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: systemd-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 systemd-tmpfiles was run. This behaviour was
14 different from what the documentation said, and not particularly
15 useful, as repeated systemd-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 systemd-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 systemd-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 systemd-networkd's .netdev and
45 .network files all gained support for a new condition
46 ConditionKernelVersion= for checking against specific kernel
49 * In systemd-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 * systemd-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. systemd-networkd and
66 systemd-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 systemd'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 * systemd-networkd gained support for natively configuring WireGuard
122 * In previous versions systemd synthesized user records both for the
123 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd 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/systemd/dont-synthesize-nobody exists synthesizing of the 65534
128 user and group record within the systemd codebase is disabled.
130 * systemd-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 "systemd-analyze service-watchdogs" command for printing
135 the current state of the service runtime watchdog, and optionally
136 enabling or disabling the per-service watchdogs system-wide if given a
137 boolean argument (i.e. the concept you configure in WatchdogSec=), for
138 debugging purposes. There's also a kernel command line option
139 systemd.service_watchdogs= for controlling the same.
141 * Two new "log-level" and "log-target" options for systemd-analyze were
142 addded that merge the now deprecated get-log-level, set-log-level and
143 get-log-target, set-log-target pairs. The deprecated options are still
144 understood for backwards compatibility. The two new options print the
145 current value when no arguments are given, and set them when a
146 level/target is given as an argument.
148 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
149 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Björn Esser,
150 bleep_blop, Bruce A. Johnson, Chris Down, Clinton Roy, Colin Walters,
151 Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov, Evgeny Vereshchagin,
152 Ewout van Mansom, Felipe Sateler, Franck Bui, Frantisek Sumsal, George
153 Gaydarov, Gianluca Boiano, Hans-Christian Noren Egtvedt, Hans de Goede,
154 Henrik Grindal Bakken, Jan Alexander Steffens, Jan Klötzke, Jason
155 A. Donenfeld, jdkbx, Jérémy Rosen, Jerónimo Borque, John Lin, John Paul
156 Herold, Jonathan Rudenberg, Jörg Thalheim, Ken (Bitsko) MacLeod, Larry
157 Bernstone, Lennart Poettering, Lucas Werkmeister, Maciej S. Szmigiero,
158 Marek Čermák, Martin Pitt, Mathieu Malaterre, Matthew Thode,
159 Matthias-Christian Ott, Max Harmathy, Michael Biebl, Michael Vogt,
160 Michal Koutný, Michal Sekletar, Michał Szczepański, Mike Gilbert,
161 Nathaniel McCallum, Nicolas Chauvet, Olaf Hering, Olivier Schwander,
162 Patrik Flykt, Paul Cercueil, Peter Hutterer, Piotr Drąg, Raphael
163 Vogelgsang, Reverend Homer, Robert Kolchmeyer, Samuel Dionne-Riel,
164 Sergey Ptashnick, Shawn Landden, Susant Sahani, Sylvain Plantefève,
165 Thomas Huth, Tomasz Bachorski, Vladislav Vishnyakov, Wieland Hoffmann,
166 Yu Watanabe, Zachary Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан
167 Георгиевски, Дилян Палаузов
173 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
174 in v235 has been extended to also set the dummy.ko module option
175 numdummies=0, preventing the kernel from automatically creating
176 dummy0. All dummy interfaces must now be explicitly created.
178 * Unknown '%' specifiers in configuration files are now rejected. This
179 applies to units and tmpfiles.d configuration. Any percent characters
180 that are followed by a letter or digit that are not supposed to be
181 interpreted as the beginning of a specifier should be escaped by
182 doubling ("%%"). (So "size=5%" is still accepted, as well as
183 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
184 valid specifiers today.)
186 * systemd-resolved now maintains a new dynamic
187 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
188 recommended to make /etc/resolv.conf a symlink to it. This file
189 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
190 includes dynamically acquired search domains, achieving more correct
191 DNS resolution by software that bypasses local DNS APIs such as NSS.
193 * The "uaccess" udev tag has been dropped from /dev/kvm and
194 /dev/dri/renderD*. These devices now have the 0666 permissions by
195 default (but this may be changed at build-time). /dev/dri/renderD*
196 will now be owned by the "render" group along with /dev/kfd.
198 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
199 systemd-journal-gatewayd.service and
200 systemd-journal-upload.service. This means "nss-systemd" must be
201 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
202 services are resolved properly.
204 * In /etc/fstab two new mount options are now understood:
205 x-systemd.makefs and x-systemd.growfs. The former has the effect that
206 the configured file system is formatted before it is mounted, the
207 latter that the file system is resized to the full block device size
208 after it is mounted (i.e. if the file system is smaller than the
209 partition it resides on, it's grown). This is similar to the fsck
210 logic in /etc/fstab, and pulls in systemd-makefs@.service and
211 systemd-growfs@.service as necessary, similar to
212 systemd-fsck@.service. Resizing is currently only supported on ext4
215 * In systemd-networkd, the IPv6 RA logic now optionally may announce
216 DNS server and domain information.
218 * Support for the LUKS2 on-disk format for encrypted partitions has
219 been added. This requires libcryptsetup2 during compilation and
222 * The systemd --user instance will now signal "readiness" when its
223 basic.target unit has been reached, instead of when the run queue ran
224 empty for the first time.
226 * Tmpfiles.d with user configuration are now also supported.
227 systemd-tmpfiles gained a new --user switch, and snippets placed in
228 ~/.config/user-tmpfiles.d/ and corresponding directories will be
229 executed by systemd-tmpfiles --user running in the new
230 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
231 running in the user session.
233 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
234 %S resolves to the top-level state directory (/var/lib for the system
235 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
236 top-level cache directory (/var/cache for the system instance,
237 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
238 logs directory (/var/log for the system instance,
239 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
240 existing %t specifier, that resolves to the top-level runtime
241 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
244 * journalctl learnt a new parameter --output-fields= for limiting the
245 set of journal fields to output in verbose and JSON output modes.
247 * systemd-timesyncd's configuration file gained a new option
248 RootDistanceMaxSec= for setting the maximum root distance of servers
249 it'll use, as well as the new options PollIntervalMinSec= and
250 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
252 * bootctl gained a new command "list" for listing all available boot
253 menu items on systems that follow the boot loader specification.
255 * systemctl gained a new --dry-run switch that shows what would be done
256 instead of doing it, and is currently supported by the shutdown and
259 * ConditionSecurity= can now detect the TOMOYO security module.
261 * Unit file [Install] sections are now also respected in unit drop-in
262 files. This is intended to be used by drop-ins under /usr/lib/.
264 * systemd-firstboot may now also set the initial keyboard mapping.
266 * Udev "changed" events for devices which are exposed as systemd
267 .device units are now propagated to units specified in
268 ReloadPropagatedFrom= as reload requests.
270 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
271 unit template name (i.e. a name in the form of 'foobar@.service',
272 without the instance component between the '@' and - the '.'), then
273 the escaped sysfs path of the device is automatically used as the
276 * SystemCallFilter= in unit files has been extended so that an "errno"
277 can be specified individually for each system call. Example:
278 SystemCallFilter=~uname:EILSEQ.
280 * The cgroup delegation logic has been substantially updated. Delegate=
281 now optionally takes a list of controllers (instead of a boolean, as
282 before), which lists the controllers to delegate at least.
284 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
286 * A new LogLevelMax= setting configures the maximum log level any
287 process of the service may log at (i.e. anything with a lesser
288 priority than what is specified is automatically dropped). A new
289 LogExtraFields= setting allows configuration of additional journal
290 fields to attach to all log records generated by any of the unit's
293 * New StandardInputData= and StandardInputText= settings along with the
294 new option StandardInput=data may be used to configure textual or
295 binary data that shall be passed to the executed service process via
296 standard input, encoded in-line in the unit file.
298 * StandardInput=, StandardOutput= and StandardError= may now be used to
299 connect stdin/stdout/stderr of executed processes directly with a
300 file or AF_UNIX socket in the file system, using the new "file:" option.
302 * A new unit file option CollectMode= has been added, that allows
303 tweaking the garbage collection logic for units. It may be used to
304 tell systemd to garbage collect units that have failed automatically
305 (normally it only GCs units that exited successfully). systemd-run
306 and systemd-mount expose this new functionality with a new -G option.
308 * "machinectl bind" may now be used to bind mount non-directories
309 (i.e. regularfiles, devices, fifos, sockets).
311 * systemd-analyze gained a new verb "calendar" for validating and
312 testing calendar time specifications to use for OnCalendar= in timer
313 units. Besides validating the expression it will calculate the next
314 time the specified expression would elapse.
316 * In addition to the pre-existing FailureAction= unit file setting
317 there's now SuccessAction=, for configuring a shutdown action to
318 execute when a unit completes successfully. This is useful in
319 particular inside containers that shall terminate after some workload
320 has been completed. Also, both options are now supported for all unit
321 types, not just services.
323 * networkds's IP rule support gained two new options
324 IncomingInterface= and OutgoingInterface= for configuring the incoming
325 and outgoing interfaces of configured rules. systemd-networkd also
326 gained support for "vxcan" network devices.
328 * networkd gained a new setting RequiredForOnline=, taking a
329 boolean. If set, systemd-wait-online will take it into consideration
330 when determining that the system is up, otherwise it will ignore the
331 interface for this purpose.
333 * The sd_notify() protocol gained support for a new operation: with
334 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
335 store again, ahead of POLLHUP or POLLERR when they are removed
338 * A new document UIDS-GIDS.md has been added to the source tree, that
339 documents the UID/GID range and assignment assumptions and
340 requirements of systemd.
342 * The watchdog device PID 1 will ping may now be configured through the
343 WatchdogDevice= configuration file setting, or by setting the
344 systemd.watchdog_service= kernel commandline option.
346 * systemd-resolved's gained support for registering DNS-SD services on
347 the local network using MulticastDNS. Services may either be
348 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
349 the same dir below /run, /usr/lib), or through its D-Bus API.
351 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
352 extend the effective start, runtime, and stop time. The service must
353 continue to send EXTEND_TIMEOUT_USEC within the period specified to
354 prevent the service manager from making the service as timedout.
356 * systemd-resolved's DNSSEC support gained support for RFC 8080
357 (Ed25519 keys and signatures).
359 * The systemd-resolve command line tool gained a new set of options
360 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
361 --set-nta= and --revert to configure per-interface DNS configuration
362 dynamically during runtime. It's useful for pushing DNS information
363 into systemd-resolved from DNS hook scripts that various interface
364 managing software supports (such as pppd).
366 * systemd-nspawn gained a new --network-namespace-path= command line
367 option, which may be used to make a container join an existing
368 network namespace, by specifying a path to a "netns" file.
370 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
371 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
372 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
373 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
374 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
375 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
376 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
377 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
378 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
379 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
380 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
381 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
382 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
383 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
384 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
385 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
386 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
387 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
388 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
389 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
390 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
391 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
392 Jędrzejewski-Szmek, Zeal Jagannatha
398 * INCOMPATIBILITY: systemd-logind.service and other long-running
399 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
400 communication with the outside. This generally improves security of
401 the system, and is in almost all cases a safe and good choice, as
402 these services do not and should not provide any network-facing
403 functionality. However, systemd-logind uses the glibc NSS API to
404 query the user database. This creates problems on systems where NSS
405 is set up to directly consult network services for user database
406 lookups. In particular, this creates incompatibilities with the
407 "nss-nis" module, which attempts to directly contact the NIS/YP
408 network servers it is configured for, and will now consistently
409 fail. In such cases, it is possible to turn off IP sandboxing for
410 systemd-logind.service (set IPAddressDeny= in its [Service] section
411 to the empty string, via a .d/ unit file drop-in). Downstream
412 distributions might want to update their nss-nis packaging to include
413 such a drop-in snippet, accordingly, to hide this incompatibility
414 from the user. Another option is to make use of glibc's nscd service
415 to proxy such network requests through a privilege-separated, minimal
416 local caching daemon, or to switch to more modern technologies such
417 sssd, whose NSS hook-ups generally do not involve direct network
418 access. In general, we think it's definitely time to question the
419 implementation choices of nss-nis, i.e. whether it's a good idea
420 today to embed a network-facing loadable module into all local
421 processes that need to query the user database, including the most
422 trivial and benign ones, such as "ls". For more details about
423 IPAddressDeny= see below.
425 * A new modprobe.d drop-in is now shipped by default that sets the
426 bonding module option max_bonds=0. This overrides the kernel default,
427 to avoid conflicts and ambiguity as to whether or not bond0 should be
428 managed by systemd-networkd or not. This resolves multiple issues
429 with bond0 properties not being applied, when bond0 is configured
430 with systemd-networkd. Distributors may choose to not package this,
431 however in that case users will be prevented from correctly managing
432 bond0 interface using systemd-networkd.
434 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
435 which print the logging level and target of the system manager. They
436 complement the existing "set-log-level" and "set-log-target" verbs
437 used to change those values.
439 * journald.conf gained a new boolean setting ReadKMsg= which defaults
440 to on. If turned off kernel log messages will not be read by
441 systemd-journald or included in the logs. It also gained a new
442 setting LineMax= for configuring the maximum line length in
443 STDOUT/STDERR log streams. The new default for this value is 48K, up
444 from the previous hardcoded 2048.
446 * A new unit setting RuntimeDirectoryPreserve= has been added, which
447 allows more detailed control of what to do with a runtime directory
448 configured with RuntimeDirectory= (i.e. a directory below /run or
449 $XDG_RUNTIME_DIR) after a unit is stopped.
451 * The RuntimeDirectory= setting for units gained support for creating
452 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
453 one top-level directory.
455 * Units gained new options StateDirectory=, CacheDirectory=,
456 LogsDirectory= and ConfigurationDirectory= which are closely related
457 to RuntimeDirectory= but manage per-service directories below
458 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
459 possible to write unit files which when activated automatically gain
460 properly owned service specific directories in these locations, thus
461 making unit files self-contained and increasing compatibility with
462 stateless systems and factory reset where /etc or /var are
463 unpopulated at boot. Matching these new settings there's also
464 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
465 ConfigurationDirectoryMode= for configuring the access mode of these
466 directories. These settings are particularly useful in combination
467 with DynamicUser=yes as they provide secure, properly-owned,
468 writable, and stateful locations for storage, excluded from the
469 sandbox that such services live in otherwise.
471 * Automake support has been removed from this release. systemd is now
474 * systemd-journald will now aggressively cache client metadata during
475 runtime, speeding up log write performance under pressure. This comes
476 at a small price though: as much of the metadata is read
477 asynchronously from /proc/ (and isn't implicitly attached to log
478 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
479 metadata stored alongside a log entry might be slightly
480 out-of-date. Previously it could only be slightly newer than the log
481 message. The time window is small however, and given that the kernel
482 is unlikely to be improved anytime soon in this regard, this appears
485 * nss-myhostname/systemd-resolved will now by default synthesize an
486 A/AAAA resource record for the "_gateway" hostname, pointing to the
487 current default IP gateway. Previously it did that for the "gateway"
488 name, hampering adoption, as some distributions wanted to leave that
489 host name open for local use. The old behaviour may still be
490 requested at build time.
492 * systemd-networkd's [Address] section in .network files gained a new
493 Scope= setting for configuring the IP address scope. The [Network]
494 section gained a new boolean setting ConfigureWithoutCarrier= that
495 tells systemd-networkd to ignore link sensing when configuring the
496 device. The [DHCP] section gained a new Anonymize= boolean option for
497 turning on a number of options suggested in RFC 7844. A new
498 [RoutingPolicyRule] section has been added for configuring the IP
499 routing policy. The [Route] section has gained support for a new
500 Type= setting which permits configuring
501 blackhole/unreachable/prohibit routes.
503 * The [VRF] section in .netdev files gained a new Table= setting for
504 configuring the routing table to use. The [Tunnel] section gained a
505 new Independent= boolean field for configuring tunnels independent of
506 an underlying network interface. The [Bridge] section gained a new
507 GroupForwardMask= option for configuration of propagation of link
508 local frames between bridge ports.
510 * The WakeOnLan= setting in .link files gained support for a number of
511 new modes. A new TCP6SegmentationOffload= setting has been added for
512 configuring TCP/IPv6 hardware segmentation offload.
514 * The IPv6 RA sender implementation may now optionally send out RDNSS
515 and RDNSSL records to supply DNS configuration to peers.
517 * systemd-nspawn gained support for a new --system-call-filter= command
518 line option for adding and removing entries in the default system
519 call filter it applies. Moreover systemd-nspawn has been changed to
520 implement a system call whitelist instead of a blacklist.
522 * systemd-run gained support for a new --pipe command line option. If
523 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
524 are directly passed on to the activated transient service
525 executable. This allows invoking arbitrary processes as systemd
526 services (for example to take benefit of dependency management,
527 accounting management, resource management or log management that is
528 done automatically for services) — while still allowing them to be
529 integrated in a classic UNIX shell pipeline.
531 * When a service sends RELOAD=1 via sd_notify() and reload propagation
532 using ReloadPropagationTo= is configured, a reload is now propagated
533 to configured units. (Previously this was only done on explicitly
534 requested reloads, using "systemctl reload" or an equivalent
537 * For each service unit a restart counter is now kept: it is increased
538 each time the service is restarted due to Restart=, and may be
539 queried using "systemctl show -p NRestarts …".
541 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
542 @signal and @timer have been added, for usage with SystemCallFilter=
543 in unit files and the new --system-call-filter= command line option
544 of systemd-nspawn (see above).
546 * ExecStart= lines in unit files gained two new modifiers: when a
547 command line is prefixed with "!" the command will be executed as
548 configured, except for the credentials applied by
549 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
550 "+", but does still apply namespacing options unlike "+". There's
551 also "!!" now, which is mostly identical, but becomes a NOP on
552 systems that support ambient capabilities. This is useful to write
553 unit files that work with ambient capabilities where possible but
554 automatically fall back to traditional privilege dropping mechanisms
555 on systems where this is not supported.
557 * ListenNetlink= settings in socket units now support RDMA netlink
560 * A new unit file setting LockPersonality= has been added which permits
561 locking down the chosen execution domain ("personality") of a service
564 * A new special target "getty-pre.target" has been added, which is
565 ordered before all text logins, and may be used to order services
566 before textual logins acquire access to the console.
568 * systemd will now attempt to load the virtio-rng.ko kernel module very
569 early on if a VM environment supporting this is detected. This should
570 improve entropy during early boot in virtualized environments.
572 * A _netdev option is now supported in /etc/crypttab that operates in a
573 similar way as the same option in /etc/fstab: it permits configuring
574 encrypted devices that need to be ordered after the network is up.
575 Following this logic, two new special targets
576 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
577 added that are to cryptsetup.target what remote-fs.target and
578 remote-fs-pre.target are to local-fs.target.
580 * Service units gained a new UnsetEnvironment= setting which permits
581 unsetting specific environment variables for services that are
582 normally passed to it (for example in order to mask out locale
583 settings for specific services that can't deal with it).
585 * Units acquired a new boolean option IPAccounting=. When turned on, IP
586 traffic accounting (packet count as well as byte count) is done for
587 the service, and shown as part of "systemctl status" or "systemd-run
590 * Service units acquired two new options IPAddressAllow= and
591 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
592 for configuring a simple IP access control list for all sockets of
593 the unit. These options are available also on .slice and .socket
594 units, permitting flexible access list configuration for individual
595 services as well as groups of services (as defined by a slice unit),
596 including system-wide. Note that IP ACLs configured this way are
597 enforced on every single IPv4 and IPv6 socket created by any process
598 of the service unit, and apply to ingress as well as egress traffic.
600 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
601 structured log message is generated each time the unit is stopped,
602 containing information about the consumed resources of this
605 * A new setting KeyringMode= has been added to unit files, which may be
606 used to control how the kernel keyring is set up for executed
609 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
610 "systemctl kexec" and "systemctl exit" are now always asynchronous in
611 behaviour (that is: these commands return immediately after the
612 operation was enqueued instead of waiting for the operation to
613 complete). Previously, "systemctl poweroff" and "systemctl reboot"
614 were asynchronous on systems using systemd-logind (i.e. almost
615 always, and like they were on sysvinit), and the other three commands
616 were unconditionally synchronous. With this release this is cleaned
617 up, and callers will see the same asynchronous behaviour on all
618 systems for all five operations.
620 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
623 * .timer units now accept calendar specifications in other timezones
624 than UTC or the local timezone.
626 * The tmpfiles snippet var.conf has been changed to create
627 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
628 the "utmp" group already, and it appears to be generally understood
629 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
630 databases. Previously this was implemented correctly for all these
631 databases excepts btmp, which has been opened up like this now
632 too. Note that while the other databases are world-readable
633 (i.e. 0644), btmp is not and remains more restrictive.
635 * The systemd-resolve tool gained a new --reset-server-features
636 switch. When invoked like this systemd-resolved will forget
637 everything it learnt about the features supported by the configured
638 upstream DNS servers, and restarts the feature probing logic on the
639 next resolver look-up for them at the highest feature level
642 * The status dump systemd-resolved sends to the logs upon receiving
643 SIGUSR1 now also includes information about all DNS servers it is
644 configured to use, and the features levels it probed for them.
646 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
647 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
648 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
649 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
650 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
651 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
652 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
653 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
654 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
655 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
656 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
657 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
658 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
659 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
660 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
661 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
662 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
663 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
664 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
665 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
671 * Meson is now supported as build system in addition to Automake. It is
672 our plan to remove Automake in one of our next releases, so that
673 Meson becomes our exclusive build system. Hence, please start using
674 the Meson build system in your downstream packaging. There's plenty
675 of documentation around how to use Meson, the extremely brief
678 ./autogen.sh && ./configure && make && sudo make install
682 meson build && ninja -C build && sudo ninja -C build install
684 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
685 which permits configuring a timeout on the time a job is
686 running. This is particularly useful for setting timeouts on jobs for
689 * Unit files gained two new options ConditionUser= and ConditionGroup=
690 for conditionalizing units based on the identity of the user/group
691 running a systemd user instance.
693 * systemd-networkd now understands a new FlowLabel= setting in the
694 [VXLAN] section of .network files, as well as a Priority= in
695 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
696 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
697 gained support for configuration of GENEVE links, and IPv6 address
698 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
700 * .link files now understand a new Port= setting.
702 * systemd-networkd's DHCP support gained support for DHCP option 119
703 (domain search list).
705 * systemd-networkd gained support for serving IPv6 address ranges using
706 the Router Advertisment protocol. The new .network configuration
707 section [IPv6Prefix] may be used to configure the ranges to
708 serve. This is implemented based on a new, minimal, native server
709 implementation of RA.
711 * journalctl's --output= switch gained support for a new parameter
712 "short-iso-precise" for a mode where timestamps are shown as precise
715 * systemd-udevd's "net_id" builtin may now generate stable network
716 interface names from IBM PowerVM VIO devices as well as ACPI platform
719 * MulticastDNS support in systemd-resolved may now be explicitly
720 enabled/disabled using the new MulticastDNS= configuration file
723 * systemd-resolved may now optionally use libidn2 instead of the libidn
724 for processing internationalized domain names. Support for libidn2
725 should be considered experimental and should not be enabled by
728 * "machinectl pull-tar" and related call may now do verification of
729 downloaded images using SUSE-style .sha256 checksum files in addition
730 to the already existing support for validating using Ubuntu-style
733 * sd-bus gained support for a new sd_bus_message_appendv() call which
734 is va_list equivalent of sd_bus_message_append().
736 * sd-boot gained support for validating images using SHIM/MOK.
738 * The SMACK code learnt support for "onlycap".
740 * systemd-mount --umount is now much smarter in figuring out how to
741 properly unmount a device given its mount or device path.
743 * The code to call libnss_dns as a fallback from libnss_resolve when
744 the communication with systemd-resolved fails was removed. This
745 fallback was redundant and interfered with the [!UNAVAIL=return]
746 suffix. See nss-resolve(8) for the recommended configuration.
748 * systemd-logind may now be restarted without losing state. It stores
749 the file descriptors for devices it manages in the system manager
750 using the FDSTORE= mechanism. Please note that further changes in
751 other components may be required to make use of this (for example
752 Xorg has code to listen for stops of systemd-logind and terminate
753 itself when logind is stopped or restarted, in order to avoid using
754 stale file descriptors for graphical devices, which is now
755 counterproductive and must be reverted in order for restarts of
756 systemd-logind to be safe. See
757 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
759 * All kernel install plugins are called with the environment variable
760 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
761 /etc/machine-id. If the file is missing or empty, the variable is
762 empty and BOOT_DIR_ABS is the path of a temporary directory which is
763 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
764 is empty, all plugins should not put anything in BOOT_DIR_ABS.
766 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
767 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
768 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
769 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
770 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
771 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
772 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
773 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
774 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
775 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
776 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
777 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
778 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
779 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
780 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
781 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
782 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
783 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
784 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
785 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
786 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
787 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
788 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
789 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
790 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
791 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
792 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
793 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
794 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
801 * The "hybrid" control group mode has been modified to improve
802 compatibility with "legacy" cgroups-v1 setups. Specifically, the
803 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
804 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
805 cgroups-v1 hierarchy), the only externally visible change being that
806 the cgroups-v2 hierarchy is also mounted, to
807 /sys/fs/cgroup/unified. This should provide a large degree of
808 compatibility with "legacy" cgroups-v1, while taking benefit of the
809 better management capabilities of cgroups-v2.
811 * The default control group setup mode may be selected both a boot-time
812 via a set of kernel command line parameters (specifically:
813 systemd.unified_cgroup_hierarchy= and
814 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
815 default selected on the configure command line
816 (--with-default-hierarchy=). The upstream default is "hybrid"
817 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
818 this will change in a future systemd version to be "unified" (pure
819 cgroups-v2 mode). The third option for the compile time option is
820 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
821 distributions to default to "hybrid" mode for release distributions,
822 starting with v233. We recommend "unified" for development
823 distributions (specifically: distributions such as Fedora's rawhide)
824 as that's where things are headed in the long run. Use "legacy" for
825 greatest stability and compatibility only.
827 * Note one current limitation of "unified" and "hybrid" control group
828 setup modes: the kernel currently does not permit the systemd --user
829 instance (i.e. unprivileged code) to migrate processes between two
830 disconnected cgroup subtrees, even if both are managed and owned by
831 the user. This effectively means "systemd-run --user --scope" doesn't
832 work when invoked from outside of any "systemd --user" service or
833 scope. Specifically, it is not supported from session scopes. We are
834 working on fixing this in a future systemd version. (See #3388 for
835 further details about this.)
837 * DBus policy files are now installed into /usr rather than /etc. Make
838 sure your system has dbus >= 1.9.18 running before upgrading to this
839 version, or override the install path with --with-dbuspolicydir= .
841 * All python scripts shipped with systemd (specifically: the various
842 tests written in Python) now require Python 3.
844 * systemd unit tests can now run standalone (without the source or
845 build directories), and can be installed into /usr/lib/systemd/tests/
846 with 'make install-tests'.
848 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
849 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
852 * Support for the %c, %r, %R specifiers in unit files has been
853 removed. Specifiers are not supposed to be dependent on configuration
854 in the unit file itself (so that they resolve the same regardless
855 where used in the unit files), but these specifiers were influenced
856 by the Slice= option.
858 * The shell invoked by debug-shell.service now defaults to /bin/sh in
859 all cases. If distributions want to use a different shell for this
860 purpose (for example Fedora's /sbin/sushell) they need to specify
861 this explicitly at configure time using --with-debug-shell=.
863 * The confirmation spawn prompt has been reworked to offer the
866 (c)ontinue, proceed without asking anymore
867 (D)ump, show the state of the unit
868 (f)ail, don't execute the command and pretend it failed
870 (i)nfo, show a short summary of the unit
871 (j)obs, show jobs that are in progress
872 (s)kip, don't execute the command and pretend it succeeded
873 (y)es, execute the command
875 The 'n' choice for the confirmation spawn prompt has been removed,
876 because its meaning was confusing.
878 The prompt may now also be redirected to an alternative console by
879 specifying the console as parameter to systemd.confirm_spawn=.
881 * Services of Type=notify require a READY=1 notification to be sent
882 during startup. If no such message is sent, the service now fails,
883 even if the main process exited with a successful exit code.
885 * Services that fail to start up correctly now always have their
886 ExecStopPost= commands executed. Previously, they'd enter "failed"
887 state directly, without executing these commands.
889 * The option MulticastDNS= of network configuration files has acquired
890 an actual implementation. With MulticastDNS=yes a host can resolve
891 names of remote hosts and reply to mDNS A and AAAA requests.
893 * When units are about to be started an additional check is now done to
894 ensure that all dependencies of type BindsTo= (when used in
895 combination with After=) have been started.
897 * systemd-analyze gained a new verb "syscall-filter" which shows which
898 system call groups are defined for the SystemCallFilter= unit file
899 setting, and which system calls they contain.
901 * A new system call filter group "@filesystem" has been added,
902 consisting of various file system related system calls. Group
903 "@reboot" has been added, covering reboot, kexec and shutdown related
904 calls. Finally, group "@swap" has been added covering swap
905 configuration related calls.
907 * A new unit file option RestrictNamespaces= has been added that may be
908 used to restrict access to the various process namespace types the
909 Linux kernel provides. Specifically, it may be used to take away the
910 right for a service unit to create additional file system, network,
911 user, and other namespaces. This sandboxing option is particularly
912 relevant due to the high amount of recently discovered namespacing
913 related vulnerabilities in the kernel.
915 * systemd-udev's .link files gained support for a new AutoNegotiation=
916 setting for configuring Ethernet auto-negotiation.
918 * systemd-networkd's .network files gained support for a new
919 ListenPort= setting in the [DHCP] section to explicitly configure the
920 UDP client port the DHCP client shall listen on.
922 * .network files gained a new Unmanaged= boolean setting for explicitly
923 excluding one or more interfaces from management by systemd-networkd.
925 * The systemd-networkd ProxyARP= option has been renamed to
926 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
927 renamed to ReduceARPProxy=. The old names continue to be available
930 * systemd-networkd gained support for configuring IPv6 Proxy NDP
931 addresses via the new IPv6ProxyNDPAddress= .network file setting.
933 * systemd-networkd's bonding device support gained support for two new
934 configuration options ActiveSlave= and PrimarySlave=.
936 * The various options in the [Match] section of .network files gained
937 support for negative matching.
939 * New systemd-specific mount options are now understood in /etc/fstab:
941 x-systemd.mount-timeout= may be used to configure the maximum
942 permitted runtime of the mount command.
944 x-systemd.device-bound may be set to bind a mount point to its
945 backing device unit, in order to automatically remove a mount point
946 if its backing device is unplugged. This option may also be
947 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
948 on the block device, which is now automatically set for all CDROM
949 drives, so that mounted CDs are automatically unmounted when they are
950 removed from the drive.
952 x-systemd.after= and x-systemd.before= may be used to explicitly
953 order a mount after or before another unit or mount point.
955 * Enqueued start jobs for device units are now automatically garbage
956 collected if there are no jobs waiting for them anymore.
958 * systemctl list-jobs gained two new switches: with --after, for every
959 queued job the jobs it's waiting for are shown; with --before the
960 jobs which it's blocking are shown.
962 * systemd-nspawn gained support for ephemeral boots from disk images
963 (or in other words: --ephemeral and --image= may now be
964 combined). Moreover, ephemeral boots are now supported for normal
965 directories, even if the backing file system is not btrfs. Of course,
966 if the file system does not support file system snapshots or
967 reflinks, the initial copy operation will be relatively expensive, but
968 this should still be suitable for many use cases.
970 * Calendar time specifications in .timer units now support
971 specifications relative to the end of a month by using "~" instead of
972 "-" as separator between month and day. For example, "*-02~03" means
973 "the third last day in February". In addition a new syntax for
974 repeated events has been added using the "/" character. For example,
975 "9..17/2:00" means "every two hours from 9am to 5pm".
977 * systemd-socket-proxyd gained a new parameter --connections-max= for
978 configuring the maximum number of concurrent connections.
980 * sd-id128 gained a new API for generating unique IDs for the host in a
981 way that does not leak the machine ID. Specifically,
982 sd_id128_get_machine_app_specific() derives an ID based on the
983 machine ID a in well-defined, non-reversible, stable way. This is
984 useful whenever an identifier for the host is needed but where the
985 identifier shall not be useful to identify the system beyond the
986 scope of the application itself. (Internally this uses HMAC-SHA256 as
987 keyed hash function using the machine ID as input.)
989 * NotifyAccess= gained a new supported value "exec". When set
990 notifications are accepted from all processes systemd itself invoked,
991 including all control processes.
993 * .nspawn files gained support for defining overlay mounts using the
994 Overlay= and OverlayReadOnly= options. Previously this functionality
995 was only available on the systemd-nspawn command line.
997 * systemd-nspawn's --bind= and --overlay= options gained support for
998 bind/overlay mounts whose source lies within the container tree by
999 prefixing the source path with "+".
1001 * systemd-nspawn's --bind= and --overlay= options gained support for
1002 automatically allocating a temporary source directory in /var/tmp
1003 that is removed when the container dies. Specifically, if the source
1004 directory is specified as empty string this mechanism is selected. An
1005 example usage is --overlay=+/var::/var, which creates an overlay
1006 mount based on the original /var contained in the image, overlayed
1007 with a temporary directory in the host's /var/tmp. This way changes
1008 to /var are automatically flushed when the container shuts down.
1010 * systemd-nspawn --image= option does now permit raw file system block
1011 devices (in addition to images containing partition tables, as
1014 * The disk image dissection logic in systemd-nspawn gained support for
1015 automatically setting up LUKS encrypted as well as Verity protected
1016 partitions. When a container is booted from an encrypted image the
1017 passphrase is queried at start-up time. When a container with Verity
1018 data is started, the root hash is search in a ".roothash" file
1019 accompanying the disk image (alternatively, pass the root hash via
1020 the new --root-hash= command line option).
1022 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1023 be used to dissect disk images the same way as systemd-nspawn does
1024 it, following the Bootable Partition Specification. It may even be
1025 used to mount disk images with complex partition setups (including
1026 LUKS and Verity partitions) to a local host directory, in order to
1027 inspect them. This tool is not considered public API (yet), and is
1028 thus not installed into /usr/bin. Please do not rely on its
1029 existence, since it might go away or be changed in later systemd
1032 * A new generator "systemd-verity-generator" has been added, similar in
1033 style to "systemd-cryptsetup-generator", permitting automatic setup of
1034 Verity root partitions when systemd boots up. In order to make use of
1035 this your partition setup should follow the Discoverable Partitions
1036 Specification, and the GPT partition ID of the root file system
1037 partition should be identical to the upper 128bit of the Verity root
1038 hash. The GPT partition ID of the Verity partition protecting it
1039 should be the lower 128bit of the Verity root hash. If the partition
1040 image follows this model it is sufficient to specify a single
1041 "roothash=" kernel command line argument to both configure which root
1042 image and verity partition to use as well as the root hash for
1043 it. Note that systemd-nspawn's Verity support follows the same
1044 semantics, meaning that disk images with proper Verity data in place
1045 may be booted in containers with systemd-nspawn as well as on
1046 physical systems via the verity generator. Also note that the "mkosi"
1047 tool available at https://github.com/systemd/mkosi has been updated
1048 to generate Verity protected disk images following this scheme. In
1049 fact, it has been updated to generate disk images that optionally
1050 implement a complete UEFI SecureBoot trust chain, involving a signed
1051 kernel and initrd image that incorporates such a root hash as well as
1052 a Verity-enabled root partition.
1054 * The hardware database (hwdb) udev supports has been updated to carry
1055 accelerometer quirks.
1057 * All system services are now run with a fresh kernel keyring set up
1058 for them. The invocation ID is stored by default in it, thus
1059 providing a safe, non-overridable way to determine the invocation
1062 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1063 options for bind mounting arbitrary paths in a service-specific
1064 way. When these options are used, arbitrary host or service files and
1065 directories may be mounted to arbitrary locations in the service's
1068 * Documentation has been added that lists all of systemd's low-level
1069 environment variables:
1071 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
1073 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1074 whether a specific socket file descriptor matches a specified socket
1077 * systemd-firstboot has been updated to check for the
1078 systemd.firstboot= kernel command line option. It accepts a boolean
1079 and when set to false the first boot questions are skipped.
1081 * systemd-fstab-generator has been updated to check for the
1082 systemd.volatile= kernel command line option, which either takes an
1083 optional boolean parameter or the special value "state". If used the
1084 system may be booted in a "volatile" boot mode. Specifically,
1085 "systemd.volatile" is used, the root directory will be mounted as
1086 tmpfs, and only /usr is mounted from the actual root file system. If
1087 "systemd.volatile=state" is used, the root directory will be mounted
1088 as usual, but /var is mounted as tmpfs. This concept provides similar
1089 functionality as systemd-nspawn's --volatile= option, but provides it
1090 on physical boots. Use this option for implementing stateless
1091 systems, or testing systems with all state and/or configuration reset
1092 to the defaults. (Note though that many distributions are not
1093 prepared to boot up without a populated /etc or /var, though.)
1095 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1096 partitions. Previously it only supported LUKS encrypted partitions
1097 for all other uses, except for the root partition itself.
1099 * Socket units gained support for listening on AF_VSOCK sockets for
1100 communication in virtualized QEMU environments.
1102 * The "configure" script gained a new option --with-fallback-hostname=
1103 for specifying the fallback hostname to use if none is configured in
1104 /etc/hostname. For example, by specifying
1105 --with-fallback-hostname=fedora it is possible to default to a
1106 hostname of "fedora" on pristine installations.
1108 * systemd-cgls gained support for a new --unit= switch for listing only
1109 the control groups of a specific unit. Similar --user-unit= has been
1110 added for listing only the control groups of a specific user unit.
1112 * systemd-mount gained a new --umount switch for unmounting a mount or
1113 automount point (and all mount/automount points below it).
1115 * systemd will now refuse full configuration reloads (via systemctl
1116 daemon-reload and related calls) unless at least 16MiB of free space
1117 are available in /run. This is a safety precaution in order to ensure
1118 that generators can safely operate after the reload completed.
1120 * A new unit file option RootImage= has been added, which has a similar
1121 effect as RootDirectory= but mounts the service's root directory from
1122 a disk image instead of plain directory. This logic reuses the same
1123 image dissection and mount logic that systemd-nspawn already uses,
1124 and hence supports any disk images systemd-nspawn supports, including
1125 those following the Discoverable Partition Specification, as well as
1126 Verity enabled images. This option enables systemd to run system
1127 services directly off disk images acting as resource bundles,
1128 possibly even including full integrity data.
1130 * A new MountAPIVFS= unit file option has been added, taking a boolean
1131 argument. If enabled /proc, /sys and /dev (collectively called the
1132 "API VFS") will be mounted for the service. This is only relevant if
1133 RootDirectory= or RootImage= is used for the service, as these mounts
1134 are of course in place in the host mount namespace anyway.
1136 * systemd-nspawn gained support for a new --pivot-root= switch. If
1137 specified the root directory within the container image is pivoted to
1138 the specified mount point, while the original root disk is moved to a
1139 different place. This option enables booting of ostree images
1140 directly with systemd-nspawn.
1142 * The systemd build scripts will no longer complain if the NTP server
1143 addresses are not changed from the defaults. Google now supports
1144 these NTP servers officially. We still recommend downstreams to
1145 properly register an NTP pool with the NTP pool project though.
1147 * coredumpctl gained a new "--reverse" option for printing the list
1148 of coredumps in reverse order.
1150 * coredumpctl will now show additional information about truncated and
1151 inaccessible coredumps, as well as coredumps that are still being
1152 processed. It also gained a new --quiet switch for suppressing
1153 additional informational message in its output.
1155 * coredumpctl gained support for only showing coredumps newer and/or
1156 older than specific timestamps, using the new --since= and --until=
1157 options, reminiscent of journalctl's options by the same name.
1159 * The systemd-coredump logic has been improved so that it may be reused
1160 to collect backtraces in non-compiled languages, for example in
1161 scripting languages such as Python.
1163 * machinectl will now show the UID shift of local containers, if user
1164 namespacing is enabled for them.
1166 * systemd will now optionally run "environment generator" binaries at
1167 configuration load time. They may be used to add environment
1168 variables to the environment block passed to services invoked. One
1169 user environment generator is shipped by default that sets up
1170 environment variables based on files dropped into /etc/environment.d
1171 and ~/.config/environment.d/.
1173 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1176 * hostnamed has been updated to report a new chassis type of
1177 "convertible" to cover "foldable" laptops that can both act as a
1178 tablet and as a laptop, such as various Lenovo Yoga devices.
1180 Contributions from: Adrián López, Alexander Galanin, Alexander
1181 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1182 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1183 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1184 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1185 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1186 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1187 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1188 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1189 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1190 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1191 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1192 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1193 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1194 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1195 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1196 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1197 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1198 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1199 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1200 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1201 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1202 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1203 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1204 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1205 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1206 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1207 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1210 — Berlin, 2017-03-01
1214 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1215 RestrictAddressFamilies= enabled. These sandboxing options should
1216 generally be compatible with the various external udev call-out
1217 binaries we are aware of, however there may be exceptions, in
1218 particular when exotic languages for these call-outs are used. In
1219 this case, consider turning off these settings locally.
1221 * The new RemoveIPC= option can be used to remove IPC objects owned by
1222 the user or group of a service when that service exits.
1224 * The new ProtectKernelModules= option can be used to disable explicit
1225 load and unload operations of kernel modules by a service. In
1226 addition access to /usr/lib/modules is removed if this option is set.
1228 * ProtectSystem= option gained a new value "strict", which causes the
1229 whole file system tree with the exception of /dev, /proc, and /sys,
1230 to be remounted read-only for a service.
1232 * The new ProtectKernelTunables= option can be used to disable
1233 modification of configuration files in /sys and /proc by a service.
1234 Various directories and files are remounted read-only, so access is
1235 restricted even if the file permissions would allow it.
1237 * The new ProtectControlGroups= option can be used to disable write
1238 access by a service to /sys/fs/cgroup.
1240 * Various systemd services have been hardened with
1241 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1242 RestrictAddressFamilies=.
1244 * Support for dynamically creating users for the lifetime of a service
1245 has been added. If DynamicUser=yes is specified, user and group IDs
1246 will be allocated from the range 61184..65519 for the lifetime of the
1247 service. They can be resolved using the new nss-systemd.so NSS
1248 module. The module must be enabled in /etc/nsswitch.conf. Services
1249 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1250 any resources allocated by the service will be cleaned up when the
1251 service exits. They also have ProtectHome=read-only and
1252 ProtectSystem=strict enabled, so they are not able to make any
1253 permanent modifications to the system.
1255 * The nss-systemd module also always resolves root and nobody, making
1256 it possible to have no /etc/passwd or /etc/group files in minimal
1257 container or chroot environments.
1259 * Services may be started with their own user namespace using the new
1260 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1261 under which the service is running are mapped. All other users are
1264 * Support for the cgroup namespace has been added to systemd-nspawn. If
1265 supported by kernel, the container system started by systemd-nspawn
1266 will have its own view of the cgroup hierarchy. This new behaviour
1267 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1269 * The new MemorySwapMax= option can be used to limit the maximum swap
1270 usage under the unified cgroup hierarchy.
1272 * Support for the CPU controller in the unified cgroup hierarchy has
1273 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1274 options. This controller requires out-of-tree patches for the kernel
1275 and the support is provisional.
1277 * Mount and automount units may now be created transiently
1278 (i.e. dynamically at runtime via the bus API, instead of requiring
1279 unit files in the file system).
1281 * systemd-mount is a new tool which may mount file systems – much like
1282 mount(8), optionally pulling in additional dependencies through
1283 transient .mount and .automount units. For example, this tool
1284 automatically runs fsck on a backing block device before mounting,
1285 and allows the automount logic to be used dynamically from the
1286 command line for establishing mount points. This tool is particularly
1287 useful when dealing with removable media, as it will ensure fsck is
1288 run – if necessary – before the first access and that the file system
1289 is quickly unmounted after each access by utilizing the automount
1290 logic. This maximizes the chance that the file system on the
1291 removable media stays in a clean state, and if it isn't in a clean
1292 state is fixed automatically.
1294 * LazyUnmount=yes option for mount units has been added to expose the
1295 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1298 * /efi will be used as the mount point of the EFI boot partition, if
1299 the directory is present, and the mount point was not configured
1300 through other means (e.g. fstab). If /efi directory does not exist,
1301 /boot will be used as before. This makes it easier to automatically
1302 mount the EFI partition on systems where /boot is used for something
1305 * When operating on GPT disk images for containers, systemd-nspawn will
1306 now mount the ESP to /boot or /efi according to the same rules as PID
1307 1 running on a host. This allows tools like "bootctl" to operate
1308 correctly within such containers, in order to make container images
1309 bootable on physical systems.
1311 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1313 * Two new user session targets have been added to support running
1314 graphical sessions under the systemd --user instance:
1315 graphical-session.target and graphical-session-pre.target. See
1316 systemd.special(7) for a description of how those targets should be
1319 * The vconsole initialization code has been significantly reworked to
1320 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1321 support unicode keymaps. Font and keymap configuration will now be
1322 copied to all allocated virtual consoles.
1324 * FreeBSD's bhyve virtualization is now detected.
1326 * Information recorded in the journal for core dumps now includes the
1327 contents of /proc/mountinfo and the command line of the process at
1328 the top of the process hierarchy (which is usually the init process
1331 * systemd-journal-gatewayd learned the --directory= option to serve
1332 files from the specified location.
1334 * journalctl --root=… can be used to peruse the journal in the
1335 /var/log/ directories inside of a container tree. This is similar to
1336 the existing --machine= option, but does not require the container to
1339 * The hardware database has been extended to support
1340 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1343 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1344 specify the click rate for mice which include a horizontal wheel with
1345 a click rate that is different than the one for the vertical wheel.
1347 * systemd-run gained a new --wait option that makes service execution
1348 synchronous. (Specifically, the command will not return until the
1349 specified service binary exited.)
1351 * systemctl gained a new --wait option that causes the start command to
1352 wait until the units being started have terminated again.
1354 * A new journal output mode "short-full" has been added which displays
1355 timestamps with abbreviated English day names and adds a timezone
1356 suffix. Those timestamps include more information than the default
1357 "short" output mode, and can be passed directly to journalctl's
1358 --since= and --until= options.
1360 * /etc/resolv.conf will be bind-mounted into containers started by
1361 systemd-nspawn, if possible, so any changes to resolv.conf contents
1362 are automatically propagated to the container.
1364 * The number of instances for socket-activated services originating
1365 from a single IP address can be limited with
1366 MaxConnectionsPerSource=, extending the existing setting of
1369 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1372 * .netdev and .network configuration can now be extended through
1375 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1376 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1377 can be enabled and disabled using the new UDPSegmentationOffload=,
1378 TCPSegmentationOffload=, GenericSegmentationOffload=,
1379 GenericReceiveOffload=, LargeReceiveOffload= options in the
1380 [Link] section of .link files.
1382 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1383 Port VLAN ID can be configured for bridge devices using the new STP=,
1384 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1385 section of .netdev files.
1387 * The route table to which routes received over DHCP or RA should be
1388 added can be configured with the new RouteTable= option in the [DHCP]
1389 and [IPv6AcceptRA] sections of .network files.
1391 * The Address Resolution Protocol can be disabled on links managed by
1392 systemd-networkd using the ARP=no setting in the [Link] section of
1395 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1396 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1397 encode information about the result and exit codes of the current
1398 service runtime cycle.
1400 * systemd-sysctl will now configure kernel parameters in the order
1401 they occur in the configuration files. This matches what sysctl
1402 has been traditionally doing.
1404 * kernel-install "plugins" that are executed to perform various
1405 tasks after a new kernel is added and before an old one is removed
1406 can now return a special value to terminate the procedure and
1407 prevent any later plugins from running.
1409 * Journald's SplitMode=login setting has been deprecated. It has been
1410 removed from documentation, and its use is discouraged. In a future
1411 release it will be completely removed, and made equivalent to current
1412 default of SplitMode=uid.
1414 * Storage=both option setting in /etc/systemd/coredump.conf has been
1415 removed. With fast LZ4 compression storing the core dump twice is not
1418 * The --share-system systemd-nspawn option has been replaced with an
1419 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1420 this functionality is discouraged. In addition the variables
1421 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1422 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1423 individual namespaces.
1425 * "machinectl list" now shows the IP address of running containers in
1426 the output, as well as OS release information.
1428 * "loginctl list" now shows the TTY of each session in the output.
1430 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1431 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1432 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1433 tracking objects in a "recursive" mode, where a single client can be
1434 counted multiple times, if it takes multiple references.
1436 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1437 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1438 process using sd-bus automatically exit if the bus connection is
1441 * Bus clients of the service manager may now "pin" loaded units into
1442 memory, by taking an explicit reference on them. This is useful to
1443 ensure the client can retrieve runtime data about the service even
1444 after the service completed execution. Taking such a reference is
1445 available only for privileged clients and should be helpful to watch
1446 running services in a race-free manner, and in particular collect
1447 information about exit statuses and results.
1449 * The nss-resolve module has been changed to strictly return UNAVAIL
1450 when communication via D-Bus with resolved failed, and NOTFOUND when
1451 a lookup completed but was negative. This means it is now possible to
1452 neatly configure fallbacks using nsswitch.conf result checking
1453 expressions. Taking benefit of this, the new recommended
1454 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1456 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1458 * A new setting CtrlAltDelBurstAction= has been added to
1459 /etc/systemd/system.conf which may be used to configure the precise
1460 behaviour if the user on the console presses Ctrl-Alt-Del more often
1461 than 7 times in 2s. Previously this would unconditionally result in
1462 an expedited, immediate reboot. With this new setting the precise
1463 operation may be configured in more detail, and also turned off
1466 * In .netdev files two new settings RemoteChecksumTx= and
1467 RemoteChecksumRx= are now understood that permit configuring the
1468 remote checksumming logic for VXLAN networks.
1470 * The service manager learnt a new "invocation ID" concept for invoked
1471 services. Each runtime cycle of a service will get a new invocation
1472 ID (a 128bit random UUID) assigned that identifies the current
1473 run of the service uniquely and globally. A new invocation ID
1474 is generated each time a service starts up. The journal will store
1475 the invocation ID of a service along with any logged messages, thus
1476 making the invocation ID useful for matching the online runtime of a
1477 service with the offline log data it generated in a safe way without
1478 relying on synchronized timestamps. In many ways this new service
1479 invocation ID concept is similar to the kernel's boot ID concept that
1480 uniquely and globally identifies the runtime of each boot. The
1481 invocation ID of a service is passed to the service itself via an
1482 environment variable ($INVOCATION_ID). A new bus call
1483 GetUnitByInvocationID() has been added that is similar to GetUnit()
1484 but instead of retrieving the bus path for a unit by its name
1485 retrieves it by its invocation ID. The returned path is valid only as
1486 long as the passed invocation ID is current.
1488 * systemd-resolved gained a new "DNSStubListener" setting in
1489 resolved.conf. It either takes a boolean value or the special values
1490 "udp" and "tcp", and configures whether to enable the stub DNS
1491 listener on 127.0.0.53:53.
1493 * IP addresses configured via networkd may now carry additional
1494 configuration settings supported by the kernel. New options include:
1495 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1496 PrefixRoute=, AutoJoin=.
1498 * The PAM configuration fragment file for "user@.service" shipped with
1499 systemd (i.e. the --user instance of systemd) has been stripped to
1500 the minimum necessary to make the system boot. Previously, it
1501 contained Fedora-specific stanzas that did not apply to other
1502 distributions. It is expected that downstream distributions add
1503 additional configuration lines, matching their needs to this file,
1504 using it only as rough template of what systemd itself needs. Note
1505 that this reduced fragment does not even include an invocation of
1506 pam_limits which most distributions probably want to add, even though
1507 systemd itself does not need it. (There's also the new build time
1508 option --with-pamconfdir=no to disable installation of the PAM
1511 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1512 capability is now also dropped from its set (in addition to
1513 CAP_SYS_MKNOD as before).
1515 * In service unit files it is now possible to connect a specific named
1516 file descriptor with stdin/stdout/stdout of an executed service. The
1517 name may be specified in matching .socket units using the
1518 FileDescriptorName= setting.
1520 * A number of journal settings may now be configured on the kernel
1521 command line. Specifically, the following options are now understood:
1522 systemd.journald.max_level_console=,
1523 systemd.journald.max_level_store=,
1524 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1525 systemd.journald.max_level_wall=.
1527 * "systemctl is-enabled --full" will now show by which symlinks a unit
1528 file is enabled in the unit dependency tree.
1530 * Support for VeraCrypt encrypted partitions has been added to the
1531 "cryptsetup" logic and /etc/crypttab.
1533 * systemd-detect-virt gained support for a new --private-users switch
1534 that checks whether the invoking processes are running inside a user
1535 namespace. Similar, a new special value "private-users" for the
1536 existing ConditionVirtualization= setting has been added, permitting
1537 skipping of specific units in user namespace environments.
1539 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1540 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1541 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1542 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1543 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1544 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1545 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1546 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1547 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1548 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1549 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1550 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1551 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1552 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1553 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1554 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1555 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1556 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1557 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1558 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1559 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1560 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1561 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1562 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1563 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1564 Jędrzejewski-Szmek, Zeal Jagannatha
1566 — Santa Fe, 2016-11-03
1570 * In service units the various ExecXYZ= settings have been extended
1571 with an additional special character as first argument of the
1572 assigned value: if the character '+' is used the specified command
1573 line it will be run with full privileges, regardless of User=,
1574 Group=, CapabilityBoundingSet= and similar options. The effect is
1575 similar to the existing PermissionsStartOnly= option, but allows
1576 configuration of this concept for each executed command line
1579 * Services may now alter the service watchdog timeout at runtime by
1580 sending a WATCHDOG_USEC= message via sd_notify().
1582 * MemoryLimit= and related unit settings now optionally take percentage
1583 specifications. The percentage is taken relative to the amount of
1584 physical memory in the system (or in case of containers, the assigned
1585 amount of memory). This allows scaling service resources neatly with
1586 the amount of RAM available on the system. Similarly, systemd-logind's
1587 RuntimeDirectorySize= option now also optionally takes percentage
1590 * In similar fashion TasksMax= takes percentage values now, too. The
1591 value is taken relative to the configured maximum number of processes
1592 on the system. The per-service task maximum has been changed to 15%
1593 using this functionality. (Effectively this is an increase of 512 →
1594 4915 for service units, given the kernel's default pid_max setting.)
1596 * Calendar time specifications in .timer units now understand a ".."
1597 syntax for time ranges. Example: "4..7:10" may now be used for
1598 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1601 * The InaccessableDirectories=, ReadOnlyDirectories= and
1602 ReadWriteDirectories= unit file settings have been renamed to
1603 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1604 applied to all kinds of file nodes, and not just directories, with
1605 the exception of symlinks. Specifically these settings may now be
1606 used on block and character device nodes, UNIX sockets and FIFOS as
1607 well as regular files. The old names of these settings remain
1608 available for compatibility.
1610 * systemd will now log about all service processes it kills forcibly
1611 (using SIGKILL) because they remained after the clean shutdown phase
1612 of the service completed. This should help identifying services that
1613 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1614 systemd-logind's configuration a similar log message is generated for
1615 processes killed at the end of each session due to this setting.
1617 * systemd will now set the $JOURNAL_STREAM environment variable for all
1618 services whose stdout/stderr are connected to the Journal (which
1619 effectively means by default: all services). The variable contains
1620 the device and inode number of the file descriptor used for
1621 stdout/stderr. This may be used by invoked programs to detect whether
1622 their stdout/stderr is connected to the Journal, in which case they
1623 can switch over to direct Journal communication, thus being able to
1624 pass extended, structured metadata along with their log messages. As
1625 one example, this is now used by glib's logging primitives.
1627 * When using systemd's default tmp.mount unit for /tmp, the mount point
1628 will now be established with the "nosuid" and "nodev" options. This
1629 avoids privilege escalation attacks that put traps and exploits into
1630 /tmp. However, this might cause problems if you e. g. put container
1631 images or overlays into /tmp; if you need this, override tmp.mount's
1632 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1635 * systemd now supports the "memory" cgroup controller also on
1638 * The systemd-cgtop tool now optionally takes a control group path as
1639 command line argument. If specified, the control group list shown is
1640 limited to subgroups of that group.
1642 * The SystemCallFilter= unit file setting gained support for
1643 pre-defined, named system call filter sets. For example
1644 SystemCallFilter=@clock is now an effective way to make all clock
1645 changing-related system calls unavailable to a service. A number of
1646 similar pre-defined groups are defined. Writing system call filters
1647 for system services is simplified substantially with this new
1648 concept. Accordingly, all of systemd's own, long-running services now
1649 enable system call filtering based on this, by default.
1651 * A new service setting MemoryDenyWriteExecute= has been added, taking
1652 a boolean value. If turned on, a service may no longer create memory
1653 mappings that are writable and executable at the same time. This
1654 enhances security for services where this is enabled as it becomes
1655 harder to dynamically write and then execute memory in exploited
1656 service processes. This option has been enabled for all of systemd's
1657 own long-running services.
1659 * A new RestrictRealtime= service setting has been added, taking a
1660 boolean argument. If set the service's processes may no longer
1661 acquire realtime scheduling. This improves security as realtime
1662 scheduling may otherwise be used to easily freeze the system.
1664 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1665 value. This may be used for requesting that the system manager inside
1666 of the container reports start-up completion to nspawn which then
1667 propagates this notification further to the service manager
1668 supervising nspawn itself. A related option NotifyReady= in .nspawn
1669 files has been added too. This functionality allows ordering of the
1670 start-up of multiple containers using the usual systemd ordering
1673 * machinectl gained a new command "stop" that is an alias for
1676 * systemd-resolved gained support for contacting DNS servers on
1677 link-local IPv6 addresses.
1679 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1680 its caches. A method call for requesting the same operation has been
1681 added to the bus API too, and is made available via "systemd-resolve
1684 * systemd-resolve gained a new --status switch. If passed a brief
1685 summary of the used DNS configuration with per-interface information
1688 * resolved.conf gained a new Cache= boolean option, defaulting to
1689 on. If turned off local DNS caching is disabled. This comes with a
1690 performance penalty in particular when DNSSEC is enabled. Note that
1691 resolved disables its internal caching implicitly anyway, when the
1692 configured DNS server is on a host-local IP address such as ::1 or
1693 127.0.0.1, thus automatically avoiding double local caching.
1695 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1696 for DNS requests. This improves compatibility with local programs
1697 that do not use the libc NSS or systemd-resolved's bus APIs for name
1698 resolution. This minimal DNS service is only available to local
1699 programs and does not implement the full DNS protocol, but enough to
1700 cover local DNS clients. A new, static resolv.conf file, listing just
1701 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1702 now recommended to make /etc/resolv.conf a symlink to this file in
1703 order to route all DNS lookups to systemd-resolved, regardless if
1704 done via NSS, the bus API or raw DNS packets. Note that this local
1705 DNS service is not as fully featured as the libc NSS or
1706 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1707 used to deliver link-local address information (as this implies
1708 sending a local interface index along), LLMNR/mDNS support via this
1709 interface is severely restricted. It is thus strongly recommended for
1710 all applications to use the libc NSS API or native systemd-resolved
1713 * systemd-networkd's bridge support learned a new setting
1714 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1715 in .network files has been added for configuring VLAN bridging in
1716 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1718 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1719 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1720 now be acquired without relying on DHCPv6. Two new options
1721 UseDomains= and UseDNS= have been added to configure this behaviour.
1723 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1724 renamed IPv6AcceptRA=, without altering its behaviour. The old
1725 setting name remains available for compatibility reasons.
1727 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1728 Key=, InputKey= and OutputKey=.
1730 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1731 interface configuration.
1733 * "systemctl edit" may now be used to create new unit files by
1734 specifying the --force switch.
1736 * sd-event gained a new function sd_event_get_iteration() for
1737 requesting the current iteration counter of the event loop. It starts
1738 at zero and is increased by one with each event loop iteration.
1740 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1741 file. It can be used in lieu of %systemd_requires in packages which
1742 don't use any systemd functionality and are intended to be installed
1743 in minimal containers without systemd present. This macro provides
1744 ordering dependencies to ensure that if the package is installed in
1745 the same rpm transaction as systemd, systemd will be installed before
1746 the scriptlets for the package are executed, allowing unit presets
1749 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1750 been added to simplify packaging of generators.
1752 * The os-release file gained VERSION_CODENAME field for the
1753 distribution nickname (e.g. VERSION_CODENAME=woody).
1755 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1756 can be set to disable parsing of metadata and the creation
1757 of persistent symlinks for that device.
1759 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1760 to make them available to logged-in users has been reverted.
1762 * Much of the common code of the various systemd components is now
1763 built into an internal shared library libsystemd-shared-231.so
1764 (incorporating the systemd version number in the name, to be updated
1765 with future releases) that the components link to. This should
1766 decrease systemd footprint both in memory during runtime and on
1767 disk. Note that the shared library is not for public use, and is
1768 neither API not ABI stable, but is likely to change with every new
1769 released update. Packagers need to make sure that binaries
1770 linking to libsystemd-shared.so are updated in step with the
1773 * Configuration for "mkosi" is now part of the systemd
1774 repository. mkosi is a tool to easily build legacy-free OS images,
1775 and is available on github: https://github.com/systemd/mkosi. If
1776 "mkosi" is invoked in the build tree a new raw OS image is generated
1777 incorporating the systemd sources currently being worked on and a
1778 clean, fresh distribution installation. The generated OS image may be
1779 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1780 UEFI PC. This functionality is particularly useful to easily test
1781 local changes made to systemd in a pristine, defined environment. See
1782 HACKING for details.
1784 * configure learned the --with-support-url= option to specify the
1785 distribution's bugtracker.
1787 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1788 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1789 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1790 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1791 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1792 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1793 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1794 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1795 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1796 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1797 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1798 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1799 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1800 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1801 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1802 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1803 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1804 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1805 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1807 — Berlin, 2016-07-25
1811 * DNSSEC is now turned on by default in systemd-resolved (in
1812 "allow-downgrade" mode), but may be turned off during compile time by
1813 passing "--with-default-dnssec=no" to "configure" (and of course,
1814 during runtime with DNSSEC= in resolved.conf). We recommend
1815 downstreams to leave this on at least during development cycles and
1816 report any issues with the DNSSEC logic upstream. We are very
1817 interested in collecting feedback about the DNSSEC validator and its
1818 limitations in the wild. Note however, that DNSSEC support is
1819 probably nothing downstreams should turn on in stable distros just
1820 yet, as it might create incompatibilities with a few DNS servers and
1821 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1822 automatically whenever we detect such incompatible setups, but there
1823 might be systems we do not cover yet. Hence: please help us testing
1824 the DNSSEC code, leave this on where you can, report back, but then
1825 again don't consider turning this on in your stable, LTS or
1826 production release just yet. (Note that you have to enable
1827 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1828 and its DNSSEC mode for host name resolution from local
1831 * systemd-resolve conveniently resolves DANE records with the --tlsa
1832 option and OPENPGPKEY records with the --openpgp option. It also
1833 supports dumping raw DNS record data via the new --raw= switch.
1835 * systemd-logind will now by default terminate user processes that are
1836 part of the user session scope unit (session-XX.scope) when the user
1837 logs out. This behavior is controlled by the KillUserProcesses=
1838 setting in logind.conf, and the previous default of "no" is now
1839 changed to "yes". This means that user sessions will be properly
1840 cleaned up after, but additional steps are necessary to allow
1841 intentionally long-running processes to survive logout.
1843 While the user is logged in at least once, user@.service is running,
1844 and any service that should survive the end of any individual login
1845 session can be started at a user service or scope using systemd-run.
1846 systemd-run(1) man page has been extended with an example which shows
1847 how to run screen in a scope unit underneath user@.service. The same
1848 command works for tmux.
1850 After the user logs out of all sessions, user@.service will be
1851 terminated too, by default, unless the user has "lingering" enabled.
1852 To effectively allow users to run long-term tasks even if they are
1853 logged out, lingering must be enabled for them. See loginctl(1) for
1854 details. The default polkit policy was modified to allow users to
1855 set lingering for themselves without authentication.
1857 Previous defaults can be restored at compile time by the
1858 --without-kill-user-processes option to "configure".
1860 * systemd-logind gained new configuration settings SessionsMax= and
1861 InhibitorsMax=, both with a default of 8192. It will not register new
1862 user sessions or inhibitors above this limit.
1864 * systemd-logind will now reload configuration on SIGHUP.
1866 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1867 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1868 enable. Also, support for the "io" cgroup controller in the unified
1869 hierarchy has been added, so that the "memory", "pids" and "io" are
1870 now the controllers that are supported on the unified hierarchy.
1872 WARNING: it is not possible to use previous systemd versions with
1873 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1874 is necessary to also update systemd in the initramfs if using the
1875 unified hierarchy. An updated SELinux policy is also required.
1877 * LLDP support has been extended, and both passive (receive-only) and
1878 active (sender) modes are supported. Passive mode ("routers-only") is
1879 enabled by default in systemd-networkd. Active LLDP mode is enabled
1880 by default for containers on the internal network. The "networkctl
1881 lldp" command may be used to list information gathered. "networkctl
1882 status" will also show basic LLDP information on connected peers now.
1884 * The IAID and DUID unique identifier sent in DHCP requests may now be
1885 configured for the system and each .network file managed by
1886 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1888 * systemd-networkd gained support for configuring proxy ARP support for
1889 each interface, via the ProxyArp= setting in .network files. It also
1890 gained support for configuring the multicast querier feature of
1891 bridge devices, via the new MulticastQuerier= setting in .netdev
1892 files. Similarly, snooping on the IGMP traffic can be controlled
1893 via the new setting MulticastSnooping=.
1895 A new setting PreferredLifetime= has been added for addresses
1896 configured in .network file to configure the lifetime intended for an
1899 The systemd-networkd DHCP server gained the option EmitRouter=, which
1900 defaults to yes, to configure whether the DHCP Option 3 (Router)
1903 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1904 systemd-socket-activate and installed into /usr/bin. It is now fully
1907 * systemd-journald now uses separate threads to flush changes to disk
1908 when closing journal files, thus reducing impact of slow disk I/O on
1909 logging performance.
1911 * The sd-journal API gained two new calls
1912 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1913 can be used to open journal files using file descriptors instead of
1914 file or directory paths. sd_journal_open_container() has been
1915 deprecated, sd_journal_open_directory_fd() should be used instead
1916 with the flag SD_JOURNAL_OS_ROOT.
1918 * journalctl learned a new output mode "-o short-unix" that outputs log
1919 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1920 UTC). It also gained support for a new --no-hostname setting to
1921 suppress the hostname column in the family of "short" output modes.
1923 * systemd-ask-password now optionally skips printing of the password to
1924 stdout with --no-output which can be useful in scripts.
1926 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1927 (devices tagged with ID_MAKER_TOOL) are now tagged with
1928 "uaccess" and are available to logged in users.
1930 * The DeviceAllow= unit setting now supports specifiers (with "%").
1932 * "systemctl show" gained a new --value switch, which allows print a
1933 only the contents of a specific unit property, without also printing
1934 the property's name. Similar support was added to "show*" verbs
1935 of loginctl and machinectl that output "key=value" lists.
1937 * A new unit type "generated" was added for files dynamically generated
1938 by generator tools. Similarly, a new unit type "transient" is used
1939 for unit files created using the runtime API. "systemctl enable" will
1940 refuse to operate on such files.
1942 * A new command "systemctl revert" has been added that may be used to
1943 revert to the vendor version of a unit file, in case local changes
1944 have been made by adding drop-ins or overriding the unit file.
1946 * "machinectl clean" gained a new verb to automatically remove all or
1947 just hidden container images.
1949 * systemd-tmpfiles gained support for a new line type "e" for emptying
1950 directories, if they exist, without creating them if they don't.
1952 * systemd-nspawn gained support for automatically patching the UID/GIDs
1953 of the owners and the ACLs of all files and directories in a
1954 container tree to match the UID/GID user namespacing range selected
1955 for the container invocation. This mode is enabled via the new
1956 --private-users-chown switch. It also gained support for
1957 automatically choosing a free, previously unused UID/GID range when
1958 starting a container, via the new --private-users=pick setting (which
1959 implies --private-users-chown). Together, these options for the first
1960 time make user namespacing for nspawn containers fully automatic and
1961 thus deployable. The systemd-nspawn@.service template unit file has
1962 been changed to use this functionality by default.
1964 * systemd-nspawn gained a new --network-zone= switch, that allows
1965 creating ad-hoc virtual Ethernet links between multiple containers,
1966 that only exist as long as at least one container referencing them is
1967 running. This allows easy connecting of multiple containers with a
1968 common link that implements an Ethernet broadcast domain. Each of
1969 these network "zones" may be named relatively freely by the user, and
1970 may be referenced by any number of containers, but each container may
1971 only reference one of these "zones". On the lower level, this is
1972 implemented by an automatically managed bridge network interface for
1973 each zone, that is created when the first container referencing its
1974 zone is created and removed when the last one referencing its zone
1977 * The default start timeout may now be configured on the kernel command
1978 line via systemd.default_timeout_start_sec=. It was already
1979 configurable via the DefaultTimeoutStartSec= option in
1980 /etc/systemd/system.conf.
1982 * Socket units gained a new TriggerLimitIntervalSec= and
1983 TriggerLimitBurst= setting to configure a limit on the activation
1984 rate of the socket unit.
1986 * The LimitNICE= setting now optionally takes normal UNIX nice values
1987 in addition to the raw integer limit value. If the specified
1988 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1989 value is understood as UNIX nice value. If not prefixed like this it
1990 is understood as raw RLIMIT_NICE limit.
1992 * Note that the effect of the PrivateDevices= unit file setting changed
1993 slightly with this release: the per-device /dev file system will be
1994 mounted read-only from this version on, and will have "noexec"
1995 set. This (minor) change of behavior might cause some (exceptional)
1996 legacy software to break, when PrivateDevices=yes is set for its
1997 service. Please leave PrivateDevices= off if you run into problems
2000 * systemd-bootchart has been split out to a separate repository:
2001 https://github.com/systemd/systemd-bootchart
2003 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2004 merged into the kernel in its current form.
2006 * The compatibility libraries libsystemd-daemon.so,
2007 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2008 which have been deprecated since systemd-209 have been removed along
2009 with the corresponding pkg-config files. All symbols provided by
2010 those libraries are provided by libsystemd.so.
2012 * The Capabilities= unit file setting has been removed (it is ignored
2013 for backwards compatibility). AmbientCapabilities= and
2014 CapabilityBoundingSet= should be used instead.
2016 * A new special target has been added, initrd-root-device.target,
2017 which creates a synchronization point for dependencies of the root
2018 device in early userspace. Initramfs builders must ensure that this
2019 target is now included in early userspace.
2021 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2022 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2023 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2024 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2025 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2026 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2027 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2028 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2029 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2030 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2031 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2032 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2033 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2034 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2035 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2036 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2037 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2038 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2039 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2040 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2041 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2042 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2043 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2044 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2045 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2048 — Fairfax, 2016-05-21
2052 * The systemd-resolved DNS resolver service has gained a substantial
2053 set of new features, most prominently it may now act as a DNSSEC
2054 validating stub resolver. DNSSEC mode is currently turned off by
2055 default, but is expected to be turned on by default in one of the
2056 next releases. For now, we invite everybody to test the DNSSEC logic
2057 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2058 service also gained a full set of D-Bus interfaces, including calls
2059 to configure DNS and DNSSEC settings per link (for use by external
2060 network management software). systemd-resolved and systemd-networkd
2061 now distinguish between "search" and "routing" domains. The former
2062 are used to qualify single-label names, the latter are used purely
2063 for routing lookups within certain domains to specific links.
2064 resolved now also synthesizes RRs for all entries from /etc/hosts.
2066 * The systemd-resolve tool (which is a client utility for
2067 systemd-resolved) has been improved considerably and is now fully
2068 supported and documented. Hence it has moved from /usr/lib/systemd to
2071 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2074 * The coredump collection logic has been reworked: when a coredump is
2075 collected it is now written to disk, compressed and processed
2076 (including stacktrace extraction) from a new instantiated service
2077 systemd-coredump@.service, instead of directly from the
2078 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2079 processing large coredumps can take up a substantial amount of
2080 resources and time, and this previously happened entirely outside of
2081 systemd's service supervision. With the new logic the core_pattern
2082 hook only does minimal metadata collection before passing off control
2083 to the new instantiated service, which is configured with a time
2084 limit, a nice level and other settings to minimize negative impact on
2085 the rest of the system. Also note that the new logic will honour the
2086 RLIMIT_CORE setting of the crashed process, which now allows users
2087 and processes to turn off coredumping for their processes by setting
2090 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2091 and all forked processes by default. Previously, PID 1 would leave
2092 the setting at "0" for all processes, as set by the kernel. Note that
2093 the resource limit traditionally has no effect on the generated
2094 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2095 logic is used. Since the limit is now honoured (see above) its
2096 default has been changed so that the coredumping logic is enabled by
2097 default for all processes, while allowing specific opt-out.
2099 * When the stacktrace is extracted from processes of system users, this
2100 is now done as "systemd-coredump" user, in order to sandbox this
2101 potentially security sensitive parsing operation. (Note that when
2102 processing coredumps of normal users this is done under the user ID
2103 of process that crashed, as before.) Packagers should take notice
2104 that it is now necessary to create the "systemd-coredump" system user
2105 and group at package installation time.
2107 * The systemd-activate socket activation testing tool gained support
2108 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2109 and --seqpacket switches. It also has been extended to support both
2110 new-style and inetd-style file descriptor passing. Use the new
2111 --inetd switch to request inetd-style file descriptor passing.
2113 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2114 variable, which takes a boolean value. If set to false, ANSI color
2115 output is disabled in the tools even when run on a terminal that
2118 * The VXLAN support in networkd now supports two new settings
2119 DestinationPort= and PortRange=.
2121 * A new systemd.machine_id= kernel command line switch has been added,
2122 that may be used to set the machine ID in /etc/machine-id if it is
2123 not initialized yet. This command line option has no effect if the
2124 file is already initialized.
2126 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2127 specified command line as PID 2 rather than PID 1 in the
2128 container. In this mode PID 1 is a minimal stub init process that
2129 implements the special POSIX and Linux semantics of PID 1 regarding
2130 signal and child process management. Note that this stub init process
2131 is implemented in nspawn itself and requires no support from the
2132 container image. This new logic is useful to support running
2133 arbitrary commands in the container, as normal processes are
2134 generally not prepared to run as PID 1.
2136 * systemd-nspawn gained a new --chdir= switch for setting the current
2137 working directory for the process started in the container.
2139 * "journalctl /dev/sda" will now output all kernel log messages for
2140 specified device from the current boot, in addition to all devices
2141 that are parents of it. This should make log output about devices
2142 pretty useful, as long as kernel drivers attach enough metadata to
2143 the log messages. (The usual SATA drivers do.)
2145 * The sd-journal API gained two new calls
2146 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2147 that report whether log data from /run or /var has been found.
2149 * journalctl gained a new switch "--fields" that prints all journal
2150 record field names currently in use in the journal. This is backed
2151 by two new sd-journal API calls sd_journal_enumerate_fields() and
2152 sd_journal_restart_fields().
2154 * Most configurable timeouts in systemd now expect an argument of
2155 "infinity" to turn them off, instead of "0" as before. The semantics
2156 from now on is that a timeout of "0" means "now", and "infinity"
2157 means "never". To maintain backwards compatibility, "0" continues to
2158 turn off previously existing timeout settings.
2160 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2161 try-reload-or-restart" to clarify what it actually does: the "try"
2162 logic applies to both reloading and restarting, not just restarting.
2163 The old name continues to be accepted for compatibility.
2165 * On boot-up, when PID 1 detects that the system clock is behind the
2166 release date of the systemd version in use, the clock is now set
2167 to the latter. Previously, this was already done in timesyncd, in order
2168 to avoid running with clocks set to the various clock epochs such as
2169 1902, 1938 or 1970. With this change the logic is now done in PID 1
2170 in addition to timesyncd during early boot-up, so that it is enforced
2171 before the first process is spawned by systemd. Note that the logic
2172 in timesyncd remains, as it is more comprehensive and ensures
2173 clock monotonicity by maintaining a persistent timestamp file in
2174 /var. Since /var is generally not available in earliest boot or the
2175 initrd, this part of the logic remains in timesyncd, and is not done
2178 * Support for tweaking details in net_cls.class_id through the
2179 NetClass= configuration directive has been removed, as the kernel
2180 people have decided to deprecate that controller in cgroup v2.
2181 Userspace tools such as nftables are moving over to setting rules
2182 that are specific to the full cgroup path of a task, which obsoletes
2183 these controllers anyway. The NetClass= directive is kept around for
2184 legacy compatibility reasons. For a more in-depth description of the
2185 kernel change, please refer to the respective upstream commit:
2187 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2189 * A new service setting RuntimeMaxSec= has been added that may be used
2190 to specify a maximum runtime for a service. If the timeout is hit, the
2191 service is terminated and put into a failure state.
2193 * A new service setting AmbientCapabilities= has been added. It allows
2194 configuration of additional Linux process capabilities that are
2195 passed to the activated processes. This is only available on very
2198 * The process resource limit settings in service units may now be used
2199 to configure hard and soft limits individually.
2201 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2202 expose support for gcc's __attribute__((cleanup())) C extension.
2203 Specifically, for many object destructor functions alternative
2204 versions have been added that have names suffixed with "p" and take a
2205 pointer to a pointer to the object to destroy, instead of just a
2206 pointer to the object itself. This is useful because these destructor
2207 functions may be used directly as parameters to the cleanup
2208 construct. Internally, systemd has been a heavy user of this GCC
2209 extension for a long time, and with this change similar support is
2210 now available to consumers of the library outside of systemd. Note
2211 that by using this extension in your sources compatibility with old
2212 and strictly ANSI compatible C compilers is lost. However, all gcc or
2213 LLVM versions of recent years support this extension.
2215 * Timer units gained support for a new setting RandomizedDelaySec= that
2216 allows configuring some additional randomized delay to the configured
2217 time. This is useful to spread out timer events to avoid load peaks in
2218 clusters or larger setups.
2220 * Calendar time specifications now support sub-second accuracy.
2222 * Socket units now support listening on SCTP and UDP-lite protocol
2225 * The sd-event API now comes with a full set of man pages.
2227 * Older versions of systemd contained experimental support for
2228 compressing journal files and coredumps with the LZ4 compressor that
2229 was not compatible with the lz4 binary (due to API limitations of the
2230 lz4 library). This support has been removed; only support for files
2231 compatible with the lz4 binary remains. This LZ4 logic is now
2232 officially supported and no longer considered experimental.
2234 * The dkr image import logic has been removed again from importd. dkr's
2235 micro-services focus doesn't fit into the machine image focus of
2236 importd, and quickly got out of date with the upstream dkr API.
2238 * Creation of the /run/lock/lockdev/ directory was dropped from
2239 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2240 been available for many years. If you still need this, you need to
2241 create your own tmpfiles.d config file with:
2243 d /run/lock/lockdev 0775 root lock -
2245 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2246 and RebootArgument= have been moved from the [Service] section of
2247 unit files to [Unit], and they are now supported on all unit types,
2248 not just service units. Of course, systemd will continue to
2249 understand these settings also at the old location, in order to
2250 maintain compatibility.
2252 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2253 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2254 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2255 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2256 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2257 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2258 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2259 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2260 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2261 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2262 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2263 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2264 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2265 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2266 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2267 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2268 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2269 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2270 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2272 — Berlin, 2016-02-11
2276 * A number of properties previously only settable in unit
2277 files are now also available as properties to set when
2278 creating transient units programmatically via the bus, as it
2279 is exposed with systemd-run's --property=
2280 setting. Specifically, these are: SyslogIdentifier=,
2281 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2282 EnvironmentFile=, ReadWriteDirectories=,
2283 ReadOnlyDirectories=, InaccessibleDirectories=,
2284 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2286 * When creating transient services via the bus API it is now
2287 possible to pass in a set of file descriptors to use as
2288 STDIN/STDOUT/STDERR for the invoked process.
2290 * Slice units may now be created transiently via the bus APIs,
2291 similar to the way service and scope units may already be
2292 created transiently.
2294 * Wherever systemd expects a calendar timestamp specification
2295 (like in journalctl's --since= and --until= switches) UTC
2296 timestamps are now supported. Timestamps suffixed with "UTC"
2297 are now considered to be in Universal Time Coordinated
2298 instead of the local timezone. Also, timestamps may now
2299 optionally be specified with sub-second accuracy. Both of
2300 these additions also apply to recurring calendar event
2301 specification, such as OnCalendar= in timer units.
2303 * journalctl gained a new "--sync" switch that asks the
2304 journal daemon to write all so far unwritten log messages to
2305 disk and sync the files, before returning.
2307 * systemd-tmpfiles learned two new line types "q" and "Q" that
2308 operate like "v", but also set up a basic btrfs quota
2309 hierarchy when used on a btrfs file system with quota
2312 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2313 instead of a subvolume (even on a btrfs file system) if the
2314 root directory is a plain directory, and not a
2315 subvolume. This should simplify things with certain chroot()
2316 environments which are not aware of the concept of btrfs
2319 * systemd-detect-virt gained a new --chroot switch to detect
2320 whether execution takes place in a chroot() environment.
2322 * CPUAffinity= now takes CPU index ranges in addition to
2325 * The various memory-related resource limit settings (such as
2326 LimitAS=) now understand the usual K, M, G, ... suffixes to
2327 the base of 1024 (IEC). Similar, the time-related resource
2328 limit settings understand the usual min, h, day, ...
2331 * There's a new system.conf setting DefaultTasksMax= to
2332 control the default TasksMax= setting for services and
2333 scopes running on the system. (TasksMax= is the primary
2334 setting that exposes the "pids" cgroup controller on systemd
2335 and was introduced in the previous systemd release.) The
2336 setting now defaults to 512, which means services that are
2337 not explicitly configured otherwise will only be able to
2338 create 512 processes or threads at maximum, from this
2339 version on. Note that this means that thread- or
2340 process-heavy services might need to be reconfigured to set
2341 TasksMax= to a higher value. It is sufficient to set
2342 TasksMax= in these specific unit files to a higher value, or
2343 even "infinity". Similar, there's now a logind.conf setting
2344 UserTasksMax= that defaults to 4096 and limits the total
2345 number of processes or tasks each user may own
2346 concurrently. nspawn containers also have the TasksMax=
2347 value set by default now, to 8192. Note that all of this
2348 only has an effect if the "pids" cgroup controller is
2349 enabled in the kernel. The general benefit of these changes
2350 should be a more robust and safer system, that provides a
2351 certain amount of per-service fork() bomb protection.
2353 * systemd-nspawn gained the new --network-veth-extra= switch
2354 to define additional and arbitrarily-named virtual Ethernet
2355 links between the host and the container.
2357 * A new service execution setting PassEnvironment= has been
2358 added that allows importing select environment variables
2359 from PID1's environment block into the environment block of
2362 * Timer units gained support for a new RemainAfterElapse=
2363 setting which takes a boolean argument. It defaults to on,
2364 exposing behaviour unchanged to previous releases. If set to
2365 off, timer units are unloaded after they elapsed if they
2366 cannot elapse again. This is particularly useful for
2367 transient timer units, which shall not stay around longer
2368 than until they first elapse.
2370 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2371 default now (the kernel default is 16). This is beneficial
2372 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2373 allows substantially larger numbers of queued
2374 datagrams. This should increase the capability of systemd to
2375 parallelize boot-up, as logging and sd_notify() are unlikely
2376 to stall execution anymore. If you need to change the value
2377 from the new defaults, use the usual sysctl.d/ snippets.
2379 * The compression framing format used by the journal or
2380 coredump processing has changed to be in line with what the
2381 official LZ4 tools generate. LZ4 compression support in
2382 systemd was considered unsupported previously, as the format
2383 was not compatible with the normal tools. With this release
2384 this has changed now, and it is hence safe for downstream
2385 distributions to turn it on. While not compressing as well
2386 as the XZ, LZ4 is substantially faster, which makes
2387 it a good default choice for the compression logic in the
2388 journal and in coredump handling.
2390 * Any reference to /etc/mtab has been dropped from
2391 systemd. The file has been obsolete since a while, but
2392 systemd refused to work on systems where it was incorrectly
2393 set up (it should be a symlink or non-existent). Please make
2394 sure to update to util-linux 2.27.1 or newer in conjunction
2395 with this systemd release, which also drops any reference to
2396 /etc/mtab. If you maintain a distribution make sure that no
2397 software you package still references it, as this is a
2398 likely source of bugs. There's also a glibc bug pending,
2399 asking for removal of any reference to this obsolete file:
2401 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2403 Note that only util-linux versions built with
2404 --enable-libmount-force-mountinfo are supported.
2406 * Support for the ".snapshot" unit type has been removed. This
2407 feature turned out to be little useful and little used, and
2408 has now been removed from the core and from systemctl.
2410 * The dependency types RequiresOverridable= and
2411 RequisiteOverridable= have been removed from systemd. They
2412 have been used only very sparingly to our knowledge and
2413 other options that provide a similar effect (such as
2414 systemctl --mode=ignore-dependencies) are much more useful
2415 and commonly used. Moreover, they were only half-way
2416 implemented as the option to control behaviour regarding
2417 these dependencies was never added to systemctl. By removing
2418 these dependency types the execution engine becomes a bit
2419 simpler. Unit files that use these dependencies should be
2420 changed to use the non-Overridable dependency types
2421 instead. In fact, when parsing unit files with these
2422 options, that's what systemd will automatically convert them
2423 too, but it will also warn, asking users to fix the unit
2424 files accordingly. Removal of these dependency types should
2425 only affect a negligible number of unit files in the wild.
2427 * Behaviour of networkd's IPForward= option changed
2428 (again). It will no longer maintain a per-interface setting,
2429 but propagate one way from interfaces where this is enabled
2430 to the global kernel setting. The global setting will be
2431 enabled when requested by a network that is set up, but
2432 never be disabled again. This change was made to make sure
2433 IPv4 and IPv6 behaviour regarding packet forwarding is
2434 similar (as the Linux IPv6 stack does not support
2435 per-interface control of this setting) and to minimize
2438 * In unit files the behaviour of %u, %U, %h, %s has
2439 changed. These specifiers will now unconditionally resolve
2440 to the various user database fields of the user that the
2441 systemd instance is running as, instead of the user
2442 configured in the specific unit via User=. Note that this
2443 effectively doesn't change much, as resolving of these
2444 specifiers was already turned off in the --system instance
2445 of systemd, as we cannot do NSS lookups from PID 1. In the
2446 --user instance of systemd these specifiers where correctly
2447 resolved, but hardly made any sense, since the user instance
2448 lacks privileges to do user switches anyway, and User= is
2449 hence useless. Moreover, even in the --user instance of
2450 systemd behaviour was awkward as it would only take settings
2451 from User= assignment placed before the specifier into
2452 account. In order to unify and simplify the logic around
2453 this the specifiers will now always resolve to the
2454 credentials of the user invoking the manager (which in case
2455 of PID 1 is the root user).
2457 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2458 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2459 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2460 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2461 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2462 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2463 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2464 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2465 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2466 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2467 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2468 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2469 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2470 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2473 — Berlin, 2015-11-18
2477 * systemd now depends on util-linux v2.27. More specifically,
2478 the newly added mount monitor feature in libmount now
2479 replaces systemd's former own implementation.
2481 * libmount mandates /etc/mtab not to be regular file, and
2482 systemd now enforces this condition at early boot.
2483 /etc/mtab has been deprecated and warned about for a very
2484 long time, so systems running systemd should already have
2485 stopped having this file around as anything else than a
2486 symlink to /proc/self/mounts.
2488 * Support for the "pids" cgroup controller has been added. It
2489 allows accounting the number of tasks in a cgroup and
2490 enforcing limits on it. This adds two new setting
2491 TasksAccounting= and TasksMax= to each unit, as well as a
2492 global option DefaultTasksAccounting=.
2494 * Support for the "net_cls" cgroup controller has been added.
2495 It allows assigning a net class ID to each task in the
2496 cgroup, which can then be used in firewall rules and traffic
2497 shaping configurations. Note that the kernel netfilter net
2498 class code does not currently work reliably for ingress
2499 packets on unestablished sockets.
2501 This adds a new config directive called NetClass= to CGroup
2502 enabled units. Allowed values are positive numbers for fixed
2503 assignments and "auto" for picking a free value
2506 * 'systemctl is-system-running' now returns 'offline' if the
2507 system is not booted with systemd. This command can now be
2508 used as a substitute for 'systemd-notify --booted'.
2510 * Watchdog timeouts have been increased to 3 minutes for all
2511 in-tree service files. Apparently, disk IO issues are more
2512 frequent than we hoped, and user reported >1 minute waiting
2515 * 'machine-id-commit' functionality has been merged into
2516 'machine-id-setup --commit'. The separate binary has been
2519 * The WorkingDirectory= directive in unit files may now be set
2520 to the special value '~'. In this case, the working
2521 directory is set to the home directory of the user
2522 configured in User=.
2524 * "machinectl shell" will now open the shell in the home
2525 directory of the selected user by default.
2527 * The CrashChVT= configuration file setting is renamed to
2528 CrashChangeVT=, following our usual logic of not
2529 abbreviating unnecessarily. The old directive is still
2530 supported for compat reasons. Also, this directive now takes
2531 an integer value between 1 and 63, or a boolean value. The
2532 formerly supported '-1' value for disabling stays around for
2535 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2536 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2537 RootDirectory= properties can now be set for transient
2540 * The systemd-analyze tool gained a new "set-log-target" verb
2541 to change the logging target the system manager logs to
2542 dynamically during runtime. This is similar to how
2543 "systemd-analyze set-log-level" already changes the log
2546 * In nspawn /sys is now mounted as tmpfs, with only a selected
2547 set of subdirectories mounted in from the real sysfs. This
2548 enhances security slightly, and is useful for ensuring user
2549 namespaces work correctly.
2551 * Support for USB FunctionFS activation has been added. This
2552 allows implementation of USB gadget services that are
2553 activated as soon as they are requested, so that they don't
2554 have to run continuously, similar to classic socket
2557 * The "systemctl exit" command now optionally takes an
2558 additional parameter that sets the exit code to return from
2559 the systemd manager when exiting. This is only relevant when
2560 running the systemd user instance, or when running the
2561 system instance in a container.
2563 * sd-bus gained the new API calls sd_bus_path_encode_many()
2564 and sd_bus_path_decode_many() that allow easy encoding and
2565 decoding of multiple identifier strings inside a D-Bus
2566 object path. Another new call sd_bus_default_flush_close()
2567 has been added to flush and close per-thread default
2570 * systemd-cgtop gained support for a -M/--machine= switch to
2571 show the control groups within a certain container only.
2573 * "systemctl kill" gained support for an optional --fail
2574 switch. If specified the requested operation will fail of no
2575 processes have been killed, because the unit had no
2576 processes attached, or similar.
2578 * A new systemd.crash_reboot=1 kernel command line option has
2579 been added that triggers a reboot after crashing. This can
2580 also be set through CrashReboot= in systemd.conf.
2582 * The RuntimeDirectory= setting now understands unit
2583 specifiers like %i or %f.
2585 * A new (still internal) library API sd-ipv4acd has been added,
2586 that implements address conflict detection for IPv4. It's
2587 based on code from sd-ipv4ll, and will be useful for
2588 detecting DHCP address conflicts.
2590 * File descriptors passed during socket activation may now be
2591 named. A new API sd_listen_fds_with_names() is added to
2592 access the names. The default names may be overridden,
2593 either in the .socket file using the FileDescriptorName=
2594 parameter, or by passing FDNAME= when storing the file
2595 descriptors using sd_notify().
2597 * systemd-networkd gained support for:
2599 - Setting the IPv6 Router Advertisement settings via
2600 IPv6AcceptRouterAdvertisements= in .network files.
2602 - Configuring the HelloTimeSec=, MaxAgeSec= and
2603 ForwardDelaySec= bridge parameters in .netdev files.
2605 - Configuring PreferredSource= for static routes in
2608 * The "ask-password" framework used to query for LUKS harddisk
2609 passwords or SSL passwords during boot gained support for
2610 caching passwords in the kernel keyring, if it is
2611 available. This makes sure that the user only has to type in
2612 a passphrase once if there are multiple objects to unlock
2613 with the same one. Previously, such password caching was
2614 available only when Plymouth was used; this moves the
2615 caching logic into the systemd codebase itself. The
2616 "systemd-ask-password" utility gained a new --keyname=
2617 switch to control which kernel keyring key to use for
2618 caching a password in. This functionality is also useful for
2619 enabling display managers such as gdm to automatically
2620 unlock the user's GNOME keyring if its passphrase, the
2621 user's password and the harddisk password are the same, if
2622 gdm-autologin is used.
2624 * When downloading tar or raw images using "machinectl
2625 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2626 file is now also downloaded, if it is available and stored
2627 next to the image file.
2629 * Units of type ".socket" gained a new boolean setting
2630 Writable= which is only useful in conjunction with
2631 ListenSpecial=. If true, enables opening the specified
2632 special file in O_RDWR mode rather than O_RDONLY mode.
2634 * systemd-rfkill has been reworked to become a singleton
2635 service that is activated through /dev/rfkill on each rfkill
2636 state change and saves the settings to disk. This way,
2637 systemd-rfkill is now compatible with devices that exist
2638 only intermittendly, and even restores state if the previous
2639 system shutdown was abrupt rather than clean.
2641 * The journal daemon gained support for vacuuming old journal
2642 files controlled by the number of files that shall remain,
2643 in addition to the already existing control by size and by
2644 date. This is useful as journal interleaving performance
2645 degrades with too many separate journal files, and allows
2646 putting an effective limit on them. The new setting defaults
2647 to 100, but this may be changed by setting SystemMaxFiles=
2648 and RuntimeMaxFiles= in journald.conf. Also, the
2649 "journalctl" tool gained the new --vacuum-files= switch to
2650 manually vacuum journal files to leave only the specified
2651 number of files in place.
2653 * udev will now create /dev/disk/by-path links for ATA devices
2654 on kernels where that is supported.
2656 * Galician, Serbian, Turkish and Korean translations were added.
2658 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2659 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2660 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2661 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2662 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2663 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2664 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2665 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2666 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2667 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2668 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2669 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2670 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2671 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2672 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2673 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2674 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2675 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2677 — Berlin, 2015-10-07
2681 * The DHCP implementation of systemd-networkd gained a set of
2684 - The DHCP server now supports emitting DNS and NTP
2685 information. It may be enabled and configured via
2686 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2687 and NTP information is enabled, but no servers are
2688 configured, the corresponding uplink information (if there
2689 is any) is propagated.
2691 - Server and client now support transmission and reception
2692 of timezone information. It can be configured via the
2693 newly introduced network options UseTimezone=,
2694 EmitTimezone=, and Timezone=. Transmission of timezone
2695 information is enabled between host and containers by
2696 default now: the container will change its local timezone
2697 to what the host has set.
2699 - Lease timeouts can now be configured via
2700 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2702 - The DHCP server improved on the stability of
2703 leases. Clients are more likely to get the same lease
2704 information back, even if the server loses state.
2706 - The DHCP server supports two new configuration options to
2707 control the lease address pool metrics, PoolOffset= and
2710 * The encapsulation limit of tunnels in systemd-networkd may
2711 now be configured via 'EncapsulationLimit='. It allows
2712 modifying the maximum additional levels of encapsulation
2713 that are permitted to be prepended to a packet.
2715 * systemd now supports the concept of user buses replacing
2716 session buses, if used with dbus-1.10 (and enabled via dbus
2717 --enable-user-session). It previously only supported this on
2718 kdbus-enabled systems, and this release expands this to
2719 'dbus-daemon' systems.
2721 * systemd-networkd now supports predictable interface names
2724 * systemd now optionally supports the new Linux kernel
2725 "unified" control group hierarchy. If enabled via the kernel
2726 command-line option 'systemd.unified_cgroup_hierarchy=1',
2727 systemd will try to mount the unified cgroup hierarchy
2728 directly on /sys/fs/cgroup. If not enabled, or not
2729 available, systemd will fall back to the legacy cgroup
2730 hierarchy setup, as before. Host system and containers can
2731 mix and match legacy and unified hierarchies as they
2732 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2733 environment variable to individually select the hierarchy to
2734 use for executed containers. By default, nspawn will use the
2735 unified hierarchy for the containers if the host uses the
2736 unified hierarchy, and the legacy hierarchy otherwise.
2737 Please note that at this point the unified hierarchy is an
2738 experimental kernel feature and is likely to change in one
2739 of the next kernel releases. Therefore, it should not be
2740 enabled by default in downstream distributions yet. The
2741 minimum required kernel version for the unified hierarchy to
2742 work is 4.2. Note that when the unified hierarchy is used
2743 for the first time delegated access to controllers is
2744 safe. Because of this systemd-nspawn containers will get
2745 access to controllers now, as will systemd user
2746 sessions. This means containers and user sessions may now
2747 manage their own resources, partitioning up what the system
2750 * A new special scope unit "init.scope" has been introduced
2751 that encapsulates PID 1 of the system. It may be used to
2752 determine resource usage and enforce resource limits on PID
2753 1 itself. PID 1 hence moved out of the root of the control
2756 * The cgtop tool gained support for filtering out kernel
2757 threads when counting tasks in a control group. Also, the
2758 count of processes is now recursively summed up by
2759 default. Two options -k and --recursive= have been added to
2760 revert to old behaviour. The tool has also been updated to
2761 work correctly in containers now.
2763 * systemd-nspawn's --bind= and --bind-ro= options have been
2764 extended to allow creation of non-recursive bind mounts.
2766 * libsystemd gained two new calls sd_pid_get_cgroup() and
2767 sd_peer_get_cgroup() which return the control group path of
2768 a process or peer of a connected AF_UNIX socket. This
2769 function call is particularly useful when implementing
2770 delegated subtrees support in the control group hierarchy.
2772 * The "sd-event" event loop API of libsystemd now supports
2773 correct dequeuing of real-time signals, without losing
2776 * When systemd requests a PolicyKit decision when managing
2777 units it will now add additional fields to the request,
2778 including unit name and desired operation. This enables more
2779 powerful PolicyKit policies, that make decisions depending
2780 on these parameters.
2782 * nspawn learnt support for .nspawn settings files, that may
2783 accompany the image files or directories of containers, and
2784 may contain additional settings for the container. This is
2785 an alternative to configuring container parameters via the
2786 nspawn command line.
2788 Contributions from: Cristian Rodríguez, Daniel Mack, David
2789 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2790 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2791 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2792 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2793 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2794 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2795 Andersen, Tom Gundersen, Torstein Husebø
2797 — Berlin, 2015-09-08
2801 * machinectl gained a new verb 'shell' which opens a fresh
2802 shell on the target container or the host. It is similar to
2803 the existing 'login' command of machinectl, but spawns the
2804 shell directly without prompting for username or
2805 password. The pseudo machine '.host' now refers to the local
2806 host and is used by default. Hence, 'machinectl shell' can
2807 be used as replacement for 'su -' which spawns a session as
2808 a fresh systemd unit in a way that is fully isolated from
2809 the originating session.
2811 * systemd-networkd learned to cope with private-zone DHCP
2812 options and allows other programs to query the values.
2814 * SELinux access control when enabling/disabling units is no
2815 longer enforced with this release. The previous
2816 implementation was incorrect, and a new corrected
2817 implementation is not yet available. As unit file operations
2818 are still protected via PolicyKit and D-Bus policy this is
2819 not a security problem. Yet, distributions which care about
2820 optimal SELinux support should probably not stabilize on
2823 * sd-bus gained support for matches of type "arg0has=", that
2824 test for membership of strings in string arrays sent in bus
2827 * systemd-resolved now dumps the contents of its DNS and LLMNR
2828 caches to the logs on reception of the SIGUSR1 signal. This
2829 is useful to debug DNS behaviour.
2831 * The coredumpctl tool gained a new --directory= option to
2832 operate on journal files in a specific directory.
2834 * "systemctl reboot" and related commands gained a new
2835 "--message=" option which may be used to set a free-text
2836 wall message when shutting down or rebooting the
2837 system. This message is also logged, which is useful for
2838 figuring out the reason for a reboot or shutdown a
2841 * The "systemd-resolve-host" tool's -i switch now takes
2842 network interface numbers as alternative to interface names.
2844 * A new unit file setting for services has been introduced:
2845 UtmpMode= allows configuration of how precisely systemd
2846 handles utmp and wtmp entries for the service if this is
2847 enabled. This allows writing services that appear similar to
2848 user sessions in the output of the "w", "who", "last" and
2851 * systemd-resolved will now locally synthesize DNS resource
2852 records for the "localhost" and "gateway" domains as well as
2853 the local hostname. This should ensure that clients querying
2854 RRs via resolved will get similar results as those going via
2855 NSS, if nss-myhostname is enabled.
2857 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2858 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2859 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2860 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2861 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2862 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2863 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2864 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2865 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2866 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2867 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2868 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2870 — Berlin, 2015-08-27
2874 * The systemd-efi-boot-generator functionality was merged into
2875 systemd-gpt-auto-generator.
2877 * systemd-networkd now supports Group Policy for vxlan
2878 devices. It can be enabled via the new boolean configuration
2879 option called 'GroupPolicyExtension='.
2881 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2882 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2883 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2885 — Berlin, 2015-07-31
2889 * The python-systemd code has been removed from the systemd repository.
2890 A new repository has been created which accommodates the code from
2891 now on, and we kindly ask distributions to create a separate package
2892 for this: https://github.com/systemd/python-systemd
2894 * The systemd daemon will now reload its main configuration
2895 (/etc/systemd/system.conf) on daemon-reload.
2897 * sd-dhcp now exposes vendor specific extensions via
2898 sd_dhcp_lease_get_vendor_specific().
2900 * systemd-networkd gained a number of new configuration options.
2902 - A new boolean configuration option for TAP devices called
2903 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2904 device, thus allowing to send and receive GSO packets.
2906 - A new tunnel configuration option called 'CopyDSCP='.
2907 If enabled, the DSCP field of ip6 tunnels is copied into the
2908 decapsulated packet.
2910 - A set of boolean bridge configuration options were added.
2911 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2912 and 'UnicastFlood=' are now parsed by networkd and applied to the
2913 respective bridge link device via the respective IFLA_BRPORT_*
2916 - A new string configuration option to override the hostname sent
2917 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2918 is true, networkd will use the configured hostname instead of the
2919 system hostname when sending DHCP requests.
2921 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2922 networkd will configure the IPv6 flow-label of the tunnel device
2923 according to RFC2460.
2925 - The 'macvtap' virtual network devices are now supported, similar to
2926 the already supported 'macvlan' devices.
2928 * systemd-resolved now implements RFC5452 to improve resilience against
2929 cache poisoning. Additionally, source port randomization is enabled
2930 by default to further protect against DNS spoofing attacks.
2932 * nss-mymachines now supports translating UIDs and GIDs of running
2933 containers with user-namespaces enabled. If a container 'foo'
2934 translates a host uid 'UID' to the container uid 'TUID', then
2935 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2936 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2937 mapped as 'vg-foo-TGID'.
2939 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2940 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2941 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2942 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2943 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2944 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2945 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2946 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2947 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2948 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2950 — Berlin, 2015-07-29
2954 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2955 There are no known issues with current sysfs, and udev does not need
2956 or should be used to work around such bugs.
2958 * udev does no longer enable USB HID power management. Several reports
2959 indicate, that some devices cannot handle that setting.
2961 * The udev accelerometer helper was removed. The functionality
2962 is now fully included in iio-sensor-proxy. But this means,
2963 older iio-sensor-proxy versions will no longer provide
2964 accelerometer/orientation data with this systemd version.
2965 Please upgrade iio-sensor-proxy to version 1.0.
2967 * networkd gained a new configuration option IPv6PrivacyExtensions=
2968 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2969 for Stateless Address") on selected networks.
2971 * For the sake of fewer build-time dependencies and less code in the
2972 main repository, the python bindings are about to be removed in the
2973 next release. A new repository has been created which accommodates
2974 the code from now on, and we kindly ask distributions to create a
2975 separate package for this. The removal will take place in v223.
2977 https://github.com/systemd/python-systemd
2979 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2980 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2981 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2982 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2983 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2984 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2985 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2986 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2987 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2988 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2990 — Berlin, 2015-07-07
2994 * The sd-bus.h and sd-event.h APIs have now been declared
2995 stable and have been added to the official interface of
2996 libsystemd.so. sd-bus implements an alternative D-Bus client
2997 library, that is relatively easy to use, very efficient and
2998 supports both classic D-Bus as well as kdbus as transport
2999 backend. sd-event is a generic event loop abstraction that
3000 is built around Linux epoll, but adds features such as event
3001 prioritization or efficient timer handling. Both APIs are good
3002 choices for C programs looking for a bus and/or event loop
3003 implementation that is minimal and does not have to be
3004 portable to other kernels.
3006 * kdbus support is no longer compile-time optional. It is now
3007 always built-in. However, it can still be disabled at
3008 runtime using the kdbus=0 kernel command line setting, and
3009 that setting may be changed to default to off, by specifying
3010 --disable-kdbus at build-time. Note though that the kernel
3011 command line setting has no effect if the kdbus.ko kernel
3012 module is not installed, in which case kdbus is (obviously)
3013 also disabled. We encourage all downstream distributions to
3014 begin testing kdbus by adding it to the kernel images in the
3015 development distributions, and leaving kdbus support in
3018 * The minimal required util-linux version has been bumped to
3021 * Support for chkconfig (--enable-chkconfig) was removed in
3022 favor of calling an abstraction tool
3023 /lib/systemd/systemd-sysv-install. This needs to be
3024 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3025 in README for details.
3027 * If there's a systemd unit and a SysV init script for the
3028 same service name, and the user executes "systemctl enable"
3029 for it (or a related call), then this will now enable both
3030 (or execute the related operation on both), not just the
3033 * The libudev API documentation has been converted from gtkdoc
3036 * gudev has been removed from the systemd tree, it is now an
3039 * The systemd-cgtop tool learnt a new --raw switch to generate
3040 "raw" (machine parsable) output.
3042 * networkd's IPForwarding= .network file setting learnt the
3043 new setting "kernel", which ensures that networkd does not
3044 change the IP forwarding sysctl from the default kernel
3047 * The systemd-logind bus API now exposes a new boolean
3048 property "Docked" that reports whether logind considers the
3049 system "docked", i.e. connected to a docking station or not.
3051 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3052 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3053 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3054 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3055 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3056 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3057 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3058 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3059 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3060 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3061 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3062 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3063 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3064 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3065 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3066 Fink, Zbigniew Jędrzejewski-Szmek
3068 — Berlin, 2015-06-19
3072 * The gudev library has been extracted into a separate repository
3073 available at: https://git.gnome.org/browse/libgudev/
3074 It is now managed as part of the Gnome project. Distributions
3075 are recommended to pass --disable-gudev to systemd and use
3076 gudev from the Gnome project instead. gudev is still included
3077 in systemd, for now. It will be removed soon, though. Please
3078 also see the announcement-thread on systemd-devel:
3079 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3081 * systemd now exposes a CPUUsageNSec= property for each
3082 service unit on the bus, that contains the overall consumed
3083 CPU time of a service (the sum of what each process of the
3084 service consumed). This value is only available if
3085 CPUAccounting= is turned on for a service, and is then shown
3086 in the "systemctl status" output.
3088 * Support for configuring alternative mappings of the old SysV
3089 runlevels to systemd targets has been removed. They are now
3090 hardcoded in a way that runlevels 2, 3, 4 all map to
3091 multi-user.target and 5 to graphical.target (which
3092 previously was already the default behaviour).
3094 * The auto-mounter logic gained support for mount point
3095 expiry, using a new TimeoutIdleSec= setting in .automount
3096 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3098 * The EFI System Partition (ESP) as mounted to /boot by
3099 systemd-efi-boot-generator will now be unmounted
3100 automatically after 2 minutes of not being used. This should
3101 minimize the risk of ESP corruptions.
3103 * New /etc/fstab options x-systemd.requires= and
3104 x-systemd.requires-mounts-for= are now supported to express
3105 additional dependencies for mounts. This is useful for
3106 journalling file systems that support external journal
3107 devices or overlay file systems that require underlying file
3108 systems to be mounted.
3110 * systemd does not support direct live-upgrades (via systemctl
3111 daemon-reexec) from versions older than v44 anymore. As no
3112 distribution we are aware of shipped such old versions in a
3113 stable release this should not be problematic.
3115 * When systemd forks off a new per-connection service instance
3116 it will now set the $REMOTE_ADDR environment variable to the
3117 remote IP address, and $REMOTE_PORT environment variable to
3118 the remote IP port. This behaviour is similar to the
3119 corresponding environment variables defined by CGI.
3121 * systemd-networkd gained support for uplink failure
3122 detection. The BindCarrier= option allows binding interface
3123 configuration dynamically to the link sense of other
3124 interfaces. This is useful to achieve behaviour like in
3127 * systemd-networkd gained support for configuring the DHCP
3128 client identifier to use when requesting leases.
3130 * systemd-networkd now has a per-network UseNTP= option to
3131 configure whether NTP server information acquired via DHCP
3132 is passed on to services like systemd-timesyncd.
3134 * systemd-networkd gained support for vti6 tunnels.
3136 * Note that systemd-networkd manages the sysctl variable
3137 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3138 it is configured for since v219. The variable controls IP
3139 forwarding, and is a per-interface alternative to the global
3140 /proc/sys/net/ipv[46]/ip_forward. This setting is
3141 configurable in the IPForward= option, which defaults to
3142 "no". This means if networkd is used for an interface it is
3143 no longer sufficient to set the global sysctl option to turn
3144 on IP forwarding! Instead, the .network file option
3145 IPForward= needs to be turned on! Note that the
3146 implementation of this behaviour was broken in v219 and has
3149 * Many bonding and vxlan options are now configurable in
3152 * systemd-nspawn gained a new --property= setting to set unit
3153 properties for the container scope. This is useful for
3154 setting resource parameters (e.g. "CPUShares=500") on
3155 containers started from the command line.
3157 * systemd-nspawn gained a new --private-users= switch to make
3158 use of user namespacing available on recent Linux kernels.
3160 * systemd-nspawn may now be called as part of a shell pipeline
3161 in which case the pipes used for stdin and stdout are passed
3162 directly to the process invoked in the container, without
3163 indirection via a pseudo tty.
3165 * systemd-nspawn gained a new switch to control the UNIX
3166 signal to use when killing the init process of the container
3169 * systemd-nspawn gained a new --overlay= switch for mounting
3170 overlay file systems into the container using the new kernel
3173 * When a container image is imported via systemd-importd and
3174 the host file system is not btrfs, a loopback block device
3175 file is created in /var/lib/machines.raw with a btrfs file
3176 system inside. It is then mounted to /var/lib/machines to
3177 enable btrfs features for container management. The loopback
3178 file and btrfs file system is grown as needed when container
3179 images are imported via systemd-importd.
3181 * systemd-machined/systemd-importd gained support for btrfs
3182 quota, to enforce container disk space limits on disk. This
3183 is exposed in "machinectl set-limit".
3185 * systemd-importd now can import containers from local .tar,
3186 .raw and .qcow2 images, and export them to .tar and .raw. It
3187 can also import dkr v2 images now from the network (on top
3190 * systemd-importd gained support for verifying downloaded
3191 images with gpg2 (previously only gpg1 was supported).
3193 * systemd-machined, systemd-logind, systemd: most bus calls
3194 are now accessible to unprivileged processes via
3195 PolicyKit. Also, systemd-logind will now allow users to kill
3196 their own sessions without further privileges or
3199 * systemd-shutdownd has been removed. This service was
3200 previously responsible for implementing scheduled shutdowns
3201 as exposed in /usr/bin/shutdown's time parameter. This
3202 functionality has now been moved into systemd-logind and is
3203 accessible via a bus interface.
3205 * "systemctl reboot" gained a new switch --firmware-setup that
3206 can be used to reboot into the EFI firmware setup, if that
3207 is available. systemd-logind now exposes an API on the bus
3208 to trigger such reboots, in case graphical desktop UIs want
3209 to cover this functionality.
3211 * "systemctl enable", "systemctl disable" and "systemctl mask"
3212 now support a new "--now" switch. If specified the units
3213 that are enabled will also be started, and the ones
3214 disabled/masked also stopped.
3216 * The Gummiboot EFI boot loader tool has been merged into
3217 systemd, and renamed to "systemd-boot". The bootctl tool has been
3218 updated to support systemd-boot.
3220 * An EFI kernel stub has been added that may be used to create
3221 kernel EFI binaries that contain not only the actual kernel,
3222 but also an initrd, boot splash, command line and OS release
3223 information. This combined binary can then be signed as a
3224 single image, so that the firmware can verify it all in one
3225 step. systemd-boot has special support for EFI binaries created
3226 like this and can extract OS release information from them
3227 and show them in the boot menu. This functionality is useful
3228 to implement cryptographically verified boot schemes.
3230 * Optional support has been added to systemd-fsck to pass
3231 fsck's progress report to an AF_UNIX socket in the file
3234 * udev will no longer create device symlinks for all block
3235 devices by default. A blacklist for excluding special block
3236 devices from this logic has been turned into a whitelist
3237 that requires picking block devices explicitly that require
3240 * A new (currently still internal) API sd-device.h has been
3241 added to libsystemd. This modernized API is supposed to
3242 replace libudev eventually. In fact, already much of libudev
3243 is now just a wrapper around sd-device.h.
3245 * A new hwdb database for storing metadata about pointing
3246 stick devices has been added.
3248 * systemd-tmpfiles gained support for setting file attributes
3249 similar to the "chattr" tool with new 'h' and 'H' lines.
3251 * systemd-journald will no longer unconditionally set the
3252 btrfs NOCOW flag on new journal files. This is instead done
3253 with tmpfiles snippet using the new 'h' line type. This
3254 allows easy disabling of this logic, by masking the
3255 journal-nocow.conf tmpfiles file.
3257 * systemd-journald will now translate audit message types to
3258 human readable identifiers when writing them to the
3259 journal. This should improve readability of audit messages.
3261 * The LUKS logic gained support for the offset= and skip=
3262 options in /etc/crypttab, as previously implemented by
3265 * /usr/lib/os-release gained a new optional field VARIANT= for
3266 distributions that support multiple variants (such as a
3267 desktop edition, a server edition, ...)
3269 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3270 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3271 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3272 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3273 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3274 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3275 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3276 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3277 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3278 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3279 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3280 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3281 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3282 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3283 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3284 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3285 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3286 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3287 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3288 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3289 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3290 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3291 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3292 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3293 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3294 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3295 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3297 — Berlin, 2015-05-22
3301 * Introduce a new API "sd-hwdb.h" for querying the hardware
3302 metadata database. With this minimal interface one can query
3303 and enumerate the udev hwdb, decoupled from the old libudev
3304 library. libudev's interface for this is now only a wrapper
3305 around sd-hwdb. A new tool systemd-hwdb has been added to
3306 interface with and update the database.
3308 * When any of systemd's tools copies files (for example due to
3309 tmpfiles' C lines) a btrfs reflink will attempted first,
3310 before bytewise copying is done.
3312 * systemd-nspawn gained a new --ephemeral switch. When
3313 specified a btrfs snapshot is taken of the container's root
3314 directory, and immediately removed when the container
3315 terminates again. Thus, a container can be started whose
3316 changes never alter the container's root directory, and are
3317 lost on container termination. This switch can also be used
3318 for starting a container off the root file system of the
3319 host without affecting the host OS. This switch is only
3320 available on btrfs file systems.
3322 * systemd-nspawn gained a new --template= switch. It takes the
3323 path to a container tree to use as template for the tree
3324 specified via --directory=, should that directory be
3325 missing. This allows instantiating containers dynamically,
3326 on first run. This switch is only available on btrfs file
3329 * When a .mount unit refers to a mount point on which multiple
3330 mounts are stacked, and the .mount unit is stopped all of
3331 the stacked mount points will now be unmounted until no
3332 mount point remains.
3334 * systemd now has an explicit notion of supported and
3335 unsupported unit types. Jobs enqueued for unsupported unit
3336 types will now fail with an "unsupported" error code. More
3337 specifically .swap, .automount and .device units are not
3338 supported in containers, .busname units are not supported on
3339 non-kdbus systems. .swap and .automount are also not
3340 supported if their respective kernel compile time options
3343 * machinectl gained support for two new "copy-from" and
3344 "copy-to" commands for copying files from a running
3345 container to the host or vice versa.
3347 * machinectl gained support for a new "bind" command to bind
3348 mount host directories into local containers. This is
3349 currently only supported for nspawn containers.
3351 * networkd gained support for configuring bridge forwarding
3352 database entries (fdb) from .network files.
3354 * A new tiny daemon "systemd-importd" has been added that can
3355 download container images in tar, raw, qcow2 or dkr formats,
3356 and make them available locally in /var/lib/machines, so
3357 that they can run as nspawn containers. The daemon can GPG
3358 verify the downloads (not supported for dkr, since it has no
3359 provisions for verifying downloads). It will transparently
3360 decompress bz2, xz, gzip compressed downloads if necessary,
3361 and restore sparse files on disk. The daemon uses privilege
3362 separation to ensure the actual download logic runs with
3363 fewer privileges than the daemon itself. machinectl has
3364 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3365 make the functionality of importd available to the
3366 user. With this in place the Fedora and Ubuntu "Cloud"
3367 images can be downloaded and booted as containers unmodified
3368 (the Fedora images lack the appropriate GPG signature files
3369 currently, so they cannot be verified, but this will change
3370 soon, hopefully). Note that downloading images is currently
3371 only fully supported on btrfs.
3373 * machinectl is now able to list container images found in
3374 /var/lib/machines, along with some metadata about sizes of
3375 disk and similar. If the directory is located on btrfs and
3376 quota is enabled, this includes quota display. A new command
3377 "image-status" has been added that shows additional
3378 information about images.
3380 * machinectl is now able to clone container images
3381 efficiently, if the underlying file system (btrfs) supports
3382 it, with the new "machinectl clone" command. It also
3383 gained commands for renaming and removing images, as well as
3384 marking them read-only or read-write (supported also on
3385 legacy file systems).
3387 * networkd gained support for collecting LLDP network
3388 announcements, from hardware that supports this. This is
3389 shown in networkctl output.
3391 * systemd-run gained support for a new -t (--pty) switch for
3392 invoking a binary on a pty whose input and output is
3393 connected to the invoking terminal. This allows executing
3394 processes as system services while interactively
3395 communicating with them via the terminal. Most interestingly
3396 this is supported across container boundaries. Invoking
3397 "systemd-run -t /bin/bash" is an alternative to running a
3398 full login session, the difference being that the former
3399 will not register a session, nor go through the PAM session
3402 * tmpfiles gained support for a new "v" line type for creating
3403 btrfs subvolumes. If the underlying file system is a legacy
3404 file system, this automatically degrades to creating a
3405 normal directory. Among others /var/lib/machines is now
3406 created like this at boot, should it be missing.
3408 * The directory /var/lib/containers/ has been deprecated and
3409 been replaced by /var/lib/machines. The term "machines" has
3410 been used in the systemd context as generic term for both
3411 VMs and containers, and hence appears more appropriate for
3412 this, as the directory can also contain raw images bootable
3415 * systemd-nspawn when invoked with -M but without --directory=
3416 or --image= is now capable of searching for the container
3417 root directory, subvolume or disk image automatically, in
3418 /var/lib/machines. systemd-nspawn@.service has been updated
3419 to make use of this, thus allowing it to be used for raw
3422 * A new machines.target unit has been introduced that is
3423 supposed to group all containers/VMs invoked as services on
3424 the system. systemd-nspawn@.service has been updated to
3425 integrate with that.
3427 * machinectl gained a new "start" command, for invoking a
3428 container as a service. "machinectl start foo" is mostly
3429 equivalent to "systemctl start systemd-nspawn@foo.service",
3430 but handles escaping in a nicer way.
3432 * systemd-nspawn will now mount most of the cgroupfs tree
3433 read-only into each container, with the exception of the
3434 container's own subtree in the name=systemd hierarchy.
3436 * journald now sets the special FS_NOCOW file flag for its
3437 journal files. This should improve performance on btrfs, by
3438 avoiding heavy fragmentation when journald's write-pattern
3439 is used on COW file systems. It degrades btrfs' data
3440 integrity guarantees for the files to the same levels as for
3441 ext3/ext4 however. This should be OK though as journald does
3442 its own data integrity checks and all its objects are
3443 checksummed on disk. Also, journald should handle btrfs disk
3444 full events a lot more gracefully now, by processing SIGBUS
3445 errors, and not relying on fallocate() anymore.
3447 * When journald detects that journal files it is writing to
3448 have been deleted it will immediately start new journal
3451 * systemd now provides a way to store file descriptors
3452 per-service in PID 1. This is useful for daemons to ensure
3453 that fds they require are not lost during a daemon
3454 restart. The fds are passed to the daemon on the next
3455 invocation in the same way socket activation fds are
3456 passed. This is now used by journald to ensure that the
3457 various sockets connected to all the system's stdout/stderr
3458 are not lost when journald is restarted. File descriptors
3459 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3460 an extension to sd_notify(). Note that a limit is enforced
3461 on the number of fds a service can store in PID 1, and it
3462 defaults to 0, so that no fds may be stored, unless this is
3463 explicitly turned on.
3465 * The default TERM variable to use for units connected to a
3466 terminal, when no other value is explicitly is set is now
3467 vt220 rather than vt102. This should be fairly safe still,
3468 but allows PgUp/PgDn work.
3470 * The /etc/crypttab option header= as known from Debian is now
3473 * "loginctl user-status" and "loginctl session-status" will
3474 now show the last 10 lines of log messages of the
3475 user/session following the status output. Similar,
3476 "machinectl status" will show the last 10 log lines
3477 associated with a virtual machine or container
3478 service. (Note that this is usually not the log messages
3479 done in the VM/container itself, but simply what the
3480 container manager logs. For nspawn this includes all console
3483 * "loginctl session-status" without further argument will now
3484 show the status of the session of the caller. Similar,
3485 "lock-session", "unlock-session", "activate",
3486 "enable-linger", "disable-linger" may now be called without
3487 session/user parameter in which case they apply to the
3488 caller's session/user.
3490 * An X11 session scriptlet is now shipped that uploads
3491 $DISPLAY and $XAUTHORITY into the environment of the systemd
3492 --user daemon if a session begins. This should improve
3493 compatibility with X11 enabled applications run as systemd
3496 * Generators are now subject to masking via /etc and /run, the
3497 same way as unit files.
3499 * networkd .network files gained support for configuring
3500 per-link IPv4/IPv6 packet forwarding as well as IPv4
3501 masquerading. This is by default turned on for veth links to
3502 containers, as registered by systemd-nspawn. This means that
3503 nspawn containers run with --network-veth will now get
3504 automatic routed access to the host's networks without any
3505 further configuration or setup, as long as networkd runs on
3508 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3509 or UDP posts of a container on the host. With this in place
3510 it is possible to run containers with private veth links
3511 (--network-veth), and have their functionality exposed on
3512 the host as if their services were running directly on the
3515 * systemd-nspawn's --network-veth switch now gained a short
3516 version "-n", since with the changes above it is now truly
3517 useful out-of-the-box. The systemd-nspawn@.service has been
3518 updated to make use of it too by default.
3520 * systemd-nspawn will now maintain a per-image R/W lock, to
3521 ensure that the same image is not started more than once
3522 writable. (It's OK to run an image multiple times
3523 simultaneously in read-only mode.)
3525 * systemd-nspawn's --image= option is now capable of
3526 dissecting and booting MBR and GPT disk images that contain
3527 only a single active Linux partition. Previously it
3528 supported only GPT disk images with proper GPT type
3529 IDs. This allows running cloud images from major
3530 distributions directly with systemd-nspawn, without
3533 * In addition to collecting mouse dpi data in the udev
3534 hardware database, there's now support for collecting angle
3535 information for mouse scroll wheels. The database is
3536 supposed to guarantee similar scrolling behavior on mice
3537 that it knows about. There's also support for collecting
3538 information about Touchpad types.
3540 * udev's input_id built-in will now also collect touch screen
3541 dimension data and attach it to probed devices.
3543 * /etc/os-release gained support for a Distribution Privacy
3546 * networkd gained support for creating "ipvlan", "gretap",
3547 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3549 * systemd-tmpfiles gained support for "a" lines for setting
3552 * systemd-nspawn will now mount /tmp in the container to
3553 tmpfs, automatically.
3555 * systemd now exposes the memory.usage_in_bytes cgroup
3556 attribute and shows it for each service in the "systemctl
3557 status" output, if available.
3559 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3560 immediate reboot is triggered. This useful if shutdown is
3561 hung and is unable to complete, to expedite the
3562 operation. Note that this kind of reboot will still unmount
3563 all file systems, and hence should not result in fsck being
3566 * A .device unit for an optical block device will now be
3567 considered active only when a medium is in the drive. Also,
3568 mount units are now bound to their backing devices thus
3569 triggering automatic unmounting when devices become
3570 unavailable. With this in place systemd will now
3571 automatically unmount left-over mounts when a CD-ROM is
3572 ejected or an USB stick is yanked from the system.
3574 * networkd-wait-online now has support for waiting for
3575 specific interfaces only (with globbing), and for giving up
3576 after a configurable timeout.
3578 * networkd now exits when idle. It will be automatically
3579 restarted as soon as interfaces show up, are removed or
3580 change state. networkd will stay around as long as there is
3581 at least one DHCP state machine or similar around, that keep
3584 * networkd may now configure IPv6 link-local addressing in
3585 addition to IPv4 link-local addressing.
3587 * The IPv6 "token" for use in SLAAC may now be configured for
3588 each .network interface in networkd.
3590 * Routes configured with networkd may now be assigned a scope
3593 * networkd's [Match] sections now support globbing and lists
3594 of multiple space-separated matches per item.
3596 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3597 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3598 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3599 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3600 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3601 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3602 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3603 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3604 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3605 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3606 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3607 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3608 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3609 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3610 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3611 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3612 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3613 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3614 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3615 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3616 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3617 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3618 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3619 Hoffmann, Zbigniew Jędrzejewski-Szmek
3621 — Berlin, 2015-02-16
3625 * When querying unit file enablement status (for example via
3626 "systemctl is-enabled"), a new state "indirect" is now known
3627 which indicates that a unit might not be enabled itself, but
3628 another unit listed in its Also= setting might be.
3630 * Similar to the various existing ConditionXYZ= settings for
3631 units, there are now matching AssertXYZ= settings. While
3632 failing conditions cause a unit to be skipped, but its job
3633 to succeed, failing assertions declared like this will cause
3634 a unit start operation and its job to fail.
3636 * hostnamed now knows a new chassis type "embedded".
3638 * systemctl gained a new "edit" command. When used on a unit
3639 file, this allows extending unit files with .d/ drop-in
3640 configuration snippets or editing the full file (after
3641 copying it from /usr/lib to /etc). This will invoke the
3642 user's editor (as configured with $EDITOR), and reload the
3643 modified configuration after editing.
3645 * "systemctl status" now shows the suggested enablement state
3646 for a unit, as declared in the (usually vendor-supplied)
3647 system preset files.
3649 * nss-myhostname will now resolve the single-label host name
3650 "gateway" to the locally configured default IP routing
3651 gateways, ordered by their metrics. This assigns a stable
3652 name to the used gateways, regardless which ones are
3653 currently configured. Note that the name will only be
3654 resolved after all other name sources (if nss-myhostname is
3655 configured properly) and should hence not negatively impact
3656 systems that use the single-label host name "gateway" in
3659 * systemd-inhibit now allows filtering by mode when listing
3662 * Scope and service units gained a new "Delegate" boolean
3663 property, which, when set, allows processes running inside the
3664 unit to further partition resources. This is primarily
3665 useful for systemd user instances as well as container
3668 * journald will now pick up audit messages directly from
3669 the kernel, and log them like any other log message. The
3670 audit fields are split up and fully indexed. This means that
3671 journalctl in many ways is now a (nicer!) alternative to
3672 ausearch, the traditional audit client. Note that this
3673 implements only a minimal audit client. If you want the
3674 special audit modes like reboot-on-log-overflow, please use
3675 the traditional auditd instead, which can be used in
3676 parallel to journald.
3678 * The ConditionSecurity= unit file option now understands the
3679 special string "audit" to check whether auditing is
3682 * journalctl gained two new commands --vacuum-size= and
3683 --vacuum-time= to delete old journal files until the
3684 remaining ones take up no more than the specified size on disk,
3685 or are not older than the specified time.
3687 * A new, native PPPoE library has been added to sd-network,
3688 systemd's library of light-weight networking protocols. This
3689 library will be used in a future version of networkd to
3690 enable PPPoE communication without an external pppd daemon.
3692 * The busctl tool now understands a new "capture" verb that
3693 works similar to "monitor", but writes a packet capture
3694 trace to STDOUT that can be redirected to a file which is
3695 compatible with libcap's capture file format. This can then
3696 be loaded in Wireshark and similar tools to inspect bus
3699 * The busctl tool now understands a new "tree" verb that shows
3700 the object trees of a specific service on the bus, or of all
3703 * The busctl tool now understands a new "introspect" verb that
3704 shows all interfaces and members of objects on the bus,
3705 including their signature and values. This is particularly
3706 useful to get more information about bus objects shown by
3707 the new "busctl tree" command.
3709 * The busctl tool now understands new verbs "call",
3710 "set-property" and "get-property" for invoking bus method
3711 calls, setting and getting bus object properties in a
3714 * busctl gained a new --augment-creds= argument that controls
3715 whether the tool shall augment credential information it
3716 gets from the bus with data from /proc, in a possibly
3719 * nspawn's --link-journal= switch gained two new values
3720 "try-guest" and "try-host" that work like "guest" and
3721 "host", but do not fail if the host has no persistent
3722 journalling enabled. -j is now equivalent to
3723 --link-journal=try-guest.
3725 * macvlan network devices created by nspawn will now have
3726 stable MAC addresses.
3728 * A new SmackProcessLabel= unit setting has been added, which
3729 controls the SMACK security label processes forked off by
3730 the respective unit shall use.
3732 * If compiled with --enable-xkbcommon, systemd-localed will
3733 verify x11 keymap settings by compiling the given keymap. It
3734 will spew out warnings if the compilation fails. This
3735 requires libxkbcommon to be installed.
3737 * When a coredump is collected, a larger number of metadata
3738 fields is now collected and included in the journal records
3739 created for it. More specifically, control group membership,
3740 environment variables, memory maps, working directory,
3741 chroot directory, /proc/$PID/status, and a list of open file
3742 descriptors is now stored in the log entry.
3744 * The udev hwdb now contains DPI information for mice. For
3747 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3749 * All systemd programs that read standalone configuration
3750 files in /etc now also support a corresponding series of
3751 .conf.d configuration directories in /etc/, /run/,
3752 /usr/local/lib/, /usr/lib/, and (if configured with
3753 --enable-split-usr) /lib/. In particular, the following
3754 configuration files now have corresponding configuration
3755 directories: system.conf user.conf, logind.conf,
3756 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3757 resolved.conf, timesyncd.conf, journal-remote.conf, and
3758 journal-upload.conf. Note that distributions should use the
3759 configuration directories in /usr/lib/; the directories in
3760 /etc/ are reserved for the system administrator.
3762 * systemd-rfkill will no longer take the rfkill device name
3763 into account when storing rfkill state on disk, as the name
3764 might be dynamically assigned and not stable. Instead, the
3765 ID_PATH udev variable combined with the rfkill type (wlan,
3766 bluetooth, ...) is used.
3768 * A new service systemd-machine-id-commit.service has been
3769 added. When used on systems where /etc is read-only during
3770 boot, and /etc/machine-id is not initialized (but an empty
3771 file), this service will copy the temporary machine ID
3772 created as replacement into /etc after the system is fully
3773 booted up. This is useful for systems that are freshly
3774 installed with a non-initialized machine ID, but should get
3775 a fixed machine ID for subsequent boots.
3777 * networkd's .netdev files now provide a large set of
3778 configuration parameters for VXLAN devices. Similarly, the
3779 bridge port cost parameter is now configurable in .network
3780 files. There's also new support for configuring IP source
3781 routing. networkd .link files gained support for a new
3782 OriginalName= match that is useful to match against the
3783 original interface name the kernel assigned. .network files
3784 may include MTU= and MACAddress= fields for altering the MTU
3785 and MAC address while being connected to a specific network
3788 * The LUKS logic gained supported for configuring
3789 UUID-specific key files. There's also new support for naming
3790 LUKS device from the kernel command line, using the new
3791 luks.name= argument.
3793 * Timer units may now be transiently created via the bus API
3794 (this was previously already available for scope and service
3795 units). In addition it is now possible to create multiple
3796 transient units at the same time with a single bus call. The
3797 "systemd-run" tool has been updated to make use of this for
3798 running commands on a specified time, in at(1)-style.
3800 * tmpfiles gained support for "t" lines, for assigning
3801 extended attributes to files. Among other uses this may be
3802 used to assign SMACK labels to files.
3804 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3805 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3806 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3807 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3808 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3809 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3810 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3811 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3812 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3813 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3814 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3815 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3816 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3817 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3818 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3819 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3820 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3821 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3823 — Berlin, 2014-12-10
3827 * journalctl gained the new options -t/--identifier= to match
3828 on the syslog identifier (aka "tag"), as well as --utc to
3829 show log timestamps in the UTC timezone. journalctl now also
3830 accepts -n/--lines=all to disable line capping in a pager.
3832 * journalctl gained a new switch, --flush, that synchronously
3833 flushes logs from /run/log/journal to /var/log/journal if
3834 persistent storage is enabled. systemd-journal-flush.service
3835 now waits until the operation is complete.
3837 * Services can notify the manager before they start a reload
3838 (by sending RELOADING=1) or shutdown (by sending
3839 STOPPING=1). This allows the manager to track and show the
3840 internal state of daemons and closes a race condition when
3841 the process is still running but has closed its D-Bus
3844 * Services with Type=oneshot do not have to have any ExecStart
3847 * User units are now loaded also from
3848 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3849 /run/systemd/user directory that was already previously
3850 supported, but is under the control of the user.
3852 * Job timeouts (i.e. time-outs on the time a job that is
3853 queued stays in the run queue) can now optionally result in
3854 immediate reboot or power-off actions (JobTimeoutAction= and
3855 JobTimeoutRebootArgument=). This is useful on ".target"
3856 units, to limit the maximum time a target remains
3857 undispatched in the run queue, and to trigger an emergency
3858 operation in such a case. This is now used by default to
3859 turn off the system if boot-up (as defined by everything in
3860 basic.target) hangs and does not complete for at least
3861 15min. Also, if power-off or reboot hang for at least 30min
3862 an immediate power-off/reboot operation is triggered. This
3863 functionality is particularly useful to increase reliability
3864 on embedded devices, but also on laptops which might
3865 accidentally get powered on when carried in a backpack and
3866 whose boot stays stuck in a hard disk encryption passphrase
3869 * systemd-logind can be configured to also handle lid switch
3870 events even when the machine is docked or multiple displays
3871 are attached (HandleLidSwitchDocked= option).
3873 * A helper binary and a service have been added which can be
3874 used to resume from hibernation in the initramfs. A
3875 generator will parse the resume= option on the kernel
3876 command line to trigger resume.
3878 * A user console daemon systemd-consoled has been
3879 added. Currently, it is a preview, and will so far open a
3880 single terminal on each session of the user marked as
3881 Desktop=systemd-console.
3883 * Route metrics can be specified for DHCP routes added by
3886 * The SELinux context of socket-activated services can be set
3887 from the information provided by the networking stack
3888 (SELinuxContextFromNet= option).
3890 * Userspace firmware loading support has been removed and
3891 the minimum supported kernel version is thus bumped to 3.7.
3893 * Timeout for udev workers has been increased from 1 to 3
3894 minutes, but a warning will be printed after 1 minute to
3895 help diagnose kernel modules that take a long time to load.
3897 * Udev rules can now remove tags on devices with TAG-="foobar".
3899 * systemd's readahead implementation has been removed. In many
3900 circumstances it didn't give expected benefits even for
3901 rotational disk drives and was becoming less relevant in the
3902 age of SSDs. As none of the developers has been using
3903 rotating media anymore, and nobody stepped up to actively
3904 maintain this component of systemd it has now been removed.
3906 * Swap units can use Options= to specify discard options.
3907 Discard options specified for swaps in /etc/fstab are now
3910 * Docker containers are now detected as a separate type of
3913 * The Password Agent protocol gained support for queries where
3914 the user input is shown, useful e.g. for user names.
3915 systemd-ask-password gained a new --echo option to turn that
3918 * The default sysctl.d/ snippets will now set:
3920 net.core.default_qdisc = fq_codel
3922 This selects Fair Queuing Controlled Delay as the default
3923 queuing discipline for network interfaces. fq_codel helps
3924 fight the network bufferbloat problem. It is believed to be
3925 a good default with no tuning required for most workloads.
3926 Downstream distributions may override this choice. On 10Gbit
3927 servers that do not do forwarding, "fq" may perform better.
3928 Systems without a good clocksource should use "pfifo_fast".
3930 * If kdbus is enabled during build a new option BusPolicy= is
3931 available for service units, that allows locking all service
3932 processes into a stricter bus policy, in order to limit
3933 access to various bus services, or even hide most of them
3934 from the service's view entirely.
3936 * networkctl will now show the .network and .link file
3937 networkd has applied to a specific interface.
3939 * sd-login gained a new API call sd_session_get_desktop() to
3940 query which desktop environment has been selected for a
3943 * UNIX utmp support is now compile-time optional to support
3944 legacy-free systems.
3946 * systemctl gained two new commands "add-wants" and
3947 "add-requires" for pulling in units from specific targets
3950 * If the word "rescue" is specified on the kernel command line
3951 the system will now boot into rescue mode (aka
3952 rescue.target), which was previously available only by
3953 specifying "1" or "systemd.unit=rescue.target" on the kernel
3954 command line. This new kernel command line option nicely
3955 mirrors the already existing "emergency" kernel command line
3958 * New kernel command line options mount.usr=, mount.usrflags=,
3959 mount.usrfstype= have been added that match root=, rootflags=,
3960 rootfstype= but allow mounting a specific file system to
3963 * The $NOTIFY_SOCKET is now also passed to control processes of
3964 services, not only the main process.
3966 * This version reenables support for fsck's -l switch. This
3967 means at least version v2.25 of util-linux is required for
3968 operation, otherwise dead-locks on device nodes may
3969 occur. Again: you need to update util-linux to at least
3970 v2.25 when updating systemd to v217.
3972 * The "multi-seat-x" tool has been removed from systemd, as
3973 its functionality has been integrated into X servers 1.16,
3974 and the tool is hence redundant. It is recommended to update
3975 display managers invoking this tool to simply invoke X
3976 directly from now on, again.
3978 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3979 message flag has been added for all of systemd's PolicyKit
3980 authenticated method calls has been added. In particular
3981 this now allows optional interactive authorization via
3982 PolicyKit for many of PID1's privileged operations such as
3983 unit file enabling and disabling.
3985 * "udevadm hwdb --update" learnt a new switch "--usr" for
3986 placing the rebuilt hardware database in /usr instead of
3987 /etc. When used only hardware database entries stored in
3988 /usr will be used, and any user database entries in /etc are
3989 ignored. This functionality is useful for vendors to ship a
3990 pre-built database on systems where local configuration is
3991 unnecessary or unlikely.
3993 * Calendar time specifications in .timer units now also
3994 understand the strings "semi-annually", "quarterly" and
3995 "minutely" as shortcuts (in addition to the preexisting
3996 "anually", "hourly", ...).
3998 * systemd-tmpfiles will now correctly create files in /dev
3999 at boot which are marked for creation only at boot. It is
4000 recommended to always create static device nodes with 'c!'
4001 and 'b!', so that they are created only at boot and not
4002 overwritten at runtime.
4004 * When the watchdog logic is used for a service (WatchdogSec=)
4005 and the watchdog timeout is hit the service will now be
4006 terminated with SIGABRT (instead of just SIGTERM), in order
4007 to make sure a proper coredump and backtrace is
4008 generated. This ensures that hanging services will result in
4009 similar coredump/backtrace behaviour as services that hit a
4012 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4013 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4014 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4015 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4016 Herrmann, David Sommerseth, David Strauss, Emil Renner
4017 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4018 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4019 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4020 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4021 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4022 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4023 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4024 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4025 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4026 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4027 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4028 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4029 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4030 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4031 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4032 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4035 — Berlin, 2014-10-28
4039 * timedated no longer reads NTP implementation unit names from
4040 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4041 implementations should add a
4043 Conflicts=systemd-timesyncd.service
4045 to their unit files to take over and replace systemd's NTP
4046 default functionality.
4048 * systemd-sysusers gained a new line type "r" for configuring
4049 which UID/GID ranges to allocate system users/groups
4050 from. Lines of type "u" may now add an additional column
4051 that specifies the home directory for the system user to be
4052 created. Also, systemd-sysusers may now optionally read user
4053 information from STDIN instead of a file. This is useful for
4054 invoking it from RPM preinst scriptlets that need to create
4055 users before the first RPM file is installed since these
4056 files might need to be owned by them. A new
4057 %sysusers_create_inline RPM macro has been introduced to do
4058 just that. systemd-sysusers now updates the shadow files as
4059 well as the user/group databases, which should enhance
4060 compatibility with certain tools like grpck.
4062 * A number of bus APIs of PID 1 now optionally consult
4063 PolicyKit to permit access for otherwise unprivileged
4064 clients under certain conditions. Note that this currently
4065 doesn't support interactive authentication yet, but this is
4066 expected to be added eventually, too.
4068 * /etc/machine-info now has new fields for configuring the
4069 deployment environment of the machine, as well as the
4070 location of the machine. hostnamectl has been updated with
4071 new command to update these fields.
4073 * systemd-timesyncd has been updated to automatically acquire
4074 NTP server information from systemd-networkd, which might
4075 have been discovered via DHCP.
4077 * systemd-resolved now includes a caching DNS stub resolver
4078 and a complete LLMNR name resolution implementation. A new
4079 NSS module "nss-resolve" has been added which can be used
4080 instead of glibc's own "nss-dns" to resolve hostnames via
4081 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4082 be resolved via systemd-resolved D-Bus APIs. In contrast to
4083 the glibc internal resolver systemd-resolved is aware of
4084 multi-homed system, and keeps DNS server and caches separate
4085 and per-interface. Queries are sent simultaneously on all
4086 interfaces that have DNS servers configured, in order to
4087 properly handle VPNs and local LANs which might resolve
4088 separate sets of domain names. systemd-resolved may acquire
4089 DNS server information from systemd-networkd automatically,
4090 which in turn might have discovered them via DHCP. A tool
4091 "systemd-resolve-host" has been added that may be used to
4092 query the DNS logic in resolved. systemd-resolved implements
4093 IDNA and automatically uses IDNA or UTF-8 encoding depending
4094 on whether classic DNS or LLMNR is used as transport. In the
4095 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4096 implementation to systemd-resolved.
4098 * A new NSS module nss-mymachines has been added, that
4099 automatically resolves the names of all local registered
4100 containers to their respective IP addresses.
4102 * A new client tool "networkctl" for systemd-networkd has been
4103 added. It currently is entirely passive and will query
4104 networking configuration from udev, rtnetlink and networkd,
4105 and present it to the user in a very friendly
4106 way. Eventually, we hope to extend it to become a full
4107 control utility for networkd.
4109 * .socket units gained a new DeferAcceptSec= setting that
4110 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4111 TCP. Similarly, support for controlling TCP keep-alive
4112 settings has been added (KeepAliveTimeSec=,
4113 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4114 turning off Nagle's algorithm on TCP has been added
4117 * logind learned a new session type "web", for use in projects
4118 like Cockpit which register web clients as PAM sessions.
4120 * timer units with at least one OnCalendar= setting will now
4121 be started only after time-sync.target has been
4122 reached. This way they will not elapse before the system
4123 clock has been corrected by a local NTP client or
4124 similar. This is particular useful on RTC-less embedded
4125 machines, that come up with an invalid system clock.
4127 * systemd-nspawn's --network-veth= switch should now result in
4128 stable MAC addresses for both the outer and the inner side
4131 * systemd-nspawn gained a new --volatile= switch for running
4132 container instances with /etc or /var unpopulated.
4134 * The kdbus client code has been updated to use the new Linux
4135 3.17 memfd subsystem instead of the old kdbus-specific one.
4137 * systemd-networkd's DHCP client and server now support
4138 FORCERENEW. There are also new configuration options to
4139 configure the vendor client identifier and broadcast mode
4142 * systemd will no longer inform the kernel about the current
4143 timezone, as this is necessarily incorrect and racy as the
4144 kernel has no understanding of DST and similar
4145 concepts. This hence means FAT timestamps will be always
4146 considered UTC, similar to what Android is already
4147 doing. Also, when the RTC is configured to the local time
4148 (rather than UTC) systemd will never synchronize back to it,
4149 as this might confuse Windows at a later boot.
4151 * systemd-analyze gained a new command "verify" for offline
4152 validation of unit files.
4154 * systemd-networkd gained support for a couple of additional
4155 settings for bonding networking setups. Also, the metric for
4156 statically configured routes may now be configured. For
4157 network interfaces where this is appropriate the peer IP
4158 address may now be configured.
4160 * systemd-networkd's DHCP client will no longer request
4161 broadcasting by default, as this tripped up some networks.
4162 For hardware where broadcast is required the feature should
4163 be switched back on using RequestBroadcast=yes.
4165 * systemd-networkd will now set up IPv4LL addresses (when
4166 enabled) even if DHCP is configured successfully.
4168 * udev will now default to respect network device names given
4169 by the kernel when the kernel indicates that these are
4170 predictable. This behavior can be tweaked by changing
4171 NamePolicy= in the relevant .link file.
4173 * A new library systemd-terminal has been added that
4174 implements full TTY stream parsing and rendering. This
4175 library is supposed to be used later on for implementing a
4176 full userspace VT subsystem, replacing the current kernel
4179 * A new tool systemd-journal-upload has been added to push
4180 journal data to a remote system running
4181 systemd-journal-remote.
4183 * journald will no longer forward all local data to another
4184 running syslog daemon. This change has been made because
4185 rsyslog (which appears to be the most commonly used syslog
4186 implementation these days) no longer makes use of this, and
4187 instead pulls the data out of the journal on its own. Since
4188 forwarding the messages to a non-existent syslog server is
4189 more expensive than we assumed we have now turned this
4190 off. If you run a syslog server that is not a recent rsyslog
4191 version, you have to turn this option on again
4192 (ForwardToSyslog= in journald.conf).
4194 * journald now optionally supports the LZ4 compressor for
4195 larger journal fields. This compressor should perform much
4196 better than XZ which was the previous default.
4198 * machinectl now shows the IP addresses of local containers,
4199 if it knows them, plus the interface name of the container.
4201 * A new tool "systemd-escape" has been added that makes it
4202 easy to escape strings to build unit names and similar.
4204 * sd_notify() messages may now include a new ERRNO= field
4205 which is parsed and collected by systemd and shown among the
4206 "systemctl status" output for a service.
4208 * A new component "systemd-firstboot" has been added that
4209 queries the most basic systemd information (timezone,
4210 hostname, root password) interactively on first
4211 boot. Alternatively it may also be used to provision these
4212 things offline on OS images installed into directories.
4214 * The default sysctl.d/ snippets will now set
4216 net.ipv4.conf.default.promote_secondaries=1
4218 This has the benefit of no flushing secondary IP addresses
4219 when primary addresses are removed.
4221 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4222 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4223 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4224 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4225 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4226 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4227 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4228 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4229 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4230 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4231 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4232 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4233 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4234 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4235 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4237 — Berlin, 2014-08-19
4241 * A new tool systemd-sysusers has been added. This tool
4242 creates system users and groups in /etc/passwd and
4243 /etc/group, based on static declarative system user/group
4244 definitions in /usr/lib/sysusers.d/. This is useful to
4245 enable factory resets and volatile systems that boot up with
4246 an empty /etc directory, and thus need system users and
4247 groups created during early boot. systemd now also ships
4248 with two default sysusers.d/ files for the most basic
4249 users and groups systemd and the core operating system
4252 * A new tmpfiles snippet has been added that rebuilds the
4253 essential files in /etc on boot, should they be missing.
4255 * A directive for ensuring automatic clean-up of
4256 /var/cache/man/ has been removed from the default
4257 configuration. This line should now be shipped by the man
4258 implementation. The necessary change has been made to the
4259 man-db implementation. Note that you need to update your man
4260 implementation to one that ships this line, otherwise no
4261 automatic clean-up of /var/cache/man will take place.
4263 * A new condition ConditionNeedsUpdate= has been added that
4264 may conditionalize services to only run when /etc or /var
4265 are "older" than the vendor operating system resources in
4266 /usr. This is useful for reconstructing or updating /etc
4267 after an offline update of /usr or a factory reset, on the
4268 next reboot. Services that want to run once after such an
4269 update or reset should use this condition and order
4270 themselves before the new systemd-update-done.service, which
4271 will mark the two directories as fully updated. A number of
4272 service files have been added making use of this, to rebuild
4273 the udev hardware database, the journald message catalog and
4274 dynamic loader cache (ldconfig). The systemd-sysusers tool
4275 described above also makes use of this now. With this in
4276 place it is now possible to start up a minimal operating
4277 system with /etc empty cleanly. For more information on the
4278 concepts involved see this recent blog story:
4280 http://0pointer.de/blog/projects/stateless.html
4282 * A new system group "input" has been introduced, and all
4283 input device nodes get this group assigned. This is useful
4284 for system-level software to get access to input devices. It
4285 complements what is already done for "audio" and "video".
4287 * systemd-networkd learnt minimal DHCPv4 server support in
4288 addition to the existing DHCPv4 client support. It also
4289 learnt DHCPv6 client and IPv6 Router Solicitation client
4290 support. The DHCPv4 client gained support for static routes
4291 passed in from the server. Note that the [DHCPv4] section
4292 known in older systemd-networkd versions has been renamed to
4293 [DHCP] and is now also used by the DHCPv6 client. Existing
4294 .network files using settings of this section should be
4295 updated, though compatibility is maintained. Optionally, the
4296 client hostname may now be sent to the DHCP server.
4298 * networkd gained support for vxlan virtual networks as well
4299 as tun/tap and dummy devices.
4301 * networkd gained support for automatic allocation of address
4302 ranges for interfaces from a system-wide pool of
4303 addresses. This is useful for dynamically managing a large
4304 number of interfaces with a single network configuration
4305 file. In particular this is useful to easily assign
4306 appropriate IP addresses to the veth links of a large number
4307 of nspawn instances.
4309 * RPM macros for processing sysusers, sysctl and binfmt
4310 drop-in snippets at package installation time have been
4313 * The /etc/os-release file should now be placed in
4314 /usr/lib/os-release. The old location is automatically
4315 created as symlink. /usr/lib is the more appropriate
4316 location of this file, since it shall actually describe the
4317 vendor operating system shipped in /usr, and not the
4318 configuration stored in /etc.
4320 * .mount units gained a new boolean SloppyOptions= setting
4321 that maps to mount(8)'s -s option which enables permissive
4322 parsing of unknown mount options.
4324 * tmpfiles learnt a new "L+" directive which creates a symlink
4325 but (unlike "L") deletes a pre-existing file first, should
4326 it already exist and not already be the correct
4327 symlink. Similarly, "b+", "c+" and "p+" directives have been
4328 added as well, which create block and character devices, as
4329 well as fifos in the filesystem, possibly removing any
4330 pre-existing files of different types.
4332 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4333 'argument' field (which so far specified the source to
4334 symlink/copy the files from) is now optional. If omitted the
4335 same file os copied from /usr/share/factory/ suffixed by the
4336 full destination path. This is useful for populating /etc
4337 with essential files, by copying them from vendor defaults
4338 shipped in /usr/share/factory/etc.
4340 * A new command "systemctl preset-all" has been added that
4341 applies the service preset settings to all installed unit
4342 files. A new switch --preset-mode= has been added that
4343 controls whether only enable or only disable operations
4346 * A new command "systemctl is-system-running" has been added
4347 that allows checking the overall state of the system, for
4348 example whether it is fully up and running.
4350 * When the system boots up with an empty /etc, the equivalent
4351 to "systemctl preset-all" is executed during early boot, to
4352 make sure all default services are enabled after a factory
4355 * systemd now contains a minimal preset file that enables the
4356 most basic services systemd ships by default.
4358 * Unit files' [Install] section gained a new DefaultInstance=
4359 field for defining the default instance to create if a
4360 template unit is enabled with no instance specified.
4362 * A new passive target cryptsetup-pre.target has been added
4363 that may be used by services that need to make they run and
4364 finish before the first LUKS cryptographic device is set up.
4366 * The /dev/loop-control and /dev/btrfs-control device nodes
4367 are now owned by the "disk" group by default, opening up
4368 access to this group.
4370 * systemd-coredump will now automatically generate a
4371 stack trace of all core dumps taking place on the system,
4372 based on elfutils' libdw library. This stack trace is logged
4375 * systemd-coredump may now optionally store coredumps directly
4376 on disk (in /var/lib/systemd/coredump, possibly compressed),
4377 instead of storing them unconditionally in the journal. This
4378 mode is the new default. A new configuration file
4379 /etc/systemd/coredump.conf has been added to configure this
4380 and other parameters of systemd-coredump.
4382 * coredumpctl gained a new "info" verb to show details about a
4383 specific coredump. A new switch "-1" has also been added
4384 that makes sure to only show information about the most
4385 recent entry instead of all entries. Also, as the tool is
4386 generally useful now the "systemd-" prefix of the binary
4387 name has been removed. Distributions that want to maintain
4388 compatibility with the old name should add a symlink from
4389 the old name to the new name.
4391 * journald's SplitMode= now defaults to "uid". This makes sure
4392 that unprivileged users can access their own coredumps with
4393 coredumpctl without restrictions.
4395 * New kernel command line options "systemd.wants=" (for
4396 pulling an additional unit during boot), "systemd.mask="
4397 (for masking a specific unit for the boot), and
4398 "systemd.debug-shell" (for enabling the debug shell on tty9)
4399 have been added. This is implemented in the new generator
4400 "systemd-debug-generator".
4402 * systemd-nspawn will now by default filter a couple of
4403 syscalls for containers, among them those required for
4404 kernel module loading, direct x86 IO port access, swap
4405 management, and kexec. Most importantly though
4406 open_by_handle_at() is now prohibited for containers,
4407 closing a hole similar to a recently discussed vulnerability
4408 in docker regarding access to files on file hierarchies the
4409 container should normally not have access to. Note that, for
4410 nspawn, we generally make no security claims anyway (and
4411 this is explicitly documented in the man page), so this is
4412 just a fix for one of the most obvious problems.
4414 * A new man page file-hierarchy(7) has been added that
4415 contains a minimized, modernized version of the file system
4416 layout systemd expects, similar in style to the FHS
4417 specification or hier(5). A new tool systemd-path(1) has
4418 been added to query many of these paths for the local
4421 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4422 longer done. Since the directory now has a per-user size
4423 limit, and is cleaned on logout this appears unnecessary,
4424 in particular since this now brings the lifecycle of this
4425 directory closer in line with how IPC objects are handled.
4427 * systemd.pc now exports a number of additional directories,
4428 including $libdir (which is useful to identify the library
4429 path for the primary architecture of the system), and a
4430 couple of drop-in directories.
4432 * udev's predictable network interface names now use the dev_port
4433 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4434 distinguish between ports of the same PCI function. dev_id should
4435 only be used for ports using the same HW address, hence the need
4438 * machined has been updated to export the OS version of a
4439 container (read from /etc/os-release and
4440 /usr/lib/os-release) on the bus. This is now shown in
4441 "machinectl status" for a machine.
4443 * A new service setting RestartForceExitStatus= has been
4444 added. If configured to a set of exit signals or process
4445 return values, the service will be restarted when the main
4446 daemon process exits with any of them, regardless of the
4449 * systemctl's -H switch for connecting to remote systemd
4450 machines has been extended so that it may be used to
4451 directly connect to a specific container on the
4452 host. "systemctl -H root@foobar:waldi" will now connect as
4453 user "root" to host "foobar", and then proceed directly to
4454 the container named "waldi". Note that currently you have to
4455 authenticate as user "root" for this to work, as entering
4456 containers is a privileged operation.
4458 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4459 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4460 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4461 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4462 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4463 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4464 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4465 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4466 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4467 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4468 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4469 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4471 — Berlin, 2014-07-03
4475 * As an experimental feature, udev now tries to lock the
4476 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4477 executes events for the disk or any of its partitions.
4478 Applications like partitioning programs can lock the
4479 disk device node (flock(LOCK_EX)) and claim temporary
4480 device ownership that way; udev will entirely skip all event
4481 handling for this disk and its partitions. If the disk
4482 was opened for writing, the close will trigger a partition
4483 table rescan in udev's "watch" facility, and if needed
4484 synthesize "change" events for the disk and all its partitions.
4485 This is now unconditionally enabled, and if it turns out to
4486 cause major problems, we might turn it on only for specific
4487 devices, or might need to disable it entirely. Device Mapper
4488 devices are excluded from this logic.
4490 * We temporarily dropped the "-l" switch for fsck invocations,
4491 since they collide with the flock() logic above. util-linux
4492 upstream has been changed already to avoid this conflict,
4493 and we will readd "-l" as soon as util-linux with this
4494 change has been released.
4496 * The dependency on libattr has been removed. Since a long
4497 time, the extended attribute calls have moved to glibc, and
4498 libattr is thus unnecessary.
4500 * Virtualization detection works without privileges now. This
4501 means the systemd-detect-virt binary no longer requires
4502 CAP_SYS_PTRACE file capabilities, and our daemons can run
4503 with fewer privileges.
4505 * systemd-networkd now runs under its own "systemd-network"
4506 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4507 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4508 loses the ability to write to files owned by root this way.
4510 * Similarly, systemd-resolved now runs under its own
4511 "systemd-resolve" user with no capabilities remaining.
4513 * Similarly, systemd-bus-proxyd now runs under its own
4514 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4516 * systemd-networkd gained support for setting up "veth"
4517 virtual Ethernet devices for container connectivity, as well
4518 as GRE and VTI tunnels.
4520 * systemd-networkd will no longer automatically attempt to
4521 manually load kernel modules necessary for certain tunnel
4522 transports. Instead, it is assumed the kernel loads them
4523 automatically when required. This only works correctly on
4524 very new kernels. On older kernels, please consider adding
4525 the kernel modules to /etc/modules-load.d/ as a work-around.
4527 * The resolv.conf file systemd-resolved generates has been
4528 moved to /run/systemd/resolve/. If you have a symlink from
4529 /etc/resolv.conf, it might be necessary to correct it.
4531 * Two new service settings, ProtectHome= and ProtectSystem=,
4532 have been added. When enabled, they will make the user data
4533 (such as /home) inaccessible or read-only and the system
4534 (such as /usr) read-only, for specific services. This allows
4535 very light-weight per-service sandboxing to avoid
4536 modifications of user data or system files from
4537 services. These two new switches have been enabled for all
4538 of systemd's long-running services, where appropriate.
4540 * Socket units gained new SocketUser= and SocketGroup=
4541 settings to set the owner user and group of AF_UNIX sockets
4542 and FIFOs in the file system.
4544 * Socket units gained a new RemoveOnStop= setting. If enabled,
4545 all FIFOS and sockets in the file system will be removed
4546 when the specific socket unit is stopped.
4548 * Socket units gained a new Symlinks= setting. It takes a list
4549 of symlinks to create to file system sockets or FIFOs
4550 created by the specific Unix sockets. This is useful to
4551 manage symlinks to socket nodes with the same life-cycle as
4554 * The /dev/log socket and /dev/initctl FIFO have been moved to
4555 /run, and have been replaced by symlinks. This allows
4556 connecting to these facilities even if PrivateDevices=yes is
4557 used for a service (which makes /dev/log itself unavailable,
4558 but /run is left). This also has the benefit of ensuring
4559 that /dev only contains device nodes, directories and
4560 symlinks, and nothing else.
4562 * sd-daemon gained two new calls sd_pid_notify() and
4563 sd_pid_notifyf(). They are similar to sd_notify() and
4564 sd_notifyf(), but allow overriding of the source PID of
4565 notification messages if permissions permit this. This is
4566 useful to send notify messages on behalf of a different
4567 process (for example, the parent process). The
4568 systemd-notify tool has been updated to make use of this
4569 when sending messages (so that notification messages now
4570 originate from the shell script invoking systemd-notify and
4571 not the systemd-notify process itself. This should minimize
4572 a race where systemd fails to associate notification
4573 messages to services when the originating process already
4576 * A new "on-abnormal" setting for Restart= has been added. If
4577 set, it will result in automatic restarts on all "abnormal"
4578 reasons for a process to exit, which includes unclean
4579 signals, core dumps, timeouts and watchdog timeouts, but
4580 does not include clean and unclean exit codes or clean
4581 signals. Restart=on-abnormal is an alternative for
4582 Restart=on-failure for services that shall be able to
4583 terminate and avoid restarts on certain errors, by
4584 indicating so with an unclean exit code. Restart=on-failure
4585 or Restart=on-abnormal is now the recommended setting for
4586 all long-running services.
4588 * If the InaccessibleDirectories= service setting points to a
4589 mount point (or if there are any submounts contained within
4590 it), it is now attempted to completely unmount it, to make
4591 the file systems truly unavailable for the respective
4594 * The ReadOnlyDirectories= service setting and
4595 systemd-nspawn's --read-only parameter are now recursively
4596 applied to all submounts, too.
4598 * Mount units may now be created transiently via the bus APIs.
4600 * The support for SysV and LSB init scripts has been removed
4601 from the systemd daemon itself. Instead, it is now
4602 implemented as a generator that creates native systemd units
4603 from these scripts when needed. This enables us to remove a
4604 substantial amount of legacy code from PID 1, following the
4605 fact that many distributions only ship a very small number
4606 of LSB/SysV init scripts nowadays.
4608 * Privileged Xen (dom0) domains are not considered
4609 virtualization anymore by the virtualization detection
4610 logic. After all, they generally have unrestricted access to
4611 the hardware and usually are used to manage the unprivileged
4614 * systemd-tmpfiles gained a new "C" line type, for copying
4615 files or entire directories.
4617 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4618 lines. So far, they have been non-globbing versions of the
4619 latter, and have thus been redundant. In future, it is
4620 recommended to only use "z". "m" has hence been removed
4621 from the documentation, even though it stays supported.
4623 * A tmpfiles snippet to recreate the most basic structure in
4624 /var has been added. This is enough to create the /var/run →
4625 /run symlink and create a couple of structural
4626 directories. This allows systems to boot up with an empty or
4627 volatile /var. Of course, while with this change, the core OS
4628 now is capable with dealing with a volatile /var, not all
4629 user services are ready for it. However, we hope that sooner
4630 or later, many service daemons will be changed upstream so
4631 that they are able to automatically create their necessary
4632 directories in /var at boot, should they be missing. This is
4633 the first step to allow state-less systems that only require
4634 the vendor image for /usr to boot.
4636 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4637 empty tmpfs instance to a specific directory. This is
4638 particularly useful for making use of the automatic
4639 reconstruction of /var (see above), by passing --tmpfs=/var.
4641 * Access modes specified in tmpfiles snippets may now be
4642 prefixed with "~", which indicates that they shall be masked
4643 by whether the existing file or directory is currently
4644 writable, readable or executable at all. Also, if specified,
4645 the sgid/suid/sticky bits will be masked for all
4648 * A new passive target unit "network-pre.target" has been
4649 added which is useful for services that shall run before any
4650 network is configured, for example firewall scripts.
4652 * The "floppy" group that previously owned the /dev/fd*
4653 devices is no longer used. The "disk" group is now used
4654 instead. Distributions should probably deprecate usage of
4657 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4658 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4659 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4660 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4661 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4662 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4665 — Berlin, 2014-06-11
4669 * A new "systemd-timesyncd" daemon has been added for
4670 synchronizing the system clock across the network. It
4671 implements an SNTP client. In contrast to NTP
4672 implementations such as chrony or the NTP reference server,
4673 this only implements a client side, and does not bother with
4674 the full NTP complexity, focusing only on querying time from
4675 one remote server and synchronizing the local clock to
4676 it. Unless you intend to serve NTP to networked clients or
4677 want to connect to local hardware clocks, this simple NTP
4678 client should be more than appropriate for most
4679 installations. The daemon runs with minimal privileges, and
4680 has been hooked up with networkd to only operate when
4681 network connectivity is available. The daemon saves the
4682 current clock to disk every time a new NTP sync has been
4683 acquired, and uses this to possibly correct the system clock
4684 early at bootup, in order to accommodate for systems that
4685 lack an RTC such as the Raspberry Pi and embedded devices,
4686 and to make sure that time monotonically progresses on these
4687 systems, even if it is not always correct. To make use of
4688 this daemon, a new system user and group "systemd-timesync"
4689 needs to be created on installation of systemd.
4691 * The queue "seqnum" interface of libudev has been disabled, as
4692 it was generally incompatible with device namespacing as
4693 sequence numbers of devices go "missing" if the devices are
4694 part of a different namespace.
4696 * "systemctl list-timers" and "systemctl list-sockets" gained
4697 a --recursive switch for showing units of these types also
4698 for all local containers, similar in style to the already
4699 supported --recursive switch for "systemctl list-units".
4701 * A new RebootArgument= setting has been added for service
4702 units, which may be used to specify a kernel reboot argument
4703 to use when triggering reboots with StartLimitAction=.
4705 * A new FailureAction= setting has been added for service
4706 units which may be used to specify an operation to trigger
4707 when a service fails. This works similarly to
4708 StartLimitAction=, but unlike it, controls what is done
4709 immediately rather than only after several attempts to
4710 restart the service in question.
4712 * hostnamed got updated to also expose the kernel name,
4713 release, and version on the bus. This is useful for
4714 executing commands like hostnamectl with the -H switch.
4715 systemd-analyze makes use of this to properly display
4716 details when running non-locally.
4718 * The bootchart tool can now show cgroup information in the
4719 graphs it generates.
4721 * The CFS CPU quota cgroup attribute is now exposed for
4722 services. The new CPUQuota= switch has been added for this
4723 which takes a percentage value. Setting this will have the
4724 result that a service may never get more CPU time than the
4725 specified percentage, even if the machine is otherwise idle.
4727 * systemd-networkd learned IPIP and SIT tunnel support.
4729 * LSB init scripts exposing a dependency on $network will now
4730 get a dependency on network-online.target rather than simply
4731 network.target. This should bring LSB handling closer to
4732 what it was on SysV systems.
4734 * A new fsck.repair= kernel option has been added to control
4735 how fsck shall deal with unclean file systems at boot.
4737 * The (.ini) configuration file parser will now silently
4738 ignore sections whose name begins with "X-". This may be
4739 used to maintain application-specific extension sections in unit
4742 * machined gained a new API to query the IP addresses of
4743 registered containers. "machinectl status" has been updated
4744 to show these addresses in its output.
4746 * A new call sd_uid_get_display() has been added to the
4747 sd-login APIs for querying the "primary" session of a
4748 user. The "primary" session of the user is elected from the
4749 user's sessions and generally a graphical session is
4750 preferred over a text one.
4752 * A minimal systemd-resolved daemon has been added. It
4753 currently simply acts as a companion to systemd-networkd and
4754 manages resolv.conf based on per-interface DNS
4755 configuration, possibly supplied via DHCP. In the long run
4756 we hope to extend this into a local DNSSEC enabled DNS and
4759 * The systemd-networkd-wait-online tool is now enabled by
4760 default. It will delay network-online.target until a network
4761 connection has been configured. The tool primarily integrates
4762 with networkd, but will also make a best effort to make sense
4763 of network configuration performed in some other way.
4765 * Two new service options StartupCPUShares= and
4766 StartupBlockIOWeight= have been added that work similarly to
4767 CPUShares= and BlockIOWeight= however only apply during
4768 system startup. This is useful to prioritize certain services
4769 differently during bootup than during normal runtime.
4771 * hostnamed has been changed to prefer the statically
4772 configured hostname in /etc/hostname (unless set to
4773 'localhost' or empty) over any dynamic one supplied by
4774 dhcp. With this change, the rules for picking the hostname
4775 match more closely the rules of other configuration settings
4776 where the local administrator's configuration in /etc always
4777 overrides any other settings.
4779 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4780 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4781 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4782 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4783 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4784 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4785 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4786 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4787 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4788 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4789 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4790 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4791 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4792 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4793 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4794 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4797 — Beijing, 2014-05-28
4801 * When restoring the screen brightness at boot, stay away from
4802 the darkest setting or from the lowest 5% of the available
4803 range, depending on which is the larger value of both. This
4804 should effectively protect the user from rebooting into a
4805 black screen, should the brightness have been set to minimum
4808 * sd-login gained a new sd_machine_get_class() call to
4809 determine the class ("vm" or "container") of a machine
4810 registered with machined.
4812 * sd-login gained new calls
4813 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4814 to query the identity of the peer of a local AF_UNIX
4815 connection. They operate similarly to their sd_pid_get_xyz()
4818 * PID 1 will now maintain a system-wide system state engine
4819 with the states "starting", "running", "degraded",
4820 "maintenance", "stopping". These states are bound to system
4821 startup, normal runtime, runtime with at least one failed
4822 service, rescue/emergency mode and system shutdown. This
4823 state is shown in the "systemctl status" output when no unit
4824 name is passed. It is useful to determine system state, in
4825 particularly when doing so for many systems or containers at
4828 * A new command "list-machines" has been added to "systemctl"
4829 that lists all local OS containers and shows their system
4830 state (see above), if systemd runs inside of them.
4832 * systemctl gained a new "-r" switch to recursively enumerate
4833 units on all local containers, when used with the
4834 "list-unit" command (which is the default one that is
4835 executed when no parameters are specified).
4837 * The GPT automatic partition discovery logic will now honour
4838 two GPT partition flags: one may be set on a partition to
4839 cause it to be mounted read-only, and the other may be set
4840 on a partition to ignore it during automatic discovery.
4842 * Two new GPT type UUIDs have been added for automatic root
4843 partition discovery, for 32-bit and 64-bit ARM. This is not
4844 particularly useful for discovering the root directory on
4845 these architectures during bare-metal boots (since UEFI is
4846 not common there), but still very useful to allow booting of
4847 ARM disk images in nspawn with the -i option.
4849 * MAC addresses of interfaces created with nspawn's
4850 --network-interface= switch will now be generated from the
4851 machine name, and thus be stable between multiple invocations
4854 * logind will now automatically remove all IPC objects owned
4855 by a user if she or he fully logs out. This makes sure that
4856 users who are logged out cannot continue to consume IPC
4857 resources. This covers SysV memory, semaphores and message
4858 queues as well as POSIX shared memory and message
4859 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4860 limits. With this functionality, that is corrected. This may
4861 be turned off by using the RemoveIPC= switch of logind.conf.
4863 * The systemd-machine-id-setup and tmpfiles tools gained a
4864 --root= switch to operate on a specific root directory,
4867 * journald can now forward logged messages to the TTYs of all
4868 logged in users ("wall"). This is the default for all
4869 emergency messages now.
4871 * A new tool systemd-journal-remote has been added to stream
4872 journal log messages across the network.
4874 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4875 controller trees are mounted into it. Note that the
4876 directories mounted beneath it are not read-only. This is a
4877 security measure and is particularly useful because glibc
4878 actually includes a search logic to pick any tmpfs it can
4879 find to implement shm_open() if /dev/shm is not available
4880 (which it might very well be in namespaced setups).
4882 * machinectl gained a new "poweroff" command to cleanly power
4883 down a local OS container.
4885 * The PrivateDevices= unit file setting will now also drop the
4886 CAP_MKNOD capability from the capability bound set, and
4887 imply DevicePolicy=closed.
4889 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4890 comprehensively on all long-running systemd services where
4891 this is appropriate.
4893 * systemd-udevd will now run in a disassociated mount
4894 namespace. To mount directories from udev rules, make sure to
4895 pull in mount units via SYSTEMD_WANTS properties.
4897 * The kdbus support gained support for uploading policy into
4898 the kernel. sd-bus gained support for creating "monitoring"
4899 connections that can eavesdrop into all bus communication
4900 for debugging purposes.
4902 * Timestamps may now be specified in seconds since the UNIX
4903 epoch Jan 1st, 1970 by specifying "@" followed by the value
4906 * Native tcpwrap support in systemd has been removed. tcpwrap
4907 is old code, not really maintained anymore and has serious
4908 shortcomings, and better options such as firewalls
4909 exist. For setups that require tcpwrap usage, please
4910 consider invoking your socket-activated service via tcpd,
4911 like on traditional inetd.
4913 * A new system.conf configuration option
4914 DefaultTimerAccuracySec= has been added that controls the
4915 default AccuracySec= setting of .timer units.
4917 * Timer units gained a new WakeSystem= switch. If enabled,
4918 timers configured this way will cause the system to resume
4919 from system suspend (if the system supports that, which most
4922 * Timer units gained a new Persistent= switch. If enabled,
4923 timers configured this way will save to disk when they have
4924 been last triggered. This information is then used on next
4925 reboot to possible execute overdue timer events, that
4926 could not take place because the system was powered off.
4927 This enables simple anacron-like behaviour for timer units.
4929 * systemctl's "list-timers" will now also list the time a
4930 timer unit was last triggered in addition to the next time
4931 it will be triggered.
4933 * systemd-networkd will now assign predictable IPv4LL
4934 addresses to its local interfaces.
4936 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4937 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4938 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4939 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4940 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4941 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4942 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4943 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4946 — Berlin, 2014-03-25
4950 * A new unit file setting RestrictAddressFamilies= has been
4951 added to restrict which socket address families unit
4952 processes gain access to. This takes address family names
4953 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4954 attack surface of services via exotic protocol stacks. This
4955 is built on seccomp system call filters.
4957 * Two new unit file settings RuntimeDirectory= and
4958 RuntimeDirectoryMode= have been added that may be used to
4959 manage a per-daemon runtime directories below /run. This is
4960 an alternative for setting up directory permissions with
4961 tmpfiles snippets, and has the advantage that the runtime
4962 directory's lifetime is bound to the daemon runtime and that
4963 the daemon starts up with an empty directory each time. This
4964 is particularly useful when writing services that drop
4965 privileges using the User= or Group= setting.
4967 * The DeviceAllow= unit setting now supports globbing for
4968 matching against device group names.
4970 * The systemd configuration file system.conf gained new
4971 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4972 DefaultMemoryAccounting= to globally turn on/off accounting
4973 for specific resources (cgroups) for all units. These
4974 settings may still be overridden individually in each unit
4977 * systemd-gpt-auto-generator is now able to discover /srv and
4978 root partitions in addition to /home and swap partitions. It
4979 also supports LUKS-encrypted partitions now. With this in
4980 place, automatic discovery of partitions to mount following
4981 the Discoverable Partitions Specification
4982 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4983 is now a lot more complete. This allows booting without
4984 /etc/fstab and without root= on the kernel command line on
4985 systems prepared appropriately.
4987 * systemd-nspawn gained a new --image= switch which allows
4988 booting up disk images and Linux installations on any block
4989 device that follow the Discoverable Partitions Specification
4990 (see above). This means that installations made with
4991 appropriately updated installers may now be started and
4992 deployed using container managers, completely
4993 unmodified. (We hope that libvirt-lxc will add support for
4994 this feature soon, too.)
4996 * systemd-nspawn gained a new --network-macvlan= setting to
4997 set up a private macvlan interface for the
4998 container. Similarly, systemd-networkd gained a new
4999 Kind=macvlan setting in .netdev files.
5001 * systemd-networkd now supports configuring local addresses
5004 * A new tool systemd-network-wait-online has been added to
5005 synchronously wait for network connectivity using
5008 * The sd-bus.h bus API gained a new sd_bus_track object for
5009 tracking the life-cycle of bus peers. Note that sd-bus.h is
5010 still not a public API though (unless you specify
5011 --enable-kdbus on the configure command line, which however
5012 voids your warranty and you get no API stability guarantee).
5014 * The $XDG_RUNTIME_DIR runtime directories for each user are
5015 now individual tmpfs instances, which has the benefit of
5016 introducing separate pools for each user, with individual
5017 size limits, and thus making sure that unprivileged clients
5018 can no longer negatively impact the system or other users by
5019 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5020 RuntimeDirectorySize= has been introduced that allows
5021 controlling the default size limit for all users. It
5022 defaults to 10% of the available physical memory. This is no
5023 replacement for quotas on tmpfs though (which the kernel
5024 still does not support), as /dev/shm and /tmp are still
5025 shared resources used by both the system and unprivileged
5028 * logind will now automatically turn off automatic suspending
5029 on laptop lid close when more than one display is
5030 connected. This was previously expected to be implemented
5031 individually in desktop environments (such as GNOME),
5032 however has been added to logind now, in order to fix a
5033 boot-time race where a desktop environment might not have
5034 been started yet and thus not been able to take an inhibitor
5035 lock at the time where logind already suspends the system
5036 due to a closed lid.
5038 * logind will now wait at least 30s after each system
5039 suspend/resume cycle, and 3min after system boot before
5040 suspending the system due to a closed laptop lid. This
5041 should give USB docking stations and similar enough time to
5042 be probed and configured after system resume and boot in
5043 order to then act as suspend blocker.
5045 * systemd-run gained a new --property= setting which allows
5046 initialization of resource control properties (and others)
5047 for the created scope or service unit. Example: "systemd-run
5048 --property=BlockIOWeight=10 updatedb" may be used to run
5049 updatedb at a low block IO scheduling weight.
5051 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5052 now also work in --scope mode.
5054 * When systemd is compiled with kdbus support, basic support
5055 for enforced policies is now in place. (Note that enabling
5056 kdbus still voids your warranty and no API compatibility
5059 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5060 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5061 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5062 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5063 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5064 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5065 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5066 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5067 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5068 Zbigniew Jędrzejewski-Szmek
5070 — Berlin, 2014-03-12
5074 * systemd will now relabel /dev after loading the SMACK policy
5075 according to SMACK rules.
5077 * A new unit file option AppArmorProfile= has been added to
5078 set the AppArmor profile for the processes of a unit.
5080 * A new condition check ConditionArchitecture= has been added
5081 to conditionalize units based on the system architecture, as
5082 reported by uname()'s "machine" field.
5084 * systemd-networkd now supports matching on the system
5085 virtualization, architecture, kernel command line, host name
5088 * logind is now a lot more aggressive when suspending the
5089 machine due to a closed laptop lid. Instead of acting only
5090 on the lid close action, it will continuously watch the lid
5091 status and act on it. This is useful for laptops where the
5092 power button is on the outside of the chassis so that it can
5093 be reached without opening the lid (such as the Lenovo
5094 Yoga). On those machines, logind will now immediately
5095 re-suspend the machine if the power button has been
5096 accidentally pressed while the laptop was suspended and in a
5097 backpack or similar.
5099 * logind will now watch SW_DOCK switches and inhibit reaction
5100 to the lid switch if it is pressed. This means that logind
5101 will not suspend the machine anymore if the lid is closed
5102 and the system is docked, if the laptop supports SW_DOCK
5103 notifications via the input layer. Note that ACPI docking
5104 stations do not generate this currently. Also note that this
5105 logic is usually not fully sufficient and Desktop
5106 Environments should take a lid switch inhibitor lock when an
5107 external display is connected, as systemd will not watch
5110 * nspawn will now make use of the devices cgroup controller by
5111 default, and only permit creation of and access to the usual
5112 API device nodes like /dev/null or /dev/random, as well as
5113 access to (but not creation of) the pty devices.
5115 * We will now ship a default .network file for
5116 systemd-networkd that automatically configures DHCP for
5117 network interfaces created by nspawn's --network-veth or
5118 --network-bridge= switches.
5120 * systemd will now understand the usual M, K, G, T suffixes
5121 according to SI conventions (i.e. to the base 1000) when
5122 referring to throughput and hardware metrics. It will stay
5123 with IEC conventions (i.e. to the base 1024) for software
5124 metrics, according to what is customary according to
5125 Wikipedia. We explicitly document which base applies for
5126 each configuration option.
5128 * The DeviceAllow= setting in unit files now supports a syntax
5129 to whitelist an entire group of devices node majors at once,
5130 based on the /proc/devices listing. For example, with the
5131 string "char-pts", it is now possible to whitelist all
5132 current and future pseudo-TTYs at once.
5134 * sd-event learned a new "post" event source. Event sources of
5135 this type are triggered by the dispatching of any event
5136 source of a type that is not "post". This is useful for
5137 implementing clean-up and check event sources that are
5138 triggered by other work being done in the program.
5140 * systemd-networkd is no longer statically enabled, but uses
5141 the usual [Install] sections so that it can be
5142 enabled/disabled using systemctl. It still is enabled by
5145 * When creating a veth interface pair with systemd-nspawn, the
5146 host side will now be prefixed with "vb-" if
5147 --network-bridge= is used, and with "ve-" if --network-veth
5148 is used. This way, it is easy to distinguish these cases on
5149 the host, for example to apply different configuration to
5150 them with systemd-networkd.
5152 * The compatibility libraries for libsystemd-journal.so,
5153 libsystem-id128.so, libsystemd-login.so and
5154 libsystemd-daemon.so do not make use of IFUNC
5155 anymore. Instead, we now build libsystemd.so multiple times
5156 under these alternative names. This means that the footprint
5157 is drastically increased, but given that these are
5158 transitional compatibility libraries, this should not matter
5159 much. This change has been made necessary to support the ARM
5160 platform for these compatibility libraries, as the ARM
5161 toolchain is not really at the same level as the toolchain
5162 for other architectures like x86 and does not support
5163 IFUNC. Please make sure to use --enable-compat-libs only
5164 during a transitional period!
5166 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5167 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5168 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5169 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5170 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5171 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5172 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5173 Zbigniew Jędrzejewski-Szmek
5175 — Berlin, 2014-02-24
5179 * A new component "systemd-networkd" has been added that can
5180 be used to configure local network interfaces statically or
5181 via DHCP. It is capable of bringing up bridges, VLANs, and
5182 bonding. Currently, no hook-ups for interactive network
5183 configuration are provided. Use this for your initrd,
5184 container, embedded, or server setup if you need a simple,
5185 yet powerful, network configuration solution. This
5186 configuration subsystem is quite nifty, as it allows wildcard
5187 hotplug matching in interfaces. For example, with a single
5188 configuration snippet, you can configure that all Ethernet
5189 interfaces showing up are automatically added to a bridge,
5190 or similar. It supports link-sensing and more.
5192 * A new tool "systemd-socket-proxyd" has been added which can
5193 act as a bidirectional proxy for TCP sockets. This is
5194 useful for adding socket activation support to services that
5195 do not actually support socket activation, including virtual
5196 machines and the like.
5198 * Add a new tool to save/restore rfkill state on
5201 * Save/restore state of keyboard backlights in addition to
5202 display backlights on shutdown/boot.
5204 * udev learned a new SECLABEL{} construct to label device
5205 nodes with a specific security label when they appear. For
5206 now, only SECLABEL{selinux} is supported, but the syntax is
5207 prepared for additional security frameworks.
5209 * udev gained a new scheme to configure link-level attributes
5210 from files in /etc/systemd/network/*.link. These files can
5211 match against MAC address, device path, driver name and type,
5212 and will apply attributes like the naming policy, link speed,
5213 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5214 address assignment policy (randomized, ...).
5216 * The configuration of network interface naming rules for
5217 "permanent interface names" has changed: a new NamePolicy=
5218 setting in the [Link] section of .link files determines the
5219 priority of possible naming schemes (onboard, slot, MAC,
5220 path). The default value of this setting is determined by
5221 /usr/lib/net/links/99-default.link. Old
5222 80-net-name-slot.rules udev configuration file has been
5223 removed, so local configuration overriding this file should
5224 be adapted to override 99-default.link instead.
5226 * When the User= switch is used in a unit file, also
5227 initialize $SHELL= based on the user database entry.
5229 * systemd no longer depends on libdbus. All communication is
5230 now done with sd-bus, systemd's low-level bus library
5233 * kdbus support has been added to PID 1 itself. When kdbus is
5234 enabled, this causes PID 1 to set up the system bus and
5235 enable support for a new ".busname" unit type that
5236 encapsulates bus name activation on kdbus. It works a little
5237 bit like ".socket" units, except for bus names. A new
5238 generator has been added that converts classic dbus1 service
5239 activation files automatically into native systemd .busname
5242 * sd-bus: add a light-weight vtable implementation that allows
5243 defining objects on the bus with a simple static const
5244 vtable array of its methods, signals and properties.
5246 * systemd will not generate or install static dbus
5247 introspection data anymore to /usr/share/dbus-1/interfaces,
5248 as the precise format of these files is unclear, and
5249 nothing makes use of it.
5251 * A proxy daemon is now provided to proxy clients connecting
5252 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5253 compatibility with classic D-Bus.
5255 * A bus driver implementation has been added that supports the
5256 classic D-Bus bus driver calls on kdbus, also for
5257 compatibility purposes.
5259 * A new API "sd-event.h" has been added that implements a
5260 minimal event loop API built around epoll. It provides a
5261 couple of features that direct epoll usage is lacking:
5262 prioritization of events, scales to large numbers of timer
5263 events, per-event timer slack (accuracy), system-wide
5264 coalescing of timer events, exit handlers, watchdog
5265 supervision support using systemd's sd_notify() API, child
5268 * A new API "sd-rntl.h" has been added that provides an API
5269 around the route netlink interface of the kernel, similar in
5270 style to "sd-bus.h".
5272 * A new API "sd-dhcp-client.h" has been added that provides a
5273 small DHCPv4 client-side implementation. This is used by
5276 * There is a new kernel command line option
5277 "systemd.restore_state=0|1". When set to "0", none of the
5278 systemd tools will restore saved runtime state to hardware
5279 devices. More specifically, the rfkill and backlight states
5282 * The FsckPassNo= compatibility option in mount/service units
5283 has been removed. The fstab generator will now add the
5284 necessary dependencies automatically, and does not require
5285 PID1's support for that anymore.
5287 * journalctl gained a new switch, --list-boots, that lists
5288 recent boots with their times and boot IDs.
5290 * The various tools like systemctl, loginctl, timedatectl,
5291 busctl, systemd-run, ... have gained a new switch "-M" to
5292 connect to a specific, local OS container (as direct
5293 connection, without requiring SSH). This works on any
5294 container that is registered with machined, such as those
5295 created by libvirt-lxc or nspawn.
5297 * systemd-run and systemd-analyze also gained support for "-H"
5298 to connect to remote hosts via SSH. This is particularly
5299 useful for systemd-run because it enables queuing of jobs
5300 onto remote systems.
5302 * machinectl gained a new command "login" to open a getty
5303 login in any local container. This works with any container
5304 that is registered with machined (such as those created by
5305 libvirt-lxc or nspawn), and which runs systemd inside.
5307 * machinectl gained a new "reboot" command that may be used to
5308 trigger a reboot on a specific container that is registered
5309 with machined. This works on any container that runs an init
5310 system of some kind.
5312 * systemctl gained a new "list-timers" command to print a nice
5313 listing of installed timer units with the times they elapse
5316 * Alternative reboot() parameters may now be specified on the
5317 "systemctl reboot" command line and are passed to the
5318 reboot() system call.
5320 * systemctl gained a new --job-mode= switch to configure the
5321 mode to queue a job with. This is a more generic version of
5322 --fail, --irreversible, and --ignore-dependencies, which are
5323 still available but not advertised anymore.
5325 * /etc/systemd/system.conf gained new settings to configure
5326 various default timeouts of units, as well as the default
5327 start limit interval and burst. These may still be overridden
5330 * PID1 will now export on the bus profile data of the security
5331 policy upload process (such as the SELinux policy upload to
5334 * journald: when forwarding logs to the console, include
5335 timestamps (following the setting in
5336 /sys/module/printk/parameters/time).
5338 * OnCalendar= in timer units now understands the special
5339 strings "yearly" and "annually". (Both are equivalent)
5341 * The accuracy of timer units is now configurable with the new
5342 AccuracySec= setting. It defaults to 1min.
5344 * A new dependency type JoinsNamespaceOf= has been added that
5345 allows running two services within the same /tmp and network
5346 namespace, if PrivateNetwork= or PrivateTmp= are used.
5348 * A new command "cat" has been added to systemctl. It outputs
5349 the original unit file of a unit, and concatenates the
5350 contents of additional "drop-in" unit file snippets, so that
5351 the full configuration is shown.
5353 * systemctl now supports globbing on the various "list-xyz"
5354 commands, like "list-units" or "list-sockets", as well as on
5355 those commands which take multiple unit names.
5357 * journalctl's --unit= switch gained support for globbing.
5359 * All systemd daemons now make use of the watchdog logic so
5360 that systemd automatically notices when they hang.
5362 * If the $container_ttys environment variable is set,
5363 getty-generator will automatically spawn a getty for each
5364 listed tty. This is useful for container managers to request
5365 login gettys to be spawned on as many ttys as needed.
5367 * %h, %s, %U specifier support is not available anymore when
5368 used in unit files for PID 1. This is because NSS calls are
5369 not safe from PID 1. They stay available for --user
5370 instances of systemd, and as special case for the root user.
5372 * loginctl gained a new "--no-legend" switch to turn off output
5375 * The "sd-login.h" API gained three new calls:
5376 sd_session_is_remote(), sd_session_get_remote_user(),
5377 sd_session_get_remote_host() to query information about
5380 * The udev hardware database now also carries vendor/product
5381 information of SDIO devices.
5383 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5384 determine whether watchdog notifications are requested by
5387 * Socket-activated per-connection services now include a
5388 short description of the connection parameters in the
5391 * tmpfiles gained a new "--boot" option. When this is not used,
5392 only lines where the command character is not suffixed with
5393 "!" are executed. When this option is specified, those
5394 options are executed too. This partitions tmpfiles
5395 directives into those that can be safely executed at any
5396 time, and those which should be run only at boot (for
5397 example, a line that creates /run/nologin).
5399 * A new API "sd-resolve.h" has been added which provides a simple
5400 asynchronous wrapper around glibc NSS host name resolution
5401 calls, such as getaddrinfo(). In contrast to glibc's
5402 getaddrinfo_a(), it does not use signals. In contrast to most
5403 other asynchronous name resolution libraries, this one does
5404 not reimplement DNS, but reuses NSS, so that alternate
5405 host name resolution systems continue to work, such as mDNS,
5406 LDAP, etc. This API is based on libasyncns, but it has been
5407 cleaned up for inclusion in systemd.
5409 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5410 "sd-daemon.h" are no longer found in individual libraries
5411 libsystemd-journal.so, libsystemd-login.so,
5412 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5413 merged them into a single library, libsystemd.so, which
5414 provides all symbols. The reason for this is cyclic
5415 dependencies, as these libraries tend to use each other's
5416 symbols. So far, we have managed to workaround that by linking
5417 a copy of a good part of our code into each of these
5418 libraries again and again, which, however, makes certain
5419 things hard to do, like sharing static variables. Also, it
5420 substantially increases footprint. With this change, there
5421 is only one library for the basic APIs systemd
5422 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5423 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5424 library as well, however are subject to the --enable-kdbus
5425 switch (see below). Note that "sd-dhcp-client.h" is not part
5426 of this library (this is because it only consumes, never
5427 provides, services of/to other APIs). To make the transition
5428 easy from the separate libraries to the unified one, we
5429 provide the --enable-compat-libs compile-time switch which
5430 will generate stub libraries that are compatible with the
5431 old ones but redirect all calls to the new one.
5433 * All of the kdbus logic and the new APIs "sd-bus.h",
5434 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5435 and "sd-utf8.h" are compile-time optional via the
5436 "--enable-kdbus" switch, and they are not compiled in by
5437 default. To make use of kdbus, you have to explicitly enable
5438 the switch. Note however, that neither the kernel nor the
5439 userspace API for all of this is considered stable yet. We
5440 want to maintain the freedom to still change the APIs for
5441 now. By specifying this build-time switch, you acknowledge
5442 that you are aware of the instability of the current
5445 * Also, note that while kdbus is pretty much complete,
5446 it lacks one thing: proper policy support. This means you
5447 can build a fully working system with all features; however,
5448 it will be highly insecure. Policy support will be added in
5449 one of the next releases, at the same time that we will
5450 declare the APIs stable.
5452 * When the kernel command line argument "kdbus" is specified,
5453 systemd will automatically load the kdbus.ko kernel module. At
5454 this stage of development, it is only useful for testing kdbus
5455 and should not be used in production. Note: if "--enable-kdbus"
5456 is specified, and the kdbus.ko kernel module is available, and
5457 "kdbus" is added to the kernel command line, the entire system
5458 runs with kdbus instead of dbus-daemon, with the above mentioned
5459 problem of missing the system policy enforcement. Also a future
5460 version of kdbus.ko or a newer systemd will not be compatible with
5461 each other, and will unlikely be able to boot the machine if only
5462 one of them is updated.
5464 * systemctl gained a new "import-environment" command which
5465 uploads the caller's environment (or parts thereof) into the
5466 service manager so that it is inherited by services started
5467 by the manager. This is useful to upload variables like
5468 $DISPLAY into the user service manager.
5470 * A new PrivateDevices= switch has been added to service units
5471 which allows running a service with a namespaced /dev
5472 directory that does not contain any device nodes for
5473 physical devices. More specifically, it only includes devices
5474 such as /dev/null, /dev/urandom, and /dev/zero which are API
5477 * logind has been extended to support behaviour like VT
5478 switching on seats that do not support a VT. This makes
5479 multi-session available on seats that are not the first seat
5480 (seat0), and on systems where kernel support for VTs has
5481 been disabled at compile-time.
5483 * If a process holds a delay lock for system sleep or shutdown
5484 and fails to release it in time, we will now log its
5485 identity. This makes it easier to identify processes that
5486 cause slow suspends or power-offs.
5488 * When parsing /etc/crypttab, support for a new key-slot=
5489 option as supported by Debian is added. It allows indicating
5490 which LUKS slot to use on disk, speeding up key loading.
5492 * The sd_journal_sendv() API call has been checked and
5493 officially declared to be async-signal-safe so that it may
5494 be invoked from signal handlers for logging purposes.
5496 * Boot-time status output is now enabled automatically after a
5497 short timeout if boot does not progress, in order to give
5498 the user an indication what she or he is waiting for.
5500 * The boot-time output has been improved to show how much time
5501 remains until jobs expire.
5503 * The KillMode= switch in service units gained a new possible
5504 value "mixed". If set, and the unit is shut down, then the
5505 initial SIGTERM signal is sent only to the main daemon
5506 process, while the following SIGKILL signal is sent to
5507 all remaining processes of the service.
5509 * When a scope unit is registered, a new property "Controller"
5510 may be set. If set to a valid bus name, systemd will send a
5511 RequestStop() signal to this name when it would like to shut
5512 down the scope. This may be used to hook manager logic into
5513 the shutdown logic of scope units. Also, scope units may now
5514 be put in a special "abandoned" state, in which case the
5515 manager process which created them takes no further
5516 responsibilities for it.
5518 * When reading unit files, systemd will now verify
5519 the access mode of these files, and warn about certain
5520 suspicious combinations. This has been added to make it
5521 easier to track down packaging bugs where unit files are
5522 marked executable or world-writable.
5524 * systemd-nspawn gained a new "--setenv=" switch to set
5525 container-wide environment variables. The similar option in
5526 systemd-activate was renamed from "--environment=" to
5527 "--setenv=" for consistency.
5529 * systemd-nspawn has been updated to create a new kdbus domain
5530 for each container that is invoked, thus allowing each
5531 container to have its own set of system and user buses,
5532 independent of the host.
5534 * systemd-nspawn gained a new --drop-capability= switch to run
5535 the container with less capabilities than the default. Both
5536 --drop-capability= and --capability= now take the special
5537 string "all" for dropping or keeping all capabilities.
5539 * systemd-nspawn gained new switches for executing containers
5540 with specific SELinux labels set.
5542 * systemd-nspawn gained a new --quiet switch to not generate
5543 any additional output but the container's own console
5546 * systemd-nspawn gained a new --share-system switch to run a
5547 container without PID namespacing enabled.
5549 * systemd-nspawn gained a new --register= switch to control
5550 whether the container is registered with systemd-machined or
5551 not. This is useful for containers that do not run full
5552 OS images, but only specific apps.
5554 * systemd-nspawn gained a new --keep-unit which may be used
5555 when invoked as the only program from a service unit, and
5556 results in registration of the unit service itself in
5557 systemd-machined, instead of a newly opened scope unit.
5559 * systemd-nspawn gained a new --network-interface= switch for
5560 moving arbitrary interfaces to the container. The new
5561 --network-veth switch creates a virtual Ethernet connection
5562 between host and container. The new --network-bridge=
5563 switch then allows assigning the host side of this virtual
5564 Ethernet connection to a bridge device.
5566 * systemd-nspawn gained a new --personality= switch for
5567 setting the kernel personality for the container. This is
5568 useful when running a 32-bit container on a 64-bit host. A
5569 similar option Personality= is now also available for service
5572 * logind will now also track a "Desktop" identifier for each
5573 session which encodes the desktop environment of it. This is
5574 useful for desktop environments that want to identify
5575 multiple running sessions of itself easily.
5577 * A new SELinuxContext= setting for service units has been
5578 added that allows setting a specific SELinux execution
5579 context for a service.
5581 * Most systemd client tools will now honour $SYSTEMD_LESS for
5582 settings of the "less" pager. By default, these tools will
5583 override $LESS to allow certain operations to work, such as
5584 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5585 influence this logic.
5587 * systemd's "seccomp" hook-up has been changed to make use of
5588 the libseccomp library instead of using its own
5589 implementation. This has benefits for portability among
5592 * For usage together with SystemCallFilter=, a new
5593 SystemCallErrorNumber= setting has been introduced that
5594 allows configuration of a system error number to be returned
5595 on filtered system calls, instead of immediately killing the
5596 process. Also, SystemCallArchitectures= has been added to
5597 limit access to system calls of a particular architecture
5598 (in order to turn off support for unused secondary
5599 architectures). There is also a global
5600 SystemCallArchitectures= setting in system.conf now to turn
5601 off support for non-native system calls system-wide.
5603 * systemd requires a kernel with a working name_to_handle_at(),
5604 please see the kernel config requirements in the README file.
5606 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5607 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5608 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5609 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5610 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5611 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5612 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5613 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5614 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5615 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5616 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5617 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5618 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5619 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5620 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5621 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5622 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5623 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5624 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5625 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5626 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5627 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5628 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5629 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5631 — Berlin, 2014-02-20
5635 * logind has gained support for facilitating privileged input
5636 and drm device access for unprivileged clients. This work is
5637 useful to allow Wayland display servers (and similar
5638 programs, such as kmscon) to run under the user's ID and
5639 access input and drm devices which are normally
5640 protected. When this is used (and the kernel is new enough)
5641 logind will "mute" IO on the file descriptors passed to
5642 Wayland as long as it is in the background and "unmute" it
5643 if it returns into the foreground. This allows secure
5644 session switching without allowing background sessions to
5645 eavesdrop on input and display data. This also introduces
5646 session switching support if VT support is turned off in the
5647 kernel, and on seats that are not seat0.
5649 * A new kernel command line option luks.options= is understood
5650 now which allows specifying LUKS options for usage for LUKS
5651 encrypted partitions specified with luks.uuid=.
5653 * tmpfiles.d(5) snippets may now use specifier expansion in
5654 path names. More specifically %m, %b, %H, %v, are now
5655 replaced by the local machine id, boot id, hostname, and
5656 kernel version number.
5658 * A new tmpfiles.d(5) command "m" has been introduced which
5659 may be used to change the owner/group/access mode of a file
5660 or directory if it exists, but do nothing if it does not.
5662 * This release removes high-level support for the
5663 MemorySoftLimit= cgroup setting. The underlying kernel
5664 cgroup attribute memory.soft_limit= is currently badly
5665 designed and likely to be removed from the kernel API in its
5666 current form, hence we should not expose it for now.
5668 * The memory.use_hierarchy cgroup attribute is now enabled for
5669 all cgroups systemd creates in the memory cgroup
5670 hierarchy. This option is likely to be come the built-in
5671 default in the kernel anyway, and the non-hierarchical mode
5672 never made much sense in the intrinsically hierarchical
5675 * A new field _SYSTEMD_SLICE= is logged along with all journal
5676 messages containing the slice a message was generated
5677 from. This is useful to allow easy per-customer filtering of
5678 logs among other things.
5680 * systemd-journald will no longer adjust the group of journal
5681 files it creates to the "systemd-journal" group. Instead we
5682 rely on the journal directory to be owned by the
5683 "systemd-journal" group, and its setgid bit set, so that the
5684 kernel file system layer will automatically enforce that
5685 journal files inherit this group assignment. The reason for
5686 this change is that we cannot allow NSS look-ups from
5687 journald which would be necessary to resolve
5688 "systemd-journal" to a numeric GID, because this might
5689 create deadlocks if NSS involves synchronous queries to
5690 other daemons (such as nscd, or sssd) which in turn are
5691 logging clients of journald and might block on it, which
5692 would then dead lock. A tmpfiles.d(5) snippet included in
5693 systemd will make sure the setgid bit and group are
5694 properly set on the journal directory if it exists on every
5695 boot. However, we recommend adjusting it manually after
5696 upgrades too (or from RPM scriptlets), so that the change is
5697 not delayed until next reboot.
5699 * Backlight and random seed files in /var/lib/ have moved into
5700 the /var/lib/systemd/ directory, in order to centralize all
5701 systemd generated files in one directory.
5703 * Boot time performance measurements (as displayed by
5704 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5705 performance information if that's available to determine how
5706 much time BIOS and boot loader initialization required. With
5707 a sufficiently new BIOS you hence no longer need to boot
5708 with Gummiboot to get access to such information.
5710 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5711 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5712 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5713 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5714 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5715 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5716 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5718 — Berlin, 2013-10-02
5722 * The Restart= option for services now understands a new
5723 on-watchdog setting, which will restart the service
5724 automatically if the service stops sending out watchdog keep
5725 alive messages (as configured with WatchdogSec=).
5727 * The getty generator (which is responsible for bringing up a
5728 getty on configured serial consoles) will no longer only
5729 start a getty on the primary kernel console but on all
5730 others, too. This makes the order in which console= is
5731 specified on the kernel command line less important.
5733 * libsystemd-logind gained a new sd_session_get_vt() call to
5734 retrieve the VT number of a session.
5736 * If the option "tries=0" is set for an entry of /etc/crypttab
5737 its passphrase is queried indefinitely instead of any
5738 maximum number of tries.
5740 * If a service with a configure PID file terminates its PID
5741 file will now be removed automatically if it still exists
5742 afterwards. This should put an end to stale PID files.
5744 * systemd-run will now also take relative binary path names
5745 for execution and no longer insists on absolute paths.
5747 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5748 paths that are optionally prefixed with "-" to indicate that
5749 it should not be considered a failure if they do not exist.
5751 * journalctl -o (and similar commands) now understands a new
5752 output mode "short-precise", it is similar to "short" but
5753 shows timestamps with usec accuracy.
5755 * The option "discard" (as known from Debian) is now
5756 synonymous to "allow-discards" in /etc/crypttab. In fact,
5757 "discard" is preferred now (since it is easier to remember
5760 * Some licensing clean-ups were made, so that more code is now
5761 LGPL-2.1 licensed than before.
5763 * A minimal tool to save/restore the display backlight
5764 brightness across reboots has been added. It will store the
5765 backlight setting as late as possible at shutdown, and
5766 restore it as early as possible during reboot.
5768 * A logic to automatically discover and enable home and swap
5769 partitions on GPT disks has been added. With this in place
5770 /etc/fstab becomes optional for many setups as systemd can
5771 discover certain partitions located on the root disk
5772 automatically. Home partitions are recognized under their
5773 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5774 partitions are recognized under their GPT type ID
5775 0657fd6da4ab43c484e50933c84b4f4f.
5777 * systemd will no longer pass any environment from the kernel
5778 or initrd to system services. If you want to set an
5779 environment for all services, do so via the kernel command
5780 line systemd.setenv= assignment.
5782 * The systemd-sysctl tool no longer natively reads the file
5783 /etc/sysctl.conf. If desired, the file should be symlinked
5784 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5785 legacy support by a symlink rather than built-in code, it
5786 also makes the otherwise hidden order of application of the
5787 different files visible. (Note that this partly reverts to a
5788 pre-198 application order of sysctl knobs!)
5790 * The "systemctl set-log-level" and "systemctl dump" commands
5791 have been moved to systemd-analyze.
5793 * systemd-run learned the new --remain-after-exit switch,
5794 which causes the scope unit not to be cleaned up
5795 automatically after the process terminated.
5797 * tmpfiles learned a new --exclude-prefix= switch to exclude
5798 certain paths from operation.
5800 * journald will now automatically flush all messages to disk
5801 as soon as a message at the log level CRIT, ALERT or EMERG
5804 Contributions from: Andrew Cook, Brandon Philips, Christian
5805 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5806 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5807 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5808 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5809 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5810 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5811 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5812 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5813 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5814 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5815 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5816 William Giokas, Zbigniew Jędrzejewski-Szmek
5818 — Berlin, 2013-09-13
5822 * The documentation has been updated to cover the various new
5823 concepts introduced with 205.
5825 * Unit files now understand the new %v specifier which
5826 resolves to the kernel version string as returned by "uname
5829 * systemctl now supports filtering the unit list output by
5830 load state, active state and sub state, using the new
5833 * "systemctl status" will now show the results of the
5834 condition checks (like ConditionPathExists= and similar) of
5835 the last start attempts of the unit. They are also logged to
5838 * "journalctl -b" may now be used to look for boot output of a
5839 specific boot. Try "journalctl -b -1" for the previous boot,
5840 but the syntax is substantially more powerful.
5842 * "journalctl --show-cursor" has been added which prints the
5843 cursor string the last shown log line. This may then be used
5844 with the new "journalctl --after-cursor=" switch to continue
5845 browsing logs from that point on.
5847 * "journalctl --force" may now be used to force regeneration
5850 * Creation of "dead" device nodes has been moved from udev
5851 into kmod and tmpfiles. Previously, udev would read the kmod
5852 databases to pre-generate dead device nodes based on meta
5853 information contained in kernel modules, so that these would
5854 be auto-loaded on access rather then at boot. As this
5855 does not really have much to do with the exposing actual
5856 kernel devices to userspace this has always been slightly
5857 alien in the udev codebase. Following the new scheme kmod
5858 will now generate a runtime snippet for tmpfiles from the
5859 module meta information and it now is tmpfiles' job to the
5860 create the nodes. This also allows overriding access and
5861 other parameters for the nodes using the usual tmpfiles
5862 facilities. As side effect this allows us to remove the
5863 CAP_SYS_MKNOD capability bit from udevd entirely.
5865 * logind's device ACLs may now be applied to these "dead"
5866 devices nodes too, thus finally allowing managed access to
5867 devices such as /dev/snd/sequencer without loading the
5868 backing module right-away.
5870 * A new RPM macro has been added that may be used to apply
5871 tmpfiles configuration during package installation.
5873 * systemd-detect-virt and ConditionVirtualization= now can
5874 detect User-Mode-Linux machines (UML).
5876 * journald will now implicitly log the effective capabilities
5877 set of processes in the message metadata.
5879 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5881 * The initrd interface has been simplified (more specifically,
5882 support for passing performance data via environment
5883 variables and fsck results via files in /run has been
5884 removed). These features were non-essential, and are
5885 nowadays available in a much nicer way by having systemd in
5886 the initrd serialize its state and have the hosts systemd
5887 deserialize it again.
5889 * The udev "keymap" data files and tools to apply keyboard
5890 specific mappings of scan to key codes, and force-release
5891 scan code lists have been entirely replaced by a udev
5892 "keyboard" builtin and a hwdb data file.
5894 * systemd will now honour the kernel's "quiet" command line
5895 argument also during late shutdown, resulting in a
5896 completely silent shutdown when used.
5898 * There's now an option to control the SO_REUSEPORT socket
5899 option in .socket units.
5901 * Instance units will now automatically get a per-template
5902 subslice of system.slice unless something else is explicitly
5903 configured. For example, instances of sshd@.service will now
5904 implicitly be placed in system-sshd.slice rather than
5905 system.slice as before.
5907 * Test coverage support may now be enabled at build time.
5909 Contributions from: Dave Reisner, Frederic Crozat, Harald
5910 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5911 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5912 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5913 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5914 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5915 Giokas, Zbigniew Jędrzejewski-Szmek
5917 — Berlin, 2013-07-23
5921 * Two new unit types have been introduced:
5923 Scope units are very similar to service units, however, are
5924 created out of pre-existing processes — instead of PID 1
5925 forking off the processes. By using scope units it is
5926 possible for system services and applications to group their
5927 own child processes (worker processes) in a powerful way
5928 which then maybe used to organize them, or kill them
5929 together, or apply resource limits on them.
5931 Slice units may be used to partition system resources in an
5932 hierarchical fashion and then assign other units to them. By
5933 default there are now three slices: system.slice (for all
5934 system services), user.slice (for all user sessions),
5935 machine.slice (for VMs and containers).
5937 Slices and scopes have been introduced primarily in
5938 context of the work to move cgroup handling to a
5939 single-writer scheme, where only PID 1
5940 creates/removes/manages cgroups.
5942 * There's a new concept of "transient" units. In contrast to
5943 normal units these units are created via an API at runtime,
5944 not from configuration from disk. More specifically this
5945 means it is now possible to run arbitrary programs as
5946 independent services, with all execution parameters passed
5947 in via bus APIs rather than read from disk. Transient units
5948 make systemd substantially more dynamic then it ever was,
5949 and useful as a general batch manager.
5951 * logind has been updated to make use of scope and slice units
5952 for managing user sessions. As a user logs in he will get
5953 his own private slice unit, to which all sessions are added
5954 as scope units. We also added support for automatically
5955 adding an instance of user@.service for the user into the
5956 slice. Effectively logind will no longer create cgroup
5957 hierarchies on its own now, it will defer entirely to PID 1
5958 for this by means of scope, service and slice units. Since
5959 user sessions this way become entities managed by PID 1
5960 the output of "systemctl" is now a lot more comprehensive.
5962 * A new mini-daemon "systemd-machined" has been added which
5963 may be used by virtualization managers to register local
5964 VMs/containers. nspawn has been updated accordingly, and
5965 libvirt will be updated shortly. machined will collect a bit
5966 of meta information about the VMs/containers, and assign
5967 them their own scope unit (see above). The collected
5968 meta-data is then made available via the "machinectl" tool,
5969 and exposed in "ps" and similar tools. machined/machinectl
5970 is compile-time optional.
5972 * As discussed earlier, the low-level cgroup configuration
5973 options ControlGroup=, ControlGroupModify=,
5974 ControlGroupPersistent=, ControlGroupAttribute= have been
5975 removed. Please use high-level attribute settings instead as
5976 well as slice units.
5978 * A new bus call SetUnitProperties() has been added to alter
5979 various runtime parameters of a unit. This is primarily
5980 useful to alter cgroup parameters dynamically in a nice way,
5981 but will be extended later on to make more properties
5982 modifiable at runtime. systemctl gained a new set-properties
5983 command that wraps this call.
5985 * A new tool "systemd-run" has been added which can be used to
5986 run arbitrary command lines as transient services or scopes,
5987 while configuring a number of settings via the command
5988 line. This tool is currently very basic, however already
5989 very useful. We plan to extend this tool to even allow
5990 queuing of execution jobs with time triggers from the
5991 command line, similar in fashion to "at".
5993 * nspawn will now inform the user explicitly that kernels with
5994 audit enabled break containers, and suggest the user to turn
5997 * Support for detecting the IMA and AppArmor security
5998 frameworks with ConditionSecurity= has been added.
6000 * journalctl gained a new "-k" switch for showing only kernel
6001 messages, mimicking dmesg output; in addition to "--user"
6002 and "--system" switches for showing only user's own logs
6005 * systemd-delta can now show information about drop-in
6006 snippets extending unit files.
6008 * libsystemd-bus has been substantially updated but is still
6009 not available as public API.
6011 * systemd will now look for the "debug" argument on the kernel
6012 command line and enable debug logging, similar to what
6013 "systemd.log_level=debug" already did before.
6015 * "systemctl set-default", "systemctl get-default" has been
6016 added to configure the default.target symlink, which
6017 controls what to boot into by default.
6019 * "systemctl set-log-level" has been added as a convenient
6020 way to raise and lower systemd logging threshold.
6022 * "systemd-analyze plot" will now show the time the various
6023 generators needed for execution, as well as information
6024 about the unit file loading.
6026 * libsystemd-journal gained a new sd_journal_open_files() call
6027 for opening specific journal files. journactl also gained a
6028 new switch to expose this new functionality. Previously we
6029 only supported opening all files from a directory, or all
6030 files from the system, as opening individual files only is
6031 racy due to journal file rotation.
6033 * systemd gained the new DefaultEnvironment= setting in
6034 /etc/systemd/system.conf to set environment variables for
6037 * If a privileged process logs a journal message with the
6038 OBJECT_PID= field set, then journald will automatically
6039 augment this with additional OBJECT_UID=, OBJECT_GID=,
6040 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6041 system services want to log events about specific client
6042 processes. journactl/systemctl has been updated to make use
6043 of this information if all log messages regarding a specific
6046 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6047 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6048 Reisner, David Coppa, David King, David Strauss, Eelco
6049 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6050 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6051 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6052 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6053 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6054 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6055 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6056 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6057 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6058 Łukasz Stelmach, 장동준
6062 * The Python bindings gained some minimal support for the APIs
6063 exposed by libsystemd-logind.
6065 * ConditionSecurity= gained support for detecting SMACK. Since
6066 this condition already supports SELinux and AppArmor we only
6067 miss IMA for this. Patches welcome!
6069 Contributions from: Karol Lewandowski, Lennart Poettering,
6070 Zbigniew Jędrzejewski-Szmek
6074 * systemd-nspawn will now create /etc/resolv.conf if
6075 necessary, before bind-mounting the host's file onto it.
6077 * systemd-nspawn will now store meta information about a
6078 container on the container's cgroup as extended attribute
6079 fields, including the root directory.
6081 * The cgroup hierarchy has been reworked in many ways. All
6082 objects any of the components systemd creates in the cgroup
6083 tree are now suffixed. More specifically, user sessions are
6084 now placed in cgroups suffixed with ".session", users in
6085 cgroups suffixed with ".user", and nspawn containers in
6086 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6087 names are now escaped in a simple scheme to avoid collision
6088 of userspace object names with kernel filenames. This work
6089 is preparation for making these objects relocatable in the
6090 cgroup tree, in order to allow easy resource partitioning of
6091 these objects without causing naming conflicts.
6093 * systemctl list-dependencies gained the new switches
6094 --plain, --reverse, --after and --before.
6096 * systemd-inhibit now shows the process name of processes that
6097 have taken an inhibitor lock.
6099 * nss-myhostname will now also resolve "localhost"
6100 implicitly. This makes /etc/hosts an optional file and
6101 nicely handles that on IPv6 ::1 maps to both "localhost" and
6104 * libsystemd-logind.so gained a new call
6105 sd_get_machine_names() to enumerate running containers and
6106 VMs (currently only supported by very new libvirt and
6107 nspawn). sd_login_monitor can now be used to watch
6108 VMs/containers coming and going.
6110 * .include is not allowed recursively anymore, and only in
6111 unit files. Usually it is better to use drop-in snippets in
6112 .d/*.conf anyway, as introduced with systemd 198.
6114 * systemd-analyze gained a new "critical-chain" command that
6115 determines the slowest chain of units run during system
6116 boot-up. It is very useful for tracking down where
6117 optimizing boot time is the most beneficial.
6119 * systemd will no longer allow manipulating service paths in
6120 the name=systemd:/system cgroup tree using ControlGroup= in
6121 units. (But is still fine with it in all other dirs.)
6123 * There's a new systemd-nspawn@.service service file that may
6124 be used to easily run nspawn containers as system
6125 services. With the container's root directory in
6126 /var/lib/container/foobar it is now sufficient to run
6127 "systemctl start systemd-nspawn@foobar.service" to boot it.
6129 * systemd-cgls gained a new parameter "--machine" to list only
6130 the processes within a certain container.
6132 * ConditionSecurity= now can check for "apparmor". We still
6133 are lacking checks for SMACK and IMA for this condition
6134 check though. Patches welcome!
6136 * A new configuration file /etc/systemd/sleep.conf has been
6137 added that may be used to configure which kernel operation
6138 systemd is supposed to execute when "suspend", "hibernate"
6139 or "hybrid-sleep" is requested. This makes the new kernel
6140 "freeze" state accessible to the user.
6142 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6143 the passed argument if applicable.
6145 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6146 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6147 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6148 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6149 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6150 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6151 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6156 * The output of 'systemctl list-jobs' got some polishing. The
6157 '--type=' argument may now be passed more than once. A new
6158 command 'systemctl list-sockets' has been added which shows
6159 a list of kernel sockets systemd is listening on with the
6160 socket units they belong to, plus the units these socket
6163 * The experimental libsystemd-bus library got substantial
6164 updates to work in conjunction with the (also experimental)
6165 kdbus kernel project. It works well enough to exchange
6166 messages with some sophistication. Note that kdbus is not
6167 ready yet, and the library is mostly an elaborate test case
6168 for now, and not installable.
6170 * systemd gained a new unit 'systemd-static-nodes.service'
6171 that generates static device nodes earlier during boot, and
6172 can run in conjunction with udev.
6174 * libsystemd-login gained a new call sd_pid_get_user_unit()
6175 to retrieve the user systemd unit a process is running
6176 in. This is useful for systems where systemd is used as
6179 * systemd-nspawn now places all containers in the new /machine
6180 top-level cgroup directory in the name=systemd
6181 hierarchy. libvirt will soon do the same, so that we get a
6182 uniform separation of /system, /user and /machine for system
6183 services, user processes and containers/virtual
6184 machines. This new cgroup hierarchy is also useful to stick
6185 stable names to specific container instances, which can be
6186 recognized later this way (this name may be controlled
6187 via systemd-nspawn's new -M switch). libsystemd-login also
6188 gained a new call sd_pid_get_machine_name() to retrieve the
6189 name of the container/VM a specific process belongs to.
6191 * bootchart can now store its data in the journal.
6193 * libsystemd-journal gained a new call
6194 sd_journal_add_conjunction() for AND expressions to the
6195 matching logic. This can be used to express more complex
6196 logical expressions.
6198 * journactl can now take multiple --unit= and --user-unit=
6201 * The cryptsetup logic now understands the "luks.key=" kernel
6202 command line switch for specifying a file to read the
6203 decryption key from. Also, if a configured key file is not
6204 found the tool will now automatically fall back to prompting
6207 * Python systemd.journal module was updated to wrap recently
6208 added functions from libsystemd-journal. The interface was
6209 changed to bring the low level interface in s.j._Reader
6210 closer to the C API, and the high level interface in
6211 s.j.Reader was updated to wrap and convert all data about
6214 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6215 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6216 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6217 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6218 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6219 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6223 * journalctl --update-catalog now understands a new --root=
6224 option to operate on catalogs found in a different root
6227 * During shutdown after systemd has terminated all running
6228 services a final killing loop kills all remaining left-over
6229 processes. We will now print the name of these processes
6230 when we send SIGKILL to them, since this usually indicates a
6233 * If /etc/crypttab refers to password files stored on
6234 configured mount points automatic dependencies will now be
6235 generated to ensure the specific mount is established first
6236 before the key file is attempted to be read.
6238 * 'systemctl status' will now show information about the
6239 network sockets a socket unit is listening on.
6241 * 'systemctl status' will also shown information about any
6242 drop-in configuration file for units. (Drop-In configuration
6243 files in this context are files such as
6244 /etc/systemd/systemd/foobar.service.d/*.conf)
6246 * systemd-cgtop now optionally shows summed up CPU times of
6247 cgroups. Press '%' while running cgtop to switch between
6248 percentage and absolute mode. This is useful to determine
6249 which cgroups use up the most CPU time over the entire
6250 runtime of the system. systemd-cgtop has also been updated
6251 to be 'pipeable' for processing with further shell tools.
6253 * 'hostnamectl set-hostname' will now allow setting of FQDN
6256 * The formatting and parsing of time span values has been
6257 changed. The parser now understands fractional expressions
6258 such as "5.5h". The formatter will now output fractional
6259 expressions for all time spans under 1min, i.e. "5.123456s"
6260 rather than "5s 123ms 456us". For time spans under 1s
6261 millisecond values are shown, for those under 1ms
6262 microsecond values are shown. This should greatly improve
6263 all time-related output of systemd.
6265 * libsystemd-login and libsystemd-journal gained new
6266 functions for querying the poll() events mask and poll()
6267 timeout value for integration into arbitrary event
6270 * localectl gained the ability to list available X11 keymaps
6271 (models, layouts, variants, options).
6273 * 'systemd-analyze dot' gained the ability to filter for
6274 specific units via shell-style globs, to create smaller,
6275 more useful graphs. I.e. it is now possible to create simple
6276 graphs of all the dependencies between only target units, or
6277 of all units that Avahi has dependencies with.
6279 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6280 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6281 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6282 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6283 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6284 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6285 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6289 * The boot-time readahead implementation for rotating media
6290 will now read the read-ahead data in multiple passes which
6291 consist of all read requests made in equidistant time
6292 intervals. This means instead of strictly reading read-ahead
6293 data in its physical order on disk we now try to find a
6294 middle ground between physical and access time order.
6296 * /etc/os-release files gained a new BUILD_ID= field for usage
6297 on operating systems that provide continuous builds of OS
6300 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6301 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6302 William Douglas, Zbigniew Jędrzejewski-Szmek
6306 * systemd-python gained an API exposing libsystemd-daemon.
6308 * The SMACK setup logic gained support for uploading CIPSO
6311 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6312 ReadOnlyDirectories= and InaccessibleDirectories= has
6313 changed. The private /tmp and /var/tmp directories are now
6314 shared by all processes of a service (which means
6315 ExecStartPre= may now leave data in /tmp that ExecStart= of
6316 the same service can still access). When a service is
6317 stopped its temporary directories are immediately deleted
6318 (normal clean-up with tmpfiles is still done in addition to
6321 * By default, systemd will now set a couple of sysctl
6322 variables in the kernel: the safe sysrq options are turned
6323 on, IP route verification is turned on, and source routing
6324 disabled. The recently added hardlink and softlink
6325 protection of the kernel is turned on. These settings should
6326 be reasonably safe, and good defaults for all new systems.
6328 * The predictable network naming logic may now be turned off
6329 with a new kernel command line switch: net.ifnames=0.
6331 * A new libsystemd-bus module has been added that implements a
6332 pretty complete D-Bus client library. For details see:
6334 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6336 * journald will now explicitly flush the journal files to disk
6337 at the latest 5min after each write. The file will then also
6338 be marked offline until the next write. This should increase
6339 reliability in case of a crash. The synchronization delay
6340 can be configured via SyncIntervalSec= in journald.conf.
6342 * There's a new remote-fs-setup.target unit that can be used
6343 to pull in specific services when at least one remote file
6344 system is to be mounted.
6346 * There are new targets timers.target and paths.target as
6347 canonical targets to pull user timer and path units in
6348 from. This complements sockets.target with a similar
6349 purpose for socket units.
6351 * libudev gained a new call udev_device_set_attribute_value()
6352 to set sysfs attributes of a device.
6354 * The udev daemon now sets the default number of worker
6355 processes executed in parallel based on the number of available
6356 CPUs instead of the amount of available RAM. This is supposed
6357 to provide a more reliable default and limit a too aggressive
6358 parallelism for setups with 1000s of devices connected.
6360 Contributions from: Auke Kok, Colin Walters, Cristian
6361 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6362 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6363 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6364 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6365 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6366 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6367 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6368 Zbigniew Jędrzejewski-Szmek
6372 * Configuration of unit files may now be extended via drop-in
6373 files without having to edit/override the unit files
6374 themselves. More specifically, if the administrator wants to
6375 change one value for a service file foobar.service he can
6376 now do so by dropping in a configuration snippet into
6377 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6378 will load all these snippets and apply them on top of the
6379 main unit configuration file, possibly extending or
6380 overriding its settings. Using these drop-in snippets is
6381 generally nicer than the two earlier options for changing
6382 unit files locally: copying the files from
6383 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6384 them there; or creating a new file in /etc/systemd/system/
6385 that incorporates the original one via ".include". Drop-in
6386 snippets into these .d/ directories can be placed in any
6387 directory systemd looks for units in, and the usual
6388 overriding semantics between /usr/lib, /etc and /run apply
6391 * Most unit file settings which take lists of items can now be
6392 reset by assigning the empty string to them. For example,
6393 normally, settings such as Environment=FOO=BAR append a new
6394 environment variable assignment to the environment block,
6395 each time they are used. By assigning Environment= the empty
6396 string the environment block can be reset to empty. This is
6397 particularly useful with the .d/*.conf drop-in snippets
6398 mentioned above, since this adds the ability to reset list
6399 settings from vendor unit files via these drop-ins.
6401 * systemctl gained a new "list-dependencies" command for
6402 listing the dependencies of a unit recursively.
6404 * Inhibitors are now honored and listed by "systemctl
6405 suspend", "systemctl poweroff" (and similar) too, not only
6406 GNOME. These commands will also list active sessions by
6409 * Resource limits (as exposed by the various control group
6410 controllers) can now be controlled dynamically at runtime
6411 for all units. More specifically, you can now use a command
6412 like "systemctl set-cgroup-attr foobar.service cpu.shares
6413 2000" to alter the CPU shares a specific service gets. These
6414 settings are stored persistently on disk, and thus allow the
6415 administrator to easily adjust the resource usage of
6416 services with a few simple commands. This dynamic resource
6417 management logic is also available to other programs via the
6418 bus. Almost any kernel cgroup attribute and controller is
6421 * systemd-vconsole-setup will now copy all font settings to
6422 all allocated VTs, where it previously applied them only to
6425 * libsystemd-login gained the new sd_session_get_tty() API
6428 * This release drops support for a few legacy or
6429 distribution-specific LSB facility names when parsing init
6430 scripts: $x-display-manager, $mail-transfer-agent,
6431 $mail-transport-agent, $mail-transfer-agent, $smtp,
6432 $null. Also, the mail-transfer-agent.target unit backing
6433 this has been removed. Distributions which want to retain
6434 compatibility with this should carry the burden for
6435 supporting this themselves and patch support for these back
6436 in, if they really need to. Also, the facilities $syslog and
6437 $local_fs are now ignored, since systemd does not support
6438 early-boot LSB init scripts anymore, and these facilities
6439 are implied anyway for normal services. syslog.target has
6442 * There are new bus calls on PID1's Manager object for
6443 cancelling jobs, and removing snapshot units. Previously,
6444 both calls were only available on the Job and Snapshot
6447 * systemd-journal-gatewayd gained SSL support.
6449 * The various "environment" files, such as /etc/locale.conf
6450 now support continuation lines with a backslash ("\") as
6451 last character in the line, similarly in style (but different)
6452 to how this is supported in shells.
6454 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6455 now implicitly appended to every log entry logged. systemctl
6456 has been updated to filter by this field when operating on a
6457 user systemd instance.
6459 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6460 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6461 the container. This makes it easier to boot unmodified
6462 Fedora systems in a container, which however still requires
6463 audit=0 to be passed on the kernel command line. Auditing in
6464 kernel and userspace is unfortunately still too broken in
6465 context of containers, hence we recommend compiling it out
6466 of the kernel or using audit=0. Hopefully this will be fixed
6467 one day for good in the kernel.
6469 * nspawn gained the new --bind= and --bind-ro= parameters to
6470 bind mount specific directories from the host into the
6473 * nspawn will now mount its own devpts file system instance
6474 into the container, in order not to leak pty devices from
6475 the host into the container.
6477 * systemd will now read the firmware boot time performance
6478 information from the EFI variables, if the used boot loader
6479 supports this, and takes it into account for boot performance
6480 analysis via "systemd-analyze". This is currently supported
6481 only in conjunction with Gummiboot, but could be supported
6482 by other boot loaders too. For details see:
6484 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6486 * A new generator has been added that automatically mounts the
6487 EFI System Partition (ESP) to /boot, if that directory
6488 exists, is empty, and no other file system has been
6489 configured to be mounted there.
6491 * logind will now send out PrepareForSleep(false) out
6492 unconditionally, after coming back from suspend. This may be
6493 used by applications as asynchronous notification for
6494 system resume events.
6496 * "systemctl unlock-sessions" has been added, that allows
6497 unlocking the screens of all user sessions at once, similar
6498 to how "systemctl lock-sessions" already locked all users
6499 sessions. This is backed by a new D-Bus call UnlockSessions().
6501 * "loginctl seat-status" will now show the master device of a
6502 seat. (i.e. the device of a seat that needs to be around for
6503 the seat to be considered available, usually the graphics
6506 * tmpfiles gained a new "X" line type, that allows
6507 configuration of files and directories (with wildcards) that
6508 shall be excluded from automatic cleanup ("aging").
6510 * udev default rules set the device node permissions now only
6511 at "add" events, and do not change them any longer with a
6512 later "change" event.
6514 * The log messages for lid events and power/sleep keypresses
6515 now carry a message ID.
6517 * We now have a substantially larger unit test suite, but this
6518 continues to be work in progress.
6520 * udevadm hwdb gained a new --root= parameter to change the
6521 root directory to operate relative to.
6523 * logind will now issue a background sync() request to the kernel
6524 early at shutdown, so that dirty buffers are flushed to disk early
6525 instead of at the last moment, in order to optimize shutdown
6528 * A new bootctl tool has been added that is an interface for
6529 certain boot loader operations. This is currently a preview
6530 and is likely to be extended into a small mechanism daemon
6531 like timedated, localed, hostnamed, and can be used by
6532 graphical UIs to enumerate available boot options, and
6533 request boot into firmware operations.
6535 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6536 the rest of the package. It also has been updated to work
6537 correctly in initrds.
6539 * Policykit previously has been runtime optional, and is now
6540 also compile time optional via a configure switch.
6542 * systemd-analyze has been reimplemented in C. Also "systemctl
6543 dot" has moved into systemd-analyze.
6545 * "systemctl status" with no further parameters will now print
6546 the status of all active or failed units.
6548 * Operations such as "systemctl start" can now be executed
6549 with a new mode "--irreversible" which may be used to queue
6550 operations that cannot accidentally be reversed by a later
6551 job queuing. This is by default used to make shutdown
6552 requests more robust.
6554 * The Python API of systemd now gained a new module for
6555 reading journal files.
6557 * A new tool kernel-install has been added that can install
6558 kernel images according to the Boot Loader Specification:
6560 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6562 * Boot time console output has been improved to provide
6563 animated boot time output for hanging jobs.
6565 * A new tool systemd-activate has been added which can be used
6566 to test socket activation with, directly from the command
6567 line. This should make it much easier to test and debug
6568 socket activation in daemons.
6570 * journalctl gained a new "--reverse" (or -r) option to show
6571 journal output in reverse order (i.e. newest line first).
6573 * journalctl gained a new "--pager-end" (or -e) option to jump
6574 to immediately jump to the end of the journal in the
6575 pager. This is only supported in conjunction with "less".
6577 * journalctl gained a new "--user-unit=" option, that works
6578 similarly to "--unit=" but filters for user units rather than
6581 * A number of unit files to ease adoption of systemd in
6582 initrds has been added. This moves some minimal logic from
6583 the various initrd implementations into systemd proper.
6585 * The journal files are now owned by a new group
6586 "systemd-journal", which exists specifically to allow access
6587 to the journal, and nothing else. Previously, we used the
6588 "adm" group for that, which however possibly covers more
6589 than just journal/log file access. This new group is now
6590 already used by systemd-journal-gatewayd to ensure this
6591 daemon gets access to the journal files and as little else
6592 as possible. Note that "make install" will also set FS ACLs
6593 up for /var/log/journal to give "adm" and "wheel" read
6594 access to it, in addition to "systemd-journal" which owns
6595 the journal files. We recommend that packaging scripts also
6596 add read access to "adm" + "wheel" to /var/log/journal, and
6597 all existing/future journal files. To normal users and
6598 administrators little changes, however packagers need to
6599 ensure to create the "systemd-journal" system group at
6600 package installation time.
6602 * The systemd-journal-gatewayd now runs as unprivileged user
6603 systemd-journal-gateway:systemd-journal-gateway. Packaging
6604 scripts need to create these system user/group at
6607 * timedated now exposes a new boolean property CanNTP that
6608 indicates whether a local NTP service is available or not.
6610 * systemd-detect-virt will now also detect xen PVs
6612 * The pstore file system is now mounted by default, if it is
6615 * In addition to the SELinux and IMA policies we will now also
6616 load SMACK policies at early boot.
6618 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6619 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6620 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6621 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6622 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6623 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6624 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6625 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6626 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6627 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6628 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6629 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6630 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6631 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6635 * Timer units now support calendar time events in addition to
6636 monotonic time events. That means you can now trigger a unit
6637 based on a calendar time specification such as "Thu,Fri
6638 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6639 or fifth day of any month of the year 2013, given that it is
6640 a thursday or friday. This brings timer event support
6641 considerably closer to cron's capabilities. For details on
6642 the supported calendar time specification language see
6645 * udev now supports a number of different naming policies for
6646 network interfaces for predictable names, and a combination
6647 of these policies is now the default. Please see this wiki
6648 document for details:
6650 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6652 * Auke Kok's bootchart implementation has been added to the
6653 systemd tree. It is an optional component that can graph the
6654 boot in quite some detail. It is one of the best bootchart
6655 implementations around and minimal in its code and
6658 * nss-myhostname has been integrated into the systemd source
6659 tree. nss-myhostname guarantees that the local hostname
6660 always stays resolvable via NSS. It has been a weak
6661 requirement of systemd-hostnamed since a long time, and
6662 since its code is actually trivial we decided to just
6663 include it in systemd's source tree. It can be turned off
6664 with a configure switch.
6666 * The read-ahead logic is now capable of properly detecting
6667 whether a btrfs file system is on SSD or rotating media, in
6668 order to optimize the read-ahead scheme. Previously, it was
6669 only capable of detecting this on traditional file systems
6672 * In udev, additional device properties are now read from the
6673 IAB in addition to the OUI database. Also, Bluetooth company
6674 identities are attached to the devices as well.
6676 * In service files %U may be used as specifier that is
6677 replaced by the configured user name of the service.
6679 * nspawn may now be invoked without a controlling TTY. This
6680 makes it suitable for invocation as its own service. This
6681 may be used to set up a simple containerized server system
6682 using only core OS tools.
6684 * systemd and nspawn can now accept socket file descriptors
6685 when they are started for socket activation. This enables
6686 implementation of socket activated nspawn
6687 containers. i.e. think about autospawning an entire OS image
6688 when the first SSH or HTTP connection is received. We expect
6689 that similar functionality will also be added to libvirt-lxc
6692 * journalctl will now suppress ANSI color codes when
6693 presenting log data.
6695 * systemctl will no longer show control group information for
6696 a unit if the control group is empty anyway.
6698 * logind can now automatically suspend/hibernate/shutdown the
6701 * /etc/machine-info and hostnamed now also expose the chassis
6702 type of the system. This can be used to determine whether
6703 the local system is a laptop, desktop, handset or
6704 tablet. This information may either be configured by the
6705 user/vendor or is automatically determined from ACPI and DMI
6706 information if possible.
6708 * A number of PolicyKit actions are now bound together with
6709 "imply" rules. This should simplify creating UIs because
6710 many actions will now authenticate similar ones as well.
6712 * Unit files learnt a new condition ConditionACPower= which
6713 may be used to conditionalize a unit depending on whether an
6714 AC power source is connected or not, of whether the system
6715 is running on battery power.
6717 * systemctl gained a new "is-failed" verb that may be used in
6718 shell scripts and suchlike to check whether a specific unit
6719 is in the "failed" state.
6721 * The EnvironmentFile= setting in unit files now supports file
6722 globbing, and can hence be used to easily read a number of
6723 environment files at once.
6725 * systemd will no longer detect and recognize specific
6726 distributions. All distribution-specific #ifdeffery has been
6727 removed, systemd is now fully generic and
6728 distribution-agnostic. Effectively, not too much is lost as
6729 a lot of the code is still accessible via explicit configure
6730 switches. However, support for some distribution specific
6731 legacy configuration file formats has been dropped. We
6732 recommend distributions to simply adopt the configuration
6733 files everybody else uses now and convert the old
6734 configuration from packaging scripts. Most distributions
6735 already did that. If that's not possible or desirable,
6736 distributions are welcome to forward port the specific
6737 pieces of code locally from the git history.
6739 * When logging a message about a unit systemd will now always
6740 log the unit name in the message meta data.
6742 * localectl will now also discover system locale data that is
6743 not stored in locale archives, but directly unpacked.
6745 * logind will no longer unconditionally use framebuffer
6746 devices as seat masters, i.e. as devices that are required
6747 to be existing before a seat is considered preset. Instead,
6748 it will now look for all devices that are tagged as
6749 "seat-master" in udev. By default, framebuffer devices will
6750 be marked as such, but depending on local systems, other
6751 devices might be marked as well. This may be used to
6752 integrate graphics cards using closed source drivers (such
6753 as NVidia ones) more nicely into logind. Note however, that
6754 we recommend using the open source NVidia drivers instead,
6755 and no udev rules for the closed-source drivers will be
6756 shipped from us upstream.
6758 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6759 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6760 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6761 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6762 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6763 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6764 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6765 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6766 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6767 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6768 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6769 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6774 * udev gained support for loading additional device properties
6775 from an indexed database that is keyed by vendor/product IDs
6776 and similar device identifiers. For the beginning this
6777 "hwdb" is populated with data from the well-known PCI and
6778 USB database, but also includes PNP, ACPI and OID data. In
6779 the longer run this indexed database shall grow into
6780 becoming the one central database for non-essential
6781 userspace device metadata. Previously, data from the PCI/USB
6782 database was only attached to select devices, since the
6783 lookup was a relatively expensive operation due to O(n) time
6784 complexity (with n being the number of entries in the
6785 database). Since this is now O(1), we decided to add in this
6786 data for all devices where this is available, by
6787 default. Note that the indexed database needs to be rebuilt
6788 when new data files are installed. To achieve this you need
6789 to update your packaging scripts to invoke "udevadm hwdb
6790 --update" after installation of hwdb data files. For
6791 RPM-based distributions we introduced the new
6792 %udev_hwdb_update macro for this purpose.
6794 * The Journal gained support for the "Message Catalog", an
6795 indexed database to link up additional information with
6796 journal entries. For further details please check:
6798 https://www.freedesktop.org/wiki/Software/systemd/catalog
6800 The indexed message catalog database also needs to be
6801 rebuilt after installation of message catalog files. Use
6802 "journalctl --update-catalog" for this. For RPM-based
6803 distributions we introduced the %journal_catalog_update
6804 macro for this purpose.
6806 * The Python Journal bindings gained support for the standard
6807 Python logging framework.
6809 * The Journal API gained new functions for checking whether
6810 the underlying file system of a journal file is capable of
6811 properly reporting file change notifications, or whether
6812 applications that want to reflect journal changes "live"
6813 need to recheck journal files continuously in appropriate
6816 * It is now possible to set the "age" field for tmpfiles
6817 entries to 0, indicating that files matching this entry
6818 shall always be removed when the directories are cleaned up.
6820 * coredumpctl gained a new "gdb" verb which invokes gdb
6821 right-away on the selected coredump.
6823 * There's now support for "hybrid sleep" on kernels that
6824 support this, in addition to "suspend" and "hibernate". Use
6825 "systemctl hybrid-sleep" to make use of this.
6827 * logind's HandleSuspendKey= setting (and related settings)
6828 now gained support for a new "lock" setting to simply
6829 request the screen lock on all local sessions, instead of
6830 actually executing a suspend or hibernation.
6832 * systemd will now mount the EFI variables file system by
6835 * Socket units now gained support for configuration of the
6836 SMACK security label.
6838 * timedatectl will now output the time of the last and next
6839 daylight saving change.
6841 * We dropped support for various legacy and distro-specific
6842 concepts, such as insserv, early-boot SysV services
6843 (i.e. those for non-standard runlevels such as 'b' or 'S')
6844 or ArchLinux /etc/rc.conf support. We recommend the
6845 distributions who still need support this to either continue
6846 to maintain the necessary patches downstream, or find a
6847 different solution. (Talk to us if you have questions!)
6849 * Various systemd components will now bypass PolicyKit checks
6850 for root and otherwise handle properly if PolicyKit is not
6851 found to be around. This should fix most issues for
6852 PolicyKit-less systems. Quite frankly this should have been
6853 this way since day one. It is absolutely our intention to
6854 make systemd work fine on PolicyKit-less systems, and we
6855 consider it a bug if something does not work as it should if
6856 PolicyKit is not around.
6858 * For embedded systems it is now possible to build udev and
6859 systemd without blkid and/or kmod support.
6861 * "systemctl switch-root" is now capable of switching root
6862 more than once. I.e. in addition to transitions from the
6863 initrd to the host OS it is now possible to transition to
6864 further OS images from the host. This is useful to implement
6865 offline updating tools.
6867 * Various other additions have been made to the RPM macros
6868 shipped with systemd. Use %udev_rules_update() after
6869 installing new udev rules files. %_udevhwdbdir,
6870 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6871 %_sysctldir are now available which resolve to the right
6872 directories for packages to place various data files in.
6874 * journalctl gained the new --full switch (in addition to
6875 --all, to disable ellipsation for long messages.
6877 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6878 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6879 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6880 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6881 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6882 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6883 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6884 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6885 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6889 * journalctl gained new --since= and --until= switches to
6890 filter by time. It also now supports nice filtering for
6891 units via --unit=/-u.
6893 * Type=oneshot services may use ExecReload= and do the
6896 * The journal daemon now supports time-based rotation and
6897 vacuuming, in addition to the usual disk-space based
6900 * The journal will now index the available field values for
6901 each field name. This enables clients to show pretty drop
6902 downs of available match values when filtering. The bash
6903 completion of journalctl has been updated
6904 accordingly. journalctl gained a new switch -F to list all
6905 values a certain field takes in the journal database.
6907 * More service events are now written as structured messages
6908 to the journal, and made recognizable via message IDs.
6910 * The timedated, localed and hostnamed mini-services which
6911 previously only provided support for changing time, locale
6912 and hostname settings from graphical DEs such as GNOME now
6913 also have a minimal (but very useful) text-based client
6914 utility each. This is probably the nicest way to changing
6915 these settings from the command line now, especially since
6916 it lists available options and is fully integrated with bash
6919 * There's now a new tool "systemd-coredumpctl" to list and
6920 extract coredumps from the journal.
6922 * We now install a README each in /var/log/ and
6923 /etc/rc.d/init.d explaining where the system logs and init
6924 scripts went. This hopefully should help folks who go to
6925 that dirs and look into the otherwise now empty void and
6926 scratch their heads.
6928 * When user-services are invoked (by systemd --user) the
6929 $MANAGERPID env var is set to the PID of systemd.
6931 * SIGRTMIN+24 when sent to a --user instance will now result
6932 in immediate termination of systemd.
6934 * gatewayd received numerous feature additions such as a
6935 "follow" mode, for live syncing and filtering.
6937 * browse.html now allows filtering and showing detailed
6938 information on specific entries. Keyboard navigation and
6939 mouse screen support has been added.
6941 * gatewayd/journalctl now supports HTML5/JSON
6942 Server-Sent-Events as output.
6944 * The SysV init script compatibility logic will now
6945 heuristically determine whether a script supports the
6946 "reload" verb, and only then make this available as
6949 * "systemctl status --follow" has been removed, use "journalctl
6952 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6953 have been removed since they are hardly useful to be
6956 * And I'd like to take the opportunity to specifically mention
6957 Zbigniew for his great contributions. Zbigniew, you rock!
6959 Contributions from: Andrew Eikum, Christian Hesse, Colin
6960 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6961 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6962 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6963 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6964 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6965 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6969 * If /etc/vconsole.conf is non-existent or empty we will no
6970 longer load any console font or key map at boot by
6971 default. Instead the kernel defaults will be left
6972 intact. This is definitely the right thing to do, as no
6973 configuration should mean no configuration, and hard-coding
6974 font names that are different on all archs is probably a bad
6975 idea. Also, the kernel default key map and font should be
6976 good enough for most cases anyway, and mostly identical to
6977 the userspace fonts/key maps we previously overloaded them
6978 with. If distributions want to continue to default to a
6979 non-kernel font or key map they should ship a default
6980 /etc/vconsole.conf with the appropriate contents.
6982 Contributions from: Colin Walters, Daniel J Walsh, Dave
6983 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6984 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6988 * journalctl gained a new --cursor= switch to show entries
6989 starting from the specified location in the journal.
6991 * We now enforce a size limit on journal entry fields exported
6992 with "-o json" in journalctl. Fields larger than 4K will be
6993 assigned null. This can be turned off with --all.
6995 * An (optional) journal gateway daemon is now available as
6996 "systemd-journal-gatewayd.service". This service provides
6997 access to the journal via HTTP and JSON. This functionality
6998 will be used to implement live log synchronization in both
6999 pull and push modes, but has various other users too, such
7000 as easy log access for debugging of embedded devices. Right
7001 now it is already useful to retrieve the journal via HTTP:
7003 # systemctl start systemd-journal-gatewayd.service
7004 # wget http://localhost:19531/entries
7006 This will download the journal contents in a
7007 /var/log/messages compatible format. The same as JSON:
7009 # curl -H"Accept: application/json" http://localhost:19531/entries
7011 This service is also accessible via a web browser where a
7012 single static HTML5 app is served that uses the JSON logic
7013 to enable the user to do some basic browsing of the
7014 journal. This will be extended later on. Here's an example
7015 screenshot of this app in its current state:
7017 http://0pointer.de/public/journal-gatewayd
7019 Contributions from: Kay Sievers, Lennart Poettering, Robert
7020 Milasan, Tom Gundersen
7024 * The bash completion logic is now available for journalctl
7027 * We do not mount the "cpuset" controller anymore together with
7028 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7029 started if no parameters are assigned to it. "cpuset" hence
7030 broke code that assumed it could create "cpu" groups and
7033 * journalctl -f will now subscribe to terminal size changes,
7034 and line break accordingly.
7036 Contributions from: Dave Reisner, Kay Sievers, Lennart
7037 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7041 * nspawn will now create a symlink /etc/localtime in the
7042 container environment, copying the host's timezone
7043 setting. Previously this has been done via a bind mount, but
7044 since symlinks cannot be bind mounted this has now been
7045 changed to create/update the appropriate symlink.
7047 * journalctl -n's line number argument is now optional, and
7048 will default to 10 if omitted.
7050 * journald will now log the maximum size the journal files may
7051 take up on disk. This is particularly useful if the default
7052 built-in logic of determining this parameter from the file
7053 system size is used. Use "systemctl status
7054 systemd-journald.service" to see this information.
7056 * The multi-seat X wrapper tool has been stripped down. As X
7057 is now capable of enumerating graphics devices via udev in a
7058 seat-aware way the wrapper is not strictly necessary
7059 anymore. A stripped down temporary stop-gap is still shipped
7060 until the upstream display managers have been updated to
7061 fully support the new X logic. Expect this wrapper to be
7062 removed entirely in one of the next releases.
7064 * HandleSleepKey= in logind.conf has been split up into
7065 HandleSuspendKey= and HandleHibernateKey=. The old setting
7066 is not available anymore. X11 and the kernel are
7067 distinguishing between these keys and we should too. This
7068 also means the inhibition lock for these keys has been split
7071 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7072 Poettering, Lukas Nykryn, Václav Pavlín
7076 * Whenever a unit changes state we will now log this to the
7077 journal and show along the unit's own log output in
7080 * ConditionPathIsMountPoint= can now properly detect bind
7081 mount points too. (Previously, a bind mount of one file
7082 system to another place in the same file system could not be
7083 detected as mount, since they shared struct stat's st_dev
7086 * We will now mount the cgroup controllers cpu, cpuacct,
7087 cpuset and the controllers net_cls, net_prio together by
7090 * nspawn containers will now have a virtualized boot
7091 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7092 over with a randomized ID at container initialization). This
7093 has the effect of making "journalctl -b" do the right thing
7096 * The JSON output journal serialization has been updated not
7097 to generate "endless" list objects anymore, but rather one
7098 JSON object per line. This is more in line how most JSON
7099 parsers expect JSON objects. The new output mode
7100 "json-pretty" has been added to provide similar output, but
7101 neatly aligned for readability by humans.
7103 * We dropped all explicit sync() invocations in the shutdown
7104 code. The kernel does this implicitly anyway in the kernel
7105 reboot() syscall. halt(8)'s -n option is now a compatibility
7108 * We now support virtualized reboot() in containers, as
7109 supported by newer kernels. We will fall back to exit() if
7110 CAP_SYS_REBOOT is not available to the container. Also,
7111 nspawn makes use of this now and will actually reboot the
7112 container if the containerized OS asks for that.
7114 * journalctl will only show local log output by default
7115 now. Use --merge (-m) to show remote log output, too.
7117 * libsystemd-journal gained the new sd_journal_get_usage()
7118 call to determine the current disk usage of all journal
7119 files. This is exposed in the new "journalctl --disk-usage"
7122 * journald gained a new configuration setting SplitMode= in
7123 journald.conf which may be used to control how user journals
7124 are split off. See journald.conf(5) for details.
7126 * A new condition type ConditionFileNotEmpty= has been added.
7128 * tmpfiles' "w" lines now support file globbing, to write
7129 multiple files at once.
7131 * We added Python bindings for the journal submission
7132 APIs. More Python APIs for a number of selected APIs will
7133 likely follow. Note that we intend to add native bindings
7134 only for the Python language, as we consider it common
7135 enough to deserve bindings shipped within systemd. There are
7136 various projects outside of systemd that provide bindings
7137 for languages such as PHP or Lua.
7139 * Many conditions will now resolve specifiers such as %i. In
7140 addition, PathChanged= and related directives of .path units
7141 now support specifiers as well.
7143 * There's now a new RPM macro definition for the system preset
7146 * journald will now warn if it ca not forward a message to the
7147 syslog daemon because its socket is full.
7149 * timedated will no longer write or process /etc/timezone,
7150 except on Debian. As we do not support late mounted /usr
7151 anymore /etc/localtime always being a symlink is now safe,
7152 and hence the information in /etc/timezone is not necessary
7155 * logind will now always reserve one VT for a text getty (VT6
7156 by default). Previously if more than 6 X sessions where
7157 started they took up all the VTs with auto-spawned gettys,
7158 so that no text gettys were available anymore.
7160 * udev will now automatically inform the btrfs kernel logic
7161 about btrfs RAID components showing up. This should make
7162 simple hotplug based btrfs RAID assembly work.
7164 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7165 (but not for its children which will stay at the kernel
7166 default). This should allow setups with a lot more listening
7169 * systemd will now always pass the configured timezone to the
7170 kernel at boot. timedated will do the same when the timezone
7173 * logind's inhibition logic has been updated. By default,
7174 logind will now handle the lid switch, the power and sleep
7175 keys all the time, even in graphical sessions. If DEs want
7176 to handle these events on their own they should take the new
7177 handle-power-key, handle-sleep-key and handle-lid-switch
7178 inhibitors during their runtime. A simple way to achieve
7179 that is to invoke the DE wrapped in an invocation of:
7181 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7183 * Access to unit operations is now checked via SELinux taking
7184 the unit file label and client process label into account.
7186 * systemd will now notify the administrator in the journal
7187 when he over-mounts a non-empty directory.
7189 * There are new specifiers that are resolved in unit files,
7190 for the host name (%H), the machine ID (%m) and the boot ID
7193 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7194 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7195 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7196 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7197 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7198 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7199 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7203 * Support for reading structured kernel messages from
7204 /dev/kmsg has now been added and is enabled by default.
7206 * Support for reading kernel messages from /proc/kmsg has now
7207 been removed. If you want kernel messages in the journal
7208 make sure to run a recent kernel (>= 3.5) that supports
7209 reading structured messages from /dev/kmsg (see
7210 above). /proc/kmsg is now exclusive property of classic
7211 syslog daemons again.
7213 * The libudev API gained the new
7214 udev_device_new_from_device_id() call.
7216 * The logic for file system namespace (ReadOnlyDirectory=,
7217 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7218 require pivot_root() anymore. This means fewer temporary
7219 directories are created below /tmp for this feature.
7221 * nspawn containers will now see and receive all submounts
7222 made on the host OS below the root file system of the
7225 * Forward Secure Sealing is now supported for Journal files,
7226 which provide cryptographical sealing of journal files so
7227 that attackers cannot alter log history anymore without this
7228 being detectable. Lennart will soon post a blog story about
7229 this explaining it in more detail.
7231 * There are two new service settings RestartPreventExitStatus=
7232 and SuccessExitStatus= which allow configuration of exit
7233 status (exit code or signal) which will be excepted from the
7234 restart logic, resp. consider successful.
7236 * journalctl gained the new --verify switch that can be used
7237 to check the integrity of the structure of journal files and
7238 (if Forward Secure Sealing is enabled) the contents of
7241 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7242 and similar symlinks pre-created. This makes running shells
7243 as container init process a lot more fun.
7245 * The fstab support can now handle PARTUUID= and PARTLABEL=
7248 * A new ConditionHost= condition has been added to match
7249 against the hostname (with globs) and machine ID. This is
7250 useful for clusters where a single OS image is used to
7251 provision a large number of hosts which shall run slightly
7252 different sets of services.
7254 * Services which hit the restart limit will now be placed in a
7257 Contributions from: Bertram Poettering, Dave Reisner, Huang
7258 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7259 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7263 * When running in --user mode systemd will now become a
7264 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7265 tree a lot more organized.
7267 * A new PartOf= unit dependency type has been introduced that
7268 may be used to group services in a natural way.
7270 * "systemctl enable" may now be used to enable instances of
7273 * journalctl now prints error log levels in red, and
7274 warning/notice log levels in bright white. It also supports
7275 filtering by log level now.
7277 * cgtop gained a new -n switch (similar to top), to configure
7278 the maximum number of iterations to run for. It also gained
7279 -b, to run in batch mode (accepting no input).
7281 * The suffix ".service" may now be omitted on most systemctl
7282 command lines involving service unit names.
7284 * There's a new bus call in logind to lock all sessions, as
7285 well as a loginctl verb for it "lock-sessions".
7287 * libsystemd-logind.so gained a new call sd_journal_perror()
7288 that works similar to libc perror() but logs to the journal
7289 and encodes structured information about the error number.
7291 * /etc/crypttab entries now understand the new keyfile-size=
7294 * shutdown(8) now can send a (configurable) wall message when
7295 a shutdown is cancelled.
7297 * The mount propagation mode for the root file system will now
7298 default to "shared", which is useful to make containers work
7299 nicely out-of-the-box so that they receive new mounts from
7300 the host. This can be undone locally by running "mount
7301 --make-rprivate /" if needed.
7303 * The prefdm.service file has been removed. Distributions
7304 should maintain this unit downstream if they intend to keep
7305 it around. However, we recommend writing normal unit files
7306 for display managers instead.
7308 * Since systemd is a crucial part of the OS we will now
7309 default to a number of compiler switches that improve
7310 security (hardening) such as read-only relocations, stack
7311 protection, and suchlike.
7313 * The TimeoutSec= setting for services is now split into
7314 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7315 of individual time outs for the start and the stop phase of
7318 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7319 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7320 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7321 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7322 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7323 Gundersen, Zbigniew Jędrzejewski-Szmek
7327 * The journal and id128 C APIs are now fully documented as man
7330 * Extra safety checks have been added when transitioning from
7331 the initial RAM disk to the main system to avoid accidental
7334 * /etc/crypttab entries now understand the new keyfile-offset=
7337 * systemctl -t can now be used to filter by unit load state.
7339 * The journal C API gained the new sd_journal_wait() call to
7340 make writing synchronous journal clients easier.
7342 * journalctl gained the new -D switch to show journals from a
7345 * journalctl now displays a special marker between log
7346 messages of two different boots.
7348 * The journal is now explicitly flushed to /var via a service
7349 systemd-journal-flush.service, rather than implicitly simply
7350 by seeing /var/log/journal to be writable.
7352 * journalctl (and the journal C APIs) can now match for much
7353 more complex expressions, with alternatives and
7356 * When transitioning from the initial RAM disk to the main
7357 system we will now kill all processes in a killing spree to
7358 ensure no processes stay around by accident.
7360 * Three new specifiers may be used in unit files: %u, %h, %s
7361 resolve to the user name, user home directory resp. user
7362 shell. This is useful for running systemd user instances.
7364 * We now automatically rotate journal files if their data
7365 object hash table gets a fill level > 75%. We also size the
7366 hash table based on the configured maximum file size. This
7367 together should lower hash collisions drastically and thus
7368 speed things up a bit.
7370 * journalctl gained the new "--header" switch to introspect
7371 header data of journal files.
7373 * A new setting SystemCallFilters= has been added to services
7374 which may be used to apply blacklists or whitelists to
7375 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7377 * nspawn gained a new --link-journal= switch (and quicker: -j)
7378 to link the container journal with the host. This makes it
7379 very easy to centralize log viewing on the host for all
7380 guests while still keeping the journal files separated.
7382 * Many bugfixes and optimizations
7384 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7385 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7386 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7391 * Several tools now understand kernel command line arguments,
7392 which are only read when run in an initial RAM disk. They
7393 usually follow closely their normal counterparts, but are
7396 * There's a new tool to analyze the readahead files that are
7397 automatically generated at boot. Use:
7399 /usr/lib/systemd/systemd-readahead analyze /.readahead
7401 * We now provide an early debug shell on tty9 if this enabled. Use:
7403 systemctl enable debug-shell.service
7405 * All plymouth related units have been moved into the Plymouth
7406 package. Please make sure to upgrade your Plymouth version
7409 * systemd-tmpfiles now supports getting passed the basename of
7410 a configuration file only, in which case it will look for it
7411 in all appropriate directories automatically.
7413 * udevadm info now takes a /dev or /sys path as argument, and
7414 does the right thing. Example:
7416 udevadm info /dev/sda
7417 udevadm info /sys/class/block/sda
7419 * systemctl now prints a warning if a unit is stopped but a
7420 unit that might trigger it continues to run. Example: a
7421 service is stopped but the socket that activates it is left
7424 * "systemctl status" will now mention if the log output was
7425 shortened due to rotation since a service has been started.
7427 * The journal API now exposes functions to determine the
7428 "cutoff" times due to rotation.
7430 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7431 immediately flushing of runtime logs to /var if possible,
7432 resp. for triggering immediate rotation of the journal
7435 * It is now considered an error if a service is attempted to
7436 be stopped that is not loaded.
7438 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7440 * systemd-analyze now supports Python 3
7442 * tmpfiles now supports cleaning up directories via aging
7443 where the first level dirs are always kept around but
7444 directories beneath it automatically aged. This is enabled
7445 by prefixing the age field with '~'.
7447 * Seat objects now expose CanGraphical, CanTTY properties
7448 which is required to deal with very fast bootups where the
7449 display manager might be running before the graphics drivers
7450 completed initialization.
7452 * Seat objects now expose a State property.
7454 * We now include RPM macros for service enabling/disabling
7455 based on the preset logic. We recommend RPM based
7456 distributions to make use of these macros if possible. This
7457 makes it simpler to reuse RPM spec files across
7460 * We now make sure that the collected systemd unit name is
7461 always valid when services log to the journal via
7464 * There's a new man page kernel-command-line(7) detailing all
7465 command line options we understand.
7467 * The fstab generator may now be disabled at boot by passing
7468 fstab=0 on the kernel command line.
7470 * A new kernel command line option modules-load= is now understood
7471 to load a specific kernel module statically, early at boot.
7473 * Unit names specified on the systemctl command line are now
7474 automatically escaped as needed. Also, if file system or
7475 device paths are specified they are automatically turned
7476 into the appropriate mount or device unit names. Example:
7478 systemctl status /home
7479 systemctl status /dev/sda
7481 * The SysVConsole= configuration option has been removed from
7482 system.conf parsing.
7484 * The SysV search path is no longer exported on the D-Bus
7487 * The Names= option has been removed from unit file parsing.
7489 * There's a new man page bootup(7) detailing the boot process.
7491 * Every unit and every generator we ship with systemd now
7492 comes with full documentation. The self-explanatory boot is
7495 * A couple of services gained "systemd-" prefixes in their
7496 name if they wrap systemd code, rather than only external
7497 code. Among them fsck@.service which is now
7498 systemd-fsck@.service.
7500 * The HaveWatchdog property has been removed from the D-Bus
7503 * systemd.confirm_spawn= on the kernel command line should now
7506 * There's a new man page crypttab(5) which details all options
7507 we actually understand.
7509 * systemd-nspawn gained a new --capability= switch to pass
7510 additional capabilities to the container.
7512 * timedated will now read known NTP implementation unit names
7513 from /usr/lib/systemd/ntp-units.d/*.list,
7514 systemd-timedated-ntp.target has been removed.
7516 * journalctl gained a new switch "-b" that lists log data of
7517 the current boot only.
7519 * The notify socket is in the abstract namespace again, in
7520 order to support daemons which chroot() at start-up.
7522 * There is a new Storage= configuration option for journald
7523 which allows configuration of where log data should go. This
7524 also provides a way to disable journal logging entirely, so
7525 that data collected is only forwarded to the console, the
7526 kernel log buffer or another syslog implementation.
7528 * Many bugfixes and optimizations
7530 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7531 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7532 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7533 Shawn Landden, Tom Gundersen
7537 * "systemctl help <unit>" now shows the man page if one is
7540 * Several new man pages have been added.
7542 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7543 MaxLevelConsole= can now be specified in
7544 journald.conf. These options allow reducing the amount of
7545 data stored on disk or forwarded by the log level.
7547 * TimerSlackNSec= can now be specified in system.conf for
7548 PID1. This allows system-wide power savings.
7550 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7551 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7556 * logind is now capable of (optionally) handling power and
7557 sleep keys as well as the lid switch.
7559 * journalctl now understands the syntax "journalctl
7560 /usr/bin/avahi-daemon" to get all log output of a specific
7563 * CapabilityBoundingSet= in system.conf now also influences
7564 the capability bound set of usermode helpers of the kernel.
7566 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7567 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7568 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7569 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7573 * Note that we skipped 139 releases here in order to set the
7574 new version to something that is greater than both udev's
7575 and systemd's most recent version number.
7577 * udev: all udev sources are merged into the systemd source tree now.
7578 All future udev development will happen in the systemd tree. It
7579 is still fully supported to use the udev daemon and tools without
7580 systemd running, like in initramfs or other init systems. Building
7581 udev though, will require the *build* of the systemd tree, but
7582 udev can be properly *run* without systemd.
7584 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7585 should be used to create dead device nodes as workarounds for broken
7588 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7589 no longer supported. udev_monitor_new_from_netlink() needs to be
7590 used to subscribe to events.
7592 * udev: when udevd is started by systemd, processes which are left
7593 behind by forking them off of udev rules, are unconditionally cleaned
7594 up and killed now after the event handling has finished. Services or
7595 daemons must be started as systemd services. Services can be
7596 pulled-in by udev to get started, but they can no longer be directly
7597 forked by udev rules.
7599 * udev: the daemon binary is called systemd-udevd now and installed
7600 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7601 to adapt to that, create symlink, or rename the binary after building
7604 * libudev no longer provides these symbols:
7605 udev_monitor_from_socket()
7606 udev_queue_get_failed_list_entry()
7607 udev_get_{dev,sys,run}_path()
7608 The versions number was bumped and symbol versioning introduced.
7610 * systemd-loginctl and systemd-journalctl have been renamed
7611 to loginctl and journalctl to match systemctl.
7613 * The config files: /etc/systemd/systemd-logind.conf and
7614 /etc/systemd/systemd-journald.conf have been renamed to
7615 logind.conf and journald.conf. Package updates should rename
7616 the files to the new names on upgrade.
7618 * For almost all files the license is now LGPL2.1+, changed
7619 from the previous GPL2.0+. Exceptions are some minor stuff
7620 of udev (which will be changed to LGPL2.1 eventually, too),
7621 and the MIT licensed sd-daemon.[ch] library that is suitable
7622 to be used as drop-in files.
7624 * systemd and logind now handle system sleep states, in
7625 particular suspending and hibernating.
7627 * logind now implements a sleep/shutdown/idle inhibiting logic
7628 suitable for a variety of uses. Soonishly Lennart will blog
7629 about this in more detail.
7631 * var-run.mount and var-lock.mount are no longer provided
7632 (which previously bind mounted these directories to their new
7633 places). Distributions which have not converted these
7634 directories to symlinks should consider stealing these files
7635 from git history and add them downstream.
7637 * We introduced the Documentation= field for units and added
7638 this to all our shipped units. This is useful to make it
7639 easier to explore the boot and the purpose of the various
7642 * All smaller setup units (such as
7643 systemd-vconsole-setup.service) now detect properly if they
7644 are run in a container and are skipped when
7645 appropriate. This guarantees an entirely noise-free boot in
7646 Linux container environments such as systemd-nspawn.
7648 * A framework for implementing offline system updates is now
7649 integrated, for details see:
7650 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7652 * A new service type Type=idle is available now which helps us
7653 avoiding ugly interleaving of getty output and boot status
7656 * There's now a system-wide CapabilityBoundingSet= option to
7657 globally reduce the set of capabilities for the
7658 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7659 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7660 even CAP_NET_ADMIN system-wide for secure systems.
7662 * There are now system-wide DefaultLimitXXX= options to
7663 globally change the defaults of the various resource limits
7664 for all units started by PID 1.
7666 * Harald Hoyer's systemd test suite has been integrated into
7667 systemd which allows easy testing of systemd builds in qemu
7668 and nspawn. (This is really awesome! Ask us for details!)
7670 * The fstab parser is now implemented as generator, not inside
7673 * systemctl will now warn you if .mount units generated from
7674 /etc/fstab are out of date due to changes in fstab that
7675 have not been read by systemd yet.
7677 * systemd is now suitable for usage in initrds. Dracut has
7678 already been updated to make use of this. With this in place
7679 initrds get a slight bit faster but primarily are much
7680 easier to introspect and debug since "systemctl status" in
7681 the host system can be used to introspect initrd services,
7682 and the journal from the initrd is kept around too.
7684 * systemd-delta has been added, a tool to explore differences
7685 between user/admin configuration and vendor defaults.
7687 * PrivateTmp= now affects both /tmp and /var/tmp.
7689 * Boot time status messages are now much prettier and feature
7690 proper english language. Booting up systemd has never been
7693 * Read-ahead pack files now include the inode number of all
7694 files to pre-cache. When the inode changes the pre-caching
7695 is not attempted. This should be nicer to deal with updated
7696 packages which might result in changes of read-ahead
7699 * We now temporaritly lower the kernel's read_ahead_kb variable
7700 when collecting read-ahead data to ensure the kernel's
7701 built-in read-ahead does not add noise to our measurements
7702 of necessary blocks to pre-cache.
7704 * There's now RequiresMountsFor= to add automatic dependencies
7705 for all mounts necessary for a specific file system path.
7707 * MountAuto= and SwapAuto= have been removed from
7708 system.conf. Mounting file systems at boot has to take place
7711 * nspawn now learned a new switch --uuid= to set the machine
7712 ID on the command line.
7714 * nspawn now learned the -b switch to automatically search
7717 * vt102 is now the default TERM for serial TTYs, upgraded from
7720 * systemd-logind now works on VT-less systems.
7722 * The build tree has been reorganized. The individual
7723 components now have directories of their own.
7725 * A new condition type ConditionPathIsReadWrite= is now available.
7727 * nspawn learned the new -C switch to create cgroups for the
7728 container in other hierarchies.
7730 * We now have support for hardware watchdogs, configurable in
7733 * The scheduled shutdown logic now has a public API.
7735 * We now mount /tmp as tmpfs by default, but this can be
7736 masked and /etc/fstab can override it.
7738 * Since udisks does not make use of /media anymore we are not
7739 mounting a tmpfs on it anymore.
7741 * journalctl gained a new --local switch to only interleave
7742 locally generated journal files.
7744 * We can now load the IMA policy at boot automatically.
7746 * The GTK tools have been split off into a systemd-ui.
7748 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7749 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7750 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7751 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7752 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7753 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7754 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7755 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7756 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7761 * This is mostly a bugfix release
7763 * Support optional initialization of the machine ID from the
7764 KVM or container configured UUID.
7766 * Support immediate reboots with "systemctl reboot -ff"
7768 * Show /etc/os-release data in systemd-analyze output
7770 * Many bugfixes for the journal, including endianness fixes and
7771 ensuring that disk space enforcement works
7773 * sd-login.h is C++ compatible again
7775 * Extend the /etc/os-release format on request of the Debian
7778 * We now refuse non-UTF8 strings used in various configuration
7779 and unit files. This is done to ensure we do not pass invalid
7780 data over D-Bus or expose it elsewhere.
7782 * Register Mimo USB Screens as suitable for automatic seat
7785 * Read SELinux client context from journal clients in a race
7788 * Reorder configuration file lookup order. /etc now always
7789 overrides /run in order to allow the administrator to always
7790 and unconditionally override vendor-supplied or
7791 automatically generated data.
7793 * The various user visible bits of the journal now have man
7794 pages. We still lack man pages for the journal API calls
7797 * We now ship all man pages in HTML format again in the
7800 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7801 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7802 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7807 * This is mostly a bugfix release
7809 * systems lacking /etc/os-release are no longer supported.
7811 * Various functionality updates to libsystemd-login.so
7813 * Track class of PAM logins to distinguish greeters from
7816 Contributions from: Kay Sievers, Lennart Poettering, Michael
7821 * This is an important bugfix release for v41.
7823 * Building man pages is now optional which should be useful
7824 for those building systemd from git but unwilling to install
7827 * Watchdog support for supervising services is now usable. In
7828 a future release support for hardware watchdogs
7829 (i.e. /dev/watchdog) will be added building on this.
7831 * Service start rate limiting is now configurable and can be
7832 turned off per service. When a start rate limit is hit a
7833 reboot can automatically be triggered.
7835 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7837 Contributions from: Benjamin Franzke, Bill Nottingham,
7838 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7839 Schmidt, Michał Górny, Piotr Drąg
7843 * The systemd binary is installed /usr/lib/systemd/systemd now;
7844 An existing /sbin/init symlink needs to be adapted with the
7847 * The code that loads kernel modules has been ported to invoke
7848 libkmod directly, instead of modprobe. This means we do not
7849 support systems with module-init-tools anymore.
7851 * Watchdog support is now already useful, but still not
7854 * A new kernel command line option systemd.setenv= is
7855 understood to set system wide environment variables
7856 dynamically at boot.
7858 * We now limit the set of capabilities of systemd-journald.
7860 * We now set SIGPIPE to ignore by default, since it only is
7861 useful in shell pipelines, and has little use in general
7862 code. This can be disabled with IgnoreSIPIPE=no in unit
7865 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7866 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7871 * This is mostly a bugfix release
7873 * We now expose the reason why a service failed in the
7874 "Result" D-Bus property.
7876 * Rudimentary service watchdog support (will be completed over
7877 the next few releases.)
7879 * When systemd forks off in order execute some service we will
7880 now immediately changes its argv[0] to reflect which process
7881 it will execute. This is useful to minimize the time window
7882 with a generic argv[0], which makes bootcharts more useful
7884 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7885 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7886 Mike Kazantsev, Ray Strode
7890 * This is mostly a test release, but incorporates many
7893 * New systemd-cgtop tool to show control groups by their
7896 * Linking against libacl for ACLs is optional again. If
7897 disabled, support tracking device access for active logins
7898 goes becomes unavailable, and so does access to the user
7899 journals by the respective users.
7901 * If a group "adm" exists, journal files are automatically
7902 owned by them, thus allow members of this group full access
7903 to the system journal as well as all user journals.
7905 * The journal now stores the SELinux context of the logging
7906 client for all entries.
7908 * Add C++ inclusion guards to all public headers
7910 * New output mode "cat" in the journal to print only text
7911 messages, without any meta data like date or time.
7913 * Include tiny X server wrapper as a temporary stop-gap to
7914 teach XOrg udev display enumeration. This is used by display
7915 managers such as gdm, and will go away as soon as XOrg
7916 learned native udev hotplugging for display devices.
7918 * Add new systemd-cat tool for executing arbitrary programs
7919 with STDERR/STDOUT connected to the journal. Can also act as
7920 BSD logger replacement, and does so by default.
7922 * Optionally store all locally generated coredumps in the
7923 journal along with meta data.
7925 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7926 writing short strings to files (for usage for /sys), and for
7927 creating symlinks, character and block device nodes.
7929 * New unit file option ControlGroupPersistent= to make cgroups
7930 persistent, following the mechanisms outlined in
7931 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7933 * Support multiple local RTCs in a sane way
7935 * No longer monopolize IO when replaying readahead data on
7936 rotating disks, since we might starve non-file-system IO to
7937 death, since fanotify() will not see accesses done by blkid,
7940 * Do not show kernel threads in systemd-cgls anymore, unless
7941 requested with new -k switch.
7943 Contributions from: Dan Horák, Kay Sievers, Lennart
7944 Poettering, Michal Schmidt
7948 * This is mostly a test release, but incorporates many
7951 * The git repository moved to:
7952 git://anongit.freedesktop.org/systemd/systemd
7953 ssh://git.freedesktop.org/git/systemd/systemd
7955 * First release with the journal
7956 http://0pointer.de/blog/projects/the-journal.html
7958 * The journal replaces both systemd-kmsg-syslogd and
7959 systemd-stdout-bridge.
7961 * New sd_pid_get_unit() API call in libsystemd-logind
7963 * Many systemadm clean-ups
7965 * Introduce remote-fs-pre.target which is ordered before all
7966 remote mounts and may be used to start services before all
7969 * Added Mageia support
7971 * Add bash completion for systemd-loginctl
7973 * Actively monitor PID file creation for daemons which exit in
7974 the parent process before having finished writing the PID
7975 file in the daemon process. Daemons which do this need to be
7976 fixed (i.e. PID file creation must have finished before the
7977 parent exits), but we now react a bit more gracefully to them.
7979 * Add colourful boot output, mimicking the well-known output
7980 of existing distributions.
7982 * New option PassCredentials= for socket units, for
7983 compatibility with a recent kernel ABI breakage.
7985 * /etc/rc.local is now hooked in via a generator binary, and
7986 thus will no longer act as synchronization point during
7989 * systemctl list-unit-files now supports --root=.
7991 * systemd-tmpfiles now understands two new commands: z, Z for
7992 relabelling files according to the SELinux database. This is
7993 useful to apply SELinux labels to specific files in /sys,
7996 * Output of SysV services is now forwarded to both the console
7997 and the journal by default, not only just the console.
7999 * New man pages for all APIs from libsystemd-login.
8001 * The build tree got reorganized and the build system is a
8002 lot more modular allowing embedded setups to specifically
8003 select the components of systemd they are interested in.
8005 * Support for Linux systems lacking the kernel VT subsystem is
8008 * configure's --with-rootdir= got renamed to
8009 --with-rootprefix= to follow the naming used by udev and
8012 * Unless specified otherwise we will now install to /usr instead
8013 of /usr/local by default.
8015 * Processes with '@' in argv[0][0] are now excluded from the
8016 final shut-down killing spree, following the logic explained
8018 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8020 * All processes remaining in a service cgroup when we enter
8021 the START or START_PRE states are now killed with
8022 SIGKILL. That means it is no longer possible to spawn
8023 background processes from ExecStart= lines (which was never
8024 supported anyway, and bad style).
8026 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8027 reloading of units together.
8029 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8030 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8031 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8032 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8033 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek