1 systemd System and Service Manager
3 CHANGES WITH 237 in spe:
5 * Some keyboards come with a zoom see-saw or rocker which until now got
6 mapped to the Linux "zoomin/out" keys in hwdb. However, these
7 keycodes are not recognized by any major desktop. They now produce
8 Up/Down key events so that they can be used for scrolling.
10 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
11 slightly: previously, if an argument was specified for lines of this
12 type (i.e. the right-most column was set) this string was appended to
13 existing files each time elogind-tmpfiles was run. This behaviour was
14 different from what the documentation said, and not particularly
15 useful, as repeated elogind-tmpfiles invocations would not be
16 idempotent and grow such files without bounds. With this release
17 behaviour has been altered slightly, to match what the documentation
18 says: lines of this type only have an effect if the indicated files
19 don't exist yet, and only then the argument string is written to the
22 * A new environment variable $SYSTEMD_OFFLINE is now understood by
23 systemctl. It takes a boolean argument. If on, systemctl assumes it
24 operates on an "offline" OS tree, and will not attempt to talk to the
25 service manager. Previously, this mode was implicitly enabled if a
26 chroot() environment was detected, and this new environment variable
27 now provides explicit control.
29 * .path and .socket units may now be created transiently,
30 too. Previously only service, mount, automount and timer units were
31 supported as transient units. The elogind-run tool has been updated
32 to expose this new functionality, you may hence use it now to bind
33 arbitrary commands to path or socket activation on-the-fly from the
34 command line. Moreover, for the unit types already supporting
35 transient operation most properties are now hooked up.
37 * The elogind-mount command gained support for a new --owner= parameter
38 which takes a user name, which is then resolved and included in uid=
39 and gid= mount options string of the file system to mount.
41 * A new unit condition ConditionControlGroupController= has been added
42 that checks whether a specific cgroup controller is available.
44 * Unit files, udev's .link files, and elogind-networkd's .netdev and
45 .network files all gained support for a new condition
46 ConditionKernelVersion= for checking against specific kernel
49 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
50 support for configuring device flags in the Flags= setting. In the
51 same files, the [Tunnel] section gained support for configuring
52 AllowLocalRemote=. The [Route] section in .network files gained
53 support for configuring InitialCongestionWindow=,
54 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
55 understands RapidCommit=.
57 * elogind-networkd's DHCPv6 support gained support for Prefix
60 * sd-bus gained support for a new "watch-bind" feature. When this
61 feature is enabled, an sd_bus connection may be set up to connect to
62 an AF_UNIX socket in the file system as soon as it is created. This
63 functionality is useful for writing early-boot services that
64 automatically connect to the system bus as soon as it is started,
65 without ugly time-based polling. elogind-networkd and
66 elogind-resolved have been updated to make use of this
67 functionality. busctl exposes this functionality in a new
68 --watch-bind= command line switch.
70 * sd-bus will now optionally synthesize a local "Connected" signal as
71 soon as a D-Bus connection is set up fully. This message mirrors the
72 already existing "Disconnected" signal which is synthesized when the
73 connection is terminated. This signal is generally useful but
74 particularly handy in combination with the "watch-bind" feature
75 described above. Synthesizing of this message has to be requested
76 explicitly through the new API call sd_bus_set_connected_signal(). In
77 addition a new call sd_bus_is_ready() has been added that checks
78 wether a connection is fully set up (i.e. between the "Connected" and
79 "Disconnected" signals).
81 * sd-bus gained two new calls sd_bus_request_name_async() and
82 sd_bus_release_name_async() for asynchronously registering bus
83 names. Similar, there is now sd_bus_add_match_async() for installing
84 a signal match asynchronously. All of elogind's own services have
85 been updated to make use of these calls. Doing these operations
86 asynchronously has two benefits: it reduces the risk of deadlocks in
87 case of cyclic dependencies between bus services, and it speeds up
88 service initialization since synchronization points for bus
89 round-trips are removed.
91 * sd-bus gained two new calls sd_bus_match_signal() and
92 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
93 and sd_bus_add_match_async() but instead of taking a D-Bus match
94 string take match fields as normal function parameters.
96 * sd-bus gained two new calls sd_bus_set_sender() and
97 sd_bus_message_set_sender() for setting the sender name of outgoing
98 messages (either for all outgoing messages or for just one specific
99 one). These calls are only useful in direct connections as on
100 brokered connections the broker fills in the sender anyway,
101 overwriting whatever the client filled in.
103 * sd-event gained a new pseudo-handle that may be specified on all API
104 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
105 used this refers to the default event loop object of the calling
106 thread. Note however that this does not implicitly allocate one —
107 which has to be done prior by using sd_event_default(). Similarly
108 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
109 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
110 to the default bus of the specified type of the calling thread. Here
111 too this does not implicitly allocate bus connection objects, this
112 has to be done prior with sd_bus_default() and friends.
114 * sd-event gained a new call pair
115 sd_event_source_{get|set}_io_fd_own(). This may be used to request
116 automatic closure of the file descriptor an IO event source watches
117 when the event source is destroyed.
119 * elogind-networkd gained support for natively configuring WireGuard
122 * In previous versions elogind synthesized user records both for the
123 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
124 internally. In order to simplify distribution-wide renames of the
125 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
126 new transitional flag file has been added: if
127 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
128 user and group record within the elogind codebase is disabled.
130 * elogind-notify gained a new --uid= option for selecting the source
131 user/UID to use for notification messages sent to the service
134 * There's now a "elogind-analyze service-watchdogs" command for turning
135 on or off the per-service watchdogs system-wide (i.e. the concept you
136 configure in WatchdogSec=), for debugging purposes. There's also a
137 kernel command line option elogind.service_watchdogs= for controlling
140 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
141 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Björn Esser,
142 bleep_blop, Bruce A. Johnson, Chris Down, Clinton Roy, Colin Walters,
143 Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov, Evgeny Vereshchagin,
144 Ewout van Mansom, Felipe Sateler, Franck Bui, Frantisek Sumsal, George
145 Gaydarov, Gianluca Boiano, Hans-Christian Noren Egtvedt, Hans de Goede,
146 Henrik Grindal Bakken, Jan Alexander Steffens, Jan Klötzke, Jason
147 A. Donenfeld, jdkbx, Jérémy Rosen, Jerónimo Borque, John Lin, John Paul
148 Herold, Jonathan Rudenberg, Jörg Thalheim, Ken (Bitsko) MacLeod, Larry
149 Bernstone, Lennart Poettering, Lucas Werkmeister, Maciej S. Szmigiero,
150 Marek Čermák, Martin Pitt, Mathieu Malaterre, Matthew Thode,
151 Matthias-Christian Ott, Max Harmathy, Michael Biebl, Michael Vogt,
152 Michal Koutný, Michal Sekletar, Michał Szczepański, Mike Gilbert,
153 Nathaniel McCallum, Nicolas Chauvet, Olaf Hering, Olivier Schwander,
154 Patrik Flykt, Paul Cercueil, Peter Hutterer, Piotr Drąg, Raphael
155 Vogelgsang, Reverend Homer, Robert Kolchmeyer, Samuel Dionne-Riel,
156 Sergey Ptashnick, Shawn Landden, Susant Sahani, Sylvain Plantefève,
157 Thomas Huth, Tomasz Bachorski, Vladislav Vishnyakov, Wieland Hoffmann,
158 Yu Watanabe, Zachary Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан
159 Георгиевски, Дилян Палаузов
165 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
166 in v235 has been extended to also set the dummy.ko module option
167 numdummies=0, preventing the kernel from automatically creating
168 dummy0. All dummy interfaces must now be explicitly created.
170 * Unknown '%' specifiers in configuration files are now rejected. This
171 applies to units and tmpfiles.d configuration. Any percent characters
172 that are followed by a letter or digit that are not supposed to be
173 interpreted as the beginning of a specifier should be escaped by
174 doubling ("%%"). (So "size=5%" is still accepted, as well as
175 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
176 valid specifiers today.)
178 * systemd-resolved now maintains a new dynamic
179 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
180 recommended to make /etc/resolv.conf a symlink to it. This file
181 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
182 includes dynamically acquired search domains, achieving more correct
183 DNS resolution by software that bypasses local DNS APIs such as NSS.
185 * The "uaccess" udev tag has been dropped from /dev/kvm and
186 /dev/dri/renderD*. These devices now have the 0666 permissions by
187 default (but this may be changed at build-time). /dev/dri/renderD*
188 will now be owned by the "render" group along with /dev/kfd.
190 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
191 systemd-journal-gatewayd.service and
192 systemd-journal-upload.service. This means "nss-systemd" must be
193 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
194 services are resolved properly.
196 * In /etc/fstab two new mount options are now understood:
197 x-systemd.makefs and x-systemd.growfs. The former has the effect that
198 the configured file system is formatted before it is mounted, the
199 latter that the file system is resized to the full block device size
200 after it is mounted (i.e. if the file system is smaller than the
201 partition it resides on, it's grown). This is similar to the fsck
202 logic in /etc/fstab, and pulls in systemd-makefs@.service and
203 systemd-growfs@.service as necessary, similar to
204 systemd-fsck@.service. Resizing is currently only supported on ext4
207 * In systemd-networkd, the IPv6 RA logic now optionally may announce
208 DNS server and domain information.
210 * Support for the LUKS2 on-disk format for encrypted partitions has
211 been added. This requires libcryptsetup2 during compilation and
214 * The systemd --user instance will now signal "readiness" when its
215 basic.target unit has been reached, instead of when the run queue ran
216 empty for the first time.
218 * Tmpfiles.d with user configuration are now also supported.
219 systemd-tmpfiles gained a new --user switch, and snippets placed in
220 ~/.config/user-tmpfiles.d/ and corresponding directories will be
221 executed by systemd-tmpfiles --user running in the new
222 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
223 running in the user session.
225 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
226 %S resolves to the top-level state directory (/var/lib for the system
227 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
228 top-level cache directory (/var/cache for the system instance,
229 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
230 logs directory (/var/log for the system instance,
231 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
232 existing %t specifier, that resolves to the top-level runtime
233 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
236 * journalctl learnt a new parameter --output-fields= for limiting the
237 set of journal fields to output in verbose and JSON output modes.
239 * systemd-timesyncd's configuration file gained a new option
240 RootDistanceMaxSec= for setting the maximum root distance of servers
241 it'll use, as well as the new options PollIntervalMinSec= and
242 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
244 * bootctl gained a new command "list" for listing all available boot
245 menu items on systems that follow the boot loader specification.
247 * systemctl gained a new --dry-run switch that shows what would be done
248 instead of doing it, and is currently supported by the shutdown and
251 * ConditionSecurity= can now detect the TOMOYO security module.
253 * Unit file [Install] sections are now also respected in unit drop-in
254 files. This is intended to be used by drop-ins under /usr/lib/.
256 * systemd-firstboot may now also set the initial keyboard mapping.
258 * Udev "changed" events for devices which are exposed as systemd
259 .device units are now propagated to units specified in
260 ReloadPropagatedFrom= as reload requests.
262 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
263 unit template name (i.e. a name in the form of 'foobar@.service',
264 without the instance component between the '@' and - the '.'), then
265 the escaped sysfs path of the device is automatically used as the
268 * SystemCallFilter= in unit files has been extended so that an "errno"
269 can be specified individually for each system call. Example:
270 SystemCallFilter=~uname:EILSEQ.
272 * The cgroup delegation logic has been substantially updated. Delegate=
273 now optionally takes a list of controllers (instead of a boolean, as
274 before), which lists the controllers to delegate at least.
276 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
278 * A new LogLevelMax= setting configures the maximum log level any
279 process of the service may log at (i.e. anything with a lesser
280 priority than what is specified is automatically dropped). A new
281 LogExtraFields= setting allows configuration of additional journal
282 fields to attach to all log records generated by any of the unit's
285 * New StandardInputData= and StandardInputText= settings along with the
286 new option StandardInput=data may be used to configure textual or
287 binary data that shall be passed to the executed service process via
288 standard input, encoded in-line in the unit file.
290 * StandardInput=, StandardOutput= and StandardError= may now be used to
291 connect stdin/stdout/stderr of executed processes directly with a
292 file or AF_UNIX socket in the file system, using the new "file:" option.
294 * A new unit file option CollectMode= has been added, that allows
295 tweaking the garbage collection logic for units. It may be used to
296 tell systemd to garbage collect units that have failed automatically
297 (normally it only GCs units that exited successfully). systemd-run
298 and systemd-mount expose this new functionality with a new -G option.
300 * "machinectl bind" may now be used to bind mount non-directories
301 (i.e. regularfiles, devices, fifos, sockets).
303 * systemd-analyze gained a new verb "calendar" for validating and
304 testing calendar time specifications to use for OnCalendar= in timer
305 units. Besides validating the expression it will calculate the next
306 time the specified expression would elapse.
308 * In addition to the pre-existing FailureAction= unit file setting
309 there's now SuccessAction=, for configuring a shutdown action to
310 execute when a unit completes successfully. This is useful in
311 particular inside containers that shall terminate after some workload
312 has been completed. Also, both options are now supported for all unit
313 types, not just services.
315 * networkds's IP rule support gained two new options
316 IncomingInterface= and OutgoingInterface= for configuring the incoming
317 and outgoing interfaces of configured rules. systemd-networkd also
318 gained support for "vxcan" network devices.
320 * networkd gained a new setting RequiredForOnline=, taking a
321 boolean. If set, systemd-wait-online will take it into consideration
322 when determining that the system is up, otherwise it will ignore the
323 interface for this purpose.
325 * The sd_notify() protocol gained support for a new operation: with
326 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
327 store again, ahead of POLLHUP or POLLERR when they are removed
330 * A new document UIDS-GIDS.md has been added to the source tree, that
331 documents the UID/GID range and assignment assumptions and
332 requirements of systemd.
334 * The watchdog device PID 1 will ping may now be configured through the
335 WatchdogDevice= configuration file setting, or by setting the
336 systemd.watchdog_service= kernel commandline option.
338 * systemd-resolved's gained support for registering DNS-SD services on
339 the local network using MulticastDNS. Services may either be
340 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
341 the same dir below /run, /usr/lib), or through its D-Bus API.
343 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
344 extend the effective start, runtime, and stop time. The service must
345 continue to send EXTEND_TIMEOUT_USEC within the period specified to
346 prevent the service manager from making the service as timedout.
348 * elogind-resolved's DNSSEC support gained support for RFC 8080
349 (Ed25519 keys and signatures).
351 * The elogind-resolve command line tool gained a new set of options
352 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
353 --set-nta= and --revert to configure per-interface DNS configuration
354 dynamically during runtime. It's useful for pushing DNS information
355 into elogind-resolved from DNS hook scripts that various interface
356 managing software supports (such as pppd).
358 * elogind-nspawn gained a new --network-namespace-path= command line
359 option, which may be used to make a container join an existing
360 network namespace, by specifying a path to a "netns" file.
362 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
363 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
364 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
365 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
366 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
367 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
368 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
369 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
370 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
371 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
372 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
373 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
374 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
375 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
376 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
377 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
378 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
379 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
380 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
381 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
382 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
383 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
384 Jędrzejewski-Szmek, Zeal Jagannatha
390 * INCOMPATIBILITY: systemd-logind.service and other long-running
391 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
392 communication with the outside. This generally improves security of
393 the system, and is in almost all cases a safe and good choice, as
394 these services do not and should not provide any network-facing
395 functionality. However, systemd-logind uses the glibc NSS API to
396 query the user database. This creates problems on systems where NSS
397 is set up to directly consult network services for user database
398 lookups. In particular, this creates incompatibilities with the
399 "nss-nis" module, which attempts to directly contact the NIS/YP
400 network servers it is configured for, and will now consistently
401 fail. In such cases, it is possible to turn off IP sandboxing for
402 systemd-logind.service (set IPAddressDeny= in its [Service] section
403 to the empty string, via a .d/ unit file drop-in). Downstream
404 distributions might want to update their nss-nis packaging to include
405 such a drop-in snippet, accordingly, to hide this incompatibility
406 from the user. Another option is to make use of glibc's nscd service
407 to proxy such network requests through a privilege-separated, minimal
408 local caching daemon, or to switch to more modern technologies such
409 sssd, whose NSS hook-ups generally do not involve direct network
410 access. In general, we think it's definitely time to question the
411 implementation choices of nss-nis, i.e. whether it's a good idea
412 today to embed a network-facing loadable module into all local
413 processes that need to query the user database, including the most
414 trivial and benign ones, such as "ls". For more details about
415 IPAddressDeny= see below.
417 * A new modprobe.d drop-in is now shipped by default that sets the
418 bonding module option max_bonds=0. This overrides the kernel default,
419 to avoid conflicts and ambiguity as to whether or not bond0 should be
420 managed by systemd-networkd or not. This resolves multiple issues
421 with bond0 properties not being applied, when bond0 is configured
422 with systemd-networkd. Distributors may choose to not package this,
423 however in that case users will be prevented from correctly managing
424 bond0 interface using systemd-networkd.
426 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
427 which print the logging level and target of the system manager. They
428 complement the existing "set-log-level" and "set-log-target" verbs
429 used to change those values.
431 * journald.conf gained a new boolean setting ReadKMsg= which defaults
432 to on. If turned off kernel log messages will not be read by
433 systemd-journald or included in the logs. It also gained a new
434 setting LineMax= for configuring the maximum line length in
435 STDOUT/STDERR log streams. The new default for this value is 48K, up
436 from the previous hardcoded 2048.
438 * A new unit setting RuntimeDirectoryPreserve= has been added, which
439 allows more detailed control of what to do with a runtime directory
440 configured with RuntimeDirectory= (i.e. a directory below /run or
441 $XDG_RUNTIME_DIR) after a unit is stopped.
443 * The RuntimeDirectory= setting for units gained support for creating
444 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
445 one top-level directory.
447 * Units gained new options StateDirectory=, CacheDirectory=,
448 LogsDirectory= and ConfigurationDirectory= which are closely related
449 to RuntimeDirectory= but manage per-service directories below
450 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
451 possible to write unit files which when activated automatically gain
452 properly owned service specific directories in these locations, thus
453 making unit files self-contained and increasing compatibility with
454 stateless systems and factory reset where /etc or /var are
455 unpopulated at boot. Matching these new settings there's also
456 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
457 ConfigurationDirectoryMode= for configuring the access mode of these
458 directories. These settings are particularly useful in combination
459 with DynamicUser=yes as they provide secure, properly-owned,
460 writable, and stateful locations for storage, excluded from the
461 sandbox that such services live in otherwise.
463 * Automake support has been removed from this release. systemd is now
466 * systemd-journald will now aggressively cache client metadata during
467 runtime, speeding up log write performance under pressure. This comes
468 at a small price though: as much of the metadata is read
469 asynchronously from /proc/ (and isn't implicitly attached to log
470 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
471 metadata stored alongside a log entry might be slightly
472 out-of-date. Previously it could only be slightly newer than the log
473 message. The time window is small however, and given that the kernel
474 is unlikely to be improved anytime soon in this regard, this appears
477 * nss-myhostname/systemd-resolved will now by default synthesize an
478 A/AAAA resource record for the "_gateway" hostname, pointing to the
479 current default IP gateway. Previously it did that for the "gateway"
480 name, hampering adoption, as some distributions wanted to leave that
481 host name open for local use. The old behaviour may still be
482 requested at build time.
484 * systemd-networkd's [Address] section in .network files gained a new
485 Scope= setting for configuring the IP address scope. The [Network]
486 section gained a new boolean setting ConfigureWithoutCarrier= that
487 tells systemd-networkd to ignore link sensing when configuring the
488 device. The [DHCP] section gained a new Anonymize= boolean option for
489 turning on a number of options suggested in RFC 7844. A new
490 [RoutingPolicyRule] section has been added for configuring the IP
491 routing policy. The [Route] section has gained support for a new
492 Type= setting which permits configuring
493 blackhole/unreachable/prohibit routes.
495 * The [VRF] section in .netdev files gained a new Table= setting for
496 configuring the routing table to use. The [Tunnel] section gained a
497 new Independent= boolean field for configuring tunnels independent of
498 an underlying network interface. The [Bridge] section gained a new
499 GroupForwardMask= option for configuration of propagation of link
500 local frames between bridge ports.
502 * The WakeOnLan= setting in .link files gained support for a number of
503 new modes. A new TCP6SegmentationOffload= setting has been added for
504 configuring TCP/IPv6 hardware segmentation offload.
506 * The IPv6 RA sender implementation may now optionally send out RDNSS
507 and RDNSSL records to supply DNS configuration to peers.
509 * systemd-nspawn gained support for a new --system-call-filter= command
510 line option for adding and removing entries in the default system
511 call filter it applies. Moreover systemd-nspawn has been changed to
512 implement a system call whitelist instead of a blacklist.
514 * systemd-run gained support for a new --pipe command line option. If
515 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
516 are directly passed on to the activated transient service
517 executable. This allows invoking arbitrary processes as systemd
518 services (for example to take benefit of dependency management,
519 accounting management, resource management or log management that is
520 done automatically for services) — while still allowing them to be
521 integrated in a classic UNIX shell pipeline.
523 * When a service sends RELOAD=1 via sd_notify() and reload propagation
524 using ReloadPropagationTo= is configured, a reload is now propagated
525 to configured units. (Previously this was only done on explicitly
526 requested reloads, using "systemctl reload" or an equivalent
529 * For each service unit a restart counter is now kept: it is increased
530 each time the service is restarted due to Restart=, and may be
531 queried using "systemctl show -p NRestarts …".
533 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
534 @signal and @timer have been added, for usage with SystemCallFilter=
535 in unit files and the new --system-call-filter= command line option
536 of systemd-nspawn (see above).
538 * ExecStart= lines in unit files gained two new modifiers: when a
539 command line is prefixed with "!" the command will be executed as
540 configured, except for the credentials applied by
541 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
542 "+", but does still apply namespacing options unlike "+". There's
543 also "!!" now, which is mostly identical, but becomes a NOP on
544 systems that support ambient capabilities. This is useful to write
545 unit files that work with ambient capabilities where possible but
546 automatically fall back to traditional privilege dropping mechanisms
547 on systems where this is not supported.
549 * ListenNetlink= settings in socket units now support RDMA netlink
552 * A new unit file setting LockPersonality= has been added which permits
553 locking down the chosen execution domain ("personality") of a service
556 * A new special target "getty-pre.target" has been added, which is
557 ordered before all text logins, and may be used to order services
558 before textual logins acquire access to the console.
560 * systemd will now attempt to load the virtio-rng.ko kernel module very
561 early on if a VM environment supporting this is detected. This should
562 improve entropy during early boot in virtualized environments.
564 * A _netdev option is now supported in /etc/crypttab that operates in a
565 similar way as the same option in /etc/fstab: it permits configuring
566 encrypted devices that need to be ordered after the network is up.
567 Following this logic, two new special targets
568 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
569 added that are to cryptsetup.target what remote-fs.target and
570 remote-fs-pre.target are to local-fs.target.
572 * Service units gained a new UnsetEnvironment= setting which permits
573 unsetting specific environment variables for services that are
574 normally passed to it (for example in order to mask out locale
575 settings for specific services that can't deal with it).
577 * Units acquired a new boolean option IPAccounting=. When turned on, IP
578 traffic accounting (packet count as well as byte count) is done for
579 the service, and shown as part of "systemctl status" or "systemd-run
582 * Service units acquired two new options IPAddressAllow= and
583 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
584 for configuring a simple IP access control list for all sockets of
585 the unit. These options are available also on .slice and .socket
586 units, permitting flexible access list configuration for individual
587 services as well as groups of services (as defined by a slice unit),
588 including system-wide. Note that IP ACLs configured this way are
589 enforced on every single IPv4 and IPv6 socket created by any process
590 of the service unit, and apply to ingress as well as egress traffic.
592 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
593 structured log message is generated each time the unit is stopped,
594 containing information about the consumed resources of this
597 * A new setting KeyringMode= has been added to unit files, which may be
598 used to control how the kernel keyring is set up for executed
601 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
602 "systemctl kexec" and "systemctl exit" are now always asynchronous in
603 behaviour (that is: these commands return immediately after the
604 operation was enqueued instead of waiting for the operation to
605 complete). Previously, "systemctl poweroff" and "systemctl reboot"
606 were asynchronous on systems using systemd-logind (i.e. almost
607 always, and like they were on sysvinit), and the other three commands
608 were unconditionally synchronous. With this release this is cleaned
609 up, and callers will see the same asynchronous behaviour on all
610 systems for all five operations.
612 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
615 * .timer units now accept calendar specifications in other timezones
616 than UTC or the local timezone.
618 * The tmpfiles snippet var.conf has been changed to create
619 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
620 the "utmp" group already, and it appears to be generally understood
621 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
622 databases. Previously this was implemented correctly for all these
623 databases excepts btmp, which has been opened up like this now
624 too. Note that while the other databases are world-readable
625 (i.e. 0644), btmp is not and remains more restrictive.
627 * The systemd-resolve tool gained a new --reset-server-features
628 switch. When invoked like this systemd-resolved will forget
629 everything it learnt about the features supported by the configured
630 upstream DNS servers, and restarts the feature probing logic on the
631 next resolver look-up for them at the highest feature level
634 * The status dump systemd-resolved sends to the logs upon receiving
635 SIGUSR1 now also includes information about all DNS servers it is
636 configured to use, and the features levels it probed for them.
638 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
639 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
640 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
641 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
642 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
643 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
644 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
645 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
646 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
647 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
648 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
649 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
650 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
651 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
652 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
653 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
654 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
655 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
656 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
657 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
663 * Meson is now supported as build system in addition to Automake. It is
664 our plan to remove Automake in one of our next releases, so that
665 Meson becomes our exclusive build system. Hence, please start using
666 the Meson build system in your downstream packaging. There's plenty
667 of documentation around how to use Meson, the extremely brief
670 ./autogen.sh && ./configure && make && sudo make install
674 meson build && ninja -C build && sudo ninja -C build install
676 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
677 which permits configuring a timeout on the time a job is
678 running. This is particularly useful for setting timeouts on jobs for
681 * Unit files gained two new options ConditionUser= and ConditionGroup=
682 for conditionalizing units based on the identity of the user/group
683 running a systemd user instance.
685 * systemd-networkd now understands a new FlowLabel= setting in the
686 [VXLAN] section of .network files, as well as a Priority= in
687 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
688 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
689 gained support for configuration of GENEVE links, and IPv6 address
690 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
692 * .link files now understand a new Port= setting.
694 * systemd-networkd's DHCP support gained support for DHCP option 119
695 (domain search list).
697 * systemd-networkd gained support for serving IPv6 address ranges using
698 the Router Advertisment protocol. The new .network configuration
699 section [IPv6Prefix] may be used to configure the ranges to
700 serve. This is implemented based on a new, minimal, native server
701 implementation of RA.
703 * journalctl's --output= switch gained support for a new parameter
704 "short-iso-precise" for a mode where timestamps are shown as precise
707 * systemd-udevd's "net_id" builtin may now generate stable network
708 interface names from IBM PowerVM VIO devices as well as ACPI platform
711 * MulticastDNS support in systemd-resolved may now be explicitly
712 enabled/disabled using the new MulticastDNS= configuration file
715 * systemd-resolved may now optionally use libidn2 instead of the libidn
716 for processing internationalized domain names. Support for libidn2
717 should be considered experimental and should not be enabled by
720 * "machinectl pull-tar" and related call may now do verification of
721 downloaded images using SUSE-style .sha256 checksum files in addition
722 to the already existing support for validating using Ubuntu-style
725 * sd-bus gained support for a new sd_bus_message_appendv() call which
726 is va_list equivalent of sd_bus_message_append().
728 * sd-boot gained support for validating images using SHIM/MOK.
730 * The SMACK code learnt support for "onlycap".
732 * systemd-mount --umount is now much smarter in figuring out how to
733 properly unmount a device given its mount or device path.
735 * The code to call libnss_dns as a fallback from libnss_resolve when
736 the communication with systemd-resolved fails was removed. This
737 fallback was redundant and interfered with the [!UNAVAIL=return]
738 suffix. See nss-resolve(8) for the recommended configuration.
740 * systemd-logind may now be restarted without losing state. It stores
741 the file descriptors for devices it manages in the system manager
742 using the FDSTORE= mechanism. Please note that further changes in
743 other components may be required to make use of this (for example
744 Xorg has code to listen for stops of systemd-logind and terminate
745 itself when logind is stopped or restarted, in order to avoid using
746 stale file descriptors for graphical devices, which is now
747 counterproductive and must be reverted in order for restarts of
748 systemd-logind to be safe. See
749 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
751 * All kernel install plugins are called with the environment variable
752 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
753 /etc/machine-id. If the file is missing or empty, the variable is
754 empty and BOOT_DIR_ABS is the path of a temporary directory which is
755 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
756 is empty, all plugins should not put anything in BOOT_DIR_ABS.
758 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
759 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
760 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
761 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
762 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
763 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
764 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
765 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
766 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
767 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
768 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
769 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
770 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
771 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
772 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
773 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
774 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
775 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
776 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
777 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
778 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
779 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
780 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
781 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
782 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
783 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
784 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
785 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
786 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
793 * The "hybrid" control group mode has been modified to improve
794 compatibility with "legacy" cgroups-v1 setups. Specifically, the
795 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
796 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
797 cgroups-v1 hierarchy), the only externally visible change being that
798 the cgroups-v2 hierarchy is also mounted, to
799 /sys/fs/cgroup/unified. This should provide a large degree of
800 compatibility with "legacy" cgroups-v1, while taking benefit of the
801 better management capabilities of cgroups-v2.
803 * The default control group setup mode may be selected both a boot-time
804 via a set of kernel command line parameters (specifically:
805 systemd.unified_cgroup_hierarchy= and
806 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
807 default selected on the configure command line
808 (--with-default-hierarchy=). The upstream default is "hybrid"
809 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
810 this will change in a future systemd version to be "unified" (pure
811 cgroups-v2 mode). The third option for the compile time option is
812 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
813 distributions to default to "hybrid" mode for release distributions,
814 starting with v233. We recommend "unified" for development
815 distributions (specifically: distributions such as Fedora's rawhide)
816 as that's where things are headed in the long run. Use "legacy" for
817 greatest stability and compatibility only.
819 * Note one current limitation of "unified" and "hybrid" control group
820 setup modes: the kernel currently does not permit the systemd --user
821 instance (i.e. unprivileged code) to migrate processes between two
822 disconnected cgroup subtrees, even if both are managed and owned by
823 the user. This effectively means "systemd-run --user --scope" doesn't
824 work when invoked from outside of any "systemd --user" service or
825 scope. Specifically, it is not supported from session scopes. We are
826 working on fixing this in a future systemd version. (See #3388 for
827 further details about this.)
829 * DBus policy files are now installed into /usr rather than /etc. Make
830 sure your system has dbus >= 1.9.18 running before upgrading to this
831 version, or override the install path with --with-dbuspolicydir= .
833 * All python scripts shipped with systemd (specifically: the various
834 tests written in Python) now require Python 3.
836 * systemd unit tests can now run standalone (without the source or
837 build directories), and can be installed into /usr/lib/systemd/tests/
838 with 'make install-tests'.
840 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
841 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
844 * Support for the %c, %r, %R specifiers in unit files has been
845 removed. Specifiers are not supposed to be dependent on configuration
846 in the unit file itself (so that they resolve the same regardless
847 where used in the unit files), but these specifiers were influenced
848 by the Slice= option.
850 * The shell invoked by debug-shell.service now defaults to /bin/sh in
851 all cases. If distributions want to use a different shell for this
852 purpose (for example Fedora's /sbin/sushell) they need to specify
853 this explicitly at configure time using --with-debug-shell=.
855 * The confirmation spawn prompt has been reworked to offer the
858 (c)ontinue, proceed without asking anymore
859 (D)ump, show the state of the unit
860 (f)ail, don't execute the command and pretend it failed
862 (i)nfo, show a short summary of the unit
863 (j)obs, show jobs that are in progress
864 (s)kip, don't execute the command and pretend it succeeded
865 (y)es, execute the command
867 The 'n' choice for the confirmation spawn prompt has been removed,
868 because its meaning was confusing.
870 The prompt may now also be redirected to an alternative console by
871 specifying the console as parameter to systemd.confirm_spawn=.
873 * Services of Type=notify require a READY=1 notification to be sent
874 during startup. If no such message is sent, the service now fails,
875 even if the main process exited with a successful exit code.
877 * Services that fail to start up correctly now always have their
878 ExecStopPost= commands executed. Previously, they'd enter "failed"
879 state directly, without executing these commands.
881 * The option MulticastDNS= of network configuration files has acquired
882 an actual implementation. With MulticastDNS=yes a host can resolve
883 names of remote hosts and reply to mDNS A and AAAA requests.
885 * When units are about to be started an additional check is now done to
886 ensure that all dependencies of type BindsTo= (when used in
887 combination with After=) have been started.
889 * systemd-analyze gained a new verb "syscall-filter" which shows which
890 system call groups are defined for the SystemCallFilter= unit file
891 setting, and which system calls they contain.
893 * A new system call filter group "@filesystem" has been added,
894 consisting of various file system related system calls. Group
895 "@reboot" has been added, covering reboot, kexec and shutdown related
896 calls. Finally, group "@swap" has been added covering swap
897 configuration related calls.
899 * A new unit file option RestrictNamespaces= has been added that may be
900 used to restrict access to the various process namespace types the
901 Linux kernel provides. Specifically, it may be used to take away the
902 right for a service unit to create additional file system, network,
903 user, and other namespaces. This sandboxing option is particularly
904 relevant due to the high amount of recently discovered namespacing
905 related vulnerabilities in the kernel.
907 * systemd-udev's .link files gained support for a new AutoNegotiation=
908 setting for configuring Ethernet auto-negotiation.
910 * systemd-networkd's .network files gained support for a new
911 ListenPort= setting in the [DHCP] section to explicitly configure the
912 UDP client port the DHCP client shall listen on.
914 * .network files gained a new Unmanaged= boolean setting for explicitly
915 excluding one or more interfaces from management by systemd-networkd.
917 * The systemd-networkd ProxyARP= option has been renamed to
918 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
919 renamed to ReduceARPProxy=. The old names continue to be available
922 * systemd-networkd gained support for configuring IPv6 Proxy NDP
923 addresses via the new IPv6ProxyNDPAddress= .network file setting.
925 * systemd-networkd's bonding device support gained support for two new
926 configuration options ActiveSlave= and PrimarySlave=.
928 * The various options in the [Match] section of .network files gained
929 support for negative matching.
931 * New systemd-specific mount options are now understood in /etc/fstab:
933 x-systemd.mount-timeout= may be used to configure the maximum
934 permitted runtime of the mount command.
936 x-systemd.device-bound may be set to bind a mount point to its
937 backing device unit, in order to automatically remove a mount point
938 if its backing device is unplugged. This option may also be
939 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
940 on the block device, which is now automatically set for all CDROM
941 drives, so that mounted CDs are automatically unmounted when they are
942 removed from the drive.
944 x-systemd.after= and x-systemd.before= may be used to explicitly
945 order a mount after or before another unit or mount point.
947 * Enqueued start jobs for device units are now automatically garbage
948 collected if there are no jobs waiting for them anymore.
950 * systemctl list-jobs gained two new switches: with --after, for every
951 queued job the jobs it's waiting for are shown; with --before the
952 jobs which it's blocking are shown.
954 * systemd-nspawn gained support for ephemeral boots from disk images
955 (or in other words: --ephemeral and --image= may now be
956 combined). Moreover, ephemeral boots are now supported for normal
957 directories, even if the backing file system is not btrfs. Of course,
958 if the file system does not support file system snapshots or
959 reflinks, the initial copy operation will be relatively expensive, but
960 this should still be suitable for many use cases.
962 * Calendar time specifications in .timer units now support
963 specifications relative to the end of a month by using "~" instead of
964 "-" as separator between month and day. For example, "*-02~03" means
965 "the third last day in February". In addition a new syntax for
966 repeated events has been added using the "/" character. For example,
967 "9..17/2:00" means "every two hours from 9am to 5pm".
969 * systemd-socket-proxyd gained a new parameter --connections-max= for
970 configuring the maximum number of concurrent connections.
972 * sd-id128 gained a new API for generating unique IDs for the host in a
973 way that does not leak the machine ID. Specifically,
974 sd_id128_get_machine_app_specific() derives an ID based on the
975 machine ID a in well-defined, non-reversible, stable way. This is
976 useful whenever an identifier for the host is needed but where the
977 identifier shall not be useful to identify the system beyond the
978 scope of the application itself. (Internally this uses HMAC-SHA256 as
979 keyed hash function using the machine ID as input.)
981 * NotifyAccess= gained a new supported value "exec". When set
982 notifications are accepted from all processes systemd itself invoked,
983 including all control processes.
985 * .nspawn files gained support for defining overlay mounts using the
986 Overlay= and OverlayReadOnly= options. Previously this functionality
987 was only available on the systemd-nspawn command line.
989 * systemd-nspawn's --bind= and --overlay= options gained support for
990 bind/overlay mounts whose source lies within the container tree by
991 prefixing the source path with "+".
993 * systemd-nspawn's --bind= and --overlay= options gained support for
994 automatically allocating a temporary source directory in /var/tmp
995 that is removed when the container dies. Specifically, if the source
996 directory is specified as empty string this mechanism is selected. An
997 example usage is --overlay=+/var::/var, which creates an overlay
998 mount based on the original /var contained in the image, overlayed
999 with a temporary directory in the host's /var/tmp. This way changes
1000 to /var are automatically flushed when the container shuts down.
1002 * systemd-nspawn --image= option does now permit raw file system block
1003 devices (in addition to images containing partition tables, as
1006 * The disk image dissection logic in systemd-nspawn gained support for
1007 automatically setting up LUKS encrypted as well as Verity protected
1008 partitions. When a container is booted from an encrypted image the
1009 passphrase is queried at start-up time. When a container with Verity
1010 data is started, the root hash is search in a ".roothash" file
1011 accompanying the disk image (alternatively, pass the root hash via
1012 the new --root-hash= command line option).
1014 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1015 be used to dissect disk images the same way as systemd-nspawn does
1016 it, following the Bootable Partition Specification. It may even be
1017 used to mount disk images with complex partition setups (including
1018 LUKS and Verity partitions) to a local host directory, in order to
1019 inspect them. This tool is not considered public API (yet), and is
1020 thus not installed into /usr/bin. Please do not rely on its
1021 existence, since it might go away or be changed in later systemd
1024 * A new generator "systemd-verity-generator" has been added, similar in
1025 style to "systemd-cryptsetup-generator", permitting automatic setup of
1026 Verity root partitions when systemd boots up. In order to make use of
1027 this your partition setup should follow the Discoverable Partitions
1028 Specification, and the GPT partition ID of the root file system
1029 partition should be identical to the upper 128bit of the Verity root
1030 hash. The GPT partition ID of the Verity partition protecting it
1031 should be the lower 128bit of the Verity root hash. If the partition
1032 image follows this model it is sufficient to specify a single
1033 "roothash=" kernel command line argument to both configure which root
1034 image and verity partition to use as well as the root hash for
1035 it. Note that systemd-nspawn's Verity support follows the same
1036 semantics, meaning that disk images with proper Verity data in place
1037 may be booted in containers with systemd-nspawn as well as on
1038 physical systems via the verity generator. Also note that the "mkosi"
1039 tool available at https://github.com/systemd/mkosi has been updated
1040 to generate Verity protected disk images following this scheme. In
1041 fact, it has been updated to generate disk images that optionally
1042 implement a complete UEFI SecureBoot trust chain, involving a signed
1043 kernel and initrd image that incorporates such a root hash as well as
1044 a Verity-enabled root partition.
1046 * The hardware database (hwdb) udev supports has been updated to carry
1047 accelerometer quirks.
1049 * All system services are now run with a fresh kernel keyring set up
1050 for them. The invocation ID is stored by default in it, thus
1051 providing a safe, non-overridable way to determine the invocation
1054 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1055 options for bind mounting arbitrary paths in a service-specific
1056 way. When these options are used, arbitrary host or service files and
1057 directories may be mounted to arbitrary locations in the service's
1060 * Documentation has been added that lists all of systemd's low-level
1061 environment variables:
1063 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
1065 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1066 whether a specific socket file descriptor matches a specified socket
1069 * systemd-firstboot has been updated to check for the
1070 systemd.firstboot= kernel command line option. It accepts a boolean
1071 and when set to false the first boot questions are skipped.
1073 * systemd-fstab-generator has been updated to check for the
1074 systemd.volatile= kernel command line option, which either takes an
1075 optional boolean parameter or the special value "state". If used the
1076 system may be booted in a "volatile" boot mode. Specifically,
1077 "systemd.volatile" is used, the root directory will be mounted as
1078 tmpfs, and only /usr is mounted from the actual root file system. If
1079 "systemd.volatile=state" is used, the root directory will be mounted
1080 as usual, but /var is mounted as tmpfs. This concept provides similar
1081 functionality as systemd-nspawn's --volatile= option, but provides it
1082 on physical boots. Use this option for implementing stateless
1083 systems, or testing systems with all state and/or configuration reset
1084 to the defaults. (Note though that many distributions are not
1085 prepared to boot up without a populated /etc or /var, though.)
1087 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1088 partitions. Previously it only supported LUKS encrypted partitions
1089 for all other uses, except for the root partition itself.
1091 * Socket units gained support for listening on AF_VSOCK sockets for
1092 communication in virtualized QEMU environments.
1094 * The "configure" script gained a new option --with-fallback-hostname=
1095 for specifying the fallback hostname to use if none is configured in
1096 /etc/hostname. For example, by specifying
1097 --with-fallback-hostname=fedora it is possible to default to a
1098 hostname of "fedora" on pristine installations.
1100 * systemd-cgls gained support for a new --unit= switch for listing only
1101 the control groups of a specific unit. Similar --user-unit= has been
1102 added for listing only the control groups of a specific user unit.
1104 * systemd-mount gained a new --umount switch for unmounting a mount or
1105 automount point (and all mount/automount points below it).
1107 * systemd will now refuse full configuration reloads (via systemctl
1108 daemon-reload and related calls) unless at least 16MiB of free space
1109 are available in /run. This is a safety precaution in order to ensure
1110 that generators can safely operate after the reload completed.
1112 * A new unit file option RootImage= has been added, which has a similar
1113 effect as RootDirectory= but mounts the service's root directory from
1114 a disk image instead of plain directory. This logic reuses the same
1115 image dissection and mount logic that systemd-nspawn already uses,
1116 and hence supports any disk images systemd-nspawn supports, including
1117 those following the Discoverable Partition Specification, as well as
1118 Verity enabled images. This option enables systemd to run system
1119 services directly off disk images acting as resource bundles,
1120 possibly even including full integrity data.
1122 * A new MountAPIVFS= unit file option has been added, taking a boolean
1123 argument. If enabled /proc, /sys and /dev (collectively called the
1124 "API VFS") will be mounted for the service. This is only relevant if
1125 RootDirectory= or RootImage= is used for the service, as these mounts
1126 are of course in place in the host mount namespace anyway.
1128 * systemd-nspawn gained support for a new --pivot-root= switch. If
1129 specified the root directory within the container image is pivoted to
1130 the specified mount point, while the original root disk is moved to a
1131 different place. This option enables booting of ostree images
1132 directly with systemd-nspawn.
1134 * The systemd build scripts will no longer complain if the NTP server
1135 addresses are not changed from the defaults. Google now supports
1136 these NTP servers officially. We still recommend downstreams to
1137 properly register an NTP pool with the NTP pool project though.
1139 * coredumpctl gained a new "--reverse" option for printing the list
1140 of coredumps in reverse order.
1142 * coredumpctl will now show additional information about truncated and
1143 inaccessible coredumps, as well as coredumps that are still being
1144 processed. It also gained a new --quiet switch for suppressing
1145 additional informational message in its output.
1147 * coredumpctl gained support for only showing coredumps newer and/or
1148 older than specific timestamps, using the new --since= and --until=
1149 options, reminiscent of journalctl's options by the same name.
1151 * The systemd-coredump logic has been improved so that it may be reused
1152 to collect backtraces in non-compiled languages, for example in
1153 scripting languages such as Python.
1155 * machinectl will now show the UID shift of local containers, if user
1156 namespacing is enabled for them.
1158 * systemd will now optionally run "environment generator" binaries at
1159 configuration load time. They may be used to add environment
1160 variables to the environment block passed to services invoked. One
1161 user environment generator is shipped by default that sets up
1162 environment variables based on files dropped into /etc/environment.d
1163 and ~/.config/environment.d/.
1165 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1168 * hostnamed has been updated to report a new chassis type of
1169 "convertible" to cover "foldable" laptops that can both act as a
1170 tablet and as a laptop, such as various Lenovo Yoga devices.
1172 Contributions from: Adrián López, Alexander Galanin, Alexander
1173 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1174 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1175 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1176 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1177 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1178 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1179 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1180 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1181 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1182 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1183 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1184 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1185 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1186 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1187 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1188 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1189 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1190 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1191 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1192 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1193 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1194 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1195 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1196 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1197 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1198 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1199 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1202 — Berlin, 2017-03-01
1206 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1207 RestrictAddressFamilies= enabled. These sandboxing options should
1208 generally be compatible with the various external udev call-out
1209 binaries we are aware of, however there may be exceptions, in
1210 particular when exotic languages for these call-outs are used. In
1211 this case, consider turning off these settings locally.
1213 * The new RemoveIPC= option can be used to remove IPC objects owned by
1214 the user or group of a service when that service exits.
1216 * The new ProtectKernelModules= option can be used to disable explicit
1217 load and unload operations of kernel modules by a service. In
1218 addition access to /usr/lib/modules is removed if this option is set.
1220 * ProtectSystem= option gained a new value "strict", which causes the
1221 whole file system tree with the exception of /dev, /proc, and /sys,
1222 to be remounted read-only for a service.
1224 * The new ProtectKernelTunables= option can be used to disable
1225 modification of configuration files in /sys and /proc by a service.
1226 Various directories and files are remounted read-only, so access is
1227 restricted even if the file permissions would allow it.
1229 * The new ProtectControlGroups= option can be used to disable write
1230 access by a service to /sys/fs/cgroup.
1232 * Various systemd services have been hardened with
1233 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1234 RestrictAddressFamilies=.
1236 * Support for dynamically creating users for the lifetime of a service
1237 has been added. If DynamicUser=yes is specified, user and group IDs
1238 will be allocated from the range 61184..65519 for the lifetime of the
1239 service. They can be resolved using the new nss-systemd.so NSS
1240 module. The module must be enabled in /etc/nsswitch.conf. Services
1241 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1242 any resources allocated by the service will be cleaned up when the
1243 service exits. They also have ProtectHome=read-only and
1244 ProtectSystem=strict enabled, so they are not able to make any
1245 permanent modifications to the system.
1247 * The nss-systemd module also always resolves root and nobody, making
1248 it possible to have no /etc/passwd or /etc/group files in minimal
1249 container or chroot environments.
1251 * Services may be started with their own user namespace using the new
1252 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1253 under which the service is running are mapped. All other users are
1256 * Support for the cgroup namespace has been added to systemd-nspawn. If
1257 supported by kernel, the container system started by systemd-nspawn
1258 will have its own view of the cgroup hierarchy. This new behaviour
1259 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1261 * The new MemorySwapMax= option can be used to limit the maximum swap
1262 usage under the unified cgroup hierarchy.
1264 * Support for the CPU controller in the unified cgroup hierarchy has
1265 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1266 options. This controller requires out-of-tree patches for the kernel
1267 and the support is provisional.
1269 * Mount and automount units may now be created transiently
1270 (i.e. dynamically at runtime via the bus API, instead of requiring
1271 unit files in the file system).
1273 * systemd-mount is a new tool which may mount file systems – much like
1274 mount(8), optionally pulling in additional dependencies through
1275 transient .mount and .automount units. For example, this tool
1276 automatically runs fsck on a backing block device before mounting,
1277 and allows the automount logic to be used dynamically from the
1278 command line for establishing mount points. This tool is particularly
1279 useful when dealing with removable media, as it will ensure fsck is
1280 run – if necessary – before the first access and that the file system
1281 is quickly unmounted after each access by utilizing the automount
1282 logic. This maximizes the chance that the file system on the
1283 removable media stays in a clean state, and if it isn't in a clean
1284 state is fixed automatically.
1286 * LazyUnmount=yes option for mount units has been added to expose the
1287 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1290 * /efi will be used as the mount point of the EFI boot partition, if
1291 the directory is present, and the mount point was not configured
1292 through other means (e.g. fstab). If /efi directory does not exist,
1293 /boot will be used as before. This makes it easier to automatically
1294 mount the EFI partition on systems where /boot is used for something
1297 * When operating on GPT disk images for containers, systemd-nspawn will
1298 now mount the ESP to /boot or /efi according to the same rules as PID
1299 1 running on a host. This allows tools like "bootctl" to operate
1300 correctly within such containers, in order to make container images
1301 bootable on physical systems.
1303 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1305 * Two new user session targets have been added to support running
1306 graphical sessions under the systemd --user instance:
1307 graphical-session.target and graphical-session-pre.target. See
1308 systemd.special(7) for a description of how those targets should be
1311 * The vconsole initialization code has been significantly reworked to
1312 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1313 support unicode keymaps. Font and keymap configuration will now be
1314 copied to all allocated virtual consoles.
1316 * FreeBSD's bhyve virtualization is now detected.
1318 * Information recorded in the journal for core dumps now includes the
1319 contents of /proc/mountinfo and the command line of the process at
1320 the top of the process hierarchy (which is usually the init process
1323 * systemd-journal-gatewayd learned the --directory= option to serve
1324 files from the specified location.
1326 * journalctl --root=… can be used to peruse the journal in the
1327 /var/log/ directories inside of a container tree. This is similar to
1328 the existing --machine= option, but does not require the container to
1331 * The hardware database has been extended to support
1332 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1335 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1336 specify the click rate for mice which include a horizontal wheel with
1337 a click rate that is different than the one for the vertical wheel.
1339 * systemd-run gained a new --wait option that makes service execution
1340 synchronous. (Specifically, the command will not return until the
1341 specified service binary exited.)
1343 * systemctl gained a new --wait option that causes the start command to
1344 wait until the units being started have terminated again.
1346 * A new journal output mode "short-full" has been added which displays
1347 timestamps with abbreviated English day names and adds a timezone
1348 suffix. Those timestamps include more information than the default
1349 "short" output mode, and can be passed directly to journalctl's
1350 --since= and --until= options.
1352 * /etc/resolv.conf will be bind-mounted into containers started by
1353 systemd-nspawn, if possible, so any changes to resolv.conf contents
1354 are automatically propagated to the container.
1356 * The number of instances for socket-activated services originating
1357 from a single IP address can be limited with
1358 MaxConnectionsPerSource=, extending the existing setting of
1361 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1364 * .netdev and .network configuration can now be extended through
1367 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1368 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1369 can be enabled and disabled using the new UDPSegmentationOffload=,
1370 TCPSegmentationOffload=, GenericSegmentationOffload=,
1371 GenericReceiveOffload=, LargeReceiveOffload= options in the
1372 [Link] section of .link files.
1374 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1375 Port VLAN ID can be configured for bridge devices using the new STP=,
1376 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1377 section of .netdev files.
1379 * The route table to which routes received over DHCP or RA should be
1380 added can be configured with the new RouteTable= option in the [DHCP]
1381 and [IPv6AcceptRA] sections of .network files.
1383 * The Address Resolution Protocol can be disabled on links managed by
1384 systemd-networkd using the ARP=no setting in the [Link] section of
1387 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1388 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1389 encode information about the result and exit codes of the current
1390 service runtime cycle.
1392 * systemd-sysctl will now configure kernel parameters in the order
1393 they occur in the configuration files. This matches what sysctl
1394 has been traditionally doing.
1396 * kernel-install "plugins" that are executed to perform various
1397 tasks after a new kernel is added and before an old one is removed
1398 can now return a special value to terminate the procedure and
1399 prevent any later plugins from running.
1401 * Journald's SplitMode=login setting has been deprecated. It has been
1402 removed from documentation, and its use is discouraged. In a future
1403 release it will be completely removed, and made equivalent to current
1404 default of SplitMode=uid.
1406 * Storage=both option setting in /etc/systemd/coredump.conf has been
1407 removed. With fast LZ4 compression storing the core dump twice is not
1410 * The --share-system systemd-nspawn option has been replaced with an
1411 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1412 this functionality is discouraged. In addition the variables
1413 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1414 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1415 individual namespaces.
1417 * "machinectl list" now shows the IP address of running containers in
1418 the output, as well as OS release information.
1420 * "loginctl list" now shows the TTY of each session in the output.
1422 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1423 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1424 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1425 tracking objects in a "recursive" mode, where a single client can be
1426 counted multiple times, if it takes multiple references.
1428 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1429 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1430 process using sd-bus automatically exit if the bus connection is
1433 * Bus clients of the service manager may now "pin" loaded units into
1434 memory, by taking an explicit reference on them. This is useful to
1435 ensure the client can retrieve runtime data about the service even
1436 after the service completed execution. Taking such a reference is
1437 available only for privileged clients and should be helpful to watch
1438 running services in a race-free manner, and in particular collect
1439 information about exit statuses and results.
1441 * The nss-resolve module has been changed to strictly return UNAVAIL
1442 when communication via D-Bus with resolved failed, and NOTFOUND when
1443 a lookup completed but was negative. This means it is now possible to
1444 neatly configure fallbacks using nsswitch.conf result checking
1445 expressions. Taking benefit of this, the new recommended
1446 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1448 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1450 * A new setting CtrlAltDelBurstAction= has been added to
1451 /etc/systemd/system.conf which may be used to configure the precise
1452 behaviour if the user on the console presses Ctrl-Alt-Del more often
1453 than 7 times in 2s. Previously this would unconditionally result in
1454 an expedited, immediate reboot. With this new setting the precise
1455 operation may be configured in more detail, and also turned off
1458 * In .netdev files two new settings RemoteChecksumTx= and
1459 RemoteChecksumRx= are now understood that permit configuring the
1460 remote checksumming logic for VXLAN networks.
1462 * The service manager learnt a new "invocation ID" concept for invoked
1463 services. Each runtime cycle of a service will get a new invocation
1464 ID (a 128bit random UUID) assigned that identifies the current
1465 run of the service uniquely and globally. A new invocation ID
1466 is generated each time a service starts up. The journal will store
1467 the invocation ID of a service along with any logged messages, thus
1468 making the invocation ID useful for matching the online runtime of a
1469 service with the offline log data it generated in a safe way without
1470 relying on synchronized timestamps. In many ways this new service
1471 invocation ID concept is similar to the kernel's boot ID concept that
1472 uniquely and globally identifies the runtime of each boot. The
1473 invocation ID of a service is passed to the service itself via an
1474 environment variable ($INVOCATION_ID). A new bus call
1475 GetUnitByInvocationID() has been added that is similar to GetUnit()
1476 but instead of retrieving the bus path for a unit by its name
1477 retrieves it by its invocation ID. The returned path is valid only as
1478 long as the passed invocation ID is current.
1480 * systemd-resolved gained a new "DNSStubListener" setting in
1481 resolved.conf. It either takes a boolean value or the special values
1482 "udp" and "tcp", and configures whether to enable the stub DNS
1483 listener on 127.0.0.53:53.
1485 * IP addresses configured via networkd may now carry additional
1486 configuration settings supported by the kernel. New options include:
1487 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1488 PrefixRoute=, AutoJoin=.
1490 * The PAM configuration fragment file for "user@.service" shipped with
1491 systemd (i.e. the --user instance of systemd) has been stripped to
1492 the minimum necessary to make the system boot. Previously, it
1493 contained Fedora-specific stanzas that did not apply to other
1494 distributions. It is expected that downstream distributions add
1495 additional configuration lines, matching their needs to this file,
1496 using it only as rough template of what systemd itself needs. Note
1497 that this reduced fragment does not even include an invocation of
1498 pam_limits which most distributions probably want to add, even though
1499 systemd itself does not need it. (There's also the new build time
1500 option --with-pamconfdir=no to disable installation of the PAM
1503 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1504 capability is now also dropped from its set (in addition to
1505 CAP_SYS_MKNOD as before).
1507 * In service unit files it is now possible to connect a specific named
1508 file descriptor with stdin/stdout/stdout of an executed service. The
1509 name may be specified in matching .socket units using the
1510 FileDescriptorName= setting.
1512 * A number of journal settings may now be configured on the kernel
1513 command line. Specifically, the following options are now understood:
1514 systemd.journald.max_level_console=,
1515 systemd.journald.max_level_store=,
1516 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1517 systemd.journald.max_level_wall=.
1519 * "systemctl is-enabled --full" will now show by which symlinks a unit
1520 file is enabled in the unit dependency tree.
1522 * Support for VeraCrypt encrypted partitions has been added to the
1523 "cryptsetup" logic and /etc/crypttab.
1525 * systemd-detect-virt gained support for a new --private-users switch
1526 that checks whether the invoking processes are running inside a user
1527 namespace. Similar, a new special value "private-users" for the
1528 existing ConditionVirtualization= setting has been added, permitting
1529 skipping of specific units in user namespace environments.
1531 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1532 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1533 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1534 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1535 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1536 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1537 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1538 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1539 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1540 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1541 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1542 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1543 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1544 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1545 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1546 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1547 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1548 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1549 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1550 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1551 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1552 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1553 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1554 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1555 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1556 Jędrzejewski-Szmek, Zeal Jagannatha
1558 — Santa Fe, 2016-11-03
1562 * In service units the various ExecXYZ= settings have been extended
1563 with an additional special character as first argument of the
1564 assigned value: if the character '+' is used the specified command
1565 line it will be run with full privileges, regardless of User=,
1566 Group=, CapabilityBoundingSet= and similar options. The effect is
1567 similar to the existing PermissionsStartOnly= option, but allows
1568 configuration of this concept for each executed command line
1571 * Services may now alter the service watchdog timeout at runtime by
1572 sending a WATCHDOG_USEC= message via sd_notify().
1574 * MemoryLimit= and related unit settings now optionally take percentage
1575 specifications. The percentage is taken relative to the amount of
1576 physical memory in the system (or in case of containers, the assigned
1577 amount of memory). This allows scaling service resources neatly with
1578 the amount of RAM available on the system. Similarly, systemd-logind's
1579 RuntimeDirectorySize= option now also optionally takes percentage
1582 * In similar fashion TasksMax= takes percentage values now, too. The
1583 value is taken relative to the configured maximum number of processes
1584 on the system. The per-service task maximum has been changed to 15%
1585 using this functionality. (Effectively this is an increase of 512 →
1586 4915 for service units, given the kernel's default pid_max setting.)
1588 * Calendar time specifications in .timer units now understand a ".."
1589 syntax for time ranges. Example: "4..7:10" may now be used for
1590 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1593 * The InaccessableDirectories=, ReadOnlyDirectories= and
1594 ReadWriteDirectories= unit file settings have been renamed to
1595 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1596 applied to all kinds of file nodes, and not just directories, with
1597 the exception of symlinks. Specifically these settings may now be
1598 used on block and character device nodes, UNIX sockets and FIFOS as
1599 well as regular files. The old names of these settings remain
1600 available for compatibility.
1602 * systemd will now log about all service processes it kills forcibly
1603 (using SIGKILL) because they remained after the clean shutdown phase
1604 of the service completed. This should help identifying services that
1605 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1606 systemd-logind's configuration a similar log message is generated for
1607 processes killed at the end of each session due to this setting.
1609 * systemd will now set the $JOURNAL_STREAM environment variable for all
1610 services whose stdout/stderr are connected to the Journal (which
1611 effectively means by default: all services). The variable contains
1612 the device and inode number of the file descriptor used for
1613 stdout/stderr. This may be used by invoked programs to detect whether
1614 their stdout/stderr is connected to the Journal, in which case they
1615 can switch over to direct Journal communication, thus being able to
1616 pass extended, structured metadata along with their log messages. As
1617 one example, this is now used by glib's logging primitives.
1619 * When using systemd's default tmp.mount unit for /tmp, the mount point
1620 will now be established with the "nosuid" and "nodev" options. This
1621 avoids privilege escalation attacks that put traps and exploits into
1622 /tmp. However, this might cause problems if you e. g. put container
1623 images or overlays into /tmp; if you need this, override tmp.mount's
1624 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1627 * systemd now supports the "memory" cgroup controller also on
1630 * The systemd-cgtop tool now optionally takes a control group path as
1631 command line argument. If specified, the control group list shown is
1632 limited to subgroups of that group.
1634 * The SystemCallFilter= unit file setting gained support for
1635 pre-defined, named system call filter sets. For example
1636 SystemCallFilter=@clock is now an effective way to make all clock
1637 changing-related system calls unavailable to a service. A number of
1638 similar pre-defined groups are defined. Writing system call filters
1639 for system services is simplified substantially with this new
1640 concept. Accordingly, all of systemd's own, long-running services now
1641 enable system call filtering based on this, by default.
1643 * A new service setting MemoryDenyWriteExecute= has been added, taking
1644 a boolean value. If turned on, a service may no longer create memory
1645 mappings that are writable and executable at the same time. This
1646 enhances security for services where this is enabled as it becomes
1647 harder to dynamically write and then execute memory in exploited
1648 service processes. This option has been enabled for all of systemd's
1649 own long-running services.
1651 * A new RestrictRealtime= service setting has been added, taking a
1652 boolean argument. If set the service's processes may no longer
1653 acquire realtime scheduling. This improves security as realtime
1654 scheduling may otherwise be used to easily freeze the system.
1656 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1657 value. This may be used for requesting that the system manager inside
1658 of the container reports start-up completion to nspawn which then
1659 propagates this notification further to the service manager
1660 supervising nspawn itself. A related option NotifyReady= in .nspawn
1661 files has been added too. This functionality allows ordering of the
1662 start-up of multiple containers using the usual systemd ordering
1665 * machinectl gained a new command "stop" that is an alias for
1668 * systemd-resolved gained support for contacting DNS servers on
1669 link-local IPv6 addresses.
1671 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1672 its caches. A method call for requesting the same operation has been
1673 added to the bus API too, and is made available via "systemd-resolve
1676 * systemd-resolve gained a new --status switch. If passed a brief
1677 summary of the used DNS configuration with per-interface information
1680 * resolved.conf gained a new Cache= boolean option, defaulting to
1681 on. If turned off local DNS caching is disabled. This comes with a
1682 performance penalty in particular when DNSSEC is enabled. Note that
1683 resolved disables its internal caching implicitly anyway, when the
1684 configured DNS server is on a host-local IP address such as ::1 or
1685 127.0.0.1, thus automatically avoiding double local caching.
1687 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1688 for DNS requests. This improves compatibility with local programs
1689 that do not use the libc NSS or systemd-resolved's bus APIs for name
1690 resolution. This minimal DNS service is only available to local
1691 programs and does not implement the full DNS protocol, but enough to
1692 cover local DNS clients. A new, static resolv.conf file, listing just
1693 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1694 now recommended to make /etc/resolv.conf a symlink to this file in
1695 order to route all DNS lookups to systemd-resolved, regardless if
1696 done via NSS, the bus API or raw DNS packets. Note that this local
1697 DNS service is not as fully featured as the libc NSS or
1698 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1699 used to deliver link-local address information (as this implies
1700 sending a local interface index along), LLMNR/mDNS support via this
1701 interface is severely restricted. It is thus strongly recommended for
1702 all applications to use the libc NSS API or native systemd-resolved
1705 * systemd-networkd's bridge support learned a new setting
1706 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1707 in .network files has been added for configuring VLAN bridging in
1708 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1710 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1711 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1712 now be acquired without relying on DHCPv6. Two new options
1713 UseDomains= and UseDNS= have been added to configure this behaviour.
1715 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1716 renamed IPv6AcceptRA=, without altering its behaviour. The old
1717 setting name remains available for compatibility reasons.
1719 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1720 Key=, InputKey= and OutputKey=.
1722 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1723 interface configuration.
1725 * "systemctl edit" may now be used to create new unit files by
1726 specifying the --force switch.
1728 * sd-event gained a new function sd_event_get_iteration() for
1729 requesting the current iteration counter of the event loop. It starts
1730 at zero and is increased by one with each event loop iteration.
1732 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1733 file. It can be used in lieu of %systemd_requires in packages which
1734 don't use any systemd functionality and are intended to be installed
1735 in minimal containers without systemd present. This macro provides
1736 ordering dependencies to ensure that if the package is installed in
1737 the same rpm transaction as systemd, systemd will be installed before
1738 the scriptlets for the package are executed, allowing unit presets
1741 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1742 been added to simplify packaging of generators.
1744 * The os-release file gained VERSION_CODENAME field for the
1745 distribution nickname (e.g. VERSION_CODENAME=woody).
1747 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1748 can be set to disable parsing of metadata and the creation
1749 of persistent symlinks for that device.
1751 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1752 to make them available to logged-in users has been reverted.
1754 * Much of the common code of the various systemd components is now
1755 built into an internal shared library libsystemd-shared-231.so
1756 (incorporating the systemd version number in the name, to be updated
1757 with future releases) that the components link to. This should
1758 decrease systemd footprint both in memory during runtime and on
1759 disk. Note that the shared library is not for public use, and is
1760 neither API not ABI stable, but is likely to change with every new
1761 released update. Packagers need to make sure that binaries
1762 linking to libsystemd-shared.so are updated in step with the
1765 * Configuration for "mkosi" is now part of the systemd
1766 repository. mkosi is a tool to easily build legacy-free OS images,
1767 and is available on github: https://github.com/systemd/mkosi. If
1768 "mkosi" is invoked in the build tree a new raw OS image is generated
1769 incorporating the systemd sources currently being worked on and a
1770 clean, fresh distribution installation. The generated OS image may be
1771 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1772 UEFI PC. This functionality is particularly useful to easily test
1773 local changes made to systemd in a pristine, defined environment. See
1774 HACKING for details.
1776 * configure learned the --with-support-url= option to specify the
1777 distribution's bugtracker.
1779 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1780 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1781 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1782 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1783 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1784 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1785 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1786 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1787 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1788 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1789 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1790 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1791 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1792 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1793 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1794 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1795 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1796 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1797 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1799 — Berlin, 2016-07-25
1803 * DNSSEC is now turned on by default in systemd-resolved (in
1804 "allow-downgrade" mode), but may be turned off during compile time by
1805 passing "--with-default-dnssec=no" to "configure" (and of course,
1806 during runtime with DNSSEC= in resolved.conf). We recommend
1807 downstreams to leave this on at least during development cycles and
1808 report any issues with the DNSSEC logic upstream. We are very
1809 interested in collecting feedback about the DNSSEC validator and its
1810 limitations in the wild. Note however, that DNSSEC support is
1811 probably nothing downstreams should turn on in stable distros just
1812 yet, as it might create incompatibilities with a few DNS servers and
1813 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1814 automatically whenever we detect such incompatible setups, but there
1815 might be systems we do not cover yet. Hence: please help us testing
1816 the DNSSEC code, leave this on where you can, report back, but then
1817 again don't consider turning this on in your stable, LTS or
1818 production release just yet. (Note that you have to enable
1819 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1820 and its DNSSEC mode for host name resolution from local
1823 * systemd-resolve conveniently resolves DANE records with the --tlsa
1824 option and OPENPGPKEY records with the --openpgp option. It also
1825 supports dumping raw DNS record data via the new --raw= switch.
1827 * systemd-logind will now by default terminate user processes that are
1828 part of the user session scope unit (session-XX.scope) when the user
1829 logs out. This behavior is controlled by the KillUserProcesses=
1830 setting in logind.conf, and the previous default of "no" is now
1831 changed to "yes". This means that user sessions will be properly
1832 cleaned up after, but additional steps are necessary to allow
1833 intentionally long-running processes to survive logout.
1835 While the user is logged in at least once, user@.service is running,
1836 and any service that should survive the end of any individual login
1837 session can be started at a user service or scope using systemd-run.
1838 systemd-run(1) man page has been extended with an example which shows
1839 how to run screen in a scope unit underneath user@.service. The same
1840 command works for tmux.
1842 After the user logs out of all sessions, user@.service will be
1843 terminated too, by default, unless the user has "lingering" enabled.
1844 To effectively allow users to run long-term tasks even if they are
1845 logged out, lingering must be enabled for them. See loginctl(1) for
1846 details. The default polkit policy was modified to allow users to
1847 set lingering for themselves without authentication.
1849 Previous defaults can be restored at compile time by the
1850 --without-kill-user-processes option to "configure".
1852 * systemd-logind gained new configuration settings SessionsMax= and
1853 InhibitorsMax=, both with a default of 8192. It will not register new
1854 user sessions or inhibitors above this limit.
1856 * systemd-logind will now reload configuration on SIGHUP.
1858 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1859 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1860 enable. Also, support for the "io" cgroup controller in the unified
1861 hierarchy has been added, so that the "memory", "pids" and "io" are
1862 now the controllers that are supported on the unified hierarchy.
1864 WARNING: it is not possible to use previous systemd versions with
1865 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1866 is necessary to also update systemd in the initramfs if using the
1867 unified hierarchy. An updated SELinux policy is also required.
1869 * LLDP support has been extended, and both passive (receive-only) and
1870 active (sender) modes are supported. Passive mode ("routers-only") is
1871 enabled by default in systemd-networkd. Active LLDP mode is enabled
1872 by default for containers on the internal network. The "networkctl
1873 lldp" command may be used to list information gathered. "networkctl
1874 status" will also show basic LLDP information on connected peers now.
1876 * The IAID and DUID unique identifier sent in DHCP requests may now be
1877 configured for the system and each .network file managed by
1878 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1880 * systemd-networkd gained support for configuring proxy ARP support for
1881 each interface, via the ProxyArp= setting in .network files. It also
1882 gained support for configuring the multicast querier feature of
1883 bridge devices, via the new MulticastQuerier= setting in .netdev
1884 files. Similarly, snooping on the IGMP traffic can be controlled
1885 via the new setting MulticastSnooping=.
1887 A new setting PreferredLifetime= has been added for addresses
1888 configured in .network file to configure the lifetime intended for an
1891 The systemd-networkd DHCP server gained the option EmitRouter=, which
1892 defaults to yes, to configure whether the DHCP Option 3 (Router)
1895 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1896 systemd-socket-activate and installed into /usr/bin. It is now fully
1899 * systemd-journald now uses separate threads to flush changes to disk
1900 when closing journal files, thus reducing impact of slow disk I/O on
1901 logging performance.
1903 * The sd-journal API gained two new calls
1904 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1905 can be used to open journal files using file descriptors instead of
1906 file or directory paths. sd_journal_open_container() has been
1907 deprecated, sd_journal_open_directory_fd() should be used instead
1908 with the flag SD_JOURNAL_OS_ROOT.
1910 * journalctl learned a new output mode "-o short-unix" that outputs log
1911 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1912 UTC). It also gained support for a new --no-hostname setting to
1913 suppress the hostname column in the family of "short" output modes.
1915 * systemd-ask-password now optionally skips printing of the password to
1916 stdout with --no-output which can be useful in scripts.
1918 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1919 (devices tagged with ID_MAKER_TOOL) are now tagged with
1920 "uaccess" and are available to logged in users.
1922 * The DeviceAllow= unit setting now supports specifiers (with "%").
1924 * "systemctl show" gained a new --value switch, which allows print a
1925 only the contents of a specific unit property, without also printing
1926 the property's name. Similar support was added to "show*" verbs
1927 of loginctl and machinectl that output "key=value" lists.
1929 * A new unit type "generated" was added for files dynamically generated
1930 by generator tools. Similarly, a new unit type "transient" is used
1931 for unit files created using the runtime API. "systemctl enable" will
1932 refuse to operate on such files.
1934 * A new command "systemctl revert" has been added that may be used to
1935 revert to the vendor version of a unit file, in case local changes
1936 have been made by adding drop-ins or overriding the unit file.
1938 * "machinectl clean" gained a new verb to automatically remove all or
1939 just hidden container images.
1941 * systemd-tmpfiles gained support for a new line type "e" for emptying
1942 directories, if they exist, without creating them if they don't.
1944 * systemd-nspawn gained support for automatically patching the UID/GIDs
1945 of the owners and the ACLs of all files and directories in a
1946 container tree to match the UID/GID user namespacing range selected
1947 for the container invocation. This mode is enabled via the new
1948 --private-users-chown switch. It also gained support for
1949 automatically choosing a free, previously unused UID/GID range when
1950 starting a container, via the new --private-users=pick setting (which
1951 implies --private-users-chown). Together, these options for the first
1952 time make user namespacing for nspawn containers fully automatic and
1953 thus deployable. The systemd-nspawn@.service template unit file has
1954 been changed to use this functionality by default.
1956 * systemd-nspawn gained a new --network-zone= switch, that allows
1957 creating ad-hoc virtual Ethernet links between multiple containers,
1958 that only exist as long as at least one container referencing them is
1959 running. This allows easy connecting of multiple containers with a
1960 common link that implements an Ethernet broadcast domain. Each of
1961 these network "zones" may be named relatively freely by the user, and
1962 may be referenced by any number of containers, but each container may
1963 only reference one of these "zones". On the lower level, this is
1964 implemented by an automatically managed bridge network interface for
1965 each zone, that is created when the first container referencing its
1966 zone is created and removed when the last one referencing its zone
1969 * The default start timeout may now be configured on the kernel command
1970 line via systemd.default_timeout_start_sec=. It was already
1971 configurable via the DefaultTimeoutStartSec= option in
1972 /etc/systemd/system.conf.
1974 * Socket units gained a new TriggerLimitIntervalSec= and
1975 TriggerLimitBurst= setting to configure a limit on the activation
1976 rate of the socket unit.
1978 * The LimitNICE= setting now optionally takes normal UNIX nice values
1979 in addition to the raw integer limit value. If the specified
1980 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1981 value is understood as UNIX nice value. If not prefixed like this it
1982 is understood as raw RLIMIT_NICE limit.
1984 * Note that the effect of the PrivateDevices= unit file setting changed
1985 slightly with this release: the per-device /dev file system will be
1986 mounted read-only from this version on, and will have "noexec"
1987 set. This (minor) change of behavior might cause some (exceptional)
1988 legacy software to break, when PrivateDevices=yes is set for its
1989 service. Please leave PrivateDevices= off if you run into problems
1992 * systemd-bootchart has been split out to a separate repository:
1993 https://github.com/systemd/systemd-bootchart
1995 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1996 merged into the kernel in its current form.
1998 * The compatibility libraries libsystemd-daemon.so,
1999 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2000 which have been deprecated since systemd-209 have been removed along
2001 with the corresponding pkg-config files. All symbols provided by
2002 those libraries are provided by libsystemd.so.
2004 * The Capabilities= unit file setting has been removed (it is ignored
2005 for backwards compatibility). AmbientCapabilities= and
2006 CapabilityBoundingSet= should be used instead.
2008 * A new special target has been added, initrd-root-device.target,
2009 which creates a synchronization point for dependencies of the root
2010 device in early userspace. Initramfs builders must ensure that this
2011 target is now included in early userspace.
2013 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2014 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2015 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2016 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2017 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2018 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2019 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2020 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2021 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2022 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2023 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2024 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2025 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2026 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2027 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2028 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2029 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2030 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2031 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2032 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2033 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2034 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2035 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2036 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2037 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2040 — Fairfax, 2016-05-21
2044 * The systemd-resolved DNS resolver service has gained a substantial
2045 set of new features, most prominently it may now act as a DNSSEC
2046 validating stub resolver. DNSSEC mode is currently turned off by
2047 default, but is expected to be turned on by default in one of the
2048 next releases. For now, we invite everybody to test the DNSSEC logic
2049 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2050 service also gained a full set of D-Bus interfaces, including calls
2051 to configure DNS and DNSSEC settings per link (for use by external
2052 network management software). systemd-resolved and systemd-networkd
2053 now distinguish between "search" and "routing" domains. The former
2054 are used to qualify single-label names, the latter are used purely
2055 for routing lookups within certain domains to specific links.
2056 resolved now also synthesizes RRs for all entries from /etc/hosts.
2058 * The systemd-resolve tool (which is a client utility for
2059 systemd-resolved) has been improved considerably and is now fully
2060 supported and documented. Hence it has moved from /usr/lib/systemd to
2063 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2066 * The coredump collection logic has been reworked: when a coredump is
2067 collected it is now written to disk, compressed and processed
2068 (including stacktrace extraction) from a new instantiated service
2069 systemd-coredump@.service, instead of directly from the
2070 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2071 processing large coredumps can take up a substantial amount of
2072 resources and time, and this previously happened entirely outside of
2073 systemd's service supervision. With the new logic the core_pattern
2074 hook only does minimal metadata collection before passing off control
2075 to the new instantiated service, which is configured with a time
2076 limit, a nice level and other settings to minimize negative impact on
2077 the rest of the system. Also note that the new logic will honour the
2078 RLIMIT_CORE setting of the crashed process, which now allows users
2079 and processes to turn off coredumping for their processes by setting
2082 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2083 and all forked processes by default. Previously, PID 1 would leave
2084 the setting at "0" for all processes, as set by the kernel. Note that
2085 the resource limit traditionally has no effect on the generated
2086 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2087 logic is used. Since the limit is now honoured (see above) its
2088 default has been changed so that the coredumping logic is enabled by
2089 default for all processes, while allowing specific opt-out.
2091 * When the stacktrace is extracted from processes of system users, this
2092 is now done as "systemd-coredump" user, in order to sandbox this
2093 potentially security sensitive parsing operation. (Note that when
2094 processing coredumps of normal users this is done under the user ID
2095 of process that crashed, as before.) Packagers should take notice
2096 that it is now necessary to create the "systemd-coredump" system user
2097 and group at package installation time.
2099 * The systemd-activate socket activation testing tool gained support
2100 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2101 and --seqpacket switches. It also has been extended to support both
2102 new-style and inetd-style file descriptor passing. Use the new
2103 --inetd switch to request inetd-style file descriptor passing.
2105 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2106 variable, which takes a boolean value. If set to false, ANSI color
2107 output is disabled in the tools even when run on a terminal that
2110 * The VXLAN support in networkd now supports two new settings
2111 DestinationPort= and PortRange=.
2113 * A new systemd.machine_id= kernel command line switch has been added,
2114 that may be used to set the machine ID in /etc/machine-id if it is
2115 not initialized yet. This command line option has no effect if the
2116 file is already initialized.
2118 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2119 specified command line as PID 2 rather than PID 1 in the
2120 container. In this mode PID 1 is a minimal stub init process that
2121 implements the special POSIX and Linux semantics of PID 1 regarding
2122 signal and child process management. Note that this stub init process
2123 is implemented in nspawn itself and requires no support from the
2124 container image. This new logic is useful to support running
2125 arbitrary commands in the container, as normal processes are
2126 generally not prepared to run as PID 1.
2128 * systemd-nspawn gained a new --chdir= switch for setting the current
2129 working directory for the process started in the container.
2131 * "journalctl /dev/sda" will now output all kernel log messages for
2132 specified device from the current boot, in addition to all devices
2133 that are parents of it. This should make log output about devices
2134 pretty useful, as long as kernel drivers attach enough metadata to
2135 the log messages. (The usual SATA drivers do.)
2137 * The sd-journal API gained two new calls
2138 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2139 that report whether log data from /run or /var has been found.
2141 * journalctl gained a new switch "--fields" that prints all journal
2142 record field names currently in use in the journal. This is backed
2143 by two new sd-journal API calls sd_journal_enumerate_fields() and
2144 sd_journal_restart_fields().
2146 * Most configurable timeouts in systemd now expect an argument of
2147 "infinity" to turn them off, instead of "0" as before. The semantics
2148 from now on is that a timeout of "0" means "now", and "infinity"
2149 means "never". To maintain backwards compatibility, "0" continues to
2150 turn off previously existing timeout settings.
2152 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2153 try-reload-or-restart" to clarify what it actually does: the "try"
2154 logic applies to both reloading and restarting, not just restarting.
2155 The old name continues to be accepted for compatibility.
2157 * On boot-up, when PID 1 detects that the system clock is behind the
2158 release date of the systemd version in use, the clock is now set
2159 to the latter. Previously, this was already done in timesyncd, in order
2160 to avoid running with clocks set to the various clock epochs such as
2161 1902, 1938 or 1970. With this change the logic is now done in PID 1
2162 in addition to timesyncd during early boot-up, so that it is enforced
2163 before the first process is spawned by systemd. Note that the logic
2164 in timesyncd remains, as it is more comprehensive and ensures
2165 clock monotonicity by maintaining a persistent timestamp file in
2166 /var. Since /var is generally not available in earliest boot or the
2167 initrd, this part of the logic remains in timesyncd, and is not done
2170 * Support for tweaking details in net_cls.class_id through the
2171 NetClass= configuration directive has been removed, as the kernel
2172 people have decided to deprecate that controller in cgroup v2.
2173 Userspace tools such as nftables are moving over to setting rules
2174 that are specific to the full cgroup path of a task, which obsoletes
2175 these controllers anyway. The NetClass= directive is kept around for
2176 legacy compatibility reasons. For a more in-depth description of the
2177 kernel change, please refer to the respective upstream commit:
2179 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2181 * A new service setting RuntimeMaxSec= has been added that may be used
2182 to specify a maximum runtime for a service. If the timeout is hit, the
2183 service is terminated and put into a failure state.
2185 * A new service setting AmbientCapabilities= has been added. It allows
2186 configuration of additional Linux process capabilities that are
2187 passed to the activated processes. This is only available on very
2190 * The process resource limit settings in service units may now be used
2191 to configure hard and soft limits individually.
2193 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2194 expose support for gcc's __attribute__((cleanup())) C extension.
2195 Specifically, for many object destructor functions alternative
2196 versions have been added that have names suffixed with "p" and take a
2197 pointer to a pointer to the object to destroy, instead of just a
2198 pointer to the object itself. This is useful because these destructor
2199 functions may be used directly as parameters to the cleanup
2200 construct. Internally, systemd has been a heavy user of this GCC
2201 extension for a long time, and with this change similar support is
2202 now available to consumers of the library outside of systemd. Note
2203 that by using this extension in your sources compatibility with old
2204 and strictly ANSI compatible C compilers is lost. However, all gcc or
2205 LLVM versions of recent years support this extension.
2207 * Timer units gained support for a new setting RandomizedDelaySec= that
2208 allows configuring some additional randomized delay to the configured
2209 time. This is useful to spread out timer events to avoid load peaks in
2210 clusters or larger setups.
2212 * Calendar time specifications now support sub-second accuracy.
2214 * Socket units now support listening on SCTP and UDP-lite protocol
2217 * The sd-event API now comes with a full set of man pages.
2219 * Older versions of systemd contained experimental support for
2220 compressing journal files and coredumps with the LZ4 compressor that
2221 was not compatible with the lz4 binary (due to API limitations of the
2222 lz4 library). This support has been removed; only support for files
2223 compatible with the lz4 binary remains. This LZ4 logic is now
2224 officially supported and no longer considered experimental.
2226 * The dkr image import logic has been removed again from importd. dkr's
2227 micro-services focus doesn't fit into the machine image focus of
2228 importd, and quickly got out of date with the upstream dkr API.
2230 * Creation of the /run/lock/lockdev/ directory was dropped from
2231 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2232 been available for many years. If you still need this, you need to
2233 create your own tmpfiles.d config file with:
2235 d /run/lock/lockdev 0775 root lock -
2237 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2238 and RebootArgument= have been moved from the [Service] section of
2239 unit files to [Unit], and they are now supported on all unit types,
2240 not just service units. Of course, elogind will continue to
2241 understand these settings also at the old location, in order to
2242 maintain compatibility.
2244 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2245 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2246 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2247 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2248 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2249 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2250 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2251 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2252 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2253 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2254 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2255 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2256 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2257 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2258 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2259 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2260 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2261 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2262 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2264 — Berlin, 2016-02-11
2268 * A number of properties previously only settable in unit
2269 files are now also available as properties to set when
2270 creating transient units programmatically via the bus, as it
2271 is exposed with systemd-run's --property=
2272 setting. Specifically, these are: SyslogIdentifier=,
2273 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2274 EnvironmentFile=, ReadWriteDirectories=,
2275 ReadOnlyDirectories=, InaccessibleDirectories=,
2276 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2278 * When creating transient services via the bus API it is now
2279 possible to pass in a set of file descriptors to use as
2280 STDIN/STDOUT/STDERR for the invoked process.
2282 * Slice units may now be created transiently via the bus APIs,
2283 similar to the way service and scope units may already be
2284 created transiently.
2286 * Wherever systemd expects a calendar timestamp specification
2287 (like in journalctl's --since= and --until= switches) UTC
2288 timestamps are now supported. Timestamps suffixed with "UTC"
2289 are now considered to be in Universal Time Coordinated
2290 instead of the local timezone. Also, timestamps may now
2291 optionally be specified with sub-second accuracy. Both of
2292 these additions also apply to recurring calendar event
2293 specification, such as OnCalendar= in timer units.
2295 * journalctl gained a new "--sync" switch that asks the
2296 journal daemon to write all so far unwritten log messages to
2297 disk and sync the files, before returning.
2299 * systemd-tmpfiles learned two new line types "q" and "Q" that
2300 operate like "v", but also set up a basic btrfs quota
2301 hierarchy when used on a btrfs file system with quota
2304 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2305 instead of a subvolume (even on a btrfs file system) if the
2306 root directory is a plain directory, and not a
2307 subvolume. This should simplify things with certain chroot()
2308 environments which are not aware of the concept of btrfs
2311 * systemd-detect-virt gained a new --chroot switch to detect
2312 whether execution takes place in a chroot() environment.
2314 * CPUAffinity= now takes CPU index ranges in addition to
2317 * The various memory-related resource limit settings (such as
2318 LimitAS=) now understand the usual K, M, G, ... suffixes to
2319 the base of 1024 (IEC). Similar, the time-related resource
2320 limit settings understand the usual min, h, day, ...
2323 * There's a new system.conf setting DefaultTasksMax= to
2324 control the default TasksMax= setting for services and
2325 scopes running on the system. (TasksMax= is the primary
2326 setting that exposes the "pids" cgroup controller on systemd
2327 and was introduced in the previous systemd release.) The
2328 setting now defaults to 512, which means services that are
2329 not explicitly configured otherwise will only be able to
2330 create 512 processes or threads at maximum, from this
2331 version on. Note that this means that thread- or
2332 process-heavy services might need to be reconfigured to set
2333 TasksMax= to a higher value. It is sufficient to set
2334 TasksMax= in these specific unit files to a higher value, or
2335 even "infinity". Similar, there's now a logind.conf setting
2336 UserTasksMax= that defaults to 4096 and limits the total
2337 number of processes or tasks each user may own
2338 concurrently. nspawn containers also have the TasksMax=
2339 value set by default now, to 8192. Note that all of this
2340 only has an effect if the "pids" cgroup controller is
2341 enabled in the kernel. The general benefit of these changes
2342 should be a more robust and safer system, that provides a
2343 certain amount of per-service fork() bomb protection.
2345 * systemd-nspawn gained the new --network-veth-extra= switch
2346 to define additional and arbitrarily-named virtual Ethernet
2347 links between the host and the container.
2349 * A new service execution setting PassEnvironment= has been
2350 added that allows importing select environment variables
2351 from PID1's environment block into the environment block of
2354 * Timer units gained support for a new RemainAfterElapse=
2355 setting which takes a boolean argument. It defaults to on,
2356 exposing behaviour unchanged to previous releases. If set to
2357 off, timer units are unloaded after they elapsed if they
2358 cannot elapse again. This is particularly useful for
2359 transient timer units, which shall not stay around longer
2360 than until they first elapse.
2362 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2363 default now (the kernel default is 16). This is beneficial
2364 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2365 allows substantially larger numbers of queued
2366 datagrams. This should increase the capability of systemd to
2367 parallelize boot-up, as logging and sd_notify() are unlikely
2368 to stall execution anymore. If you need to change the value
2369 from the new defaults, use the usual sysctl.d/ snippets.
2371 * The compression framing format used by the journal or
2372 coredump processing has changed to be in line with what the
2373 official LZ4 tools generate. LZ4 compression support in
2374 systemd was considered unsupported previously, as the format
2375 was not compatible with the normal tools. With this release
2376 this has changed now, and it is hence safe for downstream
2377 distributions to turn it on. While not compressing as well
2378 as the XZ, LZ4 is substantially faster, which makes
2379 it a good default choice for the compression logic in the
2380 journal and in coredump handling.
2382 * Any reference to /etc/mtab has been dropped from
2383 systemd. The file has been obsolete since a while, but
2384 systemd refused to work on systems where it was incorrectly
2385 set up (it should be a symlink or non-existent). Please make
2386 sure to update to util-linux 2.27.1 or newer in conjunction
2387 with this systemd release, which also drops any reference to
2388 /etc/mtab. If you maintain a distribution make sure that no
2389 software you package still references it, as this is a
2390 likely source of bugs. There's also a glibc bug pending,
2391 asking for removal of any reference to this obsolete file:
2393 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2395 Note that only util-linux versions built with
2396 --enable-libmount-force-mountinfo are supported.
2398 * Support for the ".snapshot" unit type has been removed. This
2399 feature turned out to be little useful and little used, and
2400 has now been removed from the core and from systemctl.
2402 * The dependency types RequiresOverridable= and
2403 RequisiteOverridable= have been removed from systemd. They
2404 have been used only very sparingly to our knowledge and
2405 other options that provide a similar effect (such as
2406 systemctl --mode=ignore-dependencies) are much more useful
2407 and commonly used. Moreover, they were only half-way
2408 implemented as the option to control behaviour regarding
2409 these dependencies was never added to systemctl. By removing
2410 these dependency types the execution engine becomes a bit
2411 simpler. Unit files that use these dependencies should be
2412 changed to use the non-Overridable dependency types
2413 instead. In fact, when parsing unit files with these
2414 options, that's what systemd will automatically convert them
2415 too, but it will also warn, asking users to fix the unit
2416 files accordingly. Removal of these dependency types should
2417 only affect a negligible number of unit files in the wild.
2419 * Behaviour of networkd's IPForward= option changed
2420 (again). It will no longer maintain a per-interface setting,
2421 but propagate one way from interfaces where this is enabled
2422 to the global kernel setting. The global setting will be
2423 enabled when requested by a network that is set up, but
2424 never be disabled again. This change was made to make sure
2425 IPv4 and IPv6 behaviour regarding packet forwarding is
2426 similar (as the Linux IPv6 stack does not support
2427 per-interface control of this setting) and to minimize
2430 * In unit files the behaviour of %u, %U, %h, %s has
2431 changed. These specifiers will now unconditionally resolve
2432 to the various user database fields of the user that the
2433 systemd instance is running as, instead of the user
2434 configured in the specific unit via User=. Note that this
2435 effectively doesn't change much, as resolving of these
2436 specifiers was already turned off in the --system instance
2437 of systemd, as we cannot do NSS lookups from PID 1. In the
2438 --user instance of systemd these specifiers where correctly
2439 resolved, but hardly made any sense, since the user instance
2440 lacks privileges to do user switches anyway, and User= is
2441 hence useless. Moreover, even in the --user instance of
2442 systemd behaviour was awkward as it would only take settings
2443 from User= assignment placed before the specifier into
2444 account. In order to unify and simplify the logic around
2445 this the specifiers will now always resolve to the
2446 credentials of the user invoking the manager (which in case
2447 of PID 1 is the root user).
2449 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2450 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2451 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2452 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2453 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2454 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2455 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2456 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2457 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2458 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2459 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2460 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2461 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2462 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2465 — Berlin, 2015-11-18
2469 * systemd now depends on util-linux v2.27. More specifically,
2470 the newly added mount monitor feature in libmount now
2471 replaces systemd's former own implementation.
2473 * libmount mandates /etc/mtab not to be regular file, and
2474 systemd now enforces this condition at early boot.
2475 /etc/mtab has been deprecated and warned about for a very
2476 long time, so systems running systemd should already have
2477 stopped having this file around as anything else than a
2478 symlink to /proc/self/mounts.
2480 * Support for the "pids" cgroup controller has been added. It
2481 allows accounting the number of tasks in a cgroup and
2482 enforcing limits on it. This adds two new setting
2483 TasksAccounting= and TasksMax= to each unit, as well as a
2484 global option DefaultTasksAccounting=.
2486 * Support for the "net_cls" cgroup controller has been added.
2487 It allows assigning a net class ID to each task in the
2488 cgroup, which can then be used in firewall rules and traffic
2489 shaping configurations. Note that the kernel netfilter net
2490 class code does not currently work reliably for ingress
2491 packets on unestablished sockets.
2493 This adds a new config directive called NetClass= to CGroup
2494 enabled units. Allowed values are positive numbers for fixed
2495 assignments and "auto" for picking a free value
2498 * 'systemctl is-system-running' now returns 'offline' if the
2499 system is not booted with systemd. This command can now be
2500 used as a substitute for 'systemd-notify --booted'.
2502 * Watchdog timeouts have been increased to 3 minutes for all
2503 in-tree service files. Apparently, disk IO issues are more
2504 frequent than we hoped, and user reported >1 minute waiting
2507 * 'machine-id-commit' functionality has been merged into
2508 'machine-id-setup --commit'. The separate binary has been
2511 * The WorkingDirectory= directive in unit files may now be set
2512 to the special value '~'. In this case, the working
2513 directory is set to the home directory of the user
2514 configured in User=.
2516 * "machinectl shell" will now open the shell in the home
2517 directory of the selected user by default.
2519 * The CrashChVT= configuration file setting is renamed to
2520 CrashChangeVT=, following our usual logic of not
2521 abbreviating unnecessarily. The old directive is still
2522 supported for compat reasons. Also, this directive now takes
2523 an integer value between 1 and 63, or a boolean value. The
2524 formerly supported '-1' value for disabling stays around for
2527 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2528 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2529 RootDirectory= properties can now be set for transient
2532 * The systemd-analyze tool gained a new "set-log-target" verb
2533 to change the logging target the system manager logs to
2534 dynamically during runtime. This is similar to how
2535 "systemd-analyze set-log-level" already changes the log
2538 * In nspawn /sys is now mounted as tmpfs, with only a selected
2539 set of subdirectories mounted in from the real sysfs. This
2540 enhances security slightly, and is useful for ensuring user
2541 namespaces work correctly.
2543 * Support for USB FunctionFS activation has been added. This
2544 allows implementation of USB gadget services that are
2545 activated as soon as they are requested, so that they don't
2546 have to run continuously, similar to classic socket
2549 * The "systemctl exit" command now optionally takes an
2550 additional parameter that sets the exit code to return from
2551 the systemd manager when exiting. This is only relevant when
2552 running the systemd user instance, or when running the
2553 system instance in a container.
2555 * sd-bus gained the new API calls sd_bus_path_encode_many()
2556 and sd_bus_path_decode_many() that allow easy encoding and
2557 decoding of multiple identifier strings inside a D-Bus
2558 object path. Another new call sd_bus_default_flush_close()
2559 has been added to flush and close per-thread default
2562 * systemd-cgtop gained support for a -M/--machine= switch to
2563 show the control groups within a certain container only.
2565 * "systemctl kill" gained support for an optional --fail
2566 switch. If specified the requested operation will fail of no
2567 processes have been killed, because the unit had no
2568 processes attached, or similar.
2570 * A new systemd.crash_reboot=1 kernel command line option has
2571 been added that triggers a reboot after crashing. This can
2572 also be set through CrashReboot= in systemd.conf.
2574 * The RuntimeDirectory= setting now understands unit
2575 specifiers like %i or %f.
2577 * A new (still internal) library API sd-ipv4acd has been added,
2578 that implements address conflict detection for IPv4. It's
2579 based on code from sd-ipv4ll, and will be useful for
2580 detecting DHCP address conflicts.
2582 * File descriptors passed during socket activation may now be
2583 named. A new API sd_listen_fds_with_names() is added to
2584 access the names. The default names may be overridden,
2585 either in the .socket file using the FileDescriptorName=
2586 parameter, or by passing FDNAME= when storing the file
2587 descriptors using sd_notify().
2589 * systemd-networkd gained support for:
2591 - Setting the IPv6 Router Advertisement settings via
2592 IPv6AcceptRouterAdvertisements= in .network files.
2594 - Configuring the HelloTimeSec=, MaxAgeSec= and
2595 ForwardDelaySec= bridge parameters in .netdev files.
2597 - Configuring PreferredSource= for static routes in
2600 * The "ask-password" framework used to query for LUKS harddisk
2601 passwords or SSL passwords during boot gained support for
2602 caching passwords in the kernel keyring, if it is
2603 available. This makes sure that the user only has to type in
2604 a passphrase once if there are multiple objects to unlock
2605 with the same one. Previously, such password caching was
2606 available only when Plymouth was used; this moves the
2607 caching logic into the systemd codebase itself. The
2608 "systemd-ask-password" utility gained a new --keyname=
2609 switch to control which kernel keyring key to use for
2610 caching a password in. This functionality is also useful for
2611 enabling display managers such as gdm to automatically
2612 unlock the user's GNOME keyring if its passphrase, the
2613 user's password and the harddisk password are the same, if
2614 gdm-autologin is used.
2616 * When downloading tar or raw images using "machinectl
2617 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2618 file is now also downloaded, if it is available and stored
2619 next to the image file.
2621 * Units of type ".socket" gained a new boolean setting
2622 Writable= which is only useful in conjunction with
2623 ListenSpecial=. If true, enables opening the specified
2624 special file in O_RDWR mode rather than O_RDONLY mode.
2626 * systemd-rfkill has been reworked to become a singleton
2627 service that is activated through /dev/rfkill on each rfkill
2628 state change and saves the settings to disk. This way,
2629 systemd-rfkill is now compatible with devices that exist
2630 only intermittendly, and even restores state if the previous
2631 system shutdown was abrupt rather than clean.
2633 * The journal daemon gained support for vacuuming old journal
2634 files controlled by the number of files that shall remain,
2635 in addition to the already existing control by size and by
2636 date. This is useful as journal interleaving performance
2637 degrades with too many separate journal files, and allows
2638 putting an effective limit on them. The new setting defaults
2639 to 100, but this may be changed by setting SystemMaxFiles=
2640 and RuntimeMaxFiles= in journald.conf. Also, the
2641 "journalctl" tool gained the new --vacuum-files= switch to
2642 manually vacuum journal files to leave only the specified
2643 number of files in place.
2645 * udev will now create /dev/disk/by-path links for ATA devices
2646 on kernels where that is supported.
2648 * Galician, Serbian, Turkish and Korean translations were added.
2650 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2651 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2652 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2653 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2654 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2655 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2656 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2657 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2658 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2659 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2660 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2661 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2662 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2663 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2664 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2665 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2666 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2667 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2669 — Berlin, 2015-10-07
2673 * The DHCP implementation of systemd-networkd gained a set of
2676 - The DHCP server now supports emitting DNS and NTP
2677 information. It may be enabled and configured via
2678 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2679 and NTP information is enabled, but no servers are
2680 configured, the corresponding uplink information (if there
2681 is any) is propagated.
2683 - Server and client now support transmission and reception
2684 of timezone information. It can be configured via the
2685 newly introduced network options UseTimezone=,
2686 EmitTimezone=, and Timezone=. Transmission of timezone
2687 information is enabled between host and containers by
2688 default now: the container will change its local timezone
2689 to what the host has set.
2691 - Lease timeouts can now be configured via
2692 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2694 - The DHCP server improved on the stability of
2695 leases. Clients are more likely to get the same lease
2696 information back, even if the server loses state.
2698 - The DHCP server supports two new configuration options to
2699 control the lease address pool metrics, PoolOffset= and
2702 * The encapsulation limit of tunnels in systemd-networkd may
2703 now be configured via 'EncapsulationLimit='. It allows
2704 modifying the maximum additional levels of encapsulation
2705 that are permitted to be prepended to a packet.
2707 * systemd now supports the concept of user buses replacing
2708 session buses, if used with dbus-1.10 (and enabled via dbus
2709 --enable-user-session). It previously only supported this on
2710 kdbus-enabled systems, and this release expands this to
2711 'dbus-daemon' systems.
2713 * systemd-networkd now supports predictable interface names
2716 * systemd now optionally supports the new Linux kernel
2717 "unified" control group hierarchy. If enabled via the kernel
2718 command-line option 'systemd.unified_cgroup_hierarchy=1',
2719 systemd will try to mount the unified cgroup hierarchy
2720 directly on /sys/fs/cgroup. If not enabled, or not
2721 available, systemd will fall back to the legacy cgroup
2722 hierarchy setup, as before. Host system and containers can
2723 mix and match legacy and unified hierarchies as they
2724 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2725 environment variable to individually select the hierarchy to
2726 use for executed containers. By default, nspawn will use the
2727 unified hierarchy for the containers if the host uses the
2728 unified hierarchy, and the legacy hierarchy otherwise.
2729 Please note that at this point the unified hierarchy is an
2730 experimental kernel feature and is likely to change in one
2731 of the next kernel releases. Therefore, it should not be
2732 enabled by default in downstream distributions yet. The
2733 minimum required kernel version for the unified hierarchy to
2734 work is 4.2. Note that when the unified hierarchy is used
2735 for the first time delegated access to controllers is
2736 safe. Because of this systemd-nspawn containers will get
2737 access to controllers now, as will systemd user
2738 sessions. This means containers and user sessions may now
2739 manage their own resources, partitioning up what the system
2742 * A new special scope unit "init.scope" has been introduced
2743 that encapsulates PID 1 of the system. It may be used to
2744 determine resource usage and enforce resource limits on PID
2745 1 itself. PID 1 hence moved out of the root of the control
2748 * The cgtop tool gained support for filtering out kernel
2749 threads when counting tasks in a control group. Also, the
2750 count of processes is now recursively summed up by
2751 default. Two options -k and --recursive= have been added to
2752 revert to old behaviour. The tool has also been updated to
2753 work correctly in containers now.
2755 * systemd-nspawn's --bind= and --bind-ro= options have been
2756 extended to allow creation of non-recursive bind mounts.
2758 * libsystemd gained two new calls sd_pid_get_cgroup() and
2759 sd_peer_get_cgroup() which return the control group path of
2760 a process or peer of a connected AF_UNIX socket. This
2761 function call is particularly useful when implementing
2762 delegated subtrees support in the control group hierarchy.
2764 * The "sd-event" event loop API of libsystemd now supports
2765 correct dequeuing of real-time signals, without losing
2768 * When systemd requests a PolicyKit decision when managing
2769 units it will now add additional fields to the request,
2770 including unit name and desired operation. This enables more
2771 powerful PolicyKit policies, that make decisions depending
2772 on these parameters.
2774 * nspawn learnt support for .nspawn settings files, that may
2775 accompany the image files or directories of containers, and
2776 may contain additional settings for the container. This is
2777 an alternative to configuring container parameters via the
2778 nspawn command line.
2780 Contributions from: Cristian Rodríguez, Daniel Mack, David
2781 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2782 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2783 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2784 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2785 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2786 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2787 Andersen, Tom Gundersen, Torstein Husebø
2789 — Berlin, 2015-09-08
2793 * machinectl gained a new verb 'shell' which opens a fresh
2794 shell on the target container or the host. It is similar to
2795 the existing 'login' command of machinectl, but spawns the
2796 shell directly without prompting for username or
2797 password. The pseudo machine '.host' now refers to the local
2798 host and is used by default. Hence, 'machinectl shell' can
2799 be used as replacement for 'su -' which spawns a session as
2800 a fresh systemd unit in a way that is fully isolated from
2801 the originating session.
2803 * systemd-networkd learned to cope with private-zone DHCP
2804 options and allows other programs to query the values.
2806 * SELinux access control when enabling/disabling units is no
2807 longer enforced with this release. The previous
2808 implementation was incorrect, and a new corrected
2809 implementation is not yet available. As unit file operations
2810 are still protected via PolicyKit and D-Bus policy this is
2811 not a security problem. Yet, distributions which care about
2812 optimal SELinux support should probably not stabilize on
2815 * sd-bus gained support for matches of type "arg0has=", that
2816 test for membership of strings in string arrays sent in bus
2819 * systemd-resolved now dumps the contents of its DNS and LLMNR
2820 caches to the logs on reception of the SIGUSR1 signal. This
2821 is useful to debug DNS behaviour.
2823 * The coredumpctl tool gained a new --directory= option to
2824 operate on journal files in a specific directory.
2826 * "systemctl reboot" and related commands gained a new
2827 "--message=" option which may be used to set a free-text
2828 wall message when shutting down or rebooting the
2829 system. This message is also logged, which is useful for
2830 figuring out the reason for a reboot or shutdown a
2833 * The "systemd-resolve-host" tool's -i switch now takes
2834 network interface numbers as alternative to interface names.
2836 * A new unit file setting for services has been introduced:
2837 UtmpMode= allows configuration of how precisely systemd
2838 handles utmp and wtmp entries for the service if this is
2839 enabled. This allows writing services that appear similar to
2840 user sessions in the output of the "w", "who", "last" and
2843 * systemd-resolved will now locally synthesize DNS resource
2844 records for the "localhost" and "gateway" domains as well as
2845 the local hostname. This should ensure that clients querying
2846 RRs via resolved will get similar results as those going via
2847 NSS, if nss-myhostname is enabled.
2849 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2850 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2851 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2852 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2853 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2854 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2855 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2856 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2857 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2858 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2859 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2860 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2862 — Berlin, 2015-08-27
2866 * The systemd-efi-boot-generator functionality was merged into
2867 systemd-gpt-auto-generator.
2869 * systemd-networkd now supports Group Policy for vxlan
2870 devices. It can be enabled via the new boolean configuration
2871 option called 'GroupPolicyExtension='.
2873 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2874 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2875 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2877 — Berlin, 2015-07-31
2881 * The python-systemd code has been removed from the systemd repository.
2882 A new repository has been created which accommodates the code from
2883 now on, and we kindly ask distributions to create a separate package
2884 for this: https://github.com/systemd/python-systemd
2886 * The systemd daemon will now reload its main configuration
2887 (/etc/systemd/system.conf) on daemon-reload.
2889 * sd-dhcp now exposes vendor specific extensions via
2890 sd_dhcp_lease_get_vendor_specific().
2892 * systemd-networkd gained a number of new configuration options.
2894 - A new boolean configuration option for TAP devices called
2895 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2896 device, thus allowing to send and receive GSO packets.
2898 - A new tunnel configuration option called 'CopyDSCP='.
2899 If enabled, the DSCP field of ip6 tunnels is copied into the
2900 decapsulated packet.
2902 - A set of boolean bridge configuration options were added.
2903 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2904 and 'UnicastFlood=' are now parsed by networkd and applied to the
2905 respective bridge link device via the respective IFLA_BRPORT_*
2908 - A new string configuration option to override the hostname sent
2909 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2910 is true, networkd will use the configured hostname instead of the
2911 system hostname when sending DHCP requests.
2913 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2914 networkd will configure the IPv6 flow-label of the tunnel device
2915 according to RFC2460.
2917 - The 'macvtap' virtual network devices are now supported, similar to
2918 the already supported 'macvlan' devices.
2920 * systemd-resolved now implements RFC5452 to improve resilience against
2921 cache poisoning. Additionally, source port randomization is enabled
2922 by default to further protect against DNS spoofing attacks.
2924 * nss-mymachines now supports translating UIDs and GIDs of running
2925 containers with user-namespaces enabled. If a container 'foo'
2926 translates a host uid 'UID' to the container uid 'TUID', then
2927 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2928 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2929 mapped as 'vg-foo-TGID'.
2931 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2932 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2933 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2934 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2935 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2936 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2937 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2938 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2939 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2940 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2942 — Berlin, 2015-07-29
2946 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2947 There are no known issues with current sysfs, and udev does not need
2948 or should be used to work around such bugs.
2950 * udev does no longer enable USB HID power management. Several reports
2951 indicate, that some devices cannot handle that setting.
2953 * The udev accelerometer helper was removed. The functionality
2954 is now fully included in iio-sensor-proxy. But this means,
2955 older iio-sensor-proxy versions will no longer provide
2956 accelerometer/orientation data with this systemd version.
2957 Please upgrade iio-sensor-proxy to version 1.0.
2959 * networkd gained a new configuration option IPv6PrivacyExtensions=
2960 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2961 for Stateless Address") on selected networks.
2963 * For the sake of fewer build-time dependencies and less code in the
2964 main repository, the python bindings are about to be removed in the
2965 next release. A new repository has been created which accommodates
2966 the code from now on, and we kindly ask distributions to create a
2967 separate package for this. The removal will take place in v223.
2969 https://github.com/systemd/python-systemd
2971 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2972 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2973 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2974 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2975 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2976 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2977 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2978 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2979 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2980 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2982 — Berlin, 2015-07-07
2986 * The sd-bus.h and sd-event.h APIs have now been declared
2987 stable and have been added to the official interface of
2988 libsystemd.so. sd-bus implements an alternative D-Bus client
2989 library, that is relatively easy to use, very efficient and
2990 supports both classic D-Bus as well as kdbus as transport
2991 backend. sd-event is a generic event loop abstraction that
2992 is built around Linux epoll, but adds features such as event
2993 prioritization or efficient timer handling. Both APIs are good
2994 choices for C programs looking for a bus and/or event loop
2995 implementation that is minimal and does not have to be
2996 portable to other kernels.
2998 * kdbus support is no longer compile-time optional. It is now
2999 always built-in. However, it can still be disabled at
3000 runtime using the kdbus=0 kernel command line setting, and
3001 that setting may be changed to default to off, by specifying
3002 --disable-kdbus at build-time. Note though that the kernel
3003 command line setting has no effect if the kdbus.ko kernel
3004 module is not installed, in which case kdbus is (obviously)
3005 also disabled. We encourage all downstream distributions to
3006 begin testing kdbus by adding it to the kernel images in the
3007 development distributions, and leaving kdbus support in
3010 * The minimal required util-linux version has been bumped to
3013 * Support for chkconfig (--enable-chkconfig) was removed in
3014 favor of calling an abstraction tool
3015 /lib/systemd/systemd-sysv-install. This needs to be
3016 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3017 in README for details.
3019 * If there's a systemd unit and a SysV init script for the
3020 same service name, and the user executes "systemctl enable"
3021 for it (or a related call), then this will now enable both
3022 (or execute the related operation on both), not just the
3025 * The libudev API documentation has been converted from gtkdoc
3028 * gudev has been removed from the systemd tree, it is now an
3031 * The systemd-cgtop tool learnt a new --raw switch to generate
3032 "raw" (machine parsable) output.
3034 * networkd's IPForwarding= .network file setting learnt the
3035 new setting "kernel", which ensures that networkd does not
3036 change the IP forwarding sysctl from the default kernel
3039 * The systemd-logind bus API now exposes a new boolean
3040 property "Docked" that reports whether logind considers the
3041 system "docked", i.e. connected to a docking station or not.
3043 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3044 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3045 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3046 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3047 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3048 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3049 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3050 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3051 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3052 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3053 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3054 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3055 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3056 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3057 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3058 Fink, Zbigniew Jędrzejewski-Szmek
3060 — Berlin, 2015-06-19
3064 * The gudev library has been extracted into a separate repository
3065 available at: https://git.gnome.org/browse/libgudev/
3066 It is now managed as part of the Gnome project. Distributions
3067 are recommended to pass --disable-gudev to systemd and use
3068 gudev from the Gnome project instead. gudev is still included
3069 in systemd, for now. It will be removed soon, though. Please
3070 also see the announcement-thread on systemd-devel:
3071 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3073 * systemd now exposes a CPUUsageNSec= property for each
3074 service unit on the bus, that contains the overall consumed
3075 CPU time of a service (the sum of what each process of the
3076 service consumed). This value is only available if
3077 CPUAccounting= is turned on for a service, and is then shown
3078 in the "systemctl status" output.
3080 * Support for configuring alternative mappings of the old SysV
3081 runlevels to systemd targets has been removed. They are now
3082 hardcoded in a way that runlevels 2, 3, 4 all map to
3083 multi-user.target and 5 to graphical.target (which
3084 previously was already the default behaviour).
3086 * The auto-mounter logic gained support for mount point
3087 expiry, using a new TimeoutIdleSec= setting in .automount
3088 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3090 * The EFI System Partition (ESP) as mounted to /boot by
3091 systemd-efi-boot-generator will now be unmounted
3092 automatically after 2 minutes of not being used. This should
3093 minimize the risk of ESP corruptions.
3095 * New /etc/fstab options x-systemd.requires= and
3096 x-systemd.requires-mounts-for= are now supported to express
3097 additional dependencies for mounts. This is useful for
3098 journalling file systems that support external journal
3099 devices or overlay file systems that require underlying file
3100 systems to be mounted.
3102 * systemd does not support direct live-upgrades (via systemctl
3103 daemon-reexec) from versions older than v44 anymore. As no
3104 distribution we are aware of shipped such old versions in a
3105 stable release this should not be problematic.
3107 * When systemd forks off a new per-connection service instance
3108 it will now set the $REMOTE_ADDR environment variable to the
3109 remote IP address, and $REMOTE_PORT environment variable to
3110 the remote IP port. This behaviour is similar to the
3111 corresponding environment variables defined by CGI.
3113 * systemd-networkd gained support for uplink failure
3114 detection. The BindCarrier= option allows binding interface
3115 configuration dynamically to the link sense of other
3116 interfaces. This is useful to achieve behaviour like in
3119 * systemd-networkd gained support for configuring the DHCP
3120 client identifier to use when requesting leases.
3122 * systemd-networkd now has a per-network UseNTP= option to
3123 configure whether NTP server information acquired via DHCP
3124 is passed on to services like systemd-timesyncd.
3126 * systemd-networkd gained support for vti6 tunnels.
3128 * Note that systemd-networkd manages the sysctl variable
3129 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3130 it is configured for since v219. The variable controls IP
3131 forwarding, and is a per-interface alternative to the global
3132 /proc/sys/net/ipv[46]/ip_forward. This setting is
3133 configurable in the IPForward= option, which defaults to
3134 "no". This means if networkd is used for an interface it is
3135 no longer sufficient to set the global sysctl option to turn
3136 on IP forwarding! Instead, the .network file option
3137 IPForward= needs to be turned on! Note that the
3138 implementation of this behaviour was broken in v219 and has
3141 * Many bonding and vxlan options are now configurable in
3144 * systemd-nspawn gained a new --property= setting to set unit
3145 properties for the container scope. This is useful for
3146 setting resource parameters (e.g. "CPUShares=500") on
3147 containers started from the command line.
3149 * systemd-nspawn gained a new --private-users= switch to make
3150 use of user namespacing available on recent Linux kernels.
3152 * systemd-nspawn may now be called as part of a shell pipeline
3153 in which case the pipes used for stdin and stdout are passed
3154 directly to the process invoked in the container, without
3155 indirection via a pseudo tty.
3157 * systemd-nspawn gained a new switch to control the UNIX
3158 signal to use when killing the init process of the container
3161 * systemd-nspawn gained a new --overlay= switch for mounting
3162 overlay file systems into the container using the new kernel
3165 * When a container image is imported via systemd-importd and
3166 the host file system is not btrfs, a loopback block device
3167 file is created in /var/lib/machines.raw with a btrfs file
3168 system inside. It is then mounted to /var/lib/machines to
3169 enable btrfs features for container management. The loopback
3170 file and btrfs file system is grown as needed when container
3171 images are imported via systemd-importd.
3173 * systemd-machined/systemd-importd gained support for btrfs
3174 quota, to enforce container disk space limits on disk. This
3175 is exposed in "machinectl set-limit".
3177 * systemd-importd now can import containers from local .tar,
3178 .raw and .qcow2 images, and export them to .tar and .raw. It
3179 can also import dkr v2 images now from the network (on top
3182 * systemd-importd gained support for verifying downloaded
3183 images with gpg2 (previously only gpg1 was supported).
3185 * systemd-machined, systemd-logind, systemd: most bus calls
3186 are now accessible to unprivileged processes via
3187 PolicyKit. Also, systemd-logind will now allow users to kill
3188 their own sessions without further privileges or
3191 * systemd-shutdownd has been removed. This service was
3192 previously responsible for implementing scheduled shutdowns
3193 as exposed in /usr/bin/shutdown's time parameter. This
3194 functionality has now been moved into systemd-logind and is
3195 accessible via a bus interface.
3197 * "systemctl reboot" gained a new switch --firmware-setup that
3198 can be used to reboot into the EFI firmware setup, if that
3199 is available. systemd-logind now exposes an API on the bus
3200 to trigger such reboots, in case graphical desktop UIs want
3201 to cover this functionality.
3203 * "systemctl enable", "systemctl disable" and "systemctl mask"
3204 now support a new "--now" switch. If specified the units
3205 that are enabled will also be started, and the ones
3206 disabled/masked also stopped.
3208 * The Gummiboot EFI boot loader tool has been merged into
3209 systemd, and renamed to "systemd-boot". The bootctl tool has been
3210 updated to support systemd-boot.
3212 * An EFI kernel stub has been added that may be used to create
3213 kernel EFI binaries that contain not only the actual kernel,
3214 but also an initrd, boot splash, command line and OS release
3215 information. This combined binary can then be signed as a
3216 single image, so that the firmware can verify it all in one
3217 step. systemd-boot has special support for EFI binaries created
3218 like this and can extract OS release information from them
3219 and show them in the boot menu. This functionality is useful
3220 to implement cryptographically verified boot schemes.
3222 * Optional support has been added to systemd-fsck to pass
3223 fsck's progress report to an AF_UNIX socket in the file
3226 * udev will no longer create device symlinks for all block
3227 devices by default. A blacklist for excluding special block
3228 devices from this logic has been turned into a whitelist
3229 that requires picking block devices explicitly that require
3232 * A new (currently still internal) API sd-device.h has been
3233 added to libsystemd. This modernized API is supposed to
3234 replace libudev eventually. In fact, already much of libudev
3235 is now just a wrapper around sd-device.h.
3237 * A new hwdb database for storing metadata about pointing
3238 stick devices has been added.
3240 * systemd-tmpfiles gained support for setting file attributes
3241 similar to the "chattr" tool with new 'h' and 'H' lines.
3243 * systemd-journald will no longer unconditionally set the
3244 btrfs NOCOW flag on new journal files. This is instead done
3245 with tmpfiles snippet using the new 'h' line type. This
3246 allows easy disabling of this logic, by masking the
3247 journal-nocow.conf tmpfiles file.
3249 * systemd-journald will now translate audit message types to
3250 human readable identifiers when writing them to the
3251 journal. This should improve readability of audit messages.
3253 * The LUKS logic gained support for the offset= and skip=
3254 options in /etc/crypttab, as previously implemented by
3257 * /usr/lib/os-release gained a new optional field VARIANT= for
3258 distributions that support multiple variants (such as a
3259 desktop edition, a server edition, ...)
3261 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3262 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3263 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3264 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3265 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3266 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3267 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3268 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3269 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3270 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3271 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3272 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3273 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3274 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3275 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3276 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3277 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3278 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3279 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3280 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3281 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3282 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3283 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3284 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3285 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3286 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3287 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3289 — Berlin, 2015-05-22
3293 * Introduce a new API "sd-hwdb.h" for querying the hardware
3294 metadata database. With this minimal interface one can query
3295 and enumerate the udev hwdb, decoupled from the old libudev
3296 library. libudev's interface for this is now only a wrapper
3297 around sd-hwdb. A new tool systemd-hwdb has been added to
3298 interface with and update the database.
3300 * When any of systemd's tools copies files (for example due to
3301 tmpfiles' C lines) a btrfs reflink will attempted first,
3302 before bytewise copying is done.
3304 * systemd-nspawn gained a new --ephemeral switch. When
3305 specified a btrfs snapshot is taken of the container's root
3306 directory, and immediately removed when the container
3307 terminates again. Thus, a container can be started whose
3308 changes never alter the container's root directory, and are
3309 lost on container termination. This switch can also be used
3310 for starting a container off the root file system of the
3311 host without affecting the host OS. This switch is only
3312 available on btrfs file systems.
3314 * systemd-nspawn gained a new --template= switch. It takes the
3315 path to a container tree to use as template for the tree
3316 specified via --directory=, should that directory be
3317 missing. This allows instantiating containers dynamically,
3318 on first run. This switch is only available on btrfs file
3321 * When a .mount unit refers to a mount point on which multiple
3322 mounts are stacked, and the .mount unit is stopped all of
3323 the stacked mount points will now be unmounted until no
3324 mount point remains.
3326 * systemd now has an explicit notion of supported and
3327 unsupported unit types. Jobs enqueued for unsupported unit
3328 types will now fail with an "unsupported" error code. More
3329 specifically .swap, .automount and .device units are not
3330 supported in containers, .busname units are not supported on
3331 non-kdbus systems. .swap and .automount are also not
3332 supported if their respective kernel compile time options
3335 * machinectl gained support for two new "copy-from" and
3336 "copy-to" commands for copying files from a running
3337 container to the host or vice versa.
3339 * machinectl gained support for a new "bind" command to bind
3340 mount host directories into local containers. This is
3341 currently only supported for nspawn containers.
3343 * networkd gained support for configuring bridge forwarding
3344 database entries (fdb) from .network files.
3346 * A new tiny daemon "systemd-importd" has been added that can
3347 download container images in tar, raw, qcow2 or dkr formats,
3348 and make them available locally in /var/lib/machines, so
3349 that they can run as nspawn containers. The daemon can GPG
3350 verify the downloads (not supported for dkr, since it has no
3351 provisions for verifying downloads). It will transparently
3352 decompress bz2, xz, gzip compressed downloads if necessary,
3353 and restore sparse files on disk. The daemon uses privilege
3354 separation to ensure the actual download logic runs with
3355 fewer privileges than the daemon itself. machinectl has
3356 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3357 make the functionality of importd available to the
3358 user. With this in place the Fedora and Ubuntu "Cloud"
3359 images can be downloaded and booted as containers unmodified
3360 (the Fedora images lack the appropriate GPG signature files
3361 currently, so they cannot be verified, but this will change
3362 soon, hopefully). Note that downloading images is currently
3363 only fully supported on btrfs.
3365 * machinectl is now able to list container images found in
3366 /var/lib/machines, along with some metadata about sizes of
3367 disk and similar. If the directory is located on btrfs and
3368 quota is enabled, this includes quota display. A new command
3369 "image-status" has been added that shows additional
3370 information about images.
3372 * machinectl is now able to clone container images
3373 efficiently, if the underlying file system (btrfs) supports
3374 it, with the new "machinectl clone" command. It also
3375 gained commands for renaming and removing images, as well as
3376 marking them read-only or read-write (supported also on
3377 legacy file systems).
3379 * networkd gained support for collecting LLDP network
3380 announcements, from hardware that supports this. This is
3381 shown in networkctl output.
3383 * systemd-run gained support for a new -t (--pty) switch for
3384 invoking a binary on a pty whose input and output is
3385 connected to the invoking terminal. This allows executing
3386 processes as system services while interactively
3387 communicating with them via the terminal. Most interestingly
3388 this is supported across container boundaries. Invoking
3389 "systemd-run -t /bin/bash" is an alternative to running a
3390 full login session, the difference being that the former
3391 will not register a session, nor go through the PAM session
3394 * tmpfiles gained support for a new "v" line type for creating
3395 btrfs subvolumes. If the underlying file system is a legacy
3396 file system, this automatically degrades to creating a
3397 normal directory. Among others /var/lib/machines is now
3398 created like this at boot, should it be missing.
3400 * The directory /var/lib/containers/ has been deprecated and
3401 been replaced by /var/lib/machines. The term "machines" has
3402 been used in the systemd context as generic term for both
3403 VMs and containers, and hence appears more appropriate for
3404 this, as the directory can also contain raw images bootable
3407 * systemd-nspawn when invoked with -M but without --directory=
3408 or --image= is now capable of searching for the container
3409 root directory, subvolume or disk image automatically, in
3410 /var/lib/machines. systemd-nspawn@.service has been updated
3411 to make use of this, thus allowing it to be used for raw
3414 * A new machines.target unit has been introduced that is
3415 supposed to group all containers/VMs invoked as services on
3416 the system. systemd-nspawn@.service has been updated to
3417 integrate with that.
3419 * machinectl gained a new "start" command, for invoking a
3420 container as a service. "machinectl start foo" is mostly
3421 equivalent to "systemctl start systemd-nspawn@foo.service",
3422 but handles escaping in a nicer way.
3424 * systemd-nspawn will now mount most of the cgroupfs tree
3425 read-only into each container, with the exception of the
3426 container's own subtree in the name=systemd hierarchy.
3428 * journald now sets the special FS_NOCOW file flag for its
3429 journal files. This should improve performance on btrfs, by
3430 avoiding heavy fragmentation when journald's write-pattern
3431 is used on COW file systems. It degrades btrfs' data
3432 integrity guarantees for the files to the same levels as for
3433 ext3/ext4 however. This should be OK though as journald does
3434 its own data integrity checks and all its objects are
3435 checksummed on disk. Also, journald should handle btrfs disk
3436 full events a lot more gracefully now, by processing SIGBUS
3437 errors, and not relying on fallocate() anymore.
3439 * When journald detects that journal files it is writing to
3440 have been deleted it will immediately start new journal
3443 * systemd now provides a way to store file descriptors
3444 per-service in PID 1. This is useful for daemons to ensure
3445 that fds they require are not lost during a daemon
3446 restart. The fds are passed to the daemon on the next
3447 invocation in the same way socket activation fds are
3448 passed. This is now used by journald to ensure that the
3449 various sockets connected to all the system's stdout/stderr
3450 are not lost when journald is restarted. File descriptors
3451 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3452 an extension to sd_notify(). Note that a limit is enforced
3453 on the number of fds a service can store in PID 1, and it
3454 defaults to 0, so that no fds may be stored, unless this is
3455 explicitly turned on.
3457 * The default TERM variable to use for units connected to a
3458 terminal, when no other value is explicitly is set is now
3459 vt220 rather than vt102. This should be fairly safe still,
3460 but allows PgUp/PgDn work.
3462 * The /etc/crypttab option header= as known from Debian is now
3465 * "loginctl user-status" and "loginctl session-status" will
3466 now show the last 10 lines of log messages of the
3467 user/session following the status output. Similar,
3468 "machinectl status" will show the last 10 log lines
3469 associated with a virtual machine or container
3470 service. (Note that this is usually not the log messages
3471 done in the VM/container itself, but simply what the
3472 container manager logs. For nspawn this includes all console
3475 * "loginctl session-status" without further argument will now
3476 show the status of the session of the caller. Similar,
3477 "lock-session", "unlock-session", "activate",
3478 "enable-linger", "disable-linger" may now be called without
3479 session/user parameter in which case they apply to the
3480 caller's session/user.
3482 * An X11 session scriptlet is now shipped that uploads
3483 $DISPLAY and $XAUTHORITY into the environment of the systemd
3484 --user daemon if a session begins. This should improve
3485 compatibility with X11 enabled applications run as systemd
3488 * Generators are now subject to masking via /etc and /run, the
3489 same way as unit files.
3491 * networkd .network files gained support for configuring
3492 per-link IPv4/IPv6 packet forwarding as well as IPv4
3493 masquerading. This is by default turned on for veth links to
3494 containers, as registered by systemd-nspawn. This means that
3495 nspawn containers run with --network-veth will now get
3496 automatic routed access to the host's networks without any
3497 further configuration or setup, as long as networkd runs on
3500 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3501 or UDP posts of a container on the host. With this in place
3502 it is possible to run containers with private veth links
3503 (--network-veth), and have their functionality exposed on
3504 the host as if their services were running directly on the
3507 * systemd-nspawn's --network-veth switch now gained a short
3508 version "-n", since with the changes above it is now truly
3509 useful out-of-the-box. The systemd-nspawn@.service has been
3510 updated to make use of it too by default.
3512 * systemd-nspawn will now maintain a per-image R/W lock, to
3513 ensure that the same image is not started more than once
3514 writable. (It's OK to run an image multiple times
3515 simultaneously in read-only mode.)
3517 * systemd-nspawn's --image= option is now capable of
3518 dissecting and booting MBR and GPT disk images that contain
3519 only a single active Linux partition. Previously it
3520 supported only GPT disk images with proper GPT type
3521 IDs. This allows running cloud images from major
3522 distributions directly with systemd-nspawn, without
3525 * In addition to collecting mouse dpi data in the udev
3526 hardware database, there's now support for collecting angle
3527 information for mouse scroll wheels. The database is
3528 supposed to guarantee similar scrolling behavior on mice
3529 that it knows about. There's also support for collecting
3530 information about Touchpad types.
3532 * udev's input_id built-in will now also collect touch screen
3533 dimension data and attach it to probed devices.
3535 * /etc/os-release gained support for a Distribution Privacy
3538 * networkd gained support for creating "ipvlan", "gretap",
3539 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3541 * systemd-tmpfiles gained support for "a" lines for setting
3544 * systemd-nspawn will now mount /tmp in the container to
3545 tmpfs, automatically.
3547 * systemd now exposes the memory.usage_in_bytes cgroup
3548 attribute and shows it for each service in the "systemctl
3549 status" output, if available.
3551 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3552 immediate reboot is triggered. This useful if shutdown is
3553 hung and is unable to complete, to expedite the
3554 operation. Note that this kind of reboot will still unmount
3555 all file systems, and hence should not result in fsck being
3558 * A .device unit for an optical block device will now be
3559 considered active only when a medium is in the drive. Also,
3560 mount units are now bound to their backing devices thus
3561 triggering automatic unmounting when devices become
3562 unavailable. With this in place systemd will now
3563 automatically unmount left-over mounts when a CD-ROM is
3564 ejected or an USB stick is yanked from the system.
3566 * networkd-wait-online now has support for waiting for
3567 specific interfaces only (with globbing), and for giving up
3568 after a configurable timeout.
3570 * networkd now exits when idle. It will be automatically
3571 restarted as soon as interfaces show up, are removed or
3572 change state. networkd will stay around as long as there is
3573 at least one DHCP state machine or similar around, that keep
3576 * networkd may now configure IPv6 link-local addressing in
3577 addition to IPv4 link-local addressing.
3579 * The IPv6 "token" for use in SLAAC may now be configured for
3580 each .network interface in networkd.
3582 * Routes configured with networkd may now be assigned a scope
3585 * networkd's [Match] sections now support globbing and lists
3586 of multiple space-separated matches per item.
3588 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3589 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3590 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3591 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3592 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3593 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3594 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3595 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3596 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3597 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3598 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3599 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3600 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3601 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3602 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3603 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3604 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3605 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3606 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3607 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3608 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3609 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3610 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3611 Hoffmann, Zbigniew Jędrzejewski-Szmek
3613 — Berlin, 2015-02-16
3617 * When querying unit file enablement status (for example via
3618 "systemctl is-enabled"), a new state "indirect" is now known
3619 which indicates that a unit might not be enabled itself, but
3620 another unit listed in its Also= setting might be.
3622 * Similar to the various existing ConditionXYZ= settings for
3623 units, there are now matching AssertXYZ= settings. While
3624 failing conditions cause a unit to be skipped, but its job
3625 to succeed, failing assertions declared like this will cause
3626 a unit start operation and its job to fail.
3628 * hostnamed now knows a new chassis type "embedded".
3630 * systemctl gained a new "edit" command. When used on a unit
3631 file, this allows extending unit files with .d/ drop-in
3632 configuration snippets or editing the full file (after
3633 copying it from /usr/lib to /etc). This will invoke the
3634 user's editor (as configured with $EDITOR), and reload the
3635 modified configuration after editing.
3637 * "systemctl status" now shows the suggested enablement state
3638 for a unit, as declared in the (usually vendor-supplied)
3639 system preset files.
3641 * nss-myhostname will now resolve the single-label host name
3642 "gateway" to the locally configured default IP routing
3643 gateways, ordered by their metrics. This assigns a stable
3644 name to the used gateways, regardless which ones are
3645 currently configured. Note that the name will only be
3646 resolved after all other name sources (if nss-myhostname is
3647 configured properly) and should hence not negatively impact
3648 systems that use the single-label host name "gateway" in
3651 * systemd-inhibit now allows filtering by mode when listing
3654 * Scope and service units gained a new "Delegate" boolean
3655 property, which, when set, allows processes running inside the
3656 unit to further partition resources. This is primarily
3657 useful for systemd user instances as well as container
3660 * journald will now pick up audit messages directly from
3661 the kernel, and log them like any other log message. The
3662 audit fields are split up and fully indexed. This means that
3663 journalctl in many ways is now a (nicer!) alternative to
3664 ausearch, the traditional audit client. Note that this
3665 implements only a minimal audit client. If you want the
3666 special audit modes like reboot-on-log-overflow, please use
3667 the traditional auditd instead, which can be used in
3668 parallel to journald.
3670 * The ConditionSecurity= unit file option now understands the
3671 special string "audit" to check whether auditing is
3674 * journalctl gained two new commands --vacuum-size= and
3675 --vacuum-time= to delete old journal files until the
3676 remaining ones take up no more than the specified size on disk,
3677 or are not older than the specified time.
3679 * A new, native PPPoE library has been added to sd-network,
3680 systemd's library of light-weight networking protocols. This
3681 library will be used in a future version of networkd to
3682 enable PPPoE communication without an external pppd daemon.
3684 * The busctl tool now understands a new "capture" verb that
3685 works similar to "monitor", but writes a packet capture
3686 trace to STDOUT that can be redirected to a file which is
3687 compatible with libcap's capture file format. This can then
3688 be loaded in Wireshark and similar tools to inspect bus
3691 * The busctl tool now understands a new "tree" verb that shows
3692 the object trees of a specific service on the bus, or of all
3695 * The busctl tool now understands a new "introspect" verb that
3696 shows all interfaces and members of objects on the bus,
3697 including their signature and values. This is particularly
3698 useful to get more information about bus objects shown by
3699 the new "busctl tree" command.
3701 * The busctl tool now understands new verbs "call",
3702 "set-property" and "get-property" for invoking bus method
3703 calls, setting and getting bus object properties in a
3706 * busctl gained a new --augment-creds= argument that controls
3707 whether the tool shall augment credential information it
3708 gets from the bus with data from /proc, in a possibly
3711 * nspawn's --link-journal= switch gained two new values
3712 "try-guest" and "try-host" that work like "guest" and
3713 "host", but do not fail if the host has no persistent
3714 journalling enabled. -j is now equivalent to
3715 --link-journal=try-guest.
3717 * macvlan network devices created by nspawn will now have
3718 stable MAC addresses.
3720 * A new SmackProcessLabel= unit setting has been added, which
3721 controls the SMACK security label processes forked off by
3722 the respective unit shall use.
3724 * If compiled with --enable-xkbcommon, systemd-localed will
3725 verify x11 keymap settings by compiling the given keymap. It
3726 will spew out warnings if the compilation fails. This
3727 requires libxkbcommon to be installed.
3729 * When a coredump is collected, a larger number of metadata
3730 fields is now collected and included in the journal records
3731 created for it. More specifically, control group membership,
3732 environment variables, memory maps, working directory,
3733 chroot directory, /proc/$PID/status, and a list of open file
3734 descriptors is now stored in the log entry.
3736 * The udev hwdb now contains DPI information for mice. For
3739 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3741 * All systemd programs that read standalone configuration
3742 files in /etc now also support a corresponding series of
3743 .conf.d configuration directories in /etc/, /run/,
3744 /usr/local/lib/, /usr/lib/, and (if configured with
3745 --enable-split-usr) /lib/. In particular, the following
3746 configuration files now have corresponding configuration
3747 directories: system.conf user.conf, logind.conf,
3748 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3749 resolved.conf, timesyncd.conf, journal-remote.conf, and
3750 journal-upload.conf. Note that distributions should use the
3751 configuration directories in /usr/lib/; the directories in
3752 /etc/ are reserved for the system administrator.
3754 * systemd-rfkill will no longer take the rfkill device name
3755 into account when storing rfkill state on disk, as the name
3756 might be dynamically assigned and not stable. Instead, the
3757 ID_PATH udev variable combined with the rfkill type (wlan,
3758 bluetooth, ...) is used.
3760 * A new service systemd-machine-id-commit.service has been
3761 added. When used on systems where /etc is read-only during
3762 boot, and /etc/machine-id is not initialized (but an empty
3763 file), this service will copy the temporary machine ID
3764 created as replacement into /etc after the system is fully
3765 booted up. This is useful for systems that are freshly
3766 installed with a non-initialized machine ID, but should get
3767 a fixed machine ID for subsequent boots.
3769 * networkd's .netdev files now provide a large set of
3770 configuration parameters for VXLAN devices. Similarly, the
3771 bridge port cost parameter is now configurable in .network
3772 files. There's also new support for configuring IP source
3773 routing. networkd .link files gained support for a new
3774 OriginalName= match that is useful to match against the
3775 original interface name the kernel assigned. .network files
3776 may include MTU= and MACAddress= fields for altering the MTU
3777 and MAC address while being connected to a specific network
3780 * The LUKS logic gained supported for configuring
3781 UUID-specific key files. There's also new support for naming
3782 LUKS device from the kernel command line, using the new
3783 luks.name= argument.
3785 * Timer units may now be transiently created via the bus API
3786 (this was previously already available for scope and service
3787 units). In addition it is now possible to create multiple
3788 transient units at the same time with a single bus call. The
3789 "systemd-run" tool has been updated to make use of this for
3790 running commands on a specified time, in at(1)-style.
3792 * tmpfiles gained support for "t" lines, for assigning
3793 extended attributes to files. Among other uses this may be
3794 used to assign SMACK labels to files.
3796 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3797 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3798 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3799 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3800 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3801 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3802 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3803 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3804 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3805 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3806 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3807 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3808 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3809 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3810 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3811 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3812 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3813 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3815 — Berlin, 2014-12-10
3819 * journalctl gained the new options -t/--identifier= to match
3820 on the syslog identifier (aka "tag"), as well as --utc to
3821 show log timestamps in the UTC timezone. journalctl now also
3822 accepts -n/--lines=all to disable line capping in a pager.
3824 * journalctl gained a new switch, --flush, that synchronously
3825 flushes logs from /run/log/journal to /var/log/journal if
3826 persistent storage is enabled. systemd-journal-flush.service
3827 now waits until the operation is complete.
3829 * Services can notify the manager before they start a reload
3830 (by sending RELOADING=1) or shutdown (by sending
3831 STOPPING=1). This allows the manager to track and show the
3832 internal state of daemons and closes a race condition when
3833 the process is still running but has closed its D-Bus
3836 * Services with Type=oneshot do not have to have any ExecStart
3839 * User units are now loaded also from
3840 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3841 /run/systemd/user directory that was already previously
3842 supported, but is under the control of the user.
3844 * Job timeouts (i.e. time-outs on the time a job that is
3845 queued stays in the run queue) can now optionally result in
3846 immediate reboot or power-off actions (JobTimeoutAction= and
3847 JobTimeoutRebootArgument=). This is useful on ".target"
3848 units, to limit the maximum time a target remains
3849 undispatched in the run queue, and to trigger an emergency
3850 operation in such a case. This is now used by default to
3851 turn off the system if boot-up (as defined by everything in
3852 basic.target) hangs and does not complete for at least
3853 15min. Also, if power-off or reboot hang for at least 30min
3854 an immediate power-off/reboot operation is triggered. This
3855 functionality is particularly useful to increase reliability
3856 on embedded devices, but also on laptops which might
3857 accidentally get powered on when carried in a backpack and
3858 whose boot stays stuck in a hard disk encryption passphrase
3861 * systemd-logind can be configured to also handle lid switch
3862 events even when the machine is docked or multiple displays
3863 are attached (HandleLidSwitchDocked= option).
3865 * A helper binary and a service have been added which can be
3866 used to resume from hibernation in the initramfs. A
3867 generator will parse the resume= option on the kernel
3868 command line to trigger resume.
3870 * A user console daemon systemd-consoled has been
3871 added. Currently, it is a preview, and will so far open a
3872 single terminal on each session of the user marked as
3873 Desktop=systemd-console.
3875 * Route metrics can be specified for DHCP routes added by
3878 * The SELinux context of socket-activated services can be set
3879 from the information provided by the networking stack
3880 (SELinuxContextFromNet= option).
3882 * Userspace firmware loading support has been removed and
3883 the minimum supported kernel version is thus bumped to 3.7.
3885 * Timeout for udev workers has been increased from 1 to 3
3886 minutes, but a warning will be printed after 1 minute to
3887 help diagnose kernel modules that take a long time to load.
3889 * Udev rules can now remove tags on devices with TAG-="foobar".
3891 * systemd's readahead implementation has been removed. In many
3892 circumstances it didn't give expected benefits even for
3893 rotational disk drives and was becoming less relevant in the
3894 age of SSDs. As none of the developers has been using
3895 rotating media anymore, and nobody stepped up to actively
3896 maintain this component of systemd it has now been removed.
3898 * Swap units can use Options= to specify discard options.
3899 Discard options specified for swaps in /etc/fstab are now
3902 * Docker containers are now detected as a separate type of
3905 * The Password Agent protocol gained support for queries where
3906 the user input is shown, useful e.g. for user names.
3907 systemd-ask-password gained a new --echo option to turn that
3910 * The default sysctl.d/ snippets will now set:
3912 net.core.default_qdisc = fq_codel
3914 This selects Fair Queuing Controlled Delay as the default
3915 queuing discipline for network interfaces. fq_codel helps
3916 fight the network bufferbloat problem. It is believed to be
3917 a good default with no tuning required for most workloads.
3918 Downstream distributions may override this choice. On 10Gbit
3919 servers that do not do forwarding, "fq" may perform better.
3920 Systems without a good clocksource should use "pfifo_fast".
3922 * If kdbus is enabled during build a new option BusPolicy= is
3923 available for service units, that allows locking all service
3924 processes into a stricter bus policy, in order to limit
3925 access to various bus services, or even hide most of them
3926 from the service's view entirely.
3928 * networkctl will now show the .network and .link file
3929 networkd has applied to a specific interface.
3931 * sd-login gained a new API call sd_session_get_desktop() to
3932 query which desktop environment has been selected for a
3935 * UNIX utmp support is now compile-time optional to support
3936 legacy-free systems.
3938 * systemctl gained two new commands "add-wants" and
3939 "add-requires" for pulling in units from specific targets
3942 * If the word "rescue" is specified on the kernel command line
3943 the system will now boot into rescue mode (aka
3944 rescue.target), which was previously available only by
3945 specifying "1" or "systemd.unit=rescue.target" on the kernel
3946 command line. This new kernel command line option nicely
3947 mirrors the already existing "emergency" kernel command line
3950 * New kernel command line options mount.usr=, mount.usrflags=,
3951 mount.usrfstype= have been added that match root=, rootflags=,
3952 rootfstype= but allow mounting a specific file system to
3955 * The $NOTIFY_SOCKET is now also passed to control processes of
3956 services, not only the main process.
3958 * This version reenables support for fsck's -l switch. This
3959 means at least version v2.25 of util-linux is required for
3960 operation, otherwise dead-locks on device nodes may
3961 occur. Again: you need to update util-linux to at least
3962 v2.25 when updating systemd to v217.
3964 * The "multi-seat-x" tool has been removed from systemd, as
3965 its functionality has been integrated into X servers 1.16,
3966 and the tool is hence redundant. It is recommended to update
3967 display managers invoking this tool to simply invoke X
3968 directly from now on, again.
3970 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3971 message flag has been added for all of systemd's PolicyKit
3972 authenticated method calls has been added. In particular
3973 this now allows optional interactive authorization via
3974 PolicyKit for many of PID1's privileged operations such as
3975 unit file enabling and disabling.
3977 * "udevadm hwdb --update" learnt a new switch "--usr" for
3978 placing the rebuilt hardware database in /usr instead of
3979 /etc. When used only hardware database entries stored in
3980 /usr will be used, and any user database entries in /etc are
3981 ignored. This functionality is useful for vendors to ship a
3982 pre-built database on systems where local configuration is
3983 unnecessary or unlikely.
3985 * Calendar time specifications in .timer units now also
3986 understand the strings "semi-annually", "quarterly" and
3987 "minutely" as shortcuts (in addition to the preexisting
3988 "anually", "hourly", ...).
3990 * systemd-tmpfiles will now correctly create files in /dev
3991 at boot which are marked for creation only at boot. It is
3992 recommended to always create static device nodes with 'c!'
3993 and 'b!', so that they are created only at boot and not
3994 overwritten at runtime.
3996 * When the watchdog logic is used for a service (WatchdogSec=)
3997 and the watchdog timeout is hit the service will now be
3998 terminated with SIGABRT (instead of just SIGTERM), in order
3999 to make sure a proper coredump and backtrace is
4000 generated. This ensures that hanging services will result in
4001 similar coredump/backtrace behaviour as services that hit a
4004 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4005 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4006 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4007 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4008 Herrmann, David Sommerseth, David Strauss, Emil Renner
4009 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4010 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4011 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4012 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4013 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4014 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4015 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4016 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4017 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4018 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4019 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4020 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4021 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4022 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4023 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4024 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4027 — Berlin, 2014-10-28
4031 * timedated no longer reads NTP implementation unit names from
4032 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4033 implementations should add a
4035 Conflicts=systemd-timesyncd.service
4037 to their unit files to take over and replace systemd's NTP
4038 default functionality.
4040 * systemd-sysusers gained a new line type "r" for configuring
4041 which UID/GID ranges to allocate system users/groups
4042 from. Lines of type "u" may now add an additional column
4043 that specifies the home directory for the system user to be
4044 created. Also, systemd-sysusers may now optionally read user
4045 information from STDIN instead of a file. This is useful for
4046 invoking it from RPM preinst scriptlets that need to create
4047 users before the first RPM file is installed since these
4048 files might need to be owned by them. A new
4049 %sysusers_create_inline RPM macro has been introduced to do
4050 just that. systemd-sysusers now updates the shadow files as
4051 well as the user/group databases, which should enhance
4052 compatibility with certain tools like grpck.
4054 * A number of bus APIs of PID 1 now optionally consult
4055 PolicyKit to permit access for otherwise unprivileged
4056 clients under certain conditions. Note that this currently
4057 doesn't support interactive authentication yet, but this is
4058 expected to be added eventually, too.
4060 * /etc/machine-info now has new fields for configuring the
4061 deployment environment of the machine, as well as the
4062 location of the machine. hostnamectl has been updated with
4063 new command to update these fields.
4065 * systemd-timesyncd has been updated to automatically acquire
4066 NTP server information from systemd-networkd, which might
4067 have been discovered via DHCP.
4069 * systemd-resolved now includes a caching DNS stub resolver
4070 and a complete LLMNR name resolution implementation. A new
4071 NSS module "nss-resolve" has been added which can be used
4072 instead of glibc's own "nss-dns" to resolve hostnames via
4073 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4074 be resolved via systemd-resolved D-Bus APIs. In contrast to
4075 the glibc internal resolver systemd-resolved is aware of
4076 multi-homed system, and keeps DNS server and caches separate
4077 and per-interface. Queries are sent simultaneously on all
4078 interfaces that have DNS servers configured, in order to
4079 properly handle VPNs and local LANs which might resolve
4080 separate sets of domain names. systemd-resolved may acquire
4081 DNS server information from systemd-networkd automatically,
4082 which in turn might have discovered them via DHCP. A tool
4083 "systemd-resolve-host" has been added that may be used to
4084 query the DNS logic in resolved. systemd-resolved implements
4085 IDNA and automatically uses IDNA or UTF-8 encoding depending
4086 on whether classic DNS or LLMNR is used as transport. In the
4087 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4088 implementation to systemd-resolved.
4090 * A new NSS module nss-mymachines has been added, that
4091 automatically resolves the names of all local registered
4092 containers to their respective IP addresses.
4094 * A new client tool "networkctl" for systemd-networkd has been
4095 added. It currently is entirely passive and will query
4096 networking configuration from udev, rtnetlink and networkd,
4097 and present it to the user in a very friendly
4098 way. Eventually, we hope to extend it to become a full
4099 control utility for networkd.
4101 * .socket units gained a new DeferAcceptSec= setting that
4102 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4103 TCP. Similarly, support for controlling TCP keep-alive
4104 settings has been added (KeepAliveTimeSec=,
4105 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4106 turning off Nagle's algorithm on TCP has been added
4109 * logind learned a new session type "web", for use in projects
4110 like Cockpit which register web clients as PAM sessions.
4112 * timer units with at least one OnCalendar= setting will now
4113 be started only after time-sync.target has been
4114 reached. This way they will not elapse before the system
4115 clock has been corrected by a local NTP client or
4116 similar. This is particular useful on RTC-less embedded
4117 machines, that come up with an invalid system clock.
4119 * systemd-nspawn's --network-veth= switch should now result in
4120 stable MAC addresses for both the outer and the inner side
4123 * systemd-nspawn gained a new --volatile= switch for running
4124 container instances with /etc or /var unpopulated.
4126 * The kdbus client code has been updated to use the new Linux
4127 3.17 memfd subsystem instead of the old kdbus-specific one.
4129 * systemd-networkd's DHCP client and server now support
4130 FORCERENEW. There are also new configuration options to
4131 configure the vendor client identifier and broadcast mode
4134 * systemd will no longer inform the kernel about the current
4135 timezone, as this is necessarily incorrect and racy as the
4136 kernel has no understanding of DST and similar
4137 concepts. This hence means FAT timestamps will be always
4138 considered UTC, similar to what Android is already
4139 doing. Also, when the RTC is configured to the local time
4140 (rather than UTC) systemd will never synchronize back to it,
4141 as this might confuse Windows at a later boot.
4143 * systemd-analyze gained a new command "verify" for offline
4144 validation of unit files.
4146 * systemd-networkd gained support for a couple of additional
4147 settings for bonding networking setups. Also, the metric for
4148 statically configured routes may now be configured. For
4149 network interfaces where this is appropriate the peer IP
4150 address may now be configured.
4152 * systemd-networkd's DHCP client will no longer request
4153 broadcasting by default, as this tripped up some networks.
4154 For hardware where broadcast is required the feature should
4155 be switched back on using RequestBroadcast=yes.
4157 * systemd-networkd will now set up IPv4LL addresses (when
4158 enabled) even if DHCP is configured successfully.
4160 * udev will now default to respect network device names given
4161 by the kernel when the kernel indicates that these are
4162 predictable. This behavior can be tweaked by changing
4163 NamePolicy= in the relevant .link file.
4165 * A new library systemd-terminal has been added that
4166 implements full TTY stream parsing and rendering. This
4167 library is supposed to be used later on for implementing a
4168 full userspace VT subsystem, replacing the current kernel
4171 * A new tool systemd-journal-upload has been added to push
4172 journal data to a remote system running
4173 systemd-journal-remote.
4175 * journald will no longer forward all local data to another
4176 running syslog daemon. This change has been made because
4177 rsyslog (which appears to be the most commonly used syslog
4178 implementation these days) no longer makes use of this, and
4179 instead pulls the data out of the journal on its own. Since
4180 forwarding the messages to a non-existent syslog server is
4181 more expensive than we assumed we have now turned this
4182 off. If you run a syslog server that is not a recent rsyslog
4183 version, you have to turn this option on again
4184 (ForwardToSyslog= in journald.conf).
4186 * journald now optionally supports the LZ4 compressor for
4187 larger journal fields. This compressor should perform much
4188 better than XZ which was the previous default.
4190 * machinectl now shows the IP addresses of local containers,
4191 if it knows them, plus the interface name of the container.
4193 * A new tool "systemd-escape" has been added that makes it
4194 easy to escape strings to build unit names and similar.
4196 * sd_notify() messages may now include a new ERRNO= field
4197 which is parsed and collected by systemd and shown among the
4198 "systemctl status" output for a service.
4200 * A new component "systemd-firstboot" has been added that
4201 queries the most basic systemd information (timezone,
4202 hostname, root password) interactively on first
4203 boot. Alternatively it may also be used to provision these
4204 things offline on OS images installed into directories.
4206 * The default sysctl.d/ snippets will now set
4208 net.ipv4.conf.default.promote_secondaries=1
4210 This has the benefit of no flushing secondary IP addresses
4211 when primary addresses are removed.
4213 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4214 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4215 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4216 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4217 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4218 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4219 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4220 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4221 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4222 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4223 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4224 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4225 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4226 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4227 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4229 — Berlin, 2014-08-19
4233 * A new tool systemd-sysusers has been added. This tool
4234 creates system users and groups in /etc/passwd and
4235 /etc/group, based on static declarative system user/group
4236 definitions in /usr/lib/sysusers.d/. This is useful to
4237 enable factory resets and volatile systems that boot up with
4238 an empty /etc directory, and thus need system users and
4239 groups created during early boot. systemd now also ships
4240 with two default sysusers.d/ files for the most basic
4241 users and groups systemd and the core operating system
4244 * A new tmpfiles snippet has been added that rebuilds the
4245 essential files in /etc on boot, should they be missing.
4247 * A directive for ensuring automatic clean-up of
4248 /var/cache/man/ has been removed from the default
4249 configuration. This line should now be shipped by the man
4250 implementation. The necessary change has been made to the
4251 man-db implementation. Note that you need to update your man
4252 implementation to one that ships this line, otherwise no
4253 automatic clean-up of /var/cache/man will take place.
4255 * A new condition ConditionNeedsUpdate= has been added that
4256 may conditionalize services to only run when /etc or /var
4257 are "older" than the vendor operating system resources in
4258 /usr. This is useful for reconstructing or updating /etc
4259 after an offline update of /usr or a factory reset, on the
4260 next reboot. Services that want to run once after such an
4261 update or reset should use this condition and order
4262 themselves before the new systemd-update-done.service, which
4263 will mark the two directories as fully updated. A number of
4264 service files have been added making use of this, to rebuild
4265 the udev hardware database, the journald message catalog and
4266 dynamic loader cache (ldconfig). The systemd-sysusers tool
4267 described above also makes use of this now. With this in
4268 place it is now possible to start up a minimal operating
4269 system with /etc empty cleanly. For more information on the
4270 concepts involved see this recent blog story:
4272 http://0pointer.de/blog/projects/stateless.html
4274 * A new system group "input" has been introduced, and all
4275 input device nodes get this group assigned. This is useful
4276 for system-level software to get access to input devices. It
4277 complements what is already done for "audio" and "video".
4279 * systemd-networkd learnt minimal DHCPv4 server support in
4280 addition to the existing DHCPv4 client support. It also
4281 learnt DHCPv6 client and IPv6 Router Solicitation client
4282 support. The DHCPv4 client gained support for static routes
4283 passed in from the server. Note that the [DHCPv4] section
4284 known in older systemd-networkd versions has been renamed to
4285 [DHCP] and is now also used by the DHCPv6 client. Existing
4286 .network files using settings of this section should be
4287 updated, though compatibility is maintained. Optionally, the
4288 client hostname may now be sent to the DHCP server.
4290 * networkd gained support for vxlan virtual networks as well
4291 as tun/tap and dummy devices.
4293 * networkd gained support for automatic allocation of address
4294 ranges for interfaces from a system-wide pool of
4295 addresses. This is useful for dynamically managing a large
4296 number of interfaces with a single network configuration
4297 file. In particular this is useful to easily assign
4298 appropriate IP addresses to the veth links of a large number
4299 of nspawn instances.
4301 * RPM macros for processing sysusers, sysctl and binfmt
4302 drop-in snippets at package installation time have been
4305 * The /etc/os-release file should now be placed in
4306 /usr/lib/os-release. The old location is automatically
4307 created as symlink. /usr/lib is the more appropriate
4308 location of this file, since it shall actually describe the
4309 vendor operating system shipped in /usr, and not the
4310 configuration stored in /etc.
4312 * .mount units gained a new boolean SloppyOptions= setting
4313 that maps to mount(8)'s -s option which enables permissive
4314 parsing of unknown mount options.
4316 * tmpfiles learnt a new "L+" directive which creates a symlink
4317 but (unlike "L") deletes a pre-existing file first, should
4318 it already exist and not already be the correct
4319 symlink. Similarly, "b+", "c+" and "p+" directives have been
4320 added as well, which create block and character devices, as
4321 well as fifos in the filesystem, possibly removing any
4322 pre-existing files of different types.
4324 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4325 'argument' field (which so far specified the source to
4326 symlink/copy the files from) is now optional. If omitted the
4327 same file os copied from /usr/share/factory/ suffixed by the
4328 full destination path. This is useful for populating /etc
4329 with essential files, by copying them from vendor defaults
4330 shipped in /usr/share/factory/etc.
4332 * A new command "systemctl preset-all" has been added that
4333 applies the service preset settings to all installed unit
4334 files. A new switch --preset-mode= has been added that
4335 controls whether only enable or only disable operations
4338 * A new command "systemctl is-system-running" has been added
4339 that allows checking the overall state of the system, for
4340 example whether it is fully up and running.
4342 * When the system boots up with an empty /etc, the equivalent
4343 to "systemctl preset-all" is executed during early boot, to
4344 make sure all default services are enabled after a factory
4347 * systemd now contains a minimal preset file that enables the
4348 most basic services systemd ships by default.
4350 * Unit files' [Install] section gained a new DefaultInstance=
4351 field for defining the default instance to create if a
4352 template unit is enabled with no instance specified.
4354 * A new passive target cryptsetup-pre.target has been added
4355 that may be used by services that need to make they run and
4356 finish before the first LUKS cryptographic device is set up.
4358 * The /dev/loop-control and /dev/btrfs-control device nodes
4359 are now owned by the "disk" group by default, opening up
4360 access to this group.
4362 * systemd-coredump will now automatically generate a
4363 stack trace of all core dumps taking place on the system,
4364 based on elfutils' libdw library. This stack trace is logged
4367 * systemd-coredump may now optionally store coredumps directly
4368 on disk (in /var/lib/systemd/coredump, possibly compressed),
4369 instead of storing them unconditionally in the journal. This
4370 mode is the new default. A new configuration file
4371 /etc/systemd/coredump.conf has been added to configure this
4372 and other parameters of systemd-coredump.
4374 * coredumpctl gained a new "info" verb to show details about a
4375 specific coredump. A new switch "-1" has also been added
4376 that makes sure to only show information about the most
4377 recent entry instead of all entries. Also, as the tool is
4378 generally useful now the "systemd-" prefix of the binary
4379 name has been removed. Distributions that want to maintain
4380 compatibility with the old name should add a symlink from
4381 the old name to the new name.
4383 * journald's SplitMode= now defaults to "uid". This makes sure
4384 that unprivileged users can access their own coredumps with
4385 coredumpctl without restrictions.
4387 * New kernel command line options "systemd.wants=" (for
4388 pulling an additional unit during boot), "systemd.mask="
4389 (for masking a specific unit for the boot), and
4390 "systemd.debug-shell" (for enabling the debug shell on tty9)
4391 have been added. This is implemented in the new generator
4392 "systemd-debug-generator".
4394 * systemd-nspawn will now by default filter a couple of
4395 syscalls for containers, among them those required for
4396 kernel module loading, direct x86 IO port access, swap
4397 management, and kexec. Most importantly though
4398 open_by_handle_at() is now prohibited for containers,
4399 closing a hole similar to a recently discussed vulnerability
4400 in docker regarding access to files on file hierarchies the
4401 container should normally not have access to. Note that, for
4402 nspawn, we generally make no security claims anyway (and
4403 this is explicitly documented in the man page), so this is
4404 just a fix for one of the most obvious problems.
4406 * A new man page file-hierarchy(7) has been added that
4407 contains a minimized, modernized version of the file system
4408 layout systemd expects, similar in style to the FHS
4409 specification or hier(5). A new tool systemd-path(1) has
4410 been added to query many of these paths for the local
4413 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4414 longer done. Since the directory now has a per-user size
4415 limit, and is cleaned on logout this appears unnecessary,
4416 in particular since this now brings the lifecycle of this
4417 directory closer in line with how IPC objects are handled.
4419 * systemd.pc now exports a number of additional directories,
4420 including $libdir (which is useful to identify the library
4421 path for the primary architecture of the system), and a
4422 couple of drop-in directories.
4424 * udev's predictable network interface names now use the dev_port
4425 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4426 distinguish between ports of the same PCI function. dev_id should
4427 only be used for ports using the same HW address, hence the need
4430 * machined has been updated to export the OS version of a
4431 container (read from /etc/os-release and
4432 /usr/lib/os-release) on the bus. This is now shown in
4433 "machinectl status" for a machine.
4435 * A new service setting RestartForceExitStatus= has been
4436 added. If configured to a set of exit signals or process
4437 return values, the service will be restarted when the main
4438 daemon process exits with any of them, regardless of the
4441 * systemctl's -H switch for connecting to remote systemd
4442 machines has been extended so that it may be used to
4443 directly connect to a specific container on the
4444 host. "systemctl -H root@foobar:waldi" will now connect as
4445 user "root" to host "foobar", and then proceed directly to
4446 the container named "waldi". Note that currently you have to
4447 authenticate as user "root" for this to work, as entering
4448 containers is a privileged operation.
4450 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4451 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4452 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4453 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4454 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4455 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4456 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4457 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4458 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4459 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4460 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4461 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4463 — Berlin, 2014-07-03
4467 * As an experimental feature, udev now tries to lock the
4468 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4469 executes events for the disk or any of its partitions.
4470 Applications like partitioning programs can lock the
4471 disk device node (flock(LOCK_EX)) and claim temporary
4472 device ownership that way; udev will entirely skip all event
4473 handling for this disk and its partitions. If the disk
4474 was opened for writing, the close will trigger a partition
4475 table rescan in udev's "watch" facility, and if needed
4476 synthesize "change" events for the disk and all its partitions.
4477 This is now unconditionally enabled, and if it turns out to
4478 cause major problems, we might turn it on only for specific
4479 devices, or might need to disable it entirely. Device Mapper
4480 devices are excluded from this logic.
4482 * We temporarily dropped the "-l" switch for fsck invocations,
4483 since they collide with the flock() logic above. util-linux
4484 upstream has been changed already to avoid this conflict,
4485 and we will readd "-l" as soon as util-linux with this
4486 change has been released.
4488 * The dependency on libattr has been removed. Since a long
4489 time, the extended attribute calls have moved to glibc, and
4490 libattr is thus unnecessary.
4492 * Virtualization detection works without privileges now. This
4493 means the systemd-detect-virt binary no longer requires
4494 CAP_SYS_PTRACE file capabilities, and our daemons can run
4495 with fewer privileges.
4497 * systemd-networkd now runs under its own "systemd-network"
4498 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4499 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4500 loses the ability to write to files owned by root this way.
4502 * Similarly, systemd-resolved now runs under its own
4503 "systemd-resolve" user with no capabilities remaining.
4505 * Similarly, systemd-bus-proxyd now runs under its own
4506 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4508 * systemd-networkd gained support for setting up "veth"
4509 virtual Ethernet devices for container connectivity, as well
4510 as GRE and VTI tunnels.
4512 * systemd-networkd will no longer automatically attempt to
4513 manually load kernel modules necessary for certain tunnel
4514 transports. Instead, it is assumed the kernel loads them
4515 automatically when required. This only works correctly on
4516 very new kernels. On older kernels, please consider adding
4517 the kernel modules to /etc/modules-load.d/ as a work-around.
4519 * The resolv.conf file systemd-resolved generates has been
4520 moved to /run/systemd/resolve/. If you have a symlink from
4521 /etc/resolv.conf, it might be necessary to correct it.
4523 * Two new service settings, ProtectHome= and ProtectSystem=,
4524 have been added. When enabled, they will make the user data
4525 (such as /home) inaccessible or read-only and the system
4526 (such as /usr) read-only, for specific services. This allows
4527 very light-weight per-service sandboxing to avoid
4528 modifications of user data or system files from
4529 services. These two new switches have been enabled for all
4530 of systemd's long-running services, where appropriate.
4532 * Socket units gained new SocketUser= and SocketGroup=
4533 settings to set the owner user and group of AF_UNIX sockets
4534 and FIFOs in the file system.
4536 * Socket units gained a new RemoveOnStop= setting. If enabled,
4537 all FIFOS and sockets in the file system will be removed
4538 when the specific socket unit is stopped.
4540 * Socket units gained a new Symlinks= setting. It takes a list
4541 of symlinks to create to file system sockets or FIFOs
4542 created by the specific Unix sockets. This is useful to
4543 manage symlinks to socket nodes with the same life-cycle as
4546 * The /dev/log socket and /dev/initctl FIFO have been moved to
4547 /run, and have been replaced by symlinks. This allows
4548 connecting to these facilities even if PrivateDevices=yes is
4549 used for a service (which makes /dev/log itself unavailable,
4550 but /run is left). This also has the benefit of ensuring
4551 that /dev only contains device nodes, directories and
4552 symlinks, and nothing else.
4554 * sd-daemon gained two new calls sd_pid_notify() and
4555 sd_pid_notifyf(). They are similar to sd_notify() and
4556 sd_notifyf(), but allow overriding of the source PID of
4557 notification messages if permissions permit this. This is
4558 useful to send notify messages on behalf of a different
4559 process (for example, the parent process). The
4560 systemd-notify tool has been updated to make use of this
4561 when sending messages (so that notification messages now
4562 originate from the shell script invoking systemd-notify and
4563 not the systemd-notify process itself. This should minimize
4564 a race where systemd fails to associate notification
4565 messages to services when the originating process already
4568 * A new "on-abnormal" setting for Restart= has been added. If
4569 set, it will result in automatic restarts on all "abnormal"
4570 reasons for a process to exit, which includes unclean
4571 signals, core dumps, timeouts and watchdog timeouts, but
4572 does not include clean and unclean exit codes or clean
4573 signals. Restart=on-abnormal is an alternative for
4574 Restart=on-failure for services that shall be able to
4575 terminate and avoid restarts on certain errors, by
4576 indicating so with an unclean exit code. Restart=on-failure
4577 or Restart=on-abnormal is now the recommended setting for
4578 all long-running services.
4580 * If the InaccessibleDirectories= service setting points to a
4581 mount point (or if there are any submounts contained within
4582 it), it is now attempted to completely unmount it, to make
4583 the file systems truly unavailable for the respective
4586 * The ReadOnlyDirectories= service setting and
4587 systemd-nspawn's --read-only parameter are now recursively
4588 applied to all submounts, too.
4590 * Mount units may now be created transiently via the bus APIs.
4592 * The support for SysV and LSB init scripts has been removed
4593 from the systemd daemon itself. Instead, it is now
4594 implemented as a generator that creates native systemd units
4595 from these scripts when needed. This enables us to remove a
4596 substantial amount of legacy code from PID 1, following the
4597 fact that many distributions only ship a very small number
4598 of LSB/SysV init scripts nowadays.
4600 * Privileged Xen (dom0) domains are not considered
4601 virtualization anymore by the virtualization detection
4602 logic. After all, they generally have unrestricted access to
4603 the hardware and usually are used to manage the unprivileged
4606 * systemd-tmpfiles gained a new "C" line type, for copying
4607 files or entire directories.
4609 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4610 lines. So far, they have been non-globbing versions of the
4611 latter, and have thus been redundant. In future, it is
4612 recommended to only use "z". "m" has hence been removed
4613 from the documentation, even though it stays supported.
4615 * A tmpfiles snippet to recreate the most basic structure in
4616 /var has been added. This is enough to create the /var/run →
4617 /run symlink and create a couple of structural
4618 directories. This allows systems to boot up with an empty or
4619 volatile /var. Of course, while with this change, the core OS
4620 now is capable with dealing with a volatile /var, not all
4621 user services are ready for it. However, we hope that sooner
4622 or later, many service daemons will be changed upstream so
4623 that they are able to automatically create their necessary
4624 directories in /var at boot, should they be missing. This is
4625 the first step to allow state-less systems that only require
4626 the vendor image for /usr to boot.
4628 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4629 empty tmpfs instance to a specific directory. This is
4630 particularly useful for making use of the automatic
4631 reconstruction of /var (see above), by passing --tmpfs=/var.
4633 * Access modes specified in tmpfiles snippets may now be
4634 prefixed with "~", which indicates that they shall be masked
4635 by whether the existing file or directory is currently
4636 writable, readable or executable at all. Also, if specified,
4637 the sgid/suid/sticky bits will be masked for all
4640 * A new passive target unit "network-pre.target" has been
4641 added which is useful for services that shall run before any
4642 network is configured, for example firewall scripts.
4644 * The "floppy" group that previously owned the /dev/fd*
4645 devices is no longer used. The "disk" group is now used
4646 instead. Distributions should probably deprecate usage of
4649 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4650 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4651 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4652 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4653 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4654 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4657 — Berlin, 2014-06-11
4661 * A new "systemd-timesyncd" daemon has been added for
4662 synchronizing the system clock across the network. It
4663 implements an SNTP client. In contrast to NTP
4664 implementations such as chrony or the NTP reference server,
4665 this only implements a client side, and does not bother with
4666 the full NTP complexity, focusing only on querying time from
4667 one remote server and synchronizing the local clock to
4668 it. Unless you intend to serve NTP to networked clients or
4669 want to connect to local hardware clocks, this simple NTP
4670 client should be more than appropriate for most
4671 installations. The daemon runs with minimal privileges, and
4672 has been hooked up with networkd to only operate when
4673 network connectivity is available. The daemon saves the
4674 current clock to disk every time a new NTP sync has been
4675 acquired, and uses this to possibly correct the system clock
4676 early at bootup, in order to accommodate for systems that
4677 lack an RTC such as the Raspberry Pi and embedded devices,
4678 and to make sure that time monotonically progresses on these
4679 systems, even if it is not always correct. To make use of
4680 this daemon, a new system user and group "systemd-timesync"
4681 needs to be created on installation of systemd.
4683 * The queue "seqnum" interface of libudev has been disabled, as
4684 it was generally incompatible with device namespacing as
4685 sequence numbers of devices go "missing" if the devices are
4686 part of a different namespace.
4688 * "systemctl list-timers" and "systemctl list-sockets" gained
4689 a --recursive switch for showing units of these types also
4690 for all local containers, similar in style to the already
4691 supported --recursive switch for "systemctl list-units".
4693 * A new RebootArgument= setting has been added for service
4694 units, which may be used to specify a kernel reboot argument
4695 to use when triggering reboots with StartLimitAction=.
4697 * A new FailureAction= setting has been added for service
4698 units which may be used to specify an operation to trigger
4699 when a service fails. This works similarly to
4700 StartLimitAction=, but unlike it, controls what is done
4701 immediately rather than only after several attempts to
4702 restart the service in question.
4704 * hostnamed got updated to also expose the kernel name,
4705 release, and version on the bus. This is useful for
4706 executing commands like hostnamectl with the -H switch.
4707 systemd-analyze makes use of this to properly display
4708 details when running non-locally.
4710 * The bootchart tool can now show cgroup information in the
4711 graphs it generates.
4713 * The CFS CPU quota cgroup attribute is now exposed for
4714 services. The new CPUQuota= switch has been added for this
4715 which takes a percentage value. Setting this will have the
4716 result that a service may never get more CPU time than the
4717 specified percentage, even if the machine is otherwise idle.
4719 * systemd-networkd learned IPIP and SIT tunnel support.
4721 * LSB init scripts exposing a dependency on $network will now
4722 get a dependency on network-online.target rather than simply
4723 network.target. This should bring LSB handling closer to
4724 what it was on SysV systems.
4726 * A new fsck.repair= kernel option has been added to control
4727 how fsck shall deal with unclean file systems at boot.
4729 * The (.ini) configuration file parser will now silently
4730 ignore sections whose name begins with "X-". This may be
4731 used to maintain application-specific extension sections in unit
4734 * machined gained a new API to query the IP addresses of
4735 registered containers. "machinectl status" has been updated
4736 to show these addresses in its output.
4738 * A new call sd_uid_get_display() has been added to the
4739 sd-login APIs for querying the "primary" session of a
4740 user. The "primary" session of the user is elected from the
4741 user's sessions and generally a graphical session is
4742 preferred over a text one.
4744 * A minimal systemd-resolved daemon has been added. It
4745 currently simply acts as a companion to systemd-networkd and
4746 manages resolv.conf based on per-interface DNS
4747 configuration, possibly supplied via DHCP. In the long run
4748 we hope to extend this into a local DNSSEC enabled DNS and
4751 * The systemd-networkd-wait-online tool is now enabled by
4752 default. It will delay network-online.target until a network
4753 connection has been configured. The tool primarily integrates
4754 with networkd, but will also make a best effort to make sense
4755 of network configuration performed in some other way.
4757 * Two new service options StartupCPUShares= and
4758 StartupBlockIOWeight= have been added that work similarly to
4759 CPUShares= and BlockIOWeight= however only apply during
4760 system startup. This is useful to prioritize certain services
4761 differently during bootup than during normal runtime.
4763 * hostnamed has been changed to prefer the statically
4764 configured hostname in /etc/hostname (unless set to
4765 'localhost' or empty) over any dynamic one supplied by
4766 dhcp. With this change, the rules for picking the hostname
4767 match more closely the rules of other configuration settings
4768 where the local administrator's configuration in /etc always
4769 overrides any other settings.
4771 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4772 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4773 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4774 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4775 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4776 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4777 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4778 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4779 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4780 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4781 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4782 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4783 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4784 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4785 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4786 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4789 — Beijing, 2014-05-28
4793 * When restoring the screen brightness at boot, stay away from
4794 the darkest setting or from the lowest 5% of the available
4795 range, depending on which is the larger value of both. This
4796 should effectively protect the user from rebooting into a
4797 black screen, should the brightness have been set to minimum
4800 * sd-login gained a new sd_machine_get_class() call to
4801 determine the class ("vm" or "container") of a machine
4802 registered with machined.
4804 * sd-login gained new calls
4805 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4806 to query the identity of the peer of a local AF_UNIX
4807 connection. They operate similarly to their sd_pid_get_xyz()
4810 * PID 1 will now maintain a system-wide system state engine
4811 with the states "starting", "running", "degraded",
4812 "maintenance", "stopping". These states are bound to system
4813 startup, normal runtime, runtime with at least one failed
4814 service, rescue/emergency mode and system shutdown. This
4815 state is shown in the "systemctl status" output when no unit
4816 name is passed. It is useful to determine system state, in
4817 particularly when doing so for many systems or containers at
4820 * A new command "list-machines" has been added to "systemctl"
4821 that lists all local OS containers and shows their system
4822 state (see above), if systemd runs inside of them.
4824 * systemctl gained a new "-r" switch to recursively enumerate
4825 units on all local containers, when used with the
4826 "list-unit" command (which is the default one that is
4827 executed when no parameters are specified).
4829 * The GPT automatic partition discovery logic will now honour
4830 two GPT partition flags: one may be set on a partition to
4831 cause it to be mounted read-only, and the other may be set
4832 on a partition to ignore it during automatic discovery.
4834 * Two new GPT type UUIDs have been added for automatic root
4835 partition discovery, for 32-bit and 64-bit ARM. This is not
4836 particularly useful for discovering the root directory on
4837 these architectures during bare-metal boots (since UEFI is
4838 not common there), but still very useful to allow booting of
4839 ARM disk images in nspawn with the -i option.
4841 * MAC addresses of interfaces created with nspawn's
4842 --network-interface= switch will now be generated from the
4843 machine name, and thus be stable between multiple invocations
4846 * logind will now automatically remove all IPC objects owned
4847 by a user if she or he fully logs out. This makes sure that
4848 users who are logged out cannot continue to consume IPC
4849 resources. This covers SysV memory, semaphores and message
4850 queues as well as POSIX shared memory and message
4851 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4852 limits. With this functionality, that is corrected. This may
4853 be turned off by using the RemoveIPC= switch of logind.conf.
4855 * The systemd-machine-id-setup and tmpfiles tools gained a
4856 --root= switch to operate on a specific root directory,
4859 * journald can now forward logged messages to the TTYs of all
4860 logged in users ("wall"). This is the default for all
4861 emergency messages now.
4863 * A new tool systemd-journal-remote has been added to stream
4864 journal log messages across the network.
4866 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4867 controller trees are mounted into it. Note that the
4868 directories mounted beneath it are not read-only. This is a
4869 security measure and is particularly useful because glibc
4870 actually includes a search logic to pick any tmpfs it can
4871 find to implement shm_open() if /dev/shm is not available
4872 (which it might very well be in namespaced setups).
4874 * machinectl gained a new "poweroff" command to cleanly power
4875 down a local OS container.
4877 * The PrivateDevices= unit file setting will now also drop the
4878 CAP_MKNOD capability from the capability bound set, and
4879 imply DevicePolicy=closed.
4881 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4882 comprehensively on all long-running systemd services where
4883 this is appropriate.
4885 * systemd-udevd will now run in a disassociated mount
4886 namespace. To mount directories from udev rules, make sure to
4887 pull in mount units via SYSTEMD_WANTS properties.
4889 * The kdbus support gained support for uploading policy into
4890 the kernel. sd-bus gained support for creating "monitoring"
4891 connections that can eavesdrop into all bus communication
4892 for debugging purposes.
4894 * Timestamps may now be specified in seconds since the UNIX
4895 epoch Jan 1st, 1970 by specifying "@" followed by the value
4898 * Native tcpwrap support in systemd has been removed. tcpwrap
4899 is old code, not really maintained anymore and has serious
4900 shortcomings, and better options such as firewalls
4901 exist. For setups that require tcpwrap usage, please
4902 consider invoking your socket-activated service via tcpd,
4903 like on traditional inetd.
4905 * A new system.conf configuration option
4906 DefaultTimerAccuracySec= has been added that controls the
4907 default AccuracySec= setting of .timer units.
4909 * Timer units gained a new WakeSystem= switch. If enabled,
4910 timers configured this way will cause the system to resume
4911 from system suspend (if the system supports that, which most
4914 * Timer units gained a new Persistent= switch. If enabled,
4915 timers configured this way will save to disk when they have
4916 been last triggered. This information is then used on next
4917 reboot to possible execute overdue timer events, that
4918 could not take place because the system was powered off.
4919 This enables simple anacron-like behaviour for timer units.
4921 * systemctl's "list-timers" will now also list the time a
4922 timer unit was last triggered in addition to the next time
4923 it will be triggered.
4925 * systemd-networkd will now assign predictable IPv4LL
4926 addresses to its local interfaces.
4928 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4929 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4930 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4931 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4932 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4933 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4934 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4935 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4938 — Berlin, 2014-03-25
4942 * A new unit file setting RestrictAddressFamilies= has been
4943 added to restrict which socket address families unit
4944 processes gain access to. This takes address family names
4945 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4946 attack surface of services via exotic protocol stacks. This
4947 is built on seccomp system call filters.
4949 * Two new unit file settings RuntimeDirectory= and
4950 RuntimeDirectoryMode= have been added that may be used to
4951 manage a per-daemon runtime directories below /run. This is
4952 an alternative for setting up directory permissions with
4953 tmpfiles snippets, and has the advantage that the runtime
4954 directory's lifetime is bound to the daemon runtime and that
4955 the daemon starts up with an empty directory each time. This
4956 is particularly useful when writing services that drop
4957 privileges using the User= or Group= setting.
4959 * The DeviceAllow= unit setting now supports globbing for
4960 matching against device group names.
4962 * The systemd configuration file system.conf gained new
4963 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4964 DefaultMemoryAccounting= to globally turn on/off accounting
4965 for specific resources (cgroups) for all units. These
4966 settings may still be overridden individually in each unit
4969 * systemd-gpt-auto-generator is now able to discover /srv and
4970 root partitions in addition to /home and swap partitions. It
4971 also supports LUKS-encrypted partitions now. With this in
4972 place, automatic discovery of partitions to mount following
4973 the Discoverable Partitions Specification
4974 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4975 is now a lot more complete. This allows booting without
4976 /etc/fstab and without root= on the kernel command line on
4977 systems prepared appropriately.
4979 * systemd-nspawn gained a new --image= switch which allows
4980 booting up disk images and Linux installations on any block
4981 device that follow the Discoverable Partitions Specification
4982 (see above). This means that installations made with
4983 appropriately updated installers may now be started and
4984 deployed using container managers, completely
4985 unmodified. (We hope that libvirt-lxc will add support for
4986 this feature soon, too.)
4988 * systemd-nspawn gained a new --network-macvlan= setting to
4989 set up a private macvlan interface for the
4990 container. Similarly, systemd-networkd gained a new
4991 Kind=macvlan setting in .netdev files.
4993 * systemd-networkd now supports configuring local addresses
4996 * A new tool systemd-network-wait-online has been added to
4997 synchronously wait for network connectivity using
5000 * The sd-bus.h bus API gained a new sd_bus_track object for
5001 tracking the life-cycle of bus peers. Note that sd-bus.h is
5002 still not a public API though (unless you specify
5003 --enable-kdbus on the configure command line, which however
5004 voids your warranty and you get no API stability guarantee).
5006 * The $XDG_RUNTIME_DIR runtime directories for each user are
5007 now individual tmpfs instances, which has the benefit of
5008 introducing separate pools for each user, with individual
5009 size limits, and thus making sure that unprivileged clients
5010 can no longer negatively impact the system or other users by
5011 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5012 RuntimeDirectorySize= has been introduced that allows
5013 controlling the default size limit for all users. It
5014 defaults to 10% of the available physical memory. This is no
5015 replacement for quotas on tmpfs though (which the kernel
5016 still does not support), as /dev/shm and /tmp are still
5017 shared resources used by both the system and unprivileged
5020 * logind will now automatically turn off automatic suspending
5021 on laptop lid close when more than one display is
5022 connected. This was previously expected to be implemented
5023 individually in desktop environments (such as GNOME),
5024 however has been added to logind now, in order to fix a
5025 boot-time race where a desktop environment might not have
5026 been started yet and thus not been able to take an inhibitor
5027 lock at the time where logind already suspends the system
5028 due to a closed lid.
5030 * logind will now wait at least 30s after each system
5031 suspend/resume cycle, and 3min after system boot before
5032 suspending the system due to a closed laptop lid. This
5033 should give USB docking stations and similar enough time to
5034 be probed and configured after system resume and boot in
5035 order to then act as suspend blocker.
5037 * systemd-run gained a new --property= setting which allows
5038 initialization of resource control properties (and others)
5039 for the created scope or service unit. Example: "systemd-run
5040 --property=BlockIOWeight=10 updatedb" may be used to run
5041 updatedb at a low block IO scheduling weight.
5043 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5044 now also work in --scope mode.
5046 * When systemd is compiled with kdbus support, basic support
5047 for enforced policies is now in place. (Note that enabling
5048 kdbus still voids your warranty and no API compatibility
5051 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5052 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5053 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5054 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5055 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5056 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5057 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5058 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5059 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5060 Zbigniew Jędrzejewski-Szmek
5062 — Berlin, 2014-03-12
5066 * systemd will now relabel /dev after loading the SMACK policy
5067 according to SMACK rules.
5069 * A new unit file option AppArmorProfile= has been added to
5070 set the AppArmor profile for the processes of a unit.
5072 * A new condition check ConditionArchitecture= has been added
5073 to conditionalize units based on the system architecture, as
5074 reported by uname()'s "machine" field.
5076 * systemd-networkd now supports matching on the system
5077 virtualization, architecture, kernel command line, host name
5080 * logind is now a lot more aggressive when suspending the
5081 machine due to a closed laptop lid. Instead of acting only
5082 on the lid close action, it will continuously watch the lid
5083 status and act on it. This is useful for laptops where the
5084 power button is on the outside of the chassis so that it can
5085 be reached without opening the lid (such as the Lenovo
5086 Yoga). On those machines, logind will now immediately
5087 re-suspend the machine if the power button has been
5088 accidentally pressed while the laptop was suspended and in a
5089 backpack or similar.
5091 * logind will now watch SW_DOCK switches and inhibit reaction
5092 to the lid switch if it is pressed. This means that logind
5093 will not suspend the machine anymore if the lid is closed
5094 and the system is docked, if the laptop supports SW_DOCK
5095 notifications via the input layer. Note that ACPI docking
5096 stations do not generate this currently. Also note that this
5097 logic is usually not fully sufficient and Desktop
5098 Environments should take a lid switch inhibitor lock when an
5099 external display is connected, as systemd will not watch
5102 * nspawn will now make use of the devices cgroup controller by
5103 default, and only permit creation of and access to the usual
5104 API device nodes like /dev/null or /dev/random, as well as
5105 access to (but not creation of) the pty devices.
5107 * We will now ship a default .network file for
5108 systemd-networkd that automatically configures DHCP for
5109 network interfaces created by nspawn's --network-veth or
5110 --network-bridge= switches.
5112 * systemd will now understand the usual M, K, G, T suffixes
5113 according to SI conventions (i.e. to the base 1000) when
5114 referring to throughput and hardware metrics. It will stay
5115 with IEC conventions (i.e. to the base 1024) for software
5116 metrics, according to what is customary according to
5117 Wikipedia. We explicitly document which base applies for
5118 each configuration option.
5120 * The DeviceAllow= setting in unit files now supports a syntax
5121 to whitelist an entire group of devices node majors at once,
5122 based on the /proc/devices listing. For example, with the
5123 string "char-pts", it is now possible to whitelist all
5124 current and future pseudo-TTYs at once.
5126 * sd-event learned a new "post" event source. Event sources of
5127 this type are triggered by the dispatching of any event
5128 source of a type that is not "post". This is useful for
5129 implementing clean-up and check event sources that are
5130 triggered by other work being done in the program.
5132 * systemd-networkd is no longer statically enabled, but uses
5133 the usual [Install] sections so that it can be
5134 enabled/disabled using systemctl. It still is enabled by
5137 * When creating a veth interface pair with systemd-nspawn, the
5138 host side will now be prefixed with "vb-" if
5139 --network-bridge= is used, and with "ve-" if --network-veth
5140 is used. This way, it is easy to distinguish these cases on
5141 the host, for example to apply different configuration to
5142 them with systemd-networkd.
5144 * The compatibility libraries for libsystemd-journal.so,
5145 libsystem-id128.so, libsystemd-login.so and
5146 libsystemd-daemon.so do not make use of IFUNC
5147 anymore. Instead, we now build libsystemd.so multiple times
5148 under these alternative names. This means that the footprint
5149 is drastically increased, but given that these are
5150 transitional compatibility libraries, this should not matter
5151 much. This change has been made necessary to support the ARM
5152 platform for these compatibility libraries, as the ARM
5153 toolchain is not really at the same level as the toolchain
5154 for other architectures like x86 and does not support
5155 IFUNC. Please make sure to use --enable-compat-libs only
5156 during a transitional period!
5158 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5159 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5160 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5161 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5162 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5163 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5164 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5165 Zbigniew Jędrzejewski-Szmek
5167 — Berlin, 2014-02-24
5171 * A new component "systemd-networkd" has been added that can
5172 be used to configure local network interfaces statically or
5173 via DHCP. It is capable of bringing up bridges, VLANs, and
5174 bonding. Currently, no hook-ups for interactive network
5175 configuration are provided. Use this for your initrd,
5176 container, embedded, or server setup if you need a simple,
5177 yet powerful, network configuration solution. This
5178 configuration subsystem is quite nifty, as it allows wildcard
5179 hotplug matching in interfaces. For example, with a single
5180 configuration snippet, you can configure that all Ethernet
5181 interfaces showing up are automatically added to a bridge,
5182 or similar. It supports link-sensing and more.
5184 * A new tool "systemd-socket-proxyd" has been added which can
5185 act as a bidirectional proxy for TCP sockets. This is
5186 useful for adding socket activation support to services that
5187 do not actually support socket activation, including virtual
5188 machines and the like.
5190 * Add a new tool to save/restore rfkill state on
5193 * Save/restore state of keyboard backlights in addition to
5194 display backlights on shutdown/boot.
5196 * udev learned a new SECLABEL{} construct to label device
5197 nodes with a specific security label when they appear. For
5198 now, only SECLABEL{selinux} is supported, but the syntax is
5199 prepared for additional security frameworks.
5201 * udev gained a new scheme to configure link-level attributes
5202 from files in /etc/systemd/network/*.link. These files can
5203 match against MAC address, device path, driver name and type,
5204 and will apply attributes like the naming policy, link speed,
5205 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5206 address assignment policy (randomized, ...).
5208 * The configuration of network interface naming rules for
5209 "permanent interface names" has changed: a new NamePolicy=
5210 setting in the [Link] section of .link files determines the
5211 priority of possible naming schemes (onboard, slot, MAC,
5212 path). The default value of this setting is determined by
5213 /usr/lib/net/links/99-default.link. Old
5214 80-net-name-slot.rules udev configuration file has been
5215 removed, so local configuration overriding this file should
5216 be adapted to override 99-default.link instead.
5218 * When the User= switch is used in a unit file, also
5219 initialize $SHELL= based on the user database entry.
5221 * systemd no longer depends on libdbus. All communication is
5222 now done with sd-bus, systemd's low-level bus library
5225 * kdbus support has been added to PID 1 itself. When kdbus is
5226 enabled, this causes PID 1 to set up the system bus and
5227 enable support for a new ".busname" unit type that
5228 encapsulates bus name activation on kdbus. It works a little
5229 bit like ".socket" units, except for bus names. A new
5230 generator has been added that converts classic dbus1 service
5231 activation files automatically into native systemd .busname
5234 * sd-bus: add a light-weight vtable implementation that allows
5235 defining objects on the bus with a simple static const
5236 vtable array of its methods, signals and properties.
5238 * systemd will not generate or install static dbus
5239 introspection data anymore to /usr/share/dbus-1/interfaces,
5240 as the precise format of these files is unclear, and
5241 nothing makes use of it.
5243 * A proxy daemon is now provided to proxy clients connecting
5244 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5245 compatibility with classic D-Bus.
5247 * A bus driver implementation has been added that supports the
5248 classic D-Bus bus driver calls on kdbus, also for
5249 compatibility purposes.
5251 * A new API "sd-event.h" has been added that implements a
5252 minimal event loop API built around epoll. It provides a
5253 couple of features that direct epoll usage is lacking:
5254 prioritization of events, scales to large numbers of timer
5255 events, per-event timer slack (accuracy), system-wide
5256 coalescing of timer events, exit handlers, watchdog
5257 supervision support using systemd's sd_notify() API, child
5260 * A new API "sd-rntl.h" has been added that provides an API
5261 around the route netlink interface of the kernel, similar in
5262 style to "sd-bus.h".
5264 * A new API "sd-dhcp-client.h" has been added that provides a
5265 small DHCPv4 client-side implementation. This is used by
5268 * There is a new kernel command line option
5269 "systemd.restore_state=0|1". When set to "0", none of the
5270 systemd tools will restore saved runtime state to hardware
5271 devices. More specifically, the rfkill and backlight states
5274 * The FsckPassNo= compatibility option in mount/service units
5275 has been removed. The fstab generator will now add the
5276 necessary dependencies automatically, and does not require
5277 PID1's support for that anymore.
5279 * journalctl gained a new switch, --list-boots, that lists
5280 recent boots with their times and boot IDs.
5282 * The various tools like systemctl, loginctl, timedatectl,
5283 busctl, systemd-run, ... have gained a new switch "-M" to
5284 connect to a specific, local OS container (as direct
5285 connection, without requiring SSH). This works on any
5286 container that is registered with machined, such as those
5287 created by libvirt-lxc or nspawn.
5289 * systemd-run and systemd-analyze also gained support for "-H"
5290 to connect to remote hosts via SSH. This is particularly
5291 useful for systemd-run because it enables queuing of jobs
5292 onto remote systems.
5294 * machinectl gained a new command "login" to open a getty
5295 login in any local container. This works with any container
5296 that is registered with machined (such as those created by
5297 libvirt-lxc or nspawn), and which runs systemd inside.
5299 * machinectl gained a new "reboot" command that may be used to
5300 trigger a reboot on a specific container that is registered
5301 with machined. This works on any container that runs an init
5302 system of some kind.
5304 * systemctl gained a new "list-timers" command to print a nice
5305 listing of installed timer units with the times they elapse
5308 * Alternative reboot() parameters may now be specified on the
5309 "systemctl reboot" command line and are passed to the
5310 reboot() system call.
5312 * systemctl gained a new --job-mode= switch to configure the
5313 mode to queue a job with. This is a more generic version of
5314 --fail, --irreversible, and --ignore-dependencies, which are
5315 still available but not advertised anymore.
5317 * /etc/systemd/system.conf gained new settings to configure
5318 various default timeouts of units, as well as the default
5319 start limit interval and burst. These may still be overridden
5322 * PID1 will now export on the bus profile data of the security
5323 policy upload process (such as the SELinux policy upload to
5326 * journald: when forwarding logs to the console, include
5327 timestamps (following the setting in
5328 /sys/module/printk/parameters/time).
5330 * OnCalendar= in timer units now understands the special
5331 strings "yearly" and "annually". (Both are equivalent)
5333 * The accuracy of timer units is now configurable with the new
5334 AccuracySec= setting. It defaults to 1min.
5336 * A new dependency type JoinsNamespaceOf= has been added that
5337 allows running two services within the same /tmp and network
5338 namespace, if PrivateNetwork= or PrivateTmp= are used.
5340 * A new command "cat" has been added to systemctl. It outputs
5341 the original unit file of a unit, and concatenates the
5342 contents of additional "drop-in" unit file snippets, so that
5343 the full configuration is shown.
5345 * systemctl now supports globbing on the various "list-xyz"
5346 commands, like "list-units" or "list-sockets", as well as on
5347 those commands which take multiple unit names.
5349 * journalctl's --unit= switch gained support for globbing.
5351 * All systemd daemons now make use of the watchdog logic so
5352 that systemd automatically notices when they hang.
5354 * If the $container_ttys environment variable is set,
5355 getty-generator will automatically spawn a getty for each
5356 listed tty. This is useful for container managers to request
5357 login gettys to be spawned on as many ttys as needed.
5359 * %h, %s, %U specifier support is not available anymore when
5360 used in unit files for PID 1. This is because NSS calls are
5361 not safe from PID 1. They stay available for --user
5362 instances of systemd, and as special case for the root user.
5364 * loginctl gained a new "--no-legend" switch to turn off output
5367 * The "sd-login.h" API gained three new calls:
5368 sd_session_is_remote(), sd_session_get_remote_user(),
5369 sd_session_get_remote_host() to query information about
5372 * The udev hardware database now also carries vendor/product
5373 information of SDIO devices.
5375 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5376 determine whether watchdog notifications are requested by
5379 * Socket-activated per-connection services now include a
5380 short description of the connection parameters in the
5383 * tmpfiles gained a new "--boot" option. When this is not used,
5384 only lines where the command character is not suffixed with
5385 "!" are executed. When this option is specified, those
5386 options are executed too. This partitions tmpfiles
5387 directives into those that can be safely executed at any
5388 time, and those which should be run only at boot (for
5389 example, a line that creates /run/nologin).
5391 * A new API "sd-resolve.h" has been added which provides a simple
5392 asynchronous wrapper around glibc NSS host name resolution
5393 calls, such as getaddrinfo(). In contrast to glibc's
5394 getaddrinfo_a(), it does not use signals. In contrast to most
5395 other asynchronous name resolution libraries, this one does
5396 not reimplement DNS, but reuses NSS, so that alternate
5397 host name resolution systems continue to work, such as mDNS,
5398 LDAP, etc. This API is based on libasyncns, but it has been
5399 cleaned up for inclusion in systemd.
5401 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5402 "sd-daemon.h" are no longer found in individual libraries
5403 libsystemd-journal.so, libsystemd-login.so,
5404 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5405 merged them into a single library, libsystemd.so, which
5406 provides all symbols. The reason for this is cyclic
5407 dependencies, as these libraries tend to use each other's
5408 symbols. So far, we have managed to workaround that by linking
5409 a copy of a good part of our code into each of these
5410 libraries again and again, which, however, makes certain
5411 things hard to do, like sharing static variables. Also, it
5412 substantially increases footprint. With this change, there
5413 is only one library for the basic APIs systemd
5414 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5415 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5416 library as well, however are subject to the --enable-kdbus
5417 switch (see below). Note that "sd-dhcp-client.h" is not part
5418 of this library (this is because it only consumes, never
5419 provides, services of/to other APIs). To make the transition
5420 easy from the separate libraries to the unified one, we
5421 provide the --enable-compat-libs compile-time switch which
5422 will generate stub libraries that are compatible with the
5423 old ones but redirect all calls to the new one.
5425 * All of the kdbus logic and the new APIs "sd-bus.h",
5426 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5427 and "sd-utf8.h" are compile-time optional via the
5428 "--enable-kdbus" switch, and they are not compiled in by
5429 default. To make use of kdbus, you have to explicitly enable
5430 the switch. Note however, that neither the kernel nor the
5431 userspace API for all of this is considered stable yet. We
5432 want to maintain the freedom to still change the APIs for
5433 now. By specifying this build-time switch, you acknowledge
5434 that you are aware of the instability of the current
5437 * Also, note that while kdbus is pretty much complete,
5438 it lacks one thing: proper policy support. This means you
5439 can build a fully working system with all features; however,
5440 it will be highly insecure. Policy support will be added in
5441 one of the next releases, at the same time that we will
5442 declare the APIs stable.
5444 * When the kernel command line argument "kdbus" is specified,
5445 systemd will automatically load the kdbus.ko kernel module. At
5446 this stage of development, it is only useful for testing kdbus
5447 and should not be used in production. Note: if "--enable-kdbus"
5448 is specified, and the kdbus.ko kernel module is available, and
5449 "kdbus" is added to the kernel command line, the entire system
5450 runs with kdbus instead of dbus-daemon, with the above mentioned
5451 problem of missing the system policy enforcement. Also a future
5452 version of kdbus.ko or a newer systemd will not be compatible with
5453 each other, and will unlikely be able to boot the machine if only
5454 one of them is updated.
5456 * systemctl gained a new "import-environment" command which
5457 uploads the caller's environment (or parts thereof) into the
5458 service manager so that it is inherited by services started
5459 by the manager. This is useful to upload variables like
5460 $DISPLAY into the user service manager.
5462 * A new PrivateDevices= switch has been added to service units
5463 which allows running a service with a namespaced /dev
5464 directory that does not contain any device nodes for
5465 physical devices. More specifically, it only includes devices
5466 such as /dev/null, /dev/urandom, and /dev/zero which are API
5469 * logind has been extended to support behaviour like VT
5470 switching on seats that do not support a VT. This makes
5471 multi-session available on seats that are not the first seat
5472 (seat0), and on systems where kernel support for VTs has
5473 been disabled at compile-time.
5475 * If a process holds a delay lock for system sleep or shutdown
5476 and fails to release it in time, we will now log its
5477 identity. This makes it easier to identify processes that
5478 cause slow suspends or power-offs.
5480 * When parsing /etc/crypttab, support for a new key-slot=
5481 option as supported by Debian is added. It allows indicating
5482 which LUKS slot to use on disk, speeding up key loading.
5484 * The sd_journal_sendv() API call has been checked and
5485 officially declared to be async-signal-safe so that it may
5486 be invoked from signal handlers for logging purposes.
5488 * Boot-time status output is now enabled automatically after a
5489 short timeout if boot does not progress, in order to give
5490 the user an indication what she or he is waiting for.
5492 * The boot-time output has been improved to show how much time
5493 remains until jobs expire.
5495 * The KillMode= switch in service units gained a new possible
5496 value "mixed". If set, and the unit is shut down, then the
5497 initial SIGTERM signal is sent only to the main daemon
5498 process, while the following SIGKILL signal is sent to
5499 all remaining processes of the service.
5501 * When a scope unit is registered, a new property "Controller"
5502 may be set. If set to a valid bus name, systemd will send a
5503 RequestStop() signal to this name when it would like to shut
5504 down the scope. This may be used to hook manager logic into
5505 the shutdown logic of scope units. Also, scope units may now
5506 be put in a special "abandoned" state, in which case the
5507 manager process which created them takes no further
5508 responsibilities for it.
5510 * When reading unit files, systemd will now verify
5511 the access mode of these files, and warn about certain
5512 suspicious combinations. This has been added to make it
5513 easier to track down packaging bugs where unit files are
5514 marked executable or world-writable.
5516 * systemd-nspawn gained a new "--setenv=" switch to set
5517 container-wide environment variables. The similar option in
5518 systemd-activate was renamed from "--environment=" to
5519 "--setenv=" for consistency.
5521 * systemd-nspawn has been updated to create a new kdbus domain
5522 for each container that is invoked, thus allowing each
5523 container to have its own set of system and user buses,
5524 independent of the host.
5526 * systemd-nspawn gained a new --drop-capability= switch to run
5527 the container with less capabilities than the default. Both
5528 --drop-capability= and --capability= now take the special
5529 string "all" for dropping or keeping all capabilities.
5531 * systemd-nspawn gained new switches for executing containers
5532 with specific SELinux labels set.
5534 * systemd-nspawn gained a new --quiet switch to not generate
5535 any additional output but the container's own console
5538 * systemd-nspawn gained a new --share-system switch to run a
5539 container without PID namespacing enabled.
5541 * systemd-nspawn gained a new --register= switch to control
5542 whether the container is registered with systemd-machined or
5543 not. This is useful for containers that do not run full
5544 OS images, but only specific apps.
5546 * systemd-nspawn gained a new --keep-unit which may be used
5547 when invoked as the only program from a service unit, and
5548 results in registration of the unit service itself in
5549 systemd-machined, instead of a newly opened scope unit.
5551 * systemd-nspawn gained a new --network-interface= switch for
5552 moving arbitrary interfaces to the container. The new
5553 --network-veth switch creates a virtual Ethernet connection
5554 between host and container. The new --network-bridge=
5555 switch then allows assigning the host side of this virtual
5556 Ethernet connection to a bridge device.
5558 * systemd-nspawn gained a new --personality= switch for
5559 setting the kernel personality for the container. This is
5560 useful when running a 32-bit container on a 64-bit host. A
5561 similar option Personality= is now also available for service
5564 * logind will now also track a "Desktop" identifier for each
5565 session which encodes the desktop environment of it. This is
5566 useful for desktop environments that want to identify
5567 multiple running sessions of itself easily.
5569 * A new SELinuxContext= setting for service units has been
5570 added that allows setting a specific SELinux execution
5571 context for a service.
5573 * Most systemd client tools will now honour $SYSTEMD_LESS for
5574 settings of the "less" pager. By default, these tools will
5575 override $LESS to allow certain operations to work, such as
5576 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5577 influence this logic.
5579 * systemd's "seccomp" hook-up has been changed to make use of
5580 the libseccomp library instead of using its own
5581 implementation. This has benefits for portability among
5584 * For usage together with SystemCallFilter=, a new
5585 SystemCallErrorNumber= setting has been introduced that
5586 allows configuration of a system error number to be returned
5587 on filtered system calls, instead of immediately killing the
5588 process. Also, SystemCallArchitectures= has been added to
5589 limit access to system calls of a particular architecture
5590 (in order to turn off support for unused secondary
5591 architectures). There is also a global
5592 SystemCallArchitectures= setting in system.conf now to turn
5593 off support for non-native system calls system-wide.
5595 * systemd requires a kernel with a working name_to_handle_at(),
5596 please see the kernel config requirements in the README file.
5598 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5599 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5600 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5601 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5602 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5603 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5604 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5605 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5606 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5607 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5608 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5609 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5610 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5611 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5612 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5613 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5614 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5615 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5616 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5617 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5618 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5619 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5620 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5621 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5623 — Berlin, 2014-02-20
5627 * logind has gained support for facilitating privileged input
5628 and drm device access for unprivileged clients. This work is
5629 useful to allow Wayland display servers (and similar
5630 programs, such as kmscon) to run under the user's ID and
5631 access input and drm devices which are normally
5632 protected. When this is used (and the kernel is new enough)
5633 logind will "mute" IO on the file descriptors passed to
5634 Wayland as long as it is in the background and "unmute" it
5635 if it returns into the foreground. This allows secure
5636 session switching without allowing background sessions to
5637 eavesdrop on input and display data. This also introduces
5638 session switching support if VT support is turned off in the
5639 kernel, and on seats that are not seat0.
5641 * A new kernel command line option luks.options= is understood
5642 now which allows specifying LUKS options for usage for LUKS
5643 encrypted partitions specified with luks.uuid=.
5645 * tmpfiles.d(5) snippets may now use specifier expansion in
5646 path names. More specifically %m, %b, %H, %v, are now
5647 replaced by the local machine id, boot id, hostname, and
5648 kernel version number.
5650 * A new tmpfiles.d(5) command "m" has been introduced which
5651 may be used to change the owner/group/access mode of a file
5652 or directory if it exists, but do nothing if it does not.
5654 * This release removes high-level support for the
5655 MemorySoftLimit= cgroup setting. The underlying kernel
5656 cgroup attribute memory.soft_limit= is currently badly
5657 designed and likely to be removed from the kernel API in its
5658 current form, hence we should not expose it for now.
5660 * The memory.use_hierarchy cgroup attribute is now enabled for
5661 all cgroups systemd creates in the memory cgroup
5662 hierarchy. This option is likely to be come the built-in
5663 default in the kernel anyway, and the non-hierarchical mode
5664 never made much sense in the intrinsically hierarchical
5667 * A new field _SYSTEMD_SLICE= is logged along with all journal
5668 messages containing the slice a message was generated
5669 from. This is useful to allow easy per-customer filtering of
5670 logs among other things.
5672 * systemd-journald will no longer adjust the group of journal
5673 files it creates to the "systemd-journal" group. Instead we
5674 rely on the journal directory to be owned by the
5675 "systemd-journal" group, and its setgid bit set, so that the
5676 kernel file system layer will automatically enforce that
5677 journal files inherit this group assignment. The reason for
5678 this change is that we cannot allow NSS look-ups from
5679 journald which would be necessary to resolve
5680 "systemd-journal" to a numeric GID, because this might
5681 create deadlocks if NSS involves synchronous queries to
5682 other daemons (such as nscd, or sssd) which in turn are
5683 logging clients of journald and might block on it, which
5684 would then dead lock. A tmpfiles.d(5) snippet included in
5685 systemd will make sure the setgid bit and group are
5686 properly set on the journal directory if it exists on every
5687 boot. However, we recommend adjusting it manually after
5688 upgrades too (or from RPM scriptlets), so that the change is
5689 not delayed until next reboot.
5691 * Backlight and random seed files in /var/lib/ have moved into
5692 the /var/lib/systemd/ directory, in order to centralize all
5693 systemd generated files in one directory.
5695 * Boot time performance measurements (as displayed by
5696 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5697 performance information if that's available to determine how
5698 much time BIOS and boot loader initialization required. With
5699 a sufficiently new BIOS you hence no longer need to boot
5700 with Gummiboot to get access to such information.
5702 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5703 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5704 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5705 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5706 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5707 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5708 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5710 — Berlin, 2013-10-02
5714 * The Restart= option for services now understands a new
5715 on-watchdog setting, which will restart the service
5716 automatically if the service stops sending out watchdog keep
5717 alive messages (as configured with WatchdogSec=).
5719 * The getty generator (which is responsible for bringing up a
5720 getty on configured serial consoles) will no longer only
5721 start a getty on the primary kernel console but on all
5722 others, too. This makes the order in which console= is
5723 specified on the kernel command line less important.
5725 * libsystemd-logind gained a new sd_session_get_vt() call to
5726 retrieve the VT number of a session.
5728 * If the option "tries=0" is set for an entry of /etc/crypttab
5729 its passphrase is queried indefinitely instead of any
5730 maximum number of tries.
5732 * If a service with a configure PID file terminates its PID
5733 file will now be removed automatically if it still exists
5734 afterwards. This should put an end to stale PID files.
5736 * systemd-run will now also take relative binary path names
5737 for execution and no longer insists on absolute paths.
5739 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5740 paths that are optionally prefixed with "-" to indicate that
5741 it should not be considered a failure if they do not exist.
5743 * journalctl -o (and similar commands) now understands a new
5744 output mode "short-precise", it is similar to "short" but
5745 shows timestamps with usec accuracy.
5747 * The option "discard" (as known from Debian) is now
5748 synonymous to "allow-discards" in /etc/crypttab. In fact,
5749 "discard" is preferred now (since it is easier to remember
5752 * Some licensing clean-ups were made, so that more code is now
5753 LGPL-2.1 licensed than before.
5755 * A minimal tool to save/restore the display backlight
5756 brightness across reboots has been added. It will store the
5757 backlight setting as late as possible at shutdown, and
5758 restore it as early as possible during reboot.
5760 * A logic to automatically discover and enable home and swap
5761 partitions on GPT disks has been added. With this in place
5762 /etc/fstab becomes optional for many setups as systemd can
5763 discover certain partitions located on the root disk
5764 automatically. Home partitions are recognized under their
5765 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5766 partitions are recognized under their GPT type ID
5767 0657fd6da4ab43c484e50933c84b4f4f.
5769 * systemd will no longer pass any environment from the kernel
5770 or initrd to system services. If you want to set an
5771 environment for all services, do so via the kernel command
5772 line systemd.setenv= assignment.
5774 * The systemd-sysctl tool no longer natively reads the file
5775 /etc/sysctl.conf. If desired, the file should be symlinked
5776 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5777 legacy support by a symlink rather than built-in code, it
5778 also makes the otherwise hidden order of application of the
5779 different files visible. (Note that this partly reverts to a
5780 pre-198 application order of sysctl knobs!)
5782 * The "systemctl set-log-level" and "systemctl dump" commands
5783 have been moved to systemd-analyze.
5785 * systemd-run learned the new --remain-after-exit switch,
5786 which causes the scope unit not to be cleaned up
5787 automatically after the process terminated.
5789 * tmpfiles learned a new --exclude-prefix= switch to exclude
5790 certain paths from operation.
5792 * journald will now automatically flush all messages to disk
5793 as soon as a message at the log level CRIT, ALERT or EMERG
5796 Contributions from: Andrew Cook, Brandon Philips, Christian
5797 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5798 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5799 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5800 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5801 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5802 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5803 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5804 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5805 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5806 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5807 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5808 William Giokas, Zbigniew Jędrzejewski-Szmek
5810 — Berlin, 2013-09-13
5814 * The documentation has been updated to cover the various new
5815 concepts introduced with 205.
5817 * Unit files now understand the new %v specifier which
5818 resolves to the kernel version string as returned by "uname
5821 * systemctl now supports filtering the unit list output by
5822 load state, active state and sub state, using the new
5825 * "systemctl status" will now show the results of the
5826 condition checks (like ConditionPathExists= and similar) of
5827 the last start attempts of the unit. They are also logged to
5830 * "journalctl -b" may now be used to look for boot output of a
5831 specific boot. Try "journalctl -b -1" for the previous boot,
5832 but the syntax is substantially more powerful.
5834 * "journalctl --show-cursor" has been added which prints the
5835 cursor string the last shown log line. This may then be used
5836 with the new "journalctl --after-cursor=" switch to continue
5837 browsing logs from that point on.
5839 * "journalctl --force" may now be used to force regeneration
5842 * Creation of "dead" device nodes has been moved from udev
5843 into kmod and tmpfiles. Previously, udev would read the kmod
5844 databases to pre-generate dead device nodes based on meta
5845 information contained in kernel modules, so that these would
5846 be auto-loaded on access rather then at boot. As this
5847 does not really have much to do with the exposing actual
5848 kernel devices to userspace this has always been slightly
5849 alien in the udev codebase. Following the new scheme kmod
5850 will now generate a runtime snippet for tmpfiles from the
5851 module meta information and it now is tmpfiles' job to the
5852 create the nodes. This also allows overriding access and
5853 other parameters for the nodes using the usual tmpfiles
5854 facilities. As side effect this allows us to remove the
5855 CAP_SYS_MKNOD capability bit from udevd entirely.
5857 * logind's device ACLs may now be applied to these "dead"
5858 devices nodes too, thus finally allowing managed access to
5859 devices such as /dev/snd/sequencer without loading the
5860 backing module right-away.
5862 * A new RPM macro has been added that may be used to apply
5863 tmpfiles configuration during package installation.
5865 * systemd-detect-virt and ConditionVirtualization= now can
5866 detect User-Mode-Linux machines (UML).
5868 * journald will now implicitly log the effective capabilities
5869 set of processes in the message metadata.
5871 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5873 * The initrd interface has been simplified (more specifically,
5874 support for passing performance data via environment
5875 variables and fsck results via files in /run has been
5876 removed). These features were non-essential, and are
5877 nowadays available in a much nicer way by having systemd in
5878 the initrd serialize its state and have the hosts systemd
5879 deserialize it again.
5881 * The udev "keymap" data files and tools to apply keyboard
5882 specific mappings of scan to key codes, and force-release
5883 scan code lists have been entirely replaced by a udev
5884 "keyboard" builtin and a hwdb data file.
5886 * systemd will now honour the kernel's "quiet" command line
5887 argument also during late shutdown, resulting in a
5888 completely silent shutdown when used.
5890 * There's now an option to control the SO_REUSEPORT socket
5891 option in .socket units.
5893 * Instance units will now automatically get a per-template
5894 subslice of system.slice unless something else is explicitly
5895 configured. For example, instances of sshd@.service will now
5896 implicitly be placed in system-sshd.slice rather than
5897 system.slice as before.
5899 * Test coverage support may now be enabled at build time.
5901 Contributions from: Dave Reisner, Frederic Crozat, Harald
5902 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5903 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5904 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5905 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5906 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5907 Giokas, Zbigniew Jędrzejewski-Szmek
5909 — Berlin, 2013-07-23
5913 * Two new unit types have been introduced:
5915 Scope units are very similar to service units, however, are
5916 created out of pre-existing processes — instead of PID 1
5917 forking off the processes. By using scope units it is
5918 possible for system services and applications to group their
5919 own child processes (worker processes) in a powerful way
5920 which then maybe used to organize them, or kill them
5921 together, or apply resource limits on them.
5923 Slice units may be used to partition system resources in an
5924 hierarchical fashion and then assign other units to them. By
5925 default there are now three slices: system.slice (for all
5926 system services), user.slice (for all user sessions),
5927 machine.slice (for VMs and containers).
5929 Slices and scopes have been introduced primarily in
5930 context of the work to move cgroup handling to a
5931 single-writer scheme, where only PID 1
5932 creates/removes/manages cgroups.
5934 * There's a new concept of "transient" units. In contrast to
5935 normal units these units are created via an API at runtime,
5936 not from configuration from disk. More specifically this
5937 means it is now possible to run arbitrary programs as
5938 independent services, with all execution parameters passed
5939 in via bus APIs rather than read from disk. Transient units
5940 make systemd substantially more dynamic then it ever was,
5941 and useful as a general batch manager.
5943 * logind has been updated to make use of scope and slice units
5944 for managing user sessions. As a user logs in he will get
5945 his own private slice unit, to which all sessions are added
5946 as scope units. We also added support for automatically
5947 adding an instance of user@.service for the user into the
5948 slice. Effectively logind will no longer create cgroup
5949 hierarchies on its own now, it will defer entirely to PID 1
5950 for this by means of scope, service and slice units. Since
5951 user sessions this way become entities managed by PID 1
5952 the output of "systemctl" is now a lot more comprehensive.
5954 * A new mini-daemon "systemd-machined" has been added which
5955 may be used by virtualization managers to register local
5956 VMs/containers. nspawn has been updated accordingly, and
5957 libvirt will be updated shortly. machined will collect a bit
5958 of meta information about the VMs/containers, and assign
5959 them their own scope unit (see above). The collected
5960 meta-data is then made available via the "machinectl" tool,
5961 and exposed in "ps" and similar tools. machined/machinectl
5962 is compile-time optional.
5964 * As discussed earlier, the low-level cgroup configuration
5965 options ControlGroup=, ControlGroupModify=,
5966 ControlGroupPersistent=, ControlGroupAttribute= have been
5967 removed. Please use high-level attribute settings instead as
5968 well as slice units.
5970 * A new bus call SetUnitProperties() has been added to alter
5971 various runtime parameters of a unit. This is primarily
5972 useful to alter cgroup parameters dynamically in a nice way,
5973 but will be extended later on to make more properties
5974 modifiable at runtime. systemctl gained a new set-properties
5975 command that wraps this call.
5977 * A new tool "systemd-run" has been added which can be used to
5978 run arbitrary command lines as transient services or scopes,
5979 while configuring a number of settings via the command
5980 line. This tool is currently very basic, however already
5981 very useful. We plan to extend this tool to even allow
5982 queuing of execution jobs with time triggers from the
5983 command line, similar in fashion to "at".
5985 * nspawn will now inform the user explicitly that kernels with
5986 audit enabled break containers, and suggest the user to turn
5989 * Support for detecting the IMA and AppArmor security
5990 frameworks with ConditionSecurity= has been added.
5992 * journalctl gained a new "-k" switch for showing only kernel
5993 messages, mimicking dmesg output; in addition to "--user"
5994 and "--system" switches for showing only user's own logs
5997 * systemd-delta can now show information about drop-in
5998 snippets extending unit files.
6000 * libsystemd-bus has been substantially updated but is still
6001 not available as public API.
6003 * systemd will now look for the "debug" argument on the kernel
6004 command line and enable debug logging, similar to what
6005 "systemd.log_level=debug" already did before.
6007 * "systemctl set-default", "systemctl get-default" has been
6008 added to configure the default.target symlink, which
6009 controls what to boot into by default.
6011 * "systemctl set-log-level" has been added as a convenient
6012 way to raise and lower systemd logging threshold.
6014 * "systemd-analyze plot" will now show the time the various
6015 generators needed for execution, as well as information
6016 about the unit file loading.
6018 * libsystemd-journal gained a new sd_journal_open_files() call
6019 for opening specific journal files. journactl also gained a
6020 new switch to expose this new functionality. Previously we
6021 only supported opening all files from a directory, or all
6022 files from the system, as opening individual files only is
6023 racy due to journal file rotation.
6025 * systemd gained the new DefaultEnvironment= setting in
6026 /etc/systemd/system.conf to set environment variables for
6029 * If a privileged process logs a journal message with the
6030 OBJECT_PID= field set, then journald will automatically
6031 augment this with additional OBJECT_UID=, OBJECT_GID=,
6032 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6033 system services want to log events about specific client
6034 processes. journactl/systemctl has been updated to make use
6035 of this information if all log messages regarding a specific
6038 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6039 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6040 Reisner, David Coppa, David King, David Strauss, Eelco
6041 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6042 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6043 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6044 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6045 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6046 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6047 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6048 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6049 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6050 Łukasz Stelmach, 장동준
6054 * The Python bindings gained some minimal support for the APIs
6055 exposed by libsystemd-logind.
6057 * ConditionSecurity= gained support for detecting SMACK. Since
6058 this condition already supports SELinux and AppArmor we only
6059 miss IMA for this. Patches welcome!
6061 Contributions from: Karol Lewandowski, Lennart Poettering,
6062 Zbigniew Jędrzejewski-Szmek
6066 * systemd-nspawn will now create /etc/resolv.conf if
6067 necessary, before bind-mounting the host's file onto it.
6069 * systemd-nspawn will now store meta information about a
6070 container on the container's cgroup as extended attribute
6071 fields, including the root directory.
6073 * The cgroup hierarchy has been reworked in many ways. All
6074 objects any of the components systemd creates in the cgroup
6075 tree are now suffixed. More specifically, user sessions are
6076 now placed in cgroups suffixed with ".session", users in
6077 cgroups suffixed with ".user", and nspawn containers in
6078 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6079 names are now escaped in a simple scheme to avoid collision
6080 of userspace object names with kernel filenames. This work
6081 is preparation for making these objects relocatable in the
6082 cgroup tree, in order to allow easy resource partitioning of
6083 these objects without causing naming conflicts.
6085 * systemctl list-dependencies gained the new switches
6086 --plain, --reverse, --after and --before.
6088 * systemd-inhibit now shows the process name of processes that
6089 have taken an inhibitor lock.
6091 * nss-myhostname will now also resolve "localhost"
6092 implicitly. This makes /etc/hosts an optional file and
6093 nicely handles that on IPv6 ::1 maps to both "localhost" and
6096 * libsystemd-logind.so gained a new call
6097 sd_get_machine_names() to enumerate running containers and
6098 VMs (currently only supported by very new libvirt and
6099 nspawn). sd_login_monitor can now be used to watch
6100 VMs/containers coming and going.
6102 * .include is not allowed recursively anymore, and only in
6103 unit files. Usually it is better to use drop-in snippets in
6104 .d/*.conf anyway, as introduced with systemd 198.
6106 * systemd-analyze gained a new "critical-chain" command that
6107 determines the slowest chain of units run during system
6108 boot-up. It is very useful for tracking down where
6109 optimizing boot time is the most beneficial.
6111 * systemd will no longer allow manipulating service paths in
6112 the name=systemd:/system cgroup tree using ControlGroup= in
6113 units. (But is still fine with it in all other dirs.)
6115 * There's a new systemd-nspawn@.service service file that may
6116 be used to easily run nspawn containers as system
6117 services. With the container's root directory in
6118 /var/lib/container/foobar it is now sufficient to run
6119 "systemctl start systemd-nspawn@foobar.service" to boot it.
6121 * systemd-cgls gained a new parameter "--machine" to list only
6122 the processes within a certain container.
6124 * ConditionSecurity= now can check for "apparmor". We still
6125 are lacking checks for SMACK and IMA for this condition
6126 check though. Patches welcome!
6128 * A new configuration file /etc/systemd/sleep.conf has been
6129 added that may be used to configure which kernel operation
6130 systemd is supposed to execute when "suspend", "hibernate"
6131 or "hybrid-sleep" is requested. This makes the new kernel
6132 "freeze" state accessible to the user.
6134 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6135 the passed argument if applicable.
6137 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6138 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6139 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6140 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6141 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6142 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6143 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6148 * The output of 'systemctl list-jobs' got some polishing. The
6149 '--type=' argument may now be passed more than once. A new
6150 command 'systemctl list-sockets' has been added which shows
6151 a list of kernel sockets systemd is listening on with the
6152 socket units they belong to, plus the units these socket
6155 * The experimental libsystemd-bus library got substantial
6156 updates to work in conjunction with the (also experimental)
6157 kdbus kernel project. It works well enough to exchange
6158 messages with some sophistication. Note that kdbus is not
6159 ready yet, and the library is mostly an elaborate test case
6160 for now, and not installable.
6162 * systemd gained a new unit 'systemd-static-nodes.service'
6163 that generates static device nodes earlier during boot, and
6164 can run in conjunction with udev.
6166 * libsystemd-login gained a new call sd_pid_get_user_unit()
6167 to retrieve the user systemd unit a process is running
6168 in. This is useful for systems where systemd is used as
6171 * systemd-nspawn now places all containers in the new /machine
6172 top-level cgroup directory in the name=systemd
6173 hierarchy. libvirt will soon do the same, so that we get a
6174 uniform separation of /system, /user and /machine for system
6175 services, user processes and containers/virtual
6176 machines. This new cgroup hierarchy is also useful to stick
6177 stable names to specific container instances, which can be
6178 recognized later this way (this name may be controlled
6179 via systemd-nspawn's new -M switch). libsystemd-login also
6180 gained a new call sd_pid_get_machine_name() to retrieve the
6181 name of the container/VM a specific process belongs to.
6183 * bootchart can now store its data in the journal.
6185 * libsystemd-journal gained a new call
6186 sd_journal_add_conjunction() for AND expressions to the
6187 matching logic. This can be used to express more complex
6188 logical expressions.
6190 * journactl can now take multiple --unit= and --user-unit=
6193 * The cryptsetup logic now understands the "luks.key=" kernel
6194 command line switch for specifying a file to read the
6195 decryption key from. Also, if a configured key file is not
6196 found the tool will now automatically fall back to prompting
6199 * Python systemd.journal module was updated to wrap recently
6200 added functions from libsystemd-journal. The interface was
6201 changed to bring the low level interface in s.j._Reader
6202 closer to the C API, and the high level interface in
6203 s.j.Reader was updated to wrap and convert all data about
6206 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6207 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6208 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6209 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6210 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6211 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6215 * journalctl --update-catalog now understands a new --root=
6216 option to operate on catalogs found in a different root
6219 * During shutdown after systemd has terminated all running
6220 services a final killing loop kills all remaining left-over
6221 processes. We will now print the name of these processes
6222 when we send SIGKILL to them, since this usually indicates a
6225 * If /etc/crypttab refers to password files stored on
6226 configured mount points automatic dependencies will now be
6227 generated to ensure the specific mount is established first
6228 before the key file is attempted to be read.
6230 * 'systemctl status' will now show information about the
6231 network sockets a socket unit is listening on.
6233 * 'systemctl status' will also shown information about any
6234 drop-in configuration file for units. (Drop-In configuration
6235 files in this context are files such as
6236 /etc/systemd/systemd/foobar.service.d/*.conf)
6238 * systemd-cgtop now optionally shows summed up CPU times of
6239 cgroups. Press '%' while running cgtop to switch between
6240 percentage and absolute mode. This is useful to determine
6241 which cgroups use up the most CPU time over the entire
6242 runtime of the system. systemd-cgtop has also been updated
6243 to be 'pipeable' for processing with further shell tools.
6245 * 'hostnamectl set-hostname' will now allow setting of FQDN
6248 * The formatting and parsing of time span values has been
6249 changed. The parser now understands fractional expressions
6250 such as "5.5h". The formatter will now output fractional
6251 expressions for all time spans under 1min, i.e. "5.123456s"
6252 rather than "5s 123ms 456us". For time spans under 1s
6253 millisecond values are shown, for those under 1ms
6254 microsecond values are shown. This should greatly improve
6255 all time-related output of systemd.
6257 * libsystemd-login and libsystemd-journal gained new
6258 functions for querying the poll() events mask and poll()
6259 timeout value for integration into arbitrary event
6262 * localectl gained the ability to list available X11 keymaps
6263 (models, layouts, variants, options).
6265 * 'systemd-analyze dot' gained the ability to filter for
6266 specific units via shell-style globs, to create smaller,
6267 more useful graphs. I.e. it is now possible to create simple
6268 graphs of all the dependencies between only target units, or
6269 of all units that Avahi has dependencies with.
6271 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6272 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6273 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6274 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6275 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6276 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6277 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6281 * The boot-time readahead implementation for rotating media
6282 will now read the read-ahead data in multiple passes which
6283 consist of all read requests made in equidistant time
6284 intervals. This means instead of strictly reading read-ahead
6285 data in its physical order on disk we now try to find a
6286 middle ground between physical and access time order.
6288 * /etc/os-release files gained a new BUILD_ID= field for usage
6289 on operating systems that provide continuous builds of OS
6292 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6293 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6294 William Douglas, Zbigniew Jędrzejewski-Szmek
6298 * systemd-python gained an API exposing libsystemd-daemon.
6300 * The SMACK setup logic gained support for uploading CIPSO
6303 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6304 ReadOnlyDirectories= and InaccessibleDirectories= has
6305 changed. The private /tmp and /var/tmp directories are now
6306 shared by all processes of a service (which means
6307 ExecStartPre= may now leave data in /tmp that ExecStart= of
6308 the same service can still access). When a service is
6309 stopped its temporary directories are immediately deleted
6310 (normal clean-up with tmpfiles is still done in addition to
6313 * By default, systemd will now set a couple of sysctl
6314 variables in the kernel: the safe sysrq options are turned
6315 on, IP route verification is turned on, and source routing
6316 disabled. The recently added hardlink and softlink
6317 protection of the kernel is turned on. These settings should
6318 be reasonably safe, and good defaults for all new systems.
6320 * The predictable network naming logic may now be turned off
6321 with a new kernel command line switch: net.ifnames=0.
6323 * A new libsystemd-bus module has been added that implements a
6324 pretty complete D-Bus client library. For details see:
6326 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6328 * journald will now explicitly flush the journal files to disk
6329 at the latest 5min after each write. The file will then also
6330 be marked offline until the next write. This should increase
6331 reliability in case of a crash. The synchronization delay
6332 can be configured via SyncIntervalSec= in journald.conf.
6334 * There's a new remote-fs-setup.target unit that can be used
6335 to pull in specific services when at least one remote file
6336 system is to be mounted.
6338 * There are new targets timers.target and paths.target as
6339 canonical targets to pull user timer and path units in
6340 from. This complements sockets.target with a similar
6341 purpose for socket units.
6343 * libudev gained a new call udev_device_set_attribute_value()
6344 to set sysfs attributes of a device.
6346 * The udev daemon now sets the default number of worker
6347 processes executed in parallel based on the number of available
6348 CPUs instead of the amount of available RAM. This is supposed
6349 to provide a more reliable default and limit a too aggressive
6350 parallelism for setups with 1000s of devices connected.
6352 Contributions from: Auke Kok, Colin Walters, Cristian
6353 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6354 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6355 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6356 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6357 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6358 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6359 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6360 Zbigniew Jędrzejewski-Szmek
6364 * Configuration of unit files may now be extended via drop-in
6365 files without having to edit/override the unit files
6366 themselves. More specifically, if the administrator wants to
6367 change one value for a service file foobar.service he can
6368 now do so by dropping in a configuration snippet into
6369 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6370 will load all these snippets and apply them on top of the
6371 main unit configuration file, possibly extending or
6372 overriding its settings. Using these drop-in snippets is
6373 generally nicer than the two earlier options for changing
6374 unit files locally: copying the files from
6375 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6376 them there; or creating a new file in /etc/systemd/system/
6377 that incorporates the original one via ".include". Drop-in
6378 snippets into these .d/ directories can be placed in any
6379 directory systemd looks for units in, and the usual
6380 overriding semantics between /usr/lib, /etc and /run apply
6383 * Most unit file settings which take lists of items can now be
6384 reset by assigning the empty string to them. For example,
6385 normally, settings such as Environment=FOO=BAR append a new
6386 environment variable assignment to the environment block,
6387 each time they are used. By assigning Environment= the empty
6388 string the environment block can be reset to empty. This is
6389 particularly useful with the .d/*.conf drop-in snippets
6390 mentioned above, since this adds the ability to reset list
6391 settings from vendor unit files via these drop-ins.
6393 * systemctl gained a new "list-dependencies" command for
6394 listing the dependencies of a unit recursively.
6396 * Inhibitors are now honored and listed by "systemctl
6397 suspend", "systemctl poweroff" (and similar) too, not only
6398 GNOME. These commands will also list active sessions by
6401 * Resource limits (as exposed by the various control group
6402 controllers) can now be controlled dynamically at runtime
6403 for all units. More specifically, you can now use a command
6404 like "systemctl set-cgroup-attr foobar.service cpu.shares
6405 2000" to alter the CPU shares a specific service gets. These
6406 settings are stored persistently on disk, and thus allow the
6407 administrator to easily adjust the resource usage of
6408 services with a few simple commands. This dynamic resource
6409 management logic is also available to other programs via the
6410 bus. Almost any kernel cgroup attribute and controller is
6413 * systemd-vconsole-setup will now copy all font settings to
6414 all allocated VTs, where it previously applied them only to
6417 * libsystemd-login gained the new sd_session_get_tty() API
6420 * This release drops support for a few legacy or
6421 distribution-specific LSB facility names when parsing init
6422 scripts: $x-display-manager, $mail-transfer-agent,
6423 $mail-transport-agent, $mail-transfer-agent, $smtp,
6424 $null. Also, the mail-transfer-agent.target unit backing
6425 this has been removed. Distributions which want to retain
6426 compatibility with this should carry the burden for
6427 supporting this themselves and patch support for these back
6428 in, if they really need to. Also, the facilities $syslog and
6429 $local_fs are now ignored, since systemd does not support
6430 early-boot LSB init scripts anymore, and these facilities
6431 are implied anyway for normal services. syslog.target has
6434 * There are new bus calls on PID1's Manager object for
6435 cancelling jobs, and removing snapshot units. Previously,
6436 both calls were only available on the Job and Snapshot
6439 * systemd-journal-gatewayd gained SSL support.
6441 * The various "environment" files, such as /etc/locale.conf
6442 now support continuation lines with a backslash ("\") as
6443 last character in the line, similarly in style (but different)
6444 to how this is supported in shells.
6446 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6447 now implicitly appended to every log entry logged. systemctl
6448 has been updated to filter by this field when operating on a
6449 user systemd instance.
6451 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6452 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6453 the container. This makes it easier to boot unmodified
6454 Fedora systems in a container, which however still requires
6455 audit=0 to be passed on the kernel command line. Auditing in
6456 kernel and userspace is unfortunately still too broken in
6457 context of containers, hence we recommend compiling it out
6458 of the kernel or using audit=0. Hopefully this will be fixed
6459 one day for good in the kernel.
6461 * nspawn gained the new --bind= and --bind-ro= parameters to
6462 bind mount specific directories from the host into the
6465 * nspawn will now mount its own devpts file system instance
6466 into the container, in order not to leak pty devices from
6467 the host into the container.
6469 * systemd will now read the firmware boot time performance
6470 information from the EFI variables, if the used boot loader
6471 supports this, and takes it into account for boot performance
6472 analysis via "systemd-analyze". This is currently supported
6473 only in conjunction with Gummiboot, but could be supported
6474 by other boot loaders too. For details see:
6476 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6478 * A new generator has been added that automatically mounts the
6479 EFI System Partition (ESP) to /boot, if that directory
6480 exists, is empty, and no other file system has been
6481 configured to be mounted there.
6483 * logind will now send out PrepareForSleep(false) out
6484 unconditionally, after coming back from suspend. This may be
6485 used by applications as asynchronous notification for
6486 system resume events.
6488 * "systemctl unlock-sessions" has been added, that allows
6489 unlocking the screens of all user sessions at once, similar
6490 to how "systemctl lock-sessions" already locked all users
6491 sessions. This is backed by a new D-Bus call UnlockSessions().
6493 * "loginctl seat-status" will now show the master device of a
6494 seat. (i.e. the device of a seat that needs to be around for
6495 the seat to be considered available, usually the graphics
6498 * tmpfiles gained a new "X" line type, that allows
6499 configuration of files and directories (with wildcards) that
6500 shall be excluded from automatic cleanup ("aging").
6502 * udev default rules set the device node permissions now only
6503 at "add" events, and do not change them any longer with a
6504 later "change" event.
6506 * The log messages for lid events and power/sleep keypresses
6507 now carry a message ID.
6509 * We now have a substantially larger unit test suite, but this
6510 continues to be work in progress.
6512 * udevadm hwdb gained a new --root= parameter to change the
6513 root directory to operate relative to.
6515 * logind will now issue a background sync() request to the kernel
6516 early at shutdown, so that dirty buffers are flushed to disk early
6517 instead of at the last moment, in order to optimize shutdown
6520 * A new bootctl tool has been added that is an interface for
6521 certain boot loader operations. This is currently a preview
6522 and is likely to be extended into a small mechanism daemon
6523 like timedated, localed, hostnamed, and can be used by
6524 graphical UIs to enumerate available boot options, and
6525 request boot into firmware operations.
6527 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6528 the rest of the package. It also has been updated to work
6529 correctly in initrds.
6531 * Policykit previously has been runtime optional, and is now
6532 also compile time optional via a configure switch.
6534 * systemd-analyze has been reimplemented in C. Also "systemctl
6535 dot" has moved into systemd-analyze.
6537 * "systemctl status" with no further parameters will now print
6538 the status of all active or failed units.
6540 * Operations such as "systemctl start" can now be executed
6541 with a new mode "--irreversible" which may be used to queue
6542 operations that cannot accidentally be reversed by a later
6543 job queuing. This is by default used to make shutdown
6544 requests more robust.
6546 * The Python API of systemd now gained a new module for
6547 reading journal files.
6549 * A new tool kernel-install has been added that can install
6550 kernel images according to the Boot Loader Specification:
6552 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6554 * Boot time console output has been improved to provide
6555 animated boot time output for hanging jobs.
6557 * A new tool systemd-activate has been added which can be used
6558 to test socket activation with, directly from the command
6559 line. This should make it much easier to test and debug
6560 socket activation in daemons.
6562 * journalctl gained a new "--reverse" (or -r) option to show
6563 journal output in reverse order (i.e. newest line first).
6565 * journalctl gained a new "--pager-end" (or -e) option to jump
6566 to immediately jump to the end of the journal in the
6567 pager. This is only supported in conjunction with "less".
6569 * journalctl gained a new "--user-unit=" option, that works
6570 similarly to "--unit=" but filters for user units rather than
6573 * A number of unit files to ease adoption of systemd in
6574 initrds has been added. This moves some minimal logic from
6575 the various initrd implementations into systemd proper.
6577 * The journal files are now owned by a new group
6578 "systemd-journal", which exists specifically to allow access
6579 to the journal, and nothing else. Previously, we used the
6580 "adm" group for that, which however possibly covers more
6581 than just journal/log file access. This new group is now
6582 already used by systemd-journal-gatewayd to ensure this
6583 daemon gets access to the journal files and as little else
6584 as possible. Note that "make install" will also set FS ACLs
6585 up for /var/log/journal to give "adm" and "wheel" read
6586 access to it, in addition to "systemd-journal" which owns
6587 the journal files. We recommend that packaging scripts also
6588 add read access to "adm" + "wheel" to /var/log/journal, and
6589 all existing/future journal files. To normal users and
6590 administrators little changes, however packagers need to
6591 ensure to create the "systemd-journal" system group at
6592 package installation time.
6594 * The systemd-journal-gatewayd now runs as unprivileged user
6595 systemd-journal-gateway:systemd-journal-gateway. Packaging
6596 scripts need to create these system user/group at
6599 * timedated now exposes a new boolean property CanNTP that
6600 indicates whether a local NTP service is available or not.
6602 * systemd-detect-virt will now also detect xen PVs
6604 * The pstore file system is now mounted by default, if it is
6607 * In addition to the SELinux and IMA policies we will now also
6608 load SMACK policies at early boot.
6610 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6611 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6612 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6613 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6614 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6615 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6616 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6617 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6618 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6619 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6620 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6621 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6622 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6623 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6627 * Timer units now support calendar time events in addition to
6628 monotonic time events. That means you can now trigger a unit
6629 based on a calendar time specification such as "Thu,Fri
6630 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6631 or fifth day of any month of the year 2013, given that it is
6632 a thursday or friday. This brings timer event support
6633 considerably closer to cron's capabilities. For details on
6634 the supported calendar time specification language see
6637 * udev now supports a number of different naming policies for
6638 network interfaces for predictable names, and a combination
6639 of these policies is now the default. Please see this wiki
6640 document for details:
6642 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6644 * Auke Kok's bootchart implementation has been added to the
6645 systemd tree. It is an optional component that can graph the
6646 boot in quite some detail. It is one of the best bootchart
6647 implementations around and minimal in its code and
6650 * nss-myhostname has been integrated into the systemd source
6651 tree. nss-myhostname guarantees that the local hostname
6652 always stays resolvable via NSS. It has been a weak
6653 requirement of systemd-hostnamed since a long time, and
6654 since its code is actually trivial we decided to just
6655 include it in systemd's source tree. It can be turned off
6656 with a configure switch.
6658 * The read-ahead logic is now capable of properly detecting
6659 whether a btrfs file system is on SSD or rotating media, in
6660 order to optimize the read-ahead scheme. Previously, it was
6661 only capable of detecting this on traditional file systems
6664 * In udev, additional device properties are now read from the
6665 IAB in addition to the OUI database. Also, Bluetooth company
6666 identities are attached to the devices as well.
6668 * In service files %U may be used as specifier that is
6669 replaced by the configured user name of the service.
6671 * nspawn may now be invoked without a controlling TTY. This
6672 makes it suitable for invocation as its own service. This
6673 may be used to set up a simple containerized server system
6674 using only core OS tools.
6676 * systemd and nspawn can now accept socket file descriptors
6677 when they are started for socket activation. This enables
6678 implementation of socket activated nspawn
6679 containers. i.e. think about autospawning an entire OS image
6680 when the first SSH or HTTP connection is received. We expect
6681 that similar functionality will also be added to libvirt-lxc
6684 * journalctl will now suppress ANSI color codes when
6685 presenting log data.
6687 * systemctl will no longer show control group information for
6688 a unit if the control group is empty anyway.
6690 * logind can now automatically suspend/hibernate/shutdown the
6693 * /etc/machine-info and hostnamed now also expose the chassis
6694 type of the system. This can be used to determine whether
6695 the local system is a laptop, desktop, handset or
6696 tablet. This information may either be configured by the
6697 user/vendor or is automatically determined from ACPI and DMI
6698 information if possible.
6700 * A number of PolicyKit actions are now bound together with
6701 "imply" rules. This should simplify creating UIs because
6702 many actions will now authenticate similar ones as well.
6704 * Unit files learnt a new condition ConditionACPower= which
6705 may be used to conditionalize a unit depending on whether an
6706 AC power source is connected or not, of whether the system
6707 is running on battery power.
6709 * systemctl gained a new "is-failed" verb that may be used in
6710 shell scripts and suchlike to check whether a specific unit
6711 is in the "failed" state.
6713 * The EnvironmentFile= setting in unit files now supports file
6714 globbing, and can hence be used to easily read a number of
6715 environment files at once.
6717 * systemd will no longer detect and recognize specific
6718 distributions. All distribution-specific #ifdeffery has been
6719 removed, systemd is now fully generic and
6720 distribution-agnostic. Effectively, not too much is lost as
6721 a lot of the code is still accessible via explicit configure
6722 switches. However, support for some distribution specific
6723 legacy configuration file formats has been dropped. We
6724 recommend distributions to simply adopt the configuration
6725 files everybody else uses now and convert the old
6726 configuration from packaging scripts. Most distributions
6727 already did that. If that's not possible or desirable,
6728 distributions are welcome to forward port the specific
6729 pieces of code locally from the git history.
6731 * When logging a message about a unit systemd will now always
6732 log the unit name in the message meta data.
6734 * localectl will now also discover system locale data that is
6735 not stored in locale archives, but directly unpacked.
6737 * logind will no longer unconditionally use framebuffer
6738 devices as seat masters, i.e. as devices that are required
6739 to be existing before a seat is considered preset. Instead,
6740 it will now look for all devices that are tagged as
6741 "seat-master" in udev. By default, framebuffer devices will
6742 be marked as such, but depending on local systems, other
6743 devices might be marked as well. This may be used to
6744 integrate graphics cards using closed source drivers (such
6745 as NVidia ones) more nicely into logind. Note however, that
6746 we recommend using the open source NVidia drivers instead,
6747 and no udev rules for the closed-source drivers will be
6748 shipped from us upstream.
6750 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6751 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6752 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6753 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6754 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6755 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6756 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6757 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6758 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6759 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6760 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6761 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6766 * udev gained support for loading additional device properties
6767 from an indexed database that is keyed by vendor/product IDs
6768 and similar device identifiers. For the beginning this
6769 "hwdb" is populated with data from the well-known PCI and
6770 USB database, but also includes PNP, ACPI and OID data. In
6771 the longer run this indexed database shall grow into
6772 becoming the one central database for non-essential
6773 userspace device metadata. Previously, data from the PCI/USB
6774 database was only attached to select devices, since the
6775 lookup was a relatively expensive operation due to O(n) time
6776 complexity (with n being the number of entries in the
6777 database). Since this is now O(1), we decided to add in this
6778 data for all devices where this is available, by
6779 default. Note that the indexed database needs to be rebuilt
6780 when new data files are installed. To achieve this you need
6781 to update your packaging scripts to invoke "udevadm hwdb
6782 --update" after installation of hwdb data files. For
6783 RPM-based distributions we introduced the new
6784 %udev_hwdb_update macro for this purpose.
6786 * The Journal gained support for the "Message Catalog", an
6787 indexed database to link up additional information with
6788 journal entries. For further details please check:
6790 https://www.freedesktop.org/wiki/Software/systemd/catalog
6792 The indexed message catalog database also needs to be
6793 rebuilt after installation of message catalog files. Use
6794 "journalctl --update-catalog" for this. For RPM-based
6795 distributions we introduced the %journal_catalog_update
6796 macro for this purpose.
6798 * The Python Journal bindings gained support for the standard
6799 Python logging framework.
6801 * The Journal API gained new functions for checking whether
6802 the underlying file system of a journal file is capable of
6803 properly reporting file change notifications, or whether
6804 applications that want to reflect journal changes "live"
6805 need to recheck journal files continuously in appropriate
6808 * It is now possible to set the "age" field for tmpfiles
6809 entries to 0, indicating that files matching this entry
6810 shall always be removed when the directories are cleaned up.
6812 * coredumpctl gained a new "gdb" verb which invokes gdb
6813 right-away on the selected coredump.
6815 * There's now support for "hybrid sleep" on kernels that
6816 support this, in addition to "suspend" and "hibernate". Use
6817 "systemctl hybrid-sleep" to make use of this.
6819 * logind's HandleSuspendKey= setting (and related settings)
6820 now gained support for a new "lock" setting to simply
6821 request the screen lock on all local sessions, instead of
6822 actually executing a suspend or hibernation.
6824 * systemd will now mount the EFI variables file system by
6827 * Socket units now gained support for configuration of the
6828 SMACK security label.
6830 * timedatectl will now output the time of the last and next
6831 daylight saving change.
6833 * We dropped support for various legacy and distro-specific
6834 concepts, such as insserv, early-boot SysV services
6835 (i.e. those for non-standard runlevels such as 'b' or 'S')
6836 or ArchLinux /etc/rc.conf support. We recommend the
6837 distributions who still need support this to either continue
6838 to maintain the necessary patches downstream, or find a
6839 different solution. (Talk to us if you have questions!)
6841 * Various systemd components will now bypass PolicyKit checks
6842 for root and otherwise handle properly if PolicyKit is not
6843 found to be around. This should fix most issues for
6844 PolicyKit-less systems. Quite frankly this should have been
6845 this way since day one. It is absolutely our intention to
6846 make systemd work fine on PolicyKit-less systems, and we
6847 consider it a bug if something does not work as it should if
6848 PolicyKit is not around.
6850 * For embedded systems it is now possible to build udev and
6851 systemd without blkid and/or kmod support.
6853 * "systemctl switch-root" is now capable of switching root
6854 more than once. I.e. in addition to transitions from the
6855 initrd to the host OS it is now possible to transition to
6856 further OS images from the host. This is useful to implement
6857 offline updating tools.
6859 * Various other additions have been made to the RPM macros
6860 shipped with systemd. Use %udev_rules_update() after
6861 installing new udev rules files. %_udevhwdbdir,
6862 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6863 %_sysctldir are now available which resolve to the right
6864 directories for packages to place various data files in.
6866 * journalctl gained the new --full switch (in addition to
6867 --all, to disable ellipsation for long messages.
6869 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6870 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6871 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6872 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6873 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6874 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6875 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6876 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6877 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6881 * journalctl gained new --since= and --until= switches to
6882 filter by time. It also now supports nice filtering for
6883 units via --unit=/-u.
6885 * Type=oneshot services may use ExecReload= and do the
6888 * The journal daemon now supports time-based rotation and
6889 vacuuming, in addition to the usual disk-space based
6892 * The journal will now index the available field values for
6893 each field name. This enables clients to show pretty drop
6894 downs of available match values when filtering. The bash
6895 completion of journalctl has been updated
6896 accordingly. journalctl gained a new switch -F to list all
6897 values a certain field takes in the journal database.
6899 * More service events are now written as structured messages
6900 to the journal, and made recognizable via message IDs.
6902 * The timedated, localed and hostnamed mini-services which
6903 previously only provided support for changing time, locale
6904 and hostname settings from graphical DEs such as GNOME now
6905 also have a minimal (but very useful) text-based client
6906 utility each. This is probably the nicest way to changing
6907 these settings from the command line now, especially since
6908 it lists available options and is fully integrated with bash
6911 * There's now a new tool "systemd-coredumpctl" to list and
6912 extract coredumps from the journal.
6914 * We now install a README each in /var/log/ and
6915 /etc/rc.d/init.d explaining where the system logs and init
6916 scripts went. This hopefully should help folks who go to
6917 that dirs and look into the otherwise now empty void and
6918 scratch their heads.
6920 * When user-services are invoked (by systemd --user) the
6921 $MANAGERPID env var is set to the PID of systemd.
6923 * SIGRTMIN+24 when sent to a --user instance will now result
6924 in immediate termination of systemd.
6926 * gatewayd received numerous feature additions such as a
6927 "follow" mode, for live syncing and filtering.
6929 * browse.html now allows filtering and showing detailed
6930 information on specific entries. Keyboard navigation and
6931 mouse screen support has been added.
6933 * gatewayd/journalctl now supports HTML5/JSON
6934 Server-Sent-Events as output.
6936 * The SysV init script compatibility logic will now
6937 heuristically determine whether a script supports the
6938 "reload" verb, and only then make this available as
6941 * "systemctl status --follow" has been removed, use "journalctl
6944 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6945 have been removed since they are hardly useful to be
6948 * And I'd like to take the opportunity to specifically mention
6949 Zbigniew for his great contributions. Zbigniew, you rock!
6951 Contributions from: Andrew Eikum, Christian Hesse, Colin
6952 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6953 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6954 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6955 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6956 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6957 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6961 * If /etc/vconsole.conf is non-existent or empty we will no
6962 longer load any console font or key map at boot by
6963 default. Instead the kernel defaults will be left
6964 intact. This is definitely the right thing to do, as no
6965 configuration should mean no configuration, and hard-coding
6966 font names that are different on all archs is probably a bad
6967 idea. Also, the kernel default key map and font should be
6968 good enough for most cases anyway, and mostly identical to
6969 the userspace fonts/key maps we previously overloaded them
6970 with. If distributions want to continue to default to a
6971 non-kernel font or key map they should ship a default
6972 /etc/vconsole.conf with the appropriate contents.
6974 Contributions from: Colin Walters, Daniel J Walsh, Dave
6975 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6976 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6980 * journalctl gained a new --cursor= switch to show entries
6981 starting from the specified location in the journal.
6983 * We now enforce a size limit on journal entry fields exported
6984 with "-o json" in journalctl. Fields larger than 4K will be
6985 assigned null. This can be turned off with --all.
6987 * An (optional) journal gateway daemon is now available as
6988 "systemd-journal-gatewayd.service". This service provides
6989 access to the journal via HTTP and JSON. This functionality
6990 will be used to implement live log synchronization in both
6991 pull and push modes, but has various other users too, such
6992 as easy log access for debugging of embedded devices. Right
6993 now it is already useful to retrieve the journal via HTTP:
6995 # systemctl start systemd-journal-gatewayd.service
6996 # wget http://localhost:19531/entries
6998 This will download the journal contents in a
6999 /var/log/messages compatible format. The same as JSON:
7001 # curl -H"Accept: application/json" http://localhost:19531/entries
7003 This service is also accessible via a web browser where a
7004 single static HTML5 app is served that uses the JSON logic
7005 to enable the user to do some basic browsing of the
7006 journal. This will be extended later on. Here's an example
7007 screenshot of this app in its current state:
7009 http://0pointer.de/public/journal-gatewayd
7011 Contributions from: Kay Sievers, Lennart Poettering, Robert
7012 Milasan, Tom Gundersen
7016 * The bash completion logic is now available for journalctl
7019 * We do not mount the "cpuset" controller anymore together with
7020 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7021 started if no parameters are assigned to it. "cpuset" hence
7022 broke code that assumed it could create "cpu" groups and
7025 * journalctl -f will now subscribe to terminal size changes,
7026 and line break accordingly.
7028 Contributions from: Dave Reisner, Kay Sievers, Lennart
7029 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7033 * nspawn will now create a symlink /etc/localtime in the
7034 container environment, copying the host's timezone
7035 setting. Previously this has been done via a bind mount, but
7036 since symlinks cannot be bind mounted this has now been
7037 changed to create/update the appropriate symlink.
7039 * journalctl -n's line number argument is now optional, and
7040 will default to 10 if omitted.
7042 * journald will now log the maximum size the journal files may
7043 take up on disk. This is particularly useful if the default
7044 built-in logic of determining this parameter from the file
7045 system size is used. Use "systemctl status
7046 systemd-journald.service" to see this information.
7048 * The multi-seat X wrapper tool has been stripped down. As X
7049 is now capable of enumerating graphics devices via udev in a
7050 seat-aware way the wrapper is not strictly necessary
7051 anymore. A stripped down temporary stop-gap is still shipped
7052 until the upstream display managers have been updated to
7053 fully support the new X logic. Expect this wrapper to be
7054 removed entirely in one of the next releases.
7056 * HandleSleepKey= in logind.conf has been split up into
7057 HandleSuspendKey= and HandleHibernateKey=. The old setting
7058 is not available anymore. X11 and the kernel are
7059 distinguishing between these keys and we should too. This
7060 also means the inhibition lock for these keys has been split
7063 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7064 Poettering, Lukas Nykryn, Václav Pavlín
7068 * Whenever a unit changes state we will now log this to the
7069 journal and show along the unit's own log output in
7072 * ConditionPathIsMountPoint= can now properly detect bind
7073 mount points too. (Previously, a bind mount of one file
7074 system to another place in the same file system could not be
7075 detected as mount, since they shared struct stat's st_dev
7078 * We will now mount the cgroup controllers cpu, cpuacct,
7079 cpuset and the controllers net_cls, net_prio together by
7082 * nspawn containers will now have a virtualized boot
7083 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7084 over with a randomized ID at container initialization). This
7085 has the effect of making "journalctl -b" do the right thing
7088 * The JSON output journal serialization has been updated not
7089 to generate "endless" list objects anymore, but rather one
7090 JSON object per line. This is more in line how most JSON
7091 parsers expect JSON objects. The new output mode
7092 "json-pretty" has been added to provide similar output, but
7093 neatly aligned for readability by humans.
7095 * We dropped all explicit sync() invocations in the shutdown
7096 code. The kernel does this implicitly anyway in the kernel
7097 reboot() syscall. halt(8)'s -n option is now a compatibility
7100 * We now support virtualized reboot() in containers, as
7101 supported by newer kernels. We will fall back to exit() if
7102 CAP_SYS_REBOOT is not available to the container. Also,
7103 nspawn makes use of this now and will actually reboot the
7104 container if the containerized OS asks for that.
7106 * journalctl will only show local log output by default
7107 now. Use --merge (-m) to show remote log output, too.
7109 * libsystemd-journal gained the new sd_journal_get_usage()
7110 call to determine the current disk usage of all journal
7111 files. This is exposed in the new "journalctl --disk-usage"
7114 * journald gained a new configuration setting SplitMode= in
7115 journald.conf which may be used to control how user journals
7116 are split off. See journald.conf(5) for details.
7118 * A new condition type ConditionFileNotEmpty= has been added.
7120 * tmpfiles' "w" lines now support file globbing, to write
7121 multiple files at once.
7123 * We added Python bindings for the journal submission
7124 APIs. More Python APIs for a number of selected APIs will
7125 likely follow. Note that we intend to add native bindings
7126 only for the Python language, as we consider it common
7127 enough to deserve bindings shipped within systemd. There are
7128 various projects outside of systemd that provide bindings
7129 for languages such as PHP or Lua.
7131 * Many conditions will now resolve specifiers such as %i. In
7132 addition, PathChanged= and related directives of .path units
7133 now support specifiers as well.
7135 * There's now a new RPM macro definition for the system preset
7138 * journald will now warn if it ca not forward a message to the
7139 syslog daemon because its socket is full.
7141 * timedated will no longer write or process /etc/timezone,
7142 except on Debian. As we do not support late mounted /usr
7143 anymore /etc/localtime always being a symlink is now safe,
7144 and hence the information in /etc/timezone is not necessary
7147 * logind will now always reserve one VT for a text getty (VT6
7148 by default). Previously if more than 6 X sessions where
7149 started they took up all the VTs with auto-spawned gettys,
7150 so that no text gettys were available anymore.
7152 * udev will now automatically inform the btrfs kernel logic
7153 about btrfs RAID components showing up. This should make
7154 simple hotplug based btrfs RAID assembly work.
7156 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7157 (but not for its children which will stay at the kernel
7158 default). This should allow setups with a lot more listening
7161 * systemd will now always pass the configured timezone to the
7162 kernel at boot. timedated will do the same when the timezone
7165 * logind's inhibition logic has been updated. By default,
7166 logind will now handle the lid switch, the power and sleep
7167 keys all the time, even in graphical sessions. If DEs want
7168 to handle these events on their own they should take the new
7169 handle-power-key, handle-sleep-key and handle-lid-switch
7170 inhibitors during their runtime. A simple way to achieve
7171 that is to invoke the DE wrapped in an invocation of:
7173 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7175 * Access to unit operations is now checked via SELinux taking
7176 the unit file label and client process label into account.
7178 * systemd will now notify the administrator in the journal
7179 when he over-mounts a non-empty directory.
7181 * There are new specifiers that are resolved in unit files,
7182 for the host name (%H), the machine ID (%m) and the boot ID
7185 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7186 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7187 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7188 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7189 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7190 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7191 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7195 * Support for reading structured kernel messages from
7196 /dev/kmsg has now been added and is enabled by default.
7198 * Support for reading kernel messages from /proc/kmsg has now
7199 been removed. If you want kernel messages in the journal
7200 make sure to run a recent kernel (>= 3.5) that supports
7201 reading structured messages from /dev/kmsg (see
7202 above). /proc/kmsg is now exclusive property of classic
7203 syslog daemons again.
7205 * The libudev API gained the new
7206 udev_device_new_from_device_id() call.
7208 * The logic for file system namespace (ReadOnlyDirectory=,
7209 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7210 require pivot_root() anymore. This means fewer temporary
7211 directories are created below /tmp for this feature.
7213 * nspawn containers will now see and receive all submounts
7214 made on the host OS below the root file system of the
7217 * Forward Secure Sealing is now supported for Journal files,
7218 which provide cryptographical sealing of journal files so
7219 that attackers cannot alter log history anymore without this
7220 being detectable. Lennart will soon post a blog story about
7221 this explaining it in more detail.
7223 * There are two new service settings RestartPreventExitStatus=
7224 and SuccessExitStatus= which allow configuration of exit
7225 status (exit code or signal) which will be excepted from the
7226 restart logic, resp. consider successful.
7228 * journalctl gained the new --verify switch that can be used
7229 to check the integrity of the structure of journal files and
7230 (if Forward Secure Sealing is enabled) the contents of
7233 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7234 and similar symlinks pre-created. This makes running shells
7235 as container init process a lot more fun.
7237 * The fstab support can now handle PARTUUID= and PARTLABEL=
7240 * A new ConditionHost= condition has been added to match
7241 against the hostname (with globs) and machine ID. This is
7242 useful for clusters where a single OS image is used to
7243 provision a large number of hosts which shall run slightly
7244 different sets of services.
7246 * Services which hit the restart limit will now be placed in a
7249 Contributions from: Bertram Poettering, Dave Reisner, Huang
7250 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7251 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7255 * When running in --user mode systemd will now become a
7256 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7257 tree a lot more organized.
7259 * A new PartOf= unit dependency type has been introduced that
7260 may be used to group services in a natural way.
7262 * "systemctl enable" may now be used to enable instances of
7265 * journalctl now prints error log levels in red, and
7266 warning/notice log levels in bright white. It also supports
7267 filtering by log level now.
7269 * cgtop gained a new -n switch (similar to top), to configure
7270 the maximum number of iterations to run for. It also gained
7271 -b, to run in batch mode (accepting no input).
7273 * The suffix ".service" may now be omitted on most systemctl
7274 command lines involving service unit names.
7276 * There's a new bus call in logind to lock all sessions, as
7277 well as a loginctl verb for it "lock-sessions".
7279 * libsystemd-logind.so gained a new call sd_journal_perror()
7280 that works similar to libc perror() but logs to the journal
7281 and encodes structured information about the error number.
7283 * /etc/crypttab entries now understand the new keyfile-size=
7286 * shutdown(8) now can send a (configurable) wall message when
7287 a shutdown is cancelled.
7289 * The mount propagation mode for the root file system will now
7290 default to "shared", which is useful to make containers work
7291 nicely out-of-the-box so that they receive new mounts from
7292 the host. This can be undone locally by running "mount
7293 --make-rprivate /" if needed.
7295 * The prefdm.service file has been removed. Distributions
7296 should maintain this unit downstream if they intend to keep
7297 it around. However, we recommend writing normal unit files
7298 for display managers instead.
7300 * Since systemd is a crucial part of the OS we will now
7301 default to a number of compiler switches that improve
7302 security (hardening) such as read-only relocations, stack
7303 protection, and suchlike.
7305 * The TimeoutSec= setting for services is now split into
7306 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7307 of individual time outs for the start and the stop phase of
7310 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7311 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7312 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7313 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7314 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7315 Gundersen, Zbigniew Jędrzejewski-Szmek
7319 * The journal and id128 C APIs are now fully documented as man
7322 * Extra safety checks have been added when transitioning from
7323 the initial RAM disk to the main system to avoid accidental
7326 * /etc/crypttab entries now understand the new keyfile-offset=
7329 * systemctl -t can now be used to filter by unit load state.
7331 * The journal C API gained the new sd_journal_wait() call to
7332 make writing synchronous journal clients easier.
7334 * journalctl gained the new -D switch to show journals from a
7337 * journalctl now displays a special marker between log
7338 messages of two different boots.
7340 * The journal is now explicitly flushed to /var via a service
7341 systemd-journal-flush.service, rather than implicitly simply
7342 by seeing /var/log/journal to be writable.
7344 * journalctl (and the journal C APIs) can now match for much
7345 more complex expressions, with alternatives and
7348 * When transitioning from the initial RAM disk to the main
7349 system we will now kill all processes in a killing spree to
7350 ensure no processes stay around by accident.
7352 * Three new specifiers may be used in unit files: %u, %h, %s
7353 resolve to the user name, user home directory resp. user
7354 shell. This is useful for running systemd user instances.
7356 * We now automatically rotate journal files if their data
7357 object hash table gets a fill level > 75%. We also size the
7358 hash table based on the configured maximum file size. This
7359 together should lower hash collisions drastically and thus
7360 speed things up a bit.
7362 * journalctl gained the new "--header" switch to introspect
7363 header data of journal files.
7365 * A new setting SystemCallFilters= has been added to services
7366 which may be used to apply blacklists or whitelists to
7367 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7369 * nspawn gained a new --link-journal= switch (and quicker: -j)
7370 to link the container journal with the host. This makes it
7371 very easy to centralize log viewing on the host for all
7372 guests while still keeping the journal files separated.
7374 * Many bugfixes and optimizations
7376 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7377 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7378 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7383 * Several tools now understand kernel command line arguments,
7384 which are only read when run in an initial RAM disk. They
7385 usually follow closely their normal counterparts, but are
7388 * There's a new tool to analyze the readahead files that are
7389 automatically generated at boot. Use:
7391 /usr/lib/systemd/systemd-readahead analyze /.readahead
7393 * We now provide an early debug shell on tty9 if this enabled. Use:
7395 systemctl enable debug-shell.service
7397 * All plymouth related units have been moved into the Plymouth
7398 package. Please make sure to upgrade your Plymouth version
7401 * systemd-tmpfiles now supports getting passed the basename of
7402 a configuration file only, in which case it will look for it
7403 in all appropriate directories automatically.
7405 * udevadm info now takes a /dev or /sys path as argument, and
7406 does the right thing. Example:
7408 udevadm info /dev/sda
7409 udevadm info /sys/class/block/sda
7411 * systemctl now prints a warning if a unit is stopped but a
7412 unit that might trigger it continues to run. Example: a
7413 service is stopped but the socket that activates it is left
7416 * "systemctl status" will now mention if the log output was
7417 shortened due to rotation since a service has been started.
7419 * The journal API now exposes functions to determine the
7420 "cutoff" times due to rotation.
7422 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7423 immediately flushing of runtime logs to /var if possible,
7424 resp. for triggering immediate rotation of the journal
7427 * It is now considered an error if a service is attempted to
7428 be stopped that is not loaded.
7430 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7432 * systemd-analyze now supports Python 3
7434 * tmpfiles now supports cleaning up directories via aging
7435 where the first level dirs are always kept around but
7436 directories beneath it automatically aged. This is enabled
7437 by prefixing the age field with '~'.
7439 * Seat objects now expose CanGraphical, CanTTY properties
7440 which is required to deal with very fast bootups where the
7441 display manager might be running before the graphics drivers
7442 completed initialization.
7444 * Seat objects now expose a State property.
7446 * We now include RPM macros for service enabling/disabling
7447 based on the preset logic. We recommend RPM based
7448 distributions to make use of these macros if possible. This
7449 makes it simpler to reuse RPM spec files across
7452 * We now make sure that the collected systemd unit name is
7453 always valid when services log to the journal via
7456 * There's a new man page kernel-command-line(7) detailing all
7457 command line options we understand.
7459 * The fstab generator may now be disabled at boot by passing
7460 fstab=0 on the kernel command line.
7462 * A new kernel command line option modules-load= is now understood
7463 to load a specific kernel module statically, early at boot.
7465 * Unit names specified on the systemctl command line are now
7466 automatically escaped as needed. Also, if file system or
7467 device paths are specified they are automatically turned
7468 into the appropriate mount or device unit names. Example:
7470 systemctl status /home
7471 systemctl status /dev/sda
7473 * The SysVConsole= configuration option has been removed from
7474 system.conf parsing.
7476 * The SysV search path is no longer exported on the D-Bus
7479 * The Names= option has been removed from unit file parsing.
7481 * There's a new man page bootup(7) detailing the boot process.
7483 * Every unit and every generator we ship with systemd now
7484 comes with full documentation. The self-explanatory boot is
7487 * A couple of services gained "systemd-" prefixes in their
7488 name if they wrap systemd code, rather than only external
7489 code. Among them fsck@.service which is now
7490 systemd-fsck@.service.
7492 * The HaveWatchdog property has been removed from the D-Bus
7495 * systemd.confirm_spawn= on the kernel command line should now
7498 * There's a new man page crypttab(5) which details all options
7499 we actually understand.
7501 * systemd-nspawn gained a new --capability= switch to pass
7502 additional capabilities to the container.
7504 * timedated will now read known NTP implementation unit names
7505 from /usr/lib/systemd/ntp-units.d/*.list,
7506 systemd-timedated-ntp.target has been removed.
7508 * journalctl gained a new switch "-b" that lists log data of
7509 the current boot only.
7511 * The notify socket is in the abstract namespace again, in
7512 order to support daemons which chroot() at start-up.
7514 * There is a new Storage= configuration option for journald
7515 which allows configuration of where log data should go. This
7516 also provides a way to disable journal logging entirely, so
7517 that data collected is only forwarded to the console, the
7518 kernel log buffer or another syslog implementation.
7520 * Many bugfixes and optimizations
7522 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7523 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7524 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7525 Shawn Landden, Tom Gundersen
7529 * "systemctl help <unit>" now shows the man page if one is
7532 * Several new man pages have been added.
7534 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7535 MaxLevelConsole= can now be specified in
7536 journald.conf. These options allow reducing the amount of
7537 data stored on disk or forwarded by the log level.
7539 * TimerSlackNSec= can now be specified in system.conf for
7540 PID1. This allows system-wide power savings.
7542 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7543 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7548 * logind is now capable of (optionally) handling power and
7549 sleep keys as well as the lid switch.
7551 * journalctl now understands the syntax "journalctl
7552 /usr/bin/avahi-daemon" to get all log output of a specific
7555 * CapabilityBoundingSet= in system.conf now also influences
7556 the capability bound set of usermode helpers of the kernel.
7558 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7559 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7560 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7561 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7565 * Note that we skipped 139 releases here in order to set the
7566 new version to something that is greater than both udev's
7567 and systemd's most recent version number.
7569 * udev: all udev sources are merged into the systemd source tree now.
7570 All future udev development will happen in the systemd tree. It
7571 is still fully supported to use the udev daemon and tools without
7572 systemd running, like in initramfs or other init systems. Building
7573 udev though, will require the *build* of the systemd tree, but
7574 udev can be properly *run* without systemd.
7576 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7577 should be used to create dead device nodes as workarounds for broken
7580 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7581 no longer supported. udev_monitor_new_from_netlink() needs to be
7582 used to subscribe to events.
7584 * udev: when udevd is started by systemd, processes which are left
7585 behind by forking them off of udev rules, are unconditionally cleaned
7586 up and killed now after the event handling has finished. Services or
7587 daemons must be started as systemd services. Services can be
7588 pulled-in by udev to get started, but they can no longer be directly
7589 forked by udev rules.
7591 * udev: the daemon binary is called systemd-udevd now and installed
7592 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7593 to adapt to that, create symlink, or rename the binary after building
7596 * libudev no longer provides these symbols:
7597 udev_monitor_from_socket()
7598 udev_queue_get_failed_list_entry()
7599 udev_get_{dev,sys,run}_path()
7600 The versions number was bumped and symbol versioning introduced.
7602 * systemd-loginctl and systemd-journalctl have been renamed
7603 to loginctl and journalctl to match systemctl.
7605 * The config files: /etc/systemd/systemd-logind.conf and
7606 /etc/systemd/systemd-journald.conf have been renamed to
7607 logind.conf and journald.conf. Package updates should rename
7608 the files to the new names on upgrade.
7610 * For almost all files the license is now LGPL2.1+, changed
7611 from the previous GPL2.0+. Exceptions are some minor stuff
7612 of udev (which will be changed to LGPL2.1 eventually, too),
7613 and the MIT licensed sd-daemon.[ch] library that is suitable
7614 to be used as drop-in files.
7616 * systemd and logind now handle system sleep states, in
7617 particular suspending and hibernating.
7619 * logind now implements a sleep/shutdown/idle inhibiting logic
7620 suitable for a variety of uses. Soonishly Lennart will blog
7621 about this in more detail.
7623 * var-run.mount and var-lock.mount are no longer provided
7624 (which previously bind mounted these directories to their new
7625 places). Distributions which have not converted these
7626 directories to symlinks should consider stealing these files
7627 from git history and add them downstream.
7629 * We introduced the Documentation= field for units and added
7630 this to all our shipped units. This is useful to make it
7631 easier to explore the boot and the purpose of the various
7634 * All smaller setup units (such as
7635 systemd-vconsole-setup.service) now detect properly if they
7636 are run in a container and are skipped when
7637 appropriate. This guarantees an entirely noise-free boot in
7638 Linux container environments such as systemd-nspawn.
7640 * A framework for implementing offline system updates is now
7641 integrated, for details see:
7642 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7644 * A new service type Type=idle is available now which helps us
7645 avoiding ugly interleaving of getty output and boot status
7648 * There's now a system-wide CapabilityBoundingSet= option to
7649 globally reduce the set of capabilities for the
7650 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7651 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7652 even CAP_NET_ADMIN system-wide for secure systems.
7654 * There are now system-wide DefaultLimitXXX= options to
7655 globally change the defaults of the various resource limits
7656 for all units started by PID 1.
7658 * Harald Hoyer's systemd test suite has been integrated into
7659 systemd which allows easy testing of systemd builds in qemu
7660 and nspawn. (This is really awesome! Ask us for details!)
7662 * The fstab parser is now implemented as generator, not inside
7665 * systemctl will now warn you if .mount units generated from
7666 /etc/fstab are out of date due to changes in fstab that
7667 have not been read by systemd yet.
7669 * systemd is now suitable for usage in initrds. Dracut has
7670 already been updated to make use of this. With this in place
7671 initrds get a slight bit faster but primarily are much
7672 easier to introspect and debug since "systemctl status" in
7673 the host system can be used to introspect initrd services,
7674 and the journal from the initrd is kept around too.
7676 * systemd-delta has been added, a tool to explore differences
7677 between user/admin configuration and vendor defaults.
7679 * PrivateTmp= now affects both /tmp and /var/tmp.
7681 * Boot time status messages are now much prettier and feature
7682 proper english language. Booting up systemd has never been
7685 * Read-ahead pack files now include the inode number of all
7686 files to pre-cache. When the inode changes the pre-caching
7687 is not attempted. This should be nicer to deal with updated
7688 packages which might result in changes of read-ahead
7691 * We now temporaritly lower the kernel's read_ahead_kb variable
7692 when collecting read-ahead data to ensure the kernel's
7693 built-in read-ahead does not add noise to our measurements
7694 of necessary blocks to pre-cache.
7696 * There's now RequiresMountsFor= to add automatic dependencies
7697 for all mounts necessary for a specific file system path.
7699 * MountAuto= and SwapAuto= have been removed from
7700 system.conf. Mounting file systems at boot has to take place
7703 * nspawn now learned a new switch --uuid= to set the machine
7704 ID on the command line.
7706 * nspawn now learned the -b switch to automatically search
7709 * vt102 is now the default TERM for serial TTYs, upgraded from
7712 * systemd-logind now works on VT-less systems.
7714 * The build tree has been reorganized. The individual
7715 components now have directories of their own.
7717 * A new condition type ConditionPathIsReadWrite= is now available.
7719 * nspawn learned the new -C switch to create cgroups for the
7720 container in other hierarchies.
7722 * We now have support for hardware watchdogs, configurable in
7725 * The scheduled shutdown logic now has a public API.
7727 * We now mount /tmp as tmpfs by default, but this can be
7728 masked and /etc/fstab can override it.
7730 * Since udisks does not make use of /media anymore we are not
7731 mounting a tmpfs on it anymore.
7733 * journalctl gained a new --local switch to only interleave
7734 locally generated journal files.
7736 * We can now load the IMA policy at boot automatically.
7738 * The GTK tools have been split off into a systemd-ui.
7740 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7741 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7742 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7743 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7744 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7745 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7746 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7747 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7748 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7753 * This is mostly a bugfix release
7755 * Support optional initialization of the machine ID from the
7756 KVM or container configured UUID.
7758 * Support immediate reboots with "systemctl reboot -ff"
7760 * Show /etc/os-release data in systemd-analyze output
7762 * Many bugfixes for the journal, including endianness fixes and
7763 ensuring that disk space enforcement works
7765 * sd-login.h is C++ compatible again
7767 * Extend the /etc/os-release format on request of the Debian
7770 * We now refuse non-UTF8 strings used in various configuration
7771 and unit files. This is done to ensure we do not pass invalid
7772 data over D-Bus or expose it elsewhere.
7774 * Register Mimo USB Screens as suitable for automatic seat
7777 * Read SELinux client context from journal clients in a race
7780 * Reorder configuration file lookup order. /etc now always
7781 overrides /run in order to allow the administrator to always
7782 and unconditionally override vendor-supplied or
7783 automatically generated data.
7785 * The various user visible bits of the journal now have man
7786 pages. We still lack man pages for the journal API calls
7789 * We now ship all man pages in HTML format again in the
7792 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7793 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7794 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7799 * This is mostly a bugfix release
7801 * systems lacking /etc/os-release are no longer supported.
7803 * Various functionality updates to libsystemd-login.so
7805 * Track class of PAM logins to distinguish greeters from
7808 Contributions from: Kay Sievers, Lennart Poettering, Michael
7813 * This is an important bugfix release for v41.
7815 * Building man pages is now optional which should be useful
7816 for those building systemd from git but unwilling to install
7819 * Watchdog support for supervising services is now usable. In
7820 a future release support for hardware watchdogs
7821 (i.e. /dev/watchdog) will be added building on this.
7823 * Service start rate limiting is now configurable and can be
7824 turned off per service. When a start rate limit is hit a
7825 reboot can automatically be triggered.
7827 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7829 Contributions from: Benjamin Franzke, Bill Nottingham,
7830 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7831 Schmidt, Michał Górny, Piotr Drąg
7835 * The systemd binary is installed /usr/lib/systemd/systemd now;
7836 An existing /sbin/init symlink needs to be adapted with the
7839 * The code that loads kernel modules has been ported to invoke
7840 libkmod directly, instead of modprobe. This means we do not
7841 support systems with module-init-tools anymore.
7843 * Watchdog support is now already useful, but still not
7846 * A new kernel command line option systemd.setenv= is
7847 understood to set system wide environment variables
7848 dynamically at boot.
7850 * We now limit the set of capabilities of systemd-journald.
7852 * We now set SIGPIPE to ignore by default, since it only is
7853 useful in shell pipelines, and has little use in general
7854 code. This can be disabled with IgnoreSIPIPE=no in unit
7857 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7858 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7863 * This is mostly a bugfix release
7865 * We now expose the reason why a service failed in the
7866 "Result" D-Bus property.
7868 * Rudimentary service watchdog support (will be completed over
7869 the next few releases.)
7871 * When systemd forks off in order execute some service we will
7872 now immediately changes its argv[0] to reflect which process
7873 it will execute. This is useful to minimize the time window
7874 with a generic argv[0], which makes bootcharts more useful
7876 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7877 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7878 Mike Kazantsev, Ray Strode
7882 * This is mostly a test release, but incorporates many
7885 * New systemd-cgtop tool to show control groups by their
7888 * Linking against libacl for ACLs is optional again. If
7889 disabled, support tracking device access for active logins
7890 goes becomes unavailable, and so does access to the user
7891 journals by the respective users.
7893 * If a group "adm" exists, journal files are automatically
7894 owned by them, thus allow members of this group full access
7895 to the system journal as well as all user journals.
7897 * The journal now stores the SELinux context of the logging
7898 client for all entries.
7900 * Add C++ inclusion guards to all public headers
7902 * New output mode "cat" in the journal to print only text
7903 messages, without any meta data like date or time.
7905 * Include tiny X server wrapper as a temporary stop-gap to
7906 teach XOrg udev display enumeration. This is used by display
7907 managers such as gdm, and will go away as soon as XOrg
7908 learned native udev hotplugging for display devices.
7910 * Add new systemd-cat tool for executing arbitrary programs
7911 with STDERR/STDOUT connected to the journal. Can also act as
7912 BSD logger replacement, and does so by default.
7914 * Optionally store all locally generated coredumps in the
7915 journal along with meta data.
7917 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7918 writing short strings to files (for usage for /sys), and for
7919 creating symlinks, character and block device nodes.
7921 * New unit file option ControlGroupPersistent= to make cgroups
7922 persistent, following the mechanisms outlined in
7923 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7925 * Support multiple local RTCs in a sane way
7927 * No longer monopolize IO when replaying readahead data on
7928 rotating disks, since we might starve non-file-system IO to
7929 death, since fanotify() will not see accesses done by blkid,
7932 * Do not show kernel threads in systemd-cgls anymore, unless
7933 requested with new -k switch.
7935 Contributions from: Dan Horák, Kay Sievers, Lennart
7936 Poettering, Michal Schmidt
7940 * This is mostly a test release, but incorporates many
7943 * The git repository moved to:
7944 git://anongit.freedesktop.org/systemd/systemd
7945 ssh://git.freedesktop.org/git/systemd/systemd
7947 * First release with the journal
7948 http://0pointer.de/blog/projects/the-journal.html
7950 * The journal replaces both systemd-kmsg-syslogd and
7951 systemd-stdout-bridge.
7953 * New sd_pid_get_unit() API call in libsystemd-logind
7955 * Many systemadm clean-ups
7957 * Introduce remote-fs-pre.target which is ordered before all
7958 remote mounts and may be used to start services before all
7961 * Added Mageia support
7963 * Add bash completion for systemd-loginctl
7965 * Actively monitor PID file creation for daemons which exit in
7966 the parent process before having finished writing the PID
7967 file in the daemon process. Daemons which do this need to be
7968 fixed (i.e. PID file creation must have finished before the
7969 parent exits), but we now react a bit more gracefully to them.
7971 * Add colourful boot output, mimicking the well-known output
7972 of existing distributions.
7974 * New option PassCredentials= for socket units, for
7975 compatibility with a recent kernel ABI breakage.
7977 * /etc/rc.local is now hooked in via a generator binary, and
7978 thus will no longer act as synchronization point during
7981 * systemctl list-unit-files now supports --root=.
7983 * systemd-tmpfiles now understands two new commands: z, Z for
7984 relabelling files according to the SELinux database. This is
7985 useful to apply SELinux labels to specific files in /sys,
7988 * Output of SysV services is now forwarded to both the console
7989 and the journal by default, not only just the console.
7991 * New man pages for all APIs from libsystemd-login.
7993 * The build tree got reorganized and the build system is a
7994 lot more modular allowing embedded setups to specifically
7995 select the components of systemd they are interested in.
7997 * Support for Linux systems lacking the kernel VT subsystem is
8000 * configure's --with-rootdir= got renamed to
8001 --with-rootprefix= to follow the naming used by udev and
8004 * Unless specified otherwise we will now install to /usr instead
8005 of /usr/local by default.
8007 * Processes with '@' in argv[0][0] are now excluded from the
8008 final shut-down killing spree, following the logic explained
8010 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8012 * All processes remaining in a service cgroup when we enter
8013 the START or START_PRE states are now killed with
8014 SIGKILL. That means it is no longer possible to spawn
8015 background processes from ExecStart= lines (which was never
8016 supported anyway, and bad style).
8018 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8019 reloading of units together.
8021 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8022 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8023 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8024 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8025 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek