1 systemd System and Service Manager
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 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
23 elogind-tmpfiles behaviour: previously, read-only files owned by root
24 were always excluded from the file "aging" algorithm (i.e. the
25 automatic clean-up of directories like /tmp based on
26 atime/mtime/ctime). We intend to drop this restriction, and age files
27 by default even when owned by root and read-only. This behaviour was
28 inherited from older tools, but there have been requests to remove
29 it, and it's not obvious why this restriction was made in the first
30 place. Please speak up now, if you are aware of software that reqires
31 this behaviour, otherwise we'll remove the restriction in v238.
33 * A new environment variable $SYSTEMD_OFFLINE is now understood by
34 systemctl. It takes a boolean argument. If on, systemctl assumes it
35 operates on an "offline" OS tree, and will not attempt to talk to the
36 service manager. Previously, this mode was implicitly enabled if a
37 chroot() environment was detected, and this new environment variable
38 now provides explicit control.
40 * .path and .socket units may now be created transiently,
41 too. Previously only service, mount, automount and timer units were
42 supported as transient units. The elogind-run tool has been updated
43 to expose this new functionality, you may hence use it now to bind
44 arbitrary commands to path or socket activation on-the-fly from the
45 command line. Moreover, for the unit types already supporting
46 transient operation most properties are now hooked up.
48 * The elogind-mount command gained support for a new --owner= parameter
49 which takes a user name, which is then resolved and included in uid=
50 and gid= mount options string of the file system to mount.
52 * A new unit condition ConditionControlGroupController= has been added
53 that checks whether a specific cgroup controller is available.
55 * Unit files, udev's .link files, and elogind-networkd's .netdev and
56 .network files all gained support for a new condition
57 ConditionKernelVersion= for checking against specific kernel
60 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
61 support for configuring device flags in the Flags= setting. In the
62 same files, the [Tunnel] section gained support for configuring
63 AllowLocalRemote=. The [Route] section in .network files gained
64 support for configuring InitialCongestionWindow=,
65 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
66 understands RapidCommit=.
68 * elogind-networkd's DHCPv6 support gained support for Prefix
71 * sd-bus gained support for a new "watch-bind" feature. When this
72 feature is enabled, an sd_bus connection may be set up to connect to
73 an AF_UNIX socket in the file system as soon as it is created. This
74 functionality is useful for writing early-boot services that
75 automatically connect to the system bus as soon as it is started,
76 without ugly time-based polling. elogind-networkd and
77 elogind-resolved have been updated to make use of this
78 functionality. busctl exposes this functionality in a new
79 --watch-bind= command line switch.
81 * sd-bus will now optionally synthesize a local "Connected" signal as
82 soon as a D-Bus connection is set up fully. This message mirrors the
83 already existing "Disconnected" signal which is synthesized when the
84 connection is terminated. This signal is generally useful but
85 particularly handy in combination with the "watch-bind" feature
86 described above. Synthesizing of this message has to be requested
87 explicitly through the new API call sd_bus_set_connected_signal(). In
88 addition a new call sd_bus_is_ready() has been added that checks
89 whether a connection is fully set up (i.e. between the "Connected" and
90 "Disconnected" signals).
92 * sd-bus gained two new calls sd_bus_request_name_async() and
93 sd_bus_release_name_async() for asynchronously registering bus
94 names. Similar, there is now sd_bus_add_match_async() for installing
95 a signal match asynchronously. All of elogind's own services have
96 been updated to make use of these calls. Doing these operations
97 asynchronously has two benefits: it reduces the risk of deadlocks in
98 case of cyclic dependencies between bus services, and it speeds up
99 service initialization since synchronization points for bus
100 round-trips are removed.
102 * sd-bus gained two new calls sd_bus_match_signal() and
103 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
104 and sd_bus_add_match_async() but instead of taking a D-Bus match
105 string take match fields as normal function parameters.
107 * sd-bus gained two new calls sd_bus_set_sender() and
108 sd_bus_message_set_sender() for setting the sender name of outgoing
109 messages (either for all outgoing messages or for just one specific
110 one). These calls are only useful in direct connections as on
111 brokered connections the broker fills in the sender anyway,
112 overwriting whatever the client filled in.
114 * sd-event gained a new pseudo-handle that may be specified on all API
115 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
116 used this refers to the default event loop object of the calling
117 thread. Note however that this does not implicitly allocate one —
118 which has to be done prior by using sd_event_default(). Similarly
119 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
120 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
121 to the default bus of the specified type of the calling thread. Here
122 too this does not implicitly allocate bus connection objects, this
123 has to be done prior with sd_bus_default() and friends.
125 * sd-event gained a new call pair
126 sd_event_source_{get|set}_io_fd_own(). This may be used to request
127 automatic closure of the file descriptor an IO event source watches
128 when the event source is destroyed.
130 * elogind-networkd gained support for natively configuring WireGuard
133 * In previous versions elogind synthesized user records both for the
134 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
135 internally. In order to simplify distribution-wide renames of the
136 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
137 new transitional flag file has been added: if
138 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
139 user and group record within the elogind codebase is disabled.
141 * elogind-notify gained a new --uid= option for selecting the source
142 user/UID to use for notification messages sent to the service
145 * journalctl gained a new --grep option to list only entries in which
146 the message matches a certain pattern. By default matching is case
147 insensitive if the pattern is lowercase, and case sensitive
148 otherwise. Option --case-sensitive=yes|no can be used to override
149 this an specify case sensitivity or case insensitivity.
151 * There's now a "elogind-analyze service-watchdogs" command for printing
152 the current state of the service runtime watchdog, and optionally
153 enabling or disabling the per-service watchdogs system-wide if given a
154 boolean argument (i.e. the concept you configure in WatchdogSec=), for
155 debugging purposes. There's also a kernel command line option
156 elogind.service_watchdogs= for controlling the same.
158 * Two new "log-level" and "log-target" options for elogind-analyze were
159 addded that merge the now deprecated get-log-level, set-log-level and
160 get-log-target, set-log-target pairs. The deprecated options are still
161 understood for backwards compatibility. The two new options print the
162 current value when no arguments are given, and set them when a
163 level/target is given as an argument.
165 * sysusers.d's "u" lines now optionally accept both a UID and a GID
166 specification, separated by a ":" character, in order to create users
167 where UID and GID do not match.
169 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
170 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
171 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
172 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
173 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
174 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
175 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
176 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
177 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
178 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
179 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
180 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
181 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
182 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
183 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
184 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
185 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
186 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
187 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
188 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
195 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
196 in v235 has been extended to also set the dummy.ko module option
197 numdummies=0, preventing the kernel from automatically creating
198 dummy0. All dummy interfaces must now be explicitly created.
200 * Unknown '%' specifiers in configuration files are now rejected. This
201 applies to units and tmpfiles.d configuration. Any percent characters
202 that are followed by a letter or digit that are not supposed to be
203 interpreted as the beginning of a specifier should be escaped by
204 doubling ("%%"). (So "size=5%" is still accepted, as well as
205 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
206 valid specifiers today.)
208 * systemd-resolved now maintains a new dynamic
209 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
210 recommended to make /etc/resolv.conf a symlink to it. This file
211 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
212 includes dynamically acquired search domains, achieving more correct
213 DNS resolution by software that bypasses local DNS APIs such as NSS.
215 * The "uaccess" udev tag has been dropped from /dev/kvm and
216 /dev/dri/renderD*. These devices now have the 0666 permissions by
217 default (but this may be changed at build-time). /dev/dri/renderD*
218 will now be owned by the "render" group along with /dev/kfd.
220 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
221 systemd-journal-gatewayd.service and
222 systemd-journal-upload.service. This means "nss-systemd" must be
223 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
224 services are resolved properly.
226 * In /etc/fstab two new mount options are now understood:
227 x-systemd.makefs and x-systemd.growfs. The former has the effect that
228 the configured file system is formatted before it is mounted, the
229 latter that the file system is resized to the full block device size
230 after it is mounted (i.e. if the file system is smaller than the
231 partition it resides on, it's grown). This is similar to the fsck
232 logic in /etc/fstab, and pulls in systemd-makefs@.service and
233 systemd-growfs@.service as necessary, similar to
234 systemd-fsck@.service. Resizing is currently only supported on ext4
237 * In systemd-networkd, the IPv6 RA logic now optionally may announce
238 DNS server and domain information.
240 * Support for the LUKS2 on-disk format for encrypted partitions has
241 been added. This requires libcryptsetup2 during compilation and
244 * The systemd --user instance will now signal "readiness" when its
245 basic.target unit has been reached, instead of when the run queue ran
246 empty for the first time.
248 * Tmpfiles.d with user configuration are now also supported.
249 systemd-tmpfiles gained a new --user switch, and snippets placed in
250 ~/.config/user-tmpfiles.d/ and corresponding directories will be
251 executed by systemd-tmpfiles --user running in the new
252 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
253 running in the user session.
255 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
256 %S resolves to the top-level state directory (/var/lib for the system
257 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
258 top-level cache directory (/var/cache for the system instance,
259 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
260 logs directory (/var/log for the system instance,
261 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
262 existing %t specifier, that resolves to the top-level runtime
263 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
266 * journalctl learnt a new parameter --output-fields= for limiting the
267 set of journal fields to output in verbose and JSON output modes.
269 * systemd-timesyncd's configuration file gained a new option
270 RootDistanceMaxSec= for setting the maximum root distance of servers
271 it'll use, as well as the new options PollIntervalMinSec= and
272 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
274 * bootctl gained a new command "list" for listing all available boot
275 menu items on systems that follow the boot loader specification.
277 * systemctl gained a new --dry-run switch that shows what would be done
278 instead of doing it, and is currently supported by the shutdown and
281 * ConditionSecurity= can now detect the TOMOYO security module.
283 * Unit file [Install] sections are now also respected in unit drop-in
284 files. This is intended to be used by drop-ins under /usr/lib/.
286 * systemd-firstboot may now also set the initial keyboard mapping.
288 * Udev "changed" events for devices which are exposed as systemd
289 .device units are now propagated to units specified in
290 ReloadPropagatedFrom= as reload requests.
292 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
293 unit template name (i.e. a name in the form of 'foobar@.service',
294 without the instance component between the '@' and - the '.'), then
295 the escaped sysfs path of the device is automatically used as the
298 * SystemCallFilter= in unit files has been extended so that an "errno"
299 can be specified individually for each system call. Example:
300 SystemCallFilter=~uname:EILSEQ.
302 * The cgroup delegation logic has been substantially updated. Delegate=
303 now optionally takes a list of controllers (instead of a boolean, as
304 before), which lists the controllers to delegate at least.
306 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
308 * A new LogLevelMax= setting configures the maximum log level any
309 process of the service may log at (i.e. anything with a lesser
310 priority than what is specified is automatically dropped). A new
311 LogExtraFields= setting allows configuration of additional journal
312 fields to attach to all log records generated by any of the unit's
315 * New StandardInputData= and StandardInputText= settings along with the
316 new option StandardInput=data may be used to configure textual or
317 binary data that shall be passed to the executed service process via
318 standard input, encoded in-line in the unit file.
320 * StandardInput=, StandardOutput= and StandardError= may now be used to
321 connect stdin/stdout/stderr of executed processes directly with a
322 file or AF_UNIX socket in the file system, using the new "file:" option.
324 * A new unit file option CollectMode= has been added, that allows
325 tweaking the garbage collection logic for units. It may be used to
326 tell systemd to garbage collect units that have failed automatically
327 (normally it only GCs units that exited successfully). systemd-run
328 and systemd-mount expose this new functionality with a new -G option.
330 * "machinectl bind" may now be used to bind mount non-directories
331 (i.e. regularfiles, devices, fifos, sockets).
333 * systemd-analyze gained a new verb "calendar" for validating and
334 testing calendar time specifications to use for OnCalendar= in timer
335 units. Besides validating the expression it will calculate the next
336 time the specified expression would elapse.
338 * In addition to the pre-existing FailureAction= unit file setting
339 there's now SuccessAction=, for configuring a shutdown action to
340 execute when a unit completes successfully. This is useful in
341 particular inside containers that shall terminate after some workload
342 has been completed. Also, both options are now supported for all unit
343 types, not just services.
345 * networkds's IP rule support gained two new options
346 IncomingInterface= and OutgoingInterface= for configuring the incoming
347 and outgoing interfaces of configured rules. systemd-networkd also
348 gained support for "vxcan" network devices.
350 * networkd gained a new setting RequiredForOnline=, taking a
351 boolean. If set, systemd-wait-online will take it into consideration
352 when determining that the system is up, otherwise it will ignore the
353 interface for this purpose.
355 * The sd_notify() protocol gained support for a new operation: with
356 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
357 store again, ahead of POLLHUP or POLLERR when they are removed
360 * A new document UIDS-GIDS.md has been added to the source tree, that
361 documents the UID/GID range and assignment assumptions and
362 requirements of systemd.
364 * The watchdog device PID 1 will ping may now be configured through the
365 WatchdogDevice= configuration file setting, or by setting the
366 systemd.watchdog_service= kernel commandline option.
368 * systemd-resolved's gained support for registering DNS-SD services on
369 the local network using MulticastDNS. Services may either be
370 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
371 the same dir below /run, /usr/lib), or through its D-Bus API.
373 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
374 extend the effective start, runtime, and stop time. The service must
375 continue to send EXTEND_TIMEOUT_USEC within the period specified to
376 prevent the service manager from making the service as timedout.
378 * elogind-resolved's DNSSEC support gained support for RFC 8080
379 (Ed25519 keys and signatures).
381 * The elogind-resolve command line tool gained a new set of options
382 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
383 --set-nta= and --revert to configure per-interface DNS configuration
384 dynamically during runtime. It's useful for pushing DNS information
385 into elogind-resolved from DNS hook scripts that various interface
386 managing software supports (such as pppd).
388 * elogind-nspawn gained a new --network-namespace-path= command line
389 option, which may be used to make a container join an existing
390 network namespace, by specifying a path to a "netns" file.
392 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
393 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
394 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
395 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
396 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
397 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
398 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
399 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
400 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
401 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
402 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
403 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
404 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
405 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
406 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
407 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
408 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
409 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
410 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
411 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
412 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
413 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
414 Jędrzejewski-Szmek, Zeal Jagannatha
420 * INCOMPATIBILITY: systemd-logind.service and other long-running
421 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
422 communication with the outside. This generally improves security of
423 the system, and is in almost all cases a safe and good choice, as
424 these services do not and should not provide any network-facing
425 functionality. However, systemd-logind uses the glibc NSS API to
426 query the user database. This creates problems on systems where NSS
427 is set up to directly consult network services for user database
428 lookups. In particular, this creates incompatibilities with the
429 "nss-nis" module, which attempts to directly contact the NIS/YP
430 network servers it is configured for, and will now consistently
431 fail. In such cases, it is possible to turn off IP sandboxing for
432 systemd-logind.service (set IPAddressDeny= in its [Service] section
433 to the empty string, via a .d/ unit file drop-in). Downstream
434 distributions might want to update their nss-nis packaging to include
435 such a drop-in snippet, accordingly, to hide this incompatibility
436 from the user. Another option is to make use of glibc's nscd service
437 to proxy such network requests through a privilege-separated, minimal
438 local caching daemon, or to switch to more modern technologies such
439 sssd, whose NSS hook-ups generally do not involve direct network
440 access. In general, we think it's definitely time to question the
441 implementation choices of nss-nis, i.e. whether it's a good idea
442 today to embed a network-facing loadable module into all local
443 processes that need to query the user database, including the most
444 trivial and benign ones, such as "ls". For more details about
445 IPAddressDeny= see below.
447 * A new modprobe.d drop-in is now shipped by default that sets the
448 bonding module option max_bonds=0. This overrides the kernel default,
449 to avoid conflicts and ambiguity as to whether or not bond0 should be
450 managed by systemd-networkd or not. This resolves multiple issues
451 with bond0 properties not being applied, when bond0 is configured
452 with systemd-networkd. Distributors may choose to not package this,
453 however in that case users will be prevented from correctly managing
454 bond0 interface using systemd-networkd.
456 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
457 which print the logging level and target of the system manager. They
458 complement the existing "set-log-level" and "set-log-target" verbs
459 used to change those values.
461 * journald.conf gained a new boolean setting ReadKMsg= which defaults
462 to on. If turned off kernel log messages will not be read by
463 systemd-journald or included in the logs. It also gained a new
464 setting LineMax= for configuring the maximum line length in
465 STDOUT/STDERR log streams. The new default for this value is 48K, up
466 from the previous hardcoded 2048.
468 * A new unit setting RuntimeDirectoryPreserve= has been added, which
469 allows more detailed control of what to do with a runtime directory
470 configured with RuntimeDirectory= (i.e. a directory below /run or
471 $XDG_RUNTIME_DIR) after a unit is stopped.
473 * The RuntimeDirectory= setting for units gained support for creating
474 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
475 one top-level directory.
477 * Units gained new options StateDirectory=, CacheDirectory=,
478 LogsDirectory= and ConfigurationDirectory= which are closely related
479 to RuntimeDirectory= but manage per-service directories below
480 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
481 possible to write unit files which when activated automatically gain
482 properly owned service specific directories in these locations, thus
483 making unit files self-contained and increasing compatibility with
484 stateless systems and factory reset where /etc or /var are
485 unpopulated at boot. Matching these new settings there's also
486 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
487 ConfigurationDirectoryMode= for configuring the access mode of these
488 directories. These settings are particularly useful in combination
489 with DynamicUser=yes as they provide secure, properly-owned,
490 writable, and stateful locations for storage, excluded from the
491 sandbox that such services live in otherwise.
493 * Automake support has been removed from this release. systemd is now
496 * systemd-journald will now aggressively cache client metadata during
497 runtime, speeding up log write performance under pressure. This comes
498 at a small price though: as much of the metadata is read
499 asynchronously from /proc/ (and isn't implicitly attached to log
500 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
501 metadata stored alongside a log entry might be slightly
502 out-of-date. Previously it could only be slightly newer than the log
503 message. The time window is small however, and given that the kernel
504 is unlikely to be improved anytime soon in this regard, this appears
507 * nss-myhostname/systemd-resolved will now by default synthesize an
508 A/AAAA resource record for the "_gateway" hostname, pointing to the
509 current default IP gateway. Previously it did that for the "gateway"
510 name, hampering adoption, as some distributions wanted to leave that
511 host name open for local use. The old behaviour may still be
512 requested at build time.
514 * systemd-networkd's [Address] section in .network files gained a new
515 Scope= setting for configuring the IP address scope. The [Network]
516 section gained a new boolean setting ConfigureWithoutCarrier= that
517 tells systemd-networkd to ignore link sensing when configuring the
518 device. The [DHCP] section gained a new Anonymize= boolean option for
519 turning on a number of options suggested in RFC 7844. A new
520 [RoutingPolicyRule] section has been added for configuring the IP
521 routing policy. The [Route] section has gained support for a new
522 Type= setting which permits configuring
523 blackhole/unreachable/prohibit routes.
525 * The [VRF] section in .netdev files gained a new Table= setting for
526 configuring the routing table to use. The [Tunnel] section gained a
527 new Independent= boolean field for configuring tunnels independent of
528 an underlying network interface. The [Bridge] section gained a new
529 GroupForwardMask= option for configuration of propagation of link
530 local frames between bridge ports.
532 * The WakeOnLan= setting in .link files gained support for a number of
533 new modes. A new TCP6SegmentationOffload= setting has been added for
534 configuring TCP/IPv6 hardware segmentation offload.
536 * The IPv6 RA sender implementation may now optionally send out RDNSS
537 and RDNSSL records to supply DNS configuration to peers.
539 * systemd-nspawn gained support for a new --system-call-filter= command
540 line option for adding and removing entries in the default system
541 call filter it applies. Moreover systemd-nspawn has been changed to
542 implement a system call whitelist instead of a blacklist.
544 * systemd-run gained support for a new --pipe command line option. If
545 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
546 are directly passed on to the activated transient service
547 executable. This allows invoking arbitrary processes as systemd
548 services (for example to take benefit of dependency management,
549 accounting management, resource management or log management that is
550 done automatically for services) — while still allowing them to be
551 integrated in a classic UNIX shell pipeline.
553 * When a service sends RELOAD=1 via sd_notify() and reload propagation
554 using ReloadPropagationTo= is configured, a reload is now propagated
555 to configured units. (Previously this was only done on explicitly
556 requested reloads, using "systemctl reload" or an equivalent
559 * For each service unit a restart counter is now kept: it is increased
560 each time the service is restarted due to Restart=, and may be
561 queried using "systemctl show -p NRestarts …".
563 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
564 @signal and @timer have been added, for usage with SystemCallFilter=
565 in unit files and the new --system-call-filter= command line option
566 of systemd-nspawn (see above).
568 * ExecStart= lines in unit files gained two new modifiers: when a
569 command line is prefixed with "!" the command will be executed as
570 configured, except for the credentials applied by
571 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
572 "+", but does still apply namespacing options unlike "+". There's
573 also "!!" now, which is mostly identical, but becomes a NOP on
574 systems that support ambient capabilities. This is useful to write
575 unit files that work with ambient capabilities where possible but
576 automatically fall back to traditional privilege dropping mechanisms
577 on systems where this is not supported.
579 * ListenNetlink= settings in socket units now support RDMA netlink
582 * A new unit file setting LockPersonality= has been added which permits
583 locking down the chosen execution domain ("personality") of a service
586 * A new special target "getty-pre.target" has been added, which is
587 ordered before all text logins, and may be used to order services
588 before textual logins acquire access to the console.
590 * systemd will now attempt to load the virtio-rng.ko kernel module very
591 early on if a VM environment supporting this is detected. This should
592 improve entropy during early boot in virtualized environments.
594 * A _netdev option is now supported in /etc/crypttab that operates in a
595 similar way as the same option in /etc/fstab: it permits configuring
596 encrypted devices that need to be ordered after the network is up.
597 Following this logic, two new special targets
598 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
599 added that are to cryptsetup.target what remote-fs.target and
600 remote-fs-pre.target are to local-fs.target.
602 * Service units gained a new UnsetEnvironment= setting which permits
603 unsetting specific environment variables for services that are
604 normally passed to it (for example in order to mask out locale
605 settings for specific services that can't deal with it).
607 * Units acquired a new boolean option IPAccounting=. When turned on, IP
608 traffic accounting (packet count as well as byte count) is done for
609 the service, and shown as part of "systemctl status" or "systemd-run
612 * Service units acquired two new options IPAddressAllow= and
613 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
614 for configuring a simple IP access control list for all sockets of
615 the unit. These options are available also on .slice and .socket
616 units, permitting flexible access list configuration for individual
617 services as well as groups of services (as defined by a slice unit),
618 including system-wide. Note that IP ACLs configured this way are
619 enforced on every single IPv4 and IPv6 socket created by any process
620 of the service unit, and apply to ingress as well as egress traffic.
622 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
623 structured log message is generated each time the unit is stopped,
624 containing information about the consumed resources of this
627 * A new setting KeyringMode= has been added to unit files, which may be
628 used to control how the kernel keyring is set up for executed
631 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
632 "systemctl kexec" and "systemctl exit" are now always asynchronous in
633 behaviour (that is: these commands return immediately after the
634 operation was enqueued instead of waiting for the operation to
635 complete). Previously, "systemctl poweroff" and "systemctl reboot"
636 were asynchronous on systems using systemd-logind (i.e. almost
637 always, and like they were on sysvinit), and the other three commands
638 were unconditionally synchronous. With this release this is cleaned
639 up, and callers will see the same asynchronous behaviour on all
640 systems for all five operations.
642 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
645 * .timer units now accept calendar specifications in other timezones
646 than UTC or the local timezone.
648 * The tmpfiles snippet var.conf has been changed to create
649 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
650 the "utmp" group already, and it appears to be generally understood
651 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
652 databases. Previously this was implemented correctly for all these
653 databases excepts btmp, which has been opened up like this now
654 too. Note that while the other databases are world-readable
655 (i.e. 0644), btmp is not and remains more restrictive.
657 * The systemd-resolve tool gained a new --reset-server-features
658 switch. When invoked like this systemd-resolved will forget
659 everything it learnt about the features supported by the configured
660 upstream DNS servers, and restarts the feature probing logic on the
661 next resolver look-up for them at the highest feature level
664 * The status dump systemd-resolved sends to the logs upon receiving
665 SIGUSR1 now also includes information about all DNS servers it is
666 configured to use, and the features levels it probed for them.
668 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
669 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
670 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
671 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
672 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
673 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
674 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
675 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
676 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
677 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
678 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
679 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
680 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
681 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
682 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
683 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
684 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
685 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
686 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
687 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
693 * Meson is now supported as build system in addition to Automake. It is
694 our plan to remove Automake in one of our next releases, so that
695 Meson becomes our exclusive build system. Hence, please start using
696 the Meson build system in your downstream packaging. There's plenty
697 of documentation around how to use Meson, the extremely brief
700 ./autogen.sh && ./configure && make && sudo make install
704 meson build && ninja -C build && sudo ninja -C build install
706 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
707 which permits configuring a timeout on the time a job is
708 running. This is particularly useful for setting timeouts on jobs for
711 * Unit files gained two new options ConditionUser= and ConditionGroup=
712 for conditionalizing units based on the identity of the user/group
713 running a systemd user instance.
715 * systemd-networkd now understands a new FlowLabel= setting in the
716 [VXLAN] section of .network files, as well as a Priority= in
717 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
718 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
719 gained support for configuration of GENEVE links, and IPv6 address
720 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
722 * .link files now understand a new Port= setting.
724 * systemd-networkd's DHCP support gained support for DHCP option 119
725 (domain search list).
727 * systemd-networkd gained support for serving IPv6 address ranges using
728 the Router Advertisment protocol. The new .network configuration
729 section [IPv6Prefix] may be used to configure the ranges to
730 serve. This is implemented based on a new, minimal, native server
731 implementation of RA.
733 * journalctl's --output= switch gained support for a new parameter
734 "short-iso-precise" for a mode where timestamps are shown as precise
737 * systemd-udevd's "net_id" builtin may now generate stable network
738 interface names from IBM PowerVM VIO devices as well as ACPI platform
741 * MulticastDNS support in systemd-resolved may now be explicitly
742 enabled/disabled using the new MulticastDNS= configuration file
745 * systemd-resolved may now optionally use libidn2 instead of the libidn
746 for processing internationalized domain names. Support for libidn2
747 should be considered experimental and should not be enabled by
750 * "machinectl pull-tar" and related call may now do verification of
751 downloaded images using SUSE-style .sha256 checksum files in addition
752 to the already existing support for validating using Ubuntu-style
755 * sd-bus gained support for a new sd_bus_message_appendv() call which
756 is va_list equivalent of sd_bus_message_append().
758 * sd-boot gained support for validating images using SHIM/MOK.
760 * The SMACK code learnt support for "onlycap".
762 * systemd-mount --umount is now much smarter in figuring out how to
763 properly unmount a device given its mount or device path.
765 * The code to call libnss_dns as a fallback from libnss_resolve when
766 the communication with systemd-resolved fails was removed. This
767 fallback was redundant and interfered with the [!UNAVAIL=return]
768 suffix. See nss-resolve(8) for the recommended configuration.
770 * systemd-logind may now be restarted without losing state. It stores
771 the file descriptors for devices it manages in the system manager
772 using the FDSTORE= mechanism. Please note that further changes in
773 other components may be required to make use of this (for example
774 Xorg has code to listen for stops of systemd-logind and terminate
775 itself when logind is stopped or restarted, in order to avoid using
776 stale file descriptors for graphical devices, which is now
777 counterproductive and must be reverted in order for restarts of
778 systemd-logind to be safe. See
779 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
781 * All kernel install plugins are called with the environment variable
782 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
783 /etc/machine-id. If the file is missing or empty, the variable is
784 empty and BOOT_DIR_ABS is the path of a temporary directory which is
785 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
786 is empty, all plugins should not put anything in BOOT_DIR_ABS.
788 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
789 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
790 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
791 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
792 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
793 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
794 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
795 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
796 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
797 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
798 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
799 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
800 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
801 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
802 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
803 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
804 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
805 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
806 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
807 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
808 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
809 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
810 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
811 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
812 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
813 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
814 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
815 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
816 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
823 * The "hybrid" control group mode has been modified to improve
824 compatibility with "legacy" cgroups-v1 setups. Specifically, the
825 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
826 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
827 cgroups-v1 hierarchy), the only externally visible change being that
828 the cgroups-v2 hierarchy is also mounted, to
829 /sys/fs/cgroup/unified. This should provide a large degree of
830 compatibility with "legacy" cgroups-v1, while taking benefit of the
831 better management capabilities of cgroups-v2.
833 * The default control group setup mode may be selected both a boot-time
834 via a set of kernel command line parameters (specifically:
835 systemd.unified_cgroup_hierarchy= and
836 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
837 default selected on the configure command line
838 (--with-default-hierarchy=). The upstream default is "hybrid"
839 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
840 this will change in a future systemd version to be "unified" (pure
841 cgroups-v2 mode). The third option for the compile time option is
842 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
843 distributions to default to "hybrid" mode for release distributions,
844 starting with v233. We recommend "unified" for development
845 distributions (specifically: distributions such as Fedora's rawhide)
846 as that's where things are headed in the long run. Use "legacy" for
847 greatest stability and compatibility only.
849 * Note one current limitation of "unified" and "hybrid" control group
850 setup modes: the kernel currently does not permit the systemd --user
851 instance (i.e. unprivileged code) to migrate processes between two
852 disconnected cgroup subtrees, even if both are managed and owned by
853 the user. This effectively means "systemd-run --user --scope" doesn't
854 work when invoked from outside of any "systemd --user" service or
855 scope. Specifically, it is not supported from session scopes. We are
856 working on fixing this in a future systemd version. (See #3388 for
857 further details about this.)
859 * DBus policy files are now installed into /usr rather than /etc. Make
860 sure your system has dbus >= 1.9.18 running before upgrading to this
861 version, or override the install path with --with-dbuspolicydir= .
863 * All python scripts shipped with systemd (specifically: the various
864 tests written in Python) now require Python 3.
866 * systemd unit tests can now run standalone (without the source or
867 build directories), and can be installed into /usr/lib/systemd/tests/
868 with 'make install-tests'.
870 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
871 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
874 * Support for the %c, %r, %R specifiers in unit files has been
875 removed. Specifiers are not supposed to be dependent on configuration
876 in the unit file itself (so that they resolve the same regardless
877 where used in the unit files), but these specifiers were influenced
878 by the Slice= option.
880 * The shell invoked by debug-shell.service now defaults to /bin/sh in
881 all cases. If distributions want to use a different shell for this
882 purpose (for example Fedora's /sbin/sushell) they need to specify
883 this explicitly at configure time using --with-debug-shell=.
885 * The confirmation spawn prompt has been reworked to offer the
888 (c)ontinue, proceed without asking anymore
889 (D)ump, show the state of the unit
890 (f)ail, don't execute the command and pretend it failed
892 (i)nfo, show a short summary of the unit
893 (j)obs, show jobs that are in progress
894 (s)kip, don't execute the command and pretend it succeeded
895 (y)es, execute the command
897 The 'n' choice for the confirmation spawn prompt has been removed,
898 because its meaning was confusing.
900 The prompt may now also be redirected to an alternative console by
901 specifying the console as parameter to systemd.confirm_spawn=.
903 * Services of Type=notify require a READY=1 notification to be sent
904 during startup. If no such message is sent, the service now fails,
905 even if the main process exited with a successful exit code.
907 * Services that fail to start up correctly now always have their
908 ExecStopPost= commands executed. Previously, they'd enter "failed"
909 state directly, without executing these commands.
911 * The option MulticastDNS= of network configuration files has acquired
912 an actual implementation. With MulticastDNS=yes a host can resolve
913 names of remote hosts and reply to mDNS A and AAAA requests.
915 * When units are about to be started an additional check is now done to
916 ensure that all dependencies of type BindsTo= (when used in
917 combination with After=) have been started.
919 * systemd-analyze gained a new verb "syscall-filter" which shows which
920 system call groups are defined for the SystemCallFilter= unit file
921 setting, and which system calls they contain.
923 * A new system call filter group "@filesystem" has been added,
924 consisting of various file system related system calls. Group
925 "@reboot" has been added, covering reboot, kexec and shutdown related
926 calls. Finally, group "@swap" has been added covering swap
927 configuration related calls.
929 * A new unit file option RestrictNamespaces= has been added that may be
930 used to restrict access to the various process namespace types the
931 Linux kernel provides. Specifically, it may be used to take away the
932 right for a service unit to create additional file system, network,
933 user, and other namespaces. This sandboxing option is particularly
934 relevant due to the high amount of recently discovered namespacing
935 related vulnerabilities in the kernel.
937 * systemd-udev's .link files gained support for a new AutoNegotiation=
938 setting for configuring Ethernet auto-negotiation.
940 * systemd-networkd's .network files gained support for a new
941 ListenPort= setting in the [DHCP] section to explicitly configure the
942 UDP client port the DHCP client shall listen on.
944 * .network files gained a new Unmanaged= boolean setting for explicitly
945 excluding one or more interfaces from management by systemd-networkd.
947 * The systemd-networkd ProxyARP= option has been renamed to
948 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
949 renamed to ReduceARPProxy=. The old names continue to be available
952 * systemd-networkd gained support for configuring IPv6 Proxy NDP
953 addresses via the new IPv6ProxyNDPAddress= .network file setting.
955 * systemd-networkd's bonding device support gained support for two new
956 configuration options ActiveSlave= and PrimarySlave=.
958 * The various options in the [Match] section of .network files gained
959 support for negative matching.
961 * New systemd-specific mount options are now understood in /etc/fstab:
963 x-systemd.mount-timeout= may be used to configure the maximum
964 permitted runtime of the mount command.
966 x-systemd.device-bound may be set to bind a mount point to its
967 backing device unit, in order to automatically remove a mount point
968 if its backing device is unplugged. This option may also be
969 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
970 on the block device, which is now automatically set for all CDROM
971 drives, so that mounted CDs are automatically unmounted when they are
972 removed from the drive.
974 x-systemd.after= and x-systemd.before= may be used to explicitly
975 order a mount after or before another unit or mount point.
977 * Enqueued start jobs for device units are now automatically garbage
978 collected if there are no jobs waiting for them anymore.
980 * systemctl list-jobs gained two new switches: with --after, for every
981 queued job the jobs it's waiting for are shown; with --before the
982 jobs which it's blocking are shown.
984 * systemd-nspawn gained support for ephemeral boots from disk images
985 (or in other words: --ephemeral and --image= may now be
986 combined). Moreover, ephemeral boots are now supported for normal
987 directories, even if the backing file system is not btrfs. Of course,
988 if the file system does not support file system snapshots or
989 reflinks, the initial copy operation will be relatively expensive, but
990 this should still be suitable for many use cases.
992 * Calendar time specifications in .timer units now support
993 specifications relative to the end of a month by using "~" instead of
994 "-" as separator between month and day. For example, "*-02~03" means
995 "the third last day in February". In addition a new syntax for
996 repeated events has been added using the "/" character. For example,
997 "9..17/2:00" means "every two hours from 9am to 5pm".
999 * systemd-socket-proxyd gained a new parameter --connections-max= for
1000 configuring the maximum number of concurrent connections.
1002 * sd-id128 gained a new API for generating unique IDs for the host in a
1003 way that does not leak the machine ID. Specifically,
1004 sd_id128_get_machine_app_specific() derives an ID based on the
1005 machine ID a in well-defined, non-reversible, stable way. This is
1006 useful whenever an identifier for the host is needed but where the
1007 identifier shall not be useful to identify the system beyond the
1008 scope of the application itself. (Internally this uses HMAC-SHA256 as
1009 keyed hash function using the machine ID as input.)
1011 * NotifyAccess= gained a new supported value "exec". When set
1012 notifications are accepted from all processes systemd itself invoked,
1013 including all control processes.
1015 * .nspawn files gained support for defining overlay mounts using the
1016 Overlay= and OverlayReadOnly= options. Previously this functionality
1017 was only available on the systemd-nspawn command line.
1019 * systemd-nspawn's --bind= and --overlay= options gained support for
1020 bind/overlay mounts whose source lies within the container tree by
1021 prefixing the source path with "+".
1023 * systemd-nspawn's --bind= and --overlay= options gained support for
1024 automatically allocating a temporary source directory in /var/tmp
1025 that is removed when the container dies. Specifically, if the source
1026 directory is specified as empty string this mechanism is selected. An
1027 example usage is --overlay=+/var::/var, which creates an overlay
1028 mount based on the original /var contained in the image, overlayed
1029 with a temporary directory in the host's /var/tmp. This way changes
1030 to /var are automatically flushed when the container shuts down.
1032 * systemd-nspawn --image= option does now permit raw file system block
1033 devices (in addition to images containing partition tables, as
1036 * The disk image dissection logic in systemd-nspawn gained support for
1037 automatically setting up LUKS encrypted as well as Verity protected
1038 partitions. When a container is booted from an encrypted image the
1039 passphrase is queried at start-up time. When a container with Verity
1040 data is started, the root hash is search in a ".roothash" file
1041 accompanying the disk image (alternatively, pass the root hash via
1042 the new --root-hash= command line option).
1044 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1045 be used to dissect disk images the same way as systemd-nspawn does
1046 it, following the Bootable Partition Specification. It may even be
1047 used to mount disk images with complex partition setups (including
1048 LUKS and Verity partitions) to a local host directory, in order to
1049 inspect them. This tool is not considered public API (yet), and is
1050 thus not installed into /usr/bin. Please do not rely on its
1051 existence, since it might go away or be changed in later systemd
1054 * A new generator "systemd-verity-generator" has been added, similar in
1055 style to "systemd-cryptsetup-generator", permitting automatic setup of
1056 Verity root partitions when systemd boots up. In order to make use of
1057 this your partition setup should follow the Discoverable Partitions
1058 Specification, and the GPT partition ID of the root file system
1059 partition should be identical to the upper 128bit of the Verity root
1060 hash. The GPT partition ID of the Verity partition protecting it
1061 should be the lower 128bit of the Verity root hash. If the partition
1062 image follows this model it is sufficient to specify a single
1063 "roothash=" kernel command line argument to both configure which root
1064 image and verity partition to use as well as the root hash for
1065 it. Note that systemd-nspawn's Verity support follows the same
1066 semantics, meaning that disk images with proper Verity data in place
1067 may be booted in containers with systemd-nspawn as well as on
1068 physical systems via the verity generator. Also note that the "mkosi"
1069 tool available at https://github.com/systemd/mkosi has been updated
1070 to generate Verity protected disk images following this scheme. In
1071 fact, it has been updated to generate disk images that optionally
1072 implement a complete UEFI SecureBoot trust chain, involving a signed
1073 kernel and initrd image that incorporates such a root hash as well as
1074 a Verity-enabled root partition.
1076 * The hardware database (hwdb) udev supports has been updated to carry
1077 accelerometer quirks.
1079 * All system services are now run with a fresh kernel keyring set up
1080 for them. The invocation ID is stored by default in it, thus
1081 providing a safe, non-overridable way to determine the invocation
1084 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1085 options for bind mounting arbitrary paths in a service-specific
1086 way. When these options are used, arbitrary host or service files and
1087 directories may be mounted to arbitrary locations in the service's
1090 * Documentation has been added that lists all of systemd's low-level
1091 environment variables:
1093 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
1095 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1096 whether a specific socket file descriptor matches a specified socket
1099 * systemd-firstboot has been updated to check for the
1100 systemd.firstboot= kernel command line option. It accepts a boolean
1101 and when set to false the first boot questions are skipped.
1103 * systemd-fstab-generator has been updated to check for the
1104 systemd.volatile= kernel command line option, which either takes an
1105 optional boolean parameter or the special value "state". If used the
1106 system may be booted in a "volatile" boot mode. Specifically,
1107 "systemd.volatile" is used, the root directory will be mounted as
1108 tmpfs, and only /usr is mounted from the actual root file system. If
1109 "systemd.volatile=state" is used, the root directory will be mounted
1110 as usual, but /var is mounted as tmpfs. This concept provides similar
1111 functionality as systemd-nspawn's --volatile= option, but provides it
1112 on physical boots. Use this option for implementing stateless
1113 systems, or testing systems with all state and/or configuration reset
1114 to the defaults. (Note though that many distributions are not
1115 prepared to boot up without a populated /etc or /var, though.)
1117 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1118 partitions. Previously it only supported LUKS encrypted partitions
1119 for all other uses, except for the root partition itself.
1121 * Socket units gained support for listening on AF_VSOCK sockets for
1122 communication in virtualized QEMU environments.
1124 * The "configure" script gained a new option --with-fallback-hostname=
1125 for specifying the fallback hostname to use if none is configured in
1126 /etc/hostname. For example, by specifying
1127 --with-fallback-hostname=fedora it is possible to default to a
1128 hostname of "fedora" on pristine installations.
1130 * systemd-cgls gained support for a new --unit= switch for listing only
1131 the control groups of a specific unit. Similar --user-unit= has been
1132 added for listing only the control groups of a specific user unit.
1134 * systemd-mount gained a new --umount switch for unmounting a mount or
1135 automount point (and all mount/automount points below it).
1137 * systemd will now refuse full configuration reloads (via systemctl
1138 daemon-reload and related calls) unless at least 16MiB of free space
1139 are available in /run. This is a safety precaution in order to ensure
1140 that generators can safely operate after the reload completed.
1142 * A new unit file option RootImage= has been added, which has a similar
1143 effect as RootDirectory= but mounts the service's root directory from
1144 a disk image instead of plain directory. This logic reuses the same
1145 image dissection and mount logic that systemd-nspawn already uses,
1146 and hence supports any disk images systemd-nspawn supports, including
1147 those following the Discoverable Partition Specification, as well as
1148 Verity enabled images. This option enables systemd to run system
1149 services directly off disk images acting as resource bundles,
1150 possibly even including full integrity data.
1152 * A new MountAPIVFS= unit file option has been added, taking a boolean
1153 argument. If enabled /proc, /sys and /dev (collectively called the
1154 "API VFS") will be mounted for the service. This is only relevant if
1155 RootDirectory= or RootImage= is used for the service, as these mounts
1156 are of course in place in the host mount namespace anyway.
1158 * systemd-nspawn gained support for a new --pivot-root= switch. If
1159 specified the root directory within the container image is pivoted to
1160 the specified mount point, while the original root disk is moved to a
1161 different place. This option enables booting of ostree images
1162 directly with systemd-nspawn.
1164 * The systemd build scripts will no longer complain if the NTP server
1165 addresses are not changed from the defaults. Google now supports
1166 these NTP servers officially. We still recommend downstreams to
1167 properly register an NTP pool with the NTP pool project though.
1169 * coredumpctl gained a new "--reverse" option for printing the list
1170 of coredumps in reverse order.
1172 * coredumpctl will now show additional information about truncated and
1173 inaccessible coredumps, as well as coredumps that are still being
1174 processed. It also gained a new --quiet switch for suppressing
1175 additional informational message in its output.
1177 * coredumpctl gained support for only showing coredumps newer and/or
1178 older than specific timestamps, using the new --since= and --until=
1179 options, reminiscent of journalctl's options by the same name.
1181 * The systemd-coredump logic has been improved so that it may be reused
1182 to collect backtraces in non-compiled languages, for example in
1183 scripting languages such as Python.
1185 * machinectl will now show the UID shift of local containers, if user
1186 namespacing is enabled for them.
1188 * systemd will now optionally run "environment generator" binaries at
1189 configuration load time. They may be used to add environment
1190 variables to the environment block passed to services invoked. One
1191 user environment generator is shipped by default that sets up
1192 environment variables based on files dropped into /etc/environment.d
1193 and ~/.config/environment.d/.
1195 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1198 * hostnamed has been updated to report a new chassis type of
1199 "convertible" to cover "foldable" laptops that can both act as a
1200 tablet and as a laptop, such as various Lenovo Yoga devices.
1202 Contributions from: Adrián López, Alexander Galanin, Alexander
1203 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1204 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1205 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1206 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1207 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1208 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1209 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1210 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1211 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1212 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1213 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1214 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1215 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1216 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1217 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1218 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1219 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1220 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1221 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1222 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1223 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1224 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1225 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1226 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1227 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1228 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1229 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1232 — Berlin, 2017-03-01
1236 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1237 RestrictAddressFamilies= enabled. These sandboxing options should
1238 generally be compatible with the various external udev call-out
1239 binaries we are aware of, however there may be exceptions, in
1240 particular when exotic languages for these call-outs are used. In
1241 this case, consider turning off these settings locally.
1243 * The new RemoveIPC= option can be used to remove IPC objects owned by
1244 the user or group of a service when that service exits.
1246 * The new ProtectKernelModules= option can be used to disable explicit
1247 load and unload operations of kernel modules by a service. In
1248 addition access to /usr/lib/modules is removed if this option is set.
1250 * ProtectSystem= option gained a new value "strict", which causes the
1251 whole file system tree with the exception of /dev, /proc, and /sys,
1252 to be remounted read-only for a service.
1254 * The new ProtectKernelTunables= option can be used to disable
1255 modification of configuration files in /sys and /proc by a service.
1256 Various directories and files are remounted read-only, so access is
1257 restricted even if the file permissions would allow it.
1259 * The new ProtectControlGroups= option can be used to disable write
1260 access by a service to /sys/fs/cgroup.
1262 * Various systemd services have been hardened with
1263 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1264 RestrictAddressFamilies=.
1266 * Support for dynamically creating users for the lifetime of a service
1267 has been added. If DynamicUser=yes is specified, user and group IDs
1268 will be allocated from the range 61184..65519 for the lifetime of the
1269 service. They can be resolved using the new nss-systemd.so NSS
1270 module. The module must be enabled in /etc/nsswitch.conf. Services
1271 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1272 any resources allocated by the service will be cleaned up when the
1273 service exits. They also have ProtectHome=read-only and
1274 ProtectSystem=strict enabled, so they are not able to make any
1275 permanent modifications to the system.
1277 * The nss-systemd module also always resolves root and nobody, making
1278 it possible to have no /etc/passwd or /etc/group files in minimal
1279 container or chroot environments.
1281 * Services may be started with their own user namespace using the new
1282 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1283 under which the service is running are mapped. All other users are
1286 * Support for the cgroup namespace has been added to systemd-nspawn. If
1287 supported by kernel, the container system started by systemd-nspawn
1288 will have its own view of the cgroup hierarchy. This new behaviour
1289 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1291 * The new MemorySwapMax= option can be used to limit the maximum swap
1292 usage under the unified cgroup hierarchy.
1294 * Support for the CPU controller in the unified cgroup hierarchy has
1295 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1296 options. This controller requires out-of-tree patches for the kernel
1297 and the support is provisional.
1299 * Mount and automount units may now be created transiently
1300 (i.e. dynamically at runtime via the bus API, instead of requiring
1301 unit files in the file system).
1303 * systemd-mount is a new tool which may mount file systems – much like
1304 mount(8), optionally pulling in additional dependencies through
1305 transient .mount and .automount units. For example, this tool
1306 automatically runs fsck on a backing block device before mounting,
1307 and allows the automount logic to be used dynamically from the
1308 command line for establishing mount points. This tool is particularly
1309 useful when dealing with removable media, as it will ensure fsck is
1310 run – if necessary – before the first access and that the file system
1311 is quickly unmounted after each access by utilizing the automount
1312 logic. This maximizes the chance that the file system on the
1313 removable media stays in a clean state, and if it isn't in a clean
1314 state is fixed automatically.
1316 * LazyUnmount=yes option for mount units has been added to expose the
1317 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1320 * /efi will be used as the mount point of the EFI boot partition, if
1321 the directory is present, and the mount point was not configured
1322 through other means (e.g. fstab). If /efi directory does not exist,
1323 /boot will be used as before. This makes it easier to automatically
1324 mount the EFI partition on systems where /boot is used for something
1327 * When operating on GPT disk images for containers, systemd-nspawn will
1328 now mount the ESP to /boot or /efi according to the same rules as PID
1329 1 running on a host. This allows tools like "bootctl" to operate
1330 correctly within such containers, in order to make container images
1331 bootable on physical systems.
1333 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1335 * Two new user session targets have been added to support running
1336 graphical sessions under the systemd --user instance:
1337 graphical-session.target and graphical-session-pre.target. See
1338 systemd.special(7) for a description of how those targets should be
1341 * The vconsole initialization code has been significantly reworked to
1342 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1343 support unicode keymaps. Font and keymap configuration will now be
1344 copied to all allocated virtual consoles.
1346 * FreeBSD's bhyve virtualization is now detected.
1348 * Information recorded in the journal for core dumps now includes the
1349 contents of /proc/mountinfo and the command line of the process at
1350 the top of the process hierarchy (which is usually the init process
1353 * systemd-journal-gatewayd learned the --directory= option to serve
1354 files from the specified location.
1356 * journalctl --root=… can be used to peruse the journal in the
1357 /var/log/ directories inside of a container tree. This is similar to
1358 the existing --machine= option, but does not require the container to
1361 * The hardware database has been extended to support
1362 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1365 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1366 specify the click rate for mice which include a horizontal wheel with
1367 a click rate that is different than the one for the vertical wheel.
1369 * systemd-run gained a new --wait option that makes service execution
1370 synchronous. (Specifically, the command will not return until the
1371 specified service binary exited.)
1373 * systemctl gained a new --wait option that causes the start command to
1374 wait until the units being started have terminated again.
1376 * A new journal output mode "short-full" has been added which displays
1377 timestamps with abbreviated English day names and adds a timezone
1378 suffix. Those timestamps include more information than the default
1379 "short" output mode, and can be passed directly to journalctl's
1380 --since= and --until= options.
1382 * /etc/resolv.conf will be bind-mounted into containers started by
1383 systemd-nspawn, if possible, so any changes to resolv.conf contents
1384 are automatically propagated to the container.
1386 * The number of instances for socket-activated services originating
1387 from a single IP address can be limited with
1388 MaxConnectionsPerSource=, extending the existing setting of
1391 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1394 * .netdev and .network configuration can now be extended through
1397 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1398 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1399 can be enabled and disabled using the new UDPSegmentationOffload=,
1400 TCPSegmentationOffload=, GenericSegmentationOffload=,
1401 GenericReceiveOffload=, LargeReceiveOffload= options in the
1402 [Link] section of .link files.
1404 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1405 Port VLAN ID can be configured for bridge devices using the new STP=,
1406 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1407 section of .netdev files.
1409 * The route table to which routes received over DHCP or RA should be
1410 added can be configured with the new RouteTable= option in the [DHCP]
1411 and [IPv6AcceptRA] sections of .network files.
1413 * The Address Resolution Protocol can be disabled on links managed by
1414 systemd-networkd using the ARP=no setting in the [Link] section of
1417 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1418 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1419 encode information about the result and exit codes of the current
1420 service runtime cycle.
1422 * systemd-sysctl will now configure kernel parameters in the order
1423 they occur in the configuration files. This matches what sysctl
1424 has been traditionally doing.
1426 * kernel-install "plugins" that are executed to perform various
1427 tasks after a new kernel is added and before an old one is removed
1428 can now return a special value to terminate the procedure and
1429 prevent any later plugins from running.
1431 * Journald's SplitMode=login setting has been deprecated. It has been
1432 removed from documentation, and its use is discouraged. In a future
1433 release it will be completely removed, and made equivalent to current
1434 default of SplitMode=uid.
1436 * Storage=both option setting in /etc/systemd/coredump.conf has been
1437 removed. With fast LZ4 compression storing the core dump twice is not
1440 * The --share-system systemd-nspawn option has been replaced with an
1441 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1442 this functionality is discouraged. In addition the variables
1443 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1444 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1445 individual namespaces.
1447 * "machinectl list" now shows the IP address of running containers in
1448 the output, as well as OS release information.
1450 * "loginctl list" now shows the TTY of each session in the output.
1452 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1453 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1454 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1455 tracking objects in a "recursive" mode, where a single client can be
1456 counted multiple times, if it takes multiple references.
1458 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1459 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1460 process using sd-bus automatically exit if the bus connection is
1463 * Bus clients of the service manager may now "pin" loaded units into
1464 memory, by taking an explicit reference on them. This is useful to
1465 ensure the client can retrieve runtime data about the service even
1466 after the service completed execution. Taking such a reference is
1467 available only for privileged clients and should be helpful to watch
1468 running services in a race-free manner, and in particular collect
1469 information about exit statuses and results.
1471 * The nss-resolve module has been changed to strictly return UNAVAIL
1472 when communication via D-Bus with resolved failed, and NOTFOUND when
1473 a lookup completed but was negative. This means it is now possible to
1474 neatly configure fallbacks using nsswitch.conf result checking
1475 expressions. Taking benefit of this, the new recommended
1476 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1478 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1480 * A new setting CtrlAltDelBurstAction= has been added to
1481 /etc/systemd/system.conf which may be used to configure the precise
1482 behaviour if the user on the console presses Ctrl-Alt-Del more often
1483 than 7 times in 2s. Previously this would unconditionally result in
1484 an expedited, immediate reboot. With this new setting the precise
1485 operation may be configured in more detail, and also turned off
1488 * In .netdev files two new settings RemoteChecksumTx= and
1489 RemoteChecksumRx= are now understood that permit configuring the
1490 remote checksumming logic for VXLAN networks.
1492 * The service manager learnt a new "invocation ID" concept for invoked
1493 services. Each runtime cycle of a service will get a new invocation
1494 ID (a 128bit random UUID) assigned that identifies the current
1495 run of the service uniquely and globally. A new invocation ID
1496 is generated each time a service starts up. The journal will store
1497 the invocation ID of a service along with any logged messages, thus
1498 making the invocation ID useful for matching the online runtime of a
1499 service with the offline log data it generated in a safe way without
1500 relying on synchronized timestamps. In many ways this new service
1501 invocation ID concept is similar to the kernel's boot ID concept that
1502 uniquely and globally identifies the runtime of each boot. The
1503 invocation ID of a service is passed to the service itself via an
1504 environment variable ($INVOCATION_ID). A new bus call
1505 GetUnitByInvocationID() has been added that is similar to GetUnit()
1506 but instead of retrieving the bus path for a unit by its name
1507 retrieves it by its invocation ID. The returned path is valid only as
1508 long as the passed invocation ID is current.
1510 * systemd-resolved gained a new "DNSStubListener" setting in
1511 resolved.conf. It either takes a boolean value or the special values
1512 "udp" and "tcp", and configures whether to enable the stub DNS
1513 listener on 127.0.0.53:53.
1515 * IP addresses configured via networkd may now carry additional
1516 configuration settings supported by the kernel. New options include:
1517 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1518 PrefixRoute=, AutoJoin=.
1520 * The PAM configuration fragment file for "user@.service" shipped with
1521 systemd (i.e. the --user instance of systemd) has been stripped to
1522 the minimum necessary to make the system boot. Previously, it
1523 contained Fedora-specific stanzas that did not apply to other
1524 distributions. It is expected that downstream distributions add
1525 additional configuration lines, matching their needs to this file,
1526 using it only as rough template of what systemd itself needs. Note
1527 that this reduced fragment does not even include an invocation of
1528 pam_limits which most distributions probably want to add, even though
1529 systemd itself does not need it. (There's also the new build time
1530 option --with-pamconfdir=no to disable installation of the PAM
1533 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1534 capability is now also dropped from its set (in addition to
1535 CAP_SYS_MKNOD as before).
1537 * In service unit files it is now possible to connect a specific named
1538 file descriptor with stdin/stdout/stdout of an executed service. The
1539 name may be specified in matching .socket units using the
1540 FileDescriptorName= setting.
1542 * A number of journal settings may now be configured on the kernel
1543 command line. Specifically, the following options are now understood:
1544 systemd.journald.max_level_console=,
1545 systemd.journald.max_level_store=,
1546 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1547 systemd.journald.max_level_wall=.
1549 * "systemctl is-enabled --full" will now show by which symlinks a unit
1550 file is enabled in the unit dependency tree.
1552 * Support for VeraCrypt encrypted partitions has been added to the
1553 "cryptsetup" logic and /etc/crypttab.
1555 * systemd-detect-virt gained support for a new --private-users switch
1556 that checks whether the invoking processes are running inside a user
1557 namespace. Similar, a new special value "private-users" for the
1558 existing ConditionVirtualization= setting has been added, permitting
1559 skipping of specific units in user namespace environments.
1561 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1562 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1563 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1564 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1565 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1566 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1567 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1568 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1569 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1570 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1571 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1572 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1573 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1574 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1575 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1576 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1577 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1578 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1579 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1580 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1581 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1582 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1583 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1584 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1585 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1586 Jędrzejewski-Szmek, Zeal Jagannatha
1588 — Santa Fe, 2016-11-03
1592 * In service units the various ExecXYZ= settings have been extended
1593 with an additional special character as first argument of the
1594 assigned value: if the character '+' is used the specified command
1595 line it will be run with full privileges, regardless of User=,
1596 Group=, CapabilityBoundingSet= and similar options. The effect is
1597 similar to the existing PermissionsStartOnly= option, but allows
1598 configuration of this concept for each executed command line
1601 * Services may now alter the service watchdog timeout at runtime by
1602 sending a WATCHDOG_USEC= message via sd_notify().
1604 * MemoryLimit= and related unit settings now optionally take percentage
1605 specifications. The percentage is taken relative to the amount of
1606 physical memory in the system (or in case of containers, the assigned
1607 amount of memory). This allows scaling service resources neatly with
1608 the amount of RAM available on the system. Similarly, systemd-logind's
1609 RuntimeDirectorySize= option now also optionally takes percentage
1612 * In similar fashion TasksMax= takes percentage values now, too. The
1613 value is taken relative to the configured maximum number of processes
1614 on the system. The per-service task maximum has been changed to 15%
1615 using this functionality. (Effectively this is an increase of 512 →
1616 4915 for service units, given the kernel's default pid_max setting.)
1618 * Calendar time specifications in .timer units now understand a ".."
1619 syntax for time ranges. Example: "4..7:10" may now be used for
1620 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1623 * The InaccessableDirectories=, ReadOnlyDirectories= and
1624 ReadWriteDirectories= unit file settings have been renamed to
1625 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1626 applied to all kinds of file nodes, and not just directories, with
1627 the exception of symlinks. Specifically these settings may now be
1628 used on block and character device nodes, UNIX sockets and FIFOS as
1629 well as regular files. The old names of these settings remain
1630 available for compatibility.
1632 * systemd will now log about all service processes it kills forcibly
1633 (using SIGKILL) because they remained after the clean shutdown phase
1634 of the service completed. This should help identifying services that
1635 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1636 systemd-logind's configuration a similar log message is generated for
1637 processes killed at the end of each session due to this setting.
1639 * systemd will now set the $JOURNAL_STREAM environment variable for all
1640 services whose stdout/stderr are connected to the Journal (which
1641 effectively means by default: all services). The variable contains
1642 the device and inode number of the file descriptor used for
1643 stdout/stderr. This may be used by invoked programs to detect whether
1644 their stdout/stderr is connected to the Journal, in which case they
1645 can switch over to direct Journal communication, thus being able to
1646 pass extended, structured metadata along with their log messages. As
1647 one example, this is now used by glib's logging primitives.
1649 * When using systemd's default tmp.mount unit for /tmp, the mount point
1650 will now be established with the "nosuid" and "nodev" options. This
1651 avoids privilege escalation attacks that put traps and exploits into
1652 /tmp. However, this might cause problems if you e. g. put container
1653 images or overlays into /tmp; if you need this, override tmp.mount's
1654 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1657 * systemd now supports the "memory" cgroup controller also on
1660 * The systemd-cgtop tool now optionally takes a control group path as
1661 command line argument. If specified, the control group list shown is
1662 limited to subgroups of that group.
1664 * The SystemCallFilter= unit file setting gained support for
1665 pre-defined, named system call filter sets. For example
1666 SystemCallFilter=@clock is now an effective way to make all clock
1667 changing-related system calls unavailable to a service. A number of
1668 similar pre-defined groups are defined. Writing system call filters
1669 for system services is simplified substantially with this new
1670 concept. Accordingly, all of systemd's own, long-running services now
1671 enable system call filtering based on this, by default.
1673 * A new service setting MemoryDenyWriteExecute= has been added, taking
1674 a boolean value. If turned on, a service may no longer create memory
1675 mappings that are writable and executable at the same time. This
1676 enhances security for services where this is enabled as it becomes
1677 harder to dynamically write and then execute memory in exploited
1678 service processes. This option has been enabled for all of systemd's
1679 own long-running services.
1681 * A new RestrictRealtime= service setting has been added, taking a
1682 boolean argument. If set the service's processes may no longer
1683 acquire realtime scheduling. This improves security as realtime
1684 scheduling may otherwise be used to easily freeze the system.
1686 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1687 value. This may be used for requesting that the system manager inside
1688 of the container reports start-up completion to nspawn which then
1689 propagates this notification further to the service manager
1690 supervising nspawn itself. A related option NotifyReady= in .nspawn
1691 files has been added too. This functionality allows ordering of the
1692 start-up of multiple containers using the usual systemd ordering
1695 * machinectl gained a new command "stop" that is an alias for
1698 * systemd-resolved gained support for contacting DNS servers on
1699 link-local IPv6 addresses.
1701 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1702 its caches. A method call for requesting the same operation has been
1703 added to the bus API too, and is made available via "systemd-resolve
1706 * systemd-resolve gained a new --status switch. If passed a brief
1707 summary of the used DNS configuration with per-interface information
1710 * resolved.conf gained a new Cache= boolean option, defaulting to
1711 on. If turned off local DNS caching is disabled. This comes with a
1712 performance penalty in particular when DNSSEC is enabled. Note that
1713 resolved disables its internal caching implicitly anyway, when the
1714 configured DNS server is on a host-local IP address such as ::1 or
1715 127.0.0.1, thus automatically avoiding double local caching.
1717 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1718 for DNS requests. This improves compatibility with local programs
1719 that do not use the libc NSS or systemd-resolved's bus APIs for name
1720 resolution. This minimal DNS service is only available to local
1721 programs and does not implement the full DNS protocol, but enough to
1722 cover local DNS clients. A new, static resolv.conf file, listing just
1723 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1724 now recommended to make /etc/resolv.conf a symlink to this file in
1725 order to route all DNS lookups to systemd-resolved, regardless if
1726 done via NSS, the bus API or raw DNS packets. Note that this local
1727 DNS service is not as fully featured as the libc NSS or
1728 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1729 used to deliver link-local address information (as this implies
1730 sending a local interface index along), LLMNR/mDNS support via this
1731 interface is severely restricted. It is thus strongly recommended for
1732 all applications to use the libc NSS API or native systemd-resolved
1735 * systemd-networkd's bridge support learned a new setting
1736 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1737 in .network files has been added for configuring VLAN bridging in
1738 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1740 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1741 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1742 now be acquired without relying on DHCPv6. Two new options
1743 UseDomains= and UseDNS= have been added to configure this behaviour.
1745 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1746 renamed IPv6AcceptRA=, without altering its behaviour. The old
1747 setting name remains available for compatibility reasons.
1749 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1750 Key=, InputKey= and OutputKey=.
1752 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1753 interface configuration.
1755 * "systemctl edit" may now be used to create new unit files by
1756 specifying the --force switch.
1758 * sd-event gained a new function sd_event_get_iteration() for
1759 requesting the current iteration counter of the event loop. It starts
1760 at zero and is increased by one with each event loop iteration.
1762 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1763 file. It can be used in lieu of %systemd_requires in packages which
1764 don't use any systemd functionality and are intended to be installed
1765 in minimal containers without systemd present. This macro provides
1766 ordering dependencies to ensure that if the package is installed in
1767 the same rpm transaction as systemd, systemd will be installed before
1768 the scriptlets for the package are executed, allowing unit presets
1771 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1772 been added to simplify packaging of generators.
1774 * The os-release file gained VERSION_CODENAME field for the
1775 distribution nickname (e.g. VERSION_CODENAME=woody).
1777 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1778 can be set to disable parsing of metadata and the creation
1779 of persistent symlinks for that device.
1781 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1782 to make them available to logged-in users has been reverted.
1784 * Much of the common code of the various systemd components is now
1785 built into an internal shared library libsystemd-shared-231.so
1786 (incorporating the systemd version number in the name, to be updated
1787 with future releases) that the components link to. This should
1788 decrease systemd footprint both in memory during runtime and on
1789 disk. Note that the shared library is not for public use, and is
1790 neither API not ABI stable, but is likely to change with every new
1791 released update. Packagers need to make sure that binaries
1792 linking to libsystemd-shared.so are updated in step with the
1795 * Configuration for "mkosi" is now part of the systemd
1796 repository. mkosi is a tool to easily build legacy-free OS images,
1797 and is available on github: https://github.com/systemd/mkosi. If
1798 "mkosi" is invoked in the build tree a new raw OS image is generated
1799 incorporating the systemd sources currently being worked on and a
1800 clean, fresh distribution installation. The generated OS image may be
1801 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1802 UEFI PC. This functionality is particularly useful to easily test
1803 local changes made to systemd in a pristine, defined environment. See
1804 HACKING for details.
1806 * configure learned the --with-support-url= option to specify the
1807 distribution's bugtracker.
1809 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1810 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1811 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1812 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1813 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1814 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1815 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1816 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1817 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1818 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1819 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1820 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1821 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1822 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1823 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1824 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1825 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1826 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1827 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1829 — Berlin, 2016-07-25
1833 * DNSSEC is now turned on by default in systemd-resolved (in
1834 "allow-downgrade" mode), but may be turned off during compile time by
1835 passing "--with-default-dnssec=no" to "configure" (and of course,
1836 during runtime with DNSSEC= in resolved.conf). We recommend
1837 downstreams to leave this on at least during development cycles and
1838 report any issues with the DNSSEC logic upstream. We are very
1839 interested in collecting feedback about the DNSSEC validator and its
1840 limitations in the wild. Note however, that DNSSEC support is
1841 probably nothing downstreams should turn on in stable distros just
1842 yet, as it might create incompatibilities with a few DNS servers and
1843 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1844 automatically whenever we detect such incompatible setups, but there
1845 might be systems we do not cover yet. Hence: please help us testing
1846 the DNSSEC code, leave this on where you can, report back, but then
1847 again don't consider turning this on in your stable, LTS or
1848 production release just yet. (Note that you have to enable
1849 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1850 and its DNSSEC mode for host name resolution from local
1853 * systemd-resolve conveniently resolves DANE records with the --tlsa
1854 option and OPENPGPKEY records with the --openpgp option. It also
1855 supports dumping raw DNS record data via the new --raw= switch.
1857 * systemd-logind will now by default terminate user processes that are
1858 part of the user session scope unit (session-XX.scope) when the user
1859 logs out. This behavior is controlled by the KillUserProcesses=
1860 setting in logind.conf, and the previous default of "no" is now
1861 changed to "yes". This means that user sessions will be properly
1862 cleaned up after, but additional steps are necessary to allow
1863 intentionally long-running processes to survive logout.
1865 While the user is logged in at least once, user@.service is running,
1866 and any service that should survive the end of any individual login
1867 session can be started at a user service or scope using systemd-run.
1868 systemd-run(1) man page has been extended with an example which shows
1869 how to run screen in a scope unit underneath user@.service. The same
1870 command works for tmux.
1872 After the user logs out of all sessions, user@.service will be
1873 terminated too, by default, unless the user has "lingering" enabled.
1874 To effectively allow users to run long-term tasks even if they are
1875 logged out, lingering must be enabled for them. See loginctl(1) for
1876 details. The default polkit policy was modified to allow users to
1877 set lingering for themselves without authentication.
1879 Previous defaults can be restored at compile time by the
1880 --without-kill-user-processes option to "configure".
1882 * systemd-logind gained new configuration settings SessionsMax= and
1883 InhibitorsMax=, both with a default of 8192. It will not register new
1884 user sessions or inhibitors above this limit.
1886 * systemd-logind will now reload configuration on SIGHUP.
1888 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1889 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1890 enable. Also, support for the "io" cgroup controller in the unified
1891 hierarchy has been added, so that the "memory", "pids" and "io" are
1892 now the controllers that are supported on the unified hierarchy.
1894 WARNING: it is not possible to use previous systemd versions with
1895 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1896 is necessary to also update systemd in the initramfs if using the
1897 unified hierarchy. An updated SELinux policy is also required.
1899 * LLDP support has been extended, and both passive (receive-only) and
1900 active (sender) modes are supported. Passive mode ("routers-only") is
1901 enabled by default in systemd-networkd. Active LLDP mode is enabled
1902 by default for containers on the internal network. The "networkctl
1903 lldp" command may be used to list information gathered. "networkctl
1904 status" will also show basic LLDP information on connected peers now.
1906 * The IAID and DUID unique identifier sent in DHCP requests may now be
1907 configured for the system and each .network file managed by
1908 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1910 * systemd-networkd gained support for configuring proxy ARP support for
1911 each interface, via the ProxyArp= setting in .network files. It also
1912 gained support for configuring the multicast querier feature of
1913 bridge devices, via the new MulticastQuerier= setting in .netdev
1914 files. Similarly, snooping on the IGMP traffic can be controlled
1915 via the new setting MulticastSnooping=.
1917 A new setting PreferredLifetime= has been added for addresses
1918 configured in .network file to configure the lifetime intended for an
1921 The systemd-networkd DHCP server gained the option EmitRouter=, which
1922 defaults to yes, to configure whether the DHCP Option 3 (Router)
1925 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1926 systemd-socket-activate and installed into /usr/bin. It is now fully
1929 * systemd-journald now uses separate threads to flush changes to disk
1930 when closing journal files, thus reducing impact of slow disk I/O on
1931 logging performance.
1933 * The sd-journal API gained two new calls
1934 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1935 can be used to open journal files using file descriptors instead of
1936 file or directory paths. sd_journal_open_container() has been
1937 deprecated, sd_journal_open_directory_fd() should be used instead
1938 with the flag SD_JOURNAL_OS_ROOT.
1940 * journalctl learned a new output mode "-o short-unix" that outputs log
1941 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1942 UTC). It also gained support for a new --no-hostname setting to
1943 suppress the hostname column in the family of "short" output modes.
1945 * systemd-ask-password now optionally skips printing of the password to
1946 stdout with --no-output which can be useful in scripts.
1948 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1949 (devices tagged with ID_MAKER_TOOL) are now tagged with
1950 "uaccess" and are available to logged in users.
1952 * The DeviceAllow= unit setting now supports specifiers (with "%").
1954 * "systemctl show" gained a new --value switch, which allows print a
1955 only the contents of a specific unit property, without also printing
1956 the property's name. Similar support was added to "show*" verbs
1957 of loginctl and machinectl that output "key=value" lists.
1959 * A new unit type "generated" was added for files dynamically generated
1960 by generator tools. Similarly, a new unit type "transient" is used
1961 for unit files created using the runtime API. "systemctl enable" will
1962 refuse to operate on such files.
1964 * A new command "systemctl revert" has been added that may be used to
1965 revert to the vendor version of a unit file, in case local changes
1966 have been made by adding drop-ins or overriding the unit file.
1968 * "machinectl clean" gained a new verb to automatically remove all or
1969 just hidden container images.
1971 * systemd-tmpfiles gained support for a new line type "e" for emptying
1972 directories, if they exist, without creating them if they don't.
1974 * systemd-nspawn gained support for automatically patching the UID/GIDs
1975 of the owners and the ACLs of all files and directories in a
1976 container tree to match the UID/GID user namespacing range selected
1977 for the container invocation. This mode is enabled via the new
1978 --private-users-chown switch. It also gained support for
1979 automatically choosing a free, previously unused UID/GID range when
1980 starting a container, via the new --private-users=pick setting (which
1981 implies --private-users-chown). Together, these options for the first
1982 time make user namespacing for nspawn containers fully automatic and
1983 thus deployable. The systemd-nspawn@.service template unit file has
1984 been changed to use this functionality by default.
1986 * systemd-nspawn gained a new --network-zone= switch, that allows
1987 creating ad-hoc virtual Ethernet links between multiple containers,
1988 that only exist as long as at least one container referencing them is
1989 running. This allows easy connecting of multiple containers with a
1990 common link that implements an Ethernet broadcast domain. Each of
1991 these network "zones" may be named relatively freely by the user, and
1992 may be referenced by any number of containers, but each container may
1993 only reference one of these "zones". On the lower level, this is
1994 implemented by an automatically managed bridge network interface for
1995 each zone, that is created when the first container referencing its
1996 zone is created and removed when the last one referencing its zone
1999 * The default start timeout may now be configured on the kernel command
2000 line via systemd.default_timeout_start_sec=. It was already
2001 configurable via the DefaultTimeoutStartSec= option in
2002 /etc/systemd/system.conf.
2004 * Socket units gained a new TriggerLimitIntervalSec= and
2005 TriggerLimitBurst= setting to configure a limit on the activation
2006 rate of the socket unit.
2008 * The LimitNICE= setting now optionally takes normal UNIX nice values
2009 in addition to the raw integer limit value. If the specified
2010 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2011 value is understood as UNIX nice value. If not prefixed like this it
2012 is understood as raw RLIMIT_NICE limit.
2014 * Note that the effect of the PrivateDevices= unit file setting changed
2015 slightly with this release: the per-device /dev file system will be
2016 mounted read-only from this version on, and will have "noexec"
2017 set. This (minor) change of behavior might cause some (exceptional)
2018 legacy software to break, when PrivateDevices=yes is set for its
2019 service. Please leave PrivateDevices= off if you run into problems
2022 * systemd-bootchart has been split out to a separate repository:
2023 https://github.com/systemd/systemd-bootchart
2025 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2026 merged into the kernel in its current form.
2028 * The compatibility libraries libsystemd-daemon.so,
2029 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2030 which have been deprecated since systemd-209 have been removed along
2031 with the corresponding pkg-config files. All symbols provided by
2032 those libraries are provided by libsystemd.so.
2034 * The Capabilities= unit file setting has been removed (it is ignored
2035 for backwards compatibility). AmbientCapabilities= and
2036 CapabilityBoundingSet= should be used instead.
2038 * A new special target has been added, initrd-root-device.target,
2039 which creates a synchronization point for dependencies of the root
2040 device in early userspace. Initramfs builders must ensure that this
2041 target is now included in early userspace.
2043 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2044 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2045 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2046 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2047 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2048 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2049 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2050 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2051 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2052 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2053 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2054 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2055 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2056 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2057 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2058 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2059 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2060 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2061 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2062 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2063 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2064 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2065 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2066 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2067 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2070 — Fairfax, 2016-05-21
2074 * The systemd-resolved DNS resolver service has gained a substantial
2075 set of new features, most prominently it may now act as a DNSSEC
2076 validating stub resolver. DNSSEC mode is currently turned off by
2077 default, but is expected to be turned on by default in one of the
2078 next releases. For now, we invite everybody to test the DNSSEC logic
2079 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2080 service also gained a full set of D-Bus interfaces, including calls
2081 to configure DNS and DNSSEC settings per link (for use by external
2082 network management software). systemd-resolved and systemd-networkd
2083 now distinguish between "search" and "routing" domains. The former
2084 are used to qualify single-label names, the latter are used purely
2085 for routing lookups within certain domains to specific links.
2086 resolved now also synthesizes RRs for all entries from /etc/hosts.
2088 * The systemd-resolve tool (which is a client utility for
2089 systemd-resolved) has been improved considerably and is now fully
2090 supported and documented. Hence it has moved from /usr/lib/systemd to
2093 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2096 * The coredump collection logic has been reworked: when a coredump is
2097 collected it is now written to disk, compressed and processed
2098 (including stacktrace extraction) from a new instantiated service
2099 systemd-coredump@.service, instead of directly from the
2100 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2101 processing large coredumps can take up a substantial amount of
2102 resources and time, and this previously happened entirely outside of
2103 systemd's service supervision. With the new logic the core_pattern
2104 hook only does minimal metadata collection before passing off control
2105 to the new instantiated service, which is configured with a time
2106 limit, a nice level and other settings to minimize negative impact on
2107 the rest of the system. Also note that the new logic will honour the
2108 RLIMIT_CORE setting of the crashed process, which now allows users
2109 and processes to turn off coredumping for their processes by setting
2112 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2113 and all forked processes by default. Previously, PID 1 would leave
2114 the setting at "0" for all processes, as set by the kernel. Note that
2115 the resource limit traditionally has no effect on the generated
2116 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2117 logic is used. Since the limit is now honoured (see above) its
2118 default has been changed so that the coredumping logic is enabled by
2119 default for all processes, while allowing specific opt-out.
2121 * When the stacktrace is extracted from processes of system users, this
2122 is now done as "systemd-coredump" user, in order to sandbox this
2123 potentially security sensitive parsing operation. (Note that when
2124 processing coredumps of normal users this is done under the user ID
2125 of process that crashed, as before.) Packagers should take notice
2126 that it is now necessary to create the "systemd-coredump" system user
2127 and group at package installation time.
2129 * The systemd-activate socket activation testing tool gained support
2130 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2131 and --seqpacket switches. It also has been extended to support both
2132 new-style and inetd-style file descriptor passing. Use the new
2133 --inetd switch to request inetd-style file descriptor passing.
2135 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2136 variable, which takes a boolean value. If set to false, ANSI color
2137 output is disabled in the tools even when run on a terminal that
2140 * The VXLAN support in networkd now supports two new settings
2141 DestinationPort= and PortRange=.
2143 * A new systemd.machine_id= kernel command line switch has been added,
2144 that may be used to set the machine ID in /etc/machine-id if it is
2145 not initialized yet. This command line option has no effect if the
2146 file is already initialized.
2148 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2149 specified command line as PID 2 rather than PID 1 in the
2150 container. In this mode PID 1 is a minimal stub init process that
2151 implements the special POSIX and Linux semantics of PID 1 regarding
2152 signal and child process management. Note that this stub init process
2153 is implemented in nspawn itself and requires no support from the
2154 container image. This new logic is useful to support running
2155 arbitrary commands in the container, as normal processes are
2156 generally not prepared to run as PID 1.
2158 * systemd-nspawn gained a new --chdir= switch for setting the current
2159 working directory for the process started in the container.
2161 * "journalctl /dev/sda" will now output all kernel log messages for
2162 specified device from the current boot, in addition to all devices
2163 that are parents of it. This should make log output about devices
2164 pretty useful, as long as kernel drivers attach enough metadata to
2165 the log messages. (The usual SATA drivers do.)
2167 * The sd-journal API gained two new calls
2168 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2169 that report whether log data from /run or /var has been found.
2171 * journalctl gained a new switch "--fields" that prints all journal
2172 record field names currently in use in the journal. This is backed
2173 by two new sd-journal API calls sd_journal_enumerate_fields() and
2174 sd_journal_restart_fields().
2176 * Most configurable timeouts in systemd now expect an argument of
2177 "infinity" to turn them off, instead of "0" as before. The semantics
2178 from now on is that a timeout of "0" means "now", and "infinity"
2179 means "never". To maintain backwards compatibility, "0" continues to
2180 turn off previously existing timeout settings.
2182 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2183 try-reload-or-restart" to clarify what it actually does: the "try"
2184 logic applies to both reloading and restarting, not just restarting.
2185 The old name continues to be accepted for compatibility.
2187 * On boot-up, when PID 1 detects that the system clock is behind the
2188 release date of the systemd version in use, the clock is now set
2189 to the latter. Previously, this was already done in timesyncd, in order
2190 to avoid running with clocks set to the various clock epochs such as
2191 1902, 1938 or 1970. With this change the logic is now done in PID 1
2192 in addition to timesyncd during early boot-up, so that it is enforced
2193 before the first process is spawned by systemd. Note that the logic
2194 in timesyncd remains, as it is more comprehensive and ensures
2195 clock monotonicity by maintaining a persistent timestamp file in
2196 /var. Since /var is generally not available in earliest boot or the
2197 initrd, this part of the logic remains in timesyncd, and is not done
2200 * Support for tweaking details in net_cls.class_id through the
2201 NetClass= configuration directive has been removed, as the kernel
2202 people have decided to deprecate that controller in cgroup v2.
2203 Userspace tools such as nftables are moving over to setting rules
2204 that are specific to the full cgroup path of a task, which obsoletes
2205 these controllers anyway. The NetClass= directive is kept around for
2206 legacy compatibility reasons. For a more in-depth description of the
2207 kernel change, please refer to the respective upstream commit:
2209 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2211 * A new service setting RuntimeMaxSec= has been added that may be used
2212 to specify a maximum runtime for a service. If the timeout is hit, the
2213 service is terminated and put into a failure state.
2215 * A new service setting AmbientCapabilities= has been added. It allows
2216 configuration of additional Linux process capabilities that are
2217 passed to the activated processes. This is only available on very
2220 * The process resource limit settings in service units may now be used
2221 to configure hard and soft limits individually.
2223 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2224 expose support for gcc's __attribute__((cleanup())) C extension.
2225 Specifically, for many object destructor functions alternative
2226 versions have been added that have names suffixed with "p" and take a
2227 pointer to a pointer to the object to destroy, instead of just a
2228 pointer to the object itself. This is useful because these destructor
2229 functions may be used directly as parameters to the cleanup
2230 construct. Internally, systemd has been a heavy user of this GCC
2231 extension for a long time, and with this change similar support is
2232 now available to consumers of the library outside of systemd. Note
2233 that by using this extension in your sources compatibility with old
2234 and strictly ANSI compatible C compilers is lost. However, all gcc or
2235 LLVM versions of recent years support this extension.
2237 * Timer units gained support for a new setting RandomizedDelaySec= that
2238 allows configuring some additional randomized delay to the configured
2239 time. This is useful to spread out timer events to avoid load peaks in
2240 clusters or larger setups.
2242 * Calendar time specifications now support sub-second accuracy.
2244 * Socket units now support listening on SCTP and UDP-lite protocol
2247 * The sd-event API now comes with a full set of man pages.
2249 * Older versions of systemd contained experimental support for
2250 compressing journal files and coredumps with the LZ4 compressor that
2251 was not compatible with the lz4 binary (due to API limitations of the
2252 lz4 library). This support has been removed; only support for files
2253 compatible with the lz4 binary remains. This LZ4 logic is now
2254 officially supported and no longer considered experimental.
2256 * The dkr image import logic has been removed again from importd. dkr's
2257 micro-services focus doesn't fit into the machine image focus of
2258 importd, and quickly got out of date with the upstream dkr API.
2260 * Creation of the /run/lock/lockdev/ directory was dropped from
2261 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2262 been available for many years. If you still need this, you need to
2263 create your own tmpfiles.d config file with:
2265 d /run/lock/lockdev 0775 root lock -
2267 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2268 and RebootArgument= have been moved from the [Service] section of
2269 unit files to [Unit], and they are now supported on all unit types,
2270 not just service units. Of course, elogind will continue to
2271 understand these settings also at the old location, in order to
2272 maintain compatibility.
2274 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2275 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2276 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2277 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2278 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2279 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2280 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2281 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2282 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2283 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2284 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2285 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2286 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2287 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2288 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2289 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2290 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2291 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2292 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2294 — Berlin, 2016-02-11
2298 * A number of properties previously only settable in unit
2299 files are now also available as properties to set when
2300 creating transient units programmatically via the bus, as it
2301 is exposed with systemd-run's --property=
2302 setting. Specifically, these are: SyslogIdentifier=,
2303 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2304 EnvironmentFile=, ReadWriteDirectories=,
2305 ReadOnlyDirectories=, InaccessibleDirectories=,
2306 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2308 * When creating transient services via the bus API it is now
2309 possible to pass in a set of file descriptors to use as
2310 STDIN/STDOUT/STDERR for the invoked process.
2312 * Slice units may now be created transiently via the bus APIs,
2313 similar to the way service and scope units may already be
2314 created transiently.
2316 * Wherever systemd expects a calendar timestamp specification
2317 (like in journalctl's --since= and --until= switches) UTC
2318 timestamps are now supported. Timestamps suffixed with "UTC"
2319 are now considered to be in Universal Time Coordinated
2320 instead of the local timezone. Also, timestamps may now
2321 optionally be specified with sub-second accuracy. Both of
2322 these additions also apply to recurring calendar event
2323 specification, such as OnCalendar= in timer units.
2325 * journalctl gained a new "--sync" switch that asks the
2326 journal daemon to write all so far unwritten log messages to
2327 disk and sync the files, before returning.
2329 * systemd-tmpfiles learned two new line types "q" and "Q" that
2330 operate like "v", but also set up a basic btrfs quota
2331 hierarchy when used on a btrfs file system with quota
2334 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2335 instead of a subvolume (even on a btrfs file system) if the
2336 root directory is a plain directory, and not a
2337 subvolume. This should simplify things with certain chroot()
2338 environments which are not aware of the concept of btrfs
2341 * systemd-detect-virt gained a new --chroot switch to detect
2342 whether execution takes place in a chroot() environment.
2344 * CPUAffinity= now takes CPU index ranges in addition to
2347 * The various memory-related resource limit settings (such as
2348 LimitAS=) now understand the usual K, M, G, ... suffixes to
2349 the base of 1024 (IEC). Similar, the time-related resource
2350 limit settings understand the usual min, h, day, ...
2353 * There's a new system.conf setting DefaultTasksMax= to
2354 control the default TasksMax= setting for services and
2355 scopes running on the system. (TasksMax= is the primary
2356 setting that exposes the "pids" cgroup controller on systemd
2357 and was introduced in the previous systemd release.) The
2358 setting now defaults to 512, which means services that are
2359 not explicitly configured otherwise will only be able to
2360 create 512 processes or threads at maximum, from this
2361 version on. Note that this means that thread- or
2362 process-heavy services might need to be reconfigured to set
2363 TasksMax= to a higher value. It is sufficient to set
2364 TasksMax= in these specific unit files to a higher value, or
2365 even "infinity". Similar, there's now a logind.conf setting
2366 UserTasksMax= that defaults to 4096 and limits the total
2367 number of processes or tasks each user may own
2368 concurrently. nspawn containers also have the TasksMax=
2369 value set by default now, to 8192. Note that all of this
2370 only has an effect if the "pids" cgroup controller is
2371 enabled in the kernel. The general benefit of these changes
2372 should be a more robust and safer system, that provides a
2373 certain amount of per-service fork() bomb protection.
2375 * systemd-nspawn gained the new --network-veth-extra= switch
2376 to define additional and arbitrarily-named virtual Ethernet
2377 links between the host and the container.
2379 * A new service execution setting PassEnvironment= has been
2380 added that allows importing select environment variables
2381 from PID1's environment block into the environment block of
2384 * Timer units gained support for a new RemainAfterElapse=
2385 setting which takes a boolean argument. It defaults to on,
2386 exposing behaviour unchanged to previous releases. If set to
2387 off, timer units are unloaded after they elapsed if they
2388 cannot elapse again. This is particularly useful for
2389 transient timer units, which shall not stay around longer
2390 than until they first elapse.
2392 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2393 default now (the kernel default is 16). This is beneficial
2394 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2395 allows substantially larger numbers of queued
2396 datagrams. This should increase the capability of systemd to
2397 parallelize boot-up, as logging and sd_notify() are unlikely
2398 to stall execution anymore. If you need to change the value
2399 from the new defaults, use the usual sysctl.d/ snippets.
2401 * The compression framing format used by the journal or
2402 coredump processing has changed to be in line with what the
2403 official LZ4 tools generate. LZ4 compression support in
2404 systemd was considered unsupported previously, as the format
2405 was not compatible with the normal tools. With this release
2406 this has changed now, and it is hence safe for downstream
2407 distributions to turn it on. While not compressing as well
2408 as the XZ, LZ4 is substantially faster, which makes
2409 it a good default choice for the compression logic in the
2410 journal and in coredump handling.
2412 * Any reference to /etc/mtab has been dropped from
2413 systemd. The file has been obsolete since a while, but
2414 systemd refused to work on systems where it was incorrectly
2415 set up (it should be a symlink or non-existent). Please make
2416 sure to update to util-linux 2.27.1 or newer in conjunction
2417 with this systemd release, which also drops any reference to
2418 /etc/mtab. If you maintain a distribution make sure that no
2419 software you package still references it, as this is a
2420 likely source of bugs. There's also a glibc bug pending,
2421 asking for removal of any reference to this obsolete file:
2423 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2425 Note that only util-linux versions built with
2426 --enable-libmount-force-mountinfo are supported.
2428 * Support for the ".snapshot" unit type has been removed. This
2429 feature turned out to be little useful and little used, and
2430 has now been removed from the core and from systemctl.
2432 * The dependency types RequiresOverridable= and
2433 RequisiteOverridable= have been removed from systemd. They
2434 have been used only very sparingly to our knowledge and
2435 other options that provide a similar effect (such as
2436 systemctl --mode=ignore-dependencies) are much more useful
2437 and commonly used. Moreover, they were only half-way
2438 implemented as the option to control behaviour regarding
2439 these dependencies was never added to systemctl. By removing
2440 these dependency types the execution engine becomes a bit
2441 simpler. Unit files that use these dependencies should be
2442 changed to use the non-Overridable dependency types
2443 instead. In fact, when parsing unit files with these
2444 options, that's what systemd will automatically convert them
2445 too, but it will also warn, asking users to fix the unit
2446 files accordingly. Removal of these dependency types should
2447 only affect a negligible number of unit files in the wild.
2449 * Behaviour of networkd's IPForward= option changed
2450 (again). It will no longer maintain a per-interface setting,
2451 but propagate one way from interfaces where this is enabled
2452 to the global kernel setting. The global setting will be
2453 enabled when requested by a network that is set up, but
2454 never be disabled again. This change was made to make sure
2455 IPv4 and IPv6 behaviour regarding packet forwarding is
2456 similar (as the Linux IPv6 stack does not support
2457 per-interface control of this setting) and to minimize
2460 * In unit files the behaviour of %u, %U, %h, %s has
2461 changed. These specifiers will now unconditionally resolve
2462 to the various user database fields of the user that the
2463 systemd instance is running as, instead of the user
2464 configured in the specific unit via User=. Note that this
2465 effectively doesn't change much, as resolving of these
2466 specifiers was already turned off in the --system instance
2467 of systemd, as we cannot do NSS lookups from PID 1. In the
2468 --user instance of systemd these specifiers where correctly
2469 resolved, but hardly made any sense, since the user instance
2470 lacks privileges to do user switches anyway, and User= is
2471 hence useless. Moreover, even in the --user instance of
2472 systemd behaviour was awkward as it would only take settings
2473 from User= assignment placed before the specifier into
2474 account. In order to unify and simplify the logic around
2475 this the specifiers will now always resolve to the
2476 credentials of the user invoking the manager (which in case
2477 of PID 1 is the root user).
2479 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2480 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2481 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2482 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2483 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2484 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2485 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2486 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2487 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2488 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2489 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2490 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2491 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2492 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2495 — Berlin, 2015-11-18
2499 * systemd now depends on util-linux v2.27. More specifically,
2500 the newly added mount monitor feature in libmount now
2501 replaces systemd's former own implementation.
2503 * libmount mandates /etc/mtab not to be regular file, and
2504 systemd now enforces this condition at early boot.
2505 /etc/mtab has been deprecated and warned about for a very
2506 long time, so systems running systemd should already have
2507 stopped having this file around as anything else than a
2508 symlink to /proc/self/mounts.
2510 * Support for the "pids" cgroup controller has been added. It
2511 allows accounting the number of tasks in a cgroup and
2512 enforcing limits on it. This adds two new setting
2513 TasksAccounting= and TasksMax= to each unit, as well as a
2514 global option DefaultTasksAccounting=.
2516 * Support for the "net_cls" cgroup controller has been added.
2517 It allows assigning a net class ID to each task in the
2518 cgroup, which can then be used in firewall rules and traffic
2519 shaping configurations. Note that the kernel netfilter net
2520 class code does not currently work reliably for ingress
2521 packets on unestablished sockets.
2523 This adds a new config directive called NetClass= to CGroup
2524 enabled units. Allowed values are positive numbers for fixed
2525 assignments and "auto" for picking a free value
2528 * 'systemctl is-system-running' now returns 'offline' if the
2529 system is not booted with systemd. This command can now be
2530 used as a substitute for 'systemd-notify --booted'.
2532 * Watchdog timeouts have been increased to 3 minutes for all
2533 in-tree service files. Apparently, disk IO issues are more
2534 frequent than we hoped, and user reported >1 minute waiting
2537 * 'machine-id-commit' functionality has been merged into
2538 'machine-id-setup --commit'. The separate binary has been
2541 * The WorkingDirectory= directive in unit files may now be set
2542 to the special value '~'. In this case, the working
2543 directory is set to the home directory of the user
2544 configured in User=.
2546 * "machinectl shell" will now open the shell in the home
2547 directory of the selected user by default.
2549 * The CrashChVT= configuration file setting is renamed to
2550 CrashChangeVT=, following our usual logic of not
2551 abbreviating unnecessarily. The old directive is still
2552 supported for compat reasons. Also, this directive now takes
2553 an integer value between 1 and 63, or a boolean value. The
2554 formerly supported '-1' value for disabling stays around for
2557 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2558 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2559 RootDirectory= properties can now be set for transient
2562 * The systemd-analyze tool gained a new "set-log-target" verb
2563 to change the logging target the system manager logs to
2564 dynamically during runtime. This is similar to how
2565 "systemd-analyze set-log-level" already changes the log
2568 * In nspawn /sys is now mounted as tmpfs, with only a selected
2569 set of subdirectories mounted in from the real sysfs. This
2570 enhances security slightly, and is useful for ensuring user
2571 namespaces work correctly.
2573 * Support for USB FunctionFS activation has been added. This
2574 allows implementation of USB gadget services that are
2575 activated as soon as they are requested, so that they don't
2576 have to run continuously, similar to classic socket
2579 * The "systemctl exit" command now optionally takes an
2580 additional parameter that sets the exit code to return from
2581 the systemd manager when exiting. This is only relevant when
2582 running the systemd user instance, or when running the
2583 system instance in a container.
2585 * sd-bus gained the new API calls sd_bus_path_encode_many()
2586 and sd_bus_path_decode_many() that allow easy encoding and
2587 decoding of multiple identifier strings inside a D-Bus
2588 object path. Another new call sd_bus_default_flush_close()
2589 has been added to flush and close per-thread default
2592 * systemd-cgtop gained support for a -M/--machine= switch to
2593 show the control groups within a certain container only.
2595 * "systemctl kill" gained support for an optional --fail
2596 switch. If specified the requested operation will fail of no
2597 processes have been killed, because the unit had no
2598 processes attached, or similar.
2600 * A new systemd.crash_reboot=1 kernel command line option has
2601 been added that triggers a reboot after crashing. This can
2602 also be set through CrashReboot= in systemd.conf.
2604 * The RuntimeDirectory= setting now understands unit
2605 specifiers like %i or %f.
2607 * A new (still internal) library API sd-ipv4acd has been added,
2608 that implements address conflict detection for IPv4. It's
2609 based on code from sd-ipv4ll, and will be useful for
2610 detecting DHCP address conflicts.
2612 * File descriptors passed during socket activation may now be
2613 named. A new API sd_listen_fds_with_names() is added to
2614 access the names. The default names may be overridden,
2615 either in the .socket file using the FileDescriptorName=
2616 parameter, or by passing FDNAME= when storing the file
2617 descriptors using sd_notify().
2619 * systemd-networkd gained support for:
2621 - Setting the IPv6 Router Advertisement settings via
2622 IPv6AcceptRouterAdvertisements= in .network files.
2624 - Configuring the HelloTimeSec=, MaxAgeSec= and
2625 ForwardDelaySec= bridge parameters in .netdev files.
2627 - Configuring PreferredSource= for static routes in
2630 * The "ask-password" framework used to query for LUKS harddisk
2631 passwords or SSL passwords during boot gained support for
2632 caching passwords in the kernel keyring, if it is
2633 available. This makes sure that the user only has to type in
2634 a passphrase once if there are multiple objects to unlock
2635 with the same one. Previously, such password caching was
2636 available only when Plymouth was used; this moves the
2637 caching logic into the systemd codebase itself. The
2638 "systemd-ask-password" utility gained a new --keyname=
2639 switch to control which kernel keyring key to use for
2640 caching a password in. This functionality is also useful for
2641 enabling display managers such as gdm to automatically
2642 unlock the user's GNOME keyring if its passphrase, the
2643 user's password and the harddisk password are the same, if
2644 gdm-autologin is used.
2646 * When downloading tar or raw images using "machinectl
2647 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2648 file is now also downloaded, if it is available and stored
2649 next to the image file.
2651 * Units of type ".socket" gained a new boolean setting
2652 Writable= which is only useful in conjunction with
2653 ListenSpecial=. If true, enables opening the specified
2654 special file in O_RDWR mode rather than O_RDONLY mode.
2656 * systemd-rfkill has been reworked to become a singleton
2657 service that is activated through /dev/rfkill on each rfkill
2658 state change and saves the settings to disk. This way,
2659 systemd-rfkill is now compatible with devices that exist
2660 only intermittendly, and even restores state if the previous
2661 system shutdown was abrupt rather than clean.
2663 * The journal daemon gained support for vacuuming old journal
2664 files controlled by the number of files that shall remain,
2665 in addition to the already existing control by size and by
2666 date. This is useful as journal interleaving performance
2667 degrades with too many separate journal files, and allows
2668 putting an effective limit on them. The new setting defaults
2669 to 100, but this may be changed by setting SystemMaxFiles=
2670 and RuntimeMaxFiles= in journald.conf. Also, the
2671 "journalctl" tool gained the new --vacuum-files= switch to
2672 manually vacuum journal files to leave only the specified
2673 number of files in place.
2675 * udev will now create /dev/disk/by-path links for ATA devices
2676 on kernels where that is supported.
2678 * Galician, Serbian, Turkish and Korean translations were added.
2680 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2681 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2682 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2683 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2684 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2685 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2686 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2687 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2688 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2689 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2690 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2691 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2692 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2693 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2694 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2695 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2696 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2697 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2699 — Berlin, 2015-10-07
2703 * The DHCP implementation of systemd-networkd gained a set of
2706 - The DHCP server now supports emitting DNS and NTP
2707 information. It may be enabled and configured via
2708 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2709 and NTP information is enabled, but no servers are
2710 configured, the corresponding uplink information (if there
2711 is any) is propagated.
2713 - Server and client now support transmission and reception
2714 of timezone information. It can be configured via the
2715 newly introduced network options UseTimezone=,
2716 EmitTimezone=, and Timezone=. Transmission of timezone
2717 information is enabled between host and containers by
2718 default now: the container will change its local timezone
2719 to what the host has set.
2721 - Lease timeouts can now be configured via
2722 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2724 - The DHCP server improved on the stability of
2725 leases. Clients are more likely to get the same lease
2726 information back, even if the server loses state.
2728 - The DHCP server supports two new configuration options to
2729 control the lease address pool metrics, PoolOffset= and
2732 * The encapsulation limit of tunnels in systemd-networkd may
2733 now be configured via 'EncapsulationLimit='. It allows
2734 modifying the maximum additional levels of encapsulation
2735 that are permitted to be prepended to a packet.
2737 * systemd now supports the concept of user buses replacing
2738 session buses, if used with dbus-1.10 (and enabled via dbus
2739 --enable-user-session). It previously only supported this on
2740 kdbus-enabled systems, and this release expands this to
2741 'dbus-daemon' systems.
2743 * systemd-networkd now supports predictable interface names
2746 * systemd now optionally supports the new Linux kernel
2747 "unified" control group hierarchy. If enabled via the kernel
2748 command-line option 'systemd.unified_cgroup_hierarchy=1',
2749 systemd will try to mount the unified cgroup hierarchy
2750 directly on /sys/fs/cgroup. If not enabled, or not
2751 available, systemd will fall back to the legacy cgroup
2752 hierarchy setup, as before. Host system and containers can
2753 mix and match legacy and unified hierarchies as they
2754 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2755 environment variable to individually select the hierarchy to
2756 use for executed containers. By default, nspawn will use the
2757 unified hierarchy for the containers if the host uses the
2758 unified hierarchy, and the legacy hierarchy otherwise.
2759 Please note that at this point the unified hierarchy is an
2760 experimental kernel feature and is likely to change in one
2761 of the next kernel releases. Therefore, it should not be
2762 enabled by default in downstream distributions yet. The
2763 minimum required kernel version for the unified hierarchy to
2764 work is 4.2. Note that when the unified hierarchy is used
2765 for the first time delegated access to controllers is
2766 safe. Because of this systemd-nspawn containers will get
2767 access to controllers now, as will systemd user
2768 sessions. This means containers and user sessions may now
2769 manage their own resources, partitioning up what the system
2772 * A new special scope unit "init.scope" has been introduced
2773 that encapsulates PID 1 of the system. It may be used to
2774 determine resource usage and enforce resource limits on PID
2775 1 itself. PID 1 hence moved out of the root of the control
2778 * The cgtop tool gained support for filtering out kernel
2779 threads when counting tasks in a control group. Also, the
2780 count of processes is now recursively summed up by
2781 default. Two options -k and --recursive= have been added to
2782 revert to old behaviour. The tool has also been updated to
2783 work correctly in containers now.
2785 * systemd-nspawn's --bind= and --bind-ro= options have been
2786 extended to allow creation of non-recursive bind mounts.
2788 * libsystemd gained two new calls sd_pid_get_cgroup() and
2789 sd_peer_get_cgroup() which return the control group path of
2790 a process or peer of a connected AF_UNIX socket. This
2791 function call is particularly useful when implementing
2792 delegated subtrees support in the control group hierarchy.
2794 * The "sd-event" event loop API of libsystemd now supports
2795 correct dequeuing of real-time signals, without losing
2798 * When systemd requests a PolicyKit decision when managing
2799 units it will now add additional fields to the request,
2800 including unit name and desired operation. This enables more
2801 powerful PolicyKit policies, that make decisions depending
2802 on these parameters.
2804 * nspawn learnt support for .nspawn settings files, that may
2805 accompany the image files or directories of containers, and
2806 may contain additional settings for the container. This is
2807 an alternative to configuring container parameters via the
2808 nspawn command line.
2810 Contributions from: Cristian Rodríguez, Daniel Mack, David
2811 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2812 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2813 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2814 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2815 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2816 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2817 Andersen, Tom Gundersen, Torstein Husebø
2819 — Berlin, 2015-09-08
2823 * machinectl gained a new verb 'shell' which opens a fresh
2824 shell on the target container or the host. It is similar to
2825 the existing 'login' command of machinectl, but spawns the
2826 shell directly without prompting for username or
2827 password. The pseudo machine '.host' now refers to the local
2828 host and is used by default. Hence, 'machinectl shell' can
2829 be used as replacement for 'su -' which spawns a session as
2830 a fresh systemd unit in a way that is fully isolated from
2831 the originating session.
2833 * systemd-networkd learned to cope with private-zone DHCP
2834 options and allows other programs to query the values.
2836 * SELinux access control when enabling/disabling units is no
2837 longer enforced with this release. The previous
2838 implementation was incorrect, and a new corrected
2839 implementation is not yet available. As unit file operations
2840 are still protected via PolicyKit and D-Bus policy this is
2841 not a security problem. Yet, distributions which care about
2842 optimal SELinux support should probably not stabilize on
2845 * sd-bus gained support for matches of type "arg0has=", that
2846 test for membership of strings in string arrays sent in bus
2849 * systemd-resolved now dumps the contents of its DNS and LLMNR
2850 caches to the logs on reception of the SIGUSR1 signal. This
2851 is useful to debug DNS behaviour.
2853 * The coredumpctl tool gained a new --directory= option to
2854 operate on journal files in a specific directory.
2856 * "systemctl reboot" and related commands gained a new
2857 "--message=" option which may be used to set a free-text
2858 wall message when shutting down or rebooting the
2859 system. This message is also logged, which is useful for
2860 figuring out the reason for a reboot or shutdown a
2863 * The "systemd-resolve-host" tool's -i switch now takes
2864 network interface numbers as alternative to interface names.
2866 * A new unit file setting for services has been introduced:
2867 UtmpMode= allows configuration of how precisely systemd
2868 handles utmp and wtmp entries for the service if this is
2869 enabled. This allows writing services that appear similar to
2870 user sessions in the output of the "w", "who", "last" and
2873 * systemd-resolved will now locally synthesize DNS resource
2874 records for the "localhost" and "gateway" domains as well as
2875 the local hostname. This should ensure that clients querying
2876 RRs via resolved will get similar results as those going via
2877 NSS, if nss-myhostname is enabled.
2879 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2880 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2881 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2882 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2883 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2884 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2885 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2886 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2887 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2888 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2889 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2890 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2892 — Berlin, 2015-08-27
2896 * The systemd-efi-boot-generator functionality was merged into
2897 systemd-gpt-auto-generator.
2899 * systemd-networkd now supports Group Policy for vxlan
2900 devices. It can be enabled via the new boolean configuration
2901 option called 'GroupPolicyExtension='.
2903 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2904 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2905 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2907 — Berlin, 2015-07-31
2911 * The python-systemd code has been removed from the systemd repository.
2912 A new repository has been created which accommodates the code from
2913 now on, and we kindly ask distributions to create a separate package
2914 for this: https://github.com/systemd/python-systemd
2916 * The systemd daemon will now reload its main configuration
2917 (/etc/systemd/system.conf) on daemon-reload.
2919 * sd-dhcp now exposes vendor specific extensions via
2920 sd_dhcp_lease_get_vendor_specific().
2922 * systemd-networkd gained a number of new configuration options.
2924 - A new boolean configuration option for TAP devices called
2925 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2926 device, thus allowing to send and receive GSO packets.
2928 - A new tunnel configuration option called 'CopyDSCP='.
2929 If enabled, the DSCP field of ip6 tunnels is copied into the
2930 decapsulated packet.
2932 - A set of boolean bridge configuration options were added.
2933 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2934 and 'UnicastFlood=' are now parsed by networkd and applied to the
2935 respective bridge link device via the respective IFLA_BRPORT_*
2938 - A new string configuration option to override the hostname sent
2939 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2940 is true, networkd will use the configured hostname instead of the
2941 system hostname when sending DHCP requests.
2943 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2944 networkd will configure the IPv6 flow-label of the tunnel device
2945 according to RFC2460.
2947 - The 'macvtap' virtual network devices are now supported, similar to
2948 the already supported 'macvlan' devices.
2950 * systemd-resolved now implements RFC5452 to improve resilience against
2951 cache poisoning. Additionally, source port randomization is enabled
2952 by default to further protect against DNS spoofing attacks.
2954 * nss-mymachines now supports translating UIDs and GIDs of running
2955 containers with user-namespaces enabled. If a container 'foo'
2956 translates a host uid 'UID' to the container uid 'TUID', then
2957 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2958 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2959 mapped as 'vg-foo-TGID'.
2961 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2962 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2963 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2964 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2965 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2966 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2967 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2968 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2969 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2970 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2972 — Berlin, 2015-07-29
2976 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2977 There are no known issues with current sysfs, and udev does not need
2978 or should be used to work around such bugs.
2980 * udev does no longer enable USB HID power management. Several reports
2981 indicate, that some devices cannot handle that setting.
2983 * The udev accelerometer helper was removed. The functionality
2984 is now fully included in iio-sensor-proxy. But this means,
2985 older iio-sensor-proxy versions will no longer provide
2986 accelerometer/orientation data with this systemd version.
2987 Please upgrade iio-sensor-proxy to version 1.0.
2989 * networkd gained a new configuration option IPv6PrivacyExtensions=
2990 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2991 for Stateless Address") on selected networks.
2993 * For the sake of fewer build-time dependencies and less code in the
2994 main repository, the python bindings are about to be removed in the
2995 next release. A new repository has been created which accommodates
2996 the code from now on, and we kindly ask distributions to create a
2997 separate package for this. The removal will take place in v223.
2999 https://github.com/systemd/python-systemd
3001 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3002 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3003 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3004 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3005 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3006 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3007 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3008 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3009 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3010 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3012 — Berlin, 2015-07-07
3016 * The sd-bus.h and sd-event.h APIs have now been declared
3017 stable and have been added to the official interface of
3018 libsystemd.so. sd-bus implements an alternative D-Bus client
3019 library, that is relatively easy to use, very efficient and
3020 supports both classic D-Bus as well as kdbus as transport
3021 backend. sd-event is a generic event loop abstraction that
3022 is built around Linux epoll, but adds features such as event
3023 prioritization or efficient timer handling. Both APIs are good
3024 choices for C programs looking for a bus and/or event loop
3025 implementation that is minimal and does not have to be
3026 portable to other kernels.
3028 * kdbus support is no longer compile-time optional. It is now
3029 always built-in. However, it can still be disabled at
3030 runtime using the kdbus=0 kernel command line setting, and
3031 that setting may be changed to default to off, by specifying
3032 --disable-kdbus at build-time. Note though that the kernel
3033 command line setting has no effect if the kdbus.ko kernel
3034 module is not installed, in which case kdbus is (obviously)
3035 also disabled. We encourage all downstream distributions to
3036 begin testing kdbus by adding it to the kernel images in the
3037 development distributions, and leaving kdbus support in
3040 * The minimal required util-linux version has been bumped to
3043 * Support for chkconfig (--enable-chkconfig) was removed in
3044 favor of calling an abstraction tool
3045 /lib/systemd/systemd-sysv-install. This needs to be
3046 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3047 in README for details.
3049 * If there's a systemd unit and a SysV init script for the
3050 same service name, and the user executes "systemctl enable"
3051 for it (or a related call), then this will now enable both
3052 (or execute the related operation on both), not just the
3055 * The libudev API documentation has been converted from gtkdoc
3058 * gudev has been removed from the systemd tree, it is now an
3061 * The systemd-cgtop tool learnt a new --raw switch to generate
3062 "raw" (machine parsable) output.
3064 * networkd's IPForwarding= .network file setting learnt the
3065 new setting "kernel", which ensures that networkd does not
3066 change the IP forwarding sysctl from the default kernel
3069 * The systemd-logind bus API now exposes a new boolean
3070 property "Docked" that reports whether logind considers the
3071 system "docked", i.e. connected to a docking station or not.
3073 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3074 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3075 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3076 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3077 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3078 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3079 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3080 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3081 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3082 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3083 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3084 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3085 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3086 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3087 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3088 Fink, Zbigniew Jędrzejewski-Szmek
3090 — Berlin, 2015-06-19
3094 * The gudev library has been extracted into a separate repository
3095 available at: https://git.gnome.org/browse/libgudev/
3096 It is now managed as part of the Gnome project. Distributions
3097 are recommended to pass --disable-gudev to systemd and use
3098 gudev from the Gnome project instead. gudev is still included
3099 in systemd, for now. It will be removed soon, though. Please
3100 also see the announcement-thread on systemd-devel:
3101 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3103 * systemd now exposes a CPUUsageNSec= property for each
3104 service unit on the bus, that contains the overall consumed
3105 CPU time of a service (the sum of what each process of the
3106 service consumed). This value is only available if
3107 CPUAccounting= is turned on for a service, and is then shown
3108 in the "systemctl status" output.
3110 * Support for configuring alternative mappings of the old SysV
3111 runlevels to systemd targets has been removed. They are now
3112 hardcoded in a way that runlevels 2, 3, 4 all map to
3113 multi-user.target and 5 to graphical.target (which
3114 previously was already the default behaviour).
3116 * The auto-mounter logic gained support for mount point
3117 expiry, using a new TimeoutIdleSec= setting in .automount
3118 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3120 * The EFI System Partition (ESP) as mounted to /boot by
3121 systemd-efi-boot-generator will now be unmounted
3122 automatically after 2 minutes of not being used. This should
3123 minimize the risk of ESP corruptions.
3125 * New /etc/fstab options x-systemd.requires= and
3126 x-systemd.requires-mounts-for= are now supported to express
3127 additional dependencies for mounts. This is useful for
3128 journalling file systems that support external journal
3129 devices or overlay file systems that require underlying file
3130 systems to be mounted.
3132 * systemd does not support direct live-upgrades (via systemctl
3133 daemon-reexec) from versions older than v44 anymore. As no
3134 distribution we are aware of shipped such old versions in a
3135 stable release this should not be problematic.
3137 * When systemd forks off a new per-connection service instance
3138 it will now set the $REMOTE_ADDR environment variable to the
3139 remote IP address, and $REMOTE_PORT environment variable to
3140 the remote IP port. This behaviour is similar to the
3141 corresponding environment variables defined by CGI.
3143 * systemd-networkd gained support for uplink failure
3144 detection. The BindCarrier= option allows binding interface
3145 configuration dynamically to the link sense of other
3146 interfaces. This is useful to achieve behaviour like in
3149 * systemd-networkd gained support for configuring the DHCP
3150 client identifier to use when requesting leases.
3152 * systemd-networkd now has a per-network UseNTP= option to
3153 configure whether NTP server information acquired via DHCP
3154 is passed on to services like systemd-timesyncd.
3156 * systemd-networkd gained support for vti6 tunnels.
3158 * Note that systemd-networkd manages the sysctl variable
3159 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3160 it is configured for since v219. The variable controls IP
3161 forwarding, and is a per-interface alternative to the global
3162 /proc/sys/net/ipv[46]/ip_forward. This setting is
3163 configurable in the IPForward= option, which defaults to
3164 "no". This means if networkd is used for an interface it is
3165 no longer sufficient to set the global sysctl option to turn
3166 on IP forwarding! Instead, the .network file option
3167 IPForward= needs to be turned on! Note that the
3168 implementation of this behaviour was broken in v219 and has
3171 * Many bonding and vxlan options are now configurable in
3174 * systemd-nspawn gained a new --property= setting to set unit
3175 properties for the container scope. This is useful for
3176 setting resource parameters (e.g. "CPUShares=500") on
3177 containers started from the command line.
3179 * systemd-nspawn gained a new --private-users= switch to make
3180 use of user namespacing available on recent Linux kernels.
3182 * systemd-nspawn may now be called as part of a shell pipeline
3183 in which case the pipes used for stdin and stdout are passed
3184 directly to the process invoked in the container, without
3185 indirection via a pseudo tty.
3187 * systemd-nspawn gained a new switch to control the UNIX
3188 signal to use when killing the init process of the container
3191 * systemd-nspawn gained a new --overlay= switch for mounting
3192 overlay file systems into the container using the new kernel
3195 * When a container image is imported via systemd-importd and
3196 the host file system is not btrfs, a loopback block device
3197 file is created in /var/lib/machines.raw with a btrfs file
3198 system inside. It is then mounted to /var/lib/machines to
3199 enable btrfs features for container management. The loopback
3200 file and btrfs file system is grown as needed when container
3201 images are imported via systemd-importd.
3203 * systemd-machined/systemd-importd gained support for btrfs
3204 quota, to enforce container disk space limits on disk. This
3205 is exposed in "machinectl set-limit".
3207 * systemd-importd now can import containers from local .tar,
3208 .raw and .qcow2 images, and export them to .tar and .raw. It
3209 can also import dkr v2 images now from the network (on top
3212 * systemd-importd gained support for verifying downloaded
3213 images with gpg2 (previously only gpg1 was supported).
3215 * systemd-machined, systemd-logind, systemd: most bus calls
3216 are now accessible to unprivileged processes via
3217 PolicyKit. Also, systemd-logind will now allow users to kill
3218 their own sessions without further privileges or
3221 * systemd-shutdownd has been removed. This service was
3222 previously responsible for implementing scheduled shutdowns
3223 as exposed in /usr/bin/shutdown's time parameter. This
3224 functionality has now been moved into systemd-logind and is
3225 accessible via a bus interface.
3227 * "systemctl reboot" gained a new switch --firmware-setup that
3228 can be used to reboot into the EFI firmware setup, if that
3229 is available. systemd-logind now exposes an API on the bus
3230 to trigger such reboots, in case graphical desktop UIs want
3231 to cover this functionality.
3233 * "systemctl enable", "systemctl disable" and "systemctl mask"
3234 now support a new "--now" switch. If specified the units
3235 that are enabled will also be started, and the ones
3236 disabled/masked also stopped.
3238 * The Gummiboot EFI boot loader tool has been merged into
3239 systemd, and renamed to "systemd-boot". The bootctl tool has been
3240 updated to support systemd-boot.
3242 * An EFI kernel stub has been added that may be used to create
3243 kernel EFI binaries that contain not only the actual kernel,
3244 but also an initrd, boot splash, command line and OS release
3245 information. This combined binary can then be signed as a
3246 single image, so that the firmware can verify it all in one
3247 step. systemd-boot has special support for EFI binaries created
3248 like this and can extract OS release information from them
3249 and show them in the boot menu. This functionality is useful
3250 to implement cryptographically verified boot schemes.
3252 * Optional support has been added to systemd-fsck to pass
3253 fsck's progress report to an AF_UNIX socket in the file
3256 * udev will no longer create device symlinks for all block
3257 devices by default. A blacklist for excluding special block
3258 devices from this logic has been turned into a whitelist
3259 that requires picking block devices explicitly that require
3262 * A new (currently still internal) API sd-device.h has been
3263 added to libsystemd. This modernized API is supposed to
3264 replace libudev eventually. In fact, already much of libudev
3265 is now just a wrapper around sd-device.h.
3267 * A new hwdb database for storing metadata about pointing
3268 stick devices has been added.
3270 * systemd-tmpfiles gained support for setting file attributes
3271 similar to the "chattr" tool with new 'h' and 'H' lines.
3273 * systemd-journald will no longer unconditionally set the
3274 btrfs NOCOW flag on new journal files. This is instead done
3275 with tmpfiles snippet using the new 'h' line type. This
3276 allows easy disabling of this logic, by masking the
3277 journal-nocow.conf tmpfiles file.
3279 * systemd-journald will now translate audit message types to
3280 human readable identifiers when writing them to the
3281 journal. This should improve readability of audit messages.
3283 * The LUKS logic gained support for the offset= and skip=
3284 options in /etc/crypttab, as previously implemented by
3287 * /usr/lib/os-release gained a new optional field VARIANT= for
3288 distributions that support multiple variants (such as a
3289 desktop edition, a server edition, ...)
3291 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3292 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3293 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3294 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3295 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3296 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3297 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3298 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3299 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3300 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3301 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3302 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3303 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3304 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3305 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3306 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3307 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3308 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3309 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3310 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3311 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3312 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3313 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3314 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3315 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3316 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3317 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3319 — Berlin, 2015-05-22
3323 * Introduce a new API "sd-hwdb.h" for querying the hardware
3324 metadata database. With this minimal interface one can query
3325 and enumerate the udev hwdb, decoupled from the old libudev
3326 library. libudev's interface for this is now only a wrapper
3327 around sd-hwdb. A new tool systemd-hwdb has been added to
3328 interface with and update the database.
3330 * When any of systemd's tools copies files (for example due to
3331 tmpfiles' C lines) a btrfs reflink will attempted first,
3332 before bytewise copying is done.
3334 * systemd-nspawn gained a new --ephemeral switch. When
3335 specified a btrfs snapshot is taken of the container's root
3336 directory, and immediately removed when the container
3337 terminates again. Thus, a container can be started whose
3338 changes never alter the container's root directory, and are
3339 lost on container termination. This switch can also be used
3340 for starting a container off the root file system of the
3341 host without affecting the host OS. This switch is only
3342 available on btrfs file systems.
3344 * systemd-nspawn gained a new --template= switch. It takes the
3345 path to a container tree to use as template for the tree
3346 specified via --directory=, should that directory be
3347 missing. This allows instantiating containers dynamically,
3348 on first run. This switch is only available on btrfs file
3351 * When a .mount unit refers to a mount point on which multiple
3352 mounts are stacked, and the .mount unit is stopped all of
3353 the stacked mount points will now be unmounted until no
3354 mount point remains.
3356 * systemd now has an explicit notion of supported and
3357 unsupported unit types. Jobs enqueued for unsupported unit
3358 types will now fail with an "unsupported" error code. More
3359 specifically .swap, .automount and .device units are not
3360 supported in containers, .busname units are not supported on
3361 non-kdbus systems. .swap and .automount are also not
3362 supported if their respective kernel compile time options
3365 * machinectl gained support for two new "copy-from" and
3366 "copy-to" commands for copying files from a running
3367 container to the host or vice versa.
3369 * machinectl gained support for a new "bind" command to bind
3370 mount host directories into local containers. This is
3371 currently only supported for nspawn containers.
3373 * networkd gained support for configuring bridge forwarding
3374 database entries (fdb) from .network files.
3376 * A new tiny daemon "systemd-importd" has been added that can
3377 download container images in tar, raw, qcow2 or dkr formats,
3378 and make them available locally in /var/lib/machines, so
3379 that they can run as nspawn containers. The daemon can GPG
3380 verify the downloads (not supported for dkr, since it has no
3381 provisions for verifying downloads). It will transparently
3382 decompress bz2, xz, gzip compressed downloads if necessary,
3383 and restore sparse files on disk. The daemon uses privilege
3384 separation to ensure the actual download logic runs with
3385 fewer privileges than the daemon itself. machinectl has
3386 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3387 make the functionality of importd available to the
3388 user. With this in place the Fedora and Ubuntu "Cloud"
3389 images can be downloaded and booted as containers unmodified
3390 (the Fedora images lack the appropriate GPG signature files
3391 currently, so they cannot be verified, but this will change
3392 soon, hopefully). Note that downloading images is currently
3393 only fully supported on btrfs.
3395 * machinectl is now able to list container images found in
3396 /var/lib/machines, along with some metadata about sizes of
3397 disk and similar. If the directory is located on btrfs and
3398 quota is enabled, this includes quota display. A new command
3399 "image-status" has been added that shows additional
3400 information about images.
3402 * machinectl is now able to clone container images
3403 efficiently, if the underlying file system (btrfs) supports
3404 it, with the new "machinectl clone" command. It also
3405 gained commands for renaming and removing images, as well as
3406 marking them read-only or read-write (supported also on
3407 legacy file systems).
3409 * networkd gained support for collecting LLDP network
3410 announcements, from hardware that supports this. This is
3411 shown in networkctl output.
3413 * systemd-run gained support for a new -t (--pty) switch for
3414 invoking a binary on a pty whose input and output is
3415 connected to the invoking terminal. This allows executing
3416 processes as system services while interactively
3417 communicating with them via the terminal. Most interestingly
3418 this is supported across container boundaries. Invoking
3419 "systemd-run -t /bin/bash" is an alternative to running a
3420 full login session, the difference being that the former
3421 will not register a session, nor go through the PAM session
3424 * tmpfiles gained support for a new "v" line type for creating
3425 btrfs subvolumes. If the underlying file system is a legacy
3426 file system, this automatically degrades to creating a
3427 normal directory. Among others /var/lib/machines is now
3428 created like this at boot, should it be missing.
3430 * The directory /var/lib/containers/ has been deprecated and
3431 been replaced by /var/lib/machines. The term "machines" has
3432 been used in the systemd context as generic term for both
3433 VMs and containers, and hence appears more appropriate for
3434 this, as the directory can also contain raw images bootable
3437 * systemd-nspawn when invoked with -M but without --directory=
3438 or --image= is now capable of searching for the container
3439 root directory, subvolume or disk image automatically, in
3440 /var/lib/machines. systemd-nspawn@.service has been updated
3441 to make use of this, thus allowing it to be used for raw
3444 * A new machines.target unit has been introduced that is
3445 supposed to group all containers/VMs invoked as services on
3446 the system. systemd-nspawn@.service has been updated to
3447 integrate with that.
3449 * machinectl gained a new "start" command, for invoking a
3450 container as a service. "machinectl start foo" is mostly
3451 equivalent to "systemctl start systemd-nspawn@foo.service",
3452 but handles escaping in a nicer way.
3454 * systemd-nspawn will now mount most of the cgroupfs tree
3455 read-only into each container, with the exception of the
3456 container's own subtree in the name=systemd hierarchy.
3458 * journald now sets the special FS_NOCOW file flag for its
3459 journal files. This should improve performance on btrfs, by
3460 avoiding heavy fragmentation when journald's write-pattern
3461 is used on COW file systems. It degrades btrfs' data
3462 integrity guarantees for the files to the same levels as for
3463 ext3/ext4 however. This should be OK though as journald does
3464 its own data integrity checks and all its objects are
3465 checksummed on disk. Also, journald should handle btrfs disk
3466 full events a lot more gracefully now, by processing SIGBUS
3467 errors, and not relying on fallocate() anymore.
3469 * When journald detects that journal files it is writing to
3470 have been deleted it will immediately start new journal
3473 * systemd now provides a way to store file descriptors
3474 per-service in PID 1. This is useful for daemons to ensure
3475 that fds they require are not lost during a daemon
3476 restart. The fds are passed to the daemon on the next
3477 invocation in the same way socket activation fds are
3478 passed. This is now used by journald to ensure that the
3479 various sockets connected to all the system's stdout/stderr
3480 are not lost when journald is restarted. File descriptors
3481 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3482 an extension to sd_notify(). Note that a limit is enforced
3483 on the number of fds a service can store in PID 1, and it
3484 defaults to 0, so that no fds may be stored, unless this is
3485 explicitly turned on.
3487 * The default TERM variable to use for units connected to a
3488 terminal, when no other value is explicitly is set is now
3489 vt220 rather than vt102. This should be fairly safe still,
3490 but allows PgUp/PgDn work.
3492 * The /etc/crypttab option header= as known from Debian is now
3495 * "loginctl user-status" and "loginctl session-status" will
3496 now show the last 10 lines of log messages of the
3497 user/session following the status output. Similar,
3498 "machinectl status" will show the last 10 log lines
3499 associated with a virtual machine or container
3500 service. (Note that this is usually not the log messages
3501 done in the VM/container itself, but simply what the
3502 container manager logs. For nspawn this includes all console
3505 * "loginctl session-status" without further argument will now
3506 show the status of the session of the caller. Similar,
3507 "lock-session", "unlock-session", "activate",
3508 "enable-linger", "disable-linger" may now be called without
3509 session/user parameter in which case they apply to the
3510 caller's session/user.
3512 * An X11 session scriptlet is now shipped that uploads
3513 $DISPLAY and $XAUTHORITY into the environment of the systemd
3514 --user daemon if a session begins. This should improve
3515 compatibility with X11 enabled applications run as systemd
3518 * Generators are now subject to masking via /etc and /run, the
3519 same way as unit files.
3521 * networkd .network files gained support for configuring
3522 per-link IPv4/IPv6 packet forwarding as well as IPv4
3523 masquerading. This is by default turned on for veth links to
3524 containers, as registered by systemd-nspawn. This means that
3525 nspawn containers run with --network-veth will now get
3526 automatic routed access to the host's networks without any
3527 further configuration or setup, as long as networkd runs on
3530 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3531 or UDP posts of a container on the host. With this in place
3532 it is possible to run containers with private veth links
3533 (--network-veth), and have their functionality exposed on
3534 the host as if their services were running directly on the
3537 * systemd-nspawn's --network-veth switch now gained a short
3538 version "-n", since with the changes above it is now truly
3539 useful out-of-the-box. The systemd-nspawn@.service has been
3540 updated to make use of it too by default.
3542 * systemd-nspawn will now maintain a per-image R/W lock, to
3543 ensure that the same image is not started more than once
3544 writable. (It's OK to run an image multiple times
3545 simultaneously in read-only mode.)
3547 * systemd-nspawn's --image= option is now capable of
3548 dissecting and booting MBR and GPT disk images that contain
3549 only a single active Linux partition. Previously it
3550 supported only GPT disk images with proper GPT type
3551 IDs. This allows running cloud images from major
3552 distributions directly with systemd-nspawn, without
3555 * In addition to collecting mouse dpi data in the udev
3556 hardware database, there's now support for collecting angle
3557 information for mouse scroll wheels. The database is
3558 supposed to guarantee similar scrolling behavior on mice
3559 that it knows about. There's also support for collecting
3560 information about Touchpad types.
3562 * udev's input_id built-in will now also collect touch screen
3563 dimension data and attach it to probed devices.
3565 * /etc/os-release gained support for a Distribution Privacy
3568 * networkd gained support for creating "ipvlan", "gretap",
3569 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3571 * systemd-tmpfiles gained support for "a" lines for setting
3574 * systemd-nspawn will now mount /tmp in the container to
3575 tmpfs, automatically.
3577 * systemd now exposes the memory.usage_in_bytes cgroup
3578 attribute and shows it for each service in the "systemctl
3579 status" output, if available.
3581 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3582 immediate reboot is triggered. This useful if shutdown is
3583 hung and is unable to complete, to expedite the
3584 operation. Note that this kind of reboot will still unmount
3585 all file systems, and hence should not result in fsck being
3588 * A .device unit for an optical block device will now be
3589 considered active only when a medium is in the drive. Also,
3590 mount units are now bound to their backing devices thus
3591 triggering automatic unmounting when devices become
3592 unavailable. With this in place systemd will now
3593 automatically unmount left-over mounts when a CD-ROM is
3594 ejected or an USB stick is yanked from the system.
3596 * networkd-wait-online now has support for waiting for
3597 specific interfaces only (with globbing), and for giving up
3598 after a configurable timeout.
3600 * networkd now exits when idle. It will be automatically
3601 restarted as soon as interfaces show up, are removed or
3602 change state. networkd will stay around as long as there is
3603 at least one DHCP state machine or similar around, that keep
3606 * networkd may now configure IPv6 link-local addressing in
3607 addition to IPv4 link-local addressing.
3609 * The IPv6 "token" for use in SLAAC may now be configured for
3610 each .network interface in networkd.
3612 * Routes configured with networkd may now be assigned a scope
3615 * networkd's [Match] sections now support globbing and lists
3616 of multiple space-separated matches per item.
3618 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3619 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3620 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3621 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3622 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3623 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3624 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3625 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3626 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3627 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3628 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3629 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3630 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3631 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3632 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3633 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3634 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3635 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3636 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3637 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3638 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3639 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3640 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3641 Hoffmann, Zbigniew Jędrzejewski-Szmek
3643 — Berlin, 2015-02-16
3647 * When querying unit file enablement status (for example via
3648 "systemctl is-enabled"), a new state "indirect" is now known
3649 which indicates that a unit might not be enabled itself, but
3650 another unit listed in its Also= setting might be.
3652 * Similar to the various existing ConditionXYZ= settings for
3653 units, there are now matching AssertXYZ= settings. While
3654 failing conditions cause a unit to be skipped, but its job
3655 to succeed, failing assertions declared like this will cause
3656 a unit start operation and its job to fail.
3658 * hostnamed now knows a new chassis type "embedded".
3660 * systemctl gained a new "edit" command. When used on a unit
3661 file, this allows extending unit files with .d/ drop-in
3662 configuration snippets or editing the full file (after
3663 copying it from /usr/lib to /etc). This will invoke the
3664 user's editor (as configured with $EDITOR), and reload the
3665 modified configuration after editing.
3667 * "systemctl status" now shows the suggested enablement state
3668 for a unit, as declared in the (usually vendor-supplied)
3669 system preset files.
3671 * nss-myhostname will now resolve the single-label host name
3672 "gateway" to the locally configured default IP routing
3673 gateways, ordered by their metrics. This assigns a stable
3674 name to the used gateways, regardless which ones are
3675 currently configured. Note that the name will only be
3676 resolved after all other name sources (if nss-myhostname is
3677 configured properly) and should hence not negatively impact
3678 systems that use the single-label host name "gateway" in
3681 * systemd-inhibit now allows filtering by mode when listing
3684 * Scope and service units gained a new "Delegate" boolean
3685 property, which, when set, allows processes running inside the
3686 unit to further partition resources. This is primarily
3687 useful for systemd user instances as well as container
3690 * journald will now pick up audit messages directly from
3691 the kernel, and log them like any other log message. The
3692 audit fields are split up and fully indexed. This means that
3693 journalctl in many ways is now a (nicer!) alternative to
3694 ausearch, the traditional audit client. Note that this
3695 implements only a minimal audit client. If you want the
3696 special audit modes like reboot-on-log-overflow, please use
3697 the traditional auditd instead, which can be used in
3698 parallel to journald.
3700 * The ConditionSecurity= unit file option now understands the
3701 special string "audit" to check whether auditing is
3704 * journalctl gained two new commands --vacuum-size= and
3705 --vacuum-time= to delete old journal files until the
3706 remaining ones take up no more than the specified size on disk,
3707 or are not older than the specified time.
3709 * A new, native PPPoE library has been added to sd-network,
3710 systemd's library of light-weight networking protocols. This
3711 library will be used in a future version of networkd to
3712 enable PPPoE communication without an external pppd daemon.
3714 * The busctl tool now understands a new "capture" verb that
3715 works similar to "monitor", but writes a packet capture
3716 trace to STDOUT that can be redirected to a file which is
3717 compatible with libcap's capture file format. This can then
3718 be loaded in Wireshark and similar tools to inspect bus
3721 * The busctl tool now understands a new "tree" verb that shows
3722 the object trees of a specific service on the bus, or of all
3725 * The busctl tool now understands a new "introspect" verb that
3726 shows all interfaces and members of objects on the bus,
3727 including their signature and values. This is particularly
3728 useful to get more information about bus objects shown by
3729 the new "busctl tree" command.
3731 * The busctl tool now understands new verbs "call",
3732 "set-property" and "get-property" for invoking bus method
3733 calls, setting and getting bus object properties in a
3736 * busctl gained a new --augment-creds= argument that controls
3737 whether the tool shall augment credential information it
3738 gets from the bus with data from /proc, in a possibly
3741 * nspawn's --link-journal= switch gained two new values
3742 "try-guest" and "try-host" that work like "guest" and
3743 "host", but do not fail if the host has no persistent
3744 journalling enabled. -j is now equivalent to
3745 --link-journal=try-guest.
3747 * macvlan network devices created by nspawn will now have
3748 stable MAC addresses.
3750 * A new SmackProcessLabel= unit setting has been added, which
3751 controls the SMACK security label processes forked off by
3752 the respective unit shall use.
3754 * If compiled with --enable-xkbcommon, systemd-localed will
3755 verify x11 keymap settings by compiling the given keymap. It
3756 will spew out warnings if the compilation fails. This
3757 requires libxkbcommon to be installed.
3759 * When a coredump is collected, a larger number of metadata
3760 fields is now collected and included in the journal records
3761 created for it. More specifically, control group membership,
3762 environment variables, memory maps, working directory,
3763 chroot directory, /proc/$PID/status, and a list of open file
3764 descriptors is now stored in the log entry.
3766 * The udev hwdb now contains DPI information for mice. For
3769 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3771 * All systemd programs that read standalone configuration
3772 files in /etc now also support a corresponding series of
3773 .conf.d configuration directories in /etc/, /run/,
3774 /usr/local/lib/, /usr/lib/, and (if configured with
3775 --enable-split-usr) /lib/. In particular, the following
3776 configuration files now have corresponding configuration
3777 directories: system.conf user.conf, logind.conf,
3778 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3779 resolved.conf, timesyncd.conf, journal-remote.conf, and
3780 journal-upload.conf. Note that distributions should use the
3781 configuration directories in /usr/lib/; the directories in
3782 /etc/ are reserved for the system administrator.
3784 * systemd-rfkill will no longer take the rfkill device name
3785 into account when storing rfkill state on disk, as the name
3786 might be dynamically assigned and not stable. Instead, the
3787 ID_PATH udev variable combined with the rfkill type (wlan,
3788 bluetooth, ...) is used.
3790 * A new service systemd-machine-id-commit.service has been
3791 added. When used on systems where /etc is read-only during
3792 boot, and /etc/machine-id is not initialized (but an empty
3793 file), this service will copy the temporary machine ID
3794 created as replacement into /etc after the system is fully
3795 booted up. This is useful for systems that are freshly
3796 installed with a non-initialized machine ID, but should get
3797 a fixed machine ID for subsequent boots.
3799 * networkd's .netdev files now provide a large set of
3800 configuration parameters for VXLAN devices. Similarly, the
3801 bridge port cost parameter is now configurable in .network
3802 files. There's also new support for configuring IP source
3803 routing. networkd .link files gained support for a new
3804 OriginalName= match that is useful to match against the
3805 original interface name the kernel assigned. .network files
3806 may include MTU= and MACAddress= fields for altering the MTU
3807 and MAC address while being connected to a specific network
3810 * The LUKS logic gained supported for configuring
3811 UUID-specific key files. There's also new support for naming
3812 LUKS device from the kernel command line, using the new
3813 luks.name= argument.
3815 * Timer units may now be transiently created via the bus API
3816 (this was previously already available for scope and service
3817 units). In addition it is now possible to create multiple
3818 transient units at the same time with a single bus call. The
3819 "systemd-run" tool has been updated to make use of this for
3820 running commands on a specified time, in at(1)-style.
3822 * tmpfiles gained support for "t" lines, for assigning
3823 extended attributes to files. Among other uses this may be
3824 used to assign SMACK labels to files.
3826 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3827 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3828 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3829 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3830 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3831 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3832 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3833 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3834 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3835 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3836 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3837 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3838 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3839 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3840 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3841 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3842 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3843 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3845 — Berlin, 2014-12-10
3849 * journalctl gained the new options -t/--identifier= to match
3850 on the syslog identifier (aka "tag"), as well as --utc to
3851 show log timestamps in the UTC timezone. journalctl now also
3852 accepts -n/--lines=all to disable line capping in a pager.
3854 * journalctl gained a new switch, --flush, that synchronously
3855 flushes logs from /run/log/journal to /var/log/journal if
3856 persistent storage is enabled. systemd-journal-flush.service
3857 now waits until the operation is complete.
3859 * Services can notify the manager before they start a reload
3860 (by sending RELOADING=1) or shutdown (by sending
3861 STOPPING=1). This allows the manager to track and show the
3862 internal state of daemons and closes a race condition when
3863 the process is still running but has closed its D-Bus
3866 * Services with Type=oneshot do not have to have any ExecStart
3869 * User units are now loaded also from
3870 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3871 /run/systemd/user directory that was already previously
3872 supported, but is under the control of the user.
3874 * Job timeouts (i.e. time-outs on the time a job that is
3875 queued stays in the run queue) can now optionally result in
3876 immediate reboot or power-off actions (JobTimeoutAction= and
3877 JobTimeoutRebootArgument=). This is useful on ".target"
3878 units, to limit the maximum time a target remains
3879 undispatched in the run queue, and to trigger an emergency
3880 operation in such a case. This is now used by default to
3881 turn off the system if boot-up (as defined by everything in
3882 basic.target) hangs and does not complete for at least
3883 15min. Also, if power-off or reboot hang for at least 30min
3884 an immediate power-off/reboot operation is triggered. This
3885 functionality is particularly useful to increase reliability
3886 on embedded devices, but also on laptops which might
3887 accidentally get powered on when carried in a backpack and
3888 whose boot stays stuck in a hard disk encryption passphrase
3891 * systemd-logind can be configured to also handle lid switch
3892 events even when the machine is docked or multiple displays
3893 are attached (HandleLidSwitchDocked= option).
3895 * A helper binary and a service have been added which can be
3896 used to resume from hibernation in the initramfs. A
3897 generator will parse the resume= option on the kernel
3898 command line to trigger resume.
3900 * A user console daemon systemd-consoled has been
3901 added. Currently, it is a preview, and will so far open a
3902 single terminal on each session of the user marked as
3903 Desktop=systemd-console.
3905 * Route metrics can be specified for DHCP routes added by
3908 * The SELinux context of socket-activated services can be set
3909 from the information provided by the networking stack
3910 (SELinuxContextFromNet= option).
3912 * Userspace firmware loading support has been removed and
3913 the minimum supported kernel version is thus bumped to 3.7.
3915 * Timeout for udev workers has been increased from 1 to 3
3916 minutes, but a warning will be printed after 1 minute to
3917 help diagnose kernel modules that take a long time to load.
3919 * Udev rules can now remove tags on devices with TAG-="foobar".
3921 * systemd's readahead implementation has been removed. In many
3922 circumstances it didn't give expected benefits even for
3923 rotational disk drives and was becoming less relevant in the
3924 age of SSDs. As none of the developers has been using
3925 rotating media anymore, and nobody stepped up to actively
3926 maintain this component of systemd it has now been removed.
3928 * Swap units can use Options= to specify discard options.
3929 Discard options specified for swaps in /etc/fstab are now
3932 * Docker containers are now detected as a separate type of
3935 * The Password Agent protocol gained support for queries where
3936 the user input is shown, useful e.g. for user names.
3937 systemd-ask-password gained a new --echo option to turn that
3940 * The default sysctl.d/ snippets will now set:
3942 net.core.default_qdisc = fq_codel
3944 This selects Fair Queuing Controlled Delay as the default
3945 queuing discipline for network interfaces. fq_codel helps
3946 fight the network bufferbloat problem. It is believed to be
3947 a good default with no tuning required for most workloads.
3948 Downstream distributions may override this choice. On 10Gbit
3949 servers that do not do forwarding, "fq" may perform better.
3950 Systems without a good clocksource should use "pfifo_fast".
3952 * If kdbus is enabled during build a new option BusPolicy= is
3953 available for service units, that allows locking all service
3954 processes into a stricter bus policy, in order to limit
3955 access to various bus services, or even hide most of them
3956 from the service's view entirely.
3958 * networkctl will now show the .network and .link file
3959 networkd has applied to a specific interface.
3961 * sd-login gained a new API call sd_session_get_desktop() to
3962 query which desktop environment has been selected for a
3965 * UNIX utmp support is now compile-time optional to support
3966 legacy-free systems.
3968 * systemctl gained two new commands "add-wants" and
3969 "add-requires" for pulling in units from specific targets
3972 * If the word "rescue" is specified on the kernel command line
3973 the system will now boot into rescue mode (aka
3974 rescue.target), which was previously available only by
3975 specifying "1" or "systemd.unit=rescue.target" on the kernel
3976 command line. This new kernel command line option nicely
3977 mirrors the already existing "emergency" kernel command line
3980 * New kernel command line options mount.usr=, mount.usrflags=,
3981 mount.usrfstype= have been added that match root=, rootflags=,
3982 rootfstype= but allow mounting a specific file system to
3985 * The $NOTIFY_SOCKET is now also passed to control processes of
3986 services, not only the main process.
3988 * This version reenables support for fsck's -l switch. This
3989 means at least version v2.25 of util-linux is required for
3990 operation, otherwise dead-locks on device nodes may
3991 occur. Again: you need to update util-linux to at least
3992 v2.25 when updating systemd to v217.
3994 * The "multi-seat-x" tool has been removed from systemd, as
3995 its functionality has been integrated into X servers 1.16,
3996 and the tool is hence redundant. It is recommended to update
3997 display managers invoking this tool to simply invoke X
3998 directly from now on, again.
4000 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4001 message flag has been added for all of systemd's PolicyKit
4002 authenticated method calls has been added. In particular
4003 this now allows optional interactive authorization via
4004 PolicyKit for many of PID1's privileged operations such as
4005 unit file enabling and disabling.
4007 * "udevadm hwdb --update" learnt a new switch "--usr" for
4008 placing the rebuilt hardware database in /usr instead of
4009 /etc. When used only hardware database entries stored in
4010 /usr will be used, and any user database entries in /etc are
4011 ignored. This functionality is useful for vendors to ship a
4012 pre-built database on systems where local configuration is
4013 unnecessary or unlikely.
4015 * Calendar time specifications in .timer units now also
4016 understand the strings "semi-annually", "quarterly" and
4017 "minutely" as shortcuts (in addition to the preexisting
4018 "anually", "hourly", ...).
4020 * systemd-tmpfiles will now correctly create files in /dev
4021 at boot which are marked for creation only at boot. It is
4022 recommended to always create static device nodes with 'c!'
4023 and 'b!', so that they are created only at boot and not
4024 overwritten at runtime.
4026 * When the watchdog logic is used for a service (WatchdogSec=)
4027 and the watchdog timeout is hit the service will now be
4028 terminated with SIGABRT (instead of just SIGTERM), in order
4029 to make sure a proper coredump and backtrace is
4030 generated. This ensures that hanging services will result in
4031 similar coredump/backtrace behaviour as services that hit a
4034 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4035 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4036 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4037 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4038 Herrmann, David Sommerseth, David Strauss, Emil Renner
4039 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4040 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4041 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4042 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4043 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4044 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4045 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4046 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4047 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4048 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4049 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4050 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4051 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4052 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4053 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4054 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4057 — Berlin, 2014-10-28
4061 * timedated no longer reads NTP implementation unit names from
4062 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4063 implementations should add a
4065 Conflicts=systemd-timesyncd.service
4067 to their unit files to take over and replace systemd's NTP
4068 default functionality.
4070 * systemd-sysusers gained a new line type "r" for configuring
4071 which UID/GID ranges to allocate system users/groups
4072 from. Lines of type "u" may now add an additional column
4073 that specifies the home directory for the system user to be
4074 created. Also, systemd-sysusers may now optionally read user
4075 information from STDIN instead of a file. This is useful for
4076 invoking it from RPM preinst scriptlets that need to create
4077 users before the first RPM file is installed since these
4078 files might need to be owned by them. A new
4079 %sysusers_create_inline RPM macro has been introduced to do
4080 just that. systemd-sysusers now updates the shadow files as
4081 well as the user/group databases, which should enhance
4082 compatibility with certain tools like grpck.
4084 * A number of bus APIs of PID 1 now optionally consult
4085 PolicyKit to permit access for otherwise unprivileged
4086 clients under certain conditions. Note that this currently
4087 doesn't support interactive authentication yet, but this is
4088 expected to be added eventually, too.
4090 * /etc/machine-info now has new fields for configuring the
4091 deployment environment of the machine, as well as the
4092 location of the machine. hostnamectl has been updated with
4093 new command to update these fields.
4095 * systemd-timesyncd has been updated to automatically acquire
4096 NTP server information from systemd-networkd, which might
4097 have been discovered via DHCP.
4099 * systemd-resolved now includes a caching DNS stub resolver
4100 and a complete LLMNR name resolution implementation. A new
4101 NSS module "nss-resolve" has been added which can be used
4102 instead of glibc's own "nss-dns" to resolve hostnames via
4103 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4104 be resolved via systemd-resolved D-Bus APIs. In contrast to
4105 the glibc internal resolver systemd-resolved is aware of
4106 multi-homed system, and keeps DNS server and caches separate
4107 and per-interface. Queries are sent simultaneously on all
4108 interfaces that have DNS servers configured, in order to
4109 properly handle VPNs and local LANs which might resolve
4110 separate sets of domain names. systemd-resolved may acquire
4111 DNS server information from systemd-networkd automatically,
4112 which in turn might have discovered them via DHCP. A tool
4113 "systemd-resolve-host" has been added that may be used to
4114 query the DNS logic in resolved. systemd-resolved implements
4115 IDNA and automatically uses IDNA or UTF-8 encoding depending
4116 on whether classic DNS or LLMNR is used as transport. In the
4117 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4118 implementation to systemd-resolved.
4120 * A new NSS module nss-mymachines has been added, that
4121 automatically resolves the names of all local registered
4122 containers to their respective IP addresses.
4124 * A new client tool "networkctl" for systemd-networkd has been
4125 added. It currently is entirely passive and will query
4126 networking configuration from udev, rtnetlink and networkd,
4127 and present it to the user in a very friendly
4128 way. Eventually, we hope to extend it to become a full
4129 control utility for networkd.
4131 * .socket units gained a new DeferAcceptSec= setting that
4132 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4133 TCP. Similarly, support for controlling TCP keep-alive
4134 settings has been added (KeepAliveTimeSec=,
4135 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4136 turning off Nagle's algorithm on TCP has been added
4139 * logind learned a new session type "web", for use in projects
4140 like Cockpit which register web clients as PAM sessions.
4142 * timer units with at least one OnCalendar= setting will now
4143 be started only after time-sync.target has been
4144 reached. This way they will not elapse before the system
4145 clock has been corrected by a local NTP client or
4146 similar. This is particular useful on RTC-less embedded
4147 machines, that come up with an invalid system clock.
4149 * systemd-nspawn's --network-veth= switch should now result in
4150 stable MAC addresses for both the outer and the inner side
4153 * systemd-nspawn gained a new --volatile= switch for running
4154 container instances with /etc or /var unpopulated.
4156 * The kdbus client code has been updated to use the new Linux
4157 3.17 memfd subsystem instead of the old kdbus-specific one.
4159 * systemd-networkd's DHCP client and server now support
4160 FORCERENEW. There are also new configuration options to
4161 configure the vendor client identifier and broadcast mode
4164 * systemd will no longer inform the kernel about the current
4165 timezone, as this is necessarily incorrect and racy as the
4166 kernel has no understanding of DST and similar
4167 concepts. This hence means FAT timestamps will be always
4168 considered UTC, similar to what Android is already
4169 doing. Also, when the RTC is configured to the local time
4170 (rather than UTC) systemd will never synchronize back to it,
4171 as this might confuse Windows at a later boot.
4173 * systemd-analyze gained a new command "verify" for offline
4174 validation of unit files.
4176 * systemd-networkd gained support for a couple of additional
4177 settings for bonding networking setups. Also, the metric for
4178 statically configured routes may now be configured. For
4179 network interfaces where this is appropriate the peer IP
4180 address may now be configured.
4182 * systemd-networkd's DHCP client will no longer request
4183 broadcasting by default, as this tripped up some networks.
4184 For hardware where broadcast is required the feature should
4185 be switched back on using RequestBroadcast=yes.
4187 * systemd-networkd will now set up IPv4LL addresses (when
4188 enabled) even if DHCP is configured successfully.
4190 * udev will now default to respect network device names given
4191 by the kernel when the kernel indicates that these are
4192 predictable. This behavior can be tweaked by changing
4193 NamePolicy= in the relevant .link file.
4195 * A new library systemd-terminal has been added that
4196 implements full TTY stream parsing and rendering. This
4197 library is supposed to be used later on for implementing a
4198 full userspace VT subsystem, replacing the current kernel
4201 * A new tool systemd-journal-upload has been added to push
4202 journal data to a remote system running
4203 systemd-journal-remote.
4205 * journald will no longer forward all local data to another
4206 running syslog daemon. This change has been made because
4207 rsyslog (which appears to be the most commonly used syslog
4208 implementation these days) no longer makes use of this, and
4209 instead pulls the data out of the journal on its own. Since
4210 forwarding the messages to a non-existent syslog server is
4211 more expensive than we assumed we have now turned this
4212 off. If you run a syslog server that is not a recent rsyslog
4213 version, you have to turn this option on again
4214 (ForwardToSyslog= in journald.conf).
4216 * journald now optionally supports the LZ4 compressor for
4217 larger journal fields. This compressor should perform much
4218 better than XZ which was the previous default.
4220 * machinectl now shows the IP addresses of local containers,
4221 if it knows them, plus the interface name of the container.
4223 * A new tool "systemd-escape" has been added that makes it
4224 easy to escape strings to build unit names and similar.
4226 * sd_notify() messages may now include a new ERRNO= field
4227 which is parsed and collected by systemd and shown among the
4228 "systemctl status" output for a service.
4230 * A new component "systemd-firstboot" has been added that
4231 queries the most basic systemd information (timezone,
4232 hostname, root password) interactively on first
4233 boot. Alternatively it may also be used to provision these
4234 things offline on OS images installed into directories.
4236 * The default sysctl.d/ snippets will now set
4238 net.ipv4.conf.default.promote_secondaries=1
4240 This has the benefit of no flushing secondary IP addresses
4241 when primary addresses are removed.
4243 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4244 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4245 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4246 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4247 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4248 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4249 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4250 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4251 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4252 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4253 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4254 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4255 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4256 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4257 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4259 — Berlin, 2014-08-19
4263 * A new tool systemd-sysusers has been added. This tool
4264 creates system users and groups in /etc/passwd and
4265 /etc/group, based on static declarative system user/group
4266 definitions in /usr/lib/sysusers.d/. This is useful to
4267 enable factory resets and volatile systems that boot up with
4268 an empty /etc directory, and thus need system users and
4269 groups created during early boot. systemd now also ships
4270 with two default sysusers.d/ files for the most basic
4271 users and groups systemd and the core operating system
4274 * A new tmpfiles snippet has been added that rebuilds the
4275 essential files in /etc on boot, should they be missing.
4277 * A directive for ensuring automatic clean-up of
4278 /var/cache/man/ has been removed from the default
4279 configuration. This line should now be shipped by the man
4280 implementation. The necessary change has been made to the
4281 man-db implementation. Note that you need to update your man
4282 implementation to one that ships this line, otherwise no
4283 automatic clean-up of /var/cache/man will take place.
4285 * A new condition ConditionNeedsUpdate= has been added that
4286 may conditionalize services to only run when /etc or /var
4287 are "older" than the vendor operating system resources in
4288 /usr. This is useful for reconstructing or updating /etc
4289 after an offline update of /usr or a factory reset, on the
4290 next reboot. Services that want to run once after such an
4291 update or reset should use this condition and order
4292 themselves before the new systemd-update-done.service, which
4293 will mark the two directories as fully updated. A number of
4294 service files have been added making use of this, to rebuild
4295 the udev hardware database, the journald message catalog and
4296 dynamic loader cache (ldconfig). The systemd-sysusers tool
4297 described above also makes use of this now. With this in
4298 place it is now possible to start up a minimal operating
4299 system with /etc empty cleanly. For more information on the
4300 concepts involved see this recent blog story:
4302 http://0pointer.de/blog/projects/stateless.html
4304 * A new system group "input" has been introduced, and all
4305 input device nodes get this group assigned. This is useful
4306 for system-level software to get access to input devices. It
4307 complements what is already done for "audio" and "video".
4309 * systemd-networkd learnt minimal DHCPv4 server support in
4310 addition to the existing DHCPv4 client support. It also
4311 learnt DHCPv6 client and IPv6 Router Solicitation client
4312 support. The DHCPv4 client gained support for static routes
4313 passed in from the server. Note that the [DHCPv4] section
4314 known in older systemd-networkd versions has been renamed to
4315 [DHCP] and is now also used by the DHCPv6 client. Existing
4316 .network files using settings of this section should be
4317 updated, though compatibility is maintained. Optionally, the
4318 client hostname may now be sent to the DHCP server.
4320 * networkd gained support for vxlan virtual networks as well
4321 as tun/tap and dummy devices.
4323 * networkd gained support for automatic allocation of address
4324 ranges for interfaces from a system-wide pool of
4325 addresses. This is useful for dynamically managing a large
4326 number of interfaces with a single network configuration
4327 file. In particular this is useful to easily assign
4328 appropriate IP addresses to the veth links of a large number
4329 of nspawn instances.
4331 * RPM macros for processing sysusers, sysctl and binfmt
4332 drop-in snippets at package installation time have been
4335 * The /etc/os-release file should now be placed in
4336 /usr/lib/os-release. The old location is automatically
4337 created as symlink. /usr/lib is the more appropriate
4338 location of this file, since it shall actually describe the
4339 vendor operating system shipped in /usr, and not the
4340 configuration stored in /etc.
4342 * .mount units gained a new boolean SloppyOptions= setting
4343 that maps to mount(8)'s -s option which enables permissive
4344 parsing of unknown mount options.
4346 * tmpfiles learnt a new "L+" directive which creates a symlink
4347 but (unlike "L") deletes a pre-existing file first, should
4348 it already exist and not already be the correct
4349 symlink. Similarly, "b+", "c+" and "p+" directives have been
4350 added as well, which create block and character devices, as
4351 well as fifos in the filesystem, possibly removing any
4352 pre-existing files of different types.
4354 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4355 'argument' field (which so far specified the source to
4356 symlink/copy the files from) is now optional. If omitted the
4357 same file os copied from /usr/share/factory/ suffixed by the
4358 full destination path. This is useful for populating /etc
4359 with essential files, by copying them from vendor defaults
4360 shipped in /usr/share/factory/etc.
4362 * A new command "systemctl preset-all" has been added that
4363 applies the service preset settings to all installed unit
4364 files. A new switch --preset-mode= has been added that
4365 controls whether only enable or only disable operations
4368 * A new command "systemctl is-system-running" has been added
4369 that allows checking the overall state of the system, for
4370 example whether it is fully up and running.
4372 * When the system boots up with an empty /etc, the equivalent
4373 to "systemctl preset-all" is executed during early boot, to
4374 make sure all default services are enabled after a factory
4377 * systemd now contains a minimal preset file that enables the
4378 most basic services systemd ships by default.
4380 * Unit files' [Install] section gained a new DefaultInstance=
4381 field for defining the default instance to create if a
4382 template unit is enabled with no instance specified.
4384 * A new passive target cryptsetup-pre.target has been added
4385 that may be used by services that need to make they run and
4386 finish before the first LUKS cryptographic device is set up.
4388 * The /dev/loop-control and /dev/btrfs-control device nodes
4389 are now owned by the "disk" group by default, opening up
4390 access to this group.
4392 * systemd-coredump will now automatically generate a
4393 stack trace of all core dumps taking place on the system,
4394 based on elfutils' libdw library. This stack trace is logged
4397 * systemd-coredump may now optionally store coredumps directly
4398 on disk (in /var/lib/systemd/coredump, possibly compressed),
4399 instead of storing them unconditionally in the journal. This
4400 mode is the new default. A new configuration file
4401 /etc/systemd/coredump.conf has been added to configure this
4402 and other parameters of systemd-coredump.
4404 * coredumpctl gained a new "info" verb to show details about a
4405 specific coredump. A new switch "-1" has also been added
4406 that makes sure to only show information about the most
4407 recent entry instead of all entries. Also, as the tool is
4408 generally useful now the "systemd-" prefix of the binary
4409 name has been removed. Distributions that want to maintain
4410 compatibility with the old name should add a symlink from
4411 the old name to the new name.
4413 * journald's SplitMode= now defaults to "uid". This makes sure
4414 that unprivileged users can access their own coredumps with
4415 coredumpctl without restrictions.
4417 * New kernel command line options "systemd.wants=" (for
4418 pulling an additional unit during boot), "systemd.mask="
4419 (for masking a specific unit for the boot), and
4420 "systemd.debug-shell" (for enabling the debug shell on tty9)
4421 have been added. This is implemented in the new generator
4422 "systemd-debug-generator".
4424 * systemd-nspawn will now by default filter a couple of
4425 syscalls for containers, among them those required for
4426 kernel module loading, direct x86 IO port access, swap
4427 management, and kexec. Most importantly though
4428 open_by_handle_at() is now prohibited for containers,
4429 closing a hole similar to a recently discussed vulnerability
4430 in docker regarding access to files on file hierarchies the
4431 container should normally not have access to. Note that, for
4432 nspawn, we generally make no security claims anyway (and
4433 this is explicitly documented in the man page), so this is
4434 just a fix for one of the most obvious problems.
4436 * A new man page file-hierarchy(7) has been added that
4437 contains a minimized, modernized version of the file system
4438 layout systemd expects, similar in style to the FHS
4439 specification or hier(5). A new tool systemd-path(1) has
4440 been added to query many of these paths for the local
4443 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4444 longer done. Since the directory now has a per-user size
4445 limit, and is cleaned on logout this appears unnecessary,
4446 in particular since this now brings the lifecycle of this
4447 directory closer in line with how IPC objects are handled.
4449 * systemd.pc now exports a number of additional directories,
4450 including $libdir (which is useful to identify the library
4451 path for the primary architecture of the system), and a
4452 couple of drop-in directories.
4454 * udev's predictable network interface names now use the dev_port
4455 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4456 distinguish between ports of the same PCI function. dev_id should
4457 only be used for ports using the same HW address, hence the need
4460 * machined has been updated to export the OS version of a
4461 container (read from /etc/os-release and
4462 /usr/lib/os-release) on the bus. This is now shown in
4463 "machinectl status" for a machine.
4465 * A new service setting RestartForceExitStatus= has been
4466 added. If configured to a set of exit signals or process
4467 return values, the service will be restarted when the main
4468 daemon process exits with any of them, regardless of the
4471 * systemctl's -H switch for connecting to remote systemd
4472 machines has been extended so that it may be used to
4473 directly connect to a specific container on the
4474 host. "systemctl -H root@foobar:waldi" will now connect as
4475 user "root" to host "foobar", and then proceed directly to
4476 the container named "waldi". Note that currently you have to
4477 authenticate as user "root" for this to work, as entering
4478 containers is a privileged operation.
4480 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4481 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4482 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4483 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4484 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4485 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4486 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4487 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4488 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4489 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4490 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4491 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4493 — Berlin, 2014-07-03
4497 * As an experimental feature, udev now tries to lock the
4498 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4499 executes events for the disk or any of its partitions.
4500 Applications like partitioning programs can lock the
4501 disk device node (flock(LOCK_EX)) and claim temporary
4502 device ownership that way; udev will entirely skip all event
4503 handling for this disk and its partitions. If the disk
4504 was opened for writing, the close will trigger a partition
4505 table rescan in udev's "watch" facility, and if needed
4506 synthesize "change" events for the disk and all its partitions.
4507 This is now unconditionally enabled, and if it turns out to
4508 cause major problems, we might turn it on only for specific
4509 devices, or might need to disable it entirely. Device Mapper
4510 devices are excluded from this logic.
4512 * We temporarily dropped the "-l" switch for fsck invocations,
4513 since they collide with the flock() logic above. util-linux
4514 upstream has been changed already to avoid this conflict,
4515 and we will readd "-l" as soon as util-linux with this
4516 change has been released.
4518 * The dependency on libattr has been removed. Since a long
4519 time, the extended attribute calls have moved to glibc, and
4520 libattr is thus unnecessary.
4522 * Virtualization detection works without privileges now. This
4523 means the systemd-detect-virt binary no longer requires
4524 CAP_SYS_PTRACE file capabilities, and our daemons can run
4525 with fewer privileges.
4527 * systemd-networkd now runs under its own "systemd-network"
4528 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4529 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4530 loses the ability to write to files owned by root this way.
4532 * Similarly, systemd-resolved now runs under its own
4533 "systemd-resolve" user with no capabilities remaining.
4535 * Similarly, systemd-bus-proxyd now runs under its own
4536 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4538 * systemd-networkd gained support for setting up "veth"
4539 virtual Ethernet devices for container connectivity, as well
4540 as GRE and VTI tunnels.
4542 * systemd-networkd will no longer automatically attempt to
4543 manually load kernel modules necessary for certain tunnel
4544 transports. Instead, it is assumed the kernel loads them
4545 automatically when required. This only works correctly on
4546 very new kernels. On older kernels, please consider adding
4547 the kernel modules to /etc/modules-load.d/ as a work-around.
4549 * The resolv.conf file systemd-resolved generates has been
4550 moved to /run/systemd/resolve/. If you have a symlink from
4551 /etc/resolv.conf, it might be necessary to correct it.
4553 * Two new service settings, ProtectHome= and ProtectSystem=,
4554 have been added. When enabled, they will make the user data
4555 (such as /home) inaccessible or read-only and the system
4556 (such as /usr) read-only, for specific services. This allows
4557 very light-weight per-service sandboxing to avoid
4558 modifications of user data or system files from
4559 services. These two new switches have been enabled for all
4560 of systemd's long-running services, where appropriate.
4562 * Socket units gained new SocketUser= and SocketGroup=
4563 settings to set the owner user and group of AF_UNIX sockets
4564 and FIFOs in the file system.
4566 * Socket units gained a new RemoveOnStop= setting. If enabled,
4567 all FIFOS and sockets in the file system will be removed
4568 when the specific socket unit is stopped.
4570 * Socket units gained a new Symlinks= setting. It takes a list
4571 of symlinks to create to file system sockets or FIFOs
4572 created by the specific Unix sockets. This is useful to
4573 manage symlinks to socket nodes with the same life-cycle as
4576 * The /dev/log socket and /dev/initctl FIFO have been moved to
4577 /run, and have been replaced by symlinks. This allows
4578 connecting to these facilities even if PrivateDevices=yes is
4579 used for a service (which makes /dev/log itself unavailable,
4580 but /run is left). This also has the benefit of ensuring
4581 that /dev only contains device nodes, directories and
4582 symlinks, and nothing else.
4584 * sd-daemon gained two new calls sd_pid_notify() and
4585 sd_pid_notifyf(). They are similar to sd_notify() and
4586 sd_notifyf(), but allow overriding of the source PID of
4587 notification messages if permissions permit this. This is
4588 useful to send notify messages on behalf of a different
4589 process (for example, the parent process). The
4590 systemd-notify tool has been updated to make use of this
4591 when sending messages (so that notification messages now
4592 originate from the shell script invoking systemd-notify and
4593 not the systemd-notify process itself. This should minimize
4594 a race where systemd fails to associate notification
4595 messages to services when the originating process already
4598 * A new "on-abnormal" setting for Restart= has been added. If
4599 set, it will result in automatic restarts on all "abnormal"
4600 reasons for a process to exit, which includes unclean
4601 signals, core dumps, timeouts and watchdog timeouts, but
4602 does not include clean and unclean exit codes or clean
4603 signals. Restart=on-abnormal is an alternative for
4604 Restart=on-failure for services that shall be able to
4605 terminate and avoid restarts on certain errors, by
4606 indicating so with an unclean exit code. Restart=on-failure
4607 or Restart=on-abnormal is now the recommended setting for
4608 all long-running services.
4610 * If the InaccessibleDirectories= service setting points to a
4611 mount point (or if there are any submounts contained within
4612 it), it is now attempted to completely unmount it, to make
4613 the file systems truly unavailable for the respective
4616 * The ReadOnlyDirectories= service setting and
4617 systemd-nspawn's --read-only parameter are now recursively
4618 applied to all submounts, too.
4620 * Mount units may now be created transiently via the bus APIs.
4622 * The support for SysV and LSB init scripts has been removed
4623 from the systemd daemon itself. Instead, it is now
4624 implemented as a generator that creates native systemd units
4625 from these scripts when needed. This enables us to remove a
4626 substantial amount of legacy code from PID 1, following the
4627 fact that many distributions only ship a very small number
4628 of LSB/SysV init scripts nowadays.
4630 * Privileged Xen (dom0) domains are not considered
4631 virtualization anymore by the virtualization detection
4632 logic. After all, they generally have unrestricted access to
4633 the hardware and usually are used to manage the unprivileged
4636 * systemd-tmpfiles gained a new "C" line type, for copying
4637 files or entire directories.
4639 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4640 lines. So far, they have been non-globbing versions of the
4641 latter, and have thus been redundant. In future, it is
4642 recommended to only use "z". "m" has hence been removed
4643 from the documentation, even though it stays supported.
4645 * A tmpfiles snippet to recreate the most basic structure in
4646 /var has been added. This is enough to create the /var/run →
4647 /run symlink and create a couple of structural
4648 directories. This allows systems to boot up with an empty or
4649 volatile /var. Of course, while with this change, the core OS
4650 now is capable with dealing with a volatile /var, not all
4651 user services are ready for it. However, we hope that sooner
4652 or later, many service daemons will be changed upstream so
4653 that they are able to automatically create their necessary
4654 directories in /var at boot, should they be missing. This is
4655 the first step to allow state-less systems that only require
4656 the vendor image for /usr to boot.
4658 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4659 empty tmpfs instance to a specific directory. This is
4660 particularly useful for making use of the automatic
4661 reconstruction of /var (see above), by passing --tmpfs=/var.
4663 * Access modes specified in tmpfiles snippets may now be
4664 prefixed with "~", which indicates that they shall be masked
4665 by whether the existing file or directory is currently
4666 writable, readable or executable at all. Also, if specified,
4667 the sgid/suid/sticky bits will be masked for all
4670 * A new passive target unit "network-pre.target" has been
4671 added which is useful for services that shall run before any
4672 network is configured, for example firewall scripts.
4674 * The "floppy" group that previously owned the /dev/fd*
4675 devices is no longer used. The "disk" group is now used
4676 instead. Distributions should probably deprecate usage of
4679 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4680 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4681 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4682 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4683 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4684 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4687 — Berlin, 2014-06-11
4691 * A new "systemd-timesyncd" daemon has been added for
4692 synchronizing the system clock across the network. It
4693 implements an SNTP client. In contrast to NTP
4694 implementations such as chrony or the NTP reference server,
4695 this only implements a client side, and does not bother with
4696 the full NTP complexity, focusing only on querying time from
4697 one remote server and synchronizing the local clock to
4698 it. Unless you intend to serve NTP to networked clients or
4699 want to connect to local hardware clocks, this simple NTP
4700 client should be more than appropriate for most
4701 installations. The daemon runs with minimal privileges, and
4702 has been hooked up with networkd to only operate when
4703 network connectivity is available. The daemon saves the
4704 current clock to disk every time a new NTP sync has been
4705 acquired, and uses this to possibly correct the system clock
4706 early at bootup, in order to accommodate for systems that
4707 lack an RTC such as the Raspberry Pi and embedded devices,
4708 and to make sure that time monotonically progresses on these
4709 systems, even if it is not always correct. To make use of
4710 this daemon, a new system user and group "systemd-timesync"
4711 needs to be created on installation of systemd.
4713 * The queue "seqnum" interface of libudev has been disabled, as
4714 it was generally incompatible with device namespacing as
4715 sequence numbers of devices go "missing" if the devices are
4716 part of a different namespace.
4718 * "systemctl list-timers" and "systemctl list-sockets" gained
4719 a --recursive switch for showing units of these types also
4720 for all local containers, similar in style to the already
4721 supported --recursive switch for "systemctl list-units".
4723 * A new RebootArgument= setting has been added for service
4724 units, which may be used to specify a kernel reboot argument
4725 to use when triggering reboots with StartLimitAction=.
4727 * A new FailureAction= setting has been added for service
4728 units which may be used to specify an operation to trigger
4729 when a service fails. This works similarly to
4730 StartLimitAction=, but unlike it, controls what is done
4731 immediately rather than only after several attempts to
4732 restart the service in question.
4734 * hostnamed got updated to also expose the kernel name,
4735 release, and version on the bus. This is useful for
4736 executing commands like hostnamectl with the -H switch.
4737 systemd-analyze makes use of this to properly display
4738 details when running non-locally.
4740 * The bootchart tool can now show cgroup information in the
4741 graphs it generates.
4743 * The CFS CPU quota cgroup attribute is now exposed for
4744 services. The new CPUQuota= switch has been added for this
4745 which takes a percentage value. Setting this will have the
4746 result that a service may never get more CPU time than the
4747 specified percentage, even if the machine is otherwise idle.
4749 * systemd-networkd learned IPIP and SIT tunnel support.
4751 * LSB init scripts exposing a dependency on $network will now
4752 get a dependency on network-online.target rather than simply
4753 network.target. This should bring LSB handling closer to
4754 what it was on SysV systems.
4756 * A new fsck.repair= kernel option has been added to control
4757 how fsck shall deal with unclean file systems at boot.
4759 * The (.ini) configuration file parser will now silently
4760 ignore sections whose name begins with "X-". This may be
4761 used to maintain application-specific extension sections in unit
4764 * machined gained a new API to query the IP addresses of
4765 registered containers. "machinectl status" has been updated
4766 to show these addresses in its output.
4768 * A new call sd_uid_get_display() has been added to the
4769 sd-login APIs for querying the "primary" session of a
4770 user. The "primary" session of the user is elected from the
4771 user's sessions and generally a graphical session is
4772 preferred over a text one.
4774 * A minimal systemd-resolved daemon has been added. It
4775 currently simply acts as a companion to systemd-networkd and
4776 manages resolv.conf based on per-interface DNS
4777 configuration, possibly supplied via DHCP. In the long run
4778 we hope to extend this into a local DNSSEC enabled DNS and
4781 * The systemd-networkd-wait-online tool is now enabled by
4782 default. It will delay network-online.target until a network
4783 connection has been configured. The tool primarily integrates
4784 with networkd, but will also make a best effort to make sense
4785 of network configuration performed in some other way.
4787 * Two new service options StartupCPUShares= and
4788 StartupBlockIOWeight= have been added that work similarly to
4789 CPUShares= and BlockIOWeight= however only apply during
4790 system startup. This is useful to prioritize certain services
4791 differently during bootup than during normal runtime.
4793 * hostnamed has been changed to prefer the statically
4794 configured hostname in /etc/hostname (unless set to
4795 'localhost' or empty) over any dynamic one supplied by
4796 dhcp. With this change, the rules for picking the hostname
4797 match more closely the rules of other configuration settings
4798 where the local administrator's configuration in /etc always
4799 overrides any other settings.
4801 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4802 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4803 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4804 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4805 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4806 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4807 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4808 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4809 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4810 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4811 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4812 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4813 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4814 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4815 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4816 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4819 — Beijing, 2014-05-28
4823 * When restoring the screen brightness at boot, stay away from
4824 the darkest setting or from the lowest 5% of the available
4825 range, depending on which is the larger value of both. This
4826 should effectively protect the user from rebooting into a
4827 black screen, should the brightness have been set to minimum
4830 * sd-login gained a new sd_machine_get_class() call to
4831 determine the class ("vm" or "container") of a machine
4832 registered with machined.
4834 * sd-login gained new calls
4835 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4836 to query the identity of the peer of a local AF_UNIX
4837 connection. They operate similarly to their sd_pid_get_xyz()
4840 * PID 1 will now maintain a system-wide system state engine
4841 with the states "starting", "running", "degraded",
4842 "maintenance", "stopping". These states are bound to system
4843 startup, normal runtime, runtime with at least one failed
4844 service, rescue/emergency mode and system shutdown. This
4845 state is shown in the "systemctl status" output when no unit
4846 name is passed. It is useful to determine system state, in
4847 particularly when doing so for many systems or containers at
4850 * A new command "list-machines" has been added to "systemctl"
4851 that lists all local OS containers and shows their system
4852 state (see above), if systemd runs inside of them.
4854 * systemctl gained a new "-r" switch to recursively enumerate
4855 units on all local containers, when used with the
4856 "list-unit" command (which is the default one that is
4857 executed when no parameters are specified).
4859 * The GPT automatic partition discovery logic will now honour
4860 two GPT partition flags: one may be set on a partition to
4861 cause it to be mounted read-only, and the other may be set
4862 on a partition to ignore it during automatic discovery.
4864 * Two new GPT type UUIDs have been added for automatic root
4865 partition discovery, for 32-bit and 64-bit ARM. This is not
4866 particularly useful for discovering the root directory on
4867 these architectures during bare-metal boots (since UEFI is
4868 not common there), but still very useful to allow booting of
4869 ARM disk images in nspawn with the -i option.
4871 * MAC addresses of interfaces created with nspawn's
4872 --network-interface= switch will now be generated from the
4873 machine name, and thus be stable between multiple invocations
4876 * logind will now automatically remove all IPC objects owned
4877 by a user if she or he fully logs out. This makes sure that
4878 users who are logged out cannot continue to consume IPC
4879 resources. This covers SysV memory, semaphores and message
4880 queues as well as POSIX shared memory and message
4881 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4882 limits. With this functionality, that is corrected. This may
4883 be turned off by using the RemoveIPC= switch of logind.conf.
4885 * The systemd-machine-id-setup and tmpfiles tools gained a
4886 --root= switch to operate on a specific root directory,
4889 * journald can now forward logged messages to the TTYs of all
4890 logged in users ("wall"). This is the default for all
4891 emergency messages now.
4893 * A new tool systemd-journal-remote has been added to stream
4894 journal log messages across the network.
4896 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4897 controller trees are mounted into it. Note that the
4898 directories mounted beneath it are not read-only. This is a
4899 security measure and is particularly useful because glibc
4900 actually includes a search logic to pick any tmpfs it can
4901 find to implement shm_open() if /dev/shm is not available
4902 (which it might very well be in namespaced setups).
4904 * machinectl gained a new "poweroff" command to cleanly power
4905 down a local OS container.
4907 * The PrivateDevices= unit file setting will now also drop the
4908 CAP_MKNOD capability from the capability bound set, and
4909 imply DevicePolicy=closed.
4911 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4912 comprehensively on all long-running systemd services where
4913 this is appropriate.
4915 * systemd-udevd will now run in a disassociated mount
4916 namespace. To mount directories from udev rules, make sure to
4917 pull in mount units via SYSTEMD_WANTS properties.
4919 * The kdbus support gained support for uploading policy into
4920 the kernel. sd-bus gained support for creating "monitoring"
4921 connections that can eavesdrop into all bus communication
4922 for debugging purposes.
4924 * Timestamps may now be specified in seconds since the UNIX
4925 epoch Jan 1st, 1970 by specifying "@" followed by the value
4928 * Native tcpwrap support in systemd has been removed. tcpwrap
4929 is old code, not really maintained anymore and has serious
4930 shortcomings, and better options such as firewalls
4931 exist. For setups that require tcpwrap usage, please
4932 consider invoking your socket-activated service via tcpd,
4933 like on traditional inetd.
4935 * A new system.conf configuration option
4936 DefaultTimerAccuracySec= has been added that controls the
4937 default AccuracySec= setting of .timer units.
4939 * Timer units gained a new WakeSystem= switch. If enabled,
4940 timers configured this way will cause the system to resume
4941 from system suspend (if the system supports that, which most
4944 * Timer units gained a new Persistent= switch. If enabled,
4945 timers configured this way will save to disk when they have
4946 been last triggered. This information is then used on next
4947 reboot to possible execute overdue timer events, that
4948 could not take place because the system was powered off.
4949 This enables simple anacron-like behaviour for timer units.
4951 * systemctl's "list-timers" will now also list the time a
4952 timer unit was last triggered in addition to the next time
4953 it will be triggered.
4955 * systemd-networkd will now assign predictable IPv4LL
4956 addresses to its local interfaces.
4958 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4959 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4960 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4961 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4962 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4963 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4964 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4965 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4968 — Berlin, 2014-03-25
4972 * A new unit file setting RestrictAddressFamilies= has been
4973 added to restrict which socket address families unit
4974 processes gain access to. This takes address family names
4975 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4976 attack surface of services via exotic protocol stacks. This
4977 is built on seccomp system call filters.
4979 * Two new unit file settings RuntimeDirectory= and
4980 RuntimeDirectoryMode= have been added that may be used to
4981 manage a per-daemon runtime directories below /run. This is
4982 an alternative for setting up directory permissions with
4983 tmpfiles snippets, and has the advantage that the runtime
4984 directory's lifetime is bound to the daemon runtime and that
4985 the daemon starts up with an empty directory each time. This
4986 is particularly useful when writing services that drop
4987 privileges using the User= or Group= setting.
4989 * The DeviceAllow= unit setting now supports globbing for
4990 matching against device group names.
4992 * The systemd configuration file system.conf gained new
4993 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4994 DefaultMemoryAccounting= to globally turn on/off accounting
4995 for specific resources (cgroups) for all units. These
4996 settings may still be overridden individually in each unit
4999 * systemd-gpt-auto-generator is now able to discover /srv and
5000 root partitions in addition to /home and swap partitions. It
5001 also supports LUKS-encrypted partitions now. With this in
5002 place, automatic discovery of partitions to mount following
5003 the Discoverable Partitions Specification
5004 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5005 is now a lot more complete. This allows booting without
5006 /etc/fstab and without root= on the kernel command line on
5007 systems prepared appropriately.
5009 * systemd-nspawn gained a new --image= switch which allows
5010 booting up disk images and Linux installations on any block
5011 device that follow the Discoverable Partitions Specification
5012 (see above). This means that installations made with
5013 appropriately updated installers may now be started and
5014 deployed using container managers, completely
5015 unmodified. (We hope that libvirt-lxc will add support for
5016 this feature soon, too.)
5018 * systemd-nspawn gained a new --network-macvlan= setting to
5019 set up a private macvlan interface for the
5020 container. Similarly, systemd-networkd gained a new
5021 Kind=macvlan setting in .netdev files.
5023 * systemd-networkd now supports configuring local addresses
5026 * A new tool systemd-network-wait-online has been added to
5027 synchronously wait for network connectivity using
5030 * The sd-bus.h bus API gained a new sd_bus_track object for
5031 tracking the life-cycle of bus peers. Note that sd-bus.h is
5032 still not a public API though (unless you specify
5033 --enable-kdbus on the configure command line, which however
5034 voids your warranty and you get no API stability guarantee).
5036 * The $XDG_RUNTIME_DIR runtime directories for each user are
5037 now individual tmpfs instances, which has the benefit of
5038 introducing separate pools for each user, with individual
5039 size limits, and thus making sure that unprivileged clients
5040 can no longer negatively impact the system or other users by
5041 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5042 RuntimeDirectorySize= has been introduced that allows
5043 controlling the default size limit for all users. It
5044 defaults to 10% of the available physical memory. This is no
5045 replacement for quotas on tmpfs though (which the kernel
5046 still does not support), as /dev/shm and /tmp are still
5047 shared resources used by both the system and unprivileged
5050 * logind will now automatically turn off automatic suspending
5051 on laptop lid close when more than one display is
5052 connected. This was previously expected to be implemented
5053 individually in desktop environments (such as GNOME),
5054 however has been added to logind now, in order to fix a
5055 boot-time race where a desktop environment might not have
5056 been started yet and thus not been able to take an inhibitor
5057 lock at the time where logind already suspends the system
5058 due to a closed lid.
5060 * logind will now wait at least 30s after each system
5061 suspend/resume cycle, and 3min after system boot before
5062 suspending the system due to a closed laptop lid. This
5063 should give USB docking stations and similar enough time to
5064 be probed and configured after system resume and boot in
5065 order to then act as suspend blocker.
5067 * systemd-run gained a new --property= setting which allows
5068 initialization of resource control properties (and others)
5069 for the created scope or service unit. Example: "systemd-run
5070 --property=BlockIOWeight=10 updatedb" may be used to run
5071 updatedb at a low block IO scheduling weight.
5073 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5074 now also work in --scope mode.
5076 * When systemd is compiled with kdbus support, basic support
5077 for enforced policies is now in place. (Note that enabling
5078 kdbus still voids your warranty and no API compatibility
5081 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5082 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5083 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5084 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5085 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5086 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5087 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5088 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5089 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5090 Zbigniew Jędrzejewski-Szmek
5092 — Berlin, 2014-03-12
5096 * systemd will now relabel /dev after loading the SMACK policy
5097 according to SMACK rules.
5099 * A new unit file option AppArmorProfile= has been added to
5100 set the AppArmor profile for the processes of a unit.
5102 * A new condition check ConditionArchitecture= has been added
5103 to conditionalize units based on the system architecture, as
5104 reported by uname()'s "machine" field.
5106 * systemd-networkd now supports matching on the system
5107 virtualization, architecture, kernel command line, host name
5110 * logind is now a lot more aggressive when suspending the
5111 machine due to a closed laptop lid. Instead of acting only
5112 on the lid close action, it will continuously watch the lid
5113 status and act on it. This is useful for laptops where the
5114 power button is on the outside of the chassis so that it can
5115 be reached without opening the lid (such as the Lenovo
5116 Yoga). On those machines, logind will now immediately
5117 re-suspend the machine if the power button has been
5118 accidentally pressed while the laptop was suspended and in a
5119 backpack or similar.
5121 * logind will now watch SW_DOCK switches and inhibit reaction
5122 to the lid switch if it is pressed. This means that logind
5123 will not suspend the machine anymore if the lid is closed
5124 and the system is docked, if the laptop supports SW_DOCK
5125 notifications via the input layer. Note that ACPI docking
5126 stations do not generate this currently. Also note that this
5127 logic is usually not fully sufficient and Desktop
5128 Environments should take a lid switch inhibitor lock when an
5129 external display is connected, as systemd will not watch
5132 * nspawn will now make use of the devices cgroup controller by
5133 default, and only permit creation of and access to the usual
5134 API device nodes like /dev/null or /dev/random, as well as
5135 access to (but not creation of) the pty devices.
5137 * We will now ship a default .network file for
5138 systemd-networkd that automatically configures DHCP for
5139 network interfaces created by nspawn's --network-veth or
5140 --network-bridge= switches.
5142 * systemd will now understand the usual M, K, G, T suffixes
5143 according to SI conventions (i.e. to the base 1000) when
5144 referring to throughput and hardware metrics. It will stay
5145 with IEC conventions (i.e. to the base 1024) for software
5146 metrics, according to what is customary according to
5147 Wikipedia. We explicitly document which base applies for
5148 each configuration option.
5150 * The DeviceAllow= setting in unit files now supports a syntax
5151 to whitelist an entire group of devices node majors at once,
5152 based on the /proc/devices listing. For example, with the
5153 string "char-pts", it is now possible to whitelist all
5154 current and future pseudo-TTYs at once.
5156 * sd-event learned a new "post" event source. Event sources of
5157 this type are triggered by the dispatching of any event
5158 source of a type that is not "post". This is useful for
5159 implementing clean-up and check event sources that are
5160 triggered by other work being done in the program.
5162 * systemd-networkd is no longer statically enabled, but uses
5163 the usual [Install] sections so that it can be
5164 enabled/disabled using systemctl. It still is enabled by
5167 * When creating a veth interface pair with systemd-nspawn, the
5168 host side will now be prefixed with "vb-" if
5169 --network-bridge= is used, and with "ve-" if --network-veth
5170 is used. This way, it is easy to distinguish these cases on
5171 the host, for example to apply different configuration to
5172 them with systemd-networkd.
5174 * The compatibility libraries for libsystemd-journal.so,
5175 libsystem-id128.so, libsystemd-login.so and
5176 libsystemd-daemon.so do not make use of IFUNC
5177 anymore. Instead, we now build libsystemd.so multiple times
5178 under these alternative names. This means that the footprint
5179 is drastically increased, but given that these are
5180 transitional compatibility libraries, this should not matter
5181 much. This change has been made necessary to support the ARM
5182 platform for these compatibility libraries, as the ARM
5183 toolchain is not really at the same level as the toolchain
5184 for other architectures like x86 and does not support
5185 IFUNC. Please make sure to use --enable-compat-libs only
5186 during a transitional period!
5188 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5189 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5190 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5191 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5192 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5193 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5194 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5195 Zbigniew Jędrzejewski-Szmek
5197 — Berlin, 2014-02-24
5201 * A new component "systemd-networkd" has been added that can
5202 be used to configure local network interfaces statically or
5203 via DHCP. It is capable of bringing up bridges, VLANs, and
5204 bonding. Currently, no hook-ups for interactive network
5205 configuration are provided. Use this for your initrd,
5206 container, embedded, or server setup if you need a simple,
5207 yet powerful, network configuration solution. This
5208 configuration subsystem is quite nifty, as it allows wildcard
5209 hotplug matching in interfaces. For example, with a single
5210 configuration snippet, you can configure that all Ethernet
5211 interfaces showing up are automatically added to a bridge,
5212 or similar. It supports link-sensing and more.
5214 * A new tool "systemd-socket-proxyd" has been added which can
5215 act as a bidirectional proxy for TCP sockets. This is
5216 useful for adding socket activation support to services that
5217 do not actually support socket activation, including virtual
5218 machines and the like.
5220 * Add a new tool to save/restore rfkill state on
5223 * Save/restore state of keyboard backlights in addition to
5224 display backlights on shutdown/boot.
5226 * udev learned a new SECLABEL{} construct to label device
5227 nodes with a specific security label when they appear. For
5228 now, only SECLABEL{selinux} is supported, but the syntax is
5229 prepared for additional security frameworks.
5231 * udev gained a new scheme to configure link-level attributes
5232 from files in /etc/systemd/network/*.link. These files can
5233 match against MAC address, device path, driver name and type,
5234 and will apply attributes like the naming policy, link speed,
5235 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5236 address assignment policy (randomized, ...).
5238 * The configuration of network interface naming rules for
5239 "permanent interface names" has changed: a new NamePolicy=
5240 setting in the [Link] section of .link files determines the
5241 priority of possible naming schemes (onboard, slot, MAC,
5242 path). The default value of this setting is determined by
5243 /usr/lib/net/links/99-default.link. Old
5244 80-net-name-slot.rules udev configuration file has been
5245 removed, so local configuration overriding this file should
5246 be adapted to override 99-default.link instead.
5248 * When the User= switch is used in a unit file, also
5249 initialize $SHELL= based on the user database entry.
5251 * systemd no longer depends on libdbus. All communication is
5252 now done with sd-bus, systemd's low-level bus library
5255 * kdbus support has been added to PID 1 itself. When kdbus is
5256 enabled, this causes PID 1 to set up the system bus and
5257 enable support for a new ".busname" unit type that
5258 encapsulates bus name activation on kdbus. It works a little
5259 bit like ".socket" units, except for bus names. A new
5260 generator has been added that converts classic dbus1 service
5261 activation files automatically into native systemd .busname
5264 * sd-bus: add a light-weight vtable implementation that allows
5265 defining objects on the bus with a simple static const
5266 vtable array of its methods, signals and properties.
5268 * systemd will not generate or install static dbus
5269 introspection data anymore to /usr/share/dbus-1/interfaces,
5270 as the precise format of these files is unclear, and
5271 nothing makes use of it.
5273 * A proxy daemon is now provided to proxy clients connecting
5274 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5275 compatibility with classic D-Bus.
5277 * A bus driver implementation has been added that supports the
5278 classic D-Bus bus driver calls on kdbus, also for
5279 compatibility purposes.
5281 * A new API "sd-event.h" has been added that implements a
5282 minimal event loop API built around epoll. It provides a
5283 couple of features that direct epoll usage is lacking:
5284 prioritization of events, scales to large numbers of timer
5285 events, per-event timer slack (accuracy), system-wide
5286 coalescing of timer events, exit handlers, watchdog
5287 supervision support using systemd's sd_notify() API, child
5290 * A new API "sd-rntl.h" has been added that provides an API
5291 around the route netlink interface of the kernel, similar in
5292 style to "sd-bus.h".
5294 * A new API "sd-dhcp-client.h" has been added that provides a
5295 small DHCPv4 client-side implementation. This is used by
5298 * There is a new kernel command line option
5299 "systemd.restore_state=0|1". When set to "0", none of the
5300 systemd tools will restore saved runtime state to hardware
5301 devices. More specifically, the rfkill and backlight states
5304 * The FsckPassNo= compatibility option in mount/service units
5305 has been removed. The fstab generator will now add the
5306 necessary dependencies automatically, and does not require
5307 PID1's support for that anymore.
5309 * journalctl gained a new switch, --list-boots, that lists
5310 recent boots with their times and boot IDs.
5312 * The various tools like systemctl, loginctl, timedatectl,
5313 busctl, systemd-run, ... have gained a new switch "-M" to
5314 connect to a specific, local OS container (as direct
5315 connection, without requiring SSH). This works on any
5316 container that is registered with machined, such as those
5317 created by libvirt-lxc or nspawn.
5319 * systemd-run and systemd-analyze also gained support for "-H"
5320 to connect to remote hosts via SSH. This is particularly
5321 useful for systemd-run because it enables queuing of jobs
5322 onto remote systems.
5324 * machinectl gained a new command "login" to open a getty
5325 login in any local container. This works with any container
5326 that is registered with machined (such as those created by
5327 libvirt-lxc or nspawn), and which runs systemd inside.
5329 * machinectl gained a new "reboot" command that may be used to
5330 trigger a reboot on a specific container that is registered
5331 with machined. This works on any container that runs an init
5332 system of some kind.
5334 * systemctl gained a new "list-timers" command to print a nice
5335 listing of installed timer units with the times they elapse
5338 * Alternative reboot() parameters may now be specified on the
5339 "systemctl reboot" command line and are passed to the
5340 reboot() system call.
5342 * systemctl gained a new --job-mode= switch to configure the
5343 mode to queue a job with. This is a more generic version of
5344 --fail, --irreversible, and --ignore-dependencies, which are
5345 still available but not advertised anymore.
5347 * /etc/systemd/system.conf gained new settings to configure
5348 various default timeouts of units, as well as the default
5349 start limit interval and burst. These may still be overridden
5352 * PID1 will now export on the bus profile data of the security
5353 policy upload process (such as the SELinux policy upload to
5356 * journald: when forwarding logs to the console, include
5357 timestamps (following the setting in
5358 /sys/module/printk/parameters/time).
5360 * OnCalendar= in timer units now understands the special
5361 strings "yearly" and "annually". (Both are equivalent)
5363 * The accuracy of timer units is now configurable with the new
5364 AccuracySec= setting. It defaults to 1min.
5366 * A new dependency type JoinsNamespaceOf= has been added that
5367 allows running two services within the same /tmp and network
5368 namespace, if PrivateNetwork= or PrivateTmp= are used.
5370 * A new command "cat" has been added to systemctl. It outputs
5371 the original unit file of a unit, and concatenates the
5372 contents of additional "drop-in" unit file snippets, so that
5373 the full configuration is shown.
5375 * systemctl now supports globbing on the various "list-xyz"
5376 commands, like "list-units" or "list-sockets", as well as on
5377 those commands which take multiple unit names.
5379 * journalctl's --unit= switch gained support for globbing.
5381 * All systemd daemons now make use of the watchdog logic so
5382 that systemd automatically notices when they hang.
5384 * If the $container_ttys environment variable is set,
5385 getty-generator will automatically spawn a getty for each
5386 listed tty. This is useful for container managers to request
5387 login gettys to be spawned on as many ttys as needed.
5389 * %h, %s, %U specifier support is not available anymore when
5390 used in unit files for PID 1. This is because NSS calls are
5391 not safe from PID 1. They stay available for --user
5392 instances of systemd, and as special case for the root user.
5394 * loginctl gained a new "--no-legend" switch to turn off output
5397 * The "sd-login.h" API gained three new calls:
5398 sd_session_is_remote(), sd_session_get_remote_user(),
5399 sd_session_get_remote_host() to query information about
5402 * The udev hardware database now also carries vendor/product
5403 information of SDIO devices.
5405 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5406 determine whether watchdog notifications are requested by
5409 * Socket-activated per-connection services now include a
5410 short description of the connection parameters in the
5413 * tmpfiles gained a new "--boot" option. When this is not used,
5414 only lines where the command character is not suffixed with
5415 "!" are executed. When this option is specified, those
5416 options are executed too. This partitions tmpfiles
5417 directives into those that can be safely executed at any
5418 time, and those which should be run only at boot (for
5419 example, a line that creates /run/nologin).
5421 * A new API "sd-resolve.h" has been added which provides a simple
5422 asynchronous wrapper around glibc NSS host name resolution
5423 calls, such as getaddrinfo(). In contrast to glibc's
5424 getaddrinfo_a(), it does not use signals. In contrast to most
5425 other asynchronous name resolution libraries, this one does
5426 not reimplement DNS, but reuses NSS, so that alternate
5427 host name resolution systems continue to work, such as mDNS,
5428 LDAP, etc. This API is based on libasyncns, but it has been
5429 cleaned up for inclusion in systemd.
5431 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5432 "sd-daemon.h" are no longer found in individual libraries
5433 libsystemd-journal.so, libsystemd-login.so,
5434 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5435 merged them into a single library, libsystemd.so, which
5436 provides all symbols. The reason for this is cyclic
5437 dependencies, as these libraries tend to use each other's
5438 symbols. So far, we have managed to workaround that by linking
5439 a copy of a good part of our code into each of these
5440 libraries again and again, which, however, makes certain
5441 things hard to do, like sharing static variables. Also, it
5442 substantially increases footprint. With this change, there
5443 is only one library for the basic APIs systemd
5444 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5445 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5446 library as well, however are subject to the --enable-kdbus
5447 switch (see below). Note that "sd-dhcp-client.h" is not part
5448 of this library (this is because it only consumes, never
5449 provides, services of/to other APIs). To make the transition
5450 easy from the separate libraries to the unified one, we
5451 provide the --enable-compat-libs compile-time switch which
5452 will generate stub libraries that are compatible with the
5453 old ones but redirect all calls to the new one.
5455 * All of the kdbus logic and the new APIs "sd-bus.h",
5456 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5457 and "sd-utf8.h" are compile-time optional via the
5458 "--enable-kdbus" switch, and they are not compiled in by
5459 default. To make use of kdbus, you have to explicitly enable
5460 the switch. Note however, that neither the kernel nor the
5461 userspace API for all of this is considered stable yet. We
5462 want to maintain the freedom to still change the APIs for
5463 now. By specifying this build-time switch, you acknowledge
5464 that you are aware of the instability of the current
5467 * Also, note that while kdbus is pretty much complete,
5468 it lacks one thing: proper policy support. This means you
5469 can build a fully working system with all features; however,
5470 it will be highly insecure. Policy support will be added in
5471 one of the next releases, at the same time that we will
5472 declare the APIs stable.
5474 * When the kernel command line argument "kdbus" is specified,
5475 systemd will automatically load the kdbus.ko kernel module. At
5476 this stage of development, it is only useful for testing kdbus
5477 and should not be used in production. Note: if "--enable-kdbus"
5478 is specified, and the kdbus.ko kernel module is available, and
5479 "kdbus" is added to the kernel command line, the entire system
5480 runs with kdbus instead of dbus-daemon, with the above mentioned
5481 problem of missing the system policy enforcement. Also a future
5482 version of kdbus.ko or a newer systemd will not be compatible with
5483 each other, and will unlikely be able to boot the machine if only
5484 one of them is updated.
5486 * systemctl gained a new "import-environment" command which
5487 uploads the caller's environment (or parts thereof) into the
5488 service manager so that it is inherited by services started
5489 by the manager. This is useful to upload variables like
5490 $DISPLAY into the user service manager.
5492 * A new PrivateDevices= switch has been added to service units
5493 which allows running a service with a namespaced /dev
5494 directory that does not contain any device nodes for
5495 physical devices. More specifically, it only includes devices
5496 such as /dev/null, /dev/urandom, and /dev/zero which are API
5499 * logind has been extended to support behaviour like VT
5500 switching on seats that do not support a VT. This makes
5501 multi-session available on seats that are not the first seat
5502 (seat0), and on systems where kernel support for VTs has
5503 been disabled at compile-time.
5505 * If a process holds a delay lock for system sleep or shutdown
5506 and fails to release it in time, we will now log its
5507 identity. This makes it easier to identify processes that
5508 cause slow suspends or power-offs.
5510 * When parsing /etc/crypttab, support for a new key-slot=
5511 option as supported by Debian is added. It allows indicating
5512 which LUKS slot to use on disk, speeding up key loading.
5514 * The sd_journal_sendv() API call has been checked and
5515 officially declared to be async-signal-safe so that it may
5516 be invoked from signal handlers for logging purposes.
5518 * Boot-time status output is now enabled automatically after a
5519 short timeout if boot does not progress, in order to give
5520 the user an indication what she or he is waiting for.
5522 * The boot-time output has been improved to show how much time
5523 remains until jobs expire.
5525 * The KillMode= switch in service units gained a new possible
5526 value "mixed". If set, and the unit is shut down, then the
5527 initial SIGTERM signal is sent only to the main daemon
5528 process, while the following SIGKILL signal is sent to
5529 all remaining processes of the service.
5531 * When a scope unit is registered, a new property "Controller"
5532 may be set. If set to a valid bus name, systemd will send a
5533 RequestStop() signal to this name when it would like to shut
5534 down the scope. This may be used to hook manager logic into
5535 the shutdown logic of scope units. Also, scope units may now
5536 be put in a special "abandoned" state, in which case the
5537 manager process which created them takes no further
5538 responsibilities for it.
5540 * When reading unit files, systemd will now verify
5541 the access mode of these files, and warn about certain
5542 suspicious combinations. This has been added to make it
5543 easier to track down packaging bugs where unit files are
5544 marked executable or world-writable.
5546 * systemd-nspawn gained a new "--setenv=" switch to set
5547 container-wide environment variables. The similar option in
5548 systemd-activate was renamed from "--environment=" to
5549 "--setenv=" for consistency.
5551 * systemd-nspawn has been updated to create a new kdbus domain
5552 for each container that is invoked, thus allowing each
5553 container to have its own set of system and user buses,
5554 independent of the host.
5556 * systemd-nspawn gained a new --drop-capability= switch to run
5557 the container with less capabilities than the default. Both
5558 --drop-capability= and --capability= now take the special
5559 string "all" for dropping or keeping all capabilities.
5561 * systemd-nspawn gained new switches for executing containers
5562 with specific SELinux labels set.
5564 * systemd-nspawn gained a new --quiet switch to not generate
5565 any additional output but the container's own console
5568 * systemd-nspawn gained a new --share-system switch to run a
5569 container without PID namespacing enabled.
5571 * systemd-nspawn gained a new --register= switch to control
5572 whether the container is registered with systemd-machined or
5573 not. This is useful for containers that do not run full
5574 OS images, but only specific apps.
5576 * systemd-nspawn gained a new --keep-unit which may be used
5577 when invoked as the only program from a service unit, and
5578 results in registration of the unit service itself in
5579 systemd-machined, instead of a newly opened scope unit.
5581 * systemd-nspawn gained a new --network-interface= switch for
5582 moving arbitrary interfaces to the container. The new
5583 --network-veth switch creates a virtual Ethernet connection
5584 between host and container. The new --network-bridge=
5585 switch then allows assigning the host side of this virtual
5586 Ethernet connection to a bridge device.
5588 * systemd-nspawn gained a new --personality= switch for
5589 setting the kernel personality for the container. This is
5590 useful when running a 32-bit container on a 64-bit host. A
5591 similar option Personality= is now also available for service
5594 * logind will now also track a "Desktop" identifier for each
5595 session which encodes the desktop environment of it. This is
5596 useful for desktop environments that want to identify
5597 multiple running sessions of itself easily.
5599 * A new SELinuxContext= setting for service units has been
5600 added that allows setting a specific SELinux execution
5601 context for a service.
5603 * Most systemd client tools will now honour $SYSTEMD_LESS for
5604 settings of the "less" pager. By default, these tools will
5605 override $LESS to allow certain operations to work, such as
5606 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5607 influence this logic.
5609 * systemd's "seccomp" hook-up has been changed to make use of
5610 the libseccomp library instead of using its own
5611 implementation. This has benefits for portability among
5614 * For usage together with SystemCallFilter=, a new
5615 SystemCallErrorNumber= setting has been introduced that
5616 allows configuration of a system error number to be returned
5617 on filtered system calls, instead of immediately killing the
5618 process. Also, SystemCallArchitectures= has been added to
5619 limit access to system calls of a particular architecture
5620 (in order to turn off support for unused secondary
5621 architectures). There is also a global
5622 SystemCallArchitectures= setting in system.conf now to turn
5623 off support for non-native system calls system-wide.
5625 * systemd requires a kernel with a working name_to_handle_at(),
5626 please see the kernel config requirements in the README file.
5628 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5629 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5630 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5631 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5632 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5633 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5634 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5635 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5636 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5637 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5638 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5639 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5640 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5641 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5642 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5643 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5644 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5645 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5646 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5647 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5648 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5649 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5650 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5651 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5653 — Berlin, 2014-02-20
5657 * logind has gained support for facilitating privileged input
5658 and drm device access for unprivileged clients. This work is
5659 useful to allow Wayland display servers (and similar
5660 programs, such as kmscon) to run under the user's ID and
5661 access input and drm devices which are normally
5662 protected. When this is used (and the kernel is new enough)
5663 logind will "mute" IO on the file descriptors passed to
5664 Wayland as long as it is in the background and "unmute" it
5665 if it returns into the foreground. This allows secure
5666 session switching without allowing background sessions to
5667 eavesdrop on input and display data. This also introduces
5668 session switching support if VT support is turned off in the
5669 kernel, and on seats that are not seat0.
5671 * A new kernel command line option luks.options= is understood
5672 now which allows specifying LUKS options for usage for LUKS
5673 encrypted partitions specified with luks.uuid=.
5675 * tmpfiles.d(5) snippets may now use specifier expansion in
5676 path names. More specifically %m, %b, %H, %v, are now
5677 replaced by the local machine id, boot id, hostname, and
5678 kernel version number.
5680 * A new tmpfiles.d(5) command "m" has been introduced which
5681 may be used to change the owner/group/access mode of a file
5682 or directory if it exists, but do nothing if it does not.
5684 * This release removes high-level support for the
5685 MemorySoftLimit= cgroup setting. The underlying kernel
5686 cgroup attribute memory.soft_limit= is currently badly
5687 designed and likely to be removed from the kernel API in its
5688 current form, hence we should not expose it for now.
5690 * The memory.use_hierarchy cgroup attribute is now enabled for
5691 all cgroups systemd creates in the memory cgroup
5692 hierarchy. This option is likely to be come the built-in
5693 default in the kernel anyway, and the non-hierarchical mode
5694 never made much sense in the intrinsically hierarchical
5697 * A new field _SYSTEMD_SLICE= is logged along with all journal
5698 messages containing the slice a message was generated
5699 from. This is useful to allow easy per-customer filtering of
5700 logs among other things.
5702 * systemd-journald will no longer adjust the group of journal
5703 files it creates to the "systemd-journal" group. Instead we
5704 rely on the journal directory to be owned by the
5705 "systemd-journal" group, and its setgid bit set, so that the
5706 kernel file system layer will automatically enforce that
5707 journal files inherit this group assignment. The reason for
5708 this change is that we cannot allow NSS look-ups from
5709 journald which would be necessary to resolve
5710 "systemd-journal" to a numeric GID, because this might
5711 create deadlocks if NSS involves synchronous queries to
5712 other daemons (such as nscd, or sssd) which in turn are
5713 logging clients of journald and might block on it, which
5714 would then dead lock. A tmpfiles.d(5) snippet included in
5715 systemd will make sure the setgid bit and group are
5716 properly set on the journal directory if it exists on every
5717 boot. However, we recommend adjusting it manually after
5718 upgrades too (or from RPM scriptlets), so that the change is
5719 not delayed until next reboot.
5721 * Backlight and random seed files in /var/lib/ have moved into
5722 the /var/lib/systemd/ directory, in order to centralize all
5723 systemd generated files in one directory.
5725 * Boot time performance measurements (as displayed by
5726 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5727 performance information if that's available to determine how
5728 much time BIOS and boot loader initialization required. With
5729 a sufficiently new BIOS you hence no longer need to boot
5730 with Gummiboot to get access to such information.
5732 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5733 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5734 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5735 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5736 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5737 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5738 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5740 — Berlin, 2013-10-02
5744 * The Restart= option for services now understands a new
5745 on-watchdog setting, which will restart the service
5746 automatically if the service stops sending out watchdog keep
5747 alive messages (as configured with WatchdogSec=).
5749 * The getty generator (which is responsible for bringing up a
5750 getty on configured serial consoles) will no longer only
5751 start a getty on the primary kernel console but on all
5752 others, too. This makes the order in which console= is
5753 specified on the kernel command line less important.
5755 * libsystemd-logind gained a new sd_session_get_vt() call to
5756 retrieve the VT number of a session.
5758 * If the option "tries=0" is set for an entry of /etc/crypttab
5759 its passphrase is queried indefinitely instead of any
5760 maximum number of tries.
5762 * If a service with a configure PID file terminates its PID
5763 file will now be removed automatically if it still exists
5764 afterwards. This should put an end to stale PID files.
5766 * systemd-run will now also take relative binary path names
5767 for execution and no longer insists on absolute paths.
5769 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5770 paths that are optionally prefixed with "-" to indicate that
5771 it should not be considered a failure if they do not exist.
5773 * journalctl -o (and similar commands) now understands a new
5774 output mode "short-precise", it is similar to "short" but
5775 shows timestamps with usec accuracy.
5777 * The option "discard" (as known from Debian) is now
5778 synonymous to "allow-discards" in /etc/crypttab. In fact,
5779 "discard" is preferred now (since it is easier to remember
5782 * Some licensing clean-ups were made, so that more code is now
5783 LGPL-2.1 licensed than before.
5785 * A minimal tool to save/restore the display backlight
5786 brightness across reboots has been added. It will store the
5787 backlight setting as late as possible at shutdown, and
5788 restore it as early as possible during reboot.
5790 * A logic to automatically discover and enable home and swap
5791 partitions on GPT disks has been added. With this in place
5792 /etc/fstab becomes optional for many setups as systemd can
5793 discover certain partitions located on the root disk
5794 automatically. Home partitions are recognized under their
5795 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5796 partitions are recognized under their GPT type ID
5797 0657fd6da4ab43c484e50933c84b4f4f.
5799 * systemd will no longer pass any environment from the kernel
5800 or initrd to system services. If you want to set an
5801 environment for all services, do so via the kernel command
5802 line systemd.setenv= assignment.
5804 * The systemd-sysctl tool no longer natively reads the file
5805 /etc/sysctl.conf. If desired, the file should be symlinked
5806 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5807 legacy support by a symlink rather than built-in code, it
5808 also makes the otherwise hidden order of application of the
5809 different files visible. (Note that this partly reverts to a
5810 pre-198 application order of sysctl knobs!)
5812 * The "systemctl set-log-level" and "systemctl dump" commands
5813 have been moved to systemd-analyze.
5815 * systemd-run learned the new --remain-after-exit switch,
5816 which causes the scope unit not to be cleaned up
5817 automatically after the process terminated.
5819 * tmpfiles learned a new --exclude-prefix= switch to exclude
5820 certain paths from operation.
5822 * journald will now automatically flush all messages to disk
5823 as soon as a message at the log level CRIT, ALERT or EMERG
5826 Contributions from: Andrew Cook, Brandon Philips, Christian
5827 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5828 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5829 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5830 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5831 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5832 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5833 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5834 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5835 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5836 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5837 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5838 William Giokas, Zbigniew Jędrzejewski-Szmek
5840 — Berlin, 2013-09-13
5844 * The documentation has been updated to cover the various new
5845 concepts introduced with 205.
5847 * Unit files now understand the new %v specifier which
5848 resolves to the kernel version string as returned by "uname
5851 * systemctl now supports filtering the unit list output by
5852 load state, active state and sub state, using the new
5855 * "systemctl status" will now show the results of the
5856 condition checks (like ConditionPathExists= and similar) of
5857 the last start attempts of the unit. They are also logged to
5860 * "journalctl -b" may now be used to look for boot output of a
5861 specific boot. Try "journalctl -b -1" for the previous boot,
5862 but the syntax is substantially more powerful.
5864 * "journalctl --show-cursor" has been added which prints the
5865 cursor string the last shown log line. This may then be used
5866 with the new "journalctl --after-cursor=" switch to continue
5867 browsing logs from that point on.
5869 * "journalctl --force" may now be used to force regeneration
5872 * Creation of "dead" device nodes has been moved from udev
5873 into kmod and tmpfiles. Previously, udev would read the kmod
5874 databases to pre-generate dead device nodes based on meta
5875 information contained in kernel modules, so that these would
5876 be auto-loaded on access rather then at boot. As this
5877 does not really have much to do with the exposing actual
5878 kernel devices to userspace this has always been slightly
5879 alien in the udev codebase. Following the new scheme kmod
5880 will now generate a runtime snippet for tmpfiles from the
5881 module meta information and it now is tmpfiles' job to the
5882 create the nodes. This also allows overriding access and
5883 other parameters for the nodes using the usual tmpfiles
5884 facilities. As side effect this allows us to remove the
5885 CAP_SYS_MKNOD capability bit from udevd entirely.
5887 * logind's device ACLs may now be applied to these "dead"
5888 devices nodes too, thus finally allowing managed access to
5889 devices such as /dev/snd/sequencer without loading the
5890 backing module right-away.
5892 * A new RPM macro has been added that may be used to apply
5893 tmpfiles configuration during package installation.
5895 * systemd-detect-virt and ConditionVirtualization= now can
5896 detect User-Mode-Linux machines (UML).
5898 * journald will now implicitly log the effective capabilities
5899 set of processes in the message metadata.
5901 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5903 * The initrd interface has been simplified (more specifically,
5904 support for passing performance data via environment
5905 variables and fsck results via files in /run has been
5906 removed). These features were non-essential, and are
5907 nowadays available in a much nicer way by having systemd in
5908 the initrd serialize its state and have the hosts systemd
5909 deserialize it again.
5911 * The udev "keymap" data files and tools to apply keyboard
5912 specific mappings of scan to key codes, and force-release
5913 scan code lists have been entirely replaced by a udev
5914 "keyboard" builtin and a hwdb data file.
5916 * systemd will now honour the kernel's "quiet" command line
5917 argument also during late shutdown, resulting in a
5918 completely silent shutdown when used.
5920 * There's now an option to control the SO_REUSEPORT socket
5921 option in .socket units.
5923 * Instance units will now automatically get a per-template
5924 subslice of system.slice unless something else is explicitly
5925 configured. For example, instances of sshd@.service will now
5926 implicitly be placed in system-sshd.slice rather than
5927 system.slice as before.
5929 * Test coverage support may now be enabled at build time.
5931 Contributions from: Dave Reisner, Frederic Crozat, Harald
5932 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5933 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5934 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5935 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5936 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5937 Giokas, Zbigniew Jędrzejewski-Szmek
5939 — Berlin, 2013-07-23
5943 * Two new unit types have been introduced:
5945 Scope units are very similar to service units, however, are
5946 created out of pre-existing processes — instead of PID 1
5947 forking off the processes. By using scope units it is
5948 possible for system services and applications to group their
5949 own child processes (worker processes) in a powerful way
5950 which then maybe used to organize them, or kill them
5951 together, or apply resource limits on them.
5953 Slice units may be used to partition system resources in an
5954 hierarchical fashion and then assign other units to them. By
5955 default there are now three slices: system.slice (for all
5956 system services), user.slice (for all user sessions),
5957 machine.slice (for VMs and containers).
5959 Slices and scopes have been introduced primarily in
5960 context of the work to move cgroup handling to a
5961 single-writer scheme, where only PID 1
5962 creates/removes/manages cgroups.
5964 * There's a new concept of "transient" units. In contrast to
5965 normal units these units are created via an API at runtime,
5966 not from configuration from disk. More specifically this
5967 means it is now possible to run arbitrary programs as
5968 independent services, with all execution parameters passed
5969 in via bus APIs rather than read from disk. Transient units
5970 make systemd substantially more dynamic then it ever was,
5971 and useful as a general batch manager.
5973 * logind has been updated to make use of scope and slice units
5974 for managing user sessions. As a user logs in he will get
5975 his own private slice unit, to which all sessions are added
5976 as scope units. We also added support for automatically
5977 adding an instance of user@.service for the user into the
5978 slice. Effectively logind will no longer create cgroup
5979 hierarchies on its own now, it will defer entirely to PID 1
5980 for this by means of scope, service and slice units. Since
5981 user sessions this way become entities managed by PID 1
5982 the output of "systemctl" is now a lot more comprehensive.
5984 * A new mini-daemon "systemd-machined" has been added which
5985 may be used by virtualization managers to register local
5986 VMs/containers. nspawn has been updated accordingly, and
5987 libvirt will be updated shortly. machined will collect a bit
5988 of meta information about the VMs/containers, and assign
5989 them their own scope unit (see above). The collected
5990 meta-data is then made available via the "machinectl" tool,
5991 and exposed in "ps" and similar tools. machined/machinectl
5992 is compile-time optional.
5994 * As discussed earlier, the low-level cgroup configuration
5995 options ControlGroup=, ControlGroupModify=,
5996 ControlGroupPersistent=, ControlGroupAttribute= have been
5997 removed. Please use high-level attribute settings instead as
5998 well as slice units.
6000 * A new bus call SetUnitProperties() has been added to alter
6001 various runtime parameters of a unit. This is primarily
6002 useful to alter cgroup parameters dynamically in a nice way,
6003 but will be extended later on to make more properties
6004 modifiable at runtime. systemctl gained a new set-properties
6005 command that wraps this call.
6007 * A new tool "systemd-run" has been added which can be used to
6008 run arbitrary command lines as transient services or scopes,
6009 while configuring a number of settings via the command
6010 line. This tool is currently very basic, however already
6011 very useful. We plan to extend this tool to even allow
6012 queuing of execution jobs with time triggers from the
6013 command line, similar in fashion to "at".
6015 * nspawn will now inform the user explicitly that kernels with
6016 audit enabled break containers, and suggest the user to turn
6019 * Support for detecting the IMA and AppArmor security
6020 frameworks with ConditionSecurity= has been added.
6022 * journalctl gained a new "-k" switch for showing only kernel
6023 messages, mimicking dmesg output; in addition to "--user"
6024 and "--system" switches for showing only user's own logs
6027 * systemd-delta can now show information about drop-in
6028 snippets extending unit files.
6030 * libsystemd-bus has been substantially updated but is still
6031 not available as public API.
6033 * systemd will now look for the "debug" argument on the kernel
6034 command line and enable debug logging, similar to what
6035 "systemd.log_level=debug" already did before.
6037 * "systemctl set-default", "systemctl get-default" has been
6038 added to configure the default.target symlink, which
6039 controls what to boot into by default.
6041 * "systemctl set-log-level" has been added as a convenient
6042 way to raise and lower systemd logging threshold.
6044 * "systemd-analyze plot" will now show the time the various
6045 generators needed for execution, as well as information
6046 about the unit file loading.
6048 * libsystemd-journal gained a new sd_journal_open_files() call
6049 for opening specific journal files. journactl also gained a
6050 new switch to expose this new functionality. Previously we
6051 only supported opening all files from a directory, or all
6052 files from the system, as opening individual files only is
6053 racy due to journal file rotation.
6055 * systemd gained the new DefaultEnvironment= setting in
6056 /etc/systemd/system.conf to set environment variables for
6059 * If a privileged process logs a journal message with the
6060 OBJECT_PID= field set, then journald will automatically
6061 augment this with additional OBJECT_UID=, OBJECT_GID=,
6062 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6063 system services want to log events about specific client
6064 processes. journactl/systemctl has been updated to make use
6065 of this information if all log messages regarding a specific
6068 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6069 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6070 Reisner, David Coppa, David King, David Strauss, Eelco
6071 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6072 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6073 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6074 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6075 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6076 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6077 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6078 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6079 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6080 Łukasz Stelmach, 장동준
6084 * The Python bindings gained some minimal support for the APIs
6085 exposed by libsystemd-logind.
6087 * ConditionSecurity= gained support for detecting SMACK. Since
6088 this condition already supports SELinux and AppArmor we only
6089 miss IMA for this. Patches welcome!
6091 Contributions from: Karol Lewandowski, Lennart Poettering,
6092 Zbigniew Jędrzejewski-Szmek
6096 * systemd-nspawn will now create /etc/resolv.conf if
6097 necessary, before bind-mounting the host's file onto it.
6099 * systemd-nspawn will now store meta information about a
6100 container on the container's cgroup as extended attribute
6101 fields, including the root directory.
6103 * The cgroup hierarchy has been reworked in many ways. All
6104 objects any of the components systemd creates in the cgroup
6105 tree are now suffixed. More specifically, user sessions are
6106 now placed in cgroups suffixed with ".session", users in
6107 cgroups suffixed with ".user", and nspawn containers in
6108 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6109 names are now escaped in a simple scheme to avoid collision
6110 of userspace object names with kernel filenames. This work
6111 is preparation for making these objects relocatable in the
6112 cgroup tree, in order to allow easy resource partitioning of
6113 these objects without causing naming conflicts.
6115 * systemctl list-dependencies gained the new switches
6116 --plain, --reverse, --after and --before.
6118 * systemd-inhibit now shows the process name of processes that
6119 have taken an inhibitor lock.
6121 * nss-myhostname will now also resolve "localhost"
6122 implicitly. This makes /etc/hosts an optional file and
6123 nicely handles that on IPv6 ::1 maps to both "localhost" and
6126 * libsystemd-logind.so gained a new call
6127 sd_get_machine_names() to enumerate running containers and
6128 VMs (currently only supported by very new libvirt and
6129 nspawn). sd_login_monitor can now be used to watch
6130 VMs/containers coming and going.
6132 * .include is not allowed recursively anymore, and only in
6133 unit files. Usually it is better to use drop-in snippets in
6134 .d/*.conf anyway, as introduced with systemd 198.
6136 * systemd-analyze gained a new "critical-chain" command that
6137 determines the slowest chain of units run during system
6138 boot-up. It is very useful for tracking down where
6139 optimizing boot time is the most beneficial.
6141 * systemd will no longer allow manipulating service paths in
6142 the name=systemd:/system cgroup tree using ControlGroup= in
6143 units. (But is still fine with it in all other dirs.)
6145 * There's a new systemd-nspawn@.service service file that may
6146 be used to easily run nspawn containers as system
6147 services. With the container's root directory in
6148 /var/lib/container/foobar it is now sufficient to run
6149 "systemctl start systemd-nspawn@foobar.service" to boot it.
6151 * systemd-cgls gained a new parameter "--machine" to list only
6152 the processes within a certain container.
6154 * ConditionSecurity= now can check for "apparmor". We still
6155 are lacking checks for SMACK and IMA for this condition
6156 check though. Patches welcome!
6158 * A new configuration file /etc/systemd/sleep.conf has been
6159 added that may be used to configure which kernel operation
6160 systemd is supposed to execute when "suspend", "hibernate"
6161 or "hybrid-sleep" is requested. This makes the new kernel
6162 "freeze" state accessible to the user.
6164 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6165 the passed argument if applicable.
6167 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6168 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6169 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6170 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6171 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6172 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6173 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6178 * The output of 'systemctl list-jobs' got some polishing. The
6179 '--type=' argument may now be passed more than once. A new
6180 command 'systemctl list-sockets' has been added which shows
6181 a list of kernel sockets systemd is listening on with the
6182 socket units they belong to, plus the units these socket
6185 * The experimental libsystemd-bus library got substantial
6186 updates to work in conjunction with the (also experimental)
6187 kdbus kernel project. It works well enough to exchange
6188 messages with some sophistication. Note that kdbus is not
6189 ready yet, and the library is mostly an elaborate test case
6190 for now, and not installable.
6192 * systemd gained a new unit 'systemd-static-nodes.service'
6193 that generates static device nodes earlier during boot, and
6194 can run in conjunction with udev.
6196 * libsystemd-login gained a new call sd_pid_get_user_unit()
6197 to retrieve the user systemd unit a process is running
6198 in. This is useful for systems where systemd is used as
6201 * systemd-nspawn now places all containers in the new /machine
6202 top-level cgroup directory in the name=systemd
6203 hierarchy. libvirt will soon do the same, so that we get a
6204 uniform separation of /system, /user and /machine for system
6205 services, user processes and containers/virtual
6206 machines. This new cgroup hierarchy is also useful to stick
6207 stable names to specific container instances, which can be
6208 recognized later this way (this name may be controlled
6209 via systemd-nspawn's new -M switch). libsystemd-login also
6210 gained a new call sd_pid_get_machine_name() to retrieve the
6211 name of the container/VM a specific process belongs to.
6213 * bootchart can now store its data in the journal.
6215 * libsystemd-journal gained a new call
6216 sd_journal_add_conjunction() for AND expressions to the
6217 matching logic. This can be used to express more complex
6218 logical expressions.
6220 * journactl can now take multiple --unit= and --user-unit=
6223 * The cryptsetup logic now understands the "luks.key=" kernel
6224 command line switch for specifying a file to read the
6225 decryption key from. Also, if a configured key file is not
6226 found the tool will now automatically fall back to prompting
6229 * Python systemd.journal module was updated to wrap recently
6230 added functions from libsystemd-journal. The interface was
6231 changed to bring the low level interface in s.j._Reader
6232 closer to the C API, and the high level interface in
6233 s.j.Reader was updated to wrap and convert all data about
6236 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6237 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6238 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6239 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6240 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6241 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6245 * journalctl --update-catalog now understands a new --root=
6246 option to operate on catalogs found in a different root
6249 * During shutdown after systemd has terminated all running
6250 services a final killing loop kills all remaining left-over
6251 processes. We will now print the name of these processes
6252 when we send SIGKILL to them, since this usually indicates a
6255 * If /etc/crypttab refers to password files stored on
6256 configured mount points automatic dependencies will now be
6257 generated to ensure the specific mount is established first
6258 before the key file is attempted to be read.
6260 * 'systemctl status' will now show information about the
6261 network sockets a socket unit is listening on.
6263 * 'systemctl status' will also shown information about any
6264 drop-in configuration file for units. (Drop-In configuration
6265 files in this context are files such as
6266 /etc/systemd/systemd/foobar.service.d/*.conf)
6268 * systemd-cgtop now optionally shows summed up CPU times of
6269 cgroups. Press '%' while running cgtop to switch between
6270 percentage and absolute mode. This is useful to determine
6271 which cgroups use up the most CPU time over the entire
6272 runtime of the system. systemd-cgtop has also been updated
6273 to be 'pipeable' for processing with further shell tools.
6275 * 'hostnamectl set-hostname' will now allow setting of FQDN
6278 * The formatting and parsing of time span values has been
6279 changed. The parser now understands fractional expressions
6280 such as "5.5h". The formatter will now output fractional
6281 expressions for all time spans under 1min, i.e. "5.123456s"
6282 rather than "5s 123ms 456us". For time spans under 1s
6283 millisecond values are shown, for those under 1ms
6284 microsecond values are shown. This should greatly improve
6285 all time-related output of systemd.
6287 * libsystemd-login and libsystemd-journal gained new
6288 functions for querying the poll() events mask and poll()
6289 timeout value for integration into arbitrary event
6292 * localectl gained the ability to list available X11 keymaps
6293 (models, layouts, variants, options).
6295 * 'systemd-analyze dot' gained the ability to filter for
6296 specific units via shell-style globs, to create smaller,
6297 more useful graphs. I.e. it is now possible to create simple
6298 graphs of all the dependencies between only target units, or
6299 of all units that Avahi has dependencies with.
6301 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6302 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6303 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6304 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6305 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6306 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6307 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6311 * The boot-time readahead implementation for rotating media
6312 will now read the read-ahead data in multiple passes which
6313 consist of all read requests made in equidistant time
6314 intervals. This means instead of strictly reading read-ahead
6315 data in its physical order on disk we now try to find a
6316 middle ground between physical and access time order.
6318 * /etc/os-release files gained a new BUILD_ID= field for usage
6319 on operating systems that provide continuous builds of OS
6322 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6323 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6324 William Douglas, Zbigniew Jędrzejewski-Szmek
6328 * systemd-python gained an API exposing libsystemd-daemon.
6330 * The SMACK setup logic gained support for uploading CIPSO
6333 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6334 ReadOnlyDirectories= and InaccessibleDirectories= has
6335 changed. The private /tmp and /var/tmp directories are now
6336 shared by all processes of a service (which means
6337 ExecStartPre= may now leave data in /tmp that ExecStart= of
6338 the same service can still access). When a service is
6339 stopped its temporary directories are immediately deleted
6340 (normal clean-up with tmpfiles is still done in addition to
6343 * By default, systemd will now set a couple of sysctl
6344 variables in the kernel: the safe sysrq options are turned
6345 on, IP route verification is turned on, and source routing
6346 disabled. The recently added hardlink and softlink
6347 protection of the kernel is turned on. These settings should
6348 be reasonably safe, and good defaults for all new systems.
6350 * The predictable network naming logic may now be turned off
6351 with a new kernel command line switch: net.ifnames=0.
6353 * A new libsystemd-bus module has been added that implements a
6354 pretty complete D-Bus client library. For details see:
6356 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6358 * journald will now explicitly flush the journal files to disk
6359 at the latest 5min after each write. The file will then also
6360 be marked offline until the next write. This should increase
6361 reliability in case of a crash. The synchronization delay
6362 can be configured via SyncIntervalSec= in journald.conf.
6364 * There's a new remote-fs-setup.target unit that can be used
6365 to pull in specific services when at least one remote file
6366 system is to be mounted.
6368 * There are new targets timers.target and paths.target as
6369 canonical targets to pull user timer and path units in
6370 from. This complements sockets.target with a similar
6371 purpose for socket units.
6373 * libudev gained a new call udev_device_set_attribute_value()
6374 to set sysfs attributes of a device.
6376 * The udev daemon now sets the default number of worker
6377 processes executed in parallel based on the number of available
6378 CPUs instead of the amount of available RAM. This is supposed
6379 to provide a more reliable default and limit a too aggressive
6380 parallelism for setups with 1000s of devices connected.
6382 Contributions from: Auke Kok, Colin Walters, Cristian
6383 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6384 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6385 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6386 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6387 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6388 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6389 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6390 Zbigniew Jędrzejewski-Szmek
6394 * Configuration of unit files may now be extended via drop-in
6395 files without having to edit/override the unit files
6396 themselves. More specifically, if the administrator wants to
6397 change one value for a service file foobar.service he can
6398 now do so by dropping in a configuration snippet into
6399 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6400 will load all these snippets and apply them on top of the
6401 main unit configuration file, possibly extending or
6402 overriding its settings. Using these drop-in snippets is
6403 generally nicer than the two earlier options for changing
6404 unit files locally: copying the files from
6405 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6406 them there; or creating a new file in /etc/systemd/system/
6407 that incorporates the original one via ".include". Drop-in
6408 snippets into these .d/ directories can be placed in any
6409 directory systemd looks for units in, and the usual
6410 overriding semantics between /usr/lib, /etc and /run apply
6413 * Most unit file settings which take lists of items can now be
6414 reset by assigning the empty string to them. For example,
6415 normally, settings such as Environment=FOO=BAR append a new
6416 environment variable assignment to the environment block,
6417 each time they are used. By assigning Environment= the empty
6418 string the environment block can be reset to empty. This is
6419 particularly useful with the .d/*.conf drop-in snippets
6420 mentioned above, since this adds the ability to reset list
6421 settings from vendor unit files via these drop-ins.
6423 * systemctl gained a new "list-dependencies" command for
6424 listing the dependencies of a unit recursively.
6426 * Inhibitors are now honored and listed by "systemctl
6427 suspend", "systemctl poweroff" (and similar) too, not only
6428 GNOME. These commands will also list active sessions by
6431 * Resource limits (as exposed by the various control group
6432 controllers) can now be controlled dynamically at runtime
6433 for all units. More specifically, you can now use a command
6434 like "systemctl set-cgroup-attr foobar.service cpu.shares
6435 2000" to alter the CPU shares a specific service gets. These
6436 settings are stored persistently on disk, and thus allow the
6437 administrator to easily adjust the resource usage of
6438 services with a few simple commands. This dynamic resource
6439 management logic is also available to other programs via the
6440 bus. Almost any kernel cgroup attribute and controller is
6443 * systemd-vconsole-setup will now copy all font settings to
6444 all allocated VTs, where it previously applied them only to
6447 * libsystemd-login gained the new sd_session_get_tty() API
6450 * This release drops support for a few legacy or
6451 distribution-specific LSB facility names when parsing init
6452 scripts: $x-display-manager, $mail-transfer-agent,
6453 $mail-transport-agent, $mail-transfer-agent, $smtp,
6454 $null. Also, the mail-transfer-agent.target unit backing
6455 this has been removed. Distributions which want to retain
6456 compatibility with this should carry the burden for
6457 supporting this themselves and patch support for these back
6458 in, if they really need to. Also, the facilities $syslog and
6459 $local_fs are now ignored, since systemd does not support
6460 early-boot LSB init scripts anymore, and these facilities
6461 are implied anyway for normal services. syslog.target has
6464 * There are new bus calls on PID1's Manager object for
6465 cancelling jobs, and removing snapshot units. Previously,
6466 both calls were only available on the Job and Snapshot
6469 * systemd-journal-gatewayd gained SSL support.
6471 * The various "environment" files, such as /etc/locale.conf
6472 now support continuation lines with a backslash ("\") as
6473 last character in the line, similarly in style (but different)
6474 to how this is supported in shells.
6476 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6477 now implicitly appended to every log entry logged. systemctl
6478 has been updated to filter by this field when operating on a
6479 user systemd instance.
6481 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6482 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6483 the container. This makes it easier to boot unmodified
6484 Fedora systems in a container, which however still requires
6485 audit=0 to be passed on the kernel command line. Auditing in
6486 kernel and userspace is unfortunately still too broken in
6487 context of containers, hence we recommend compiling it out
6488 of the kernel or using audit=0. Hopefully this will be fixed
6489 one day for good in the kernel.
6491 * nspawn gained the new --bind= and --bind-ro= parameters to
6492 bind mount specific directories from the host into the
6495 * nspawn will now mount its own devpts file system instance
6496 into the container, in order not to leak pty devices from
6497 the host into the container.
6499 * systemd will now read the firmware boot time performance
6500 information from the EFI variables, if the used boot loader
6501 supports this, and takes it into account for boot performance
6502 analysis via "systemd-analyze". This is currently supported
6503 only in conjunction with Gummiboot, but could be supported
6504 by other boot loaders too. For details see:
6506 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6508 * A new generator has been added that automatically mounts the
6509 EFI System Partition (ESP) to /boot, if that directory
6510 exists, is empty, and no other file system has been
6511 configured to be mounted there.
6513 * logind will now send out PrepareForSleep(false) out
6514 unconditionally, after coming back from suspend. This may be
6515 used by applications as asynchronous notification for
6516 system resume events.
6518 * "systemctl unlock-sessions" has been added, that allows
6519 unlocking the screens of all user sessions at once, similar
6520 to how "systemctl lock-sessions" already locked all users
6521 sessions. This is backed by a new D-Bus call UnlockSessions().
6523 * "loginctl seat-status" will now show the master device of a
6524 seat. (i.e. the device of a seat that needs to be around for
6525 the seat to be considered available, usually the graphics
6528 * tmpfiles gained a new "X" line type, that allows
6529 configuration of files and directories (with wildcards) that
6530 shall be excluded from automatic cleanup ("aging").
6532 * udev default rules set the device node permissions now only
6533 at "add" events, and do not change them any longer with a
6534 later "change" event.
6536 * The log messages for lid events and power/sleep keypresses
6537 now carry a message ID.
6539 * We now have a substantially larger unit test suite, but this
6540 continues to be work in progress.
6542 * udevadm hwdb gained a new --root= parameter to change the
6543 root directory to operate relative to.
6545 * logind will now issue a background sync() request to the kernel
6546 early at shutdown, so that dirty buffers are flushed to disk early
6547 instead of at the last moment, in order to optimize shutdown
6550 * A new bootctl tool has been added that is an interface for
6551 certain boot loader operations. This is currently a preview
6552 and is likely to be extended into a small mechanism daemon
6553 like timedated, localed, hostnamed, and can be used by
6554 graphical UIs to enumerate available boot options, and
6555 request boot into firmware operations.
6557 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6558 the rest of the package. It also has been updated to work
6559 correctly in initrds.
6561 * Policykit previously has been runtime optional, and is now
6562 also compile time optional via a configure switch.
6564 * systemd-analyze has been reimplemented in C. Also "systemctl
6565 dot" has moved into systemd-analyze.
6567 * "systemctl status" with no further parameters will now print
6568 the status of all active or failed units.
6570 * Operations such as "systemctl start" can now be executed
6571 with a new mode "--irreversible" which may be used to queue
6572 operations that cannot accidentally be reversed by a later
6573 job queuing. This is by default used to make shutdown
6574 requests more robust.
6576 * The Python API of systemd now gained a new module for
6577 reading journal files.
6579 * A new tool kernel-install has been added that can install
6580 kernel images according to the Boot Loader Specification:
6582 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6584 * Boot time console output has been improved to provide
6585 animated boot time output for hanging jobs.
6587 * A new tool systemd-activate has been added which can be used
6588 to test socket activation with, directly from the command
6589 line. This should make it much easier to test and debug
6590 socket activation in daemons.
6592 * journalctl gained a new "--reverse" (or -r) option to show
6593 journal output in reverse order (i.e. newest line first).
6595 * journalctl gained a new "--pager-end" (or -e) option to jump
6596 to immediately jump to the end of the journal in the
6597 pager. This is only supported in conjunction with "less".
6599 * journalctl gained a new "--user-unit=" option, that works
6600 similarly to "--unit=" but filters for user units rather than
6603 * A number of unit files to ease adoption of systemd in
6604 initrds has been added. This moves some minimal logic from
6605 the various initrd implementations into systemd proper.
6607 * The journal files are now owned by a new group
6608 "systemd-journal", which exists specifically to allow access
6609 to the journal, and nothing else. Previously, we used the
6610 "adm" group for that, which however possibly covers more
6611 than just journal/log file access. This new group is now
6612 already used by systemd-journal-gatewayd to ensure this
6613 daemon gets access to the journal files and as little else
6614 as possible. Note that "make install" will also set FS ACLs
6615 up for /var/log/journal to give "adm" and "wheel" read
6616 access to it, in addition to "systemd-journal" which owns
6617 the journal files. We recommend that packaging scripts also
6618 add read access to "adm" + "wheel" to /var/log/journal, and
6619 all existing/future journal files. To normal users and
6620 administrators little changes, however packagers need to
6621 ensure to create the "systemd-journal" system group at
6622 package installation time.
6624 * The systemd-journal-gatewayd now runs as unprivileged user
6625 systemd-journal-gateway:systemd-journal-gateway. Packaging
6626 scripts need to create these system user/group at
6629 * timedated now exposes a new boolean property CanNTP that
6630 indicates whether a local NTP service is available or not.
6632 * systemd-detect-virt will now also detect xen PVs
6634 * The pstore file system is now mounted by default, if it is
6637 * In addition to the SELinux and IMA policies we will now also
6638 load SMACK policies at early boot.
6640 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6641 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6642 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6643 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6644 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6645 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6646 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6647 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6648 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6649 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6650 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6651 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6652 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6653 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6657 * Timer units now support calendar time events in addition to
6658 monotonic time events. That means you can now trigger a unit
6659 based on a calendar time specification such as "Thu,Fri
6660 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6661 or fifth day of any month of the year 2013, given that it is
6662 a thursday or friday. This brings timer event support
6663 considerably closer to cron's capabilities. For details on
6664 the supported calendar time specification language see
6667 * udev now supports a number of different naming policies for
6668 network interfaces for predictable names, and a combination
6669 of these policies is now the default. Please see this wiki
6670 document for details:
6672 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6674 * Auke Kok's bootchart implementation has been added to the
6675 systemd tree. It is an optional component that can graph the
6676 boot in quite some detail. It is one of the best bootchart
6677 implementations around and minimal in its code and
6680 * nss-myhostname has been integrated into the systemd source
6681 tree. nss-myhostname guarantees that the local hostname
6682 always stays resolvable via NSS. It has been a weak
6683 requirement of systemd-hostnamed since a long time, and
6684 since its code is actually trivial we decided to just
6685 include it in systemd's source tree. It can be turned off
6686 with a configure switch.
6688 * The read-ahead logic is now capable of properly detecting
6689 whether a btrfs file system is on SSD or rotating media, in
6690 order to optimize the read-ahead scheme. Previously, it was
6691 only capable of detecting this on traditional file systems
6694 * In udev, additional device properties are now read from the
6695 IAB in addition to the OUI database. Also, Bluetooth company
6696 identities are attached to the devices as well.
6698 * In service files %U may be used as specifier that is
6699 replaced by the configured user name of the service.
6701 * nspawn may now be invoked without a controlling TTY. This
6702 makes it suitable for invocation as its own service. This
6703 may be used to set up a simple containerized server system
6704 using only core OS tools.
6706 * systemd and nspawn can now accept socket file descriptors
6707 when they are started for socket activation. This enables
6708 implementation of socket activated nspawn
6709 containers. i.e. think about autospawning an entire OS image
6710 when the first SSH or HTTP connection is received. We expect
6711 that similar functionality will also be added to libvirt-lxc
6714 * journalctl will now suppress ANSI color codes when
6715 presenting log data.
6717 * systemctl will no longer show control group information for
6718 a unit if the control group is empty anyway.
6720 * logind can now automatically suspend/hibernate/shutdown the
6723 * /etc/machine-info and hostnamed now also expose the chassis
6724 type of the system. This can be used to determine whether
6725 the local system is a laptop, desktop, handset or
6726 tablet. This information may either be configured by the
6727 user/vendor or is automatically determined from ACPI and DMI
6728 information if possible.
6730 * A number of PolicyKit actions are now bound together with
6731 "imply" rules. This should simplify creating UIs because
6732 many actions will now authenticate similar ones as well.
6734 * Unit files learnt a new condition ConditionACPower= which
6735 may be used to conditionalize a unit depending on whether an
6736 AC power source is connected or not, of whether the system
6737 is running on battery power.
6739 * systemctl gained a new "is-failed" verb that may be used in
6740 shell scripts and suchlike to check whether a specific unit
6741 is in the "failed" state.
6743 * The EnvironmentFile= setting in unit files now supports file
6744 globbing, and can hence be used to easily read a number of
6745 environment files at once.
6747 * systemd will no longer detect and recognize specific
6748 distributions. All distribution-specific #ifdeffery has been
6749 removed, systemd is now fully generic and
6750 distribution-agnostic. Effectively, not too much is lost as
6751 a lot of the code is still accessible via explicit configure
6752 switches. However, support for some distribution specific
6753 legacy configuration file formats has been dropped. We
6754 recommend distributions to simply adopt the configuration
6755 files everybody else uses now and convert the old
6756 configuration from packaging scripts. Most distributions
6757 already did that. If that's not possible or desirable,
6758 distributions are welcome to forward port the specific
6759 pieces of code locally from the git history.
6761 * When logging a message about a unit systemd will now always
6762 log the unit name in the message meta data.
6764 * localectl will now also discover system locale data that is
6765 not stored in locale archives, but directly unpacked.
6767 * logind will no longer unconditionally use framebuffer
6768 devices as seat masters, i.e. as devices that are required
6769 to be existing before a seat is considered preset. Instead,
6770 it will now look for all devices that are tagged as
6771 "seat-master" in udev. By default, framebuffer devices will
6772 be marked as such, but depending on local systems, other
6773 devices might be marked as well. This may be used to
6774 integrate graphics cards using closed source drivers (such
6775 as NVidia ones) more nicely into logind. Note however, that
6776 we recommend using the open source NVidia drivers instead,
6777 and no udev rules for the closed-source drivers will be
6778 shipped from us upstream.
6780 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6781 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6782 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6783 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6784 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6785 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6786 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6787 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6788 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6789 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6790 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6791 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6796 * udev gained support for loading additional device properties
6797 from an indexed database that is keyed by vendor/product IDs
6798 and similar device identifiers. For the beginning this
6799 "hwdb" is populated with data from the well-known PCI and
6800 USB database, but also includes PNP, ACPI and OID data. In
6801 the longer run this indexed database shall grow into
6802 becoming the one central database for non-essential
6803 userspace device metadata. Previously, data from the PCI/USB
6804 database was only attached to select devices, since the
6805 lookup was a relatively expensive operation due to O(n) time
6806 complexity (with n being the number of entries in the
6807 database). Since this is now O(1), we decided to add in this
6808 data for all devices where this is available, by
6809 default. Note that the indexed database needs to be rebuilt
6810 when new data files are installed. To achieve this you need
6811 to update your packaging scripts to invoke "udevadm hwdb
6812 --update" after installation of hwdb data files. For
6813 RPM-based distributions we introduced the new
6814 %udev_hwdb_update macro for this purpose.
6816 * The Journal gained support for the "Message Catalog", an
6817 indexed database to link up additional information with
6818 journal entries. For further details please check:
6820 https://www.freedesktop.org/wiki/Software/systemd/catalog
6822 The indexed message catalog database also needs to be
6823 rebuilt after installation of message catalog files. Use
6824 "journalctl --update-catalog" for this. For RPM-based
6825 distributions we introduced the %journal_catalog_update
6826 macro for this purpose.
6828 * The Python Journal bindings gained support for the standard
6829 Python logging framework.
6831 * The Journal API gained new functions for checking whether
6832 the underlying file system of a journal file is capable of
6833 properly reporting file change notifications, or whether
6834 applications that want to reflect journal changes "live"
6835 need to recheck journal files continuously in appropriate
6838 * It is now possible to set the "age" field for tmpfiles
6839 entries to 0, indicating that files matching this entry
6840 shall always be removed when the directories are cleaned up.
6842 * coredumpctl gained a new "gdb" verb which invokes gdb
6843 right-away on the selected coredump.
6845 * There's now support for "hybrid sleep" on kernels that
6846 support this, in addition to "suspend" and "hibernate". Use
6847 "systemctl hybrid-sleep" to make use of this.
6849 * logind's HandleSuspendKey= setting (and related settings)
6850 now gained support for a new "lock" setting to simply
6851 request the screen lock on all local sessions, instead of
6852 actually executing a suspend or hibernation.
6854 * systemd will now mount the EFI variables file system by
6857 * Socket units now gained support for configuration of the
6858 SMACK security label.
6860 * timedatectl will now output the time of the last and next
6861 daylight saving change.
6863 * We dropped support for various legacy and distro-specific
6864 concepts, such as insserv, early-boot SysV services
6865 (i.e. those for non-standard runlevels such as 'b' or 'S')
6866 or ArchLinux /etc/rc.conf support. We recommend the
6867 distributions who still need support this to either continue
6868 to maintain the necessary patches downstream, or find a
6869 different solution. (Talk to us if you have questions!)
6871 * Various systemd components will now bypass PolicyKit checks
6872 for root and otherwise handle properly if PolicyKit is not
6873 found to be around. This should fix most issues for
6874 PolicyKit-less systems. Quite frankly this should have been
6875 this way since day one. It is absolutely our intention to
6876 make systemd work fine on PolicyKit-less systems, and we
6877 consider it a bug if something does not work as it should if
6878 PolicyKit is not around.
6880 * For embedded systems it is now possible to build udev and
6881 systemd without blkid and/or kmod support.
6883 * "systemctl switch-root" is now capable of switching root
6884 more than once. I.e. in addition to transitions from the
6885 initrd to the host OS it is now possible to transition to
6886 further OS images from the host. This is useful to implement
6887 offline updating tools.
6889 * Various other additions have been made to the RPM macros
6890 shipped with systemd. Use %udev_rules_update() after
6891 installing new udev rules files. %_udevhwdbdir,
6892 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6893 %_sysctldir are now available which resolve to the right
6894 directories for packages to place various data files in.
6896 * journalctl gained the new --full switch (in addition to
6897 --all, to disable ellipsation for long messages.
6899 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6900 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6901 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6902 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6903 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6904 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6905 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6906 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6907 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6911 * journalctl gained new --since= and --until= switches to
6912 filter by time. It also now supports nice filtering for
6913 units via --unit=/-u.
6915 * Type=oneshot services may use ExecReload= and do the
6918 * The journal daemon now supports time-based rotation and
6919 vacuuming, in addition to the usual disk-space based
6922 * The journal will now index the available field values for
6923 each field name. This enables clients to show pretty drop
6924 downs of available match values when filtering. The bash
6925 completion of journalctl has been updated
6926 accordingly. journalctl gained a new switch -F to list all
6927 values a certain field takes in the journal database.
6929 * More service events are now written as structured messages
6930 to the journal, and made recognizable via message IDs.
6932 * The timedated, localed and hostnamed mini-services which
6933 previously only provided support for changing time, locale
6934 and hostname settings from graphical DEs such as GNOME now
6935 also have a minimal (but very useful) text-based client
6936 utility each. This is probably the nicest way to changing
6937 these settings from the command line now, especially since
6938 it lists available options and is fully integrated with bash
6941 * There's now a new tool "systemd-coredumpctl" to list and
6942 extract coredumps from the journal.
6944 * We now install a README each in /var/log/ and
6945 /etc/rc.d/init.d explaining where the system logs and init
6946 scripts went. This hopefully should help folks who go to
6947 that dirs and look into the otherwise now empty void and
6948 scratch their heads.
6950 * When user-services are invoked (by systemd --user) the
6951 $MANAGERPID env var is set to the PID of systemd.
6953 * SIGRTMIN+24 when sent to a --user instance will now result
6954 in immediate termination of systemd.
6956 * gatewayd received numerous feature additions such as a
6957 "follow" mode, for live syncing and filtering.
6959 * browse.html now allows filtering and showing detailed
6960 information on specific entries. Keyboard navigation and
6961 mouse screen support has been added.
6963 * gatewayd/journalctl now supports HTML5/JSON
6964 Server-Sent-Events as output.
6966 * The SysV init script compatibility logic will now
6967 heuristically determine whether a script supports the
6968 "reload" verb, and only then make this available as
6971 * "systemctl status --follow" has been removed, use "journalctl
6974 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6975 have been removed since they are hardly useful to be
6978 * And I'd like to take the opportunity to specifically mention
6979 Zbigniew for his great contributions. Zbigniew, you rock!
6981 Contributions from: Andrew Eikum, Christian Hesse, Colin
6982 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6983 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6984 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6985 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6986 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6987 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6991 * If /etc/vconsole.conf is non-existent or empty we will no
6992 longer load any console font or key map at boot by
6993 default. Instead the kernel defaults will be left
6994 intact. This is definitely the right thing to do, as no
6995 configuration should mean no configuration, and hard-coding
6996 font names that are different on all archs is probably a bad
6997 idea. Also, the kernel default key map and font should be
6998 good enough for most cases anyway, and mostly identical to
6999 the userspace fonts/key maps we previously overloaded them
7000 with. If distributions want to continue to default to a
7001 non-kernel font or key map they should ship a default
7002 /etc/vconsole.conf with the appropriate contents.
7004 Contributions from: Colin Walters, Daniel J Walsh, Dave
7005 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7006 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7010 * journalctl gained a new --cursor= switch to show entries
7011 starting from the specified location in the journal.
7013 * We now enforce a size limit on journal entry fields exported
7014 with "-o json" in journalctl. Fields larger than 4K will be
7015 assigned null. This can be turned off with --all.
7017 * An (optional) journal gateway daemon is now available as
7018 "systemd-journal-gatewayd.service". This service provides
7019 access to the journal via HTTP and JSON. This functionality
7020 will be used to implement live log synchronization in both
7021 pull and push modes, but has various other users too, such
7022 as easy log access for debugging of embedded devices. Right
7023 now it is already useful to retrieve the journal via HTTP:
7025 # systemctl start systemd-journal-gatewayd.service
7026 # wget http://localhost:19531/entries
7028 This will download the journal contents in a
7029 /var/log/messages compatible format. The same as JSON:
7031 # curl -H"Accept: application/json" http://localhost:19531/entries
7033 This service is also accessible via a web browser where a
7034 single static HTML5 app is served that uses the JSON logic
7035 to enable the user to do some basic browsing of the
7036 journal. This will be extended later on. Here's an example
7037 screenshot of this app in its current state:
7039 http://0pointer.de/public/journal-gatewayd
7041 Contributions from: Kay Sievers, Lennart Poettering, Robert
7042 Milasan, Tom Gundersen
7046 * The bash completion logic is now available for journalctl
7049 * We do not mount the "cpuset" controller anymore together with
7050 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7051 started if no parameters are assigned to it. "cpuset" hence
7052 broke code that assumed it could create "cpu" groups and
7055 * journalctl -f will now subscribe to terminal size changes,
7056 and line break accordingly.
7058 Contributions from: Dave Reisner, Kay Sievers, Lennart
7059 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7063 * nspawn will now create a symlink /etc/localtime in the
7064 container environment, copying the host's timezone
7065 setting. Previously this has been done via a bind mount, but
7066 since symlinks cannot be bind mounted this has now been
7067 changed to create/update the appropriate symlink.
7069 * journalctl -n's line number argument is now optional, and
7070 will default to 10 if omitted.
7072 * journald will now log the maximum size the journal files may
7073 take up on disk. This is particularly useful if the default
7074 built-in logic of determining this parameter from the file
7075 system size is used. Use "systemctl status
7076 systemd-journald.service" to see this information.
7078 * The multi-seat X wrapper tool has been stripped down. As X
7079 is now capable of enumerating graphics devices via udev in a
7080 seat-aware way the wrapper is not strictly necessary
7081 anymore. A stripped down temporary stop-gap is still shipped
7082 until the upstream display managers have been updated to
7083 fully support the new X logic. Expect this wrapper to be
7084 removed entirely in one of the next releases.
7086 * HandleSleepKey= in logind.conf has been split up into
7087 HandleSuspendKey= and HandleHibernateKey=. The old setting
7088 is not available anymore. X11 and the kernel are
7089 distinguishing between these keys and we should too. This
7090 also means the inhibition lock for these keys has been split
7093 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7094 Poettering, Lukas Nykryn, Václav Pavlín
7098 * Whenever a unit changes state we will now log this to the
7099 journal and show along the unit's own log output in
7102 * ConditionPathIsMountPoint= can now properly detect bind
7103 mount points too. (Previously, a bind mount of one file
7104 system to another place in the same file system could not be
7105 detected as mount, since they shared struct stat's st_dev
7108 * We will now mount the cgroup controllers cpu, cpuacct,
7109 cpuset and the controllers net_cls, net_prio together by
7112 * nspawn containers will now have a virtualized boot
7113 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7114 over with a randomized ID at container initialization). This
7115 has the effect of making "journalctl -b" do the right thing
7118 * The JSON output journal serialization has been updated not
7119 to generate "endless" list objects anymore, but rather one
7120 JSON object per line. This is more in line how most JSON
7121 parsers expect JSON objects. The new output mode
7122 "json-pretty" has been added to provide similar output, but
7123 neatly aligned for readability by humans.
7125 * We dropped all explicit sync() invocations in the shutdown
7126 code. The kernel does this implicitly anyway in the kernel
7127 reboot() syscall. halt(8)'s -n option is now a compatibility
7130 * We now support virtualized reboot() in containers, as
7131 supported by newer kernels. We will fall back to exit() if
7132 CAP_SYS_REBOOT is not available to the container. Also,
7133 nspawn makes use of this now and will actually reboot the
7134 container if the containerized OS asks for that.
7136 * journalctl will only show local log output by default
7137 now. Use --merge (-m) to show remote log output, too.
7139 * libsystemd-journal gained the new sd_journal_get_usage()
7140 call to determine the current disk usage of all journal
7141 files. This is exposed in the new "journalctl --disk-usage"
7144 * journald gained a new configuration setting SplitMode= in
7145 journald.conf which may be used to control how user journals
7146 are split off. See journald.conf(5) for details.
7148 * A new condition type ConditionFileNotEmpty= has been added.
7150 * tmpfiles' "w" lines now support file globbing, to write
7151 multiple files at once.
7153 * We added Python bindings for the journal submission
7154 APIs. More Python APIs for a number of selected APIs will
7155 likely follow. Note that we intend to add native bindings
7156 only for the Python language, as we consider it common
7157 enough to deserve bindings shipped within systemd. There are
7158 various projects outside of systemd that provide bindings
7159 for languages such as PHP or Lua.
7161 * Many conditions will now resolve specifiers such as %i. In
7162 addition, PathChanged= and related directives of .path units
7163 now support specifiers as well.
7165 * There's now a new RPM macro definition for the system preset
7168 * journald will now warn if it ca not forward a message to the
7169 syslog daemon because its socket is full.
7171 * timedated will no longer write or process /etc/timezone,
7172 except on Debian. As we do not support late mounted /usr
7173 anymore /etc/localtime always being a symlink is now safe,
7174 and hence the information in /etc/timezone is not necessary
7177 * logind will now always reserve one VT for a text getty (VT6
7178 by default). Previously if more than 6 X sessions where
7179 started they took up all the VTs with auto-spawned gettys,
7180 so that no text gettys were available anymore.
7182 * udev will now automatically inform the btrfs kernel logic
7183 about btrfs RAID components showing up. This should make
7184 simple hotplug based btrfs RAID assembly work.
7186 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7187 (but not for its children which will stay at the kernel
7188 default). This should allow setups with a lot more listening
7191 * systemd will now always pass the configured timezone to the
7192 kernel at boot. timedated will do the same when the timezone
7195 * logind's inhibition logic has been updated. By default,
7196 logind will now handle the lid switch, the power and sleep
7197 keys all the time, even in graphical sessions. If DEs want
7198 to handle these events on their own they should take the new
7199 handle-power-key, handle-sleep-key and handle-lid-switch
7200 inhibitors during their runtime. A simple way to achieve
7201 that is to invoke the DE wrapped in an invocation of:
7203 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7205 * Access to unit operations is now checked via SELinux taking
7206 the unit file label and client process label into account.
7208 * systemd will now notify the administrator in the journal
7209 when he over-mounts a non-empty directory.
7211 * There are new specifiers that are resolved in unit files,
7212 for the host name (%H), the machine ID (%m) and the boot ID
7215 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7216 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7217 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7218 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7219 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7220 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7221 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7225 * Support for reading structured kernel messages from
7226 /dev/kmsg has now been added and is enabled by default.
7228 * Support for reading kernel messages from /proc/kmsg has now
7229 been removed. If you want kernel messages in the journal
7230 make sure to run a recent kernel (>= 3.5) that supports
7231 reading structured messages from /dev/kmsg (see
7232 above). /proc/kmsg is now exclusive property of classic
7233 syslog daemons again.
7235 * The libudev API gained the new
7236 udev_device_new_from_device_id() call.
7238 * The logic for file system namespace (ReadOnlyDirectory=,
7239 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7240 require pivot_root() anymore. This means fewer temporary
7241 directories are created below /tmp for this feature.
7243 * nspawn containers will now see and receive all submounts
7244 made on the host OS below the root file system of the
7247 * Forward Secure Sealing is now supported for Journal files,
7248 which provide cryptographical sealing of journal files so
7249 that attackers cannot alter log history anymore without this
7250 being detectable. Lennart will soon post a blog story about
7251 this explaining it in more detail.
7253 * There are two new service settings RestartPreventExitStatus=
7254 and SuccessExitStatus= which allow configuration of exit
7255 status (exit code or signal) which will be excepted from the
7256 restart logic, resp. consider successful.
7258 * journalctl gained the new --verify switch that can be used
7259 to check the integrity of the structure of journal files and
7260 (if Forward Secure Sealing is enabled) the contents of
7263 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7264 and similar symlinks pre-created. This makes running shells
7265 as container init process a lot more fun.
7267 * The fstab support can now handle PARTUUID= and PARTLABEL=
7270 * A new ConditionHost= condition has been added to match
7271 against the hostname (with globs) and machine ID. This is
7272 useful for clusters where a single OS image is used to
7273 provision a large number of hosts which shall run slightly
7274 different sets of services.
7276 * Services which hit the restart limit will now be placed in a
7279 Contributions from: Bertram Poettering, Dave Reisner, Huang
7280 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7281 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7285 * When running in --user mode systemd will now become a
7286 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7287 tree a lot more organized.
7289 * A new PartOf= unit dependency type has been introduced that
7290 may be used to group services in a natural way.
7292 * "systemctl enable" may now be used to enable instances of
7295 * journalctl now prints error log levels in red, and
7296 warning/notice log levels in bright white. It also supports
7297 filtering by log level now.
7299 * cgtop gained a new -n switch (similar to top), to configure
7300 the maximum number of iterations to run for. It also gained
7301 -b, to run in batch mode (accepting no input).
7303 * The suffix ".service" may now be omitted on most systemctl
7304 command lines involving service unit names.
7306 * There's a new bus call in logind to lock all sessions, as
7307 well as a loginctl verb for it "lock-sessions".
7309 * libsystemd-logind.so gained a new call sd_journal_perror()
7310 that works similar to libc perror() but logs to the journal
7311 and encodes structured information about the error number.
7313 * /etc/crypttab entries now understand the new keyfile-size=
7316 * shutdown(8) now can send a (configurable) wall message when
7317 a shutdown is cancelled.
7319 * The mount propagation mode for the root file system will now
7320 default to "shared", which is useful to make containers work
7321 nicely out-of-the-box so that they receive new mounts from
7322 the host. This can be undone locally by running "mount
7323 --make-rprivate /" if needed.
7325 * The prefdm.service file has been removed. Distributions
7326 should maintain this unit downstream if they intend to keep
7327 it around. However, we recommend writing normal unit files
7328 for display managers instead.
7330 * Since systemd is a crucial part of the OS we will now
7331 default to a number of compiler switches that improve
7332 security (hardening) such as read-only relocations, stack
7333 protection, and suchlike.
7335 * The TimeoutSec= setting for services is now split into
7336 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7337 of individual time outs for the start and the stop phase of
7340 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7341 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7342 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7343 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7344 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7345 Gundersen, Zbigniew Jędrzejewski-Szmek
7349 * The journal and id128 C APIs are now fully documented as man
7352 * Extra safety checks have been added when transitioning from
7353 the initial RAM disk to the main system to avoid accidental
7356 * /etc/crypttab entries now understand the new keyfile-offset=
7359 * systemctl -t can now be used to filter by unit load state.
7361 * The journal C API gained the new sd_journal_wait() call to
7362 make writing synchronous journal clients easier.
7364 * journalctl gained the new -D switch to show journals from a
7367 * journalctl now displays a special marker between log
7368 messages of two different boots.
7370 * The journal is now explicitly flushed to /var via a service
7371 systemd-journal-flush.service, rather than implicitly simply
7372 by seeing /var/log/journal to be writable.
7374 * journalctl (and the journal C APIs) can now match for much
7375 more complex expressions, with alternatives and
7378 * When transitioning from the initial RAM disk to the main
7379 system we will now kill all processes in a killing spree to
7380 ensure no processes stay around by accident.
7382 * Three new specifiers may be used in unit files: %u, %h, %s
7383 resolve to the user name, user home directory resp. user
7384 shell. This is useful for running systemd user instances.
7386 * We now automatically rotate journal files if their data
7387 object hash table gets a fill level > 75%. We also size the
7388 hash table based on the configured maximum file size. This
7389 together should lower hash collisions drastically and thus
7390 speed things up a bit.
7392 * journalctl gained the new "--header" switch to introspect
7393 header data of journal files.
7395 * A new setting SystemCallFilters= has been added to services
7396 which may be used to apply blacklists or whitelists to
7397 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7399 * nspawn gained a new --link-journal= switch (and quicker: -j)
7400 to link the container journal with the host. This makes it
7401 very easy to centralize log viewing on the host for all
7402 guests while still keeping the journal files separated.
7404 * Many bugfixes and optimizations
7406 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7407 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7408 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7413 * Several tools now understand kernel command line arguments,
7414 which are only read when run in an initial RAM disk. They
7415 usually follow closely their normal counterparts, but are
7418 * There's a new tool to analyze the readahead files that are
7419 automatically generated at boot. Use:
7421 /usr/lib/systemd/systemd-readahead analyze /.readahead
7423 * We now provide an early debug shell on tty9 if this enabled. Use:
7425 systemctl enable debug-shell.service
7427 * All plymouth related units have been moved into the Plymouth
7428 package. Please make sure to upgrade your Plymouth version
7431 * systemd-tmpfiles now supports getting passed the basename of
7432 a configuration file only, in which case it will look for it
7433 in all appropriate directories automatically.
7435 * udevadm info now takes a /dev or /sys path as argument, and
7436 does the right thing. Example:
7438 udevadm info /dev/sda
7439 udevadm info /sys/class/block/sda
7441 * systemctl now prints a warning if a unit is stopped but a
7442 unit that might trigger it continues to run. Example: a
7443 service is stopped but the socket that activates it is left
7446 * "systemctl status" will now mention if the log output was
7447 shortened due to rotation since a service has been started.
7449 * The journal API now exposes functions to determine the
7450 "cutoff" times due to rotation.
7452 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7453 immediately flushing of runtime logs to /var if possible,
7454 resp. for triggering immediate rotation of the journal
7457 * It is now considered an error if a service is attempted to
7458 be stopped that is not loaded.
7460 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7462 * systemd-analyze now supports Python 3
7464 * tmpfiles now supports cleaning up directories via aging
7465 where the first level dirs are always kept around but
7466 directories beneath it automatically aged. This is enabled
7467 by prefixing the age field with '~'.
7469 * Seat objects now expose CanGraphical, CanTTY properties
7470 which is required to deal with very fast bootups where the
7471 display manager might be running before the graphics drivers
7472 completed initialization.
7474 * Seat objects now expose a State property.
7476 * We now include RPM macros for service enabling/disabling
7477 based on the preset logic. We recommend RPM based
7478 distributions to make use of these macros if possible. This
7479 makes it simpler to reuse RPM spec files across
7482 * We now make sure that the collected systemd unit name is
7483 always valid when services log to the journal via
7486 * There's a new man page kernel-command-line(7) detailing all
7487 command line options we understand.
7489 * The fstab generator may now be disabled at boot by passing
7490 fstab=0 on the kernel command line.
7492 * A new kernel command line option modules-load= is now understood
7493 to load a specific kernel module statically, early at boot.
7495 * Unit names specified on the systemctl command line are now
7496 automatically escaped as needed. Also, if file system or
7497 device paths are specified they are automatically turned
7498 into the appropriate mount or device unit names. Example:
7500 systemctl status /home
7501 systemctl status /dev/sda
7503 * The SysVConsole= configuration option has been removed from
7504 system.conf parsing.
7506 * The SysV search path is no longer exported on the D-Bus
7509 * The Names= option has been removed from unit file parsing.
7511 * There's a new man page bootup(7) detailing the boot process.
7513 * Every unit and every generator we ship with systemd now
7514 comes with full documentation. The self-explanatory boot is
7517 * A couple of services gained "systemd-" prefixes in their
7518 name if they wrap systemd code, rather than only external
7519 code. Among them fsck@.service which is now
7520 systemd-fsck@.service.
7522 * The HaveWatchdog property has been removed from the D-Bus
7525 * systemd.confirm_spawn= on the kernel command line should now
7528 * There's a new man page crypttab(5) which details all options
7529 we actually understand.
7531 * systemd-nspawn gained a new --capability= switch to pass
7532 additional capabilities to the container.
7534 * timedated will now read known NTP implementation unit names
7535 from /usr/lib/systemd/ntp-units.d/*.list,
7536 systemd-timedated-ntp.target has been removed.
7538 * journalctl gained a new switch "-b" that lists log data of
7539 the current boot only.
7541 * The notify socket is in the abstract namespace again, in
7542 order to support daemons which chroot() at start-up.
7544 * There is a new Storage= configuration option for journald
7545 which allows configuration of where log data should go. This
7546 also provides a way to disable journal logging entirely, so
7547 that data collected is only forwarded to the console, the
7548 kernel log buffer or another syslog implementation.
7550 * Many bugfixes and optimizations
7552 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7553 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7554 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7555 Shawn Landden, Tom Gundersen
7559 * "systemctl help <unit>" now shows the man page if one is
7562 * Several new man pages have been added.
7564 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7565 MaxLevelConsole= can now be specified in
7566 journald.conf. These options allow reducing the amount of
7567 data stored on disk or forwarded by the log level.
7569 * TimerSlackNSec= can now be specified in system.conf for
7570 PID1. This allows system-wide power savings.
7572 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7573 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7578 * logind is now capable of (optionally) handling power and
7579 sleep keys as well as the lid switch.
7581 * journalctl now understands the syntax "journalctl
7582 /usr/bin/avahi-daemon" to get all log output of a specific
7585 * CapabilityBoundingSet= in system.conf now also influences
7586 the capability bound set of usermode helpers of the kernel.
7588 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7589 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7590 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7591 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7595 * Note that we skipped 139 releases here in order to set the
7596 new version to something that is greater than both udev's
7597 and systemd's most recent version number.
7599 * udev: all udev sources are merged into the systemd source tree now.
7600 All future udev development will happen in the systemd tree. It
7601 is still fully supported to use the udev daemon and tools without
7602 systemd running, like in initramfs or other init systems. Building
7603 udev though, will require the *build* of the systemd tree, but
7604 udev can be properly *run* without systemd.
7606 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7607 should be used to create dead device nodes as workarounds for broken
7610 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7611 no longer supported. udev_monitor_new_from_netlink() needs to be
7612 used to subscribe to events.
7614 * udev: when udevd is started by systemd, processes which are left
7615 behind by forking them off of udev rules, are unconditionally cleaned
7616 up and killed now after the event handling has finished. Services or
7617 daemons must be started as systemd services. Services can be
7618 pulled-in by udev to get started, but they can no longer be directly
7619 forked by udev rules.
7621 * udev: the daemon binary is called systemd-udevd now and installed
7622 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7623 to adapt to that, create symlink, or rename the binary after building
7626 * libudev no longer provides these symbols:
7627 udev_monitor_from_socket()
7628 udev_queue_get_failed_list_entry()
7629 udev_get_{dev,sys,run}_path()
7630 The versions number was bumped and symbol versioning introduced.
7632 * systemd-loginctl and systemd-journalctl have been renamed
7633 to loginctl and journalctl to match systemctl.
7635 * The config files: /etc/systemd/systemd-logind.conf and
7636 /etc/systemd/systemd-journald.conf have been renamed to
7637 logind.conf and journald.conf. Package updates should rename
7638 the files to the new names on upgrade.
7640 * For almost all files the license is now LGPL2.1+, changed
7641 from the previous GPL2.0+. Exceptions are some minor stuff
7642 of udev (which will be changed to LGPL2.1 eventually, too),
7643 and the MIT licensed sd-daemon.[ch] library that is suitable
7644 to be used as drop-in files.
7646 * systemd and logind now handle system sleep states, in
7647 particular suspending and hibernating.
7649 * logind now implements a sleep/shutdown/idle inhibiting logic
7650 suitable for a variety of uses. Soonishly Lennart will blog
7651 about this in more detail.
7653 * var-run.mount and var-lock.mount are no longer provided
7654 (which previously bind mounted these directories to their new
7655 places). Distributions which have not converted these
7656 directories to symlinks should consider stealing these files
7657 from git history and add them downstream.
7659 * We introduced the Documentation= field for units and added
7660 this to all our shipped units. This is useful to make it
7661 easier to explore the boot and the purpose of the various
7664 * All smaller setup units (such as
7665 systemd-vconsole-setup.service) now detect properly if they
7666 are run in a container and are skipped when
7667 appropriate. This guarantees an entirely noise-free boot in
7668 Linux container environments such as systemd-nspawn.
7670 * A framework for implementing offline system updates is now
7671 integrated, for details see:
7672 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7674 * A new service type Type=idle is available now which helps us
7675 avoiding ugly interleaving of getty output and boot status
7678 * There's now a system-wide CapabilityBoundingSet= option to
7679 globally reduce the set of capabilities for the
7680 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7681 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7682 even CAP_NET_ADMIN system-wide for secure systems.
7684 * There are now system-wide DefaultLimitXXX= options to
7685 globally change the defaults of the various resource limits
7686 for all units started by PID 1.
7688 * Harald Hoyer's systemd test suite has been integrated into
7689 systemd which allows easy testing of systemd builds in qemu
7690 and nspawn. (This is really awesome! Ask us for details!)
7692 * The fstab parser is now implemented as generator, not inside
7695 * systemctl will now warn you if .mount units generated from
7696 /etc/fstab are out of date due to changes in fstab that
7697 have not been read by systemd yet.
7699 * systemd is now suitable for usage in initrds. Dracut has
7700 already been updated to make use of this. With this in place
7701 initrds get a slight bit faster but primarily are much
7702 easier to introspect and debug since "systemctl status" in
7703 the host system can be used to introspect initrd services,
7704 and the journal from the initrd is kept around too.
7706 * systemd-delta has been added, a tool to explore differences
7707 between user/admin configuration and vendor defaults.
7709 * PrivateTmp= now affects both /tmp and /var/tmp.
7711 * Boot time status messages are now much prettier and feature
7712 proper english language. Booting up systemd has never been
7715 * Read-ahead pack files now include the inode number of all
7716 files to pre-cache. When the inode changes the pre-caching
7717 is not attempted. This should be nicer to deal with updated
7718 packages which might result in changes of read-ahead
7721 * We now temporaritly lower the kernel's read_ahead_kb variable
7722 when collecting read-ahead data to ensure the kernel's
7723 built-in read-ahead does not add noise to our measurements
7724 of necessary blocks to pre-cache.
7726 * There's now RequiresMountsFor= to add automatic dependencies
7727 for all mounts necessary for a specific file system path.
7729 * MountAuto= and SwapAuto= have been removed from
7730 system.conf. Mounting file systems at boot has to take place
7733 * nspawn now learned a new switch --uuid= to set the machine
7734 ID on the command line.
7736 * nspawn now learned the -b switch to automatically search
7739 * vt102 is now the default TERM for serial TTYs, upgraded from
7742 * systemd-logind now works on VT-less systems.
7744 * The build tree has been reorganized. The individual
7745 components now have directories of their own.
7747 * A new condition type ConditionPathIsReadWrite= is now available.
7749 * nspawn learned the new -C switch to create cgroups for the
7750 container in other hierarchies.
7752 * We now have support for hardware watchdogs, configurable in
7755 * The scheduled shutdown logic now has a public API.
7757 * We now mount /tmp as tmpfs by default, but this can be
7758 masked and /etc/fstab can override it.
7760 * Since udisks does not make use of /media anymore we are not
7761 mounting a tmpfs on it anymore.
7763 * journalctl gained a new --local switch to only interleave
7764 locally generated journal files.
7766 * We can now load the IMA policy at boot automatically.
7768 * The GTK tools have been split off into a systemd-ui.
7770 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7771 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7772 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7773 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7774 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7775 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7776 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7777 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7778 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7783 * This is mostly a bugfix release
7785 * Support optional initialization of the machine ID from the
7786 KVM or container configured UUID.
7788 * Support immediate reboots with "systemctl reboot -ff"
7790 * Show /etc/os-release data in systemd-analyze output
7792 * Many bugfixes for the journal, including endianness fixes and
7793 ensuring that disk space enforcement works
7795 * sd-login.h is C++ compatible again
7797 * Extend the /etc/os-release format on request of the Debian
7800 * We now refuse non-UTF8 strings used in various configuration
7801 and unit files. This is done to ensure we do not pass invalid
7802 data over D-Bus or expose it elsewhere.
7804 * Register Mimo USB Screens as suitable for automatic seat
7807 * Read SELinux client context from journal clients in a race
7810 * Reorder configuration file lookup order. /etc now always
7811 overrides /run in order to allow the administrator to always
7812 and unconditionally override vendor-supplied or
7813 automatically generated data.
7815 * The various user visible bits of the journal now have man
7816 pages. We still lack man pages for the journal API calls
7819 * We now ship all man pages in HTML format again in the
7822 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7823 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7824 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7829 * This is mostly a bugfix release
7831 * systems lacking /etc/os-release are no longer supported.
7833 * Various functionality updates to libsystemd-login.so
7835 * Track class of PAM logins to distinguish greeters from
7838 Contributions from: Kay Sievers, Lennart Poettering, Michael
7843 * This is an important bugfix release for v41.
7845 * Building man pages is now optional which should be useful
7846 for those building systemd from git but unwilling to install
7849 * Watchdog support for supervising services is now usable. In
7850 a future release support for hardware watchdogs
7851 (i.e. /dev/watchdog) will be added building on this.
7853 * Service start rate limiting is now configurable and can be
7854 turned off per service. When a start rate limit is hit a
7855 reboot can automatically be triggered.
7857 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7859 Contributions from: Benjamin Franzke, Bill Nottingham,
7860 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7861 Schmidt, Michał Górny, Piotr Drąg
7865 * The systemd binary is installed /usr/lib/systemd/systemd now;
7866 An existing /sbin/init symlink needs to be adapted with the
7869 * The code that loads kernel modules has been ported to invoke
7870 libkmod directly, instead of modprobe. This means we do not
7871 support systems with module-init-tools anymore.
7873 * Watchdog support is now already useful, but still not
7876 * A new kernel command line option systemd.setenv= is
7877 understood to set system wide environment variables
7878 dynamically at boot.
7880 * We now limit the set of capabilities of systemd-journald.
7882 * We now set SIGPIPE to ignore by default, since it only is
7883 useful in shell pipelines, and has little use in general
7884 code. This can be disabled with IgnoreSIPIPE=no in unit
7887 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7888 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7893 * This is mostly a bugfix release
7895 * We now expose the reason why a service failed in the
7896 "Result" D-Bus property.
7898 * Rudimentary service watchdog support (will be completed over
7899 the next few releases.)
7901 * When systemd forks off in order execute some service we will
7902 now immediately changes its argv[0] to reflect which process
7903 it will execute. This is useful to minimize the time window
7904 with a generic argv[0], which makes bootcharts more useful
7906 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7907 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7908 Mike Kazantsev, Ray Strode
7912 * This is mostly a test release, but incorporates many
7915 * New systemd-cgtop tool to show control groups by their
7918 * Linking against libacl for ACLs is optional again. If
7919 disabled, support tracking device access for active logins
7920 goes becomes unavailable, and so does access to the user
7921 journals by the respective users.
7923 * If a group "adm" exists, journal files are automatically
7924 owned by them, thus allow members of this group full access
7925 to the system journal as well as all user journals.
7927 * The journal now stores the SELinux context of the logging
7928 client for all entries.
7930 * Add C++ inclusion guards to all public headers
7932 * New output mode "cat" in the journal to print only text
7933 messages, without any meta data like date or time.
7935 * Include tiny X server wrapper as a temporary stop-gap to
7936 teach XOrg udev display enumeration. This is used by display
7937 managers such as gdm, and will go away as soon as XOrg
7938 learned native udev hotplugging for display devices.
7940 * Add new systemd-cat tool for executing arbitrary programs
7941 with STDERR/STDOUT connected to the journal. Can also act as
7942 BSD logger replacement, and does so by default.
7944 * Optionally store all locally generated coredumps in the
7945 journal along with meta data.
7947 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7948 writing short strings to files (for usage for /sys), and for
7949 creating symlinks, character and block device nodes.
7951 * New unit file option ControlGroupPersistent= to make cgroups
7952 persistent, following the mechanisms outlined in
7953 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7955 * Support multiple local RTCs in a sane way
7957 * No longer monopolize IO when replaying readahead data on
7958 rotating disks, since we might starve non-file-system IO to
7959 death, since fanotify() will not see accesses done by blkid,
7962 * Do not show kernel threads in systemd-cgls anymore, unless
7963 requested with new -k switch.
7965 Contributions from: Dan Horák, Kay Sievers, Lennart
7966 Poettering, Michal Schmidt
7970 * This is mostly a test release, but incorporates many
7973 * The git repository moved to:
7974 git://anongit.freedesktop.org/systemd/systemd
7975 ssh://git.freedesktop.org/git/systemd/systemd
7977 * First release with the journal
7978 http://0pointer.de/blog/projects/the-journal.html
7980 * The journal replaces both systemd-kmsg-syslogd and
7981 systemd-stdout-bridge.
7983 * New sd_pid_get_unit() API call in libsystemd-logind
7985 * Many systemadm clean-ups
7987 * Introduce remote-fs-pre.target which is ordered before all
7988 remote mounts and may be used to start services before all
7991 * Added Mageia support
7993 * Add bash completion for systemd-loginctl
7995 * Actively monitor PID file creation for daemons which exit in
7996 the parent process before having finished writing the PID
7997 file in the daemon process. Daemons which do this need to be
7998 fixed (i.e. PID file creation must have finished before the
7999 parent exits), but we now react a bit more gracefully to them.
8001 * Add colourful boot output, mimicking the well-known output
8002 of existing distributions.
8004 * New option PassCredentials= for socket units, for
8005 compatibility with a recent kernel ABI breakage.
8007 * /etc/rc.local is now hooked in via a generator binary, and
8008 thus will no longer act as synchronization point during
8011 * systemctl list-unit-files now supports --root=.
8013 * systemd-tmpfiles now understands two new commands: z, Z for
8014 relabelling files according to the SELinux database. This is
8015 useful to apply SELinux labels to specific files in /sys,
8018 * Output of SysV services is now forwarded to both the console
8019 and the journal by default, not only just the console.
8021 * New man pages for all APIs from libsystemd-login.
8023 * The build tree got reorganized and the build system is a
8024 lot more modular allowing embedded setups to specifically
8025 select the components of systemd they are interested in.
8027 * Support for Linux systems lacking the kernel VT subsystem is
8030 * configure's --with-rootdir= got renamed to
8031 --with-rootprefix= to follow the naming used by udev and
8034 * Unless specified otherwise we will now install to /usr instead
8035 of /usr/local by default.
8037 * Processes with '@' in argv[0][0] are now excluded from the
8038 final shut-down killing spree, following the logic explained
8040 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8042 * All processes remaining in a service cgroup when we enter
8043 the START or START_PRE states are now killed with
8044 SIGKILL. That means it is no longer possible to spawn
8045 background processes from ExecStart= lines (which was never
8046 supported anyway, and bad style).
8048 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8049 reloading of units together.
8051 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8052 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8053 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8054 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8055 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob