1 systemd System and Service Manager
5 * The MemoryAccounting= unit property now defaults to on. After
6 discussions with the upstream control group maintainers we learnt
7 that the negative impact of cgroup memory accounting on current
8 kernels is finally relatively minimal, so that it should be safe to
9 enable this by default without affecting system performance. Besides
10 memory accounting only task accounting is turned on by default, all
11 other forms of resource accounting (CPU, IO, IP) remain off for now,
12 because it's not clear yet that their impact is small enough to move
13 from opt-in to opt-out. We recommend downstreams to leave memory
14 accounting on by default if kernel 4.14 or higher is are primarily
15 used. On very resource constrained systems or when support for old
16 kernels is a necessity, -Dmemory-accounting-default=false can be used
17 to revert this change.
19 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
20 %udev_rules_update) and the journal catalog (%journal_catalog_update)
21 from the upgrade scriptlets of individual packages now do nothing.
22 Transfiletriggers have been added which will perform those updates
23 once at the end of the transaction.
25 Similar transfiletriggers have been added to execute any sysctl.d
26 and binfmt.d rules. Thus, it should be unnecessary to provide any
27 scriptlets to execute this configuration from package installation
30 * elogind-sysusers gained a mode where the configuration to execute is
31 specified on the command line, but this configuration is not executed
32 directly, but instead it is merged with the configuration on disk,
33 and the result is executed. This is useful for package installation
34 scripts which want to create the user before installing any files on
35 disk (in case some of those files are owned by that user), while
36 still allowing local admin overrides.
38 This functionality is exposed to rpm scriplets through a new
39 %sysusers_create_package macro. Old %sysusers_create and
40 %sysusers_create_inline macros are deprecated.
42 A transfiletrigger for sysusers.d configuration is now installed,
43 which means that it should be uncessary to call elogind-sysusers from
44 package installation scripts, unless the package installs any files
45 owned by those newly-created users, in which case
46 %sysusers_create_package should be used.
48 * Analogous change has been done for elogind-tmpfiles: it gained a mode
49 where the command-line configuration is merged with the configuration
50 on disk. This is exposed as the new %tmpfiles_create_package macro,
51 and %tmpfiles_create is deprecated. A transfiletrigger is installed
52 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
53 from package installation scripts.
55 * sysusers.d configuration for a user may now also specify the group
56 number, in addition to the user number ("u username 123:456"), or
57 without the user number ("u username -:456").
59 * Configution items for elogind-sysusers can now be specified as
60 positional arguments when the new --inline switch is used.
62 * The login shell of users created through sysusers.d may now be
63 specified (previously, it was always /bin/sh for root and
64 /sbin/nologin for other users).
66 * elogind-analyze gained a new --global switch to look at global user
67 configuration. It also gained a unit-paths verb to list the unit load
68 paths that are compiled into elogind (which can be used with
69 --elogind, --user, or --global).
71 * udevadm trigger gained a new --settle/-w option to wait for any
72 triggered events to finish (but just those, and not any other events
73 which are triggered meanwhile).
75 * The action that elogind takes when the lid is closed and the
76 machine is connected to external power can now be configured using
77 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
78 was determined by HandleLidSwitch=, and, for backwards compatibility,
79 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
81 * journalctl will periodically call sd_journal_process() to make it
82 resilient against inotify queue overruns when journal files are
85 * Two new functions in libelogind — sd_bus_get_n_queued_read and
86 sd_bus_get_n_queued_write — may be used to check the number of
89 * elogind gained a new
90 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
91 which can be used to migrate foreign processes to scope and service
92 units. The primary user for this new API is elogind itself: the
93 elogind --user instance uses this call of the elogind --system
94 instance to migrate processes if it itself gets the request to
95 migrate processes and the kernel refuses this due to access
96 restrictions. Thanks to this "elogind-run --scope --user …" works
97 again in pure cgroups v2 environments when invoked from the user
100 * A new TemporaryFileSystem= setting can be used to mask out part of
101 the real file system tree with tmpfs mounts. This may be combined
102 with BindPaths= and BindReadOnlyPaths= to hide files or directories
103 not relevant to the unit, while still allowing some paths lower in
104 the tree to be accessed.
106 ProtectHome=tmpfs may now be used to hide user home and runtime
107 directories from units, in a way that is mostly equivalent to
108 "TemporaryFileSystem=/home /run/user /root".
110 * Non-service units are now started with KeyringMode=shared by default.
111 This means that mount and swapon and other mount tools have access
112 to keys in the main keyring.
114 * /sys/fs/bpf is now mounted automatically.
116 * QNX virtualization is now detected by elogind-detect-virt and may
117 be used in ConditionVirtualization=.
119 * IPAccounting= may now be enabled also for slice units.
121 * A new -Dsplit-bin= build configuration switch may be used to specify
122 whether bin and sbin directories are merged, or if they should be
123 included separately in $PATH and various listings of executable
124 directories. The build configuration scripts will try to autodetect
125 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
126 system, but distributions are encouraged to configure this
129 * A new -Dok-color= build configuration switch may be used to change
130 the colour of "OK" status messages.
132 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
133 PrivateNetwork=yes was buggy in previous versions of elogind. This
134 means that after the upgrade and daemon-reexec, any such units must
137 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
138 will not exclude read-only files owned by root from cleanup.
140 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
141 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
142 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
143 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
144 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
145 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
146 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
147 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
148 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
149 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
150 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
151 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
152 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
153 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
154 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
155 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
161 * Some keyboards come with a zoom see-saw or rocker which until now got
162 mapped to the Linux "zoomin/out" keys in hwdb. However, these
163 keycodes are not recognized by any major desktop. They now produce
164 Up/Down key events so that they can be used for scrolling.
166 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
167 slightly: previously, if an argument was specified for lines of this
168 type (i.e. the right-most column was set) this string was appended to
169 existing files each time elogind-tmpfiles was run. This behaviour was
170 different from what the documentation said, and not particularly
171 useful, as repeated elogind-tmpfiles invocations would not be
172 idempotent and grow such files without bounds. With this release
173 behaviour has been altered slightly, to match what the documentation
174 says: lines of this type only have an effect if the indicated files
175 don't exist yet, and only then the argument string is written to the
178 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
179 elogind-tmpfiles behaviour: previously, read-only files owned by root
180 were always excluded from the file "aging" algorithm (i.e. the
181 automatic clean-up of directories like /tmp based on
182 atime/mtime/ctime). We intend to drop this restriction, and age files
183 by default even when owned by root and read-only. This behaviour was
184 inherited from older tools, but there have been requests to remove
185 it, and it's not obvious why this restriction was made in the first
186 place. Please speak up now, if you are aware of software that reqires
187 this behaviour, otherwise we'll remove the restriction in v238.
189 * A new environment variable $SYSTEMD_OFFLINE is now understood by
190 systemctl. It takes a boolean argument. If on, systemctl assumes it
191 operates on an "offline" OS tree, and will not attempt to talk to the
192 service manager. Previously, this mode was implicitly enabled if a
193 chroot() environment was detected, and this new environment variable
194 now provides explicit control.
196 * .path and .socket units may now be created transiently, too.
197 Previously only service, mount, automount and timer units were
198 supported as transient units. The elogind-run tool has been updated
199 to expose this new functionality, you may hence use it now to bind
200 arbitrary commands to path or socket activation on-the-fly from the
201 command line. Moreover, almost all properties are now exposed for the
202 unit types that already supported transient operation.
204 * The elogind-mount command gained support for a new --owner= parameter
205 which takes a user name, which is then resolved and included in uid=
206 and gid= mount options string of the file system to mount.
208 * A new unit condition ConditionControlGroupController= has been added
209 that checks whether a specific cgroup controller is available.
211 * Unit files, udev's .link files, and elogind-networkd's .netdev and
212 .network files all gained support for a new condition
213 ConditionKernelVersion= for checking against specific kernel
216 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
217 support for configuring device flags in the Flags= setting. In the
218 same files, the [Tunnel] section gained support for configuring
219 AllowLocalRemote=. The [Route] section in .network files gained
220 support for configuring InitialCongestionWindow=,
221 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
222 understands RapidCommit=.
224 * elogind-networkd's DHCPv6 support gained support for Prefix
227 * sd-bus gained support for a new "watch-bind" feature. When this
228 feature is enabled, an sd_bus connection may be set up to connect to
229 an AF_UNIX socket in the file system as soon as it is created. This
230 functionality is useful for writing early-boot services that
231 automatically connect to the system bus as soon as it is started,
232 without ugly time-based polling. elogind-networkd and
233 elogind-resolved have been updated to make use of this
234 functionality. busctl exposes this functionality in a new
235 --watch-bind= command line switch.
237 * sd-bus will now optionally synthesize a local "Connected" signal as
238 soon as a D-Bus connection is set up fully. This message mirrors the
239 already existing "Disconnected" signal which is synthesized when the
240 connection is terminated. This signal is generally useful but
241 particularly handy in combination with the "watch-bind" feature
242 described above. Synthesizing of this message has to be requested
243 explicitly through the new API call sd_bus_set_connected_signal(). In
244 addition a new call sd_bus_is_ready() has been added that checks
245 whether a connection is fully set up (i.e. between the "Connected" and
246 "Disconnected" signals).
248 * sd-bus gained two new calls sd_bus_request_name_async() and
249 sd_bus_release_name_async() for asynchronously registering bus
250 names. Similar, there is now sd_bus_add_match_async() for installing
251 a signal match asynchronously. All of elogind's own services have
252 been updated to make use of these calls. Doing these operations
253 asynchronously has two benefits: it reduces the risk of deadlocks in
254 case of cyclic dependencies between bus services, and it speeds up
255 service initialization since synchronization points for bus
256 round-trips are removed.
258 * sd-bus gained two new calls sd_bus_match_signal() and
259 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
260 and sd_bus_add_match_async() but instead of taking a D-Bus match
261 string take match fields as normal function parameters.
263 * sd-bus gained two new calls sd_bus_set_sender() and
264 sd_bus_message_set_sender() for setting the sender name of outgoing
265 messages (either for all outgoing messages or for just one specific
266 one). These calls are only useful in direct connections as on
267 brokered connections the broker fills in the sender anyway,
268 overwriting whatever the client filled in.
270 * sd-event gained a new pseudo-handle that may be specified on all API
271 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
272 used this refers to the default event loop object of the calling
273 thread. Note however that this does not implicitly allocate one —
274 which has to be done prior by using sd_event_default(). Similarly
275 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
276 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
277 to the default bus of the specified type of the calling thread. Here
278 too this does not implicitly allocate bus connection objects, this
279 has to be done prior with sd_bus_default() and friends.
281 * sd-event gained a new call pair
282 sd_event_source_{get|set}_io_fd_own(). This may be used to request
283 automatic closure of the file descriptor an IO event source watches
284 when the event source is destroyed.
286 * elogind-networkd gained support for natively configuring WireGuard
289 * In previous versions elogind synthesized user records both for the
290 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
291 internally. In order to simplify distribution-wide renames of the
292 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
293 new transitional flag file has been added: if
294 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
295 user and group record within the elogind codebase is disabled.
297 * elogind-notify gained a new --uid= option for selecting the source
298 user/UID to use for notification messages sent to the service
301 * journalctl gained a new --grep= option to list only entries in which
302 the message matches a certain pattern. By default matching is case
303 insensitive if the pattern is lowercase, and case sensitive
304 otherwise. Option --case-sensitive=yes|no can be used to override
305 this an specify case sensitivity or case insensitivity.
307 * There's now a "elogind-analyze service-watchdogs" command for printing
308 the current state of the service runtime watchdog, and optionally
309 enabling or disabling the per-service watchdogs system-wide if given a
310 boolean argument (i.e. the concept you configure in WatchdogSec=), for
311 debugging purposes. There's also a kernel command line option
312 elogind.service_watchdogs= for controlling the same.
314 * Two new "log-level" and "log-target" options for elogind-analyze were
315 addded that merge the now deprecated get-log-level, set-log-level and
316 get-log-target, set-log-target pairs. The deprecated options are still
317 understood for backwards compatibility. The two new options print the
318 current value when no arguments are given, and set them when a
319 level/target is given as an argument.
321 * sysusers.d's "u" lines now optionally accept both a UID and a GID
322 specification, separated by a ":" character, in order to create users
323 where UID and GID do not match.
325 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
326 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
327 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
328 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
329 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
330 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
331 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
332 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
333 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
334 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
335 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
336 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
337 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
338 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
339 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
340 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
341 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
342 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
343 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
344 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
351 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
352 in v235 has been extended to also set the dummy.ko module option
353 numdummies=0, preventing the kernel from automatically creating
354 dummy0. All dummy interfaces must now be explicitly created.
356 * Unknown '%' specifiers in configuration files are now rejected. This
357 applies to units and tmpfiles.d configuration. Any percent characters
358 that are followed by a letter or digit that are not supposed to be
359 interpreted as the beginning of a specifier should be escaped by
360 doubling ("%%"). (So "size=5%" is still accepted, as well as
361 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
362 valid specifiers today.)
364 * systemd-resolved now maintains a new dynamic
365 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
366 recommended to make /etc/resolv.conf a symlink to it. This file
367 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
368 includes dynamically acquired search domains, achieving more correct
369 DNS resolution by software that bypasses local DNS APIs such as NSS.
371 * The "uaccess" udev tag has been dropped from /dev/kvm and
372 /dev/dri/renderD*. These devices now have the 0666 permissions by
373 default (but this may be changed at build-time). /dev/dri/renderD*
374 will now be owned by the "render" group along with /dev/kfd.
376 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
377 systemd-journal-gatewayd.service and
378 systemd-journal-upload.service. This means "nss-systemd" must be
379 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
380 services are resolved properly.
382 * In /etc/fstab two new mount options are now understood:
383 x-systemd.makefs and x-systemd.growfs. The former has the effect that
384 the configured file system is formatted before it is mounted, the
385 latter that the file system is resized to the full block device size
386 after it is mounted (i.e. if the file system is smaller than the
387 partition it resides on, it's grown). This is similar to the fsck
388 logic in /etc/fstab, and pulls in systemd-makefs@.service and
389 systemd-growfs@.service as necessary, similar to
390 systemd-fsck@.service. Resizing is currently only supported on ext4
393 * In systemd-networkd, the IPv6 RA logic now optionally may announce
394 DNS server and domain information.
396 * Support for the LUKS2 on-disk format for encrypted partitions has
397 been added. This requires libcryptsetup2 during compilation and
400 * The systemd --user instance will now signal "readiness" when its
401 basic.target unit has been reached, instead of when the run queue ran
402 empty for the first time.
404 * Tmpfiles.d with user configuration are now also supported.
405 systemd-tmpfiles gained a new --user switch, and snippets placed in
406 ~/.config/user-tmpfiles.d/ and corresponding directories will be
407 executed by systemd-tmpfiles --user running in the new
408 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
409 running in the user session.
411 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
412 %S resolves to the top-level state directory (/var/lib for the system
413 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
414 top-level cache directory (/var/cache for the system instance,
415 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
416 logs directory (/var/log for the system instance,
417 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
418 existing %t specifier, that resolves to the top-level runtime
419 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
422 * journalctl learnt a new parameter --output-fields= for limiting the
423 set of journal fields to output in verbose and JSON output modes.
425 * systemd-timesyncd's configuration file gained a new option
426 RootDistanceMaxSec= for setting the maximum root distance of servers
427 it'll use, as well as the new options PollIntervalMinSec= and
428 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
430 * bootctl gained a new command "list" for listing all available boot
431 menu items on systems that follow the boot loader specification.
433 * systemctl gained a new --dry-run switch that shows what would be done
434 instead of doing it, and is currently supported by the shutdown and
437 * ConditionSecurity= can now detect the TOMOYO security module.
439 * Unit file [Install] sections are now also respected in unit drop-in
440 files. This is intended to be used by drop-ins under /usr/lib/.
442 * systemd-firstboot may now also set the initial keyboard mapping.
444 * Udev "changed" events for devices which are exposed as systemd
445 .device units are now propagated to units specified in
446 ReloadPropagatedFrom= as reload requests.
448 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
449 unit template name (i.e. a name in the form of 'foobar@.service',
450 without the instance component between the '@' and - the '.'), then
451 the escaped sysfs path of the device is automatically used as the
454 * SystemCallFilter= in unit files has been extended so that an "errno"
455 can be specified individually for each system call. Example:
456 SystemCallFilter=~uname:EILSEQ.
458 * The cgroup delegation logic has been substantially updated. Delegate=
459 now optionally takes a list of controllers (instead of a boolean, as
460 before), which lists the controllers to delegate at least.
462 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
464 * A new LogLevelMax= setting configures the maximum log level any
465 process of the service may log at (i.e. anything with a lesser
466 priority than what is specified is automatically dropped). A new
467 LogExtraFields= setting allows configuration of additional journal
468 fields to attach to all log records generated by any of the unit's
471 * New StandardInputData= and StandardInputText= settings along with the
472 new option StandardInput=data may be used to configure textual or
473 binary data that shall be passed to the executed service process via
474 standard input, encoded in-line in the unit file.
476 * StandardInput=, StandardOutput= and StandardError= may now be used to
477 connect stdin/stdout/stderr of executed processes directly with a
478 file or AF_UNIX socket in the file system, using the new "file:" option.
480 * A new unit file option CollectMode= has been added, that allows
481 tweaking the garbage collection logic for units. It may be used to
482 tell systemd to garbage collect units that have failed automatically
483 (normally it only GCs units that exited successfully). systemd-run
484 and systemd-mount expose this new functionality with a new -G option.
486 * "machinectl bind" may now be used to bind mount non-directories
487 (i.e. regularfiles, devices, fifos, sockets).
489 * systemd-analyze gained a new verb "calendar" for validating and
490 testing calendar time specifications to use for OnCalendar= in timer
491 units. Besides validating the expression it will calculate the next
492 time the specified expression would elapse.
494 * In addition to the pre-existing FailureAction= unit file setting
495 there's now SuccessAction=, for configuring a shutdown action to
496 execute when a unit completes successfully. This is useful in
497 particular inside containers that shall terminate after some workload
498 has been completed. Also, both options are now supported for all unit
499 types, not just services.
501 * networkds's IP rule support gained two new options
502 IncomingInterface= and OutgoingInterface= for configuring the incoming
503 and outgoing interfaces of configured rules. systemd-networkd also
504 gained support for "vxcan" network devices.
506 * networkd gained a new setting RequiredForOnline=, taking a
507 boolean. If set, systemd-wait-online will take it into consideration
508 when determining that the system is up, otherwise it will ignore the
509 interface for this purpose.
511 * The sd_notify() protocol gained support for a new operation: with
512 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
513 store again, ahead of POLLHUP or POLLERR when they are removed
516 * A new document doc/UIDS-GIDS.md has been added to the source tree,
517 that documents the UID/GID range and assignment assumptions and
518 requirements of systemd.
520 * The watchdog device PID 1 will ping may now be configured through the
521 WatchdogDevice= configuration file setting, or by setting the
522 systemd.watchdog_service= kernel commandline option.
524 * systemd-resolved's gained support for registering DNS-SD services on
525 the local network using MulticastDNS. Services may either be
526 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
527 the same dir below /run, /usr/lib), or through its D-Bus API.
529 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
530 extend the effective start, runtime, and stop time. The service must
531 continue to send EXTEND_TIMEOUT_USEC within the period specified to
532 prevent the service manager from making the service as timedout.
534 * elogind-resolved's DNSSEC support gained support for RFC 8080
535 (Ed25519 keys and signatures).
537 * The elogind-resolve command line tool gained a new set of options
538 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
539 --set-nta= and --revert to configure per-interface DNS configuration
540 dynamically during runtime. It's useful for pushing DNS information
541 into elogind-resolved from DNS hook scripts that various interface
542 managing software supports (such as pppd).
544 * elogind-nspawn gained a new --network-namespace-path= command line
545 option, which may be used to make a container join an existing
546 network namespace, by specifying a path to a "netns" file.
548 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
549 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
550 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
551 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
552 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
553 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
554 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
555 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
556 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
557 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
558 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
559 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
560 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
561 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
562 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
563 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
564 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
565 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
566 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
567 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
568 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
569 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
570 Jędrzejewski-Szmek, Zeal Jagannatha
576 * INCOMPATIBILITY: systemd-logind.service and other long-running
577 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
578 communication with the outside. This generally improves security of
579 the system, and is in almost all cases a safe and good choice, as
580 these services do not and should not provide any network-facing
581 functionality. However, systemd-logind uses the glibc NSS API to
582 query the user database. This creates problems on systems where NSS
583 is set up to directly consult network services for user database
584 lookups. In particular, this creates incompatibilities with the
585 "nss-nis" module, which attempts to directly contact the NIS/YP
586 network servers it is configured for, and will now consistently
587 fail. In such cases, it is possible to turn off IP sandboxing for
588 systemd-logind.service (set IPAddressDeny= in its [Service] section
589 to the empty string, via a .d/ unit file drop-in). Downstream
590 distributions might want to update their nss-nis packaging to include
591 such a drop-in snippet, accordingly, to hide this incompatibility
592 from the user. Another option is to make use of glibc's nscd service
593 to proxy such network requests through a privilege-separated, minimal
594 local caching daemon, or to switch to more modern technologies such
595 sssd, whose NSS hook-ups generally do not involve direct network
596 access. In general, we think it's definitely time to question the
597 implementation choices of nss-nis, i.e. whether it's a good idea
598 today to embed a network-facing loadable module into all local
599 processes that need to query the user database, including the most
600 trivial and benign ones, such as "ls". For more details about
601 IPAddressDeny= see below.
603 * A new modprobe.d drop-in is now shipped by default that sets the
604 bonding module option max_bonds=0. This overrides the kernel default,
605 to avoid conflicts and ambiguity as to whether or not bond0 should be
606 managed by systemd-networkd or not. This resolves multiple issues
607 with bond0 properties not being applied, when bond0 is configured
608 with systemd-networkd. Distributors may choose to not package this,
609 however in that case users will be prevented from correctly managing
610 bond0 interface using systemd-networkd.
612 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
613 which print the logging level and target of the system manager. They
614 complement the existing "set-log-level" and "set-log-target" verbs
615 used to change those values.
617 * journald.conf gained a new boolean setting ReadKMsg= which defaults
618 to on. If turned off kernel log messages will not be read by
619 systemd-journald or included in the logs. It also gained a new
620 setting LineMax= for configuring the maximum line length in
621 STDOUT/STDERR log streams. The new default for this value is 48K, up
622 from the previous hardcoded 2048.
624 * A new unit setting RuntimeDirectoryPreserve= has been added, which
625 allows more detailed control of what to do with a runtime directory
626 configured with RuntimeDirectory= (i.e. a directory below /run or
627 $XDG_RUNTIME_DIR) after a unit is stopped.
629 * The RuntimeDirectory= setting for units gained support for creating
630 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
631 one top-level directory.
633 * Units gained new options StateDirectory=, CacheDirectory=,
634 LogsDirectory= and ConfigurationDirectory= which are closely related
635 to RuntimeDirectory= but manage per-service directories below
636 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
637 possible to write unit files which when activated automatically gain
638 properly owned service specific directories in these locations, thus
639 making unit files self-contained and increasing compatibility with
640 stateless systems and factory reset where /etc or /var are
641 unpopulated at boot. Matching these new settings there's also
642 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
643 ConfigurationDirectoryMode= for configuring the access mode of these
644 directories. These settings are particularly useful in combination
645 with DynamicUser=yes as they provide secure, properly-owned,
646 writable, and stateful locations for storage, excluded from the
647 sandbox that such services live in otherwise.
649 * Automake support has been removed from this release. systemd is now
652 * systemd-journald will now aggressively cache client metadata during
653 runtime, speeding up log write performance under pressure. This comes
654 at a small price though: as much of the metadata is read
655 asynchronously from /proc/ (and isn't implicitly attached to log
656 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
657 metadata stored alongside a log entry might be slightly
658 out-of-date. Previously it could only be slightly newer than the log
659 message. The time window is small however, and given that the kernel
660 is unlikely to be improved anytime soon in this regard, this appears
663 * nss-myhostname/systemd-resolved will now by default synthesize an
664 A/AAAA resource record for the "_gateway" hostname, pointing to the
665 current default IP gateway. Previously it did that for the "gateway"
666 name, hampering adoption, as some distributions wanted to leave that
667 host name open for local use. The old behaviour may still be
668 requested at build time.
670 * systemd-networkd's [Address] section in .network files gained a new
671 Scope= setting for configuring the IP address scope. The [Network]
672 section gained a new boolean setting ConfigureWithoutCarrier= that
673 tells systemd-networkd to ignore link sensing when configuring the
674 device. The [DHCP] section gained a new Anonymize= boolean option for
675 turning on a number of options suggested in RFC 7844. A new
676 [RoutingPolicyRule] section has been added for configuring the IP
677 routing policy. The [Route] section has gained support for a new
678 Type= setting which permits configuring
679 blackhole/unreachable/prohibit routes.
681 * The [VRF] section in .netdev files gained a new Table= setting for
682 configuring the routing table to use. The [Tunnel] section gained a
683 new Independent= boolean field for configuring tunnels independent of
684 an underlying network interface. The [Bridge] section gained a new
685 GroupForwardMask= option for configuration of propagation of link
686 local frames between bridge ports.
688 * The WakeOnLan= setting in .link files gained support for a number of
689 new modes. A new TCP6SegmentationOffload= setting has been added for
690 configuring TCP/IPv6 hardware segmentation offload.
692 * The IPv6 RA sender implementation may now optionally send out RDNSS
693 and RDNSSL records to supply DNS configuration to peers.
695 * systemd-nspawn gained support for a new --system-call-filter= command
696 line option for adding and removing entries in the default system
697 call filter it applies. Moreover systemd-nspawn has been changed to
698 implement a system call whitelist instead of a blacklist.
700 * systemd-run gained support for a new --pipe command line option. If
701 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
702 are directly passed on to the activated transient service
703 executable. This allows invoking arbitrary processes as systemd
704 services (for example to take benefit of dependency management,
705 accounting management, resource management or log management that is
706 done automatically for services) — while still allowing them to be
707 integrated in a classic UNIX shell pipeline.
709 * When a service sends RELOAD=1 via sd_notify() and reload propagation
710 using ReloadPropagationTo= is configured, a reload is now propagated
711 to configured units. (Previously this was only done on explicitly
712 requested reloads, using "systemctl reload" or an equivalent
715 * For each service unit a restart counter is now kept: it is increased
716 each time the service is restarted due to Restart=, and may be
717 queried using "systemctl show -p NRestarts …".
719 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
720 @signal and @timer have been added, for usage with SystemCallFilter=
721 in unit files and the new --system-call-filter= command line option
722 of systemd-nspawn (see above).
724 * ExecStart= lines in unit files gained two new modifiers: when a
725 command line is prefixed with "!" the command will be executed as
726 configured, except for the credentials applied by
727 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
728 "+", but does still apply namespacing options unlike "+". There's
729 also "!!" now, which is mostly identical, but becomes a NOP on
730 systems that support ambient capabilities. This is useful to write
731 unit files that work with ambient capabilities where possible but
732 automatically fall back to traditional privilege dropping mechanisms
733 on systems where this is not supported.
735 * ListenNetlink= settings in socket units now support RDMA netlink
738 * A new unit file setting LockPersonality= has been added which permits
739 locking down the chosen execution domain ("personality") of a service
742 * A new special target "getty-pre.target" has been added, which is
743 ordered before all text logins, and may be used to order services
744 before textual logins acquire access to the console.
746 * systemd will now attempt to load the virtio-rng.ko kernel module very
747 early on if a VM environment supporting this is detected. This should
748 improve entropy during early boot in virtualized environments.
750 * A _netdev option is now supported in /etc/crypttab that operates in a
751 similar way as the same option in /etc/fstab: it permits configuring
752 encrypted devices that need to be ordered after the network is up.
753 Following this logic, two new special targets
754 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
755 added that are to cryptsetup.target what remote-fs.target and
756 remote-fs-pre.target are to local-fs.target.
758 * Service units gained a new UnsetEnvironment= setting which permits
759 unsetting specific environment variables for services that are
760 normally passed to it (for example in order to mask out locale
761 settings for specific services that can't deal with it).
763 * Units acquired a new boolean option IPAccounting=. When turned on, IP
764 traffic accounting (packet count as well as byte count) is done for
765 the service, and shown as part of "systemctl status" or "systemd-run
768 * Service units acquired two new options IPAddressAllow= and
769 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
770 for configuring a simple IP access control list for all sockets of
771 the unit. These options are available also on .slice and .socket
772 units, permitting flexible access list configuration for individual
773 services as well as groups of services (as defined by a slice unit),
774 including system-wide. Note that IP ACLs configured this way are
775 enforced on every single IPv4 and IPv6 socket created by any process
776 of the service unit, and apply to ingress as well as egress traffic.
778 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
779 structured log message is generated each time the unit is stopped,
780 containing information about the consumed resources of this
783 * A new setting KeyringMode= has been added to unit files, which may be
784 used to control how the kernel keyring is set up for executed
787 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
788 "systemctl kexec" and "systemctl exit" are now always asynchronous in
789 behaviour (that is: these commands return immediately after the
790 operation was enqueued instead of waiting for the operation to
791 complete). Previously, "systemctl poweroff" and "systemctl reboot"
792 were asynchronous on systems using systemd-logind (i.e. almost
793 always, and like they were on sysvinit), and the other three commands
794 were unconditionally synchronous. With this release this is cleaned
795 up, and callers will see the same asynchronous behaviour on all
796 systems for all five operations.
798 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
801 * .timer units now accept calendar specifications in other timezones
802 than UTC or the local timezone.
804 * The tmpfiles snippet var.conf has been changed to create
805 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
806 the "utmp" group already, and it appears to be generally understood
807 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
808 databases. Previously this was implemented correctly for all these
809 databases excepts btmp, which has been opened up like this now
810 too. Note that while the other databases are world-readable
811 (i.e. 0644), btmp is not and remains more restrictive.
813 * The systemd-resolve tool gained a new --reset-server-features
814 switch. When invoked like this systemd-resolved will forget
815 everything it learnt about the features supported by the configured
816 upstream DNS servers, and restarts the feature probing logic on the
817 next resolver look-up for them at the highest feature level
820 * The status dump systemd-resolved sends to the logs upon receiving
821 SIGUSR1 now also includes information about all DNS servers it is
822 configured to use, and the features levels it probed for them.
824 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
825 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
826 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
827 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
828 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
829 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
830 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
831 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
832 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
833 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
834 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
835 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
836 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
837 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
838 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
839 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
840 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
841 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
842 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
843 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
849 * Meson is now supported as build system in addition to Automake. It is
850 our plan to remove Automake in one of our next releases, so that
851 Meson becomes our exclusive build system. Hence, please start using
852 the Meson build system in your downstream packaging. There's plenty
853 of documentation around how to use Meson, the extremely brief
856 ./autogen.sh && ./configure && make && sudo make install
860 meson build && ninja -C build && sudo ninja -C build install
862 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
863 which permits configuring a timeout on the time a job is
864 running. This is particularly useful for setting timeouts on jobs for
867 * Unit files gained two new options ConditionUser= and ConditionGroup=
868 for conditionalizing units based on the identity of the user/group
869 running a systemd user instance.
871 * systemd-networkd now understands a new FlowLabel= setting in the
872 [VXLAN] section of .network files, as well as a Priority= in
873 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
874 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
875 gained support for configuration of GENEVE links, and IPv6 address
876 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
878 * .link files now understand a new Port= setting.
880 * systemd-networkd's DHCP support gained support for DHCP option 119
881 (domain search list).
883 * systemd-networkd gained support for serving IPv6 address ranges using
884 the Router Advertisment protocol. The new .network configuration
885 section [IPv6Prefix] may be used to configure the ranges to
886 serve. This is implemented based on a new, minimal, native server
887 implementation of RA.
889 * journalctl's --output= switch gained support for a new parameter
890 "short-iso-precise" for a mode where timestamps are shown as precise
893 * systemd-udevd's "net_id" builtin may now generate stable network
894 interface names from IBM PowerVM VIO devices as well as ACPI platform
897 * MulticastDNS support in systemd-resolved may now be explicitly
898 enabled/disabled using the new MulticastDNS= configuration file
901 * systemd-resolved may now optionally use libidn2 instead of the libidn
902 for processing internationalized domain names. Support for libidn2
903 should be considered experimental and should not be enabled by
906 * "machinectl pull-tar" and related call may now do verification of
907 downloaded images using SUSE-style .sha256 checksum files in addition
908 to the already existing support for validating using Ubuntu-style
911 * sd-bus gained support for a new sd_bus_message_appendv() call which
912 is va_list equivalent of sd_bus_message_append().
914 * sd-boot gained support for validating images using SHIM/MOK.
916 * The SMACK code learnt support for "onlycap".
918 * systemd-mount --umount is now much smarter in figuring out how to
919 properly unmount a device given its mount or device path.
921 * The code to call libnss_dns as a fallback from libnss_resolve when
922 the communication with systemd-resolved fails was removed. This
923 fallback was redundant and interfered with the [!UNAVAIL=return]
924 suffix. See nss-resolve(8) for the recommended configuration.
926 * systemd-logind may now be restarted without losing state. It stores
927 the file descriptors for devices it manages in the system manager
928 using the FDSTORE= mechanism. Please note that further changes in
929 other components may be required to make use of this (for example
930 Xorg has code to listen for stops of systemd-logind and terminate
931 itself when logind is stopped or restarted, in order to avoid using
932 stale file descriptors for graphical devices, which is now
933 counterproductive and must be reverted in order for restarts of
934 systemd-logind to be safe. See
935 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
937 * All kernel install plugins are called with the environment variable
938 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
939 /etc/machine-id. If the file is missing or empty, the variable is
940 empty and BOOT_DIR_ABS is the path of a temporary directory which is
941 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
942 is empty, all plugins should not put anything in BOOT_DIR_ABS.
944 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
945 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
946 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
947 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
948 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
949 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
950 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
951 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
952 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
953 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
954 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
955 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
956 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
957 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
958 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
959 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
960 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
961 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
962 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
963 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
964 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
965 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
966 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
967 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
968 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
969 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
970 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
971 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
972 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
979 * The "hybrid" control group mode has been modified to improve
980 compatibility with "legacy" cgroups-v1 setups. Specifically, the
981 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
982 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
983 cgroups-v1 hierarchy), the only externally visible change being that
984 the cgroups-v2 hierarchy is also mounted, to
985 /sys/fs/cgroup/unified. This should provide a large degree of
986 compatibility with "legacy" cgroups-v1, while taking benefit of the
987 better management capabilities of cgroups-v2.
989 * The default control group setup mode may be selected both a boot-time
990 via a set of kernel command line parameters (specifically:
991 systemd.unified_cgroup_hierarchy= and
992 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
993 default selected on the configure command line
994 (--with-default-hierarchy=). The upstream default is "hybrid"
995 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
996 this will change in a future systemd version to be "unified" (pure
997 cgroups-v2 mode). The third option for the compile time option is
998 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
999 distributions to default to "hybrid" mode for release distributions,
1000 starting with v233. We recommend "unified" for development
1001 distributions (specifically: distributions such as Fedora's rawhide)
1002 as that's where things are headed in the long run. Use "legacy" for
1003 greatest stability and compatibility only.
1005 * Note one current limitation of "unified" and "hybrid" control group
1006 setup modes: the kernel currently does not permit the systemd --user
1007 instance (i.e. unprivileged code) to migrate processes between two
1008 disconnected cgroup subtrees, even if both are managed and owned by
1009 the user. This effectively means "systemd-run --user --scope" doesn't
1010 work when invoked from outside of any "systemd --user" service or
1011 scope. Specifically, it is not supported from session scopes. We are
1012 working on fixing this in a future systemd version. (See #3388 for
1013 further details about this.)
1015 * DBus policy files are now installed into /usr rather than /etc. Make
1016 sure your system has dbus >= 1.9.18 running before upgrading to this
1017 version, or override the install path with --with-dbuspolicydir= .
1019 * All python scripts shipped with systemd (specifically: the various
1020 tests written in Python) now require Python 3.
1022 * systemd unit tests can now run standalone (without the source or
1023 build directories), and can be installed into /usr/lib/systemd/tests/
1024 with 'make install-tests'.
1026 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1027 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1030 * Support for the %c, %r, %R specifiers in unit files has been
1031 removed. Specifiers are not supposed to be dependent on configuration
1032 in the unit file itself (so that they resolve the same regardless
1033 where used in the unit files), but these specifiers were influenced
1034 by the Slice= option.
1036 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1037 all cases. If distributions want to use a different shell for this
1038 purpose (for example Fedora's /sbin/sushell) they need to specify
1039 this explicitly at configure time using --with-debug-shell=.
1041 * The confirmation spawn prompt has been reworked to offer the
1044 (c)ontinue, proceed without asking anymore
1045 (D)ump, show the state of the unit
1046 (f)ail, don't execute the command and pretend it failed
1048 (i)nfo, show a short summary of the unit
1049 (j)obs, show jobs that are in progress
1050 (s)kip, don't execute the command and pretend it succeeded
1051 (y)es, execute the command
1053 The 'n' choice for the confirmation spawn prompt has been removed,
1054 because its meaning was confusing.
1056 The prompt may now also be redirected to an alternative console by
1057 specifying the console as parameter to systemd.confirm_spawn=.
1059 * Services of Type=notify require a READY=1 notification to be sent
1060 during startup. If no such message is sent, the service now fails,
1061 even if the main process exited with a successful exit code.
1063 * Services that fail to start up correctly now always have their
1064 ExecStopPost= commands executed. Previously, they'd enter "failed"
1065 state directly, without executing these commands.
1067 * The option MulticastDNS= of network configuration files has acquired
1068 an actual implementation. With MulticastDNS=yes a host can resolve
1069 names of remote hosts and reply to mDNS A and AAAA requests.
1071 * When units are about to be started an additional check is now done to
1072 ensure that all dependencies of type BindsTo= (when used in
1073 combination with After=) have been started.
1075 * systemd-analyze gained a new verb "syscall-filter" which shows which
1076 system call groups are defined for the SystemCallFilter= unit file
1077 setting, and which system calls they contain.
1079 * A new system call filter group "@filesystem" has been added,
1080 consisting of various file system related system calls. Group
1081 "@reboot" has been added, covering reboot, kexec and shutdown related
1082 calls. Finally, group "@swap" has been added covering swap
1083 configuration related calls.
1085 * A new unit file option RestrictNamespaces= has been added that may be
1086 used to restrict access to the various process namespace types the
1087 Linux kernel provides. Specifically, it may be used to take away the
1088 right for a service unit to create additional file system, network,
1089 user, and other namespaces. This sandboxing option is particularly
1090 relevant due to the high amount of recently discovered namespacing
1091 related vulnerabilities in the kernel.
1093 * systemd-udev's .link files gained support for a new AutoNegotiation=
1094 setting for configuring Ethernet auto-negotiation.
1096 * systemd-networkd's .network files gained support for a new
1097 ListenPort= setting in the [DHCP] section to explicitly configure the
1098 UDP client port the DHCP client shall listen on.
1100 * .network files gained a new Unmanaged= boolean setting for explicitly
1101 excluding one or more interfaces from management by systemd-networkd.
1103 * The systemd-networkd ProxyARP= option has been renamed to
1104 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1105 renamed to ReduceARPProxy=. The old names continue to be available
1108 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1109 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1111 * systemd-networkd's bonding device support gained support for two new
1112 configuration options ActiveSlave= and PrimarySlave=.
1114 * The various options in the [Match] section of .network files gained
1115 support for negative matching.
1117 * New systemd-specific mount options are now understood in /etc/fstab:
1119 x-systemd.mount-timeout= may be used to configure the maximum
1120 permitted runtime of the mount command.
1122 x-systemd.device-bound may be set to bind a mount point to its
1123 backing device unit, in order to automatically remove a mount point
1124 if its backing device is unplugged. This option may also be
1125 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1126 on the block device, which is now automatically set for all CDROM
1127 drives, so that mounted CDs are automatically unmounted when they are
1128 removed from the drive.
1130 x-systemd.after= and x-systemd.before= may be used to explicitly
1131 order a mount after or before another unit or mount point.
1133 * Enqueued start jobs for device units are now automatically garbage
1134 collected if there are no jobs waiting for them anymore.
1136 * systemctl list-jobs gained two new switches: with --after, for every
1137 queued job the jobs it's waiting for are shown; with --before the
1138 jobs which it's blocking are shown.
1140 * systemd-nspawn gained support for ephemeral boots from disk images
1141 (or in other words: --ephemeral and --image= may now be
1142 combined). Moreover, ephemeral boots are now supported for normal
1143 directories, even if the backing file system is not btrfs. Of course,
1144 if the file system does not support file system snapshots or
1145 reflinks, the initial copy operation will be relatively expensive, but
1146 this should still be suitable for many use cases.
1148 * Calendar time specifications in .timer units now support
1149 specifications relative to the end of a month by using "~" instead of
1150 "-" as separator between month and day. For example, "*-02~03" means
1151 "the third last day in February". In addition a new syntax for
1152 repeated events has been added using the "/" character. For example,
1153 "9..17/2:00" means "every two hours from 9am to 5pm".
1155 * systemd-socket-proxyd gained a new parameter --connections-max= for
1156 configuring the maximum number of concurrent connections.
1158 * sd-id128 gained a new API for generating unique IDs for the host in a
1159 way that does not leak the machine ID. Specifically,
1160 sd_id128_get_machine_app_specific() derives an ID based on the
1161 machine ID a in well-defined, non-reversible, stable way. This is
1162 useful whenever an identifier for the host is needed but where the
1163 identifier shall not be useful to identify the system beyond the
1164 scope of the application itself. (Internally this uses HMAC-SHA256 as
1165 keyed hash function using the machine ID as input.)
1167 * NotifyAccess= gained a new supported value "exec". When set
1168 notifications are accepted from all processes systemd itself invoked,
1169 including all control processes.
1171 * .nspawn files gained support for defining overlay mounts using the
1172 Overlay= and OverlayReadOnly= options. Previously this functionality
1173 was only available on the systemd-nspawn command line.
1175 * systemd-nspawn's --bind= and --overlay= options gained support for
1176 bind/overlay mounts whose source lies within the container tree by
1177 prefixing the source path with "+".
1179 * systemd-nspawn's --bind= and --overlay= options gained support for
1180 automatically allocating a temporary source directory in /var/tmp
1181 that is removed when the container dies. Specifically, if the source
1182 directory is specified as empty string this mechanism is selected. An
1183 example usage is --overlay=+/var::/var, which creates an overlay
1184 mount based on the original /var contained in the image, overlayed
1185 with a temporary directory in the host's /var/tmp. This way changes
1186 to /var are automatically flushed when the container shuts down.
1188 * systemd-nspawn --image= option does now permit raw file system block
1189 devices (in addition to images containing partition tables, as
1192 * The disk image dissection logic in systemd-nspawn gained support for
1193 automatically setting up LUKS encrypted as well as Verity protected
1194 partitions. When a container is booted from an encrypted image the
1195 passphrase is queried at start-up time. When a container with Verity
1196 data is started, the root hash is search in a ".roothash" file
1197 accompanying the disk image (alternatively, pass the root hash via
1198 the new --root-hash= command line option).
1200 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1201 be used to dissect disk images the same way as systemd-nspawn does
1202 it, following the Bootable Partition Specification. It may even be
1203 used to mount disk images with complex partition setups (including
1204 LUKS and Verity partitions) to a local host directory, in order to
1205 inspect them. This tool is not considered public API (yet), and is
1206 thus not installed into /usr/bin. Please do not rely on its
1207 existence, since it might go away or be changed in later systemd
1210 * A new generator "systemd-verity-generator" has been added, similar in
1211 style to "systemd-cryptsetup-generator", permitting automatic setup of
1212 Verity root partitions when systemd boots up. In order to make use of
1213 this your partition setup should follow the Discoverable Partitions
1214 Specification, and the GPT partition ID of the root file system
1215 partition should be identical to the upper 128bit of the Verity root
1216 hash. The GPT partition ID of the Verity partition protecting it
1217 should be the lower 128bit of the Verity root hash. If the partition
1218 image follows this model it is sufficient to specify a single
1219 "roothash=" kernel command line argument to both configure which root
1220 image and verity partition to use as well as the root hash for
1221 it. Note that systemd-nspawn's Verity support follows the same
1222 semantics, meaning that disk images with proper Verity data in place
1223 may be booted in containers with systemd-nspawn as well as on
1224 physical systems via the verity generator. Also note that the "mkosi"
1225 tool available at https://github.com/systemd/mkosi has been updated
1226 to generate Verity protected disk images following this scheme. In
1227 fact, it has been updated to generate disk images that optionally
1228 implement a complete UEFI SecureBoot trust chain, involving a signed
1229 kernel and initrd image that incorporates such a root hash as well as
1230 a Verity-enabled root partition.
1232 * The hardware database (hwdb) udev supports has been updated to carry
1233 accelerometer quirks.
1235 * All system services are now run with a fresh kernel keyring set up
1236 for them. The invocation ID is stored by default in it, thus
1237 providing a safe, non-overridable way to determine the invocation
1240 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1241 options for bind mounting arbitrary paths in a service-specific
1242 way. When these options are used, arbitrary host or service files and
1243 directories may be mounted to arbitrary locations in the service's
1246 * Documentation has been added that lists all of systemd's low-level
1247 environment variables:
1249 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1251 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1252 whether a specific socket file descriptor matches a specified socket
1255 * systemd-firstboot has been updated to check for the
1256 systemd.firstboot= kernel command line option. It accepts a boolean
1257 and when set to false the first boot questions are skipped.
1259 * systemd-fstab-generator has been updated to check for the
1260 systemd.volatile= kernel command line option, which either takes an
1261 optional boolean parameter or the special value "state". If used the
1262 system may be booted in a "volatile" boot mode. Specifically,
1263 "systemd.volatile" is used, the root directory will be mounted as
1264 tmpfs, and only /usr is mounted from the actual root file system. If
1265 "systemd.volatile=state" is used, the root directory will be mounted
1266 as usual, but /var is mounted as tmpfs. This concept provides similar
1267 functionality as systemd-nspawn's --volatile= option, but provides it
1268 on physical boots. Use this option for implementing stateless
1269 systems, or testing systems with all state and/or configuration reset
1270 to the defaults. (Note though that many distributions are not
1271 prepared to boot up without a populated /etc or /var, though.)
1273 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1274 partitions. Previously it only supported LUKS encrypted partitions
1275 for all other uses, except for the root partition itself.
1277 * Socket units gained support for listening on AF_VSOCK sockets for
1278 communication in virtualized QEMU environments.
1280 * The "configure" script gained a new option --with-fallback-hostname=
1281 for specifying the fallback hostname to use if none is configured in
1282 /etc/hostname. For example, by specifying
1283 --with-fallback-hostname=fedora it is possible to default to a
1284 hostname of "fedora" on pristine installations.
1286 * systemd-cgls gained support for a new --unit= switch for listing only
1287 the control groups of a specific unit. Similar --user-unit= has been
1288 added for listing only the control groups of a specific user unit.
1290 * systemd-mount gained a new --umount switch for unmounting a mount or
1291 automount point (and all mount/automount points below it).
1293 * systemd will now refuse full configuration reloads (via systemctl
1294 daemon-reload and related calls) unless at least 16MiB of free space
1295 are available in /run. This is a safety precaution in order to ensure
1296 that generators can safely operate after the reload completed.
1298 * A new unit file option RootImage= has been added, which has a similar
1299 effect as RootDirectory= but mounts the service's root directory from
1300 a disk image instead of plain directory. This logic reuses the same
1301 image dissection and mount logic that systemd-nspawn already uses,
1302 and hence supports any disk images systemd-nspawn supports, including
1303 those following the Discoverable Partition Specification, as well as
1304 Verity enabled images. This option enables systemd to run system
1305 services directly off disk images acting as resource bundles,
1306 possibly even including full integrity data.
1308 * A new MountAPIVFS= unit file option has been added, taking a boolean
1309 argument. If enabled /proc, /sys and /dev (collectively called the
1310 "API VFS") will be mounted for the service. This is only relevant if
1311 RootDirectory= or RootImage= is used for the service, as these mounts
1312 are of course in place in the host mount namespace anyway.
1314 * systemd-nspawn gained support for a new --pivot-root= switch. If
1315 specified the root directory within the container image is pivoted to
1316 the specified mount point, while the original root disk is moved to a
1317 different place. This option enables booting of ostree images
1318 directly with systemd-nspawn.
1320 * The systemd build scripts will no longer complain if the NTP server
1321 addresses are not changed from the defaults. Google now supports
1322 these NTP servers officially. We still recommend downstreams to
1323 properly register an NTP pool with the NTP pool project though.
1325 * coredumpctl gained a new "--reverse" option for printing the list
1326 of coredumps in reverse order.
1328 * coredumpctl will now show additional information about truncated and
1329 inaccessible coredumps, as well as coredumps that are still being
1330 processed. It also gained a new --quiet switch for suppressing
1331 additional informational message in its output.
1333 * coredumpctl gained support for only showing coredumps newer and/or
1334 older than specific timestamps, using the new --since= and --until=
1335 options, reminiscent of journalctl's options by the same name.
1337 * The systemd-coredump logic has been improved so that it may be reused
1338 to collect backtraces in non-compiled languages, for example in
1339 scripting languages such as Python.
1341 * machinectl will now show the UID shift of local containers, if user
1342 namespacing is enabled for them.
1344 * systemd will now optionally run "environment generator" binaries at
1345 configuration load time. They may be used to add environment
1346 variables to the environment block passed to services invoked. One
1347 user environment generator is shipped by default that sets up
1348 environment variables based on files dropped into /etc/environment.d
1349 and ~/.config/environment.d/.
1351 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1354 * hostnamed has been updated to report a new chassis type of
1355 "convertible" to cover "foldable" laptops that can both act as a
1356 tablet and as a laptop, such as various Lenovo Yoga devices.
1358 Contributions from: Adrián López, Alexander Galanin, Alexander
1359 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1360 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1361 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1362 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1363 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1364 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1365 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1366 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1367 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1368 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1369 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1370 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1371 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1372 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1373 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1374 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1375 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1376 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1377 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1378 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1379 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1380 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1381 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1382 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1383 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1384 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1385 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1388 — Berlin, 2017-03-01
1392 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1393 RestrictAddressFamilies= enabled. These sandboxing options should
1394 generally be compatible with the various external udev call-out
1395 binaries we are aware of, however there may be exceptions, in
1396 particular when exotic languages for these call-outs are used. In
1397 this case, consider turning off these settings locally.
1399 * The new RemoveIPC= option can be used to remove IPC objects owned by
1400 the user or group of a service when that service exits.
1402 * The new ProtectKernelModules= option can be used to disable explicit
1403 load and unload operations of kernel modules by a service. In
1404 addition access to /usr/lib/modules is removed if this option is set.
1406 * ProtectSystem= option gained a new value "strict", which causes the
1407 whole file system tree with the exception of /dev, /proc, and /sys,
1408 to be remounted read-only for a service.
1410 * The new ProtectKernelTunables= option can be used to disable
1411 modification of configuration files in /sys and /proc by a service.
1412 Various directories and files are remounted read-only, so access is
1413 restricted even if the file permissions would allow it.
1415 * The new ProtectControlGroups= option can be used to disable write
1416 access by a service to /sys/fs/cgroup.
1418 * Various systemd services have been hardened with
1419 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1420 RestrictAddressFamilies=.
1422 * Support for dynamically creating users for the lifetime of a service
1423 has been added. If DynamicUser=yes is specified, user and group IDs
1424 will be allocated from the range 61184..65519 for the lifetime of the
1425 service. They can be resolved using the new nss-systemd.so NSS
1426 module. The module must be enabled in /etc/nsswitch.conf. Services
1427 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1428 any resources allocated by the service will be cleaned up when the
1429 service exits. They also have ProtectHome=read-only and
1430 ProtectSystem=strict enabled, so they are not able to make any
1431 permanent modifications to the system.
1433 * The nss-systemd module also always resolves root and nobody, making
1434 it possible to have no /etc/passwd or /etc/group files in minimal
1435 container or chroot environments.
1437 * Services may be started with their own user namespace using the new
1438 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1439 under which the service is running are mapped. All other users are
1442 * Support for the cgroup namespace has been added to systemd-nspawn. If
1443 supported by kernel, the container system started by systemd-nspawn
1444 will have its own view of the cgroup hierarchy. This new behaviour
1445 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1447 * The new MemorySwapMax= option can be used to limit the maximum swap
1448 usage under the unified cgroup hierarchy.
1450 * Support for the CPU controller in the unified cgroup hierarchy has
1451 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1452 options. This controller requires out-of-tree patches for the kernel
1453 and the support is provisional.
1455 * Mount and automount units may now be created transiently
1456 (i.e. dynamically at runtime via the bus API, instead of requiring
1457 unit files in the file system).
1459 * systemd-mount is a new tool which may mount file systems – much like
1460 mount(8), optionally pulling in additional dependencies through
1461 transient .mount and .automount units. For example, this tool
1462 automatically runs fsck on a backing block device before mounting,
1463 and allows the automount logic to be used dynamically from the
1464 command line for establishing mount points. This tool is particularly
1465 useful when dealing with removable media, as it will ensure fsck is
1466 run – if necessary – before the first access and that the file system
1467 is quickly unmounted after each access by utilizing the automount
1468 logic. This maximizes the chance that the file system on the
1469 removable media stays in a clean state, and if it isn't in a clean
1470 state is fixed automatically.
1472 * LazyUnmount=yes option for mount units has been added to expose the
1473 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1476 * /efi will be used as the mount point of the EFI boot partition, if
1477 the directory is present, and the mount point was not configured
1478 through other means (e.g. fstab). If /efi directory does not exist,
1479 /boot will be used as before. This makes it easier to automatically
1480 mount the EFI partition on systems where /boot is used for something
1483 * When operating on GPT disk images for containers, systemd-nspawn will
1484 now mount the ESP to /boot or /efi according to the same rules as PID
1485 1 running on a host. This allows tools like "bootctl" to operate
1486 correctly within such containers, in order to make container images
1487 bootable on physical systems.
1489 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1491 * Two new user session targets have been added to support running
1492 graphical sessions under the systemd --user instance:
1493 graphical-session.target and graphical-session-pre.target. See
1494 systemd.special(7) for a description of how those targets should be
1497 * The vconsole initialization code has been significantly reworked to
1498 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1499 support unicode keymaps. Font and keymap configuration will now be
1500 copied to all allocated virtual consoles.
1502 * FreeBSD's bhyve virtualization is now detected.
1504 * Information recorded in the journal for core dumps now includes the
1505 contents of /proc/mountinfo and the command line of the process at
1506 the top of the process hierarchy (which is usually the init process
1509 * systemd-journal-gatewayd learned the --directory= option to serve
1510 files from the specified location.
1512 * journalctl --root=… can be used to peruse the journal in the
1513 /var/log/ directories inside of a container tree. This is similar to
1514 the existing --machine= option, but does not require the container to
1517 * The hardware database has been extended to support
1518 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1521 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1522 specify the click rate for mice which include a horizontal wheel with
1523 a click rate that is different than the one for the vertical wheel.
1525 * systemd-run gained a new --wait option that makes service execution
1526 synchronous. (Specifically, the command will not return until the
1527 specified service binary exited.)
1529 * systemctl gained a new --wait option that causes the start command to
1530 wait until the units being started have terminated again.
1532 * A new journal output mode "short-full" has been added which displays
1533 timestamps with abbreviated English day names and adds a timezone
1534 suffix. Those timestamps include more information than the default
1535 "short" output mode, and can be passed directly to journalctl's
1536 --since= and --until= options.
1538 * /etc/resolv.conf will be bind-mounted into containers started by
1539 systemd-nspawn, if possible, so any changes to resolv.conf contents
1540 are automatically propagated to the container.
1542 * The number of instances for socket-activated services originating
1543 from a single IP address can be limited with
1544 MaxConnectionsPerSource=, extending the existing setting of
1547 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1550 * .netdev and .network configuration can now be extended through
1553 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1554 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1555 can be enabled and disabled using the new UDPSegmentationOffload=,
1556 TCPSegmentationOffload=, GenericSegmentationOffload=,
1557 GenericReceiveOffload=, LargeReceiveOffload= options in the
1558 [Link] section of .link files.
1560 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1561 Port VLAN ID can be configured for bridge devices using the new STP=,
1562 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1563 section of .netdev files.
1565 * The route table to which routes received over DHCP or RA should be
1566 added can be configured with the new RouteTable= option in the [DHCP]
1567 and [IPv6AcceptRA] sections of .network files.
1569 * The Address Resolution Protocol can be disabled on links managed by
1570 systemd-networkd using the ARP=no setting in the [Link] section of
1573 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1574 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1575 encode information about the result and exit codes of the current
1576 service runtime cycle.
1578 * systemd-sysctl will now configure kernel parameters in the order
1579 they occur in the configuration files. This matches what sysctl
1580 has been traditionally doing.
1582 * kernel-install "plugins" that are executed to perform various
1583 tasks after a new kernel is added and before an old one is removed
1584 can now return a special value to terminate the procedure and
1585 prevent any later plugins from running.
1587 * Journald's SplitMode=login setting has been deprecated. It has been
1588 removed from documentation, and its use is discouraged. In a future
1589 release it will be completely removed, and made equivalent to current
1590 default of SplitMode=uid.
1592 * Storage=both option setting in /etc/systemd/coredump.conf has been
1593 removed. With fast LZ4 compression storing the core dump twice is not
1596 * The --share-system systemd-nspawn option has been replaced with an
1597 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1598 this functionality is discouraged. In addition the variables
1599 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1600 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1601 individual namespaces.
1603 * "machinectl list" now shows the IP address of running containers in
1604 the output, as well as OS release information.
1606 * "loginctl list" now shows the TTY of each session in the output.
1608 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1609 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1610 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1611 tracking objects in a "recursive" mode, where a single client can be
1612 counted multiple times, if it takes multiple references.
1614 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1615 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1616 process using sd-bus automatically exit if the bus connection is
1619 * Bus clients of the service manager may now "pin" loaded units into
1620 memory, by taking an explicit reference on them. This is useful to
1621 ensure the client can retrieve runtime data about the service even
1622 after the service completed execution. Taking such a reference is
1623 available only for privileged clients and should be helpful to watch
1624 running services in a race-free manner, and in particular collect
1625 information about exit statuses and results.
1627 * The nss-resolve module has been changed to strictly return UNAVAIL
1628 when communication via D-Bus with resolved failed, and NOTFOUND when
1629 a lookup completed but was negative. This means it is now possible to
1630 neatly configure fallbacks using nsswitch.conf result checking
1631 expressions. Taking benefit of this, the new recommended
1632 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1634 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1636 * A new setting CtrlAltDelBurstAction= has been added to
1637 /etc/systemd/system.conf which may be used to configure the precise
1638 behaviour if the user on the console presses Ctrl-Alt-Del more often
1639 than 7 times in 2s. Previously this would unconditionally result in
1640 an expedited, immediate reboot. With this new setting the precise
1641 operation may be configured in more detail, and also turned off
1644 * In .netdev files two new settings RemoteChecksumTx= and
1645 RemoteChecksumRx= are now understood that permit configuring the
1646 remote checksumming logic for VXLAN networks.
1648 * The service manager learnt a new "invocation ID" concept for invoked
1649 services. Each runtime cycle of a service will get a new invocation
1650 ID (a 128bit random UUID) assigned that identifies the current
1651 run of the service uniquely and globally. A new invocation ID
1652 is generated each time a service starts up. The journal will store
1653 the invocation ID of a service along with any logged messages, thus
1654 making the invocation ID useful for matching the online runtime of a
1655 service with the offline log data it generated in a safe way without
1656 relying on synchronized timestamps. In many ways this new service
1657 invocation ID concept is similar to the kernel's boot ID concept that
1658 uniquely and globally identifies the runtime of each boot. The
1659 invocation ID of a service is passed to the service itself via an
1660 environment variable ($INVOCATION_ID). A new bus call
1661 GetUnitByInvocationID() has been added that is similar to GetUnit()
1662 but instead of retrieving the bus path for a unit by its name
1663 retrieves it by its invocation ID. The returned path is valid only as
1664 long as the passed invocation ID is current.
1666 * systemd-resolved gained a new "DNSStubListener" setting in
1667 resolved.conf. It either takes a boolean value or the special values
1668 "udp" and "tcp", and configures whether to enable the stub DNS
1669 listener on 127.0.0.53:53.
1671 * IP addresses configured via networkd may now carry additional
1672 configuration settings supported by the kernel. New options include:
1673 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1674 PrefixRoute=, AutoJoin=.
1676 * The PAM configuration fragment file for "user@.service" shipped with
1677 systemd (i.e. the --user instance of systemd) has been stripped to
1678 the minimum necessary to make the system boot. Previously, it
1679 contained Fedora-specific stanzas that did not apply to other
1680 distributions. It is expected that downstream distributions add
1681 additional configuration lines, matching their needs to this file,
1682 using it only as rough template of what systemd itself needs. Note
1683 that this reduced fragment does not even include an invocation of
1684 pam_limits which most distributions probably want to add, even though
1685 systemd itself does not need it. (There's also the new build time
1686 option --with-pamconfdir=no to disable installation of the PAM
1689 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1690 capability is now also dropped from its set (in addition to
1691 CAP_SYS_MKNOD as before).
1693 * In service unit files it is now possible to connect a specific named
1694 file descriptor with stdin/stdout/stdout of an executed service. The
1695 name may be specified in matching .socket units using the
1696 FileDescriptorName= setting.
1698 * A number of journal settings may now be configured on the kernel
1699 command line. Specifically, the following options are now understood:
1700 systemd.journald.max_level_console=,
1701 systemd.journald.max_level_store=,
1702 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1703 systemd.journald.max_level_wall=.
1705 * "systemctl is-enabled --full" will now show by which symlinks a unit
1706 file is enabled in the unit dependency tree.
1708 * Support for VeraCrypt encrypted partitions has been added to the
1709 "cryptsetup" logic and /etc/crypttab.
1711 * systemd-detect-virt gained support for a new --private-users switch
1712 that checks whether the invoking processes are running inside a user
1713 namespace. Similar, a new special value "private-users" for the
1714 existing ConditionVirtualization= setting has been added, permitting
1715 skipping of specific units in user namespace environments.
1717 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1718 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1719 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1720 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1721 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1722 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1723 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1724 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1725 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1726 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1727 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1728 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1729 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1730 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1731 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1732 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1733 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1734 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1735 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1736 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1737 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1738 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1739 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1740 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1741 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1742 Jędrzejewski-Szmek, Zeal Jagannatha
1744 — Santa Fe, 2016-11-03
1748 * In service units the various ExecXYZ= settings have been extended
1749 with an additional special character as first argument of the
1750 assigned value: if the character '+' is used the specified command
1751 line it will be run with full privileges, regardless of User=,
1752 Group=, CapabilityBoundingSet= and similar options. The effect is
1753 similar to the existing PermissionsStartOnly= option, but allows
1754 configuration of this concept for each executed command line
1757 * Services may now alter the service watchdog timeout at runtime by
1758 sending a WATCHDOG_USEC= message via sd_notify().
1760 * MemoryLimit= and related unit settings now optionally take percentage
1761 specifications. The percentage is taken relative to the amount of
1762 physical memory in the system (or in case of containers, the assigned
1763 amount of memory). This allows scaling service resources neatly with
1764 the amount of RAM available on the system. Similarly, systemd-logind's
1765 RuntimeDirectorySize= option now also optionally takes percentage
1768 * In similar fashion TasksMax= takes percentage values now, too. The
1769 value is taken relative to the configured maximum number of processes
1770 on the system. The per-service task maximum has been changed to 15%
1771 using this functionality. (Effectively this is an increase of 512 →
1772 4915 for service units, given the kernel's default pid_max setting.)
1774 * Calendar time specifications in .timer units now understand a ".."
1775 syntax for time ranges. Example: "4..7:10" may now be used for
1776 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1779 * The InaccessableDirectories=, ReadOnlyDirectories= and
1780 ReadWriteDirectories= unit file settings have been renamed to
1781 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1782 applied to all kinds of file nodes, and not just directories, with
1783 the exception of symlinks. Specifically these settings may now be
1784 used on block and character device nodes, UNIX sockets and FIFOS as
1785 well as regular files. The old names of these settings remain
1786 available for compatibility.
1788 * systemd will now log about all service processes it kills forcibly
1789 (using SIGKILL) because they remained after the clean shutdown phase
1790 of the service completed. This should help identifying services that
1791 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1792 systemd-logind's configuration a similar log message is generated for
1793 processes killed at the end of each session due to this setting.
1795 * systemd will now set the $JOURNAL_STREAM environment variable for all
1796 services whose stdout/stderr are connected to the Journal (which
1797 effectively means by default: all services). The variable contains
1798 the device and inode number of the file descriptor used for
1799 stdout/stderr. This may be used by invoked programs to detect whether
1800 their stdout/stderr is connected to the Journal, in which case they
1801 can switch over to direct Journal communication, thus being able to
1802 pass extended, structured metadata along with their log messages. As
1803 one example, this is now used by glib's logging primitives.
1805 * When using systemd's default tmp.mount unit for /tmp, the mount point
1806 will now be established with the "nosuid" and "nodev" options. This
1807 avoids privilege escalation attacks that put traps and exploits into
1808 /tmp. However, this might cause problems if you e. g. put container
1809 images or overlays into /tmp; if you need this, override tmp.mount's
1810 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1813 * systemd now supports the "memory" cgroup controller also on
1816 * The systemd-cgtop tool now optionally takes a control group path as
1817 command line argument. If specified, the control group list shown is
1818 limited to subgroups of that group.
1820 * The SystemCallFilter= unit file setting gained support for
1821 pre-defined, named system call filter sets. For example
1822 SystemCallFilter=@clock is now an effective way to make all clock
1823 changing-related system calls unavailable to a service. A number of
1824 similar pre-defined groups are defined. Writing system call filters
1825 for system services is simplified substantially with this new
1826 concept. Accordingly, all of systemd's own, long-running services now
1827 enable system call filtering based on this, by default.
1829 * A new service setting MemoryDenyWriteExecute= has been added, taking
1830 a boolean value. If turned on, a service may no longer create memory
1831 mappings that are writable and executable at the same time. This
1832 enhances security for services where this is enabled as it becomes
1833 harder to dynamically write and then execute memory in exploited
1834 service processes. This option has been enabled for all of systemd's
1835 own long-running services.
1837 * A new RestrictRealtime= service setting has been added, taking a
1838 boolean argument. If set the service's processes may no longer
1839 acquire realtime scheduling. This improves security as realtime
1840 scheduling may otherwise be used to easily freeze the system.
1842 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1843 value. This may be used for requesting that the system manager inside
1844 of the container reports start-up completion to nspawn which then
1845 propagates this notification further to the service manager
1846 supervising nspawn itself. A related option NotifyReady= in .nspawn
1847 files has been added too. This functionality allows ordering of the
1848 start-up of multiple containers using the usual systemd ordering
1851 * machinectl gained a new command "stop" that is an alias for
1854 * systemd-resolved gained support for contacting DNS servers on
1855 link-local IPv6 addresses.
1857 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1858 its caches. A method call for requesting the same operation has been
1859 added to the bus API too, and is made available via "systemd-resolve
1862 * systemd-resolve gained a new --status switch. If passed a brief
1863 summary of the used DNS configuration with per-interface information
1866 * resolved.conf gained a new Cache= boolean option, defaulting to
1867 on. If turned off local DNS caching is disabled. This comes with a
1868 performance penalty in particular when DNSSEC is enabled. Note that
1869 resolved disables its internal caching implicitly anyway, when the
1870 configured DNS server is on a host-local IP address such as ::1 or
1871 127.0.0.1, thus automatically avoiding double local caching.
1873 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1874 for DNS requests. This improves compatibility with local programs
1875 that do not use the libc NSS or systemd-resolved's bus APIs for name
1876 resolution. This minimal DNS service is only available to local
1877 programs and does not implement the full DNS protocol, but enough to
1878 cover local DNS clients. A new, static resolv.conf file, listing just
1879 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1880 now recommended to make /etc/resolv.conf a symlink to this file in
1881 order to route all DNS lookups to systemd-resolved, regardless if
1882 done via NSS, the bus API or raw DNS packets. Note that this local
1883 DNS service is not as fully featured as the libc NSS or
1884 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1885 used to deliver link-local address information (as this implies
1886 sending a local interface index along), LLMNR/mDNS support via this
1887 interface is severely restricted. It is thus strongly recommended for
1888 all applications to use the libc NSS API or native systemd-resolved
1891 * systemd-networkd's bridge support learned a new setting
1892 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1893 in .network files has been added for configuring VLAN bridging in
1894 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1896 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1897 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1898 now be acquired without relying on DHCPv6. Two new options
1899 UseDomains= and UseDNS= have been added to configure this behaviour.
1901 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1902 renamed IPv6AcceptRA=, without altering its behaviour. The old
1903 setting name remains available for compatibility reasons.
1905 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1906 Key=, InputKey= and OutputKey=.
1908 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1909 interface configuration.
1911 * "systemctl edit" may now be used to create new unit files by
1912 specifying the --force switch.
1914 * sd-event gained a new function sd_event_get_iteration() for
1915 requesting the current iteration counter of the event loop. It starts
1916 at zero and is increased by one with each event loop iteration.
1918 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1919 file. It can be used in lieu of %systemd_requires in packages which
1920 don't use any systemd functionality and are intended to be installed
1921 in minimal containers without systemd present. This macro provides
1922 ordering dependencies to ensure that if the package is installed in
1923 the same rpm transaction as systemd, systemd will be installed before
1924 the scriptlets for the package are executed, allowing unit presets
1927 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1928 been added to simplify packaging of generators.
1930 * The os-release file gained VERSION_CODENAME field for the
1931 distribution nickname (e.g. VERSION_CODENAME=woody).
1933 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1934 can be set to disable parsing of metadata and the creation
1935 of persistent symlinks for that device.
1937 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1938 to make them available to logged-in users has been reverted.
1940 * Much of the common code of the various systemd components is now
1941 built into an internal shared library libsystemd-shared-231.so
1942 (incorporating the systemd version number in the name, to be updated
1943 with future releases) that the components link to. This should
1944 decrease systemd footprint both in memory during runtime and on
1945 disk. Note that the shared library is not for public use, and is
1946 neither API not ABI stable, but is likely to change with every new
1947 released update. Packagers need to make sure that binaries
1948 linking to libsystemd-shared.so are updated in step with the
1951 * Configuration for "mkosi" is now part of the systemd
1952 repository. mkosi is a tool to easily build legacy-free OS images,
1953 and is available on github: https://github.com/systemd/mkosi. If
1954 "mkosi" is invoked in the build tree a new raw OS image is generated
1955 incorporating the systemd sources currently being worked on and a
1956 clean, fresh distribution installation. The generated OS image may be
1957 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1958 UEFI PC. This functionality is particularly useful to easily test
1959 local changes made to systemd in a pristine, defined environment. See
1960 doc/HACKING for details.
1962 * configure learned the --with-support-url= option to specify the
1963 distribution's bugtracker.
1965 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1966 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1967 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1968 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1969 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1970 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1971 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1972 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1973 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1974 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1975 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1976 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1977 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1978 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1979 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1980 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1981 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1982 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1983 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1985 — Berlin, 2016-07-25
1989 * DNSSEC is now turned on by default in systemd-resolved (in
1990 "allow-downgrade" mode), but may be turned off during compile time by
1991 passing "--with-default-dnssec=no" to "configure" (and of course,
1992 during runtime with DNSSEC= in resolved.conf). We recommend
1993 downstreams to leave this on at least during development cycles and
1994 report any issues with the DNSSEC logic upstream. We are very
1995 interested in collecting feedback about the DNSSEC validator and its
1996 limitations in the wild. Note however, that DNSSEC support is
1997 probably nothing downstreams should turn on in stable distros just
1998 yet, as it might create incompatibilities with a few DNS servers and
1999 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2000 automatically whenever we detect such incompatible setups, but there
2001 might be systems we do not cover yet. Hence: please help us testing
2002 the DNSSEC code, leave this on where you can, report back, but then
2003 again don't consider turning this on in your stable, LTS or
2004 production release just yet. (Note that you have to enable
2005 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2006 and its DNSSEC mode for host name resolution from local
2009 * systemd-resolve conveniently resolves DANE records with the --tlsa
2010 option and OPENPGPKEY records with the --openpgp option. It also
2011 supports dumping raw DNS record data via the new --raw= switch.
2013 * systemd-logind will now by default terminate user processes that are
2014 part of the user session scope unit (session-XX.scope) when the user
2015 logs out. This behavior is controlled by the KillUserProcesses=
2016 setting in logind.conf, and the previous default of "no" is now
2017 changed to "yes". This means that user sessions will be properly
2018 cleaned up after, but additional steps are necessary to allow
2019 intentionally long-running processes to survive logout.
2021 While the user is logged in at least once, user@.service is running,
2022 and any service that should survive the end of any individual login
2023 session can be started at a user service or scope using systemd-run.
2024 systemd-run(1) man page has been extended with an example which shows
2025 how to run screen in a scope unit underneath user@.service. The same
2026 command works for tmux.
2028 After the user logs out of all sessions, user@.service will be
2029 terminated too, by default, unless the user has "lingering" enabled.
2030 To effectively allow users to run long-term tasks even if they are
2031 logged out, lingering must be enabled for them. See loginctl(1) for
2032 details. The default polkit policy was modified to allow users to
2033 set lingering for themselves without authentication.
2035 Previous defaults can be restored at compile time by the
2036 --without-kill-user-processes option to "configure".
2038 * systemd-logind gained new configuration settings SessionsMax= and
2039 InhibitorsMax=, both with a default of 8192. It will not register new
2040 user sessions or inhibitors above this limit.
2042 * systemd-logind will now reload configuration on SIGHUP.
2044 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2045 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2046 enable. Also, support for the "io" cgroup controller in the unified
2047 hierarchy has been added, so that the "memory", "pids" and "io" are
2048 now the controllers that are supported on the unified hierarchy.
2050 WARNING: it is not possible to use previous systemd versions with
2051 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2052 is necessary to also update systemd in the initramfs if using the
2053 unified hierarchy. An updated SELinux policy is also required.
2055 * LLDP support has been extended, and both passive (receive-only) and
2056 active (sender) modes are supported. Passive mode ("routers-only") is
2057 enabled by default in systemd-networkd. Active LLDP mode is enabled
2058 by default for containers on the internal network. The "networkctl
2059 lldp" command may be used to list information gathered. "networkctl
2060 status" will also show basic LLDP information on connected peers now.
2062 * The IAID and DUID unique identifier sent in DHCP requests may now be
2063 configured for the system and each .network file managed by
2064 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2066 * systemd-networkd gained support for configuring proxy ARP support for
2067 each interface, via the ProxyArp= setting in .network files. It also
2068 gained support for configuring the multicast querier feature of
2069 bridge devices, via the new MulticastQuerier= setting in .netdev
2070 files. Similarly, snooping on the IGMP traffic can be controlled
2071 via the new setting MulticastSnooping=.
2073 A new setting PreferredLifetime= has been added for addresses
2074 configured in .network file to configure the lifetime intended for an
2077 The systemd-networkd DHCP server gained the option EmitRouter=, which
2078 defaults to yes, to configure whether the DHCP Option 3 (Router)
2081 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2082 systemd-socket-activate and installed into /usr/bin. It is now fully
2085 * systemd-journald now uses separate threads to flush changes to disk
2086 when closing journal files, thus reducing impact of slow disk I/O on
2087 logging performance.
2089 * The sd-journal API gained two new calls
2090 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2091 can be used to open journal files using file descriptors instead of
2092 file or directory paths. sd_journal_open_container() has been
2093 deprecated, sd_journal_open_directory_fd() should be used instead
2094 with the flag SD_JOURNAL_OS_ROOT.
2096 * journalctl learned a new output mode "-o short-unix" that outputs log
2097 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2098 UTC). It also gained support for a new --no-hostname setting to
2099 suppress the hostname column in the family of "short" output modes.
2101 * systemd-ask-password now optionally skips printing of the password to
2102 stdout with --no-output which can be useful in scripts.
2104 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2105 (devices tagged with ID_MAKER_TOOL) are now tagged with
2106 "uaccess" and are available to logged in users.
2108 * The DeviceAllow= unit setting now supports specifiers (with "%").
2110 * "systemctl show" gained a new --value switch, which allows print a
2111 only the contents of a specific unit property, without also printing
2112 the property's name. Similar support was added to "show*" verbs
2113 of loginctl and machinectl that output "key=value" lists.
2115 * A new unit type "generated" was added for files dynamically generated
2116 by generator tools. Similarly, a new unit type "transient" is used
2117 for unit files created using the runtime API. "systemctl enable" will
2118 refuse to operate on such files.
2120 * A new command "systemctl revert" has been added that may be used to
2121 revert to the vendor version of a unit file, in case local changes
2122 have been made by adding drop-ins or overriding the unit file.
2124 * "machinectl clean" gained a new verb to automatically remove all or
2125 just hidden container images.
2127 * systemd-tmpfiles gained support for a new line type "e" for emptying
2128 directories, if they exist, without creating them if they don't.
2130 * systemd-nspawn gained support for automatically patching the UID/GIDs
2131 of the owners and the ACLs of all files and directories in a
2132 container tree to match the UID/GID user namespacing range selected
2133 for the container invocation. This mode is enabled via the new
2134 --private-users-chown switch. It also gained support for
2135 automatically choosing a free, previously unused UID/GID range when
2136 starting a container, via the new --private-users=pick setting (which
2137 implies --private-users-chown). Together, these options for the first
2138 time make user namespacing for nspawn containers fully automatic and
2139 thus deployable. The systemd-nspawn@.service template unit file has
2140 been changed to use this functionality by default.
2142 * systemd-nspawn gained a new --network-zone= switch, that allows
2143 creating ad-hoc virtual Ethernet links between multiple containers,
2144 that only exist as long as at least one container referencing them is
2145 running. This allows easy connecting of multiple containers with a
2146 common link that implements an Ethernet broadcast domain. Each of
2147 these network "zones" may be named relatively freely by the user, and
2148 may be referenced by any number of containers, but each container may
2149 only reference one of these "zones". On the lower level, this is
2150 implemented by an automatically managed bridge network interface for
2151 each zone, that is created when the first container referencing its
2152 zone is created and removed when the last one referencing its zone
2155 * The default start timeout may now be configured on the kernel command
2156 line via systemd.default_timeout_start_sec=. It was already
2157 configurable via the DefaultTimeoutStartSec= option in
2158 /etc/systemd/system.conf.
2160 * Socket units gained a new TriggerLimitIntervalSec= and
2161 TriggerLimitBurst= setting to configure a limit on the activation
2162 rate of the socket unit.
2164 * The LimitNICE= setting now optionally takes normal UNIX nice values
2165 in addition to the raw integer limit value. If the specified
2166 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2167 value is understood as UNIX nice value. If not prefixed like this it
2168 is understood as raw RLIMIT_NICE limit.
2170 * Note that the effect of the PrivateDevices= unit file setting changed
2171 slightly with this release: the per-device /dev file system will be
2172 mounted read-only from this version on, and will have "noexec"
2173 set. This (minor) change of behavior might cause some (exceptional)
2174 legacy software to break, when PrivateDevices=yes is set for its
2175 service. Please leave PrivateDevices= off if you run into problems
2178 * systemd-bootchart has been split out to a separate repository:
2179 https://github.com/systemd/systemd-bootchart
2181 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2182 merged into the kernel in its current form.
2184 * The compatibility libraries libsystemd-daemon.so,
2185 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2186 which have been deprecated since systemd-209 have been removed along
2187 with the corresponding pkg-config files. All symbols provided by
2188 those libraries are provided by libsystemd.so.
2190 * The Capabilities= unit file setting has been removed (it is ignored
2191 for backwards compatibility). AmbientCapabilities= and
2192 CapabilityBoundingSet= should be used instead.
2194 * A new special target has been added, initrd-root-device.target,
2195 which creates a synchronization point for dependencies of the root
2196 device in early userspace. Initramfs builders must ensure that this
2197 target is now included in early userspace.
2199 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2200 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2201 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2202 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2203 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2204 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2205 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2206 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2207 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2208 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2209 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2210 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2211 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2212 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2213 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2214 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2215 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2216 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2217 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2218 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2219 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2220 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2221 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2222 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2223 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2226 — Fairfax, 2016-05-21
2230 * The systemd-resolved DNS resolver service has gained a substantial
2231 set of new features, most prominently it may now act as a DNSSEC
2232 validating stub resolver. DNSSEC mode is currently turned off by
2233 default, but is expected to be turned on by default in one of the
2234 next releases. For now, we invite everybody to test the DNSSEC logic
2235 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2236 service also gained a full set of D-Bus interfaces, including calls
2237 to configure DNS and DNSSEC settings per link (for use by external
2238 network management software). systemd-resolved and systemd-networkd
2239 now distinguish between "search" and "routing" domains. The former
2240 are used to qualify single-label names, the latter are used purely
2241 for routing lookups within certain domains to specific links.
2242 resolved now also synthesizes RRs for all entries from /etc/hosts.
2244 * The systemd-resolve tool (which is a client utility for
2245 systemd-resolved) has been improved considerably and is now fully
2246 supported and documented. Hence it has moved from /usr/lib/systemd to
2249 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2252 * The coredump collection logic has been reworked: when a coredump is
2253 collected it is now written to disk, compressed and processed
2254 (including stacktrace extraction) from a new instantiated service
2255 systemd-coredump@.service, instead of directly from the
2256 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2257 processing large coredumps can take up a substantial amount of
2258 resources and time, and this previously happened entirely outside of
2259 systemd's service supervision. With the new logic the core_pattern
2260 hook only does minimal metadata collection before passing off control
2261 to the new instantiated service, which is configured with a time
2262 limit, a nice level and other settings to minimize negative impact on
2263 the rest of the system. Also note that the new logic will honour the
2264 RLIMIT_CORE setting of the crashed process, which now allows users
2265 and processes to turn off coredumping for their processes by setting
2268 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2269 and all forked processes by default. Previously, PID 1 would leave
2270 the setting at "0" for all processes, as set by the kernel. Note that
2271 the resource limit traditionally has no effect on the generated
2272 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2273 logic is used. Since the limit is now honoured (see above) its
2274 default has been changed so that the coredumping logic is enabled by
2275 default for all processes, while allowing specific opt-out.
2277 * When the stacktrace is extracted from processes of system users, this
2278 is now done as "systemd-coredump" user, in order to sandbox this
2279 potentially security sensitive parsing operation. (Note that when
2280 processing coredumps of normal users this is done under the user ID
2281 of process that crashed, as before.) Packagers should take notice
2282 that it is now necessary to create the "systemd-coredump" system user
2283 and group at package installation time.
2285 * The systemd-activate socket activation testing tool gained support
2286 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2287 and --seqpacket switches. It also has been extended to support both
2288 new-style and inetd-style file descriptor passing. Use the new
2289 --inetd switch to request inetd-style file descriptor passing.
2291 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2292 variable, which takes a boolean value. If set to false, ANSI color
2293 output is disabled in the tools even when run on a terminal that
2296 * The VXLAN support in networkd now supports two new settings
2297 DestinationPort= and PortRange=.
2299 * A new systemd.machine_id= kernel command line switch has been added,
2300 that may be used to set the machine ID in /etc/machine-id if it is
2301 not initialized yet. This command line option has no effect if the
2302 file is already initialized.
2304 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2305 specified command line as PID 2 rather than PID 1 in the
2306 container. In this mode PID 1 is a minimal stub init process that
2307 implements the special POSIX and Linux semantics of PID 1 regarding
2308 signal and child process management. Note that this stub init process
2309 is implemented in nspawn itself and requires no support from the
2310 container image. This new logic is useful to support running
2311 arbitrary commands in the container, as normal processes are
2312 generally not prepared to run as PID 1.
2314 * systemd-nspawn gained a new --chdir= switch for setting the current
2315 working directory for the process started in the container.
2317 * "journalctl /dev/sda" will now output all kernel log messages for
2318 specified device from the current boot, in addition to all devices
2319 that are parents of it. This should make log output about devices
2320 pretty useful, as long as kernel drivers attach enough metadata to
2321 the log messages. (The usual SATA drivers do.)
2323 * The sd-journal API gained two new calls
2324 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2325 that report whether log data from /run or /var has been found.
2327 * journalctl gained a new switch "--fields" that prints all journal
2328 record field names currently in use in the journal. This is backed
2329 by two new sd-journal API calls sd_journal_enumerate_fields() and
2330 sd_journal_restart_fields().
2332 * Most configurable timeouts in systemd now expect an argument of
2333 "infinity" to turn them off, instead of "0" as before. The semantics
2334 from now on is that a timeout of "0" means "now", and "infinity"
2335 means "never". To maintain backwards compatibility, "0" continues to
2336 turn off previously existing timeout settings.
2338 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2339 try-reload-or-restart" to clarify what it actually does: the "try"
2340 logic applies to both reloading and restarting, not just restarting.
2341 The old name continues to be accepted for compatibility.
2343 * On boot-up, when PID 1 detects that the system clock is behind the
2344 release date of the systemd version in use, the clock is now set
2345 to the latter. Previously, this was already done in timesyncd, in order
2346 to avoid running with clocks set to the various clock epochs such as
2347 1902, 1938 or 1970. With this change the logic is now done in PID 1
2348 in addition to timesyncd during early boot-up, so that it is enforced
2349 before the first process is spawned by systemd. Note that the logic
2350 in timesyncd remains, as it is more comprehensive and ensures
2351 clock monotonicity by maintaining a persistent timestamp file in
2352 /var. Since /var is generally not available in earliest boot or the
2353 initrd, this part of the logic remains in timesyncd, and is not done
2356 * Support for tweaking details in net_cls.class_id through the
2357 NetClass= configuration directive has been removed, as the kernel
2358 people have decided to deprecate that controller in cgroup v2.
2359 Userspace tools such as nftables are moving over to setting rules
2360 that are specific to the full cgroup path of a task, which obsoletes
2361 these controllers anyway. The NetClass= directive is kept around for
2362 legacy compatibility reasons. For a more in-depth description of the
2363 kernel change, please refer to the respective upstream commit:
2365 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2367 * A new service setting RuntimeMaxSec= has been added that may be used
2368 to specify a maximum runtime for a service. If the timeout is hit, the
2369 service is terminated and put into a failure state.
2371 * A new service setting AmbientCapabilities= has been added. It allows
2372 configuration of additional Linux process capabilities that are
2373 passed to the activated processes. This is only available on very
2376 * The process resource limit settings in service units may now be used
2377 to configure hard and soft limits individually.
2379 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2380 expose support for gcc's __attribute__((cleanup())) C extension.
2381 Specifically, for many object destructor functions alternative
2382 versions have been added that have names suffixed with "p" and take a
2383 pointer to a pointer to the object to destroy, instead of just a
2384 pointer to the object itself. This is useful because these destructor
2385 functions may be used directly as parameters to the cleanup
2386 construct. Internally, systemd has been a heavy user of this GCC
2387 extension for a long time, and with this change similar support is
2388 now available to consumers of the library outside of systemd. Note
2389 that by using this extension in your sources compatibility with old
2390 and strictly ANSI compatible C compilers is lost. However, all gcc or
2391 LLVM versions of recent years support this extension.
2393 * Timer units gained support for a new setting RandomizedDelaySec= that
2394 allows configuring some additional randomized delay to the configured
2395 time. This is useful to spread out timer events to avoid load peaks in
2396 clusters or larger setups.
2398 * Calendar time specifications now support sub-second accuracy.
2400 * Socket units now support listening on SCTP and UDP-lite protocol
2403 * The sd-event API now comes with a full set of man pages.
2405 * Older versions of systemd contained experimental support for
2406 compressing journal files and coredumps with the LZ4 compressor that
2407 was not compatible with the lz4 binary (due to API limitations of the
2408 lz4 library). This support has been removed; only support for files
2409 compatible with the lz4 binary remains. This LZ4 logic is now
2410 officially supported and no longer considered experimental.
2412 * The dkr image import logic has been removed again from importd. dkr's
2413 micro-services focus doesn't fit into the machine image focus of
2414 importd, and quickly got out of date with the upstream dkr API.
2416 * Creation of the /run/lock/lockdev/ directory was dropped from
2417 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2418 been available for many years. If you still need this, you need to
2419 create your own tmpfiles.d config file with:
2421 d /run/lock/lockdev 0775 root lock -
2423 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2424 and RebootArgument= have been moved from the [Service] section of
2425 unit files to [Unit], and they are now supported on all unit types,
2426 not just service units. Of course, elogind will continue to
2427 understand these settings also at the old location, in order to
2428 maintain compatibility.
2430 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2431 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2432 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2433 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2434 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2435 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2436 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2437 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2438 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2439 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2440 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2441 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2442 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2443 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2444 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2445 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2446 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2447 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2448 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2450 — Berlin, 2016-02-11
2454 * A number of properties previously only settable in unit
2455 files are now also available as properties to set when
2456 creating transient units programmatically via the bus, as it
2457 is exposed with systemd-run's --property=
2458 setting. Specifically, these are: SyslogIdentifier=,
2459 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2460 EnvironmentFile=, ReadWriteDirectories=,
2461 ReadOnlyDirectories=, InaccessibleDirectories=,
2462 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2464 * When creating transient services via the bus API it is now
2465 possible to pass in a set of file descriptors to use as
2466 STDIN/STDOUT/STDERR for the invoked process.
2468 * Slice units may now be created transiently via the bus APIs,
2469 similar to the way service and scope units may already be
2470 created transiently.
2472 * Wherever systemd expects a calendar timestamp specification
2473 (like in journalctl's --since= and --until= switches) UTC
2474 timestamps are now supported. Timestamps suffixed with "UTC"
2475 are now considered to be in Universal Time Coordinated
2476 instead of the local timezone. Also, timestamps may now
2477 optionally be specified with sub-second accuracy. Both of
2478 these additions also apply to recurring calendar event
2479 specification, such as OnCalendar= in timer units.
2481 * journalctl gained a new "--sync" switch that asks the
2482 journal daemon to write all so far unwritten log messages to
2483 disk and sync the files, before returning.
2485 * systemd-tmpfiles learned two new line types "q" and "Q" that
2486 operate like "v", but also set up a basic btrfs quota
2487 hierarchy when used on a btrfs file system with quota
2490 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2491 instead of a subvolume (even on a btrfs file system) if the
2492 root directory is a plain directory, and not a
2493 subvolume. This should simplify things with certain chroot()
2494 environments which are not aware of the concept of btrfs
2497 * systemd-detect-virt gained a new --chroot switch to detect
2498 whether execution takes place in a chroot() environment.
2500 * CPUAffinity= now takes CPU index ranges in addition to
2503 * The various memory-related resource limit settings (such as
2504 LimitAS=) now understand the usual K, M, G, ... suffixes to
2505 the base of 1024 (IEC). Similar, the time-related resource
2506 limit settings understand the usual min, h, day, ...
2509 * There's a new system.conf setting DefaultTasksMax= to
2510 control the default TasksMax= setting for services and
2511 scopes running on the system. (TasksMax= is the primary
2512 setting that exposes the "pids" cgroup controller on systemd
2513 and was introduced in the previous systemd release.) The
2514 setting now defaults to 512, which means services that are
2515 not explicitly configured otherwise will only be able to
2516 create 512 processes or threads at maximum, from this
2517 version on. Note that this means that thread- or
2518 process-heavy services might need to be reconfigured to set
2519 TasksMax= to a higher value. It is sufficient to set
2520 TasksMax= in these specific unit files to a higher value, or
2521 even "infinity". Similar, there's now a logind.conf setting
2522 UserTasksMax= that defaults to 4096 and limits the total
2523 number of processes or tasks each user may own
2524 concurrently. nspawn containers also have the TasksMax=
2525 value set by default now, to 8192. Note that all of this
2526 only has an effect if the "pids" cgroup controller is
2527 enabled in the kernel. The general benefit of these changes
2528 should be a more robust and safer system, that provides a
2529 certain amount of per-service fork() bomb protection.
2531 * systemd-nspawn gained the new --network-veth-extra= switch
2532 to define additional and arbitrarily-named virtual Ethernet
2533 links between the host and the container.
2535 * A new service execution setting PassEnvironment= has been
2536 added that allows importing select environment variables
2537 from PID1's environment block into the environment block of
2540 * Timer units gained support for a new RemainAfterElapse=
2541 setting which takes a boolean argument. It defaults to on,
2542 exposing behaviour unchanged to previous releases. If set to
2543 off, timer units are unloaded after they elapsed if they
2544 cannot elapse again. This is particularly useful for
2545 transient timer units, which shall not stay around longer
2546 than until they first elapse.
2548 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2549 default now (the kernel default is 16). This is beneficial
2550 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2551 allows substantially larger numbers of queued
2552 datagrams. This should increase the capability of systemd to
2553 parallelize boot-up, as logging and sd_notify() are unlikely
2554 to stall execution anymore. If you need to change the value
2555 from the new defaults, use the usual sysctl.d/ snippets.
2557 * The compression framing format used by the journal or
2558 coredump processing has changed to be in line with what the
2559 official LZ4 tools generate. LZ4 compression support in
2560 systemd was considered unsupported previously, as the format
2561 was not compatible with the normal tools. With this release
2562 this has changed now, and it is hence safe for downstream
2563 distributions to turn it on. While not compressing as well
2564 as the XZ, LZ4 is substantially faster, which makes
2565 it a good default choice for the compression logic in the
2566 journal and in coredump handling.
2568 * Any reference to /etc/mtab has been dropped from
2569 systemd. The file has been obsolete since a while, but
2570 systemd refused to work on systems where it was incorrectly
2571 set up (it should be a symlink or non-existent). Please make
2572 sure to update to util-linux 2.27.1 or newer in conjunction
2573 with this systemd release, which also drops any reference to
2574 /etc/mtab. If you maintain a distribution make sure that no
2575 software you package still references it, as this is a
2576 likely source of bugs. There's also a glibc bug pending,
2577 asking for removal of any reference to this obsolete file:
2579 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2581 Note that only util-linux versions built with
2582 --enable-libmount-force-mountinfo are supported.
2584 * Support for the ".snapshot" unit type has been removed. This
2585 feature turned out to be little useful and little used, and
2586 has now been removed from the core and from systemctl.
2588 * The dependency types RequiresOverridable= and
2589 RequisiteOverridable= have been removed from systemd. They
2590 have been used only very sparingly to our knowledge and
2591 other options that provide a similar effect (such as
2592 systemctl --mode=ignore-dependencies) are much more useful
2593 and commonly used. Moreover, they were only half-way
2594 implemented as the option to control behaviour regarding
2595 these dependencies was never added to systemctl. By removing
2596 these dependency types the execution engine becomes a bit
2597 simpler. Unit files that use these dependencies should be
2598 changed to use the non-Overridable dependency types
2599 instead. In fact, when parsing unit files with these
2600 options, that's what systemd will automatically convert them
2601 too, but it will also warn, asking users to fix the unit
2602 files accordingly. Removal of these dependency types should
2603 only affect a negligible number of unit files in the wild.
2605 * Behaviour of networkd's IPForward= option changed
2606 (again). It will no longer maintain a per-interface setting,
2607 but propagate one way from interfaces where this is enabled
2608 to the global kernel setting. The global setting will be
2609 enabled when requested by a network that is set up, but
2610 never be disabled again. This change was made to make sure
2611 IPv4 and IPv6 behaviour regarding packet forwarding is
2612 similar (as the Linux IPv6 stack does not support
2613 per-interface control of this setting) and to minimize
2616 * In unit files the behaviour of %u, %U, %h, %s has
2617 changed. These specifiers will now unconditionally resolve
2618 to the various user database fields of the user that the
2619 systemd instance is running as, instead of the user
2620 configured in the specific unit via User=. Note that this
2621 effectively doesn't change much, as resolving of these
2622 specifiers was already turned off in the --system instance
2623 of systemd, as we cannot do NSS lookups from PID 1. In the
2624 --user instance of systemd these specifiers where correctly
2625 resolved, but hardly made any sense, since the user instance
2626 lacks privileges to do user switches anyway, and User= is
2627 hence useless. Moreover, even in the --user instance of
2628 systemd behaviour was awkward as it would only take settings
2629 from User= assignment placed before the specifier into
2630 account. In order to unify and simplify the logic around
2631 this the specifiers will now always resolve to the
2632 credentials of the user invoking the manager (which in case
2633 of PID 1 is the root user).
2635 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2636 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2637 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2638 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2639 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2640 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2641 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2642 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2643 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2644 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2645 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2646 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2647 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2648 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2651 — Berlin, 2015-11-18
2655 * systemd now depends on util-linux v2.27. More specifically,
2656 the newly added mount monitor feature in libmount now
2657 replaces systemd's former own implementation.
2659 * libmount mandates /etc/mtab not to be regular file, and
2660 systemd now enforces this condition at early boot.
2661 /etc/mtab has been deprecated and warned about for a very
2662 long time, so systems running systemd should already have
2663 stopped having this file around as anything else than a
2664 symlink to /proc/self/mounts.
2666 * Support for the "pids" cgroup controller has been added. It
2667 allows accounting the number of tasks in a cgroup and
2668 enforcing limits on it. This adds two new setting
2669 TasksAccounting= and TasksMax= to each unit, as well as a
2670 global option DefaultTasksAccounting=.
2672 * Support for the "net_cls" cgroup controller has been added.
2673 It allows assigning a net class ID to each task in the
2674 cgroup, which can then be used in firewall rules and traffic
2675 shaping configurations. Note that the kernel netfilter net
2676 class code does not currently work reliably for ingress
2677 packets on unestablished sockets.
2679 This adds a new config directive called NetClass= to CGroup
2680 enabled units. Allowed values are positive numbers for fixed
2681 assignments and "auto" for picking a free value
2684 * 'systemctl is-system-running' now returns 'offline' if the
2685 system is not booted with systemd. This command can now be
2686 used as a substitute for 'systemd-notify --booted'.
2688 * Watchdog timeouts have been increased to 3 minutes for all
2689 in-tree service files. Apparently, disk IO issues are more
2690 frequent than we hoped, and user reported >1 minute waiting
2693 * 'machine-id-commit' functionality has been merged into
2694 'machine-id-setup --commit'. The separate binary has been
2697 * The WorkingDirectory= directive in unit files may now be set
2698 to the special value '~'. In this case, the working
2699 directory is set to the home directory of the user
2700 configured in User=.
2702 * "machinectl shell" will now open the shell in the home
2703 directory of the selected user by default.
2705 * The CrashChVT= configuration file setting is renamed to
2706 CrashChangeVT=, following our usual logic of not
2707 abbreviating unnecessarily. The old directive is still
2708 supported for compat reasons. Also, this directive now takes
2709 an integer value between 1 and 63, or a boolean value. The
2710 formerly supported '-1' value for disabling stays around for
2713 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2714 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2715 RootDirectory= properties can now be set for transient
2718 * The systemd-analyze tool gained a new "set-log-target" verb
2719 to change the logging target the system manager logs to
2720 dynamically during runtime. This is similar to how
2721 "systemd-analyze set-log-level" already changes the log
2724 * In nspawn /sys is now mounted as tmpfs, with only a selected
2725 set of subdirectories mounted in from the real sysfs. This
2726 enhances security slightly, and is useful for ensuring user
2727 namespaces work correctly.
2729 * Support for USB FunctionFS activation has been added. This
2730 allows implementation of USB gadget services that are
2731 activated as soon as they are requested, so that they don't
2732 have to run continuously, similar to classic socket
2735 * The "systemctl exit" command now optionally takes an
2736 additional parameter that sets the exit code to return from
2737 the systemd manager when exiting. This is only relevant when
2738 running the systemd user instance, or when running the
2739 system instance in a container.
2741 * sd-bus gained the new API calls sd_bus_path_encode_many()
2742 and sd_bus_path_decode_many() that allow easy encoding and
2743 decoding of multiple identifier strings inside a D-Bus
2744 object path. Another new call sd_bus_default_flush_close()
2745 has been added to flush and close per-thread default
2748 * systemd-cgtop gained support for a -M/--machine= switch to
2749 show the control groups within a certain container only.
2751 * "systemctl kill" gained support for an optional --fail
2752 switch. If specified the requested operation will fail of no
2753 processes have been killed, because the unit had no
2754 processes attached, or similar.
2756 * A new systemd.crash_reboot=1 kernel command line option has
2757 been added that triggers a reboot after crashing. This can
2758 also be set through CrashReboot= in systemd.conf.
2760 * The RuntimeDirectory= setting now understands unit
2761 specifiers like %i or %f.
2763 * A new (still internal) library API sd-ipv4acd has been added,
2764 that implements address conflict detection for IPv4. It's
2765 based on code from sd-ipv4ll, and will be useful for
2766 detecting DHCP address conflicts.
2768 * File descriptors passed during socket activation may now be
2769 named. A new API sd_listen_fds_with_names() is added to
2770 access the names. The default names may be overridden,
2771 either in the .socket file using the FileDescriptorName=
2772 parameter, or by passing FDNAME= when storing the file
2773 descriptors using sd_notify().
2775 * systemd-networkd gained support for:
2777 - Setting the IPv6 Router Advertisement settings via
2778 IPv6AcceptRouterAdvertisements= in .network files.
2780 - Configuring the HelloTimeSec=, MaxAgeSec= and
2781 ForwardDelaySec= bridge parameters in .netdev files.
2783 - Configuring PreferredSource= for static routes in
2786 * The "ask-password" framework used to query for LUKS harddisk
2787 passwords or SSL passwords during boot gained support for
2788 caching passwords in the kernel keyring, if it is
2789 available. This makes sure that the user only has to type in
2790 a passphrase once if there are multiple objects to unlock
2791 with the same one. Previously, such password caching was
2792 available only when Plymouth was used; this moves the
2793 caching logic into the systemd codebase itself. The
2794 "systemd-ask-password" utility gained a new --keyname=
2795 switch to control which kernel keyring key to use for
2796 caching a password in. This functionality is also useful for
2797 enabling display managers such as gdm to automatically
2798 unlock the user's GNOME keyring if its passphrase, the
2799 user's password and the harddisk password are the same, if
2800 gdm-autologin is used.
2802 * When downloading tar or raw images using "machinectl
2803 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2804 file is now also downloaded, if it is available and stored
2805 next to the image file.
2807 * Units of type ".socket" gained a new boolean setting
2808 Writable= which is only useful in conjunction with
2809 ListenSpecial=. If true, enables opening the specified
2810 special file in O_RDWR mode rather than O_RDONLY mode.
2812 * systemd-rfkill has been reworked to become a singleton
2813 service that is activated through /dev/rfkill on each rfkill
2814 state change and saves the settings to disk. This way,
2815 systemd-rfkill is now compatible with devices that exist
2816 only intermittendly, and even restores state if the previous
2817 system shutdown was abrupt rather than clean.
2819 * The journal daemon gained support for vacuuming old journal
2820 files controlled by the number of files that shall remain,
2821 in addition to the already existing control by size and by
2822 date. This is useful as journal interleaving performance
2823 degrades with too many separate journal files, and allows
2824 putting an effective limit on them. The new setting defaults
2825 to 100, but this may be changed by setting SystemMaxFiles=
2826 and RuntimeMaxFiles= in journald.conf. Also, the
2827 "journalctl" tool gained the new --vacuum-files= switch to
2828 manually vacuum journal files to leave only the specified
2829 number of files in place.
2831 * udev will now create /dev/disk/by-path links for ATA devices
2832 on kernels where that is supported.
2834 * Galician, Serbian, Turkish and Korean translations were added.
2836 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2837 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2838 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2839 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2840 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2841 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2842 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2843 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2844 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2845 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2846 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2847 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2848 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2849 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2850 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2851 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2852 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2853 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2855 — Berlin, 2015-10-07
2859 * The DHCP implementation of systemd-networkd gained a set of
2862 - The DHCP server now supports emitting DNS and NTP
2863 information. It may be enabled and configured via
2864 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2865 and NTP information is enabled, but no servers are
2866 configured, the corresponding uplink information (if there
2867 is any) is propagated.
2869 - Server and client now support transmission and reception
2870 of timezone information. It can be configured via the
2871 newly introduced network options UseTimezone=,
2872 EmitTimezone=, and Timezone=. Transmission of timezone
2873 information is enabled between host and containers by
2874 default now: the container will change its local timezone
2875 to what the host has set.
2877 - Lease timeouts can now be configured via
2878 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2880 - The DHCP server improved on the stability of
2881 leases. Clients are more likely to get the same lease
2882 information back, even if the server loses state.
2884 - The DHCP server supports two new configuration options to
2885 control the lease address pool metrics, PoolOffset= and
2888 * The encapsulation limit of tunnels in systemd-networkd may
2889 now be configured via 'EncapsulationLimit='. It allows
2890 modifying the maximum additional levels of encapsulation
2891 that are permitted to be prepended to a packet.
2893 * systemd now supports the concept of user buses replacing
2894 session buses, if used with dbus-1.10 (and enabled via dbus
2895 --enable-user-session). It previously only supported this on
2896 kdbus-enabled systems, and this release expands this to
2897 'dbus-daemon' systems.
2899 * systemd-networkd now supports predictable interface names
2902 * systemd now optionally supports the new Linux kernel
2903 "unified" control group hierarchy. If enabled via the kernel
2904 command-line option 'systemd.unified_cgroup_hierarchy=1',
2905 systemd will try to mount the unified cgroup hierarchy
2906 directly on /sys/fs/cgroup. If not enabled, or not
2907 available, systemd will fall back to the legacy cgroup
2908 hierarchy setup, as before. Host system and containers can
2909 mix and match legacy and unified hierarchies as they
2910 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2911 environment variable to individually select the hierarchy to
2912 use for executed containers. By default, nspawn will use the
2913 unified hierarchy for the containers if the host uses the
2914 unified hierarchy, and the legacy hierarchy otherwise.
2915 Please note that at this point the unified hierarchy is an
2916 experimental kernel feature and is likely to change in one
2917 of the next kernel releases. Therefore, it should not be
2918 enabled by default in downstream distributions yet. The
2919 minimum required kernel version for the unified hierarchy to
2920 work is 4.2. Note that when the unified hierarchy is used
2921 for the first time delegated access to controllers is
2922 safe. Because of this systemd-nspawn containers will get
2923 access to controllers now, as will systemd user
2924 sessions. This means containers and user sessions may now
2925 manage their own resources, partitioning up what the system
2928 * A new special scope unit "init.scope" has been introduced
2929 that encapsulates PID 1 of the system. It may be used to
2930 determine resource usage and enforce resource limits on PID
2931 1 itself. PID 1 hence moved out of the root of the control
2934 * The cgtop tool gained support for filtering out kernel
2935 threads when counting tasks in a control group. Also, the
2936 count of processes is now recursively summed up by
2937 default. Two options -k and --recursive= have been added to
2938 revert to old behaviour. The tool has also been updated to
2939 work correctly in containers now.
2941 * systemd-nspawn's --bind= and --bind-ro= options have been
2942 extended to allow creation of non-recursive bind mounts.
2944 * libsystemd gained two new calls sd_pid_get_cgroup() and
2945 sd_peer_get_cgroup() which return the control group path of
2946 a process or peer of a connected AF_UNIX socket. This
2947 function call is particularly useful when implementing
2948 delegated subtrees support in the control group hierarchy.
2950 * The "sd-event" event loop API of libsystemd now supports
2951 correct dequeuing of real-time signals, without losing
2954 * When systemd requests a PolicyKit decision when managing
2955 units it will now add additional fields to the request,
2956 including unit name and desired operation. This enables more
2957 powerful PolicyKit policies, that make decisions depending
2958 on these parameters.
2960 * nspawn learnt support for .nspawn settings files, that may
2961 accompany the image files or directories of containers, and
2962 may contain additional settings for the container. This is
2963 an alternative to configuring container parameters via the
2964 nspawn command line.
2966 Contributions from: Cristian Rodríguez, Daniel Mack, David
2967 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2968 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2969 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2970 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2971 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2972 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2973 Andersen, Tom Gundersen, Torstein Husebø
2975 — Berlin, 2015-09-08
2979 * machinectl gained a new verb 'shell' which opens a fresh
2980 shell on the target container or the host. It is similar to
2981 the existing 'login' command of machinectl, but spawns the
2982 shell directly without prompting for username or
2983 password. The pseudo machine '.host' now refers to the local
2984 host and is used by default. Hence, 'machinectl shell' can
2985 be used as replacement for 'su -' which spawns a session as
2986 a fresh systemd unit in a way that is fully isolated from
2987 the originating session.
2989 * systemd-networkd learned to cope with private-zone DHCP
2990 options and allows other programs to query the values.
2992 * SELinux access control when enabling/disabling units is no
2993 longer enforced with this release. The previous
2994 implementation was incorrect, and a new corrected
2995 implementation is not yet available. As unit file operations
2996 are still protected via PolicyKit and D-Bus policy this is
2997 not a security problem. Yet, distributions which care about
2998 optimal SELinux support should probably not stabilize on
3001 * sd-bus gained support for matches of type "arg0has=", that
3002 test for membership of strings in string arrays sent in bus
3005 * systemd-resolved now dumps the contents of its DNS and LLMNR
3006 caches to the logs on reception of the SIGUSR1 signal. This
3007 is useful to debug DNS behaviour.
3009 * The coredumpctl tool gained a new --directory= option to
3010 operate on journal files in a specific directory.
3012 * "systemctl reboot" and related commands gained a new
3013 "--message=" option which may be used to set a free-text
3014 wall message when shutting down or rebooting the
3015 system. This message is also logged, which is useful for
3016 figuring out the reason for a reboot or shutdown a
3019 * The "systemd-resolve-host" tool's -i switch now takes
3020 network interface numbers as alternative to interface names.
3022 * A new unit file setting for services has been introduced:
3023 UtmpMode= allows configuration of how precisely systemd
3024 handles utmp and wtmp entries for the service if this is
3025 enabled. This allows writing services that appear similar to
3026 user sessions in the output of the "w", "who", "last" and
3029 * systemd-resolved will now locally synthesize DNS resource
3030 records for the "localhost" and "gateway" domains as well as
3031 the local hostname. This should ensure that clients querying
3032 RRs via resolved will get similar results as those going via
3033 NSS, if nss-myhostname is enabled.
3035 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3036 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3037 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3038 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3039 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3040 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3041 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3042 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3043 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3044 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3045 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3046 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3048 — Berlin, 2015-08-27
3052 * The systemd-efi-boot-generator functionality was merged into
3053 systemd-gpt-auto-generator.
3055 * systemd-networkd now supports Group Policy for vxlan
3056 devices. It can be enabled via the new boolean configuration
3057 option called 'GroupPolicyExtension='.
3059 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3060 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3061 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3063 — Berlin, 2015-07-31
3067 * The python-systemd code has been removed from the systemd repository.
3068 A new repository has been created which accommodates the code from
3069 now on, and we kindly ask distributions to create a separate package
3070 for this: https://github.com/systemd/python-systemd
3072 * The systemd daemon will now reload its main configuration
3073 (/etc/systemd/system.conf) on daemon-reload.
3075 * sd-dhcp now exposes vendor specific extensions via
3076 sd_dhcp_lease_get_vendor_specific().
3078 * systemd-networkd gained a number of new configuration options.
3080 - A new boolean configuration option for TAP devices called
3081 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3082 device, thus allowing to send and receive GSO packets.
3084 - A new tunnel configuration option called 'CopyDSCP='.
3085 If enabled, the DSCP field of ip6 tunnels is copied into the
3086 decapsulated packet.
3088 - A set of boolean bridge configuration options were added.
3089 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3090 and 'UnicastFlood=' are now parsed by networkd and applied to the
3091 respective bridge link device via the respective IFLA_BRPORT_*
3094 - A new string configuration option to override the hostname sent
3095 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3096 is true, networkd will use the configured hostname instead of the
3097 system hostname when sending DHCP requests.
3099 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3100 networkd will configure the IPv6 flow-label of the tunnel device
3101 according to RFC2460.
3103 - The 'macvtap' virtual network devices are now supported, similar to
3104 the already supported 'macvlan' devices.
3106 * systemd-resolved now implements RFC5452 to improve resilience against
3107 cache poisoning. Additionally, source port randomization is enabled
3108 by default to further protect against DNS spoofing attacks.
3110 * nss-mymachines now supports translating UIDs and GIDs of running
3111 containers with user-namespaces enabled. If a container 'foo'
3112 translates a host uid 'UID' to the container uid 'TUID', then
3113 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3114 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3115 mapped as 'vg-foo-TGID'.
3117 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3118 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3119 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3120 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3121 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3122 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3123 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3124 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3125 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3126 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3128 — Berlin, 2015-07-29
3132 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3133 There are no known issues with current sysfs, and udev does not need
3134 or should be used to work around such bugs.
3136 * udev does no longer enable USB HID power management. Several reports
3137 indicate, that some devices cannot handle that setting.
3139 * The udev accelerometer helper was removed. The functionality
3140 is now fully included in iio-sensor-proxy. But this means,
3141 older iio-sensor-proxy versions will no longer provide
3142 accelerometer/orientation data with this systemd version.
3143 Please upgrade iio-sensor-proxy to version 1.0.
3145 * networkd gained a new configuration option IPv6PrivacyExtensions=
3146 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3147 for Stateless Address") on selected networks.
3149 * For the sake of fewer build-time dependencies and less code in the
3150 main repository, the python bindings are about to be removed in the
3151 next release. A new repository has been created which accommodates
3152 the code from now on, and we kindly ask distributions to create a
3153 separate package for this. The removal will take place in v223.
3155 https://github.com/systemd/python-systemd
3157 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3158 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3159 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3160 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3161 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3162 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3163 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3164 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3165 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3166 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3168 — Berlin, 2015-07-07
3172 * The sd-bus.h and sd-event.h APIs have now been declared
3173 stable and have been added to the official interface of
3174 libsystemd.so. sd-bus implements an alternative D-Bus client
3175 library, that is relatively easy to use, very efficient and
3176 supports both classic D-Bus as well as kdbus as transport
3177 backend. sd-event is a generic event loop abstraction that
3178 is built around Linux epoll, but adds features such as event
3179 prioritization or efficient timer handling. Both APIs are good
3180 choices for C programs looking for a bus and/or event loop
3181 implementation that is minimal and does not have to be
3182 portable to other kernels.
3184 * kdbus support is no longer compile-time optional. It is now
3185 always built-in. However, it can still be disabled at
3186 runtime using the kdbus=0 kernel command line setting, and
3187 that setting may be changed to default to off, by specifying
3188 --disable-kdbus at build-time. Note though that the kernel
3189 command line setting has no effect if the kdbus.ko kernel
3190 module is not installed, in which case kdbus is (obviously)
3191 also disabled. We encourage all downstream distributions to
3192 begin testing kdbus by adding it to the kernel images in the
3193 development distributions, and leaving kdbus support in
3196 * The minimal required util-linux version has been bumped to
3199 * Support for chkconfig (--enable-chkconfig) was removed in
3200 favor of calling an abstraction tool
3201 /lib/systemd/systemd-sysv-install. This needs to be
3202 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3203 in README for details.
3205 * If there's a systemd unit and a SysV init script for the
3206 same service name, and the user executes "systemctl enable"
3207 for it (or a related call), then this will now enable both
3208 (or execute the related operation on both), not just the
3211 * The libudev API documentation has been converted from gtkdoc
3214 * gudev has been removed from the systemd tree, it is now an
3217 * The systemd-cgtop tool learnt a new --raw switch to generate
3218 "raw" (machine parsable) output.
3220 * networkd's IPForwarding= .network file setting learnt the
3221 new setting "kernel", which ensures that networkd does not
3222 change the IP forwarding sysctl from the default kernel
3225 * The systemd-logind bus API now exposes a new boolean
3226 property "Docked" that reports whether logind considers the
3227 system "docked", i.e. connected to a docking station or not.
3229 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3230 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3231 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3232 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3233 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3234 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3235 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3236 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3237 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3238 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3239 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3240 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3241 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3242 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3243 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3244 Fink, Zbigniew Jędrzejewski-Szmek
3246 — Berlin, 2015-06-19
3250 * The gudev library has been extracted into a separate repository
3251 available at: https://git.gnome.org/browse/libgudev/
3252 It is now managed as part of the Gnome project. Distributions
3253 are recommended to pass --disable-gudev to systemd and use
3254 gudev from the Gnome project instead. gudev is still included
3255 in systemd, for now. It will be removed soon, though. Please
3256 also see the announcement-thread on systemd-devel:
3257 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3259 * systemd now exposes a CPUUsageNSec= property for each
3260 service unit on the bus, that contains the overall consumed
3261 CPU time of a service (the sum of what each process of the
3262 service consumed). This value is only available if
3263 CPUAccounting= is turned on for a service, and is then shown
3264 in the "systemctl status" output.
3266 * Support for configuring alternative mappings of the old SysV
3267 runlevels to systemd targets has been removed. They are now
3268 hardcoded in a way that runlevels 2, 3, 4 all map to
3269 multi-user.target and 5 to graphical.target (which
3270 previously was already the default behaviour).
3272 * The auto-mounter logic gained support for mount point
3273 expiry, using a new TimeoutIdleSec= setting in .automount
3274 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3276 * The EFI System Partition (ESP) as mounted to /boot by
3277 systemd-efi-boot-generator will now be unmounted
3278 automatically after 2 minutes of not being used. This should
3279 minimize the risk of ESP corruptions.
3281 * New /etc/fstab options x-systemd.requires= and
3282 x-systemd.requires-mounts-for= are now supported to express
3283 additional dependencies for mounts. This is useful for
3284 journalling file systems that support external journal
3285 devices or overlay file systems that require underlying file
3286 systems to be mounted.
3288 * systemd does not support direct live-upgrades (via systemctl
3289 daemon-reexec) from versions older than v44 anymore. As no
3290 distribution we are aware of shipped such old versions in a
3291 stable release this should not be problematic.
3293 * When systemd forks off a new per-connection service instance
3294 it will now set the $REMOTE_ADDR environment variable to the
3295 remote IP address, and $REMOTE_PORT environment variable to
3296 the remote IP port. This behaviour is similar to the
3297 corresponding environment variables defined by CGI.
3299 * systemd-networkd gained support for uplink failure
3300 detection. The BindCarrier= option allows binding interface
3301 configuration dynamically to the link sense of other
3302 interfaces. This is useful to achieve behaviour like in
3305 * systemd-networkd gained support for configuring the DHCP
3306 client identifier to use when requesting leases.
3308 * systemd-networkd now has a per-network UseNTP= option to
3309 configure whether NTP server information acquired via DHCP
3310 is passed on to services like systemd-timesyncd.
3312 * systemd-networkd gained support for vti6 tunnels.
3314 * Note that systemd-networkd manages the sysctl variable
3315 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3316 it is configured for since v219. The variable controls IP
3317 forwarding, and is a per-interface alternative to the global
3318 /proc/sys/net/ipv[46]/ip_forward. This setting is
3319 configurable in the IPForward= option, which defaults to
3320 "no". This means if networkd is used for an interface it is
3321 no longer sufficient to set the global sysctl option to turn
3322 on IP forwarding! Instead, the .network file option
3323 IPForward= needs to be turned on! Note that the
3324 implementation of this behaviour was broken in v219 and has
3327 * Many bonding and vxlan options are now configurable in
3330 * systemd-nspawn gained a new --property= setting to set unit
3331 properties for the container scope. This is useful for
3332 setting resource parameters (e.g. "CPUShares=500") on
3333 containers started from the command line.
3335 * systemd-nspawn gained a new --private-users= switch to make
3336 use of user namespacing available on recent Linux kernels.
3338 * systemd-nspawn may now be called as part of a shell pipeline
3339 in which case the pipes used for stdin and stdout are passed
3340 directly to the process invoked in the container, without
3341 indirection via a pseudo tty.
3343 * systemd-nspawn gained a new switch to control the UNIX
3344 signal to use when killing the init process of the container
3347 * systemd-nspawn gained a new --overlay= switch for mounting
3348 overlay file systems into the container using the new kernel
3351 * When a container image is imported via systemd-importd and
3352 the host file system is not btrfs, a loopback block device
3353 file is created in /var/lib/machines.raw with a btrfs file
3354 system inside. It is then mounted to /var/lib/machines to
3355 enable btrfs features for container management. The loopback
3356 file and btrfs file system is grown as needed when container
3357 images are imported via systemd-importd.
3359 * systemd-machined/systemd-importd gained support for btrfs
3360 quota, to enforce container disk space limits on disk. This
3361 is exposed in "machinectl set-limit".
3363 * systemd-importd now can import containers from local .tar,
3364 .raw and .qcow2 images, and export them to .tar and .raw. It
3365 can also import dkr v2 images now from the network (on top
3368 * systemd-importd gained support for verifying downloaded
3369 images with gpg2 (previously only gpg1 was supported).
3371 * systemd-machined, systemd-logind, systemd: most bus calls
3372 are now accessible to unprivileged processes via
3373 PolicyKit. Also, systemd-logind will now allow users to kill
3374 their own sessions without further privileges or
3377 * systemd-shutdownd has been removed. This service was
3378 previously responsible for implementing scheduled shutdowns
3379 as exposed in /usr/bin/shutdown's time parameter. This
3380 functionality has now been moved into systemd-logind and is
3381 accessible via a bus interface.
3383 * "systemctl reboot" gained a new switch --firmware-setup that
3384 can be used to reboot into the EFI firmware setup, if that
3385 is available. systemd-logind now exposes an API on the bus
3386 to trigger such reboots, in case graphical desktop UIs want
3387 to cover this functionality.
3389 * "systemctl enable", "systemctl disable" and "systemctl mask"
3390 now support a new "--now" switch. If specified the units
3391 that are enabled will also be started, and the ones
3392 disabled/masked also stopped.
3394 * The Gummiboot EFI boot loader tool has been merged into
3395 systemd, and renamed to "systemd-boot". The bootctl tool has been
3396 updated to support systemd-boot.
3398 * An EFI kernel stub has been added that may be used to create
3399 kernel EFI binaries that contain not only the actual kernel,
3400 but also an initrd, boot splash, command line and OS release
3401 information. This combined binary can then be signed as a
3402 single image, so that the firmware can verify it all in one
3403 step. systemd-boot has special support for EFI binaries created
3404 like this and can extract OS release information from them
3405 and show them in the boot menu. This functionality is useful
3406 to implement cryptographically verified boot schemes.
3408 * Optional support has been added to systemd-fsck to pass
3409 fsck's progress report to an AF_UNIX socket in the file
3412 * udev will no longer create device symlinks for all block
3413 devices by default. A blacklist for excluding special block
3414 devices from this logic has been turned into a whitelist
3415 that requires picking block devices explicitly that require
3418 * A new (currently still internal) API sd-device.h has been
3419 added to libsystemd. This modernized API is supposed to
3420 replace libudev eventually. In fact, already much of libudev
3421 is now just a wrapper around sd-device.h.
3423 * A new hwdb database for storing metadata about pointing
3424 stick devices has been added.
3426 * systemd-tmpfiles gained support for setting file attributes
3427 similar to the "chattr" tool with new 'h' and 'H' lines.
3429 * systemd-journald will no longer unconditionally set the
3430 btrfs NOCOW flag on new journal files. This is instead done
3431 with tmpfiles snippet using the new 'h' line type. This
3432 allows easy disabling of this logic, by masking the
3433 journal-nocow.conf tmpfiles file.
3435 * systemd-journald will now translate audit message types to
3436 human readable identifiers when writing them to the
3437 journal. This should improve readability of audit messages.
3439 * The LUKS logic gained support for the offset= and skip=
3440 options in /etc/crypttab, as previously implemented by
3443 * /usr/lib/os-release gained a new optional field VARIANT= for
3444 distributions that support multiple variants (such as a
3445 desktop edition, a server edition, ...)
3447 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3448 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3449 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3450 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3451 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3452 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3453 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3454 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3455 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3456 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3457 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3458 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3459 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3460 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3461 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3462 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3463 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3464 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3465 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3466 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3467 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3468 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3469 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3470 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3471 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3472 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3473 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3475 — Berlin, 2015-05-22
3479 * Introduce a new API "sd-hwdb.h" for querying the hardware
3480 metadata database. With this minimal interface one can query
3481 and enumerate the udev hwdb, decoupled from the old libudev
3482 library. libudev's interface for this is now only a wrapper
3483 around sd-hwdb. A new tool systemd-hwdb has been added to
3484 interface with and update the database.
3486 * When any of systemd's tools copies files (for example due to
3487 tmpfiles' C lines) a btrfs reflink will attempted first,
3488 before bytewise copying is done.
3490 * systemd-nspawn gained a new --ephemeral switch. When
3491 specified a btrfs snapshot is taken of the container's root
3492 directory, and immediately removed when the container
3493 terminates again. Thus, a container can be started whose
3494 changes never alter the container's root directory, and are
3495 lost on container termination. This switch can also be used
3496 for starting a container off the root file system of the
3497 host without affecting the host OS. This switch is only
3498 available on btrfs file systems.
3500 * systemd-nspawn gained a new --template= switch. It takes the
3501 path to a container tree to use as template for the tree
3502 specified via --directory=, should that directory be
3503 missing. This allows instantiating containers dynamically,
3504 on first run. This switch is only available on btrfs file
3507 * When a .mount unit refers to a mount point on which multiple
3508 mounts are stacked, and the .mount unit is stopped all of
3509 the stacked mount points will now be unmounted until no
3510 mount point remains.
3512 * systemd now has an explicit notion of supported and
3513 unsupported unit types. Jobs enqueued for unsupported unit
3514 types will now fail with an "unsupported" error code. More
3515 specifically .swap, .automount and .device units are not
3516 supported in containers, .busname units are not supported on
3517 non-kdbus systems. .swap and .automount are also not
3518 supported if their respective kernel compile time options
3521 * machinectl gained support for two new "copy-from" and
3522 "copy-to" commands for copying files from a running
3523 container to the host or vice versa.
3525 * machinectl gained support for a new "bind" command to bind
3526 mount host directories into local containers. This is
3527 currently only supported for nspawn containers.
3529 * networkd gained support for configuring bridge forwarding
3530 database entries (fdb) from .network files.
3532 * A new tiny daemon "systemd-importd" has been added that can
3533 download container images in tar, raw, qcow2 or dkr formats,
3534 and make them available locally in /var/lib/machines, so
3535 that they can run as nspawn containers. The daemon can GPG
3536 verify the downloads (not supported for dkr, since it has no
3537 provisions for verifying downloads). It will transparently
3538 decompress bz2, xz, gzip compressed downloads if necessary,
3539 and restore sparse files on disk. The daemon uses privilege
3540 separation to ensure the actual download logic runs with
3541 fewer privileges than the daemon itself. machinectl has
3542 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3543 make the functionality of importd available to the
3544 user. With this in place the Fedora and Ubuntu "Cloud"
3545 images can be downloaded and booted as containers unmodified
3546 (the Fedora images lack the appropriate GPG signature files
3547 currently, so they cannot be verified, but this will change
3548 soon, hopefully). Note that downloading images is currently
3549 only fully supported on btrfs.
3551 * machinectl is now able to list container images found in
3552 /var/lib/machines, along with some metadata about sizes of
3553 disk and similar. If the directory is located on btrfs and
3554 quota is enabled, this includes quota display. A new command
3555 "image-status" has been added that shows additional
3556 information about images.
3558 * machinectl is now able to clone container images
3559 efficiently, if the underlying file system (btrfs) supports
3560 it, with the new "machinectl clone" command. It also
3561 gained commands for renaming and removing images, as well as
3562 marking them read-only or read-write (supported also on
3563 legacy file systems).
3565 * networkd gained support for collecting LLDP network
3566 announcements, from hardware that supports this. This is
3567 shown in networkctl output.
3569 * systemd-run gained support for a new -t (--pty) switch for
3570 invoking a binary on a pty whose input and output is
3571 connected to the invoking terminal. This allows executing
3572 processes as system services while interactively
3573 communicating with them via the terminal. Most interestingly
3574 this is supported across container boundaries. Invoking
3575 "systemd-run -t /bin/bash" is an alternative to running a
3576 full login session, the difference being that the former
3577 will not register a session, nor go through the PAM session
3580 * tmpfiles gained support for a new "v" line type for creating
3581 btrfs subvolumes. If the underlying file system is a legacy
3582 file system, this automatically degrades to creating a
3583 normal directory. Among others /var/lib/machines is now
3584 created like this at boot, should it be missing.
3586 * The directory /var/lib/containers/ has been deprecated and
3587 been replaced by /var/lib/machines. The term "machines" has
3588 been used in the systemd context as generic term for both
3589 VMs and containers, and hence appears more appropriate for
3590 this, as the directory can also contain raw images bootable
3593 * systemd-nspawn when invoked with -M but without --directory=
3594 or --image= is now capable of searching for the container
3595 root directory, subvolume or disk image automatically, in
3596 /var/lib/machines. systemd-nspawn@.service has been updated
3597 to make use of this, thus allowing it to be used for raw
3600 * A new machines.target unit has been introduced that is
3601 supposed to group all containers/VMs invoked as services on
3602 the system. systemd-nspawn@.service has been updated to
3603 integrate with that.
3605 * machinectl gained a new "start" command, for invoking a
3606 container as a service. "machinectl start foo" is mostly
3607 equivalent to "systemctl start systemd-nspawn@foo.service",
3608 but handles escaping in a nicer way.
3610 * systemd-nspawn will now mount most of the cgroupfs tree
3611 read-only into each container, with the exception of the
3612 container's own subtree in the name=systemd hierarchy.
3614 * journald now sets the special FS_NOCOW file flag for its
3615 journal files. This should improve performance on btrfs, by
3616 avoiding heavy fragmentation when journald's write-pattern
3617 is used on COW file systems. It degrades btrfs' data
3618 integrity guarantees for the files to the same levels as for
3619 ext3/ext4 however. This should be OK though as journald does
3620 its own data integrity checks and all its objects are
3621 checksummed on disk. Also, journald should handle btrfs disk
3622 full events a lot more gracefully now, by processing SIGBUS
3623 errors, and not relying on fallocate() anymore.
3625 * When journald detects that journal files it is writing to
3626 have been deleted it will immediately start new journal
3629 * systemd now provides a way to store file descriptors
3630 per-service in PID 1. This is useful for daemons to ensure
3631 that fds they require are not lost during a daemon
3632 restart. The fds are passed to the daemon on the next
3633 invocation in the same way socket activation fds are
3634 passed. This is now used by journald to ensure that the
3635 various sockets connected to all the system's stdout/stderr
3636 are not lost when journald is restarted. File descriptors
3637 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3638 an extension to sd_notify(). Note that a limit is enforced
3639 on the number of fds a service can store in PID 1, and it
3640 defaults to 0, so that no fds may be stored, unless this is
3641 explicitly turned on.
3643 * The default TERM variable to use for units connected to a
3644 terminal, when no other value is explicitly is set is now
3645 vt220 rather than vt102. This should be fairly safe still,
3646 but allows PgUp/PgDn work.
3648 * The /etc/crypttab option header= as known from Debian is now
3651 * "loginctl user-status" and "loginctl session-status" will
3652 now show the last 10 lines of log messages of the
3653 user/session following the status output. Similar,
3654 "machinectl status" will show the last 10 log lines
3655 associated with a virtual machine or container
3656 service. (Note that this is usually not the log messages
3657 done in the VM/container itself, but simply what the
3658 container manager logs. For nspawn this includes all console
3661 * "loginctl session-status" without further argument will now
3662 show the status of the session of the caller. Similar,
3663 "lock-session", "unlock-session", "activate",
3664 "enable-linger", "disable-linger" may now be called without
3665 session/user parameter in which case they apply to the
3666 caller's session/user.
3668 * An X11 session scriptlet is now shipped that uploads
3669 $DISPLAY and $XAUTHORITY into the environment of the systemd
3670 --user daemon if a session begins. This should improve
3671 compatibility with X11 enabled applications run as systemd
3674 * Generators are now subject to masking via /etc and /run, the
3675 same way as unit files.
3677 * networkd .network files gained support for configuring
3678 per-link IPv4/IPv6 packet forwarding as well as IPv4
3679 masquerading. This is by default turned on for veth links to
3680 containers, as registered by systemd-nspawn. This means that
3681 nspawn containers run with --network-veth will now get
3682 automatic routed access to the host's networks without any
3683 further configuration or setup, as long as networkd runs on
3686 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3687 or UDP posts of a container on the host. With this in place
3688 it is possible to run containers with private veth links
3689 (--network-veth), and have their functionality exposed on
3690 the host as if their services were running directly on the
3693 * systemd-nspawn's --network-veth switch now gained a short
3694 version "-n", since with the changes above it is now truly
3695 useful out-of-the-box. The systemd-nspawn@.service has been
3696 updated to make use of it too by default.
3698 * systemd-nspawn will now maintain a per-image R/W lock, to
3699 ensure that the same image is not started more than once
3700 writable. (It's OK to run an image multiple times
3701 simultaneously in read-only mode.)
3703 * systemd-nspawn's --image= option is now capable of
3704 dissecting and booting MBR and GPT disk images that contain
3705 only a single active Linux partition. Previously it
3706 supported only GPT disk images with proper GPT type
3707 IDs. This allows running cloud images from major
3708 distributions directly with systemd-nspawn, without
3711 * In addition to collecting mouse dpi data in the udev
3712 hardware database, there's now support for collecting angle
3713 information for mouse scroll wheels. The database is
3714 supposed to guarantee similar scrolling behavior on mice
3715 that it knows about. There's also support for collecting
3716 information about Touchpad types.
3718 * udev's input_id built-in will now also collect touch screen
3719 dimension data and attach it to probed devices.
3721 * /etc/os-release gained support for a Distribution Privacy
3724 * networkd gained support for creating "ipvlan", "gretap",
3725 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3727 * systemd-tmpfiles gained support for "a" lines for setting
3730 * systemd-nspawn will now mount /tmp in the container to
3731 tmpfs, automatically.
3733 * systemd now exposes the memory.usage_in_bytes cgroup
3734 attribute and shows it for each service in the "systemctl
3735 status" output, if available.
3737 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3738 immediate reboot is triggered. This useful if shutdown is
3739 hung and is unable to complete, to expedite the
3740 operation. Note that this kind of reboot will still unmount
3741 all file systems, and hence should not result in fsck being
3744 * A .device unit for an optical block device will now be
3745 considered active only when a medium is in the drive. Also,
3746 mount units are now bound to their backing devices thus
3747 triggering automatic unmounting when devices become
3748 unavailable. With this in place systemd will now
3749 automatically unmount left-over mounts when a CD-ROM is
3750 ejected or an USB stick is yanked from the system.
3752 * networkd-wait-online now has support for waiting for
3753 specific interfaces only (with globbing), and for giving up
3754 after a configurable timeout.
3756 * networkd now exits when idle. It will be automatically
3757 restarted as soon as interfaces show up, are removed or
3758 change state. networkd will stay around as long as there is
3759 at least one DHCP state machine or similar around, that keep
3762 * networkd may now configure IPv6 link-local addressing in
3763 addition to IPv4 link-local addressing.
3765 * The IPv6 "token" for use in SLAAC may now be configured for
3766 each .network interface in networkd.
3768 * Routes configured with networkd may now be assigned a scope
3771 * networkd's [Match] sections now support globbing and lists
3772 of multiple space-separated matches per item.
3774 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3775 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3776 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3777 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3778 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3779 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3780 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3781 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3782 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3783 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3784 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3785 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3786 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3787 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3788 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3789 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3790 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3791 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3792 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3793 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3794 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3795 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3796 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3797 Hoffmann, Zbigniew Jędrzejewski-Szmek
3799 — Berlin, 2015-02-16
3803 * When querying unit file enablement status (for example via
3804 "systemctl is-enabled"), a new state "indirect" is now known
3805 which indicates that a unit might not be enabled itself, but
3806 another unit listed in its Also= setting might be.
3808 * Similar to the various existing ConditionXYZ= settings for
3809 units, there are now matching AssertXYZ= settings. While
3810 failing conditions cause a unit to be skipped, but its job
3811 to succeed, failing assertions declared like this will cause
3812 a unit start operation and its job to fail.
3814 * hostnamed now knows a new chassis type "embedded".
3816 * systemctl gained a new "edit" command. When used on a unit
3817 file, this allows extending unit files with .d/ drop-in
3818 configuration snippets or editing the full file (after
3819 copying it from /usr/lib to /etc). This will invoke the
3820 user's editor (as configured with $EDITOR), and reload the
3821 modified configuration after editing.
3823 * "systemctl status" now shows the suggested enablement state
3824 for a unit, as declared in the (usually vendor-supplied)
3825 system preset files.
3827 * nss-myhostname will now resolve the single-label host name
3828 "gateway" to the locally configured default IP routing
3829 gateways, ordered by their metrics. This assigns a stable
3830 name to the used gateways, regardless which ones are
3831 currently configured. Note that the name will only be
3832 resolved after all other name sources (if nss-myhostname is
3833 configured properly) and should hence not negatively impact
3834 systems that use the single-label host name "gateway" in
3837 * systemd-inhibit now allows filtering by mode when listing
3840 * Scope and service units gained a new "Delegate" boolean
3841 property, which, when set, allows processes running inside the
3842 unit to further partition resources. This is primarily
3843 useful for systemd user instances as well as container
3846 * journald will now pick up audit messages directly from
3847 the kernel, and log them like any other log message. The
3848 audit fields are split up and fully indexed. This means that
3849 journalctl in many ways is now a (nicer!) alternative to
3850 ausearch, the traditional audit client. Note that this
3851 implements only a minimal audit client. If you want the
3852 special audit modes like reboot-on-log-overflow, please use
3853 the traditional auditd instead, which can be used in
3854 parallel to journald.
3856 * The ConditionSecurity= unit file option now understands the
3857 special string "audit" to check whether auditing is
3860 * journalctl gained two new commands --vacuum-size= and
3861 --vacuum-time= to delete old journal files until the
3862 remaining ones take up no more than the specified size on disk,
3863 or are not older than the specified time.
3865 * A new, native PPPoE library has been added to sd-network,
3866 systemd's library of light-weight networking protocols. This
3867 library will be used in a future version of networkd to
3868 enable PPPoE communication without an external pppd daemon.
3870 * The busctl tool now understands a new "capture" verb that
3871 works similar to "monitor", but writes a packet capture
3872 trace to STDOUT that can be redirected to a file which is
3873 compatible with libcap's capture file format. This can then
3874 be loaded in Wireshark and similar tools to inspect bus
3877 * The busctl tool now understands a new "tree" verb that shows
3878 the object trees of a specific service on the bus, or of all
3881 * The busctl tool now understands a new "introspect" verb that
3882 shows all interfaces and members of objects on the bus,
3883 including their signature and values. This is particularly
3884 useful to get more information about bus objects shown by
3885 the new "busctl tree" command.
3887 * The busctl tool now understands new verbs "call",
3888 "set-property" and "get-property" for invoking bus method
3889 calls, setting and getting bus object properties in a
3892 * busctl gained a new --augment-creds= argument that controls
3893 whether the tool shall augment credential information it
3894 gets from the bus with data from /proc, in a possibly
3897 * nspawn's --link-journal= switch gained two new values
3898 "try-guest" and "try-host" that work like "guest" and
3899 "host", but do not fail if the host has no persistent
3900 journalling enabled. -j is now equivalent to
3901 --link-journal=try-guest.
3903 * macvlan network devices created by nspawn will now have
3904 stable MAC addresses.
3906 * A new SmackProcessLabel= unit setting has been added, which
3907 controls the SMACK security label processes forked off by
3908 the respective unit shall use.
3910 * If compiled with --enable-xkbcommon, systemd-localed will
3911 verify x11 keymap settings by compiling the given keymap. It
3912 will spew out warnings if the compilation fails. This
3913 requires libxkbcommon to be installed.
3915 * When a coredump is collected, a larger number of metadata
3916 fields is now collected and included in the journal records
3917 created for it. More specifically, control group membership,
3918 environment variables, memory maps, working directory,
3919 chroot directory, /proc/$PID/status, and a list of open file
3920 descriptors is now stored in the log entry.
3922 * The udev hwdb now contains DPI information for mice. For
3925 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3927 * All systemd programs that read standalone configuration
3928 files in /etc now also support a corresponding series of
3929 .conf.d configuration directories in /etc/, /run/,
3930 /usr/local/lib/, /usr/lib/, and (if configured with
3931 --enable-split-usr) /lib/. In particular, the following
3932 configuration files now have corresponding configuration
3933 directories: system.conf user.conf, logind.conf,
3934 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3935 resolved.conf, timesyncd.conf, journal-remote.conf, and
3936 journal-upload.conf. Note that distributions should use the
3937 configuration directories in /usr/lib/; the directories in
3938 /etc/ are reserved for the system administrator.
3940 * systemd-rfkill will no longer take the rfkill device name
3941 into account when storing rfkill state on disk, as the name
3942 might be dynamically assigned and not stable. Instead, the
3943 ID_PATH udev variable combined with the rfkill type (wlan,
3944 bluetooth, ...) is used.
3946 * A new service systemd-machine-id-commit.service has been
3947 added. When used on systems where /etc is read-only during
3948 boot, and /etc/machine-id is not initialized (but an empty
3949 file), this service will copy the temporary machine ID
3950 created as replacement into /etc after the system is fully
3951 booted up. This is useful for systems that are freshly
3952 installed with a non-initialized machine ID, but should get
3953 a fixed machine ID for subsequent boots.
3955 * networkd's .netdev files now provide a large set of
3956 configuration parameters for VXLAN devices. Similarly, the
3957 bridge port cost parameter is now configurable in .network
3958 files. There's also new support for configuring IP source
3959 routing. networkd .link files gained support for a new
3960 OriginalName= match that is useful to match against the
3961 original interface name the kernel assigned. .network files
3962 may include MTU= and MACAddress= fields for altering the MTU
3963 and MAC address while being connected to a specific network
3966 * The LUKS logic gained supported for configuring
3967 UUID-specific key files. There's also new support for naming
3968 LUKS device from the kernel command line, using the new
3969 luks.name= argument.
3971 * Timer units may now be transiently created via the bus API
3972 (this was previously already available for scope and service
3973 units). In addition it is now possible to create multiple
3974 transient units at the same time with a single bus call. The
3975 "systemd-run" tool has been updated to make use of this for
3976 running commands on a specified time, in at(1)-style.
3978 * tmpfiles gained support for "t" lines, for assigning
3979 extended attributes to files. Among other uses this may be
3980 used to assign SMACK labels to files.
3982 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3983 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3984 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3985 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3986 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3987 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3988 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3989 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3990 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3991 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3992 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3993 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3994 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3995 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3996 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3997 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3998 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3999 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4001 — Berlin, 2014-12-10
4005 * journalctl gained the new options -t/--identifier= to match
4006 on the syslog identifier (aka "tag"), as well as --utc to
4007 show log timestamps in the UTC timezone. journalctl now also
4008 accepts -n/--lines=all to disable line capping in a pager.
4010 * journalctl gained a new switch, --flush, that synchronously
4011 flushes logs from /run/log/journal to /var/log/journal if
4012 persistent storage is enabled. systemd-journal-flush.service
4013 now waits until the operation is complete.
4015 * Services can notify the manager before they start a reload
4016 (by sending RELOADING=1) or shutdown (by sending
4017 STOPPING=1). This allows the manager to track and show the
4018 internal state of daemons and closes a race condition when
4019 the process is still running but has closed its D-Bus
4022 * Services with Type=oneshot do not have to have any ExecStart
4025 * User units are now loaded also from
4026 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4027 /run/systemd/user directory that was already previously
4028 supported, but is under the control of the user.
4030 * Job timeouts (i.e. time-outs on the time a job that is
4031 queued stays in the run queue) can now optionally result in
4032 immediate reboot or power-off actions (JobTimeoutAction= and
4033 JobTimeoutRebootArgument=). This is useful on ".target"
4034 units, to limit the maximum time a target remains
4035 undispatched in the run queue, and to trigger an emergency
4036 operation in such a case. This is now used by default to
4037 turn off the system if boot-up (as defined by everything in
4038 basic.target) hangs and does not complete for at least
4039 15min. Also, if power-off or reboot hang for at least 30min
4040 an immediate power-off/reboot operation is triggered. This
4041 functionality is particularly useful to increase reliability
4042 on embedded devices, but also on laptops which might
4043 accidentally get powered on when carried in a backpack and
4044 whose boot stays stuck in a hard disk encryption passphrase
4047 * systemd-logind can be configured to also handle lid switch
4048 events even when the machine is docked or multiple displays
4049 are attached (HandleLidSwitchDocked= option).
4051 * A helper binary and a service have been added which can be
4052 used to resume from hibernation in the initramfs. A
4053 generator will parse the resume= option on the kernel
4054 command line to trigger resume.
4056 * A user console daemon systemd-consoled has been
4057 added. Currently, it is a preview, and will so far open a
4058 single terminal on each session of the user marked as
4059 Desktop=systemd-console.
4061 * Route metrics can be specified for DHCP routes added by
4064 * The SELinux context of socket-activated services can be set
4065 from the information provided by the networking stack
4066 (SELinuxContextFromNet= option).
4068 * Userspace firmware loading support has been removed and
4069 the minimum supported kernel version is thus bumped to 3.7.
4071 * Timeout for udev workers has been increased from 1 to 3
4072 minutes, but a warning will be printed after 1 minute to
4073 help diagnose kernel modules that take a long time to load.
4075 * Udev rules can now remove tags on devices with TAG-="foobar".
4077 * systemd's readahead implementation has been removed. In many
4078 circumstances it didn't give expected benefits even for
4079 rotational disk drives and was becoming less relevant in the
4080 age of SSDs. As none of the developers has been using
4081 rotating media anymore, and nobody stepped up to actively
4082 maintain this component of systemd it has now been removed.
4084 * Swap units can use Options= to specify discard options.
4085 Discard options specified for swaps in /etc/fstab are now
4088 * Docker containers are now detected as a separate type of
4091 * The Password Agent protocol gained support for queries where
4092 the user input is shown, useful e.g. for user names.
4093 systemd-ask-password gained a new --echo option to turn that
4096 * The default sysctl.d/ snippets will now set:
4098 net.core.default_qdisc = fq_codel
4100 This selects Fair Queuing Controlled Delay as the default
4101 queuing discipline for network interfaces. fq_codel helps
4102 fight the network bufferbloat problem. It is believed to be
4103 a good default with no tuning required for most workloads.
4104 Downstream distributions may override this choice. On 10Gbit
4105 servers that do not do forwarding, "fq" may perform better.
4106 Systems without a good clocksource should use "pfifo_fast".
4108 * If kdbus is enabled during build a new option BusPolicy= is
4109 available for service units, that allows locking all service
4110 processes into a stricter bus policy, in order to limit
4111 access to various bus services, or even hide most of them
4112 from the service's view entirely.
4114 * networkctl will now show the .network and .link file
4115 networkd has applied to a specific interface.
4117 * sd-login gained a new API call sd_session_get_desktop() to
4118 query which desktop environment has been selected for a
4121 * UNIX utmp support is now compile-time optional to support
4122 legacy-free systems.
4124 * systemctl gained two new commands "add-wants" and
4125 "add-requires" for pulling in units from specific targets
4128 * If the word "rescue" is specified on the kernel command line
4129 the system will now boot into rescue mode (aka
4130 rescue.target), which was previously available only by
4131 specifying "1" or "systemd.unit=rescue.target" on the kernel
4132 command line. This new kernel command line option nicely
4133 mirrors the already existing "emergency" kernel command line
4136 * New kernel command line options mount.usr=, mount.usrflags=,
4137 mount.usrfstype= have been added that match root=, rootflags=,
4138 rootfstype= but allow mounting a specific file system to
4141 * The $NOTIFY_SOCKET is now also passed to control processes of
4142 services, not only the main process.
4144 * This version reenables support for fsck's -l switch. This
4145 means at least version v2.25 of util-linux is required for
4146 operation, otherwise dead-locks on device nodes may
4147 occur. Again: you need to update util-linux to at least
4148 v2.25 when updating systemd to v217.
4150 * The "multi-seat-x" tool has been removed from systemd, as
4151 its functionality has been integrated into X servers 1.16,
4152 and the tool is hence redundant. It is recommended to update
4153 display managers invoking this tool to simply invoke X
4154 directly from now on, again.
4156 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4157 message flag has been added for all of systemd's PolicyKit
4158 authenticated method calls has been added. In particular
4159 this now allows optional interactive authorization via
4160 PolicyKit for many of PID1's privileged operations such as
4161 unit file enabling and disabling.
4163 * "udevadm hwdb --update" learnt a new switch "--usr" for
4164 placing the rebuilt hardware database in /usr instead of
4165 /etc. When used only hardware database entries stored in
4166 /usr will be used, and any user database entries in /etc are
4167 ignored. This functionality is useful for vendors to ship a
4168 pre-built database on systems where local configuration is
4169 unnecessary or unlikely.
4171 * Calendar time specifications in .timer units now also
4172 understand the strings "semi-annually", "quarterly" and
4173 "minutely" as shortcuts (in addition to the preexisting
4174 "anually", "hourly", ...).
4176 * systemd-tmpfiles will now correctly create files in /dev
4177 at boot which are marked for creation only at boot. It is
4178 recommended to always create static device nodes with 'c!'
4179 and 'b!', so that they are created only at boot and not
4180 overwritten at runtime.
4182 * When the watchdog logic is used for a service (WatchdogSec=)
4183 and the watchdog timeout is hit the service will now be
4184 terminated with SIGABRT (instead of just SIGTERM), in order
4185 to make sure a proper coredump and backtrace is
4186 generated. This ensures that hanging services will result in
4187 similar coredump/backtrace behaviour as services that hit a
4190 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4191 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4192 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4193 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4194 Herrmann, David Sommerseth, David Strauss, Emil Renner
4195 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4196 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4197 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4198 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4199 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4200 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4201 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4202 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4203 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4204 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4205 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4206 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4207 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4208 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4209 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4210 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4213 — Berlin, 2014-10-28
4217 * timedated no longer reads NTP implementation unit names from
4218 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4219 implementations should add a
4221 Conflicts=systemd-timesyncd.service
4223 to their unit files to take over and replace systemd's NTP
4224 default functionality.
4226 * systemd-sysusers gained a new line type "r" for configuring
4227 which UID/GID ranges to allocate system users/groups
4228 from. Lines of type "u" may now add an additional column
4229 that specifies the home directory for the system user to be
4230 created. Also, systemd-sysusers may now optionally read user
4231 information from STDIN instead of a file. This is useful for
4232 invoking it from RPM preinst scriptlets that need to create
4233 users before the first RPM file is installed since these
4234 files might need to be owned by them. A new
4235 %sysusers_create_inline RPM macro has been introduced to do
4236 just that. systemd-sysusers now updates the shadow files as
4237 well as the user/group databases, which should enhance
4238 compatibility with certain tools like grpck.
4240 * A number of bus APIs of PID 1 now optionally consult
4241 PolicyKit to permit access for otherwise unprivileged
4242 clients under certain conditions. Note that this currently
4243 doesn't support interactive authentication yet, but this is
4244 expected to be added eventually, too.
4246 * /etc/machine-info now has new fields for configuring the
4247 deployment environment of the machine, as well as the
4248 location of the machine. hostnamectl has been updated with
4249 new command to update these fields.
4251 * systemd-timesyncd has been updated to automatically acquire
4252 NTP server information from systemd-networkd, which might
4253 have been discovered via DHCP.
4255 * systemd-resolved now includes a caching DNS stub resolver
4256 and a complete LLMNR name resolution implementation. A new
4257 NSS module "nss-resolve" has been added which can be used
4258 instead of glibc's own "nss-dns" to resolve hostnames via
4259 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4260 be resolved via systemd-resolved D-Bus APIs. In contrast to
4261 the glibc internal resolver systemd-resolved is aware of
4262 multi-homed system, and keeps DNS server and caches separate
4263 and per-interface. Queries are sent simultaneously on all
4264 interfaces that have DNS servers configured, in order to
4265 properly handle VPNs and local LANs which might resolve
4266 separate sets of domain names. systemd-resolved may acquire
4267 DNS server information from systemd-networkd automatically,
4268 which in turn might have discovered them via DHCP. A tool
4269 "systemd-resolve-host" has been added that may be used to
4270 query the DNS logic in resolved. systemd-resolved implements
4271 IDNA and automatically uses IDNA or UTF-8 encoding depending
4272 on whether classic DNS or LLMNR is used as transport. In the
4273 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4274 implementation to systemd-resolved.
4276 * A new NSS module nss-mymachines has been added, that
4277 automatically resolves the names of all local registered
4278 containers to their respective IP addresses.
4280 * A new client tool "networkctl" for systemd-networkd has been
4281 added. It currently is entirely passive and will query
4282 networking configuration from udev, rtnetlink and networkd,
4283 and present it to the user in a very friendly
4284 way. Eventually, we hope to extend it to become a full
4285 control utility for networkd.
4287 * .socket units gained a new DeferAcceptSec= setting that
4288 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4289 TCP. Similarly, support for controlling TCP keep-alive
4290 settings has been added (KeepAliveTimeSec=,
4291 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4292 turning off Nagle's algorithm on TCP has been added
4295 * logind learned a new session type "web", for use in projects
4296 like Cockpit which register web clients as PAM sessions.
4298 * timer units with at least one OnCalendar= setting will now
4299 be started only after time-sync.target has been
4300 reached. This way they will not elapse before the system
4301 clock has been corrected by a local NTP client or
4302 similar. This is particular useful on RTC-less embedded
4303 machines, that come up with an invalid system clock.
4305 * systemd-nspawn's --network-veth= switch should now result in
4306 stable MAC addresses for both the outer and the inner side
4309 * systemd-nspawn gained a new --volatile= switch for running
4310 container instances with /etc or /var unpopulated.
4312 * The kdbus client code has been updated to use the new Linux
4313 3.17 memfd subsystem instead of the old kdbus-specific one.
4315 * systemd-networkd's DHCP client and server now support
4316 FORCERENEW. There are also new configuration options to
4317 configure the vendor client identifier and broadcast mode
4320 * systemd will no longer inform the kernel about the current
4321 timezone, as this is necessarily incorrect and racy as the
4322 kernel has no understanding of DST and similar
4323 concepts. This hence means FAT timestamps will be always
4324 considered UTC, similar to what Android is already
4325 doing. Also, when the RTC is configured to the local time
4326 (rather than UTC) systemd will never synchronize back to it,
4327 as this might confuse Windows at a later boot.
4329 * systemd-analyze gained a new command "verify" for offline
4330 validation of unit files.
4332 * systemd-networkd gained support for a couple of additional
4333 settings for bonding networking setups. Also, the metric for
4334 statically configured routes may now be configured. For
4335 network interfaces where this is appropriate the peer IP
4336 address may now be configured.
4338 * systemd-networkd's DHCP client will no longer request
4339 broadcasting by default, as this tripped up some networks.
4340 For hardware where broadcast is required the feature should
4341 be switched back on using RequestBroadcast=yes.
4343 * systemd-networkd will now set up IPv4LL addresses (when
4344 enabled) even if DHCP is configured successfully.
4346 * udev will now default to respect network device names given
4347 by the kernel when the kernel indicates that these are
4348 predictable. This behavior can be tweaked by changing
4349 NamePolicy= in the relevant .link file.
4351 * A new library systemd-terminal has been added that
4352 implements full TTY stream parsing and rendering. This
4353 library is supposed to be used later on for implementing a
4354 full userspace VT subsystem, replacing the current kernel
4357 * A new tool systemd-journal-upload has been added to push
4358 journal data to a remote system running
4359 systemd-journal-remote.
4361 * journald will no longer forward all local data to another
4362 running syslog daemon. This change has been made because
4363 rsyslog (which appears to be the most commonly used syslog
4364 implementation these days) no longer makes use of this, and
4365 instead pulls the data out of the journal on its own. Since
4366 forwarding the messages to a non-existent syslog server is
4367 more expensive than we assumed we have now turned this
4368 off. If you run a syslog server that is not a recent rsyslog
4369 version, you have to turn this option on again
4370 (ForwardToSyslog= in journald.conf).
4372 * journald now optionally supports the LZ4 compressor for
4373 larger journal fields. This compressor should perform much
4374 better than XZ which was the previous default.
4376 * machinectl now shows the IP addresses of local containers,
4377 if it knows them, plus the interface name of the container.
4379 * A new tool "systemd-escape" has been added that makes it
4380 easy to escape strings to build unit names and similar.
4382 * sd_notify() messages may now include a new ERRNO= field
4383 which is parsed and collected by systemd and shown among the
4384 "systemctl status" output for a service.
4386 * A new component "systemd-firstboot" has been added that
4387 queries the most basic systemd information (timezone,
4388 hostname, root password) interactively on first
4389 boot. Alternatively it may also be used to provision these
4390 things offline on OS images installed into directories.
4392 * The default sysctl.d/ snippets will now set
4394 net.ipv4.conf.default.promote_secondaries=1
4396 This has the benefit of no flushing secondary IP addresses
4397 when primary addresses are removed.
4399 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4400 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4401 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4402 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4403 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4404 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4405 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4406 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4407 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4408 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4409 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4410 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4411 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4412 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4413 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4415 — Berlin, 2014-08-19
4419 * A new tool systemd-sysusers has been added. This tool
4420 creates system users and groups in /etc/passwd and
4421 /etc/group, based on static declarative system user/group
4422 definitions in /usr/lib/sysusers.d/. This is useful to
4423 enable factory resets and volatile systems that boot up with
4424 an empty /etc directory, and thus need system users and
4425 groups created during early boot. systemd now also ships
4426 with two default sysusers.d/ files for the most basic
4427 users and groups systemd and the core operating system
4430 * A new tmpfiles snippet has been added that rebuilds the
4431 essential files in /etc on boot, should they be missing.
4433 * A directive for ensuring automatic clean-up of
4434 /var/cache/man/ has been removed from the default
4435 configuration. This line should now be shipped by the man
4436 implementation. The necessary change has been made to the
4437 man-db implementation. Note that you need to update your man
4438 implementation to one that ships this line, otherwise no
4439 automatic clean-up of /var/cache/man will take place.
4441 * A new condition ConditionNeedsUpdate= has been added that
4442 may conditionalize services to only run when /etc or /var
4443 are "older" than the vendor operating system resources in
4444 /usr. This is useful for reconstructing or updating /etc
4445 after an offline update of /usr or a factory reset, on the
4446 next reboot. Services that want to run once after such an
4447 update or reset should use this condition and order
4448 themselves before the new systemd-update-done.service, which
4449 will mark the two directories as fully updated. A number of
4450 service files have been added making use of this, to rebuild
4451 the udev hardware database, the journald message catalog and
4452 dynamic loader cache (ldconfig). The systemd-sysusers tool
4453 described above also makes use of this now. With this in
4454 place it is now possible to start up a minimal operating
4455 system with /etc empty cleanly. For more information on the
4456 concepts involved see this recent blog story:
4458 http://0pointer.de/blog/projects/stateless.html
4460 * A new system group "input" has been introduced, and all
4461 input device nodes get this group assigned. This is useful
4462 for system-level software to get access to input devices. It
4463 complements what is already done for "audio" and "video".
4465 * systemd-networkd learnt minimal DHCPv4 server support in
4466 addition to the existing DHCPv4 client support. It also
4467 learnt DHCPv6 client and IPv6 Router Solicitation client
4468 support. The DHCPv4 client gained support for static routes
4469 passed in from the server. Note that the [DHCPv4] section
4470 known in older systemd-networkd versions has been renamed to
4471 [DHCP] and is now also used by the DHCPv6 client. Existing
4472 .network files using settings of this section should be
4473 updated, though compatibility is maintained. Optionally, the
4474 client hostname may now be sent to the DHCP server.
4476 * networkd gained support for vxlan virtual networks as well
4477 as tun/tap and dummy devices.
4479 * networkd gained support for automatic allocation of address
4480 ranges for interfaces from a system-wide pool of
4481 addresses. This is useful for dynamically managing a large
4482 number of interfaces with a single network configuration
4483 file. In particular this is useful to easily assign
4484 appropriate IP addresses to the veth links of a large number
4485 of nspawn instances.
4487 * RPM macros for processing sysusers, sysctl and binfmt
4488 drop-in snippets at package installation time have been
4491 * The /etc/os-release file should now be placed in
4492 /usr/lib/os-release. The old location is automatically
4493 created as symlink. /usr/lib is the more appropriate
4494 location of this file, since it shall actually describe the
4495 vendor operating system shipped in /usr, and not the
4496 configuration stored in /etc.
4498 * .mount units gained a new boolean SloppyOptions= setting
4499 that maps to mount(8)'s -s option which enables permissive
4500 parsing of unknown mount options.
4502 * tmpfiles learnt a new "L+" directive which creates a symlink
4503 but (unlike "L") deletes a pre-existing file first, should
4504 it already exist and not already be the correct
4505 symlink. Similarly, "b+", "c+" and "p+" directives have been
4506 added as well, which create block and character devices, as
4507 well as fifos in the filesystem, possibly removing any
4508 pre-existing files of different types.
4510 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4511 'argument' field (which so far specified the source to
4512 symlink/copy the files from) is now optional. If omitted the
4513 same file os copied from /usr/share/factory/ suffixed by the
4514 full destination path. This is useful for populating /etc
4515 with essential files, by copying them from vendor defaults
4516 shipped in /usr/share/factory/etc.
4518 * A new command "systemctl preset-all" has been added that
4519 applies the service preset settings to all installed unit
4520 files. A new switch --preset-mode= has been added that
4521 controls whether only enable or only disable operations
4524 * A new command "systemctl is-system-running" has been added
4525 that allows checking the overall state of the system, for
4526 example whether it is fully up and running.
4528 * When the system boots up with an empty /etc, the equivalent
4529 to "systemctl preset-all" is executed during early boot, to
4530 make sure all default services are enabled after a factory
4533 * systemd now contains a minimal preset file that enables the
4534 most basic services systemd ships by default.
4536 * Unit files' [Install] section gained a new DefaultInstance=
4537 field for defining the default instance to create if a
4538 template unit is enabled with no instance specified.
4540 * A new passive target cryptsetup-pre.target has been added
4541 that may be used by services that need to make they run and
4542 finish before the first LUKS cryptographic device is set up.
4544 * The /dev/loop-control and /dev/btrfs-control device nodes
4545 are now owned by the "disk" group by default, opening up
4546 access to this group.
4548 * systemd-coredump will now automatically generate a
4549 stack trace of all core dumps taking place on the system,
4550 based on elfutils' libdw library. This stack trace is logged
4553 * systemd-coredump may now optionally store coredumps directly
4554 on disk (in /var/lib/systemd/coredump, possibly compressed),
4555 instead of storing them unconditionally in the journal. This
4556 mode is the new default. A new configuration file
4557 /etc/systemd/coredump.conf has been added to configure this
4558 and other parameters of systemd-coredump.
4560 * coredumpctl gained a new "info" verb to show details about a
4561 specific coredump. A new switch "-1" has also been added
4562 that makes sure to only show information about the most
4563 recent entry instead of all entries. Also, as the tool is
4564 generally useful now the "systemd-" prefix of the binary
4565 name has been removed. Distributions that want to maintain
4566 compatibility with the old name should add a symlink from
4567 the old name to the new name.
4569 * journald's SplitMode= now defaults to "uid". This makes sure
4570 that unprivileged users can access their own coredumps with
4571 coredumpctl without restrictions.
4573 * New kernel command line options "systemd.wants=" (for
4574 pulling an additional unit during boot), "systemd.mask="
4575 (for masking a specific unit for the boot), and
4576 "systemd.debug-shell" (for enabling the debug shell on tty9)
4577 have been added. This is implemented in the new generator
4578 "systemd-debug-generator".
4580 * systemd-nspawn will now by default filter a couple of
4581 syscalls for containers, among them those required for
4582 kernel module loading, direct x86 IO port access, swap
4583 management, and kexec. Most importantly though
4584 open_by_handle_at() is now prohibited for containers,
4585 closing a hole similar to a recently discussed vulnerability
4586 in docker regarding access to files on file hierarchies the
4587 container should normally not have access to. Note that, for
4588 nspawn, we generally make no security claims anyway (and
4589 this is explicitly documented in the man page), so this is
4590 just a fix for one of the most obvious problems.
4592 * A new man page file-hierarchy(7) has been added that
4593 contains a minimized, modernized version of the file system
4594 layout systemd expects, similar in style to the FHS
4595 specification or hier(5). A new tool systemd-path(1) has
4596 been added to query many of these paths for the local
4599 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4600 longer done. Since the directory now has a per-user size
4601 limit, and is cleaned on logout this appears unnecessary,
4602 in particular since this now brings the lifecycle of this
4603 directory closer in line with how IPC objects are handled.
4605 * systemd.pc now exports a number of additional directories,
4606 including $libdir (which is useful to identify the library
4607 path for the primary architecture of the system), and a
4608 couple of drop-in directories.
4610 * udev's predictable network interface names now use the dev_port
4611 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4612 distinguish between ports of the same PCI function. dev_id should
4613 only be used for ports using the same HW address, hence the need
4616 * machined has been updated to export the OS version of a
4617 container (read from /etc/os-release and
4618 /usr/lib/os-release) on the bus. This is now shown in
4619 "machinectl status" for a machine.
4621 * A new service setting RestartForceExitStatus= has been
4622 added. If configured to a set of exit signals or process
4623 return values, the service will be restarted when the main
4624 daemon process exits with any of them, regardless of the
4627 * systemctl's -H switch for connecting to remote systemd
4628 machines has been extended so that it may be used to
4629 directly connect to a specific container on the
4630 host. "systemctl -H root@foobar:waldi" will now connect as
4631 user "root" to host "foobar", and then proceed directly to
4632 the container named "waldi". Note that currently you have to
4633 authenticate as user "root" for this to work, as entering
4634 containers is a privileged operation.
4636 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4637 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4638 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4639 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4640 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4641 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4642 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4643 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4644 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4645 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4646 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4647 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4649 — Berlin, 2014-07-03
4653 * As an experimental feature, udev now tries to lock the
4654 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4655 executes events for the disk or any of its partitions.
4656 Applications like partitioning programs can lock the
4657 disk device node (flock(LOCK_EX)) and claim temporary
4658 device ownership that way; udev will entirely skip all event
4659 handling for this disk and its partitions. If the disk
4660 was opened for writing, the close will trigger a partition
4661 table rescan in udev's "watch" facility, and if needed
4662 synthesize "change" events for the disk and all its partitions.
4663 This is now unconditionally enabled, and if it turns out to
4664 cause major problems, we might turn it on only for specific
4665 devices, or might need to disable it entirely. Device Mapper
4666 devices are excluded from this logic.
4668 * We temporarily dropped the "-l" switch for fsck invocations,
4669 since they collide with the flock() logic above. util-linux
4670 upstream has been changed already to avoid this conflict,
4671 and we will readd "-l" as soon as util-linux with this
4672 change has been released.
4674 * The dependency on libattr has been removed. Since a long
4675 time, the extended attribute calls have moved to glibc, and
4676 libattr is thus unnecessary.
4678 * Virtualization detection works without privileges now. This
4679 means the systemd-detect-virt binary no longer requires
4680 CAP_SYS_PTRACE file capabilities, and our daemons can run
4681 with fewer privileges.
4683 * systemd-networkd now runs under its own "systemd-network"
4684 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4685 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4686 loses the ability to write to files owned by root this way.
4688 * Similarly, systemd-resolved now runs under its own
4689 "systemd-resolve" user with no capabilities remaining.
4691 * Similarly, systemd-bus-proxyd now runs under its own
4692 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4694 * systemd-networkd gained support for setting up "veth"
4695 virtual Ethernet devices for container connectivity, as well
4696 as GRE and VTI tunnels.
4698 * systemd-networkd will no longer automatically attempt to
4699 manually load kernel modules necessary for certain tunnel
4700 transports. Instead, it is assumed the kernel loads them
4701 automatically when required. This only works correctly on
4702 very new kernels. On older kernels, please consider adding
4703 the kernel modules to /etc/modules-load.d/ as a work-around.
4705 * The resolv.conf file systemd-resolved generates has been
4706 moved to /run/systemd/resolve/. If you have a symlink from
4707 /etc/resolv.conf, it might be necessary to correct it.
4709 * Two new service settings, ProtectHome= and ProtectSystem=,
4710 have been added. When enabled, they will make the user data
4711 (such as /home) inaccessible or read-only and the system
4712 (such as /usr) read-only, for specific services. This allows
4713 very light-weight per-service sandboxing to avoid
4714 modifications of user data or system files from
4715 services. These two new switches have been enabled for all
4716 of systemd's long-running services, where appropriate.
4718 * Socket units gained new SocketUser= and SocketGroup=
4719 settings to set the owner user and group of AF_UNIX sockets
4720 and FIFOs in the file system.
4722 * Socket units gained a new RemoveOnStop= setting. If enabled,
4723 all FIFOS and sockets in the file system will be removed
4724 when the specific socket unit is stopped.
4726 * Socket units gained a new Symlinks= setting. It takes a list
4727 of symlinks to create to file system sockets or FIFOs
4728 created by the specific Unix sockets. This is useful to
4729 manage symlinks to socket nodes with the same life-cycle as
4732 * The /dev/log socket and /dev/initctl FIFO have been moved to
4733 /run, and have been replaced by symlinks. This allows
4734 connecting to these facilities even if PrivateDevices=yes is
4735 used for a service (which makes /dev/log itself unavailable,
4736 but /run is left). This also has the benefit of ensuring
4737 that /dev only contains device nodes, directories and
4738 symlinks, and nothing else.
4740 * sd-daemon gained two new calls sd_pid_notify() and
4741 sd_pid_notifyf(). They are similar to sd_notify() and
4742 sd_notifyf(), but allow overriding of the source PID of
4743 notification messages if permissions permit this. This is
4744 useful to send notify messages on behalf of a different
4745 process (for example, the parent process). The
4746 systemd-notify tool has been updated to make use of this
4747 when sending messages (so that notification messages now
4748 originate from the shell script invoking systemd-notify and
4749 not the systemd-notify process itself. This should minimize
4750 a race where systemd fails to associate notification
4751 messages to services when the originating process already
4754 * A new "on-abnormal" setting for Restart= has been added. If
4755 set, it will result in automatic restarts on all "abnormal"
4756 reasons for a process to exit, which includes unclean
4757 signals, core dumps, timeouts and watchdog timeouts, but
4758 does not include clean and unclean exit codes or clean
4759 signals. Restart=on-abnormal is an alternative for
4760 Restart=on-failure for services that shall be able to
4761 terminate and avoid restarts on certain errors, by
4762 indicating so with an unclean exit code. Restart=on-failure
4763 or Restart=on-abnormal is now the recommended setting for
4764 all long-running services.
4766 * If the InaccessibleDirectories= service setting points to a
4767 mount point (or if there are any submounts contained within
4768 it), it is now attempted to completely unmount it, to make
4769 the file systems truly unavailable for the respective
4772 * The ReadOnlyDirectories= service setting and
4773 systemd-nspawn's --read-only parameter are now recursively
4774 applied to all submounts, too.
4776 * Mount units may now be created transiently via the bus APIs.
4778 * The support for SysV and LSB init scripts has been removed
4779 from the systemd daemon itself. Instead, it is now
4780 implemented as a generator that creates native systemd units
4781 from these scripts when needed. This enables us to remove a
4782 substantial amount of legacy code from PID 1, following the
4783 fact that many distributions only ship a very small number
4784 of LSB/SysV init scripts nowadays.
4786 * Privileged Xen (dom0) domains are not considered
4787 virtualization anymore by the virtualization detection
4788 logic. After all, they generally have unrestricted access to
4789 the hardware and usually are used to manage the unprivileged
4792 * systemd-tmpfiles gained a new "C" line type, for copying
4793 files or entire directories.
4795 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4796 lines. So far, they have been non-globbing versions of the
4797 latter, and have thus been redundant. In future, it is
4798 recommended to only use "z". "m" has hence been removed
4799 from the documentation, even though it stays supported.
4801 * A tmpfiles snippet to recreate the most basic structure in
4802 /var has been added. This is enough to create the /var/run →
4803 /run symlink and create a couple of structural
4804 directories. This allows systems to boot up with an empty or
4805 volatile /var. Of course, while with this change, the core OS
4806 now is capable with dealing with a volatile /var, not all
4807 user services are ready for it. However, we hope that sooner
4808 or later, many service daemons will be changed upstream so
4809 that they are able to automatically create their necessary
4810 directories in /var at boot, should they be missing. This is
4811 the first step to allow state-less systems that only require
4812 the vendor image for /usr to boot.
4814 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4815 empty tmpfs instance to a specific directory. This is
4816 particularly useful for making use of the automatic
4817 reconstruction of /var (see above), by passing --tmpfs=/var.
4819 * Access modes specified in tmpfiles snippets may now be
4820 prefixed with "~", which indicates that they shall be masked
4821 by whether the existing file or directory is currently
4822 writable, readable or executable at all. Also, if specified,
4823 the sgid/suid/sticky bits will be masked for all
4826 * A new passive target unit "network-pre.target" has been
4827 added which is useful for services that shall run before any
4828 network is configured, for example firewall scripts.
4830 * The "floppy" group that previously owned the /dev/fd*
4831 devices is no longer used. The "disk" group is now used
4832 instead. Distributions should probably deprecate usage of
4835 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4836 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4837 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4838 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4839 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4840 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4843 — Berlin, 2014-06-11
4847 * A new "systemd-timesyncd" daemon has been added for
4848 synchronizing the system clock across the network. It
4849 implements an SNTP client. In contrast to NTP
4850 implementations such as chrony or the NTP reference server,
4851 this only implements a client side, and does not bother with
4852 the full NTP complexity, focusing only on querying time from
4853 one remote server and synchronizing the local clock to
4854 it. Unless you intend to serve NTP to networked clients or
4855 want to connect to local hardware clocks, this simple NTP
4856 client should be more than appropriate for most
4857 installations. The daemon runs with minimal privileges, and
4858 has been hooked up with networkd to only operate when
4859 network connectivity is available. The daemon saves the
4860 current clock to disk every time a new NTP sync has been
4861 acquired, and uses this to possibly correct the system clock
4862 early at bootup, in order to accommodate for systems that
4863 lack an RTC such as the Raspberry Pi and embedded devices,
4864 and to make sure that time monotonically progresses on these
4865 systems, even if it is not always correct. To make use of
4866 this daemon, a new system user and group "systemd-timesync"
4867 needs to be created on installation of systemd.
4869 * The queue "seqnum" interface of libudev has been disabled, as
4870 it was generally incompatible with device namespacing as
4871 sequence numbers of devices go "missing" if the devices are
4872 part of a different namespace.
4874 * "systemctl list-timers" and "systemctl list-sockets" gained
4875 a --recursive switch for showing units of these types also
4876 for all local containers, similar in style to the already
4877 supported --recursive switch for "systemctl list-units".
4879 * A new RebootArgument= setting has been added for service
4880 units, which may be used to specify a kernel reboot argument
4881 to use when triggering reboots with StartLimitAction=.
4883 * A new FailureAction= setting has been added for service
4884 units which may be used to specify an operation to trigger
4885 when a service fails. This works similarly to
4886 StartLimitAction=, but unlike it, controls what is done
4887 immediately rather than only after several attempts to
4888 restart the service in question.
4890 * hostnamed got updated to also expose the kernel name,
4891 release, and version on the bus. This is useful for
4892 executing commands like hostnamectl with the -H switch.
4893 systemd-analyze makes use of this to properly display
4894 details when running non-locally.
4896 * The bootchart tool can now show cgroup information in the
4897 graphs it generates.
4899 * The CFS CPU quota cgroup attribute is now exposed for
4900 services. The new CPUQuota= switch has been added for this
4901 which takes a percentage value. Setting this will have the
4902 result that a service may never get more CPU time than the
4903 specified percentage, even if the machine is otherwise idle.
4905 * systemd-networkd learned IPIP and SIT tunnel support.
4907 * LSB init scripts exposing a dependency on $network will now
4908 get a dependency on network-online.target rather than simply
4909 network.target. This should bring LSB handling closer to
4910 what it was on SysV systems.
4912 * A new fsck.repair= kernel option has been added to control
4913 how fsck shall deal with unclean file systems at boot.
4915 * The (.ini) configuration file parser will now silently
4916 ignore sections whose name begins with "X-". This may be
4917 used to maintain application-specific extension sections in unit
4920 * machined gained a new API to query the IP addresses of
4921 registered containers. "machinectl status" has been updated
4922 to show these addresses in its output.
4924 * A new call sd_uid_get_display() has been added to the
4925 sd-login APIs for querying the "primary" session of a
4926 user. The "primary" session of the user is elected from the
4927 user's sessions and generally a graphical session is
4928 preferred over a text one.
4930 * A minimal systemd-resolved daemon has been added. It
4931 currently simply acts as a companion to systemd-networkd and
4932 manages resolv.conf based on per-interface DNS
4933 configuration, possibly supplied via DHCP. In the long run
4934 we hope to extend this into a local DNSSEC enabled DNS and
4937 * The systemd-networkd-wait-online tool is now enabled by
4938 default. It will delay network-online.target until a network
4939 connection has been configured. The tool primarily integrates
4940 with networkd, but will also make a best effort to make sense
4941 of network configuration performed in some other way.
4943 * Two new service options StartupCPUShares= and
4944 StartupBlockIOWeight= have been added that work similarly to
4945 CPUShares= and BlockIOWeight= however only apply during
4946 system startup. This is useful to prioritize certain services
4947 differently during bootup than during normal runtime.
4949 * hostnamed has been changed to prefer the statically
4950 configured hostname in /etc/hostname (unless set to
4951 'localhost' or empty) over any dynamic one supplied by
4952 dhcp. With this change, the rules for picking the hostname
4953 match more closely the rules of other configuration settings
4954 where the local administrator's configuration in /etc always
4955 overrides any other settings.
4957 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4958 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4959 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4960 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4961 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4962 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4963 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4964 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4965 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4966 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4967 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4968 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4969 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4970 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4971 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4972 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4975 — Beijing, 2014-05-28
4979 * When restoring the screen brightness at boot, stay away from
4980 the darkest setting or from the lowest 5% of the available
4981 range, depending on which is the larger value of both. This
4982 should effectively protect the user from rebooting into a
4983 black screen, should the brightness have been set to minimum
4986 * sd-login gained a new sd_machine_get_class() call to
4987 determine the class ("vm" or "container") of a machine
4988 registered with machined.
4990 * sd-login gained new calls
4991 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4992 to query the identity of the peer of a local AF_UNIX
4993 connection. They operate similarly to their sd_pid_get_xyz()
4996 * PID 1 will now maintain a system-wide system state engine
4997 with the states "starting", "running", "degraded",
4998 "maintenance", "stopping". These states are bound to system
4999 startup, normal runtime, runtime with at least one failed
5000 service, rescue/emergency mode and system shutdown. This
5001 state is shown in the "systemctl status" output when no unit
5002 name is passed. It is useful to determine system state, in
5003 particularly when doing so for many systems or containers at
5006 * A new command "list-machines" has been added to "systemctl"
5007 that lists all local OS containers and shows their system
5008 state (see above), if systemd runs inside of them.
5010 * systemctl gained a new "-r" switch to recursively enumerate
5011 units on all local containers, when used with the
5012 "list-unit" command (which is the default one that is
5013 executed when no parameters are specified).
5015 * The GPT automatic partition discovery logic will now honour
5016 two GPT partition flags: one may be set on a partition to
5017 cause it to be mounted read-only, and the other may be set
5018 on a partition to ignore it during automatic discovery.
5020 * Two new GPT type UUIDs have been added for automatic root
5021 partition discovery, for 32-bit and 64-bit ARM. This is not
5022 particularly useful for discovering the root directory on
5023 these architectures during bare-metal boots (since UEFI is
5024 not common there), but still very useful to allow booting of
5025 ARM disk images in nspawn with the -i option.
5027 * MAC addresses of interfaces created with nspawn's
5028 --network-interface= switch will now be generated from the
5029 machine name, and thus be stable between multiple invocations
5032 * logind will now automatically remove all IPC objects owned
5033 by a user if she or he fully logs out. This makes sure that
5034 users who are logged out cannot continue to consume IPC
5035 resources. This covers SysV memory, semaphores and message
5036 queues as well as POSIX shared memory and message
5037 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5038 limits. With this functionality, that is corrected. This may
5039 be turned off by using the RemoveIPC= switch of logind.conf.
5041 * The systemd-machine-id-setup and tmpfiles tools gained a
5042 --root= switch to operate on a specific root directory,
5045 * journald can now forward logged messages to the TTYs of all
5046 logged in users ("wall"). This is the default for all
5047 emergency messages now.
5049 * A new tool systemd-journal-remote has been added to stream
5050 journal log messages across the network.
5052 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5053 controller trees are mounted into it. Note that the
5054 directories mounted beneath it are not read-only. This is a
5055 security measure and is particularly useful because glibc
5056 actually includes a search logic to pick any tmpfs it can
5057 find to implement shm_open() if /dev/shm is not available
5058 (which it might very well be in namespaced setups).
5060 * machinectl gained a new "poweroff" command to cleanly power
5061 down a local OS container.
5063 * The PrivateDevices= unit file setting will now also drop the
5064 CAP_MKNOD capability from the capability bound set, and
5065 imply DevicePolicy=closed.
5067 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5068 comprehensively on all long-running systemd services where
5069 this is appropriate.
5071 * systemd-udevd will now run in a disassociated mount
5072 namespace. To mount directories from udev rules, make sure to
5073 pull in mount units via SYSTEMD_WANTS properties.
5075 * The kdbus support gained support for uploading policy into
5076 the kernel. sd-bus gained support for creating "monitoring"
5077 connections that can eavesdrop into all bus communication
5078 for debugging purposes.
5080 * Timestamps may now be specified in seconds since the UNIX
5081 epoch Jan 1st, 1970 by specifying "@" followed by the value
5084 * Native tcpwrap support in systemd has been removed. tcpwrap
5085 is old code, not really maintained anymore and has serious
5086 shortcomings, and better options such as firewalls
5087 exist. For setups that require tcpwrap usage, please
5088 consider invoking your socket-activated service via tcpd,
5089 like on traditional inetd.
5091 * A new system.conf configuration option
5092 DefaultTimerAccuracySec= has been added that controls the
5093 default AccuracySec= setting of .timer units.
5095 * Timer units gained a new WakeSystem= switch. If enabled,
5096 timers configured this way will cause the system to resume
5097 from system suspend (if the system supports that, which most
5100 * Timer units gained a new Persistent= switch. If enabled,
5101 timers configured this way will save to disk when they have
5102 been last triggered. This information is then used on next
5103 reboot to possible execute overdue timer events, that
5104 could not take place because the system was powered off.
5105 This enables simple anacron-like behaviour for timer units.
5107 * systemctl's "list-timers" will now also list the time a
5108 timer unit was last triggered in addition to the next time
5109 it will be triggered.
5111 * systemd-networkd will now assign predictable IPv4LL
5112 addresses to its local interfaces.
5114 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5115 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5116 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5117 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5118 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5119 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5120 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5121 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5124 — Berlin, 2014-03-25
5128 * A new unit file setting RestrictAddressFamilies= has been
5129 added to restrict which socket address families unit
5130 processes gain access to. This takes address family names
5131 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5132 attack surface of services via exotic protocol stacks. This
5133 is built on seccomp system call filters.
5135 * Two new unit file settings RuntimeDirectory= and
5136 RuntimeDirectoryMode= have been added that may be used to
5137 manage a per-daemon runtime directories below /run. This is
5138 an alternative for setting up directory permissions with
5139 tmpfiles snippets, and has the advantage that the runtime
5140 directory's lifetime is bound to the daemon runtime and that
5141 the daemon starts up with an empty directory each time. This
5142 is particularly useful when writing services that drop
5143 privileges using the User= or Group= setting.
5145 * The DeviceAllow= unit setting now supports globbing for
5146 matching against device group names.
5148 * The systemd configuration file system.conf gained new
5149 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5150 DefaultMemoryAccounting= to globally turn on/off accounting
5151 for specific resources (cgroups) for all units. These
5152 settings may still be overridden individually in each unit
5155 * systemd-gpt-auto-generator is now able to discover /srv and
5156 root partitions in addition to /home and swap partitions. It
5157 also supports LUKS-encrypted partitions now. With this in
5158 place, automatic discovery of partitions to mount following
5159 the Discoverable Partitions Specification
5160 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5161 is now a lot more complete. This allows booting without
5162 /etc/fstab and without root= on the kernel command line on
5163 systems prepared appropriately.
5165 * systemd-nspawn gained a new --image= switch which allows
5166 booting up disk images and Linux installations on any block
5167 device that follow the Discoverable Partitions Specification
5168 (see above). This means that installations made with
5169 appropriately updated installers may now be started and
5170 deployed using container managers, completely
5171 unmodified. (We hope that libvirt-lxc will add support for
5172 this feature soon, too.)
5174 * systemd-nspawn gained a new --network-macvlan= setting to
5175 set up a private macvlan interface for the
5176 container. Similarly, systemd-networkd gained a new
5177 Kind=macvlan setting in .netdev files.
5179 * systemd-networkd now supports configuring local addresses
5182 * A new tool systemd-network-wait-online has been added to
5183 synchronously wait for network connectivity using
5186 * The sd-bus.h bus API gained a new sd_bus_track object for
5187 tracking the life-cycle of bus peers. Note that sd-bus.h is
5188 still not a public API though (unless you specify
5189 --enable-kdbus on the configure command line, which however
5190 voids your warranty and you get no API stability guarantee).
5192 * The $XDG_RUNTIME_DIR runtime directories for each user are
5193 now individual tmpfs instances, which has the benefit of
5194 introducing separate pools for each user, with individual
5195 size limits, and thus making sure that unprivileged clients
5196 can no longer negatively impact the system or other users by
5197 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5198 RuntimeDirectorySize= has been introduced that allows
5199 controlling the default size limit for all users. It
5200 defaults to 10% of the available physical memory. This is no
5201 replacement for quotas on tmpfs though (which the kernel
5202 still does not support), as /dev/shm and /tmp are still
5203 shared resources used by both the system and unprivileged
5206 * logind will now automatically turn off automatic suspending
5207 on laptop lid close when more than one display is
5208 connected. This was previously expected to be implemented
5209 individually in desktop environments (such as GNOME),
5210 however has been added to logind now, in order to fix a
5211 boot-time race where a desktop environment might not have
5212 been started yet and thus not been able to take an inhibitor
5213 lock at the time where logind already suspends the system
5214 due to a closed lid.
5216 * logind will now wait at least 30s after each system
5217 suspend/resume cycle, and 3min after system boot before
5218 suspending the system due to a closed laptop lid. This
5219 should give USB docking stations and similar enough time to
5220 be probed and configured after system resume and boot in
5221 order to then act as suspend blocker.
5223 * systemd-run gained a new --property= setting which allows
5224 initialization of resource control properties (and others)
5225 for the created scope or service unit. Example: "systemd-run
5226 --property=BlockIOWeight=10 updatedb" may be used to run
5227 updatedb at a low block IO scheduling weight.
5229 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5230 now also work in --scope mode.
5232 * When systemd is compiled with kdbus support, basic support
5233 for enforced policies is now in place. (Note that enabling
5234 kdbus still voids your warranty and no API compatibility
5237 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5238 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5239 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5240 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5241 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5242 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5243 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5244 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5245 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5246 Zbigniew Jędrzejewski-Szmek
5248 — Berlin, 2014-03-12
5252 * systemd will now relabel /dev after loading the SMACK policy
5253 according to SMACK rules.
5255 * A new unit file option AppArmorProfile= has been added to
5256 set the AppArmor profile for the processes of a unit.
5258 * A new condition check ConditionArchitecture= has been added
5259 to conditionalize units based on the system architecture, as
5260 reported by uname()'s "machine" field.
5262 * systemd-networkd now supports matching on the system
5263 virtualization, architecture, kernel command line, host name
5266 * logind is now a lot more aggressive when suspending the
5267 machine due to a closed laptop lid. Instead of acting only
5268 on the lid close action, it will continuously watch the lid
5269 status and act on it. This is useful for laptops where the
5270 power button is on the outside of the chassis so that it can
5271 be reached without opening the lid (such as the Lenovo
5272 Yoga). On those machines, logind will now immediately
5273 re-suspend the machine if the power button has been
5274 accidentally pressed while the laptop was suspended and in a
5275 backpack or similar.
5277 * logind will now watch SW_DOCK switches and inhibit reaction
5278 to the lid switch if it is pressed. This means that logind
5279 will not suspend the machine anymore if the lid is closed
5280 and the system is docked, if the laptop supports SW_DOCK
5281 notifications via the input layer. Note that ACPI docking
5282 stations do not generate this currently. Also note that this
5283 logic is usually not fully sufficient and Desktop
5284 Environments should take a lid switch inhibitor lock when an
5285 external display is connected, as systemd will not watch
5288 * nspawn will now make use of the devices cgroup controller by
5289 default, and only permit creation of and access to the usual
5290 API device nodes like /dev/null or /dev/random, as well as
5291 access to (but not creation of) the pty devices.
5293 * We will now ship a default .network file for
5294 systemd-networkd that automatically configures DHCP for
5295 network interfaces created by nspawn's --network-veth or
5296 --network-bridge= switches.
5298 * systemd will now understand the usual M, K, G, T suffixes
5299 according to SI conventions (i.e. to the base 1000) when
5300 referring to throughput and hardware metrics. It will stay
5301 with IEC conventions (i.e. to the base 1024) for software
5302 metrics, according to what is customary according to
5303 Wikipedia. We explicitly document which base applies for
5304 each configuration option.
5306 * The DeviceAllow= setting in unit files now supports a syntax
5307 to whitelist an entire group of devices node majors at once,
5308 based on the /proc/devices listing. For example, with the
5309 string "char-pts", it is now possible to whitelist all
5310 current and future pseudo-TTYs at once.
5312 * sd-event learned a new "post" event source. Event sources of
5313 this type are triggered by the dispatching of any event
5314 source of a type that is not "post". This is useful for
5315 implementing clean-up and check event sources that are
5316 triggered by other work being done in the program.
5318 * systemd-networkd is no longer statically enabled, but uses
5319 the usual [Install] sections so that it can be
5320 enabled/disabled using systemctl. It still is enabled by
5323 * When creating a veth interface pair with systemd-nspawn, the
5324 host side will now be prefixed with "vb-" if
5325 --network-bridge= is used, and with "ve-" if --network-veth
5326 is used. This way, it is easy to distinguish these cases on
5327 the host, for example to apply different configuration to
5328 them with systemd-networkd.
5330 * The compatibility libraries for libsystemd-journal.so,
5331 libsystem-id128.so, libsystemd-login.so and
5332 libsystemd-daemon.so do not make use of IFUNC
5333 anymore. Instead, we now build libsystemd.so multiple times
5334 under these alternative names. This means that the footprint
5335 is drastically increased, but given that these are
5336 transitional compatibility libraries, this should not matter
5337 much. This change has been made necessary to support the ARM
5338 platform for these compatibility libraries, as the ARM
5339 toolchain is not really at the same level as the toolchain
5340 for other architectures like x86 and does not support
5341 IFUNC. Please make sure to use --enable-compat-libs only
5342 during a transitional period!
5344 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5345 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5346 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5347 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5348 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5349 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5350 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5351 Zbigniew Jędrzejewski-Szmek
5353 — Berlin, 2014-02-24
5357 * A new component "systemd-networkd" has been added that can
5358 be used to configure local network interfaces statically or
5359 via DHCP. It is capable of bringing up bridges, VLANs, and
5360 bonding. Currently, no hook-ups for interactive network
5361 configuration are provided. Use this for your initrd,
5362 container, embedded, or server setup if you need a simple,
5363 yet powerful, network configuration solution. This
5364 configuration subsystem is quite nifty, as it allows wildcard
5365 hotplug matching in interfaces. For example, with a single
5366 configuration snippet, you can configure that all Ethernet
5367 interfaces showing up are automatically added to a bridge,
5368 or similar. It supports link-sensing and more.
5370 * A new tool "systemd-socket-proxyd" has been added which can
5371 act as a bidirectional proxy for TCP sockets. This is
5372 useful for adding socket activation support to services that
5373 do not actually support socket activation, including virtual
5374 machines and the like.
5376 * Add a new tool to save/restore rfkill state on
5379 * Save/restore state of keyboard backlights in addition to
5380 display backlights on shutdown/boot.
5382 * udev learned a new SECLABEL{} construct to label device
5383 nodes with a specific security label when they appear. For
5384 now, only SECLABEL{selinux} is supported, but the syntax is
5385 prepared for additional security frameworks.
5387 * udev gained a new scheme to configure link-level attributes
5388 from files in /etc/systemd/network/*.link. These files can
5389 match against MAC address, device path, driver name and type,
5390 and will apply attributes like the naming policy, link speed,
5391 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5392 address assignment policy (randomized, ...).
5394 * The configuration of network interface naming rules for
5395 "permanent interface names" has changed: a new NamePolicy=
5396 setting in the [Link] section of .link files determines the
5397 priority of possible naming schemes (onboard, slot, MAC,
5398 path). The default value of this setting is determined by
5399 /usr/lib/net/links/99-default.link. Old
5400 80-net-name-slot.rules udev configuration file has been
5401 removed, so local configuration overriding this file should
5402 be adapted to override 99-default.link instead.
5404 * When the User= switch is used in a unit file, also
5405 initialize $SHELL= based on the user database entry.
5407 * systemd no longer depends on libdbus. All communication is
5408 now done with sd-bus, systemd's low-level bus library
5411 * kdbus support has been added to PID 1 itself. When kdbus is
5412 enabled, this causes PID 1 to set up the system bus and
5413 enable support for a new ".busname" unit type that
5414 encapsulates bus name activation on kdbus. It works a little
5415 bit like ".socket" units, except for bus names. A new
5416 generator has been added that converts classic dbus1 service
5417 activation files automatically into native systemd .busname
5420 * sd-bus: add a light-weight vtable implementation that allows
5421 defining objects on the bus with a simple static const
5422 vtable array of its methods, signals and properties.
5424 * systemd will not generate or install static dbus
5425 introspection data anymore to /usr/share/dbus-1/interfaces,
5426 as the precise format of these files is unclear, and
5427 nothing makes use of it.
5429 * A proxy daemon is now provided to proxy clients connecting
5430 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5431 compatibility with classic D-Bus.
5433 * A bus driver implementation has been added that supports the
5434 classic D-Bus bus driver calls on kdbus, also for
5435 compatibility purposes.
5437 * A new API "sd-event.h" has been added that implements a
5438 minimal event loop API built around epoll. It provides a
5439 couple of features that direct epoll usage is lacking:
5440 prioritization of events, scales to large numbers of timer
5441 events, per-event timer slack (accuracy), system-wide
5442 coalescing of timer events, exit handlers, watchdog
5443 supervision support using systemd's sd_notify() API, child
5446 * A new API "sd-rntl.h" has been added that provides an API
5447 around the route netlink interface of the kernel, similar in
5448 style to "sd-bus.h".
5450 * A new API "sd-dhcp-client.h" has been added that provides a
5451 small DHCPv4 client-side implementation. This is used by
5454 * There is a new kernel command line option
5455 "systemd.restore_state=0|1". When set to "0", none of the
5456 systemd tools will restore saved runtime state to hardware
5457 devices. More specifically, the rfkill and backlight states
5460 * The FsckPassNo= compatibility option in mount/service units
5461 has been removed. The fstab generator will now add the
5462 necessary dependencies automatically, and does not require
5463 PID1's support for that anymore.
5465 * journalctl gained a new switch, --list-boots, that lists
5466 recent boots with their times and boot IDs.
5468 * The various tools like systemctl, loginctl, timedatectl,
5469 busctl, systemd-run, ... have gained a new switch "-M" to
5470 connect to a specific, local OS container (as direct
5471 connection, without requiring SSH). This works on any
5472 container that is registered with machined, such as those
5473 created by libvirt-lxc or nspawn.
5475 * systemd-run and systemd-analyze also gained support for "-H"
5476 to connect to remote hosts via SSH. This is particularly
5477 useful for systemd-run because it enables queuing of jobs
5478 onto remote systems.
5480 * machinectl gained a new command "login" to open a getty
5481 login in any local container. This works with any container
5482 that is registered with machined (such as those created by
5483 libvirt-lxc or nspawn), and which runs systemd inside.
5485 * machinectl gained a new "reboot" command that may be used to
5486 trigger a reboot on a specific container that is registered
5487 with machined. This works on any container that runs an init
5488 system of some kind.
5490 * systemctl gained a new "list-timers" command to print a nice
5491 listing of installed timer units with the times they elapse
5494 * Alternative reboot() parameters may now be specified on the
5495 "systemctl reboot" command line and are passed to the
5496 reboot() system call.
5498 * systemctl gained a new --job-mode= switch to configure the
5499 mode to queue a job with. This is a more generic version of
5500 --fail, --irreversible, and --ignore-dependencies, which are
5501 still available but not advertised anymore.
5503 * /etc/systemd/system.conf gained new settings to configure
5504 various default timeouts of units, as well as the default
5505 start limit interval and burst. These may still be overridden
5508 * PID1 will now export on the bus profile data of the security
5509 policy upload process (such as the SELinux policy upload to
5512 * journald: when forwarding logs to the console, include
5513 timestamps (following the setting in
5514 /sys/module/printk/parameters/time).
5516 * OnCalendar= in timer units now understands the special
5517 strings "yearly" and "annually". (Both are equivalent)
5519 * The accuracy of timer units is now configurable with the new
5520 AccuracySec= setting. It defaults to 1min.
5522 * A new dependency type JoinsNamespaceOf= has been added that
5523 allows running two services within the same /tmp and network
5524 namespace, if PrivateNetwork= or PrivateTmp= are used.
5526 * A new command "cat" has been added to systemctl. It outputs
5527 the original unit file of a unit, and concatenates the
5528 contents of additional "drop-in" unit file snippets, so that
5529 the full configuration is shown.
5531 * systemctl now supports globbing on the various "list-xyz"
5532 commands, like "list-units" or "list-sockets", as well as on
5533 those commands which take multiple unit names.
5535 * journalctl's --unit= switch gained support for globbing.
5537 * All systemd daemons now make use of the watchdog logic so
5538 that systemd automatically notices when they hang.
5540 * If the $container_ttys environment variable is set,
5541 getty-generator will automatically spawn a getty for each
5542 listed tty. This is useful for container managers to request
5543 login gettys to be spawned on as many ttys as needed.
5545 * %h, %s, %U specifier support is not available anymore when
5546 used in unit files for PID 1. This is because NSS calls are
5547 not safe from PID 1. They stay available for --user
5548 instances of systemd, and as special case for the root user.
5550 * loginctl gained a new "--no-legend" switch to turn off output
5553 * The "sd-login.h" API gained three new calls:
5554 sd_session_is_remote(), sd_session_get_remote_user(),
5555 sd_session_get_remote_host() to query information about
5558 * The udev hardware database now also carries vendor/product
5559 information of SDIO devices.
5561 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5562 determine whether watchdog notifications are requested by
5565 * Socket-activated per-connection services now include a
5566 short description of the connection parameters in the
5569 * tmpfiles gained a new "--boot" option. When this is not used,
5570 only lines where the command character is not suffixed with
5571 "!" are executed. When this option is specified, those
5572 options are executed too. This partitions tmpfiles
5573 directives into those that can be safely executed at any
5574 time, and those which should be run only at boot (for
5575 example, a line that creates /run/nologin).
5577 * A new API "sd-resolve.h" has been added which provides a simple
5578 asynchronous wrapper around glibc NSS host name resolution
5579 calls, such as getaddrinfo(). In contrast to glibc's
5580 getaddrinfo_a(), it does not use signals. In contrast to most
5581 other asynchronous name resolution libraries, this one does
5582 not reimplement DNS, but reuses NSS, so that alternate
5583 host name resolution systems continue to work, such as mDNS,
5584 LDAP, etc. This API is based on libasyncns, but it has been
5585 cleaned up for inclusion in systemd.
5587 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5588 "sd-daemon.h" are no longer found in individual libraries
5589 libsystemd-journal.so, libsystemd-login.so,
5590 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5591 merged them into a single library, libsystemd.so, which
5592 provides all symbols. The reason for this is cyclic
5593 dependencies, as these libraries tend to use each other's
5594 symbols. So far, we have managed to workaround that by linking
5595 a copy of a good part of our code into each of these
5596 libraries again and again, which, however, makes certain
5597 things hard to do, like sharing static variables. Also, it
5598 substantially increases footprint. With this change, there
5599 is only one library for the basic APIs systemd
5600 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5601 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5602 library as well, however are subject to the --enable-kdbus
5603 switch (see below). Note that "sd-dhcp-client.h" is not part
5604 of this library (this is because it only consumes, never
5605 provides, services of/to other APIs). To make the transition
5606 easy from the separate libraries to the unified one, we
5607 provide the --enable-compat-libs compile-time switch which
5608 will generate stub libraries that are compatible with the
5609 old ones but redirect all calls to the new one.
5611 * All of the kdbus logic and the new APIs "sd-bus.h",
5612 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5613 and "sd-utf8.h" are compile-time optional via the
5614 "--enable-kdbus" switch, and they are not compiled in by
5615 default. To make use of kdbus, you have to explicitly enable
5616 the switch. Note however, that neither the kernel nor the
5617 userspace API for all of this is considered stable yet. We
5618 want to maintain the freedom to still change the APIs for
5619 now. By specifying this build-time switch, you acknowledge
5620 that you are aware of the instability of the current
5623 * Also, note that while kdbus is pretty much complete,
5624 it lacks one thing: proper policy support. This means you
5625 can build a fully working system with all features; however,
5626 it will be highly insecure. Policy support will be added in
5627 one of the next releases, at the same time that we will
5628 declare the APIs stable.
5630 * When the kernel command line argument "kdbus" is specified,
5631 systemd will automatically load the kdbus.ko kernel module. At
5632 this stage of development, it is only useful for testing kdbus
5633 and should not be used in production. Note: if "--enable-kdbus"
5634 is specified, and the kdbus.ko kernel module is available, and
5635 "kdbus" is added to the kernel command line, the entire system
5636 runs with kdbus instead of dbus-daemon, with the above mentioned
5637 problem of missing the system policy enforcement. Also a future
5638 version of kdbus.ko or a newer systemd will not be compatible with
5639 each other, and will unlikely be able to boot the machine if only
5640 one of them is updated.
5642 * systemctl gained a new "import-environment" command which
5643 uploads the caller's environment (or parts thereof) into the
5644 service manager so that it is inherited by services started
5645 by the manager. This is useful to upload variables like
5646 $DISPLAY into the user service manager.
5648 * A new PrivateDevices= switch has been added to service units
5649 which allows running a service with a namespaced /dev
5650 directory that does not contain any device nodes for
5651 physical devices. More specifically, it only includes devices
5652 such as /dev/null, /dev/urandom, and /dev/zero which are API
5655 * logind has been extended to support behaviour like VT
5656 switching on seats that do not support a VT. This makes
5657 multi-session available on seats that are not the first seat
5658 (seat0), and on systems where kernel support for VTs has
5659 been disabled at compile-time.
5661 * If a process holds a delay lock for system sleep or shutdown
5662 and fails to release it in time, we will now log its
5663 identity. This makes it easier to identify processes that
5664 cause slow suspends or power-offs.
5666 * When parsing /etc/crypttab, support for a new key-slot=
5667 option as supported by Debian is added. It allows indicating
5668 which LUKS slot to use on disk, speeding up key loading.
5670 * The sd_journal_sendv() API call has been checked and
5671 officially declared to be async-signal-safe so that it may
5672 be invoked from signal handlers for logging purposes.
5674 * Boot-time status output is now enabled automatically after a
5675 short timeout if boot does not progress, in order to give
5676 the user an indication what she or he is waiting for.
5678 * The boot-time output has been improved to show how much time
5679 remains until jobs expire.
5681 * The KillMode= switch in service units gained a new possible
5682 value "mixed". If set, and the unit is shut down, then the
5683 initial SIGTERM signal is sent only to the main daemon
5684 process, while the following SIGKILL signal is sent to
5685 all remaining processes of the service.
5687 * When a scope unit is registered, a new property "Controller"
5688 may be set. If set to a valid bus name, systemd will send a
5689 RequestStop() signal to this name when it would like to shut
5690 down the scope. This may be used to hook manager logic into
5691 the shutdown logic of scope units. Also, scope units may now
5692 be put in a special "abandoned" state, in which case the
5693 manager process which created them takes no further
5694 responsibilities for it.
5696 * When reading unit files, systemd will now verify
5697 the access mode of these files, and warn about certain
5698 suspicious combinations. This has been added to make it
5699 easier to track down packaging bugs where unit files are
5700 marked executable or world-writable.
5702 * systemd-nspawn gained a new "--setenv=" switch to set
5703 container-wide environment variables. The similar option in
5704 systemd-activate was renamed from "--environment=" to
5705 "--setenv=" for consistency.
5707 * systemd-nspawn has been updated to create a new kdbus domain
5708 for each container that is invoked, thus allowing each
5709 container to have its own set of system and user buses,
5710 independent of the host.
5712 * systemd-nspawn gained a new --drop-capability= switch to run
5713 the container with less capabilities than the default. Both
5714 --drop-capability= and --capability= now take the special
5715 string "all" for dropping or keeping all capabilities.
5717 * systemd-nspawn gained new switches for executing containers
5718 with specific SELinux labels set.
5720 * systemd-nspawn gained a new --quiet switch to not generate
5721 any additional output but the container's own console
5724 * systemd-nspawn gained a new --share-system switch to run a
5725 container without PID namespacing enabled.
5727 * systemd-nspawn gained a new --register= switch to control
5728 whether the container is registered with systemd-machined or
5729 not. This is useful for containers that do not run full
5730 OS images, but only specific apps.
5732 * systemd-nspawn gained a new --keep-unit which may be used
5733 when invoked as the only program from a service unit, and
5734 results in registration of the unit service itself in
5735 systemd-machined, instead of a newly opened scope unit.
5737 * systemd-nspawn gained a new --network-interface= switch for
5738 moving arbitrary interfaces to the container. The new
5739 --network-veth switch creates a virtual Ethernet connection
5740 between host and container. The new --network-bridge=
5741 switch then allows assigning the host side of this virtual
5742 Ethernet connection to a bridge device.
5744 * systemd-nspawn gained a new --personality= switch for
5745 setting the kernel personality for the container. This is
5746 useful when running a 32-bit container on a 64-bit host. A
5747 similar option Personality= is now also available for service
5750 * logind will now also track a "Desktop" identifier for each
5751 session which encodes the desktop environment of it. This is
5752 useful for desktop environments that want to identify
5753 multiple running sessions of itself easily.
5755 * A new SELinuxContext= setting for service units has been
5756 added that allows setting a specific SELinux execution
5757 context for a service.
5759 * Most systemd client tools will now honour $SYSTEMD_LESS for
5760 settings of the "less" pager. By default, these tools will
5761 override $LESS to allow certain operations to work, such as
5762 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5763 influence this logic.
5765 * systemd's "seccomp" hook-up has been changed to make use of
5766 the libseccomp library instead of using its own
5767 implementation. This has benefits for portability among
5770 * For usage together with SystemCallFilter=, a new
5771 SystemCallErrorNumber= setting has been introduced that
5772 allows configuration of a system error number to be returned
5773 on filtered system calls, instead of immediately killing the
5774 process. Also, SystemCallArchitectures= has been added to
5775 limit access to system calls of a particular architecture
5776 (in order to turn off support for unused secondary
5777 architectures). There is also a global
5778 SystemCallArchitectures= setting in system.conf now to turn
5779 off support for non-native system calls system-wide.
5781 * systemd requires a kernel with a working name_to_handle_at(),
5782 please see the kernel config requirements in the README file.
5784 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5785 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5786 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5787 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5788 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5789 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5790 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5791 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5792 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5793 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5794 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5795 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5796 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5797 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5798 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5799 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5800 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5801 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5802 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5803 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5804 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5805 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5806 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5807 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5809 — Berlin, 2014-02-20
5813 * logind has gained support for facilitating privileged input
5814 and drm device access for unprivileged clients. This work is
5815 useful to allow Wayland display servers (and similar
5816 programs, such as kmscon) to run under the user's ID and
5817 access input and drm devices which are normally
5818 protected. When this is used (and the kernel is new enough)
5819 logind will "mute" IO on the file descriptors passed to
5820 Wayland as long as it is in the background and "unmute" it
5821 if it returns into the foreground. This allows secure
5822 session switching without allowing background sessions to
5823 eavesdrop on input and display data. This also introduces
5824 session switching support if VT support is turned off in the
5825 kernel, and on seats that are not seat0.
5827 * A new kernel command line option luks.options= is understood
5828 now which allows specifying LUKS options for usage for LUKS
5829 encrypted partitions specified with luks.uuid=.
5831 * tmpfiles.d(5) snippets may now use specifier expansion in
5832 path names. More specifically %m, %b, %H, %v, are now
5833 replaced by the local machine id, boot id, hostname, and
5834 kernel version number.
5836 * A new tmpfiles.d(5) command "m" has been introduced which
5837 may be used to change the owner/group/access mode of a file
5838 or directory if it exists, but do nothing if it does not.
5840 * This release removes high-level support for the
5841 MemorySoftLimit= cgroup setting. The underlying kernel
5842 cgroup attribute memory.soft_limit= is currently badly
5843 designed and likely to be removed from the kernel API in its
5844 current form, hence we should not expose it for now.
5846 * The memory.use_hierarchy cgroup attribute is now enabled for
5847 all cgroups systemd creates in the memory cgroup
5848 hierarchy. This option is likely to be come the built-in
5849 default in the kernel anyway, and the non-hierarchical mode
5850 never made much sense in the intrinsically hierarchical
5853 * A new field _SYSTEMD_SLICE= is logged along with all journal
5854 messages containing the slice a message was generated
5855 from. This is useful to allow easy per-customer filtering of
5856 logs among other things.
5858 * systemd-journald will no longer adjust the group of journal
5859 files it creates to the "systemd-journal" group. Instead we
5860 rely on the journal directory to be owned by the
5861 "systemd-journal" group, and its setgid bit set, so that the
5862 kernel file system layer will automatically enforce that
5863 journal files inherit this group assignment. The reason for
5864 this change is that we cannot allow NSS look-ups from
5865 journald which would be necessary to resolve
5866 "systemd-journal" to a numeric GID, because this might
5867 create deadlocks if NSS involves synchronous queries to
5868 other daemons (such as nscd, or sssd) which in turn are
5869 logging clients of journald and might block on it, which
5870 would then dead lock. A tmpfiles.d(5) snippet included in
5871 systemd will make sure the setgid bit and group are
5872 properly set on the journal directory if it exists on every
5873 boot. However, we recommend adjusting it manually after
5874 upgrades too (or from RPM scriptlets), so that the change is
5875 not delayed until next reboot.
5877 * Backlight and random seed files in /var/lib/ have moved into
5878 the /var/lib/systemd/ directory, in order to centralize all
5879 systemd generated files in one directory.
5881 * Boot time performance measurements (as displayed by
5882 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5883 performance information if that's available to determine how
5884 much time BIOS and boot loader initialization required. With
5885 a sufficiently new BIOS you hence no longer need to boot
5886 with Gummiboot to get access to such information.
5888 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5889 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5890 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5891 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5892 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5893 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5894 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5896 — Berlin, 2013-10-02
5900 * The Restart= option for services now understands a new
5901 on-watchdog setting, which will restart the service
5902 automatically if the service stops sending out watchdog keep
5903 alive messages (as configured with WatchdogSec=).
5905 * The getty generator (which is responsible for bringing up a
5906 getty on configured serial consoles) will no longer only
5907 start a getty on the primary kernel console but on all
5908 others, too. This makes the order in which console= is
5909 specified on the kernel command line less important.
5911 * libsystemd-logind gained a new sd_session_get_vt() call to
5912 retrieve the VT number of a session.
5914 * If the option "tries=0" is set for an entry of /etc/crypttab
5915 its passphrase is queried indefinitely instead of any
5916 maximum number of tries.
5918 * If a service with a configure PID file terminates its PID
5919 file will now be removed automatically if it still exists
5920 afterwards. This should put an end to stale PID files.
5922 * systemd-run will now also take relative binary path names
5923 for execution and no longer insists on absolute paths.
5925 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5926 paths that are optionally prefixed with "-" to indicate that
5927 it should not be considered a failure if they do not exist.
5929 * journalctl -o (and similar commands) now understands a new
5930 output mode "short-precise", it is similar to "short" but
5931 shows timestamps with usec accuracy.
5933 * The option "discard" (as known from Debian) is now
5934 synonymous to "allow-discards" in /etc/crypttab. In fact,
5935 "discard" is preferred now (since it is easier to remember
5938 * Some licensing clean-ups were made, so that more code is now
5939 LGPL-2.1 licensed than before.
5941 * A minimal tool to save/restore the display backlight
5942 brightness across reboots has been added. It will store the
5943 backlight setting as late as possible at shutdown, and
5944 restore it as early as possible during reboot.
5946 * A logic to automatically discover and enable home and swap
5947 partitions on GPT disks has been added. With this in place
5948 /etc/fstab becomes optional for many setups as systemd can
5949 discover certain partitions located on the root disk
5950 automatically. Home partitions are recognized under their
5951 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5952 partitions are recognized under their GPT type ID
5953 0657fd6da4ab43c484e50933c84b4f4f.
5955 * systemd will no longer pass any environment from the kernel
5956 or initrd to system services. If you want to set an
5957 environment for all services, do so via the kernel command
5958 line systemd.setenv= assignment.
5960 * The systemd-sysctl tool no longer natively reads the file
5961 /etc/sysctl.conf. If desired, the file should be symlinked
5962 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5963 legacy support by a symlink rather than built-in code, it
5964 also makes the otherwise hidden order of application of the
5965 different files visible. (Note that this partly reverts to a
5966 pre-198 application order of sysctl knobs!)
5968 * The "systemctl set-log-level" and "systemctl dump" commands
5969 have been moved to systemd-analyze.
5971 * systemd-run learned the new --remain-after-exit switch,
5972 which causes the scope unit not to be cleaned up
5973 automatically after the process terminated.
5975 * tmpfiles learned a new --exclude-prefix= switch to exclude
5976 certain paths from operation.
5978 * journald will now automatically flush all messages to disk
5979 as soon as a message at the log level CRIT, ALERT or EMERG
5982 Contributions from: Andrew Cook, Brandon Philips, Christian
5983 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5984 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5985 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5986 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5987 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5988 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5989 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5990 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5991 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5992 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5993 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5994 William Giokas, Zbigniew Jędrzejewski-Szmek
5996 — Berlin, 2013-09-13
6000 * The documentation has been updated to cover the various new
6001 concepts introduced with 205.
6003 * Unit files now understand the new %v specifier which
6004 resolves to the kernel version string as returned by "uname
6007 * systemctl now supports filtering the unit list output by
6008 load state, active state and sub state, using the new
6011 * "systemctl status" will now show the results of the
6012 condition checks (like ConditionPathExists= and similar) of
6013 the last start attempts of the unit. They are also logged to
6016 * "journalctl -b" may now be used to look for boot output of a
6017 specific boot. Try "journalctl -b -1" for the previous boot,
6018 but the syntax is substantially more powerful.
6020 * "journalctl --show-cursor" has been added which prints the
6021 cursor string the last shown log line. This may then be used
6022 with the new "journalctl --after-cursor=" switch to continue
6023 browsing logs from that point on.
6025 * "journalctl --force" may now be used to force regeneration
6028 * Creation of "dead" device nodes has been moved from udev
6029 into kmod and tmpfiles. Previously, udev would read the kmod
6030 databases to pre-generate dead device nodes based on meta
6031 information contained in kernel modules, so that these would
6032 be auto-loaded on access rather then at boot. As this
6033 does not really have much to do with the exposing actual
6034 kernel devices to userspace this has always been slightly
6035 alien in the udev codebase. Following the new scheme kmod
6036 will now generate a runtime snippet for tmpfiles from the
6037 module meta information and it now is tmpfiles' job to the
6038 create the nodes. This also allows overriding access and
6039 other parameters for the nodes using the usual tmpfiles
6040 facilities. As side effect this allows us to remove the
6041 CAP_SYS_MKNOD capability bit from udevd entirely.
6043 * logind's device ACLs may now be applied to these "dead"
6044 devices nodes too, thus finally allowing managed access to
6045 devices such as /dev/snd/sequencer without loading the
6046 backing module right-away.
6048 * A new RPM macro has been added that may be used to apply
6049 tmpfiles configuration during package installation.
6051 * systemd-detect-virt and ConditionVirtualization= now can
6052 detect User-Mode-Linux machines (UML).
6054 * journald will now implicitly log the effective capabilities
6055 set of processes in the message metadata.
6057 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6059 * The initrd interface has been simplified (more specifically,
6060 support for passing performance data via environment
6061 variables and fsck results via files in /run has been
6062 removed). These features were non-essential, and are
6063 nowadays available in a much nicer way by having systemd in
6064 the initrd serialize its state and have the hosts systemd
6065 deserialize it again.
6067 * The udev "keymap" data files and tools to apply keyboard
6068 specific mappings of scan to key codes, and force-release
6069 scan code lists have been entirely replaced by a udev
6070 "keyboard" builtin and a hwdb data file.
6072 * systemd will now honour the kernel's "quiet" command line
6073 argument also during late shutdown, resulting in a
6074 completely silent shutdown when used.
6076 * There's now an option to control the SO_REUSEPORT socket
6077 option in .socket units.
6079 * Instance units will now automatically get a per-template
6080 subslice of system.slice unless something else is explicitly
6081 configured. For example, instances of sshd@.service will now
6082 implicitly be placed in system-sshd.slice rather than
6083 system.slice as before.
6085 * Test coverage support may now be enabled at build time.
6087 Contributions from: Dave Reisner, Frederic Crozat, Harald
6088 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6089 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6090 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6091 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6092 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6093 Giokas, Zbigniew Jędrzejewski-Szmek
6095 — Berlin, 2013-07-23
6099 * Two new unit types have been introduced:
6101 Scope units are very similar to service units, however, are
6102 created out of pre-existing processes — instead of PID 1
6103 forking off the processes. By using scope units it is
6104 possible for system services and applications to group their
6105 own child processes (worker processes) in a powerful way
6106 which then maybe used to organize them, or kill them
6107 together, or apply resource limits on them.
6109 Slice units may be used to partition system resources in an
6110 hierarchical fashion and then assign other units to them. By
6111 default there are now three slices: system.slice (for all
6112 system services), user.slice (for all user sessions),
6113 machine.slice (for VMs and containers).
6115 Slices and scopes have been introduced primarily in
6116 context of the work to move cgroup handling to a
6117 single-writer scheme, where only PID 1
6118 creates/removes/manages cgroups.
6120 * There's a new concept of "transient" units. In contrast to
6121 normal units these units are created via an API at runtime,
6122 not from configuration from disk. More specifically this
6123 means it is now possible to run arbitrary programs as
6124 independent services, with all execution parameters passed
6125 in via bus APIs rather than read from disk. Transient units
6126 make systemd substantially more dynamic then it ever was,
6127 and useful as a general batch manager.
6129 * logind has been updated to make use of scope and slice units
6130 for managing user sessions. As a user logs in he will get
6131 his own private slice unit, to which all sessions are added
6132 as scope units. We also added support for automatically
6133 adding an instance of user@.service for the user into the
6134 slice. Effectively logind will no longer create cgroup
6135 hierarchies on its own now, it will defer entirely to PID 1
6136 for this by means of scope, service and slice units. Since
6137 user sessions this way become entities managed by PID 1
6138 the output of "systemctl" is now a lot more comprehensive.
6140 * A new mini-daemon "systemd-machined" has been added which
6141 may be used by virtualization managers to register local
6142 VMs/containers. nspawn has been updated accordingly, and
6143 libvirt will be updated shortly. machined will collect a bit
6144 of meta information about the VMs/containers, and assign
6145 them their own scope unit (see above). The collected
6146 meta-data is then made available via the "machinectl" tool,
6147 and exposed in "ps" and similar tools. machined/machinectl
6148 is compile-time optional.
6150 * As discussed earlier, the low-level cgroup configuration
6151 options ControlGroup=, ControlGroupModify=,
6152 ControlGroupPersistent=, ControlGroupAttribute= have been
6153 removed. Please use high-level attribute settings instead as
6154 well as slice units.
6156 * A new bus call SetUnitProperties() has been added to alter
6157 various runtime parameters of a unit. This is primarily
6158 useful to alter cgroup parameters dynamically in a nice way,
6159 but will be extended later on to make more properties
6160 modifiable at runtime. systemctl gained a new set-properties
6161 command that wraps this call.
6163 * A new tool "systemd-run" has been added which can be used to
6164 run arbitrary command lines as transient services or scopes,
6165 while configuring a number of settings via the command
6166 line. This tool is currently very basic, however already
6167 very useful. We plan to extend this tool to even allow
6168 queuing of execution jobs with time triggers from the
6169 command line, similar in fashion to "at".
6171 * nspawn will now inform the user explicitly that kernels with
6172 audit enabled break containers, and suggest the user to turn
6175 * Support for detecting the IMA and AppArmor security
6176 frameworks with ConditionSecurity= has been added.
6178 * journalctl gained a new "-k" switch for showing only kernel
6179 messages, mimicking dmesg output; in addition to "--user"
6180 and "--system" switches for showing only user's own logs
6183 * systemd-delta can now show information about drop-in
6184 snippets extending unit files.
6186 * libsystemd-bus has been substantially updated but is still
6187 not available as public API.
6189 * systemd will now look for the "debug" argument on the kernel
6190 command line and enable debug logging, similar to what
6191 "systemd.log_level=debug" already did before.
6193 * "systemctl set-default", "systemctl get-default" has been
6194 added to configure the default.target symlink, which
6195 controls what to boot into by default.
6197 * "systemctl set-log-level" has been added as a convenient
6198 way to raise and lower systemd logging threshold.
6200 * "systemd-analyze plot" will now show the time the various
6201 generators needed for execution, as well as information
6202 about the unit file loading.
6204 * libsystemd-journal gained a new sd_journal_open_files() call
6205 for opening specific journal files. journactl also gained a
6206 new switch to expose this new functionality. Previously we
6207 only supported opening all files from a directory, or all
6208 files from the system, as opening individual files only is
6209 racy due to journal file rotation.
6211 * systemd gained the new DefaultEnvironment= setting in
6212 /etc/systemd/system.conf to set environment variables for
6215 * If a privileged process logs a journal message with the
6216 OBJECT_PID= field set, then journald will automatically
6217 augment this with additional OBJECT_UID=, OBJECT_GID=,
6218 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6219 system services want to log events about specific client
6220 processes. journactl/systemctl has been updated to make use
6221 of this information if all log messages regarding a specific
6224 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6225 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6226 Reisner, David Coppa, David King, David Strauss, Eelco
6227 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6228 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6229 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6230 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6231 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6232 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6233 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6234 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6235 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6236 Łukasz Stelmach, 장동준
6240 * The Python bindings gained some minimal support for the APIs
6241 exposed by libsystemd-logind.
6243 * ConditionSecurity= gained support for detecting SMACK. Since
6244 this condition already supports SELinux and AppArmor we only
6245 miss IMA for this. Patches welcome!
6247 Contributions from: Karol Lewandowski, Lennart Poettering,
6248 Zbigniew Jędrzejewski-Szmek
6252 * systemd-nspawn will now create /etc/resolv.conf if
6253 necessary, before bind-mounting the host's file onto it.
6255 * systemd-nspawn will now store meta information about a
6256 container on the container's cgroup as extended attribute
6257 fields, including the root directory.
6259 * The cgroup hierarchy has been reworked in many ways. All
6260 objects any of the components systemd creates in the cgroup
6261 tree are now suffixed. More specifically, user sessions are
6262 now placed in cgroups suffixed with ".session", users in
6263 cgroups suffixed with ".user", and nspawn containers in
6264 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6265 names are now escaped in a simple scheme to avoid collision
6266 of userspace object names with kernel filenames. This work
6267 is preparation for making these objects relocatable in the
6268 cgroup tree, in order to allow easy resource partitioning of
6269 these objects without causing naming conflicts.
6271 * systemctl list-dependencies gained the new switches
6272 --plain, --reverse, --after and --before.
6274 * systemd-inhibit now shows the process name of processes that
6275 have taken an inhibitor lock.
6277 * nss-myhostname will now also resolve "localhost"
6278 implicitly. This makes /etc/hosts an optional file and
6279 nicely handles that on IPv6 ::1 maps to both "localhost" and
6282 * libsystemd-logind.so gained a new call
6283 sd_get_machine_names() to enumerate running containers and
6284 VMs (currently only supported by very new libvirt and
6285 nspawn). sd_login_monitor can now be used to watch
6286 VMs/containers coming and going.
6288 * .include is not allowed recursively anymore, and only in
6289 unit files. Usually it is better to use drop-in snippets in
6290 .d/*.conf anyway, as introduced with systemd 198.
6292 * systemd-analyze gained a new "critical-chain" command that
6293 determines the slowest chain of units run during system
6294 boot-up. It is very useful for tracking down where
6295 optimizing boot time is the most beneficial.
6297 * systemd will no longer allow manipulating service paths in
6298 the name=systemd:/system cgroup tree using ControlGroup= in
6299 units. (But is still fine with it in all other dirs.)
6301 * There's a new systemd-nspawn@.service service file that may
6302 be used to easily run nspawn containers as system
6303 services. With the container's root directory in
6304 /var/lib/container/foobar it is now sufficient to run
6305 "systemctl start systemd-nspawn@foobar.service" to boot it.
6307 * systemd-cgls gained a new parameter "--machine" to list only
6308 the processes within a certain container.
6310 * ConditionSecurity= now can check for "apparmor". We still
6311 are lacking checks for SMACK and IMA for this condition
6312 check though. Patches welcome!
6314 * A new configuration file /etc/systemd/sleep.conf has been
6315 added that may be used to configure which kernel operation
6316 systemd is supposed to execute when "suspend", "hibernate"
6317 or "hybrid-sleep" is requested. This makes the new kernel
6318 "freeze" state accessible to the user.
6320 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6321 the passed argument if applicable.
6323 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6324 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6325 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6326 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6327 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6328 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6329 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6334 * The output of 'systemctl list-jobs' got some polishing. The
6335 '--type=' argument may now be passed more than once. A new
6336 command 'systemctl list-sockets' has been added which shows
6337 a list of kernel sockets systemd is listening on with the
6338 socket units they belong to, plus the units these socket
6341 * The experimental libsystemd-bus library got substantial
6342 updates to work in conjunction with the (also experimental)
6343 kdbus kernel project. It works well enough to exchange
6344 messages with some sophistication. Note that kdbus is not
6345 ready yet, and the library is mostly an elaborate test case
6346 for now, and not installable.
6348 * systemd gained a new unit 'systemd-static-nodes.service'
6349 that generates static device nodes earlier during boot, and
6350 can run in conjunction with udev.
6352 * libsystemd-login gained a new call sd_pid_get_user_unit()
6353 to retrieve the user systemd unit a process is running
6354 in. This is useful for systems where systemd is used as
6357 * systemd-nspawn now places all containers in the new /machine
6358 top-level cgroup directory in the name=systemd
6359 hierarchy. libvirt will soon do the same, so that we get a
6360 uniform separation of /system, /user and /machine for system
6361 services, user processes and containers/virtual
6362 machines. This new cgroup hierarchy is also useful to stick
6363 stable names to specific container instances, which can be
6364 recognized later this way (this name may be controlled
6365 via systemd-nspawn's new -M switch). libsystemd-login also
6366 gained a new call sd_pid_get_machine_name() to retrieve the
6367 name of the container/VM a specific process belongs to.
6369 * bootchart can now store its data in the journal.
6371 * libsystemd-journal gained a new call
6372 sd_journal_add_conjunction() for AND expressions to the
6373 matching logic. This can be used to express more complex
6374 logical expressions.
6376 * journactl can now take multiple --unit= and --user-unit=
6379 * The cryptsetup logic now understands the "luks.key=" kernel
6380 command line switch for specifying a file to read the
6381 decryption key from. Also, if a configured key file is not
6382 found the tool will now automatically fall back to prompting
6385 * Python systemd.journal module was updated to wrap recently
6386 added functions from libsystemd-journal. The interface was
6387 changed to bring the low level interface in s.j._Reader
6388 closer to the C API, and the high level interface in
6389 s.j.Reader was updated to wrap and convert all data about
6392 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6393 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6394 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6395 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6396 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6397 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6401 * journalctl --update-catalog now understands a new --root=
6402 option to operate on catalogs found in a different root
6405 * During shutdown after systemd has terminated all running
6406 services a final killing loop kills all remaining left-over
6407 processes. We will now print the name of these processes
6408 when we send SIGKILL to them, since this usually indicates a
6411 * If /etc/crypttab refers to password files stored on
6412 configured mount points automatic dependencies will now be
6413 generated to ensure the specific mount is established first
6414 before the key file is attempted to be read.
6416 * 'systemctl status' will now show information about the
6417 network sockets a socket unit is listening on.
6419 * 'systemctl status' will also shown information about any
6420 drop-in configuration file for units. (Drop-In configuration
6421 files in this context are files such as
6422 /etc/systemd/systemd/foobar.service.d/*.conf)
6424 * systemd-cgtop now optionally shows summed up CPU times of
6425 cgroups. Press '%' while running cgtop to switch between
6426 percentage and absolute mode. This is useful to determine
6427 which cgroups use up the most CPU time over the entire
6428 runtime of the system. systemd-cgtop has also been updated
6429 to be 'pipeable' for processing with further shell tools.
6431 * 'hostnamectl set-hostname' will now allow setting of FQDN
6434 * The formatting and parsing of time span values has been
6435 changed. The parser now understands fractional expressions
6436 such as "5.5h". The formatter will now output fractional
6437 expressions for all time spans under 1min, i.e. "5.123456s"
6438 rather than "5s 123ms 456us". For time spans under 1s
6439 millisecond values are shown, for those under 1ms
6440 microsecond values are shown. This should greatly improve
6441 all time-related output of systemd.
6443 * libsystemd-login and libsystemd-journal gained new
6444 functions for querying the poll() events mask and poll()
6445 timeout value for integration into arbitrary event
6448 * localectl gained the ability to list available X11 keymaps
6449 (models, layouts, variants, options).
6451 * 'systemd-analyze dot' gained the ability to filter for
6452 specific units via shell-style globs, to create smaller,
6453 more useful graphs. I.e. it is now possible to create simple
6454 graphs of all the dependencies between only target units, or
6455 of all units that Avahi has dependencies with.
6457 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6458 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6459 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6460 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6461 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6462 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6463 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6467 * The boot-time readahead implementation for rotating media
6468 will now read the read-ahead data in multiple passes which
6469 consist of all read requests made in equidistant time
6470 intervals. This means instead of strictly reading read-ahead
6471 data in its physical order on disk we now try to find a
6472 middle ground between physical and access time order.
6474 * /etc/os-release files gained a new BUILD_ID= field for usage
6475 on operating systems that provide continuous builds of OS
6478 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6479 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6480 William Douglas, Zbigniew Jędrzejewski-Szmek
6484 * systemd-python gained an API exposing libsystemd-daemon.
6486 * The SMACK setup logic gained support for uploading CIPSO
6489 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6490 ReadOnlyDirectories= and InaccessibleDirectories= has
6491 changed. The private /tmp and /var/tmp directories are now
6492 shared by all processes of a service (which means
6493 ExecStartPre= may now leave data in /tmp that ExecStart= of
6494 the same service can still access). When a service is
6495 stopped its temporary directories are immediately deleted
6496 (normal clean-up with tmpfiles is still done in addition to
6499 * By default, systemd will now set a couple of sysctl
6500 variables in the kernel: the safe sysrq options are turned
6501 on, IP route verification is turned on, and source routing
6502 disabled. The recently added hardlink and softlink
6503 protection of the kernel is turned on. These settings should
6504 be reasonably safe, and good defaults for all new systems.
6506 * The predictable network naming logic may now be turned off
6507 with a new kernel command line switch: net.ifnames=0.
6509 * A new libsystemd-bus module has been added that implements a
6510 pretty complete D-Bus client library. For details see:
6512 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6514 * journald will now explicitly flush the journal files to disk
6515 at the latest 5min after each write. The file will then also
6516 be marked offline until the next write. This should increase
6517 reliability in case of a crash. The synchronization delay
6518 can be configured via SyncIntervalSec= in journald.conf.
6520 * There's a new remote-fs-setup.target unit that can be used
6521 to pull in specific services when at least one remote file
6522 system is to be mounted.
6524 * There are new targets timers.target and paths.target as
6525 canonical targets to pull user timer and path units in
6526 from. This complements sockets.target with a similar
6527 purpose for socket units.
6529 * libudev gained a new call udev_device_set_attribute_value()
6530 to set sysfs attributes of a device.
6532 * The udev daemon now sets the default number of worker
6533 processes executed in parallel based on the number of available
6534 CPUs instead of the amount of available RAM. This is supposed
6535 to provide a more reliable default and limit a too aggressive
6536 parallelism for setups with 1000s of devices connected.
6538 Contributions from: Auke Kok, Colin Walters, Cristian
6539 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6540 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6541 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6542 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6543 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6544 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6545 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6546 Zbigniew Jędrzejewski-Szmek
6550 * Configuration of unit files may now be extended via drop-in
6551 files without having to edit/override the unit files
6552 themselves. More specifically, if the administrator wants to
6553 change one value for a service file foobar.service he can
6554 now do so by dropping in a configuration snippet into
6555 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6556 will load all these snippets and apply them on top of the
6557 main unit configuration file, possibly extending or
6558 overriding its settings. Using these drop-in snippets is
6559 generally nicer than the two earlier options for changing
6560 unit files locally: copying the files from
6561 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6562 them there; or creating a new file in /etc/systemd/system/
6563 that incorporates the original one via ".include". Drop-in
6564 snippets into these .d/ directories can be placed in any
6565 directory systemd looks for units in, and the usual
6566 overriding semantics between /usr/lib, /etc and /run apply
6569 * Most unit file settings which take lists of items can now be
6570 reset by assigning the empty string to them. For example,
6571 normally, settings such as Environment=FOO=BAR append a new
6572 environment variable assignment to the environment block,
6573 each time they are used. By assigning Environment= the empty
6574 string the environment block can be reset to empty. This is
6575 particularly useful with the .d/*.conf drop-in snippets
6576 mentioned above, since this adds the ability to reset list
6577 settings from vendor unit files via these drop-ins.
6579 * systemctl gained a new "list-dependencies" command for
6580 listing the dependencies of a unit recursively.
6582 * Inhibitors are now honored and listed by "systemctl
6583 suspend", "systemctl poweroff" (and similar) too, not only
6584 GNOME. These commands will also list active sessions by
6587 * Resource limits (as exposed by the various control group
6588 controllers) can now be controlled dynamically at runtime
6589 for all units. More specifically, you can now use a command
6590 like "systemctl set-cgroup-attr foobar.service cpu.shares
6591 2000" to alter the CPU shares a specific service gets. These
6592 settings are stored persistently on disk, and thus allow the
6593 administrator to easily adjust the resource usage of
6594 services with a few simple commands. This dynamic resource
6595 management logic is also available to other programs via the
6596 bus. Almost any kernel cgroup attribute and controller is
6599 * systemd-vconsole-setup will now copy all font settings to
6600 all allocated VTs, where it previously applied them only to
6603 * libsystemd-login gained the new sd_session_get_tty() API
6606 * This release drops support for a few legacy or
6607 distribution-specific LSB facility names when parsing init
6608 scripts: $x-display-manager, $mail-transfer-agent,
6609 $mail-transport-agent, $mail-transfer-agent, $smtp,
6610 $null. Also, the mail-transfer-agent.target unit backing
6611 this has been removed. Distributions which want to retain
6612 compatibility with this should carry the burden for
6613 supporting this themselves and patch support for these back
6614 in, if they really need to. Also, the facilities $syslog and
6615 $local_fs are now ignored, since systemd does not support
6616 early-boot LSB init scripts anymore, and these facilities
6617 are implied anyway for normal services. syslog.target has
6620 * There are new bus calls on PID1's Manager object for
6621 cancelling jobs, and removing snapshot units. Previously,
6622 both calls were only available on the Job and Snapshot
6625 * systemd-journal-gatewayd gained SSL support.
6627 * The various "environment" files, such as /etc/locale.conf
6628 now support continuation lines with a backslash ("\") as
6629 last character in the line, similarly in style (but different)
6630 to how this is supported in shells.
6632 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6633 now implicitly appended to every log entry logged. systemctl
6634 has been updated to filter by this field when operating on a
6635 user systemd instance.
6637 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6638 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6639 the container. This makes it easier to boot unmodified
6640 Fedora systems in a container, which however still requires
6641 audit=0 to be passed on the kernel command line. Auditing in
6642 kernel and userspace is unfortunately still too broken in
6643 context of containers, hence we recommend compiling it out
6644 of the kernel or using audit=0. Hopefully this will be fixed
6645 one day for good in the kernel.
6647 * nspawn gained the new --bind= and --bind-ro= parameters to
6648 bind mount specific directories from the host into the
6651 * nspawn will now mount its own devpts file system instance
6652 into the container, in order not to leak pty devices from
6653 the host into the container.
6655 * systemd will now read the firmware boot time performance
6656 information from the EFI variables, if the used boot loader
6657 supports this, and takes it into account for boot performance
6658 analysis via "systemd-analyze". This is currently supported
6659 only in conjunction with Gummiboot, but could be supported
6660 by other boot loaders too. For details see:
6662 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6664 * A new generator has been added that automatically mounts the
6665 EFI System Partition (ESP) to /boot, if that directory
6666 exists, is empty, and no other file system has been
6667 configured to be mounted there.
6669 * logind will now send out PrepareForSleep(false) out
6670 unconditionally, after coming back from suspend. This may be
6671 used by applications as asynchronous notification for
6672 system resume events.
6674 * "systemctl unlock-sessions" has been added, that allows
6675 unlocking the screens of all user sessions at once, similar
6676 to how "systemctl lock-sessions" already locked all users
6677 sessions. This is backed by a new D-Bus call UnlockSessions().
6679 * "loginctl seat-status" will now show the master device of a
6680 seat. (i.e. the device of a seat that needs to be around for
6681 the seat to be considered available, usually the graphics
6684 * tmpfiles gained a new "X" line type, that allows
6685 configuration of files and directories (with wildcards) that
6686 shall be excluded from automatic cleanup ("aging").
6688 * udev default rules set the device node permissions now only
6689 at "add" events, and do not change them any longer with a
6690 later "change" event.
6692 * The log messages for lid events and power/sleep keypresses
6693 now carry a message ID.
6695 * We now have a substantially larger unit test suite, but this
6696 continues to be work in progress.
6698 * udevadm hwdb gained a new --root= parameter to change the
6699 root directory to operate relative to.
6701 * logind will now issue a background sync() request to the kernel
6702 early at shutdown, so that dirty buffers are flushed to disk early
6703 instead of at the last moment, in order to optimize shutdown
6706 * A new bootctl tool has been added that is an interface for
6707 certain boot loader operations. This is currently a preview
6708 and is likely to be extended into a small mechanism daemon
6709 like timedated, localed, hostnamed, and can be used by
6710 graphical UIs to enumerate available boot options, and
6711 request boot into firmware operations.
6713 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6714 the rest of the package. It also has been updated to work
6715 correctly in initrds.
6717 * Policykit previously has been runtime optional, and is now
6718 also compile time optional via a configure switch.
6720 * systemd-analyze has been reimplemented in C. Also "systemctl
6721 dot" has moved into systemd-analyze.
6723 * "systemctl status" with no further parameters will now print
6724 the status of all active or failed units.
6726 * Operations such as "systemctl start" can now be executed
6727 with a new mode "--irreversible" which may be used to queue
6728 operations that cannot accidentally be reversed by a later
6729 job queuing. This is by default used to make shutdown
6730 requests more robust.
6732 * The Python API of systemd now gained a new module for
6733 reading journal files.
6735 * A new tool kernel-install has been added that can install
6736 kernel images according to the Boot Loader Specification:
6738 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6740 * Boot time console output has been improved to provide
6741 animated boot time output for hanging jobs.
6743 * A new tool systemd-activate has been added which can be used
6744 to test socket activation with, directly from the command
6745 line. This should make it much easier to test and debug
6746 socket activation in daemons.
6748 * journalctl gained a new "--reverse" (or -r) option to show
6749 journal output in reverse order (i.e. newest line first).
6751 * journalctl gained a new "--pager-end" (or -e) option to jump
6752 to immediately jump to the end of the journal in the
6753 pager. This is only supported in conjunction with "less".
6755 * journalctl gained a new "--user-unit=" option, that works
6756 similarly to "--unit=" but filters for user units rather than
6759 * A number of unit files to ease adoption of systemd in
6760 initrds has been added. This moves some minimal logic from
6761 the various initrd implementations into systemd proper.
6763 * The journal files are now owned by a new group
6764 "systemd-journal", which exists specifically to allow access
6765 to the journal, and nothing else. Previously, we used the
6766 "adm" group for that, which however possibly covers more
6767 than just journal/log file access. This new group is now
6768 already used by systemd-journal-gatewayd to ensure this
6769 daemon gets access to the journal files and as little else
6770 as possible. Note that "make install" will also set FS ACLs
6771 up for /var/log/journal to give "adm" and "wheel" read
6772 access to it, in addition to "systemd-journal" which owns
6773 the journal files. We recommend that packaging scripts also
6774 add read access to "adm" + "wheel" to /var/log/journal, and
6775 all existing/future journal files. To normal users and
6776 administrators little changes, however packagers need to
6777 ensure to create the "systemd-journal" system group at
6778 package installation time.
6780 * The systemd-journal-gatewayd now runs as unprivileged user
6781 systemd-journal-gateway:systemd-journal-gateway. Packaging
6782 scripts need to create these system user/group at
6785 * timedated now exposes a new boolean property CanNTP that
6786 indicates whether a local NTP service is available or not.
6788 * systemd-detect-virt will now also detect xen PVs
6790 * The pstore file system is now mounted by default, if it is
6793 * In addition to the SELinux and IMA policies we will now also
6794 load SMACK policies at early boot.
6796 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6797 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6798 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6799 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6800 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6801 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6802 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6803 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6804 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6805 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6806 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6807 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6808 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6809 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6813 * Timer units now support calendar time events in addition to
6814 monotonic time events. That means you can now trigger a unit
6815 based on a calendar time specification such as "Thu,Fri
6816 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6817 or fifth day of any month of the year 2013, given that it is
6818 a thursday or friday. This brings timer event support
6819 considerably closer to cron's capabilities. For details on
6820 the supported calendar time specification language see
6823 * udev now supports a number of different naming policies for
6824 network interfaces for predictable names, and a combination
6825 of these policies is now the default. Please see this wiki
6826 document for details:
6828 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6830 * Auke Kok's bootchart implementation has been added to the
6831 systemd tree. It is an optional component that can graph the
6832 boot in quite some detail. It is one of the best bootchart
6833 implementations around and minimal in its code and
6836 * nss-myhostname has been integrated into the systemd source
6837 tree. nss-myhostname guarantees that the local hostname
6838 always stays resolvable via NSS. It has been a weak
6839 requirement of systemd-hostnamed since a long time, and
6840 since its code is actually trivial we decided to just
6841 include it in systemd's source tree. It can be turned off
6842 with a configure switch.
6844 * The read-ahead logic is now capable of properly detecting
6845 whether a btrfs file system is on SSD or rotating media, in
6846 order to optimize the read-ahead scheme. Previously, it was
6847 only capable of detecting this on traditional file systems
6850 * In udev, additional device properties are now read from the
6851 IAB in addition to the OUI database. Also, Bluetooth company
6852 identities are attached to the devices as well.
6854 * In service files %U may be used as specifier that is
6855 replaced by the configured user name of the service.
6857 * nspawn may now be invoked without a controlling TTY. This
6858 makes it suitable for invocation as its own service. This
6859 may be used to set up a simple containerized server system
6860 using only core OS tools.
6862 * systemd and nspawn can now accept socket file descriptors
6863 when they are started for socket activation. This enables
6864 implementation of socket activated nspawn
6865 containers. i.e. think about autospawning an entire OS image
6866 when the first SSH or HTTP connection is received. We expect
6867 that similar functionality will also be added to libvirt-lxc
6870 * journalctl will now suppress ANSI color codes when
6871 presenting log data.
6873 * systemctl will no longer show control group information for
6874 a unit if the control group is empty anyway.
6876 * logind can now automatically suspend/hibernate/shutdown the
6879 * /etc/machine-info and hostnamed now also expose the chassis
6880 type of the system. This can be used to determine whether
6881 the local system is a laptop, desktop, handset or
6882 tablet. This information may either be configured by the
6883 user/vendor or is automatically determined from ACPI and DMI
6884 information if possible.
6886 * A number of PolicyKit actions are now bound together with
6887 "imply" rules. This should simplify creating UIs because
6888 many actions will now authenticate similar ones as well.
6890 * Unit files learnt a new condition ConditionACPower= which
6891 may be used to conditionalize a unit depending on whether an
6892 AC power source is connected or not, of whether the system
6893 is running on battery power.
6895 * systemctl gained a new "is-failed" verb that may be used in
6896 shell scripts and suchlike to check whether a specific unit
6897 is in the "failed" state.
6899 * The EnvironmentFile= setting in unit files now supports file
6900 globbing, and can hence be used to easily read a number of
6901 environment files at once.
6903 * systemd will no longer detect and recognize specific
6904 distributions. All distribution-specific #ifdeffery has been
6905 removed, systemd is now fully generic and
6906 distribution-agnostic. Effectively, not too much is lost as
6907 a lot of the code is still accessible via explicit configure
6908 switches. However, support for some distribution specific
6909 legacy configuration file formats has been dropped. We
6910 recommend distributions to simply adopt the configuration
6911 files everybody else uses now and convert the old
6912 configuration from packaging scripts. Most distributions
6913 already did that. If that's not possible or desirable,
6914 distributions are welcome to forward port the specific
6915 pieces of code locally from the git history.
6917 * When logging a message about a unit systemd will now always
6918 log the unit name in the message meta data.
6920 * localectl will now also discover system locale data that is
6921 not stored in locale archives, but directly unpacked.
6923 * logind will no longer unconditionally use framebuffer
6924 devices as seat masters, i.e. as devices that are required
6925 to be existing before a seat is considered preset. Instead,
6926 it will now look for all devices that are tagged as
6927 "seat-master" in udev. By default, framebuffer devices will
6928 be marked as such, but depending on local systems, other
6929 devices might be marked as well. This may be used to
6930 integrate graphics cards using closed source drivers (such
6931 as NVidia ones) more nicely into logind. Note however, that
6932 we recommend using the open source NVidia drivers instead,
6933 and no udev rules for the closed-source drivers will be
6934 shipped from us upstream.
6936 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6937 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6938 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6939 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6940 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6941 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6942 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6943 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6944 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6945 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6946 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6947 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6952 * udev gained support for loading additional device properties
6953 from an indexed database that is keyed by vendor/product IDs
6954 and similar device identifiers. For the beginning this
6955 "hwdb" is populated with data from the well-known PCI and
6956 USB database, but also includes PNP, ACPI and OID data. In
6957 the longer run this indexed database shall grow into
6958 becoming the one central database for non-essential
6959 userspace device metadata. Previously, data from the PCI/USB
6960 database was only attached to select devices, since the
6961 lookup was a relatively expensive operation due to O(n) time
6962 complexity (with n being the number of entries in the
6963 database). Since this is now O(1), we decided to add in this
6964 data for all devices where this is available, by
6965 default. Note that the indexed database needs to be rebuilt
6966 when new data files are installed. To achieve this you need
6967 to update your packaging scripts to invoke "udevadm hwdb
6968 --update" after installation of hwdb data files. For
6969 RPM-based distributions we introduced the new
6970 %udev_hwdb_update macro for this purpose.
6972 * The Journal gained support for the "Message Catalog", an
6973 indexed database to link up additional information with
6974 journal entries. For further details please check:
6976 https://www.freedesktop.org/wiki/Software/systemd/catalog
6978 The indexed message catalog database also needs to be
6979 rebuilt after installation of message catalog files. Use
6980 "journalctl --update-catalog" for this. For RPM-based
6981 distributions we introduced the %journal_catalog_update
6982 macro for this purpose.
6984 * The Python Journal bindings gained support for the standard
6985 Python logging framework.
6987 * The Journal API gained new functions for checking whether
6988 the underlying file system of a journal file is capable of
6989 properly reporting file change notifications, or whether
6990 applications that want to reflect journal changes "live"
6991 need to recheck journal files continuously in appropriate
6994 * It is now possible to set the "age" field for tmpfiles
6995 entries to 0, indicating that files matching this entry
6996 shall always be removed when the directories are cleaned up.
6998 * coredumpctl gained a new "gdb" verb which invokes gdb
6999 right-away on the selected coredump.
7001 * There's now support for "hybrid sleep" on kernels that
7002 support this, in addition to "suspend" and "hibernate". Use
7003 "systemctl hybrid-sleep" to make use of this.
7005 * logind's HandleSuspendKey= setting (and related settings)
7006 now gained support for a new "lock" setting to simply
7007 request the screen lock on all local sessions, instead of
7008 actually executing a suspend or hibernation.
7010 * systemd will now mount the EFI variables file system by
7013 * Socket units now gained support for configuration of the
7014 SMACK security label.
7016 * timedatectl will now output the time of the last and next
7017 daylight saving change.
7019 * We dropped support for various legacy and distro-specific
7020 concepts, such as insserv, early-boot SysV services
7021 (i.e. those for non-standard runlevels such as 'b' or 'S')
7022 or ArchLinux /etc/rc.conf support. We recommend the
7023 distributions who still need support this to either continue
7024 to maintain the necessary patches downstream, or find a
7025 different solution. (Talk to us if you have questions!)
7027 * Various systemd components will now bypass PolicyKit checks
7028 for root and otherwise handle properly if PolicyKit is not
7029 found to be around. This should fix most issues for
7030 PolicyKit-less systems. Quite frankly this should have been
7031 this way since day one. It is absolutely our intention to
7032 make systemd work fine on PolicyKit-less systems, and we
7033 consider it a bug if something does not work as it should if
7034 PolicyKit is not around.
7036 * For embedded systems it is now possible to build udev and
7037 systemd without blkid and/or kmod support.
7039 * "systemctl switch-root" is now capable of switching root
7040 more than once. I.e. in addition to transitions from the
7041 initrd to the host OS it is now possible to transition to
7042 further OS images from the host. This is useful to implement
7043 offline updating tools.
7045 * Various other additions have been made to the RPM macros
7046 shipped with systemd. Use %udev_rules_update() after
7047 installing new udev rules files. %_udevhwdbdir,
7048 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7049 %_sysctldir are now available which resolve to the right
7050 directories for packages to place various data files in.
7052 * journalctl gained the new --full switch (in addition to
7053 --all, to disable ellipsation for long messages.
7055 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7056 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7057 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7058 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7059 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7060 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7061 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7062 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7063 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7067 * journalctl gained new --since= and --until= switches to
7068 filter by time. It also now supports nice filtering for
7069 units via --unit=/-u.
7071 * Type=oneshot services may use ExecReload= and do the
7074 * The journal daemon now supports time-based rotation and
7075 vacuuming, in addition to the usual disk-space based
7078 * The journal will now index the available field values for
7079 each field name. This enables clients to show pretty drop
7080 downs of available match values when filtering. The bash
7081 completion of journalctl has been updated
7082 accordingly. journalctl gained a new switch -F to list all
7083 values a certain field takes in the journal database.
7085 * More service events are now written as structured messages
7086 to the journal, and made recognizable via message IDs.
7088 * The timedated, localed and hostnamed mini-services which
7089 previously only provided support for changing time, locale
7090 and hostname settings from graphical DEs such as GNOME now
7091 also have a minimal (but very useful) text-based client
7092 utility each. This is probably the nicest way to changing
7093 these settings from the command line now, especially since
7094 it lists available options and is fully integrated with bash
7097 * There's now a new tool "systemd-coredumpctl" to list and
7098 extract coredumps from the journal.
7100 * We now install a README each in /var/log/ and
7101 /etc/rc.d/init.d explaining where the system logs and init
7102 scripts went. This hopefully should help folks who go to
7103 that dirs and look into the otherwise now empty void and
7104 scratch their heads.
7106 * When user-services are invoked (by systemd --user) the
7107 $MANAGERPID env var is set to the PID of systemd.
7109 * SIGRTMIN+24 when sent to a --user instance will now result
7110 in immediate termination of systemd.
7112 * gatewayd received numerous feature additions such as a
7113 "follow" mode, for live syncing and filtering.
7115 * browse.html now allows filtering and showing detailed
7116 information on specific entries. Keyboard navigation and
7117 mouse screen support has been added.
7119 * gatewayd/journalctl now supports HTML5/JSON
7120 Server-Sent-Events as output.
7122 * The SysV init script compatibility logic will now
7123 heuristically determine whether a script supports the
7124 "reload" verb, and only then make this available as
7127 * "systemctl status --follow" has been removed, use "journalctl
7130 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7131 have been removed since they are hardly useful to be
7134 * And I'd like to take the opportunity to specifically mention
7135 Zbigniew for his great contributions. Zbigniew, you rock!
7137 Contributions from: Andrew Eikum, Christian Hesse, Colin
7138 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7139 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7140 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7141 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7142 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7143 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7147 * If /etc/vconsole.conf is non-existent or empty we will no
7148 longer load any console font or key map at boot by
7149 default. Instead the kernel defaults will be left
7150 intact. This is definitely the right thing to do, as no
7151 configuration should mean no configuration, and hard-coding
7152 font names that are different on all archs is probably a bad
7153 idea. Also, the kernel default key map and font should be
7154 good enough for most cases anyway, and mostly identical to
7155 the userspace fonts/key maps we previously overloaded them
7156 with. If distributions want to continue to default to a
7157 non-kernel font or key map they should ship a default
7158 /etc/vconsole.conf with the appropriate contents.
7160 Contributions from: Colin Walters, Daniel J Walsh, Dave
7161 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7162 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7166 * journalctl gained a new --cursor= switch to show entries
7167 starting from the specified location in the journal.
7169 * We now enforce a size limit on journal entry fields exported
7170 with "-o json" in journalctl. Fields larger than 4K will be
7171 assigned null. This can be turned off with --all.
7173 * An (optional) journal gateway daemon is now available as
7174 "systemd-journal-gatewayd.service". This service provides
7175 access to the journal via HTTP and JSON. This functionality
7176 will be used to implement live log synchronization in both
7177 pull and push modes, but has various other users too, such
7178 as easy log access for debugging of embedded devices. Right
7179 now it is already useful to retrieve the journal via HTTP:
7181 # systemctl start systemd-journal-gatewayd.service
7182 # wget http://localhost:19531/entries
7184 This will download the journal contents in a
7185 /var/log/messages compatible format. The same as JSON:
7187 # curl -H"Accept: application/json" http://localhost:19531/entries
7189 This service is also accessible via a web browser where a
7190 single static HTML5 app is served that uses the JSON logic
7191 to enable the user to do some basic browsing of the
7192 journal. This will be extended later on. Here's an example
7193 screenshot of this app in its current state:
7195 http://0pointer.de/public/journal-gatewayd
7197 Contributions from: Kay Sievers, Lennart Poettering, Robert
7198 Milasan, Tom Gundersen
7202 * The bash completion logic is now available for journalctl
7205 * We do not mount the "cpuset" controller anymore together with
7206 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7207 started if no parameters are assigned to it. "cpuset" hence
7208 broke code that assumed it could create "cpu" groups and
7211 * journalctl -f will now subscribe to terminal size changes,
7212 and line break accordingly.
7214 Contributions from: Dave Reisner, Kay Sievers, Lennart
7215 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7219 * nspawn will now create a symlink /etc/localtime in the
7220 container environment, copying the host's timezone
7221 setting. Previously this has been done via a bind mount, but
7222 since symlinks cannot be bind mounted this has now been
7223 changed to create/update the appropriate symlink.
7225 * journalctl -n's line number argument is now optional, and
7226 will default to 10 if omitted.
7228 * journald will now log the maximum size the journal files may
7229 take up on disk. This is particularly useful if the default
7230 built-in logic of determining this parameter from the file
7231 system size is used. Use "systemctl status
7232 systemd-journald.service" to see this information.
7234 * The multi-seat X wrapper tool has been stripped down. As X
7235 is now capable of enumerating graphics devices via udev in a
7236 seat-aware way the wrapper is not strictly necessary
7237 anymore. A stripped down temporary stop-gap is still shipped
7238 until the upstream display managers have been updated to
7239 fully support the new X logic. Expect this wrapper to be
7240 removed entirely in one of the next releases.
7242 * HandleSleepKey= in logind.conf has been split up into
7243 HandleSuspendKey= and HandleHibernateKey=. The old setting
7244 is not available anymore. X11 and the kernel are
7245 distinguishing between these keys and we should too. This
7246 also means the inhibition lock for these keys has been split
7249 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7250 Poettering, Lukas Nykryn, Václav Pavlín
7254 * Whenever a unit changes state we will now log this to the
7255 journal and show along the unit's own log output in
7258 * ConditionPathIsMountPoint= can now properly detect bind
7259 mount points too. (Previously, a bind mount of one file
7260 system to another place in the same file system could not be
7261 detected as mount, since they shared struct stat's st_dev
7264 * We will now mount the cgroup controllers cpu, cpuacct,
7265 cpuset and the controllers net_cls, net_prio together by
7268 * nspawn containers will now have a virtualized boot
7269 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7270 over with a randomized ID at container initialization). This
7271 has the effect of making "journalctl -b" do the right thing
7274 * The JSON output journal serialization has been updated not
7275 to generate "endless" list objects anymore, but rather one
7276 JSON object per line. This is more in line how most JSON
7277 parsers expect JSON objects. The new output mode
7278 "json-pretty" has been added to provide similar output, but
7279 neatly aligned for readability by humans.
7281 * We dropped all explicit sync() invocations in the shutdown
7282 code. The kernel does this implicitly anyway in the kernel
7283 reboot() syscall. halt(8)'s -n option is now a compatibility
7286 * We now support virtualized reboot() in containers, as
7287 supported by newer kernels. We will fall back to exit() if
7288 CAP_SYS_REBOOT is not available to the container. Also,
7289 nspawn makes use of this now and will actually reboot the
7290 container if the containerized OS asks for that.
7292 * journalctl will only show local log output by default
7293 now. Use --merge (-m) to show remote log output, too.
7295 * libsystemd-journal gained the new sd_journal_get_usage()
7296 call to determine the current disk usage of all journal
7297 files. This is exposed in the new "journalctl --disk-usage"
7300 * journald gained a new configuration setting SplitMode= in
7301 journald.conf which may be used to control how user journals
7302 are split off. See journald.conf(5) for details.
7304 * A new condition type ConditionFileNotEmpty= has been added.
7306 * tmpfiles' "w" lines now support file globbing, to write
7307 multiple files at once.
7309 * We added Python bindings for the journal submission
7310 APIs. More Python APIs for a number of selected APIs will
7311 likely follow. Note that we intend to add native bindings
7312 only for the Python language, as we consider it common
7313 enough to deserve bindings shipped within systemd. There are
7314 various projects outside of systemd that provide bindings
7315 for languages such as PHP or Lua.
7317 * Many conditions will now resolve specifiers such as %i. In
7318 addition, PathChanged= and related directives of .path units
7319 now support specifiers as well.
7321 * There's now a new RPM macro definition for the system preset
7324 * journald will now warn if it ca not forward a message to the
7325 syslog daemon because its socket is full.
7327 * timedated will no longer write or process /etc/timezone,
7328 except on Debian. As we do not support late mounted /usr
7329 anymore /etc/localtime always being a symlink is now safe,
7330 and hence the information in /etc/timezone is not necessary
7333 * logind will now always reserve one VT for a text getty (VT6
7334 by default). Previously if more than 6 X sessions where
7335 started they took up all the VTs with auto-spawned gettys,
7336 so that no text gettys were available anymore.
7338 * udev will now automatically inform the btrfs kernel logic
7339 about btrfs RAID components showing up. This should make
7340 simple hotplug based btrfs RAID assembly work.
7342 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7343 (but not for its children which will stay at the kernel
7344 default). This should allow setups with a lot more listening
7347 * systemd will now always pass the configured timezone to the
7348 kernel at boot. timedated will do the same when the timezone
7351 * logind's inhibition logic has been updated. By default,
7352 logind will now handle the lid switch, the power and sleep
7353 keys all the time, even in graphical sessions. If DEs want
7354 to handle these events on their own they should take the new
7355 handle-power-key, handle-sleep-key and handle-lid-switch
7356 inhibitors during their runtime. A simple way to achieve
7357 that is to invoke the DE wrapped in an invocation of:
7359 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7361 * Access to unit operations is now checked via SELinux taking
7362 the unit file label and client process label into account.
7364 * systemd will now notify the administrator in the journal
7365 when he over-mounts a non-empty directory.
7367 * There are new specifiers that are resolved in unit files,
7368 for the host name (%H), the machine ID (%m) and the boot ID
7371 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7372 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7373 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7374 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7375 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7376 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7377 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7381 * Support for reading structured kernel messages from
7382 /dev/kmsg has now been added and is enabled by default.
7384 * Support for reading kernel messages from /proc/kmsg has now
7385 been removed. If you want kernel messages in the journal
7386 make sure to run a recent kernel (>= 3.5) that supports
7387 reading structured messages from /dev/kmsg (see
7388 above). /proc/kmsg is now exclusive property of classic
7389 syslog daemons again.
7391 * The libudev API gained the new
7392 udev_device_new_from_device_id() call.
7394 * The logic for file system namespace (ReadOnlyDirectory=,
7395 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7396 require pivot_root() anymore. This means fewer temporary
7397 directories are created below /tmp for this feature.
7399 * nspawn containers will now see and receive all submounts
7400 made on the host OS below the root file system of the
7403 * Forward Secure Sealing is now supported for Journal files,
7404 which provide cryptographical sealing of journal files so
7405 that attackers cannot alter log history anymore without this
7406 being detectable. Lennart will soon post a blog story about
7407 this explaining it in more detail.
7409 * There are two new service settings RestartPreventExitStatus=
7410 and SuccessExitStatus= which allow configuration of exit
7411 status (exit code or signal) which will be excepted from the
7412 restart logic, resp. consider successful.
7414 * journalctl gained the new --verify switch that can be used
7415 to check the integrity of the structure of journal files and
7416 (if Forward Secure Sealing is enabled) the contents of
7419 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7420 and similar symlinks pre-created. This makes running shells
7421 as container init process a lot more fun.
7423 * The fstab support can now handle PARTUUID= and PARTLABEL=
7426 * A new ConditionHost= condition has been added to match
7427 against the hostname (with globs) and machine ID. This is
7428 useful for clusters where a single OS image is used to
7429 provision a large number of hosts which shall run slightly
7430 different sets of services.
7432 * Services which hit the restart limit will now be placed in a
7435 Contributions from: Bertram Poettering, Dave Reisner, Huang
7436 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7437 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7441 * When running in --user mode systemd will now become a
7442 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7443 tree a lot more organized.
7445 * A new PartOf= unit dependency type has been introduced that
7446 may be used to group services in a natural way.
7448 * "systemctl enable" may now be used to enable instances of
7451 * journalctl now prints error log levels in red, and
7452 warning/notice log levels in bright white. It also supports
7453 filtering by log level now.
7455 * cgtop gained a new -n switch (similar to top), to configure
7456 the maximum number of iterations to run for. It also gained
7457 -b, to run in batch mode (accepting no input).
7459 * The suffix ".service" may now be omitted on most systemctl
7460 command lines involving service unit names.
7462 * There's a new bus call in logind to lock all sessions, as
7463 well as a loginctl verb for it "lock-sessions".
7465 * libsystemd-logind.so gained a new call sd_journal_perror()
7466 that works similar to libc perror() but logs to the journal
7467 and encodes structured information about the error number.
7469 * /etc/crypttab entries now understand the new keyfile-size=
7472 * shutdown(8) now can send a (configurable) wall message when
7473 a shutdown is cancelled.
7475 * The mount propagation mode for the root file system will now
7476 default to "shared", which is useful to make containers work
7477 nicely out-of-the-box so that they receive new mounts from
7478 the host. This can be undone locally by running "mount
7479 --make-rprivate /" if needed.
7481 * The prefdm.service file has been removed. Distributions
7482 should maintain this unit downstream if they intend to keep
7483 it around. However, we recommend writing normal unit files
7484 for display managers instead.
7486 * Since systemd is a crucial part of the OS we will now
7487 default to a number of compiler switches that improve
7488 security (hardening) such as read-only relocations, stack
7489 protection, and suchlike.
7491 * The TimeoutSec= setting for services is now split into
7492 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7493 of individual time outs for the start and the stop phase of
7496 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7497 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7498 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7499 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7500 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7501 Gundersen, Zbigniew Jędrzejewski-Szmek
7505 * The journal and id128 C APIs are now fully documented as man
7508 * Extra safety checks have been added when transitioning from
7509 the initial RAM disk to the main system to avoid accidental
7512 * /etc/crypttab entries now understand the new keyfile-offset=
7515 * systemctl -t can now be used to filter by unit load state.
7517 * The journal C API gained the new sd_journal_wait() call to
7518 make writing synchronous journal clients easier.
7520 * journalctl gained the new -D switch to show journals from a
7523 * journalctl now displays a special marker between log
7524 messages of two different boots.
7526 * The journal is now explicitly flushed to /var via a service
7527 systemd-journal-flush.service, rather than implicitly simply
7528 by seeing /var/log/journal to be writable.
7530 * journalctl (and the journal C APIs) can now match for much
7531 more complex expressions, with alternatives and
7534 * When transitioning from the initial RAM disk to the main
7535 system we will now kill all processes in a killing spree to
7536 ensure no processes stay around by accident.
7538 * Three new specifiers may be used in unit files: %u, %h, %s
7539 resolve to the user name, user home directory resp. user
7540 shell. This is useful for running systemd user instances.
7542 * We now automatically rotate journal files if their data
7543 object hash table gets a fill level > 75%. We also size the
7544 hash table based on the configured maximum file size. This
7545 together should lower hash collisions drastically and thus
7546 speed things up a bit.
7548 * journalctl gained the new "--header" switch to introspect
7549 header data of journal files.
7551 * A new setting SystemCallFilters= has been added to services
7552 which may be used to apply blacklists or whitelists to
7553 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7555 * nspawn gained a new --link-journal= switch (and quicker: -j)
7556 to link the container journal with the host. This makes it
7557 very easy to centralize log viewing on the host for all
7558 guests while still keeping the journal files separated.
7560 * Many bugfixes and optimizations
7562 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7563 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7564 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7569 * Several tools now understand kernel command line arguments,
7570 which are only read when run in an initial RAM disk. They
7571 usually follow closely their normal counterparts, but are
7574 * There's a new tool to analyze the readahead files that are
7575 automatically generated at boot. Use:
7577 /usr/lib/systemd/systemd-readahead analyze /.readahead
7579 * We now provide an early debug shell on tty9 if this enabled. Use:
7581 systemctl enable debug-shell.service
7583 * All plymouth related units have been moved into the Plymouth
7584 package. Please make sure to upgrade your Plymouth version
7587 * systemd-tmpfiles now supports getting passed the basename of
7588 a configuration file only, in which case it will look for it
7589 in all appropriate directories automatically.
7591 * udevadm info now takes a /dev or /sys path as argument, and
7592 does the right thing. Example:
7594 udevadm info /dev/sda
7595 udevadm info /sys/class/block/sda
7597 * systemctl now prints a warning if a unit is stopped but a
7598 unit that might trigger it continues to run. Example: a
7599 service is stopped but the socket that activates it is left
7602 * "systemctl status" will now mention if the log output was
7603 shortened due to rotation since a service has been started.
7605 * The journal API now exposes functions to determine the
7606 "cutoff" times due to rotation.
7608 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7609 immediately flushing of runtime logs to /var if possible,
7610 resp. for triggering immediate rotation of the journal
7613 * It is now considered an error if a service is attempted to
7614 be stopped that is not loaded.
7616 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7618 * systemd-analyze now supports Python 3
7620 * tmpfiles now supports cleaning up directories via aging
7621 where the first level dirs are always kept around but
7622 directories beneath it automatically aged. This is enabled
7623 by prefixing the age field with '~'.
7625 * Seat objects now expose CanGraphical, CanTTY properties
7626 which is required to deal with very fast bootups where the
7627 display manager might be running before the graphics drivers
7628 completed initialization.
7630 * Seat objects now expose a State property.
7632 * We now include RPM macros for service enabling/disabling
7633 based on the preset logic. We recommend RPM based
7634 distributions to make use of these macros if possible. This
7635 makes it simpler to reuse RPM spec files across
7638 * We now make sure that the collected systemd unit name is
7639 always valid when services log to the journal via
7642 * There's a new man page kernel-command-line(7) detailing all
7643 command line options we understand.
7645 * The fstab generator may now be disabled at boot by passing
7646 fstab=0 on the kernel command line.
7648 * A new kernel command line option modules-load= is now understood
7649 to load a specific kernel module statically, early at boot.
7651 * Unit names specified on the systemctl command line are now
7652 automatically escaped as needed. Also, if file system or
7653 device paths are specified they are automatically turned
7654 into the appropriate mount or device unit names. Example:
7656 systemctl status /home
7657 systemctl status /dev/sda
7659 * The SysVConsole= configuration option has been removed from
7660 system.conf parsing.
7662 * The SysV search path is no longer exported on the D-Bus
7665 * The Names= option has been removed from unit file parsing.
7667 * There's a new man page bootup(7) detailing the boot process.
7669 * Every unit and every generator we ship with systemd now
7670 comes with full documentation. The self-explanatory boot is
7673 * A couple of services gained "systemd-" prefixes in their
7674 name if they wrap systemd code, rather than only external
7675 code. Among them fsck@.service which is now
7676 systemd-fsck@.service.
7678 * The HaveWatchdog property has been removed from the D-Bus
7681 * systemd.confirm_spawn= on the kernel command line should now
7684 * There's a new man page crypttab(5) which details all options
7685 we actually understand.
7687 * systemd-nspawn gained a new --capability= switch to pass
7688 additional capabilities to the container.
7690 * timedated will now read known NTP implementation unit names
7691 from /usr/lib/systemd/ntp-units.d/*.list,
7692 systemd-timedated-ntp.target has been removed.
7694 * journalctl gained a new switch "-b" that lists log data of
7695 the current boot only.
7697 * The notify socket is in the abstract namespace again, in
7698 order to support daemons which chroot() at start-up.
7700 * There is a new Storage= configuration option for journald
7701 which allows configuration of where log data should go. This
7702 also provides a way to disable journal logging entirely, so
7703 that data collected is only forwarded to the console, the
7704 kernel log buffer or another syslog implementation.
7706 * Many bugfixes and optimizations
7708 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7709 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7710 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7711 Shawn Landden, Tom Gundersen
7715 * "systemctl help <unit>" now shows the man page if one is
7718 * Several new man pages have been added.
7720 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7721 MaxLevelConsole= can now be specified in
7722 journald.conf. These options allow reducing the amount of
7723 data stored on disk or forwarded by the log level.
7725 * TimerSlackNSec= can now be specified in system.conf for
7726 PID1. This allows system-wide power savings.
7728 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7729 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7734 * logind is now capable of (optionally) handling power and
7735 sleep keys as well as the lid switch.
7737 * journalctl now understands the syntax "journalctl
7738 /usr/bin/avahi-daemon" to get all log output of a specific
7741 * CapabilityBoundingSet= in system.conf now also influences
7742 the capability bound set of usermode helpers of the kernel.
7744 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7745 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7746 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7747 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7751 * Note that we skipped 139 releases here in order to set the
7752 new version to something that is greater than both udev's
7753 and systemd's most recent version number.
7755 * udev: all udev sources are merged into the systemd source tree now.
7756 All future udev development will happen in the systemd tree. It
7757 is still fully supported to use the udev daemon and tools without
7758 systemd running, like in initramfs or other init systems. Building
7759 udev though, will require the *build* of the systemd tree, but
7760 udev can be properly *run* without systemd.
7762 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7763 should be used to create dead device nodes as workarounds for broken
7766 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7767 no longer supported. udev_monitor_new_from_netlink() needs to be
7768 used to subscribe to events.
7770 * udev: when udevd is started by systemd, processes which are left
7771 behind by forking them off of udev rules, are unconditionally cleaned
7772 up and killed now after the event handling has finished. Services or
7773 daemons must be started as systemd services. Services can be
7774 pulled-in by udev to get started, but they can no longer be directly
7775 forked by udev rules.
7777 * udev: the daemon binary is called systemd-udevd now and installed
7778 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7779 to adapt to that, create symlink, or rename the binary after building
7782 * libudev no longer provides these symbols:
7783 udev_monitor_from_socket()
7784 udev_queue_get_failed_list_entry()
7785 udev_get_{dev,sys,run}_path()
7786 The versions number was bumped and symbol versioning introduced.
7788 * systemd-loginctl and systemd-journalctl have been renamed
7789 to loginctl and journalctl to match systemctl.
7791 * The config files: /etc/systemd/systemd-logind.conf and
7792 /etc/systemd/systemd-journald.conf have been renamed to
7793 logind.conf and journald.conf. Package updates should rename
7794 the files to the new names on upgrade.
7796 * For almost all files the license is now LGPL2.1+, changed
7797 from the previous GPL2.0+. Exceptions are some minor stuff
7798 of udev (which will be changed to LGPL2.1 eventually, too),
7799 and the MIT licensed sd-daemon.[ch] library that is suitable
7800 to be used as drop-in files.
7802 * systemd and logind now handle system sleep states, in
7803 particular suspending and hibernating.
7805 * logind now implements a sleep/shutdown/idle inhibiting logic
7806 suitable for a variety of uses. Soonishly Lennart will blog
7807 about this in more detail.
7809 * var-run.mount and var-lock.mount are no longer provided
7810 (which previously bind mounted these directories to their new
7811 places). Distributions which have not converted these
7812 directories to symlinks should consider stealing these files
7813 from git history and add them downstream.
7815 * We introduced the Documentation= field for units and added
7816 this to all our shipped units. This is useful to make it
7817 easier to explore the boot and the purpose of the various
7820 * All smaller setup units (such as
7821 systemd-vconsole-setup.service) now detect properly if they
7822 are run in a container and are skipped when
7823 appropriate. This guarantees an entirely noise-free boot in
7824 Linux container environments such as systemd-nspawn.
7826 * A framework for implementing offline system updates is now
7827 integrated, for details see:
7828 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7830 * A new service type Type=idle is available now which helps us
7831 avoiding ugly interleaving of getty output and boot status
7834 * There's now a system-wide CapabilityBoundingSet= option to
7835 globally reduce the set of capabilities for the
7836 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7837 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7838 even CAP_NET_ADMIN system-wide for secure systems.
7840 * There are now system-wide DefaultLimitXXX= options to
7841 globally change the defaults of the various resource limits
7842 for all units started by PID 1.
7844 * Harald Hoyer's systemd test suite has been integrated into
7845 systemd which allows easy testing of systemd builds in qemu
7846 and nspawn. (This is really awesome! Ask us for details!)
7848 * The fstab parser is now implemented as generator, not inside
7851 * systemctl will now warn you if .mount units generated from
7852 /etc/fstab are out of date due to changes in fstab that
7853 have not been read by systemd yet.
7855 * systemd is now suitable for usage in initrds. Dracut has
7856 already been updated to make use of this. With this in place
7857 initrds get a slight bit faster but primarily are much
7858 easier to introspect and debug since "systemctl status" in
7859 the host system can be used to introspect initrd services,
7860 and the journal from the initrd is kept around too.
7862 * systemd-delta has been added, a tool to explore differences
7863 between user/admin configuration and vendor defaults.
7865 * PrivateTmp= now affects both /tmp and /var/tmp.
7867 * Boot time status messages are now much prettier and feature
7868 proper english language. Booting up systemd has never been
7871 * Read-ahead pack files now include the inode number of all
7872 files to pre-cache. When the inode changes the pre-caching
7873 is not attempted. This should be nicer to deal with updated
7874 packages which might result in changes of read-ahead
7877 * We now temporaritly lower the kernel's read_ahead_kb variable
7878 when collecting read-ahead data to ensure the kernel's
7879 built-in read-ahead does not add noise to our measurements
7880 of necessary blocks to pre-cache.
7882 * There's now RequiresMountsFor= to add automatic dependencies
7883 for all mounts necessary for a specific file system path.
7885 * MountAuto= and SwapAuto= have been removed from
7886 system.conf. Mounting file systems at boot has to take place
7889 * nspawn now learned a new switch --uuid= to set the machine
7890 ID on the command line.
7892 * nspawn now learned the -b switch to automatically search
7895 * vt102 is now the default TERM for serial TTYs, upgraded from
7898 * systemd-logind now works on VT-less systems.
7900 * The build tree has been reorganized. The individual
7901 components now have directories of their own.
7903 * A new condition type ConditionPathIsReadWrite= is now available.
7905 * nspawn learned the new -C switch to create cgroups for the
7906 container in other hierarchies.
7908 * We now have support for hardware watchdogs, configurable in
7911 * The scheduled shutdown logic now has a public API.
7913 * We now mount /tmp as tmpfs by default, but this can be
7914 masked and /etc/fstab can override it.
7916 * Since udisks does not make use of /media anymore we are not
7917 mounting a tmpfs on it anymore.
7919 * journalctl gained a new --local switch to only interleave
7920 locally generated journal files.
7922 * We can now load the IMA policy at boot automatically.
7924 * The GTK tools have been split off into a systemd-ui.
7926 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7927 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7928 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7929 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7930 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7931 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7932 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7933 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7934 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7939 * This is mostly a bugfix release
7941 * Support optional initialization of the machine ID from the
7942 KVM or container configured UUID.
7944 * Support immediate reboots with "systemctl reboot -ff"
7946 * Show /etc/os-release data in systemd-analyze output
7948 * Many bugfixes for the journal, including endianness fixes and
7949 ensuring that disk space enforcement works
7951 * sd-login.h is C++ compatible again
7953 * Extend the /etc/os-release format on request of the Debian
7956 * We now refuse non-UTF8 strings used in various configuration
7957 and unit files. This is done to ensure we do not pass invalid
7958 data over D-Bus or expose it elsewhere.
7960 * Register Mimo USB Screens as suitable for automatic seat
7963 * Read SELinux client context from journal clients in a race
7966 * Reorder configuration file lookup order. /etc now always
7967 overrides /run in order to allow the administrator to always
7968 and unconditionally override vendor-supplied or
7969 automatically generated data.
7971 * The various user visible bits of the journal now have man
7972 pages. We still lack man pages for the journal API calls
7975 * We now ship all man pages in HTML format again in the
7978 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7979 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7980 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7985 * This is mostly a bugfix release
7987 * systems lacking /etc/os-release are no longer supported.
7989 * Various functionality updates to libsystemd-login.so
7991 * Track class of PAM logins to distinguish greeters from
7994 Contributions from: Kay Sievers, Lennart Poettering, Michael
7999 * This is an important bugfix release for v41.
8001 * Building man pages is now optional which should be useful
8002 for those building systemd from git but unwilling to install
8005 * Watchdog support for supervising services is now usable. In
8006 a future release support for hardware watchdogs
8007 (i.e. /dev/watchdog) will be added building on this.
8009 * Service start rate limiting is now configurable and can be
8010 turned off per service. When a start rate limit is hit a
8011 reboot can automatically be triggered.
8013 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8015 Contributions from: Benjamin Franzke, Bill Nottingham,
8016 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8017 Schmidt, Michał Górny, Piotr Drąg
8021 * The systemd binary is installed /usr/lib/systemd/systemd now;
8022 An existing /sbin/init symlink needs to be adapted with the
8025 * The code that loads kernel modules has been ported to invoke
8026 libkmod directly, instead of modprobe. This means we do not
8027 support systems with module-init-tools anymore.
8029 * Watchdog support is now already useful, but still not
8032 * A new kernel command line option systemd.setenv= is
8033 understood to set system wide environment variables
8034 dynamically at boot.
8036 * We now limit the set of capabilities of systemd-journald.
8038 * We now set SIGPIPE to ignore by default, since it only is
8039 useful in shell pipelines, and has little use in general
8040 code. This can be disabled with IgnoreSIPIPE=no in unit
8043 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8044 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8049 * This is mostly a bugfix release
8051 * We now expose the reason why a service failed in the
8052 "Result" D-Bus property.
8054 * Rudimentary service watchdog support (will be completed over
8055 the next few releases.)
8057 * When systemd forks off in order execute some service we will
8058 now immediately changes its argv[0] to reflect which process
8059 it will execute. This is useful to minimize the time window
8060 with a generic argv[0], which makes bootcharts more useful
8062 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8063 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8064 Mike Kazantsev, Ray Strode
8068 * This is mostly a test release, but incorporates many
8071 * New systemd-cgtop tool to show control groups by their
8074 * Linking against libacl for ACLs is optional again. If
8075 disabled, support tracking device access for active logins
8076 goes becomes unavailable, and so does access to the user
8077 journals by the respective users.
8079 * If a group "adm" exists, journal files are automatically
8080 owned by them, thus allow members of this group full access
8081 to the system journal as well as all user journals.
8083 * The journal now stores the SELinux context of the logging
8084 client for all entries.
8086 * Add C++ inclusion guards to all public headers
8088 * New output mode "cat" in the journal to print only text
8089 messages, without any meta data like date or time.
8091 * Include tiny X server wrapper as a temporary stop-gap to
8092 teach XOrg udev display enumeration. This is used by display
8093 managers such as gdm, and will go away as soon as XOrg
8094 learned native udev hotplugging for display devices.
8096 * Add new systemd-cat tool for executing arbitrary programs
8097 with STDERR/STDOUT connected to the journal. Can also act as
8098 BSD logger replacement, and does so by default.
8100 * Optionally store all locally generated coredumps in the
8101 journal along with meta data.
8103 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8104 writing short strings to files (for usage for /sys), and for
8105 creating symlinks, character and block device nodes.
8107 * New unit file option ControlGroupPersistent= to make cgroups
8108 persistent, following the mechanisms outlined in
8109 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8111 * Support multiple local RTCs in a sane way
8113 * No longer monopolize IO when replaying readahead data on
8114 rotating disks, since we might starve non-file-system IO to
8115 death, since fanotify() will not see accesses done by blkid,
8118 * Do not show kernel threads in systemd-cgls anymore, unless
8119 requested with new -k switch.
8121 Contributions from: Dan Horák, Kay Sievers, Lennart
8122 Poettering, Michal Schmidt
8126 * This is mostly a test release, but incorporates many
8129 * The git repository moved to:
8130 git://anongit.freedesktop.org/systemd/systemd
8131 ssh://git.freedesktop.org/git/systemd/systemd
8133 * First release with the journal
8134 http://0pointer.de/blog/projects/the-journal.html
8136 * The journal replaces both systemd-kmsg-syslogd and
8137 systemd-stdout-bridge.
8139 * New sd_pid_get_unit() API call in libsystemd-logind
8141 * Many systemadm clean-ups
8143 * Introduce remote-fs-pre.target which is ordered before all
8144 remote mounts and may be used to start services before all
8147 * Added Mageia support
8149 * Add bash completion for systemd-loginctl
8151 * Actively monitor PID file creation for daemons which exit in
8152 the parent process before having finished writing the PID
8153 file in the daemon process. Daemons which do this need to be
8154 fixed (i.e. PID file creation must have finished before the
8155 parent exits), but we now react a bit more gracefully to them.
8157 * Add colourful boot output, mimicking the well-known output
8158 of existing distributions.
8160 * New option PassCredentials= for socket units, for
8161 compatibility with a recent kernel ABI breakage.
8163 * /etc/rc.local is now hooked in via a generator binary, and
8164 thus will no longer act as synchronization point during
8167 * systemctl list-unit-files now supports --root=.
8169 * systemd-tmpfiles now understands two new commands: z, Z for
8170 relabelling files according to the SELinux database. This is
8171 useful to apply SELinux labels to specific files in /sys,
8174 * Output of SysV services is now forwarded to both the console
8175 and the journal by default, not only just the console.
8177 * New man pages for all APIs from libsystemd-login.
8179 * The build tree got reorganized and the build system is a
8180 lot more modular allowing embedded setups to specifically
8181 select the components of systemd they are interested in.
8183 * Support for Linux systems lacking the kernel VT subsystem is
8186 * configure's --with-rootdir= got renamed to
8187 --with-rootprefix= to follow the naming used by udev and
8190 * Unless specified otherwise we will now install to /usr instead
8191 of /usr/local by default.
8193 * Processes with '@' in argv[0][0] are now excluded from the
8194 final shut-down killing spree, following the logic explained
8196 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8198 * All processes remaining in a service cgroup when we enter
8199 the START or START_PRE states are now killed with
8200 SIGKILL. That means it is no longer possible to spawn
8201 background processes from ExecStart= lines (which was never
8202 supported anyway, and bad style).
8204 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8205 reloading of units together.
8207 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8208 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8209 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8210 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8211 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob