernel command line option
5288 "systemd.restore_state=0|1". When set to "0", none of the
5289 systemd tools will restore saved runtime state to hardware
5290 devices. More specifically, the rfkill and backlight states
5293 * The FsckPassNo= compatibility option in mount/service units
5294 has been removed. The fstab generator will now add the
5295 necessary dependencies automatically, and does not require
5296 PID1's support for that anymore.
5298 * journalctl gained a new switch, --list-boots, that lists
5299 recent boots with their times and boot IDs.
5301 * The various tools like systemctl, loginctl, timedatectl,
5302 busctl, systemd-run, ... have gained a new switch "-M" to
5303 connect to a specific, local OS container (as direct
5304 connection, without requiring SSH). This works on any
5305 container that is registered with machined, such as those
5306 created by libvirt-lxc or nspawn.
5308 * systemd-run and systemd-analyze also gained support for "-H"
5309 to connect to remote hosts via SSH. This is particularly
5310 useful for systemd-run because it enables queuing of jobs
5311 onto remote systems.
5313 * machinectl gained a new command "login" to open a getty
5314 login in any local container. This works with any container
5315 that is registered with machined (such as those created by
5316 libvirt-lxc or nspawn), and which runs systemd inside.
5318 * machinectl gained a new "reboot" command that may be used to
5319 trigger a reboot on a specific container that is registered
5320 with machined. This works on any container that runs an init
5321 system of some kind.
5323 * systemctl gained a new "list-timers" command to print a nice
5324 listing of installed timer units with the times they elapse
5327 * Alternative reboot() parameters may now be specified on the
5328 "systemctl reboot" command line and are passed to the
5329 reboot() system call.
5331 * systemctl gained a new --job-mode= switch to configure the
5332 mode to queue a job with. This is a more generic version of
5333 --fail, --irreversible, and --ignore-dependencies, which are
5334 still available but not advertised anymore.
5336 * /etc/systemd/system.conf gained new settings to configure
5337 various default timeouts of units, as well as the default
5338 start limit interval and burst. These may still be overridden
5341 * PID1 will now export on the bus profile data of the security
5342 policy upload process (such as the SELinux policy upload to
5345 * journald: when forwarding logs to the console, include
5346 timestamps (following the setting in
5347 /sys/module/printk/parameters/time).
5349 * OnCalendar= in timer units now understands the special
5350 strings "yearly" and "annually". (Both are equivalent)
5352 * The accuracy of timer units is now configurable with the new
5353 AccuracySec= setting. It defaults to 1min.
5355 * A new dependency type JoinsNamespaceOf= has been added that
5356 allows running two services within the same /tmp and network
5357 namespace, if PrivateNetwork= or PrivateTmp= are used.
5359 * A new command "cat" has been added to systemctl. It outputs
5360 the original unit file of a unit, and concatenates the
5361 contents of additional "drop-in" unit file snippets, so that
5362 the full configuration is shown.
5364 * systemctl now supports globbing on the various "list-xyz"
5365 commands, like "list-units" or "list-sockets", as well as on
5366 those commands which take multiple unit names.
5368 * journalctl's --unit= switch gained support for globbing.
5370 * All systemd daemons now make use of the watchdog logic so
5371 that systemd automatically notices when they hang.
5373 * If the $container_ttys environment variable is set,
5374 getty-generator will automatically spawn a getty for each
5375 listed tty. This is useful for container managers to request
5376 login gettys to be spawned on as many ttys as needed.
5378 * %h, %s, %U specifier support is not available anymore when
5379 used in unit files for PID 1. This is because NSS calls are
5380 not safe from PID 1. They stay available for --user
5381 instances of systemd, and as special case for the root user.
5383 * loginctl gained a new "--no-legend" switch to turn off output
5386 * The "sd-login.h" API gained three new calls:
5387 sd_session_is_remote(), sd_session_get_remote_user(),
5388 sd_session_get_remote_host() to query information about
5391 * The udev hardware database now also carries vendor/product
5392 information of SDIO devices.
5394 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5395 determine whether watchdog notifications are requested by
5398 * Socket-activated per-connection services now include a
5399 short description of the connection parameters in the
5402 * tmpfiles gained a new "--boot" option. When this is not used,
5403 only lines where the command character is not suffixed with
5404 "!" are executed. When this option is specified, those
5405 options are executed too. This partitions tmpfiles
5406 directives into those that can be safely executed at any
5407 time, and those which should be run only at boot (for
5408 example, a line that creates /run/nologin).
5410 * A new API "sd-resolve.h" has been added which provides a simple
5411 asynchronous wrapper around glibc NSS host name resolution
5412 calls, such as getaddrinfo(). In contrast to glibc's
5413 getaddrinfo_a(), it does not use signals. In contrast to most
5414 other asynchronous name resolution libraries, this one does
5415 not reimplement DNS, but reuses NSS, so that alternate
5416 host name resolution systems continue to work, such as mDNS,
5417 LDAP, etc. This API is based on libasyncns, but it has been
5418 cleaned up for inclusion in systemd.
5420 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5421 "sd-daemon.h" are no longer found in individual libraries
5422 libsystemd-journal.so, libsystemd-login.so,
5423 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5424 merged them into a single library, libsystemd.so, which
5425 provides all symbols. The reason for this is cyclic
5426 dependencies, as these libraries tend to use each other's
5427 symbols. So far, we have managed to workaround that by linking
5428 a copy of a good part of our code into each of these
5429 libraries again and again, which, however, makes certain
5430 things hard to do, like sharing static variables. Also, it
5431 substantially increases footprint. With this change, there
5432 is only one library for the basic APIs systemd
5433 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5434 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5435 library as well, however are subject to the --enable-kdbus
5436 switch (see below). Note that "sd-dhcp-client.h" is not part
5437 of this library (this is because it only consumes, never
5438 provides, services of/to other APIs). To make the transition
5439 easy from the separate libraries to the unified one, we
5440 provide the --enable-compat-libs compile-time switch which
5441 will generate stub libraries that are compatible with the
5442 old ones but redirect all calls to the new one.
5444 * All of the kdbus logic and the new APIs "sd-bus.h",
5445 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5446 and "sd-utf8.h" are compile-time optional via the
5447 "--enable-kdbus" switch, and they are not compiled in by
5448 default. To make use of kdbus, you have to explicitly enable
5449 the switch. Note however, that neither the kernel nor the
5450 userspace API for all of this is considered stable yet. We
5451 want to maintain the freedom to still change the APIs for
5452 now. By specifying this build-time switch, you acknowledge
5453 that you are aware of the instability of the current
5456 * Also, note that while kdbus is pretty much complete,
5457 it lacks one thing: proper policy support. This means you
5458 can build a fully working system with all features; however,
5459 it will be highly insecure. Policy support will be added in
5460 one of the next releases, at the same time that we will
5461 declare the APIs stable.
5463 * When the kernel command line argument "kdbus" is specified,
5464 systemd will automatically load the kdbus.ko kernel module. At
5465 this stage of development, it is only useful for testing kdbus
5466 and should not be used in production. Note: if "--enable-kdbus"
5467 is specified, and the kdbus.ko kernel module is available, and
5468 "kdbus" is added to the kernel command line, the entire system
5469 runs with kdbus instead of dbus-daemon, with the above mentioned
5470 problem of missing the system policy enforcement. Also a future
5471 version of kdbus.ko or a newer systemd will not be compatible with
5472 each other, and will unlikely be able to boot the machine if only
5473 one of them is updated.
5475 * systemctl gained a new "import-environment" command which
5476 uploads the caller's environment (or parts thereof) into the
5477 service manager so that it is inherited by services started
5478 by the manager. This is useful to upload variables like
5479 $DISPLAY into the user service manager.
5481 * A new PrivateDevices= switch has been added to service units
5482 which allows running a service with a namespaced /dev
5483 directory that does not contain any device nodes for
5484 physical devices. More specifically, it only includes devices
5485 such as /dev/null, /dev/urandom, and /dev/zero which are API
5488 * logind has been extended to support behaviour like VT
5489 switching on seats that do not support a VT. This makes
5490 multi-session available on seats that are not the first seat
5491 (seat0), and on systems where kernel support for VTs has
5492 been disabled at compile-time.
5494 * If a process holds a delay lock for system sleep or shutdown
5495 and fails to release it in time, we will now log its
5496 identity. This makes it easier to identify processes that
5497 cause slow suspends or power-offs.
5499 * When parsing /etc/crypttab, support for a new key-slot=
5500 option as supported by Debian is added. It allows indicating
5501 which LUKS slot to use on disk, speeding up key loading.
5503 * The sd_journal_sendv() API call has been checked and
5504 officially declared to be async-signal-safe so that it may
5505 be invoked from signal handlers for logging purposes.
5507 * Boot-time status output is now enabled automatically after a
5508 short timeout if boot does not progress, in order to give
5509 the user an indication what she or he is waiting for.
5511 * The boot-time output has been improved to show how much time
5512 remains until jobs expire.
5514 * The KillMode= switch in service units gained a new possible
5515 value "mixed". If set, and the unit is shut down, then the
5516 initial SIGTERM signal is sent only to the main daemon
5517 process, while the following SIGKILL signal is sent to
5518 all remaining processes of the service.
5520 * When a scope unit is registered, a new property "Controller"
5521 may be set. If set to a valid bus name, systemd will send a
5522 RequestStop() signal to this name when it would like to shut
5523 down the scope. This may be used to hook manager logic into
5524 the shutdown logic of scope units. Also, scope units may now
5525 be put in a special "abandoned" state, in which case the
5526 manager process which created them takes no further
5527 responsibilities for it.
5529 * When reading unit files, systemd will now verify
5530 the access mode of these files, and warn about certain
5531 suspicious combinations. This has been added to make it
5532 easier to track down packaging bugs where unit files are
5533 marked executable or world-writable.
5535 * systemd-nspawn gained a new "--setenv=" switch to set
5536 container-wide environment variables. The similar option in
5537 systemd-activate was renamed from "--environment=" to
5538 "--setenv=" for consistency.
5540 * systemd-nspawn has been updated to create a new kdbus domain
5541 for each container that is invoked, thus allowing each
5542 container to have its own set of system and user buses,
5543 independent of the host.
5545 * systemd-nspawn gained a new --drop-capability= switch to run
5546 the container with less capabilities than the default. Both
5547 --drop-capability= and --capability= now take the special
5548 string "all" for dropping or keeping all capabilities.
5550 * systemd-nspawn gained new switches for executing containers
5551 with specific SELinux labels set.
5553 * systemd-nspawn gained a new --quiet switch to not generate
5554 any additional output but the container's own console
5557 * systemd-nspawn gained a new --share-system switch to run a
5558 container without PID namespacing enabled.
5560 * systemd-nspawn gained a new --register= switch to control
5561 whether the container is registered with systemd-machined or
5562 not. This is useful for containers that do not run full
5563 OS images, but only specific apps.
5565 * systemd-nspawn gained a new --keep-unit which may be used
5566 when invoked as the only program from a service unit, and
5567 results in registration of the unit service itself in
5568 systemd-machined, instead of a newly opened scope unit.
5570 * systemd-nspawn gained a new --network-interface= switch for
5571 moving arbitrary interfaces to the container. The new
5572 --network-veth switch creates a virtual Ethernet connection
5573 between host and container. The new --network-bridge=
5574 switch then allows assigning the host side of this virtual
5575 Ethernet connection to a bridge device.
5577 * systemd-nspawn gained a new --personality= switch for
5578 setting the kernel personality for the container. This is
5579 useful when running a 32-bit container on a 64-bit host. A
5580 similar option Personality= is now also available for service
5583 * logind will now also track a "Desktop" identifier for each
5584 session which encodes the desktop environment of it. This is
5585 useful for desktop environments that want to identify
5586 multiple running sessions of itself easily.
5588 * A new SELinuxContext= setting for service units has been
5589 added that allows setting a specific SELinux execution
5590 context for a service.
5592 * Most systemd client tools will now honour $SYSTEMD_LESS for
5593 settings of the "less" pager. By default, these tools will
5594 override $LESS to allow certain operations to work, such as
5595 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5596 influence this logic.
5598 * systemd's "seccomp" hook-up has been changed to make use of
5599 the libseccomp library instead of using its own
5600 implementation. This has benefits for portability among
5603 * For usage together with SystemCallFilter=, a new
5604 SystemCallErrorNumber= setting has been introduced that
5605 allows configuration of a system error number to be returned
5606 on filtered system calls, instead of immediately killing the
5607 process. Also, SystemCallArchitectures= has been added to
5608 limit access to system calls of a particular architecture
5609 (in order to turn off support for unused secondary
5610 architectures). There is also a global
5611 SystemCallArchitectures= setting in system.conf now to turn
5612 off support for non-native system calls system-wide.
5614 * systemd requires a kernel with a working name_to_handle_at(),
5615 please see the kernel config requirements in the README file.
5617 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5618 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5619 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5620 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5621 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5622 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5623 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5624 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5625 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5626 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5627 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5628 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5629 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5630 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5631 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5632 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5633 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5634 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5635 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5636 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5637 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5638 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5639 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5640 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5642 — Berlin, 2014-02-20
5646 * logind has gained support for facilitating privileged input
5647 and drm device access for unprivileged clients. This work is
5648 useful to allow Wayland display servers (and similar
5649 programs, such as kmscon) to run under the user's ID and
5650 access input and drm devices which are normally
5651 protected. When this is used (and the kernel is new enough)
5652 logind will "mute" IO on the file descriptors passed to
5653 Wayland as long as it is in the background and "unmute" it
5654 if it returns into the foreground. This allows secure
5655 session switching without allowing background sessions to
5656 eavesdrop on input and display data. This also introduces
5657 session switching support if VT support is turned off in the
5658 kernel, and on seats that are not seat0.
5660 * A new kernel command line option luks.options= is understood
5661 now which allows specifying LUKS options for usage for LUKS
5662 encrypted partitions specified with luks.uuid=.
5664 * tmpfiles.d(5) snippets may now use specifier expansion in
5665 path names. More specifically %m, %b, %H, %v, are now
5666 replaced by the local machine id, boot id, hostname, and
5667 kernel version number.
5669 * A new tmpfiles.d(5) command "m" has been introduced which
5670 may be used to change the owner/group/access mode of a file
5671 or directory if it exists, but do nothing if it does not.
5673 * This release removes high-level support for the
5674 MemorySoftLimit= cgroup setting. The underlying kernel
5675 cgroup attribute memory.soft_limit= is currently badly
5676 designed and likely to be removed from the kernel API in its
5677 current form, hence we should not expose it for now.
5679 * The memory.use_hierarchy cgroup attribute is now enabled for
5680 all cgroups systemd creates in the memory cgroup
5681 hierarchy. This option is likely to be come the built-in
5682 default in the kernel anyway, and the non-hierarchical mode
5683 never made much sense in the intrinsically hierarchical
5686 * A new field _SYSTEMD_SLICE= is logged along with all journal
5687 messages containing the slice a message was generated
5688 from. This is useful to allow easy per-customer filtering of
5689 logs among other things.
5691 * systemd-journald will no longer adjust the group of journal
5692 files it creates to the "systemd-journal" group. Instead we
5693 rely on the journal directory to be owned by the
5694 "systemd-journal" group, and its setgid bit set, so that the
5695 kernel file system layer will automatically enforce that
5696 journal files inherit this group assignment. The reason for
5697 this change is that we cannot allow NSS look-ups from
5698 journald which would be necessary to resolve
5699 "systemd-journal" to a numeric GID, because this might
5700 create deadlocks if NSS involves synchronous queries to
5701 other daemons (such as nscd, or sssd) which in turn are
5702 logging clients of journald and might block on it, which
5703 would then dead lock. A tmpfiles.d(5) snippet included in
5704 systemd will make sure the setgid bit and group are
5705 properly set on the journal directory if it exists on every
5706 boot. However, we recommend adjusting it manually after
5707 upgrades too (or from RPM scriptlets), so that the change is
5708 not delayed until next reboot.
5710 * Backlight and random seed files in /var/lib/ have moved into
5711 the /var/lib/systemd/ directory, in order to centralize all
5712 systemd generated files in one directory.
5714 * Boot time performance measurements (as displayed by
5715 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5716 performance information if that's available to determine how
5717 much time BIOS and boot loader initialization required. With
5718 a sufficiently new BIOS you hence no longer need to boot
5719 with Gummiboot to get access to such information.
5721 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5722 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5723 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5724 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5725 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5726 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5727 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5729 — Berlin, 2013-10-02
5733 * The Restart= option for services now understands a new
5734 on-watchdog setting, which will restart the service
5735 automatically if the service stops sending out watchdog keep
5736 alive messages (as configured with WatchdogSec=).
5738 * The getty generator (which is responsible for bringing up a
5739 getty on configured serial consoles) will no longer only
5740 start a getty on the primary kernel console but on all
5741 others, too. This makes the order in which console= is
5742 specified on the kernel command line less important.
5744 * libsystemd-logind gained a new sd_session_get_vt() call to
5745 retrieve the VT number of a session.
5747 * If the option "tries=0" is set for an entry of /etc/crypttab
5748 its passphrase is queried indefinitely instead of any
5749 maximum number of tries.
5751 * If a service with a configure PID file terminates its PID
5752 file will now be removed automatically if it still exists
5753 afterwards. This should put an end to stale PID files.
5755 * systemd-run will now also take relative binary path names
5756 for execution and no longer insists on absolute paths.
5758 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5759 paths that are optionally prefixed with "-" to indicate that
5760 it should not be considered a failure if they do not exist.
5762 * journalctl -o (and similar commands) now understands a new
5763 output mode "short-precise", it is similar to "short" but
5764 shows timestamps with usec accuracy.
5766 * The option "discard" (as known from Debian) is now
5767 synonymous to "allow-discards" in /etc/crypttab. In fact,
5768 "discard" is preferred now (since it is easier to remember
5771 * Some licensing clean-ups were made, so that more code is now
5772 LGPL-2.1 licensed than before.
5774 * A minimal tool to save/restore the display backlight
5775 brightness across reboots has been added. It will store the
5776 backlight setting as late as possible at shutdown, and
5777 restore it as early as possible during reboot.
5779 * A logic to automatically discover and enable home and swap
5780 partitions on GPT disks has been added. With this in place
5781 /etc/fstab becomes optional for many setups as systemd can
5782 discover certain partitions located on the root disk
5783 automatically. Home partitions are recognized under their
5784 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5785 partitions are recognized under their GPT type ID
5786 0657fd6da4ab43c484e50933c84b4f4f.
5788 * systemd will no longer pass any environment from the kernel
5789 or initrd to system services. If you want to set an
5790 environment for all services, do so via the kernel command
5791 line systemd.setenv= assignment.
5793 * The systemd-sysctl tool no longer natively reads the file
5794 /etc/sysctl.conf. If desired, the file should be symlinked
5795 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5796 legacy support by a symlink rather than built-in code, it
5797 also makes the otherwise hidden order of application of the
5798 different files visible. (Note that this partly reverts to a
5799 pre-198 application order of sysctl knobs!)
5801 * The "systemctl set-log-level" and "systemctl dump" commands
5802 have been moved to systemd-analyze.
5804 * systemd-run learned the new --remain-after-exit switch,
5805 which causes the scope unit not to be cleaned up
5806 automatically after the process terminated.
5808 * tmpfiles learned a new --exclude-prefix= switch to exclude
5809 certain paths from operation.
5811 * journald will now automatically flush all messages to disk
5812 as soon as a message at the log level CRIT, ALERT or EMERG
5815 Contributions from: Andrew Cook, Brandon Philips, Christian
5816 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5817 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5818 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5819 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5820 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5821 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5822 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5823 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5824 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5825 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5826 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5827 William Giokas, Zbigniew Jędrzejewski-Szmek
5829 — Berlin, 2013-09-13
5833 * The documentation has been updated to cover the various new
5834 concepts introduced with 205.
5836 * Unit files now understand the new %v specifier which
5837 resolves to the kernel version string as returned by "uname
5840 * systemctl now supports filtering the unit list output by
5841 load state, active state and sub state, using the new
5844 * "systemctl status" will now show the results of the
5845 condition checks (like ConditionPathExists= and similar) of
5846 the last start attempts of the unit. They are also logged to
5849 * "journalctl -b" may now be used to look for boot output of a
5850 specific boot. Try "journalctl -b -1" for the previous boot,
5851 but the syntax is substantially more powerful.
5853 * "journalctl --show-cursor" has been added which prints the
5854 cursor string the last shown log line. This may then be used
5855 with the new "journalctl --after-cursor=" switch to continue
5856 browsing logs from that point on.
5858 * "journalctl --force" may now be used to force regeneration
5861 * Creation of "dead" device nodes has been moved from udev
5862 into kmod and tmpfiles. Previously, udev would read the kmod
5863 databases to pre-generate dead device nodes based on meta
5864 information contained in kernel modules, so that these would
5865 be auto-loaded on access rather then at boot. As this
5866 does not really have much to do with the exposing actual
5867 kernel devices to userspace this has always been slightly
5868 alien in the udev codebase. Following the new scheme kmod
5869 will now generate a runtime snippet for tmpfiles from the
5870 module meta information and it now is tmpfiles' job to the
5871 create the nodes. This also allows overriding access and
5872 other parameters for the nodes using the usual tmpfiles
5873 facilities. As side effect this allows us to remove the
5874 CAP_SYS_MKNOD capability bit from udevd entirely.
5876 * logind's device ACLs may now be applied to these "dead"
5877 devices nodes too, thus finally allowing managed access to
5878 devices such as /dev/snd/sequencer without loading the
5879 backing module right-away.
5881 * A new RPM macro has been added that may be used to apply
5882 tmpfiles configuration during package installation.
5884 * systemd-detect-virt and ConditionVirtualization= now can
5885 detect User-Mode-Linux machines (UML).
5887 * journald will now implicitly log the effective capabilities
5888 set of processes in the message metadata.
5890 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5892 * The initrd interface has been simplified (more specifically,
5893 support for passing performance data via environment
5894 variables and fsck results via files in /run has been
5895 removed). These features were non-essential, and are
5896 nowadays available in a much nicer way by having systemd in
5897 the initrd serialize its state and have the hosts systemd
5898 deserialize it again.
5900 * The udev "keymap" data files and tools to apply keyboard
5901 specific mappings of scan to key codes, and force-release
5902 scan code lists have been entirely replaced by a udev
5903 "keyboard" builtin and a hwdb data file.
5905 * systemd will now honour the kernel's "quiet" command line
5906 argument also during late shutdown, resulting in a
5907 completely silent shutdown when used.
5909 * There's now an option to control the SO_REUSEPORT socket
5910 option in .socket units.
5912 * Instance units will now automatically get a per-template
5913 subslice of system.slice unless something else is explicitly
5914 configured. For example, instances of sshd@.service will now
5915 implicitly be placed in system-sshd.slice rather than
5916 system.slice as before.
5918 * Test coverage support may now be enabled at build time.
5920 Contributions from: Dave Reisner, Frederic Crozat, Harald
5921 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5922 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5923 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5924 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5925 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5926 Giokas, Zbigniew Jędrzejewski-Szmek
5928 — Berlin, 2013-07-23
5932 * Two new unit types have been introduced:
5934 Scope units are very similar to service units, however, are
5935 created out of pre-existing processes — instead of PID 1
5936 forking off the processes. By using scope units it is
5937 possible for system services and applications to group their
5938 own child processes (worker processes) in a powerful way
5939 which then maybe used to organize them, or kill them
5940 together, or apply resource limits on them.
5942 Slice units may be used to partition system resources in an
5943 hierarchical fashion and then assign other units to them. By
5944 default there are now three slices: system.slice (for all
5945 system services), user.slice (for all user sessions),
5946 machine.slice (for VMs and containers).
5948 Slices and scopes have been introduced primarily in
5949 context of the work to move cgroup handling to a
5950 single-writer scheme, where only PID 1
5951 creates/removes/manages cgroups.
5953 * There's a new concept of "transient" units. In contrast to
5954 normal units these units are created via an API at runtime,
5955 not from configuration from disk. More specifically this
5956 means it is now possible to run arbitrary programs as
5957 independent services, with all execution parameters passed
5958 in via bus APIs rather than read from disk. Transient units
5959 make systemd substantially more dynamic then it ever was,
5960 and useful as a general batch manager.
5962 * logind has been updated to make use of scope and slice units
5963 for managing user sessions. As a user logs in he will get
5964 his own private slice unit, to which all sessions are added
5965 as scope units. We also added support for automatically
5966 adding an instance of user@.service for the user into the
5967 slice. Effectively logind will no longer create cgroup
5968 hierarchies on its own now, it will defer entirely to PID 1
5969 for this by means of scope, service and slice units. Since
5970 user sessions this way become entities managed by PID 1
5971 the output of "systemctl" is now a lot more comprehensive.
5973 * A new mini-daemon "systemd-machined" has been added which
5974 may be used by virtualization managers to register local
5975 VMs/containers. nspawn has been updated accordingly, and
5976 libvirt will be updated shortly. machined will collect a bit
5977 of meta information about the VMs/containers, and assign
5978 them their own scope unit (see above). The collected
5979 meta-data is then made available via the "machinectl" tool,
5980 and exposed in "ps" and similar tools. machined/machinectl
5981 is compile-time optional.
5983 * As discussed earlier, the low-level cgroup configuration
5984 options ControlGroup=, ControlGroupModify=,
5985 ControlGroupPersistent=, ControlGroupAttribute= have been
5986 removed. Please use high-level attribute settings instead as
5987 well as slice units.
5989 * A new bus call SetUnitProperties() has been added to alter
5990 various runtime parameters of a unit. This is primarily
5991 useful to alter cgroup parameters dynamically in a nice way,
5992 but will be extended later on to make more properties
5993 modifiable at runtime. systemctl gained a new set-properties
5994 command that wraps this call.
5996 * A new tool "systemd-run" has been added which can be used to
5997 run arbitrary command lines as transient services or scopes,
5998 while configuring a number of settings via the command
5999 line. This tool is currently very basic, however already
6000 very useful. We plan to extend this tool to even allow
6001 queuing of execution jobs with time triggers from the
6002 command line, similar in fashion to "at".
6004 * nspawn will now inform the user explicitly that kernels with
6005 audit enabled break containers, and suggest the user to turn
6008 * Support for detecting the IMA and AppArmor security
6009 frameworks with ConditionSecurity= has been added.
6011 * journalctl gained a new "-k" switch for showing only kernel
6012 messages, mimicking dmesg output; in addition to "--user"
6013 and "--system" switches for showing only user's own logs
6016 * systemd-delta can now show information about drop-in
6017 snippets extending unit files.
6019 * libsystemd-bus has been substantially updated but is still
6020 not available as public API.
6022 * systemd will now look for the "debug" argument on the kernel
6023 command line and enable debug logging, similar to what
6024 "systemd.log_level=debug" already did before.
6026 * "systemctl set-default", "systemctl get-default" has been
6027 added to configure the default.target symlink, which
6028 controls what to boot into by default.
6030 * "systemctl set-log-level" has been added as a convenient
6031 way to raise and lower systemd logging threshold.
6033 * "systemd-analyze plot" will now show the time the various
6034 generators needed for execution, as well as information
6035 about the unit file loading.
6037 * libsystemd-journal gained a new sd_journal_open_files() call
6038 for opening specific journal files. journactl also gained a
6039 new switch to expose this new functionality. Previously we
6040 only supported opening all files from a directory, or all
6041 files from the system, as opening individual files only is
6042 racy due to journal file rotation.
6044 * systemd gained the new DefaultEnvironment= setting in
6045 /etc/systemd/system.conf to set environment variables for
6048 * If a privileged process logs a journal message with the
6049 OBJECT_PID= field set, then journald will automatically
6050 augment this with additional OBJECT_UID=, OBJECT_GID=,
6051 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6052 system services want to log events about specific client
6053 processes. journactl/systemctl has been updated to make use
6054 of this information if all log messages regarding a specific
6057 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6058 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6059 Reisner, David Coppa, David King, David Strauss, Eelco
6060 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6061 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6062 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6063 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6064 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6065 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6066 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6067 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6068 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6069 Łukasz Stelmach, 장동준
6073 * The Python bindings gained some minimal support for the APIs
6074 exposed by libsystemd-logind.
6076 * ConditionSecurity= gained support for detecting SMACK. Since
6077 this condition already supports SELinux and AppArmor we only
6078 miss IMA for this. Patches welcome!
6080 Contributions from: Karol Lewandowski, Lennart Poettering,
6081 Zbigniew Jędrzejewski-Szmek
6085 * systemd-nspawn will now create /etc/resolv.conf if
6086 necessary, before bind-mounting the host's file onto it.
6088 * systemd-nspawn will now store meta information about a
6089 container on the container's cgroup as extended attribute
6090 fields, including the root directory.
6092 * The cgroup hierarchy has been reworked in many ways. All
6093 objects any of the components systemd creates in the cgroup
6094 tree are now suffixed. More specifically, user sessions are
6095 now placed in cgroups suffixed with ".session", users in
6096 cgroups suffixed with ".user", and nspawn containers in
6097 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6098 names are now escaped in a simple scheme to avoid collision
6099 of userspace object names with kernel filenames. This work
6100 is preparation for making these objects relocatable in the
6101 cgroup tree, in order to allow easy resource partitioning of
6102 these objects without causing naming conflicts.
6104 * systemctl list-dependencies gained the new switches
6105 --plain, --reverse, --after and --before.
6107 * systemd-inhibit now shows the process name of processes that
6108 have taken an inhibitor lock.
6110 * nss-myhostname will now also resolve "localhost"
6111 implicitly. This makes /etc/hosts an optional file and
6112 nicely handles that on IPv6 ::1 maps to both "localhost" and
6115 * libsystemd-logind.so gained a new call
6116 sd_get_machine_names() to enumerate running containers and
6117 VMs (currently only supported by very new libvirt and
6118 nspawn). sd_login_monitor can now be used to watch
6119 VMs/containers coming and going.
6121 * .include is not allowed recursively anymore, and only in
6122 unit files. Usually it is better to use drop-in snippets in
6123 .d/*.conf anyway, as introduced with systemd 198.
6125 * systemd-analyze gained a new "critical-chain" command that
6126 determines the slowest chain of units run during system
6127 boot-up. It is very useful for tracking down where
6128 optimizing boot time is the most beneficial.
6130 * systemd will no longer allow manipulating service paths in
6131 the name=systemd:/system cgroup tree using ControlGroup= in
6132 units. (But is still fine with it in all other dirs.)
6134 * There's a new systemd-nspawn@.service service file that may
6135 be used to easily run nspawn containers as system
6136 services. With the container's root directory in
6137 /var/lib/container/foobar it is now sufficient to run
6138 "systemctl start systemd-nspawn@foobar.service" to boot it.
6140 * systemd-cgls gained a new parameter "--machine" to list only
6141 the processes within a certain container.
6143 * ConditionSecurity= now can check for "apparmor". We still
6144 are lacking checks for SMACK and IMA for this condition
6145 check though. Patches welcome!
6147 * A new configuration file /etc/systemd/sleep.conf has been
6148 added that may be used to configure which kernel operation
6149 systemd is supposed to execute when "suspend", "hibernate"
6150 or "hybrid-sleep" is requested. This makes the new kernel
6151 "freeze" state accessible to the user.
6153 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6154 the passed argument if applicable.
6156 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6157 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6158 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6159 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6160 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6161 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6162 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6167 * The output of 'systemctl list-jobs' got some polishing. The
6168 '--type=' argument may now be passed more than once. A new
6169 command 'systemctl list-sockets' has been added which shows
6170 a list of kernel sockets systemd is listening on with the
6171 socket units they belong to, plus the units these socket
6174 * The experimental libsystemd-bus library got substantial
6175 updates to work in conjunction with the (also experimental)
6176 kdbus kernel project. It works well enough to exchange
6177 messages with some sophistication. Note that kdbus is not
6178 ready yet, and the library is mostly an elaborate test case
6179 for now, and not installable.
6181 * systemd gained a new unit 'systemd-static-nodes.service'
6182 that generates static device nodes earlier during boot, and
6183 can run in conjunction with udev.
6185 * libsystemd-login gained a new call sd_pid_get_user_unit()
6186 to retrieve the user systemd unit a process is running
6187 in. This is useful for systems where systemd is used as
6190 * systemd-nspawn now places all containers in the new /machine
6191 top-level cgroup directory in the name=systemd
6192 hierarchy. libvirt will soon do the same, so that we get a
6193 uniform separation of /system, /user and /machine for system
6194 services, user processes and containers/virtual
6195 machines. This new cgroup hierarchy is also useful to stick
6196 stable names to specific container instances, which can be
6197 recognized later this way (this name may be controlled
6198 via systemd-nspawn's new -M switch). libsystemd-login also
6199 gained a new call sd_pid_get_machine_name() to retrieve the
6200 name of the container/VM a specific process belongs to.
6202 * bootchart can now store its data in the journal.
6204 * libsystemd-journal gained a new call
6205 sd_journal_add_conjunction() for AND expressions to the
6206 matching logic. This can be used to express more complex
6207 logical expressions.
6209 * journactl can now take multiple --unit= and --user-unit=
6212 * The cryptsetup logic now understands the "luks.key=" kernel
6213 command line switch for specifying a file to read the
6214 decryption key from. Also, if a configured key file is not
6215 found the tool will now automatically fall back to prompting
6218 * Python systemd.journal module was updated to wrap recently
6219 added functions from libsystemd-journal. The interface was
6220 changed to bring the low level interface in s.j._Reader
6221 closer to the C API, and the high level interface in
6222 s.j.Reader was updated to wrap and convert all data about
6225 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6226 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6227 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6228 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6229 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6230 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6234 * journalctl --update-catalog now understands a new --root=
6235 option to operate on catalogs found in a different root
6238 * During shutdown after systemd has terminated all running
6239 services a final killing loop kills all remaining left-over
6240 processes. We will now print the name of these processes
6241 when we send SIGKILL to them, since this usually indicates a
6244 * If /etc/crypttab refers to password files stored on
6245 configured mount points automatic dependencies will now be
6246 generated to ensure the specific mount is established first
6247 before the key file is attempted to be read.
6249 * 'systemctl status' will now show information about the
6250 network sockets a socket unit is listening on.
6252 * 'systemctl status' will also shown information about any
6253 drop-in configuration file for units. (Drop-In configuration
6254 files in this context are files such as
6255 /etc/systemd/systemd/foobar.service.d/*.conf)
6257 * systemd-cgtop now optionally shows summed up CPU times of
6258 cgroups. Press '%' while running cgtop to switch between
6259 percentage and absolute mode. This is useful to determine
6260 which cgroups use up the most CPU time over the entire
6261 runtime of the system. systemd-cgtop has also been updated
6262 to be 'pipeable' for processing with further shell tools.
6264 * 'hostnamectl set-hostname' will now allow setting of FQDN
6267 * The formatting and parsing of time span values has been
6268 changed. The parser now understands fractional expressions
6269 such as "5.5h". The formatter will now output fractional
6270 expressions for all time spans under 1min, i.e. "5.123456s"
6271 rather than "5s 123ms 456us". For time spans under 1s
6272 millisecond values are shown, for those under 1ms
6273 microsecond values are shown. This should greatly improve
6274 all time-related output of systemd.
6276 * libsystemd-login and libsystemd-journal gained new
6277 functions for querying the poll() events mask and poll()
6278 timeout value for integration into arbitrary event
6281 * localectl gained the ability to list available X11 keymaps
6282 (models, layouts, variants, options).
6284 * 'systemd-analyze dot' gained the ability to filter for
6285 specific units via shell-style globs, to create smaller,
6286 more useful graphs. I.e. it is now possible to create simple
6287 graphs of all the dependencies between only target units, or
6288 of all units that Avahi has dependencies with.
6290 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6291 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6292 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6293 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6294 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6295 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6296 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6300 * The boot-time readahead implementation for rotating media
6301 will now read the read-ahead data in multiple passes which
6302 consist of all read requests made in equidistant time
6303 intervals. This means instead of strictly reading read-ahead
6304 data in its physical order on disk we now try to find a
6305 middle ground between physical and access time order.
6307 * /etc/os-release files gained a new BUILD_ID= field for usage
6308 on operating systems that provide continuous builds of OS
6311 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6312 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6313 William Douglas, Zbigniew Jędrzejewski-Szmek
6317 * systemd-python gained an API exposing libsystemd-daemon.
6319 * The SMACK setup logic gained support for uploading CIPSO
6322 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6323 ReadOnlyDirectories= and InaccessibleDirectories= has
6324 changed. The private /tmp and /var/tmp directories are now
6325 shared by all processes of a service (which means
6326 ExecStartPre= may now leave data in /tmp that ExecStart= of
6327 the same service can still access). When a service is
6328 stopped its temporary directories are immediately deleted
6329 (normal clean-up with tmpfiles is still done in addition to
6332 * By default, systemd will now set a couple of sysctl
6333 variables in the kernel: the safe sysrq options are turned
6334 on, IP route verification is turned on, and source routing
6335 disabled. The recently added hardlink and softlink
6336 protection of the kernel is turned on. These settings should
6337 be reasonably safe, and good defaults for all new systems.
6339 * The predictable network naming logic may now be turned off
6340 with a new kernel command line switch: net.ifnames=0.
6342 * A new libsystemd-bus module has been added that implements a
6343 pretty complete D-Bus client library. For details see:
6345 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6347 * journald will now explicitly flush the journal files to disk
6348 at the latest 5min after each write. The file will then also
6349 be marked offline until the next write. This should increase
6350 reliability in case of a crash. The synchronization delay
6351 can be configured via SyncIntervalSec= in journald.conf.
6353 * There's a new remote-fs-setup.target unit that can be used
6354 to pull in specific services when at least one remote file
6355 system is to be mounted.
6357 * There are new targets timers.target and paths.target as
6358 canonical targets to pull user timer and path units in
6359 from. This complements sockets.target with a similar
6360 purpose for socket units.
6362 * libudev gained a new call udev_device_set_attribute_value()
6363 to set sysfs attributes of a device.
6365 * The udev daemon now sets the default number of worker
6366 processes executed in parallel based on the number of available
6367 CPUs instead of the amount of available RAM. This is supposed
6368 to provide a more reliable default and limit a too aggressive
6369 parallelism for setups with 1000s of devices connected.
6371 Contributions from: Auke Kok, Colin Walters, Cristian
6372 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6373 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6374 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6375 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6376 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6377 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6378 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6379 Zbigniew Jędrzejewski-Szmek
6383 * Configuration of unit files may now be extended via drop-in
6384 files without having to edit/override the unit files
6385 themselves. More specifically, if the administrator wants to
6386 change one value for a service file foobar.service he can
6387 now do so by dropping in a configuration snippet into
6388 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6389 will load all these snippets and apply them on top of the
6390 main unit configuration file, possibly extending or
6391 overriding its settings. Using these drop-in snippets is
6392 generally nicer than the two earlier options for changing
6393 unit files locally: copying the files from
6394 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6395 them there; or creating a new file in /etc/systemd/system/
6396 that incorporates the original one via ".include". Drop-in
6397 snippets into these .d/ directories can be placed in any
6398 directory systemd looks for units in, and the usual
6399 overriding semantics between /usr/lib, /etc and /run apply
6402 * Most unit file settings which take lists of items can now be
6403 reset by assigning the empty string to them. For example,
6404 normally, settings such as Environment=FOO=BAR append a new
6405 environment variable assignment to the environment block,
6406 each time they are used. By assigning Environment= the empty
6407 string the environment block can be reset to empty. This is
6408 particularly useful with the .d/*.conf drop-in snippets
6409 mentioned above, since this adds the ability to reset list
6410 settings from vendor unit files via these drop-ins.
6412 * systemctl gained a new "list-dependencies" command for
6413 listing the dependencies of a unit recursively.
6415 * Inhibitors are now honored and listed by "systemctl
6416 suspend", "systemctl poweroff" (and similar) too, not only
6417 GNOME. These commands will also list active sessions by
6420 * Resource limits (as exposed by the various control group
6421 controllers) can now be controlled dynamically at runtime
6422 for all units. More specifically, you can now use a command
6423 like "systemctl set-cgroup-attr foobar.service cpu.shares
6424 2000" to alter the CPU shares a specific service gets. These
6425 settings are stored persistently on disk, and thus allow the
6426 administrator to easily adjust the resource usage of
6427 services with a few simple commands. This dynamic resource
6428 management logic is also available to other programs via the
6429 bus. Almost any kernel cgroup attribute and controller is
6432 * systemd-vconsole-setup will now copy all font settings to
6433 all allocated VTs, where it previously applied them only to
6436 * libsystemd-login gained the new sd_session_get_tty() API
6439 * This release drops support for a few legacy or
6440 distribution-specific LSB facility names when parsing init
6441 scripts: $x-display-manager, $mail-transfer-agent,
6442 $mail-transport-agent, $mail-transfer-agent, $smtp,
6443 $null. Also, the mail-transfer-agent.target unit backing
6444 this has been removed. Distributions which want to retain
6445 compatibility with this should carry the burden for
6446 supporting this themselves and patch support for these back
6447 in, if they really need to. Also, the facilities $syslog and
6448 $local_fs are now ignored, since systemd does not support
6449 early-boot LSB init scripts anymore, and these facilities
6450 are implied anyway for normal services. syslog.target has
6453 * There are new bus calls on PID1's Manager object for
6454 cancelling jobs, and removing snapshot units. Previously,
6455 both calls were only available on the Job and Snapshot
6458 * systemd-journal-gatewayd gained SSL support.
6460 * The various "environment" files, such as /etc/locale.conf
6461 now support continuation lines with a backslash ("\") as
6462 last character in the line, similarly in style (but different)
6463 to how this is supported in shells.
6465 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6466 now implicitly appended to every log entry logged. systemctl
6467 has been updated to filter by this field when operating on a
6468 user systemd instance.
6470 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6471 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6472 the container. This makes it easier to boot unmodified
6473 Fedora systems in a container, which however still requires
6474 audit=0 to be passed on the kernel command line. Auditing in
6475 kernel and userspace is unfortunately still too broken in
6476 context of containers, hence we recommend compiling it out
6477 of the kernel or using audit=0. Hopefully this will be fixed
6478 one day for good in the kernel.
6480 * nspawn gained the new --bind= and --bind-ro= parameters to
6481 bind mount specific directories from the host into the
6484 * nspawn will now mount its own devpts file system instance
6485 into the container, in order not to leak pty devices from
6486 the host into the container.
6488 * systemd will now read the firmware boot time performance
6489 information from the EFI variables, if the used boot loader
6490 supports this, and takes it into account for boot performance
6491 analysis via "systemd-analyze". This is currently supported
6492 only in conjunction with Gummiboot, but could be supported
6493 by other boot loaders too. For details see:
6495 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6497 * A new generator has been added that automatically mounts the
6498 EFI System Partition (ESP) to /boot, if that directory
6499 exists, is empty, and no other file system has been
6500 configured to be mounted there.
6502 * logind will now send out PrepareForSleep(false) out
6503 unconditionally, after coming back from suspend. This may be
6504 used by applications as asynchronous notification for
6505 system resume events.
6507 * "systemctl unlock-sessions" has been added, that allows
6508 unlocking the screens of all user sessions at once, similar
6509 to how "systemctl lock-sessions" already locked all users
6510 sessions. This is backed by a new D-Bus call UnlockSessions().
6512 * "loginctl seat-status" will now show the master device of a
6513 seat. (i.e. the device of a seat that needs to be around for
6514 the seat to be considered available, usually the graphics
6517 * tmpfiles gained a new "X" line type, that allows
6518 configuration of files and directories (with wildcards) that
6519 shall be excluded from automatic cleanup ("aging").
6521 * udev default rules set the device node permissions now only
6522 at "add" events, and do not change them any longer with a
6523 later "change" event.
6525 * The log messages for lid events and power/sleep keypresses
6526 now carry a message ID.
6528 * We now have a substantially larger unit test suite, but this
6529 continues to be work in progress.
6531 * udevadm hwdb gained a new --root= parameter to change the
6532 root directory to operate relative to.
6534 * logind will now issue a background sync() request to the kernel
6535 early at shutdown, so that dirty buffers are flushed to disk early
6536 instead of at the last moment, in order to optimize shutdown
6539 * A new bootctl tool has been added that is an interface for
6540 certain boot loader operations. This is currently a preview
6541 and is likely to be extended into a small mechanism daemon
6542 like timedated, localed, hostnamed, and can be used by
6543 graphical UIs to enumerate available boot options, and
6544 request boot into firmware operations.
6546 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6547 the rest of the package. It also has been updated to work
6548 correctly in initrds.
6550 * Policykit previously has been runtime optional, and is now
6551 also compile time optional via a configure switch.
6553 * systemd-analyze has been reimplemented in C. Also "systemctl
6554 dot" has moved into systemd-analyze.
6556 * "systemctl status" with no further parameters will now print
6557 the status of all active or failed units.
6559 * Operations such as "systemctl start" can now be executed
6560 with a new mode "--irreversible" which may be used to queue
6561 operations that cannot accidentally be reversed by a later
6562 job queuing. This is by default used to make shutdown
6563 requests more robust.
6565 * The Python API of systemd now gained a new module for
6566 reading journal files.
6568 * A new tool kernel-install has been added that can install
6569 kernel images according to the Boot Loader Specification:
6571 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6573 * Boot time console output has been improved to provide
6574 animated boot time output for hanging jobs.
6576 * A new tool systemd-activate has been added which can be used
6577 to test socket activation with, directly from the command
6578 line. This should make it much easier to test and debug
6579 socket activation in daemons.
6581 * journalctl gained a new "--reverse" (or -r) option to show
6582 journal output in reverse order (i.e. newest line first).
6584 * journalctl gained a new "--pager-end" (or -e) option to jump
6585 to immediately jump to the end of the journal in the
6586 pager. This is only supported in conjunction with "less".
6588 * journalctl gained a new "--user-unit=" option, that works
6589 similarly to "--unit=" but filters for user units rather than
6592 * A number of unit files to ease adoption of systemd in
6593 initrds has been added. This moves some minimal logic from
6594 the various initrd implementations into systemd proper.
6596 * The journal files are now owned by a new group
6597 "systemd-journal", which exists specifically to allow access
6598 to the journal, and nothing else. Previously, we used the
6599 "adm" group for that, which however possibly covers more
6600 than just journal/log file access. This new group is now
6601 already used by systemd-journal-gatewayd to ensure this
6602 daemon gets access to the journal files and as little else
6603 as possible. Note that "make install" will also set FS ACLs
6604 up for /var/log/journal to give "adm" and "wheel" read
6605 access to it, in addition to "systemd-journal" which owns
6606 the journal files. We recommend that packaging scripts also
6607 add read access to "adm" + "wheel" to /var/log/journal, and
6608 all existing/future journal files. To normal users and
6609 administrators little changes, however packagers need to
6610 ensure to create the "systemd-journal" system group at
6611 package installation time.
6613 * The systemd-journal-gatewayd now runs as unprivileged user
6614 systemd-journal-gateway:systemd-journal-gateway. Packaging
6615 scripts need to create these system user/group at
6618 * timedated now exposes a new boolean property CanNTP that
6619 indicates whether a local NTP service is available or not.
6621 * systemd-detect-virt will now also detect xen PVs
6623 * The pstore file system is now mounted by default, if it is
6626 * In addition to the SELinux and IMA policies we will now also
6627 load SMACK policies at early boot.
6629 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6630 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6631 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6632 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6633 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6634 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6635 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6636 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6637 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6638 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6639 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6640 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6641 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6642 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6646 * Timer units now support calendar time events in addition to
6647 monotonic time events. That means you can now trigger a unit
6648 based on a calendar time specification such as "Thu,Fri
6649 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6650 or fifth day of any month of the year 2013, given that it is
6651 a thursday or friday. This brings timer event support
6652 considerably closer to cron's capabilities. For details on
6653 the supported calendar time specification language see
6656 * udev now supports a number of different naming policies for
6657 network interfaces for predictable names, and a combination
6658 of these policies is now the default. Please see this wiki
6659 document for details:
6661 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6663 * Auke Kok's bootchart implementation has been added to the
6664 systemd tree. It is an optional component that can graph the
6665 boot in quite some detail. It is one of the best bootchart
6666 implementations around and minimal in its code and
6669 * nss-myhostname has been integrated into the systemd source
6670 tree. nss-myhostname guarantees that the local hostname
6671 always stays resolvable via NSS. It has been a weak
6672 requirement of systemd-hostnamed since a long time, and
6673 since its code is actually trivial we decided to just
6674 include it in systemd's source tree. It can be turned off
6675 with a configure switch.
6677 * The read-ahead logic is now capable of properly detecting
6678 whether a btrfs file system is on SSD or rotating media, in
6679 order to optimize the read-ahead scheme. Previously, it was
6680 only capable of detecting this on traditional file systems
6683 * In udev, additional device properties are now read from the
6684 IAB in addition to the OUI database. Also, Bluetooth company
6685 identities are attached to the devices as well.
6687 * In service files %U may be used as specifier that is
6688 replaced by the configured user name of the service.
6690 * nspawn may now be invoked without a controlling TTY. This
6691 makes it suitable for invocation as its own service. This
6692 may be used to set up a simple containerized server system
6693 using only core OS tools.
6695 * systemd and nspawn can now accept socket file descriptors
6696 when they are started for socket activation. This enables
6697 implementation of socket activated nspawn
6698 containers. i.e. think about autospawning an entire OS image
6699 when the first SSH or HTTP connection is received. We expect
6700 that similar functionality will also be added to libvirt-lxc
6703 * journalctl will now suppress ANSI color codes when
6704 presenting log data.
6706 * systemctl will no longer show control group information for
6707 a unit if the control group is empty anyway.
6709 * logind can now automatically suspend/hibernate/shutdown the
6712 * /etc/machine-info and hostnamed now also expose the chassis
6713 type of the system. This can be used to determine whether
6714 the local system is a laptop, desktop, handset or
6715 tablet. This information may either be configured by the
6716 user/vendor or is automatically determined from ACPI and DMI
6717 information if possible.
6719 * A number of PolicyKit actions are now bound together with
6720 "imply" rules. This should simplify creating UIs because
6721 many actions will now authenticate similar ones as well.
6723 * Unit files learnt a new condition ConditionACPower= which
6724 may be used to conditionalize a unit depending on whether an
6725 AC power source is connected or not, of whether the system
6726 is running on battery power.
6728 * systemctl gained a new "is-failed" verb that may be used in
6729 shell scripts and suchlike to check whether a specific unit
6730 is in the "failed" state.
6732 * The EnvironmentFile= setting in unit files now supports file
6733 globbing, and can hence be used to easily read a number of
6734 environment files at once.
6736 * systemd will no longer detect and recognize specific
6737 distributions. All distribution-specific #ifdeffery has been
6738 removed, systemd is now fully generic and
6739 distribution-agnostic. Effectively, not too much is lost as
6740 a lot of the code is still accessible via explicit configure
6741 switches. However, support for some distribution specific
6742 legacy configuration file formats has been dropped. We
6743 recommend distributions to simply adopt the configuration
6744 files everybody else uses now and convert the old
6745 configuration from packaging scripts. Most distributions
6746 already did that. If that's not possible or desirable,
6747 distributions are welcome to forward port the specific
6748 pieces of code locally from the git history.
6750 * When logging a message about a unit systemd will now always
6751 log the unit name in the message meta data.
6753 * localectl will now also discover system locale data that is
6754 not stored in locale archives, but directly unpacked.
6756 * logind will no longer unconditionally use framebuffer
6757 devices as seat masters, i.e. as devices that are required
6758 to be existing before a seat is considered preset. Instead,
6759 it will now look for all devices that are tagged as
6760 "seat-master" in udev. By default, framebuffer devices will
6761 be marked as such, but depending on local systems, other
6762 devices might be marked as well. This may be used to
6763 integrate graphics cards using closed source drivers (such
6764 as NVidia ones) more nicely into logind. Note however, that
6765 we recommend using the open source NVidia drivers instead,
6766 and no udev rules for the closed-source drivers will be
6767 shipped from us upstream.
6769 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6770 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6771 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6772 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6773 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6774 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6775 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6776 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6777 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6778 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6779 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6780 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6785 * udev gained support for loading additional device properties
6786 from an indexed database that is keyed by vendor/product IDs
6787 and similar device identifiers. For the beginning this
6788 "hwdb" is populated with data from the well-known PCI and
6789 USB database, but also includes PNP, ACPI and OID data. In
6790 the longer run this indexed database shall grow into
6791 becoming the one central database for non-essential
6792 userspace device metadata. Previously, data from the PCI/USB
6793 database was only attached to select devices, since the
6794 lookup was a relatively expensive operation due to O(n) time
6795 complexity (with n being the number of entries in the
6796 database). Since this is now O(1), we decided to add in this
6797 data for all devices where this is available, by
6798 default. Note that the indexed database needs to be rebuilt
6799 when new data files are installed. To achieve this you need
6800 to update your packaging scripts to invoke "udevadm hwdb
6801 --update" after installation of hwdb data files. For
6802 RPM-based distributions we introduced the new
6803 %udev_hwdb_update macro for this purpose.
6805 * The Journal gained support for the "Message Catalog", an
6806 indexed database to link up additional information with
6807 journal entries. For further details please check:
6809 https://www.freedesktop.org/wiki/Software/systemd/catalog
6811 The indexed message catalog database also needs to be
6812 rebuilt after installation of message catalog files. Use
6813 "journalctl --update-catalog" for this. For RPM-based
6814 distributions we introduced the %journal_catalog_update
6815 macro for this purpose.
6817 * The Python Journal bindings gained support for the standard
6818 Python logging framework.
6820 * The Journal API gained new functions for checking whether
6821 the underlying file system of a journal file is capable of
6822 properly reporting file change notifications, or whether
6823 applications that want to reflect journal changes "live"
6824 need to recheck journal files continuously in appropriate
6827 * It is now possible to set the "age" field for tmpfiles
6828 entries to 0, indicating that files matching this entry
6829 shall always be removed when the directories are cleaned up.
6831 * coredumpctl gained a new "gdb" verb which invokes gdb
6832 right-away on the selected coredump.
6834 * There's now support for "hybrid sleep" on kernels that
6835 support this, in addition to "suspend" and "hibernate". Use
6836 "systemctl hybrid-sleep" to make use of this.
6838 * logind's HandleSuspendKey= setting (and related settings)
6839 now gained support for a new "lock" setting to simply
6840 request the screen lock on all local sessions, instead of
6841 actually executing a suspend or hibernation.
6843 * systemd will now mount the EFI variables file system by
6846 * Socket units now gained support for configuration of the
6847 SMACK security label.
6849 * timedatectl will now output the time of the last and next
6850 daylight saving change.
6852 * We dropped support for various legacy and distro-specific
6853 concepts, such as insserv, early-boot SysV services
6854 (i.e. those for non-standard runlevels such as 'b' or 'S')
6855 or ArchLinux /etc/rc.conf support. We recommend the
6856 distributions who still need support this to either continue
6857 to maintain the necessary patches downstream, or find a
6858 different solution. (Talk to us if you have questions!)
6860 * Various systemd components will now bypass PolicyKit checks
6861 for root and otherwise handle properly if PolicyKit is not
6862 found to be around. This should fix most issues for
6863 PolicyKit-less systems. Quite frankly this should have been
6864 this way since day one. It is absolutely our intention to
6865 make systemd work fine on PolicyKit-less systems, and we
6866 consider it a bug if something does not work as it should if
6867 PolicyKit is not around.
6869 * For embedded systems it is now possible to build udev and
6870 systemd without blkid and/or kmod support.
6872 * "systemctl switch-root" is now capable of switching root
6873 more than once. I.e. in addition to transitions from the
6874 initrd to the host OS it is now possible to transition to
6875 further OS images from the host. This is useful to implement
6876 offline updating tools.
6878 * Various other additions have been made to the RPM macros
6879 shipped with systemd. Use %udev_rules_update() after
6880 installing new udev rules files. %_udevhwdbdir,
6881 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6882 %_sysctldir are now available which resolve to the right
6883 directories for packages to place various data files in.
6885 * journalctl gained the new --full switch (in addition to
6886 --all, to disable ellipsation for long messages.
6888 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6889 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6890 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6891 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6892 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6893 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6894 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6895 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6896 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6900 * journalctl gained new --since= and --until= switches to
6901 filter by time. It also now supports nice filtering for
6902 units via --unit=/-u.
6904 * Type=oneshot services may use ExecReload= and do the
6907 * The journal daemon now supports time-based rotation and
6908 vacuuming, in addition to the usual disk-space based
6911 * The journal will now index the available field values for
6912 each field name. This enables clients to show pretty drop
6913 downs of available match values when filtering. The bash
6914 completion of journalctl has been updated
6915 accordingly. journalctl gained a new switch -F to list all
6916 values a certain field takes in the journal database.
6918 * More service events are now written as structured messages
6919 to the journal, and made recognizable via message IDs.
6921 * The timedated, localed and hostnamed mini-services which
6922 previously only provided support for changing time, locale
6923 and hostname settings from graphical DEs such as GNOME now
6924 also have a minimal (but very useful) text-based client
6925 utility each. This is probably the nicest way to changing
6926 these settings from the command line now, especially since
6927 it lists available options and is fully integrated with bash
6930 * There's now a new tool "systemd-coredumpctl" to list and
6931 extract coredumps from the journal.
6933 * We now install a README each in /var/log/ and
6934 /etc/rc.d/init.d explaining where the system logs and init
6935 scripts went. This hopefully should help folks who go to
6936 that dirs and look into the otherwise now empty void and
6937 scratch their heads.
6939 * When user-services are invoked (by systemd --user) the
6940 $MANAGERPID env var is set to the PID of systemd.
6942 * SIGRTMIN+24 when sent to a --user instance will now result
6943 in immediate termination of systemd.
6945 * gatewayd received numerous feature additions such as a
6946 "follow" mode, for live syncing and filtering.
6948 * browse.html now allows filtering and showing detailed
6949 information on specific entries. Keyboard navigation and
6950 mouse screen support has been added.
6952 * gatewayd/journalctl now supports HTML5/JSON
6953 Server-Sent-Events as output.
6955 * The SysV init script compatibility logic will now
6956 heuristically determine whether a script supports the
6957 "reload" verb, and only then make this available as
6960 * "systemctl status --follow" has been removed, use "journalctl
6963 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6964 have been removed since they are hardly useful to be
6967 * And I'd like to take the opportunity to specifically mention
6968 Zbigniew for his great contributions. Zbigniew, you rock!
6970 Contributions from: Andrew Eikum, Christian Hesse, Colin
6971 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6972 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6973 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6974 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6975 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6976 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6980 * If /etc/vconsole.conf is non-existent or empty we will no
6981 longer load any console font or key map at boot by
6982 default. Instead the kernel defaults will be left
6983 intact. This is definitely the right thing to do, as no
6984 configuration should mean no configuration, and hard-coding
6985 font names that are different on all archs is probably a bad
6986 idea. Also, the kernel default key map and font should be
6987 good enough for most cases anyway, and mostly identical to
6988 the userspace fonts/key maps we previously overloaded them
6989 with. If distributions want to continue to default to a
6990 non-kernel font or key map they should ship a default
6991 /etc/vconsole.conf with the appropriate contents.
6993 Contributions from: Colin Walters, Daniel J Walsh, Dave
6994 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6995 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6999 * journalctl gained a new --cursor= switch to show entries
7000 starting from the specified location in the journal.
7002 * We now enforce a size limit on journal entry fields exported
7003 with "-o json" in journalctl. Fields larger than 4K will be
7004 assigned null. This can be turned off with --all.
7006 * An (optional) journal gateway daemon is now available as
7007 "systemd-journal-gatewayd.service". This service provides
7008 access to the journal via HTTP and JSON. This functionality
7009 will be used to implement live log synchronization in both
7010 pull and push modes, but has various other users too, such
7011 as easy log access for debugging of embedded devices. Right
7012 now it is already useful to retrieve the journal via HTTP:
7014 # systemctl start systemd-journal-gatewayd.service
7015 # wget http://localhost:19531/entries
7017 This will download the journal contents in a
7018 /var/log/messages compatible format. The same as JSON:
7020 # curl -H"Accept: application/json" http://localhost:19531/entries
7022 This service is also accessible via a web browser where a
7023 single static HTML5 app is served that uses the JSON logic
7024 to enable the user to do some basic browsing of the
7025 journal. This will be extended later on. Here's an example
7026 screenshot of this app in its current state:
7028 http://0pointer.de/public/journal-gatewayd
7030 Contributions from: Kay Sievers, Lennart Poettering, Robert
7031 Milasan, Tom Gundersen
7035 * The bash completion logic is now available for journalctl
7038 * We do not mount the "cpuset" controller anymore together with
7039 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7040 started if no parameters are assigned to it. "cpuset" hence
7041 broke code that assumed it could create "cpu" groups and
7044 * journalctl -f will now subscribe to terminal size changes,
7045 and line break accordingly.
7047 Contributions from: Dave Reisner, Kay Sievers, Lennart
7048 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7052 * nspawn will now create a symlink /etc/localtime in the
7053 container environment, copying the host's timezone
7054 setting. Previously this has been done via a bind mount, but
7055 since symlinks cannot be bind mounted this has now been
7056 changed to create/update the appropriate symlink.
7058 * journalctl -n's line number argument is now optional, and
7059 will default to 10 if omitted.
7061 * journald will now log the maximum size the journal files may
7062 take up on disk. This is particularly useful if the default
7063 built-in logic of determining this parameter from the file
7064 system size is used. Use "systemctl status
7065 systemd-journald.service" to see this information.
7067 * The multi-seat X wrapper tool has been stripped down. As X
7068 is now capable of enumerating graphics devices via udev in a
7069 seat-aware way the wrapper is not strictly necessary
7070 anymore. A stripped down temporary stop-gap is still shipped
7071 until the upstream display managers have been updated to
7072 fully support the new X logic. Expect this wrapper to be
7073 removed entirely in one of the next releases.
7075 * HandleSleepKey= in logind.conf has been split up into
7076 HandleSuspendKey= and HandleHibernateKey=. The old setting
7077 is not available anymore. X11 and the kernel are
7078 distinguishing between these keys and we should too. This
7079 also means the inhibition lock for these keys has been split
7082 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7083 Poettering, Lukas Nykryn, Václav Pavlín
7087 * Whenever a unit changes state we will now log this to the
7088 journal and show along the unit's own log output in
7091 * ConditionPathIsMountPoint= can now properly detect bind
7092 mount points too. (Previously, a bind mount of one file
7093 system to another place in the same file system could not be
7094 detected as mount, since they shared struct stat's st_dev
7097 * We will now mount the cgroup controllers cpu, cpuacct,
7098 cpuset and the controllers net_cls, net_prio together by
7101 * nspawn containers will now have a virtualized boot
7102 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7103 over with a randomized ID at container initialization). This
7104 has the effect of making "journalctl -b" do the right thing
7107 * The JSON output journal serialization has been updated not
7108 to generate "endless" list objects anymore, but rather one
7109 JSON object per line. This is more in line how most JSON
7110 parsers expect JSON objects. The new output mode
7111 "json-pretty" has been added to provide similar output, but
7112 neatly aligned for readability by humans.
7114 * We dropped all explicit sync() invocations in the shutdown
7115 code. The kernel does this implicitly anyway in the kernel
7116 reboot() syscall. halt(8)'s -n option is now a compatibility
7119 * We now support virtualized reboot() in containers, as
7120 supported by newer kernels. We will fall back to exit() if
7121 CAP_SYS_REBOOT is not available to the container. Also,
7122 nspawn makes use of this now and will actually reboot the
7123 container if the containerized OS asks for that.
7125 * journalctl will only show local log output by default
7126 now. Use --merge (-m) to show remote log output, too.
7128 * libsystemd-journal gained the new sd_journal_get_usage()
7129 call to determine the current disk usage of all journal
7130 files. This is exposed in the new "journalctl --disk-usage"
7133 * journald gained a new configuration setting SplitMode= in
7134 journald.conf which may be used to control how user journals
7135 are split off. See journald.conf(5) for details.
7137 * A new condition type ConditionFileNotEmpty= has been added.
7139 * tmpfiles' "w" lines now support file globbing, to write
7140 multiple files at once.
7142 * We added Python bindings for the journal submission
7143 APIs. More Python APIs for a number of selected APIs will
7144 likely follow. Note that we intend to add native bindings
7145 only for the Python language, as we consider it common
7146 enough to deserve bindings shipped within systemd. There are
7147 various projects outside of systemd that provide bindings
7148 for languages such as PHP or Lua.
7150 * Many conditions will now resolve specifiers such as %i. In
7151 addition, PathChanged= and related directives of .path units
7152 now support specifiers as well.
7154 * There's now a new RPM macro definition for the system preset
7157 * journald will now warn if it ca not forward a message to the
7158 syslog daemon because its socket is full.
7160 * timedated will no longer write or process /etc/timezone,
7161 except on Debian. As we do not support late mounted /usr
7162 anymore /etc/localtime always being a symlink is now safe,
7163 and hence the information in /etc/timezone is not necessary
7166 * logind will now always reserve one VT for a text getty (VT6
7167 by default). Previously if more than 6 X sessions where
7168 started they took up all the VTs with auto-spawned gettys,
7169 so that no text gettys were available anymore.
7171 * udev will now automatically inform the btrfs kernel logic
7172 about btrfs RAID components showing up. This should make
7173 simple hotplug based btrfs RAID assembly work.
7175 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7176 (but not for its children which will stay at the kernel
7177 default). This should allow setups with a lot more listening
7180 * systemd will now always pass the configured timezone to the
7181 kernel at boot. timedated will do the same when the timezone
7184 * logind's inhibition logic has been updated. By default,
7185 logind will now handle the lid switch, the power and sleep
7186 keys all the time, even in graphical sessions. If DEs want
7187 to handle these events on their own they should take the new
7188 handle-power-key, handle-sleep-key and handle-lid-switch
7189 inhibitors during their runtime. A simple way to achieve
7190 that is to invoke the DE wrapped in an invocation of:
7192 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7194 * Access to unit operations is now checked via SELinux taking
7195 the unit file label and client process label into account.
7197 * systemd will now notify the administrator in the journal
7198 when he over-mounts a non-empty directory.
7200 * There are new specifiers that are resolved in unit files,
7201 for the host name (%H), the machine ID (%m) and the boot ID
7204 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7205 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7206 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7207 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7208 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7209 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7210 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7214 * Support for reading structured kernel messages from
7215 /dev/kmsg has now been added and is enabled by default.
7217 * Support for reading kernel messages from /proc/kmsg has now
7218 been removed. If you want kernel messages in the journal
7219 make sure to run a recent kernel (>= 3.5) that supports
7220 reading structured messages from /dev/kmsg (see
7221 above). /proc/kmsg is now exclusive property of classic
7222 syslog daemons again.
7224 * The libudev API gained the new
7225 udev_device_new_from_device_id() call.
7227 * The logic for file system namespace (ReadOnlyDirectory=,
7228 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7229 require pivot_root() anymore. This means fewer temporary
7230 directories are created below /tmp for this feature.
7232 * nspawn containers will now see and receive all submounts
7233 made on the host OS below the root file system of the
7236 * Forward Secure Sealing is now supported for Journal files,
7237 which provide cryptographical sealing of journal files so
7238 that attackers cannot alter log history anymore without this
7239 being detectable. Lennart will soon post a blog story about
7240 this explaining it in more detail.
7242 * There are two new service settings RestartPreventExitStatus=
7243 and SuccessExitStatus= which allow configuration of exit
7244 status (exit code or signal) which will be excepted from the
7245 restart logic, resp. consider successful.
7247 * journalctl gained the new --verify switch that can be used
7248 to check the integrity of the structure of journal files and
7249 (if Forward Secure Sealing is enabled) the contents of
7252 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7253 and similar symlinks pre-created. This makes running shells
7254 as container init process a lot more fun.
7256 * The fstab support can now handle PARTUUID= and PARTLABEL=
7259 * A new ConditionHost= condition has been added to match
7260 against the hostname (with globs) and machine ID. This is
7261 useful for clusters where a single OS image is used to
7262 provision a large number of hosts which shall run slightly
7263 different sets of services.
7265 * Services which hit the restart limit will now be placed in a
7268 Contributions from: Bertram Poettering, Dave Reisner, Huang
7269 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7270 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7274 * When running in --user mode systemd will now become a
7275 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7276 tree a lot more organized.
7278 * A new PartOf= unit dependency type has been introduced that
7279 may be used to group services in a natural way.
7281 * "systemctl enable" may now be used to enable instances of
7284 * journalctl now prints error log levels in red, and
7285 warning/notice log levels in bright white. It also supports
7286 filtering by log level now.
7288 * cgtop gained a new -n switch (similar to top), to configure
7289 the maximum number of iterations to run for. It also gained
7290 -b, to run in batch mode (accepting no input).
7292 * The suffix ".service" may now be omitted on most systemctl
7293 command lines involving service unit names.
7295 * There's a new bus call in logind to lock all sessions, as
7296 well as a loginctl verb for it "lock-sessions".
7298 * libsystemd-logind.so gained a new call sd_journal_perror()
7299 that works similar to libc perror() but logs to the journal
7300 and encodes structured information about the error number.
7302 * /etc/crypttab entries now understand the new keyfile-size=
7305 * shutdown(8) now can send a (configurable) wall message when
7306 a shutdown is cancelled.
7308 * The mount propagation mode for the root file system will now
7309 default to "shared", which is useful to make containers work
7310 nicely out-of-the-box so that they receive new mounts from
7311 the host. This can be undone locally by running "mount
7312 --make-rprivate /" if needed.
7314 * The prefdm.service file has been removed. Distributions
7315 should maintain this unit downstream if they intend to keep
7316 it around. However, we recommend writing normal unit files
7317 for display managers instead.
7319 * Since systemd is a crucial part of the OS we will now
7320 default to a number of compiler switches that improve
7321 security (hardening) such as read-only relocations, stack
7322 protection, and suchlike.
7324 * The TimeoutSec= setting for services is now split into
7325 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7326 of individual time outs for the start and the stop phase of
7329 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7330 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7331 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7332 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7333 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7334 Gundersen, Zbigniew Jędrzejewski-Szmek
7338 * The journal and id128 C APIs are now fully documented as man
7341 * Extra safety checks have been added when transitioning from
7342 the initial RAM disk to the main system to avoid accidental
7345 * /etc/crypttab entries now understand the new keyfile-offset=
7348 * systemctl -t can now be used to filter by unit load state.
7350 * The journal C API gained the new sd_journal_wait() call to
7351 make writing synchronous journal clients easier.
7353 * journalctl gained the new -D switch to show journals from a
7356 * journalctl now displays a special marker between log
7357 messages of two different boots.
7359 * The journal is now explicitly flushed to /var via a service
7360 systemd-journal-flush.service, rather than implicitly simply
7361 by seeing /var/log/journal to be writable.
7363 * journalctl (and the journal C APIs) can now match for much
7364 more complex expressions, with alternatives and
7367 * When transitioning from the initial RAM disk to the main
7368 system we will now kill all processes in a killing spree to
7369 ensure no processes stay around by accident.
7371 * Three new specifiers may be used in unit files: %u, %h, %s
7372 resolve to the user name, user home directory resp. user
7373 shell. This is useful for running systemd user instances.
7375 * We now automatically rotate journal files if their data
7376 object hash table gets a fill level > 75%. We also size the
7377 hash table based on the configured maximum file size. This
7378 together should lower hash collisions drastically and thus
7379 speed things up a bit.
7381 * journalctl gained the new "--header" switch to introspect
7382 header data of journal files.
7384 * A new setting SystemCallFilters= has been added to services
7385 which may be used to apply blacklists or whitelists to
7386 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7388 * nspawn gained a new --link-journal= switch (and quicker: -j)
7389 to link the container journal with the host. This makes it
7390 very easy to centralize log viewing on the host for all
7391 guests while still keeping the journal files separated.
7393 * Many bugfixes and optimizations
7395 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7396 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7397 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7402 * Several tools now understand kernel command line arguments,
7403 which are only read when run in an initial RAM disk. They
7404 usually follow closely their normal counterparts, but are
7407 * There's a new tool to analyze the readahead files that are
7408 automatically generated at boot. Use:
7410 /usr/lib/systemd/systemd-readahead analyze /.readahead
7412 * We now provide an early debug shell on tty9 if this enabled. Use:
7414 systemctl enable debug-shell.service
7416 * All plymouth related units have been moved into the Plymouth
7417 package. Please make sure to upgrade your Plymouth version
7420 * systemd-tmpfiles now supports getting passed the basename of
7421 a configuration file only, in which case it will look for it
7422 in all appropriate directories automatically.
7424 * udevadm info now takes a /dev or /sys path as argument, and
7425 does the right thing. Example:
7427 udevadm info /dev/sda
7428 udevadm info /sys/class/block/sda
7430 * systemctl now prints a warning if a unit is stopped but a
7431 unit that might trigger it continues to run. Example: a
7432 service is stopped but the socket that activates it is left
7435 * "systemctl status" will now mention if the log output was
7436 shortened due to rotation since a service has been started.
7438 * The journal API now exposes functions to determine the
7439 "cutoff" times due to rotation.
7441 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7442 immediately flushing of runtime logs to /var if possible,
7443 resp. for triggering immediate rotation of the journal
7446 * It is now considered an error if a service is attempted to
7447 be stopped that is not loaded.
7449 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7451 * systemd-analyze now supports Python 3
7453 * tmpfiles now supports cleaning up directories via aging
7454 where the first level dirs are always kept around but
7455 directories beneath it automatically aged. This is enabled
7456 by prefixing the age field with '~'.
7458 * Seat objects now expose CanGraphical, CanTTY properties
7459 which is required to deal with very fast bootups where the
7460 display manager might be running before the graphics drivers
7461 completed initialization.
7463 * Seat objects now expose a State property.
7465 * We now include RPM macros for service enabling/disabling
7466 based on the preset logic. We recommend RPM based
7467 distributions to make use of these macros if possible. This
7468 makes it simpler to reuse RPM spec files across
7471 * We now make sure that the collected systemd unit name is
7472 always valid when services log to the journal via
7475 * There's a new man page kernel-command-line(7) detailing all
7476 command line options we understand.
7478 * The fstab generator may now be disabled at boot by passing
7479 fstab=0 on the kernel command line.
7481 * A new kernel command line option modules-load= is now understood
7482 to load a specific kernel module statically, early at boot.
7484 * Unit names specified on the systemctl command line are now
7485 automatically escaped as needed. Also, if file system or
7486 device paths are specified they are automatically turned
7487 into the appropriate mount or device unit names. Example:
7489 systemctl status /home
7490 systemctl status /dev/sda
7492 * The SysVConsole= configuration option has been removed from
7493 system.conf parsing.
7495 * The SysV search path is no longer exported on the D-Bus
7498 * The Names= option has been removed from unit file parsing.
7500 * There's a new man page bootup(7) detailing the boot process.
7502 * Every unit and every generator we ship with systemd now
7503 comes with full documentation. The self-explanatory boot is
7506 * A couple of services gained "systemd-" prefixes in their
7507 name if they wrap systemd code, rather than only external
7508 code. Among them fsck@.service which is now
7509 systemd-fsck@.service.
7511 * The HaveWatchdog property has been removed from the D-Bus
7514 * systemd.confirm_spawn= on the kernel command line should now
7517 * There's a new man page crypttab(5) which details all options
7518 we actually understand.
7520 * systemd-nspawn gained a new --capability= switch to pass
7521 additional capabilities to the container.
7523 * timedated will now read known NTP implementation unit names
7524 from /usr/lib/systemd/ntp-units.d/*.list,
7525 systemd-timedated-ntp.target has been removed.
7527 * journalctl gained a new switch "-b" that lists log data of
7528 the current boot only.
7530 * The notify socket is in the abstract namespace again, in
7531 order to support daemons which chroot() at start-up.
7533 * There is a new Storage= configuration option for journald
7534 which allows configuration of where log data should go. This
7535 also provides a way to disable journal logging entirely, so
7536 that data collected is only forwarded to the console, the
7537 kernel log buffer or another syslog implementation.
7539 * Many bugfixes and optimizations
7541 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7542 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7543 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7544 Shawn Landden, Tom Gundersen
7548 * "systemctl help <unit>" now shows the man page if one is
7551 * Several new man pages have been added.
7553 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7554 MaxLevelConsole= can now be specified in
7555 journald.conf. These options allow reducing the amount of
7556 data stored on disk or forwarded by the log level.
7558 * TimerSlackNSec= can now be specified in system.conf for
7559 PID1. This allows system-wide power savings.
7561 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7562 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7567 * logind is now capable of (optionally) handling power and
7568 sleep keys as well as the lid switch.
7570 * journalctl now understands the syntax "journalctl
7571 /usr/bin/avahi-daemon" to get all log output of a specific
7574 * CapabilityBoundingSet= in system.conf now also influences
7575 the capability bound set of usermode helpers of the kernel.
7577 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7578 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7579 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7580 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7584 * Note that we skipped 139 releases here in order to set the
7585 new version to something that is greater than both udev's
7586 and systemd's most recent version number.
7588 * udev: all udev sources are merged into the systemd source tree now.
7589 All future udev development will happen in the systemd tree. It
7590 is still fully supported to use the udev daemon and tools without
7591 systemd running, like in initramfs or other init systems. Building
7592 udev though, will require the *build* of the systemd tree, but
7593 udev can be properly *run* without systemd.
7595 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7596 should be used to create dead device nodes as workarounds for broken
7599 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7600 no longer supported. udev_monitor_new_from_netlink() needs to be
7601 used to subscribe to events.
7603 * udev: when udevd is started by systemd, processes which are left
7604 behind by forking them off of udev rules, are unconditionally cleaned
7605 up and killed now after the event handling has finished. Services or
7606 daemons must be started as systemd services. Services can be
7607 pulled-in by udev to get started, but they can no longer be directly
7608 forked by udev rules.
7610 * udev: the daemon binary is called systemd-udevd now and installed
7611 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7612 to adapt to that, create symlink, or rename the binary after building
7615 * libudev no longer provides these symbols:
7616 udev_monitor_from_socket()
7617 udev_queue_get_failed_list_entry()
7618 udev_get_{dev,sys,run}_path()
7619 The versions number was bumped and symbol versioning introduced.
7621 * systemd-loginctl and systemd-journalctl have been renamed
7622 to loginctl and journalctl to match systemctl.
7624 * The config files: /etc/systemd/systemd-logind.conf and
7625 /etc/systemd/systemd-journald.conf have been renamed to
7626 logind.conf and journald.conf. Package updates should rename
7627 the files to the new names on upgrade.
7629 * For almost all files the license is now LGPL2.1+, changed
7630 from the previous GPL2.0+. Exceptions are some minor stuff
7631 of udev (which will be changed to LGPL2.1 eventually, too),
7632 and the MIT licensed sd-daemon.[ch] library that is suitable
7633 to be used as drop-in files.
7635 * systemd and logind now handle system sleep states, in
7636 particular suspending and hibernating.
7638 * logind now implements a sleep/shutdown/idle inhibiting logic
7639 suitable for a variety of uses. Soonishly Lennart will blog
7640 about this in more detail.
7642 * var-run.mount and var-lock.mount are no longer provided
7643 (which previously bind mounted these directories to their new
7644 places). Distributions which have not converted these
7645 directories to symlinks should consider stealing these files
7646 from git history and add them downstream.
7648 * We introduced the Documentation= field for units and added
7649 this to all our shipped units. This is useful to make it
7650 easier to explore the boot and the purpose of the various
7653 * All smaller setup units (such as
7654 systemd-vconsole-setup.service) now detect properly if they
7655 are run in a container and are skipped when
7656 appropriate. This guarantees an entirely noise-free boot in
7657 Linux container environments such as systemd-nspawn.
7659 * A framework for implementing offline system updates is now
7660 integrated, for details see:
7661 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7663 * A new service type Type=idle is available now which helps us
7664 avoiding ugly interleaving of getty output and boot status
7667 * There's now a system-wide CapabilityBoundingSet= option to
7668 globally reduce the set of capabilities for the
7669 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7670 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7671 even CAP_NET_ADMIN system-wide for secure systems.
7673 * There are now system-wide DefaultLimitXXX= options to
7674 globally change the defaults of the various resource limits
7675 for all units started by PID 1.
7677 * Harald Hoyer's systemd test suite has been integrated into
7678 systemd which allows easy testing of systemd builds in qemu
7679 and nspawn. (This is really awesome! Ask us for details!)
7681 * The fstab parser is now implemented as generator, not inside
7684 * systemctl will now warn you if .mount units generated from
7685 /etc/fstab are out of date due to changes in fstab that
7686 have not been read by systemd yet.
7688 * systemd is now suitable for usage in initrds. Dracut has
7689 already been updated to make use of this. With this in place
7690 initrds get a slight bit faster but primarily are much
7691 easier to introspect and debug since "systemctl status" in
7692 the host system can be used to introspect initrd services,
7693 and the journal from the initrd is kept around too.
7695 * systemd-delta has been added, a tool to explore differences
7696 between user/admin configuration and vendor defaults.
7698 * PrivateTmp= now affects both /tmp and /var/tmp.
7700 * Boot time status messages are now much prettier and feature
7701 proper english language. Booting up systemd has never been
7704 * Read-ahead pack files now include the inode number of all
7705 files to pre-cache. When the inode changes the pre-caching
7706 is not attempted. This should be nicer to deal with updated
7707 packages which might result in changes of read-ahead
7710 * We now temporaritly lower the kernel's read_ahead_kb variable
7711 when collecting read-ahead data to ensure the kernel's
7712 built-in read-ahead does not add noise to our measurements
7713 of necessary blocks to pre-cache.
7715 * There's now RequiresMountsFor= to add automatic dependencies
7716 for all mounts necessary for a specific file system path.
7718 * MountAuto= and SwapAuto= have been removed from
7719 system.conf. Mounting file systems at boot has to take place
7722 * nspawn now learned a new switch --uuid= to set the machine
7723 ID on the command line.
7725 * nspawn now learned the -b switch to automatically search
7728 * vt102 is now the default TERM for serial TTYs, upgraded from
7731 * systemd-logind now works on VT-less systems.
7733 * The build tree has been reorganized. The individual
7734 components now have directories of their own.
7736 * A new condition type ConditionPathIsReadWrite= is now available.
7738 * nspawn learned the new -C switch to create cgroups for the
7739 container in other hierarchies.
7741 * We now have support for hardware watchdogs, configurable in
7744 * The scheduled shutdown logic now has a public API.
7746 * We now mount /tmp as tmpfs by default, but this can be
7747 masked and /etc/fstab can override it.
7749 * Since udisks does not make use of /media anymore we are not
7750 mounting a tmpfs on it anymore.
7752 * journalctl gained a new --local switch to only interleave
7753 locally generated journal files.
7755 * We can now load the IMA policy at boot automatically.
7757 * The GTK tools have been split off into a systemd-ui.
7759 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7760 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7761 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7762 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7763 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7764 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7765 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7766 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7767 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7772 * This is mostly a bugfix release
7774 * Support optional initialization of the machine ID from the
7775 KVM or container configured UUID.
7777 * Support immediate reboots with "systemctl reboot -ff"
7779 * Show /etc/os-release data in systemd-analyze output
7781 * Many bugfixes for the journal, including endianness fixes and
7782 ensuring that disk space enforcement works
7784 * sd-login.h is C++ compatible again
7786 * Extend the /etc/os-release format on request of the Debian
7789 * We now refuse non-UTF8 strings used in various configuration
7790 and unit files. This is done to ensure we do not pass invalid
7791 data over D-Bus or expose it elsewhere.
7793 * Register Mimo USB Screens as suitable for automatic seat
7796 * Read SELinux client context from journal clients in a race
7799 * Reorder configuration file lookup order. /etc now always
7800 overrides /run in order to allow the administrator to always
7801 and unconditionally override vendor-supplied or
7802 automatically generated data.
7804 * The various user visible bits of the journal now have man
7805 pages. We still lack man pages for the journal API calls
7808 * We now ship all man pages in HTML format again in the
7811 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7812 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7813 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7818 * This is mostly a bugfix release
7820 * systems lacking /etc/os-release are no longer supported.
7822 * Various functionality updates to libsystemd-login.so
7824 * Track class of PAM logins to distinguish greeters from
7827 Contributions from: Kay Sievers, Lennart Poettering, Michael
7832 * This is an important bugfix release for v41.
7834 * Building man pages is now optional which should be useful
7835 for those building systemd from git but unwilling to install
7838 * Watchdog support for supervising services is now usable. In
7839 a future release support for hardware watchdogs
7840 (i.e. /dev/watchdog) will be added building on this.
7842 * Service start rate limiting is now configurable and can be
7843 turned off per service. When a start rate limit is hit a
7844 reboot can automatically be triggered.
7846 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7848 Contributions from: Benjamin Franzke, Bill Nottingham,
7849 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7850 Schmidt, Michał Górny, Piotr Drąg
7854 * The systemd binary is installed /usr/lib/systemd/systemd now;
7855 An existing /sbin/init symlink needs to be adapted with the
7858 * The code that loads kernel modules has been ported to invoke
7859 libkmod directly, instead of modprobe. This means we do not
7860 support systems with module-init-tools anymore.
7862 * Watchdog support is now already useful, but still not
7865 * A new kernel command line option systemd.setenv= is
7866 understood to set system wide environment variables
7867 dynamically at boot.
7869 * We now limit the set of capabilities of systemd-journald.
7871 * We now set SIGPIPE to ignore by default, since it only is
7872 useful in shell pipelines, and has little use in general
7873 code. This can be disabled with IgnoreSIPIPE=no in unit
7876 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7877 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7882 * This is mostly a bugfix release
7884 * We now expose the reason why a service failed in the
7885 "Result" D-Bus property.
7887 * Rudimentary service watchdog support (will be completed over
7888 the next few releases.)
7890 * When systemd forks off in order execute some service we will
7891 now immediately changes its argv[0] to reflect which process
7892 it will execute. This is useful to minimize the time window
7893 with a generic argv[0], which makes bootcharts more useful
7895 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7896 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7897 Mike Kazantsev, Ray Strode
7901 * This is mostly a test release, but incorporates many
7904 * New systemd-cgtop tool to show control groups by their
7907 * Linking against libacl for ACLs is optional again. If
7908 disabled, support tracking device access for active logins
7909 goes becomes unavailable, and so does access to the user
7910 journals by the respective users.
7912 * If a group "adm" exists, journal files are automatically
7913 owned by them, thus allow members of this group full access
7914 to the system journal as well as all user journals.
7916 * The journal now stores the SELinux context of the logging
7917 client for all entries.
7919 * Add C++ inclusion guards to all public headers
7921 * New output mode "cat" in the journal to print only text
7922 messages, without any meta data like date or time.
7924 * Include tiny X server wrapper as a temporary stop-gap to
7925 teach XOrg udev display enumeration. This is used by display
7926 managers such as gdm, and will go away as soon as XOrg
7927 learned native udev hotplugging for display devices.
7929 * Add new systemd-cat tool for executing arbitrary programs
7930 with STDERR/STDOUT connected to the journal. Can also act as
7931 BSD logger replacement, and does so by default.
7933 * Optionally store all locally generated coredumps in the
7934 journal along with meta data.
7936 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7937 writing short strings to files (for usage for /sys), and for
7938 creating symlinks, character and block device nodes.
7940 * New unit file option ControlGroupPersistent= to make cgroups
7941 persistent, following the mechanisms outlined in
7942 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7944 * Support multiple local RTCs in a sane way
7946 * No longer monopolize IO when replaying readahead data on
7947 rotating disks, since we might starve non-file-system IO to
7948 death, since fanotify() will not see accesses done by blkid,
7951 * Do not show kernel threads in systemd-cgls anymore, unless
7952 requested with new -k switch.
7954 Contributions from: Dan Horák, Kay Sievers, Lennart
7955 Poettering, Michal Schmidt
7959 * This is mostly a test release, but incorporates many
7962 * The git repository moved to:
7963 git://anongit.freedesktop.org/systemd/systemd
7964 ssh://git.freedesktop.org/git/systemd/systemd
7966 * First release with the journal
7967 http://0pointer.de/blog/projects/the-journal.html
7969 * The journal replaces both systemd-kmsg-syslogd and
7970 systemd-stdout-bridge.
7972 * New sd_pid_get_unit() API call in libsystemd-logind
7974 * Many systemadm clean-ups
7976 * Introduce remote-fs-pre.target which is ordered before all
7977 remote mounts and may be used to start services before all
7980 * Added Mageia support
7982 * Add bash completion for systemd-loginctl
7984 * Actively monitor PID file creation for daemons which exit in
7985 the parent process before having finished writing the PID
7986 file in the daemon process. Daemons which do this need to be
7987 fixed (i.e. PID file creation must have finished before the
7988 parent exits), but we now react a bit more gracefully to them.
7990 * Add colourful boot output, mimicking the well-known output
7991 of existing distributions.
7993 * New option PassCredentials= for socket units, for
7994 compatibility with a recent kernel ABI breakage.
7996 * /etc/rc.local is now hooked in via a generator binary, and
7997 thus will no longer act as synchronization point during
8000 * systemctl list-unit-files now supports --root=.
8002 * systemd-tmpfiles now understands two new commands: z, Z for
8003 relabelling files according to the SELinux database. This is
8004 useful to apply SELinux labels to specific files in /sys,
8007 * Output of SysV services is now forwarded to both the console
8008 and the journal by default, not only just the console.
8010 * New man pages for all APIs from libsystemd-login.
8012 * The build tree got reorganized and the build system is a
8013 lot more modular allowing embedded setups to specifically
8014 select the components of systemd they are interested in.
8016 * Support for Linux systems lacking the kernel VT subsystem is
8019 * configure's --with-rootdir= got renamed to
8020 --with-rootprefix= to follow the naming used by udev and
8023 * Unless specified otherwise we will now install to /usr instead
8024 of /usr/local by default.
8026 * Processes with '@' in argv[0][0] are now excluded from the
8027 final shut-down killing spree, following the logic explained
8029 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8031 * All processes remaining in a service cgroup when we enter
8032 the START or START_PRE states are now killed with
8033 SIGKILL. That means it is no longer possible to spawn
8034 background processes from ExecStart= lines (which was never
8035 supported anyway, and bad style).
8037 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8038 reloading of units together.
8040 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8041 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8042 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8043 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8044 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob