1 systemd System and Service Manager
3 CHANGES WITH 239 in spe:
5 * NETWORK INTERFACE DEVICE NAMING CHANGES: elogind-udevd's "net_id"
6 builtin may name network interfaces differently than in previous
7 versions. SR-IOV virtual functions and NPAR partitions with PCI
8 function numbers of 8 and above will be named more predictably,
9 and udev may generate names based on PCI slot number in some cases
10 where it previously did not.
12 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
13 elogind.service. Since v235, IPAddressDeny=any has been set to
14 the unit. So, it is expected that the default behavior of elogind
15 is not changed. However, if distribution packagers or administrators
16 disabled or modified IPAddressDeny= setting by a drop-in config file,
17 then it may be necessary to update the file to re-enable AF_INET and
18 AF_INET6 to support network user name services, e.g. NIS.
20 * When the RestrictNamespaces= unit property is specified multiple times,
21 then the specified types are merged now. Previously, only the last
22 assignment was used. So, if distribution packagers or administrators
23 modified the setting by a drop-in config file, then it may be necessary
28 * The MemoryAccounting= unit property now defaults to on. After
29 discussions with the upstream control group maintainers we learnt
30 that the negative impact of cgroup memory accounting on current
31 kernels is finally relatively minimal, so that it should be safe to
32 enable this by default without affecting system performance. Besides
33 memory accounting only task accounting is turned on by default, all
34 other forms of resource accounting (CPU, IO, IP) remain off for now,
35 because it's not clear yet that their impact is small enough to move
36 from opt-in to opt-out. We recommend downstreams to leave memory
37 accounting on by default if kernel 4.14 or higher is primarily
38 used. On very resource constrained systems or when support for old
39 kernels is a necessity, -Dmemory-accounting-default=false can be used
40 to revert this change.
42 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
43 %udev_rules_update) and the journal catalog (%journal_catalog_update)
44 from the upgrade scriptlets of individual packages now do nothing.
45 Transfiletriggers have been added which will perform those updates
46 once at the end of the transaction.
48 Similar transfiletriggers have been added to execute any sysctl.d
49 and binfmt.d rules. Thus, it should be unnecessary to provide any
50 scriptlets to execute this configuration from package installation
53 * elogind-sysusers gained a mode where the configuration to execute is
54 specified on the command line, but this configuration is not executed
55 directly, but instead it is merged with the configuration on disk,
56 and the result is executed. This is useful for package installation
57 scripts which want to create the user before installing any files on
58 disk (in case some of those files are owned by that user), while
59 still allowing local admin overrides.
61 This functionality is exposed to rpm scriptlets through a new
62 %sysusers_create_package macro. Old %sysusers_create and
63 %sysusers_create_inline macros are deprecated.
65 A transfiletrigger for sysusers.d configuration is now installed,
66 which means that it should be unnecessary to call elogind-sysusers from
67 package installation scripts, unless the package installs any files
68 owned by those newly-created users, in which case
69 %sysusers_create_package should be used.
71 * Analogous change has been done for elogind-tmpfiles: it gained a mode
72 where the command-line configuration is merged with the configuration
73 on disk. This is exposed as the new %tmpfiles_create_package macro,
74 and %tmpfiles_create is deprecated. A transfiletrigger is installed
75 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
76 from package installation scripts.
78 * sysusers.d configuration for a user may now also specify the group
79 number, in addition to the user number ("u username 123:456"), or
80 without the user number ("u username -:456").
82 * Configution items for elogind-sysusers can now be specified as
83 positional arguments when the new --inline switch is used.
85 * The login shell of users created through sysusers.d may now be
86 specified (previously, it was always /bin/sh for root and
87 /sbin/nologin for other users).
89 * elogind-analyze gained a new --global switch to look at global user
90 configuration. It also gained a unit-paths verb to list the unit load
91 paths that are compiled into elogind (which can be used with
92 --elogind, --user, or --global).
94 * udevadm trigger gained a new --settle/-w option to wait for any
95 triggered events to finish (but just those, and not any other events
96 which are triggered meanwhile).
98 * The action that elogind takes when the lid is closed and the
99 machine is connected to external power can now be configured using
100 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
101 was determined by HandleLidSwitch=, and, for backwards compatibility,
102 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
104 * journalctl will periodically call sd_journal_process() to make it
105 resilient against inotify queue overruns when journal files are
106 rotated very quickly.
108 * Two new functions in libelogind — sd_bus_get_n_queued_read and
109 sd_bus_get_n_queued_write — may be used to check the number of
110 pending bus messages.
112 * elogind gained a new
113 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
114 which can be used to migrate foreign processes to scope and service
115 units. The primary user for this new API is elogind itself: the
116 elogind --user instance uses this call of the elogind --system
117 instance to migrate processes if it itself gets the request to
118 migrate processes and the kernel refuses this due to access
119 restrictions. Thanks to this "elogind-run --scope --user …" works
120 again in pure cgroups v2 environments when invoked from the user
123 * A new TemporaryFileSystem= setting can be used to mask out part of
124 the real file system tree with tmpfs mounts. This may be combined
125 with BindPaths= and BindReadOnlyPaths= to hide files or directories
126 not relevant to the unit, while still allowing some paths lower in
127 the tree to be accessed.
129 ProtectHome=tmpfs may now be used to hide user home and runtime
130 directories from units, in a way that is mostly equivalent to
131 "TemporaryFileSystem=/home /run/user /root".
133 * Non-service units are now started with KeyringMode=shared by default.
134 This means that mount and swapon and other mount tools have access
135 to keys in the main keyring.
137 * /sys/fs/bpf is now mounted automatically.
139 * QNX virtualization is now detected by elogind-detect-virt and may
140 be used in ConditionVirtualization=.
142 * IPAccounting= may now be enabled also for slice units.
144 * A new -Dsplit-bin= build configuration switch may be used to specify
145 whether bin and sbin directories are merged, or if they should be
146 included separately in $PATH and various listings of executable
147 directories. The build configuration scripts will try to autodetect
148 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
149 system, but distributions are encouraged to configure this
152 * A new -Dok-color= build configuration switch may be used to change
153 the colour of "OK" status messages.
155 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
156 PrivateNetwork=yes was buggy in previous versions of elogind. This
157 means that after the upgrade and daemon-reexec, any such units must
160 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
161 will not exclude read-only files owned by root from cleanup.
163 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
164 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
165 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
166 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
167 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
168 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
169 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
170 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
171 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
172 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
173 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
174 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
175 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
176 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
177 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
178 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
184 * Some keyboards come with a zoom see-saw or rocker which until now got
185 mapped to the Linux "zoomin/out" keys in hwdb. However, these
186 keycodes are not recognized by any major desktop. They now produce
187 Up/Down key events so that they can be used for scrolling.
189 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
190 slightly: previously, if an argument was specified for lines of this
191 type (i.e. the right-most column was set) this string was appended to
192 existing files each time elogind-tmpfiles was run. This behaviour was
193 different from what the documentation said, and not particularly
194 useful, as repeated elogind-tmpfiles invocations would not be
195 idempotent and grow such files without bounds. With this release
196 behaviour has been altered slightly, to match what the documentation
197 says: lines of this type only have an effect if the indicated files
198 don't exist yet, and only then the argument string is written to the
201 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
202 elogind-tmpfiles behaviour: previously, read-only files owned by root
203 were always excluded from the file "aging" algorithm (i.e. the
204 automatic clean-up of directories like /tmp based on
205 atime/mtime/ctime). We intend to drop this restriction, and age files
206 by default even when owned by root and read-only. This behaviour was
207 inherited from older tools, but there have been requests to remove
208 it, and it's not obvious why this restriction was made in the first
209 place. Please speak up now, if you are aware of software that reqires
210 this behaviour, otherwise we'll remove the restriction in v238.
212 * A new environment variable $SYSTEMD_OFFLINE is now understood by
213 systemctl. It takes a boolean argument. If on, systemctl assumes it
214 operates on an "offline" OS tree, and will not attempt to talk to the
215 service manager. Previously, this mode was implicitly enabled if a
216 chroot() environment was detected, and this new environment variable
217 now provides explicit control.
219 * .path and .socket units may now be created transiently, too.
220 Previously only service, mount, automount and timer units were
221 supported as transient units. The elogind-run tool has been updated
222 to expose this new functionality, you may hence use it now to bind
223 arbitrary commands to path or socket activation on-the-fly from the
224 command line. Moreover, almost all properties are now exposed for the
225 unit types that already supported transient operation.
227 * The elogind-mount command gained support for a new --owner= parameter
228 which takes a user name, which is then resolved and included in uid=
229 and gid= mount options string of the file system to mount.
231 * A new unit condition ConditionControlGroupController= has been added
232 that checks whether a specific cgroup controller is available.
234 * Unit files, udev's .link files, and elogind-networkd's .netdev and
235 .network files all gained support for a new condition
236 ConditionKernelVersion= for checking against specific kernel
239 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
240 support for configuring device flags in the Flags= setting. In the
241 same files, the [Tunnel] section gained support for configuring
242 AllowLocalRemote=. The [Route] section in .network files gained
243 support for configuring InitialCongestionWindow=,
244 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
245 understands RapidCommit=.
247 * elogind-networkd's DHCPv6 support gained support for Prefix
250 * sd-bus gained support for a new "watch-bind" feature. When this
251 feature is enabled, an sd_bus connection may be set up to connect to
252 an AF_UNIX socket in the file system as soon as it is created. This
253 functionality is useful for writing early-boot services that
254 automatically connect to the system bus as soon as it is started,
255 without ugly time-based polling. elogind-networkd and
256 elogind-resolved have been updated to make use of this
257 functionality. busctl exposes this functionality in a new
258 --watch-bind= command line switch.
260 * sd-bus will now optionally synthesize a local "Connected" signal as
261 soon as a D-Bus connection is set up fully. This message mirrors the
262 already existing "Disconnected" signal which is synthesized when the
263 connection is terminated. This signal is generally useful but
264 particularly handy in combination with the "watch-bind" feature
265 described above. Synthesizing of this message has to be requested
266 explicitly through the new API call sd_bus_set_connected_signal(). In
267 addition a new call sd_bus_is_ready() has been added that checks
268 whether a connection is fully set up (i.e. between the "Connected" and
269 "Disconnected" signals).
271 * sd-bus gained two new calls sd_bus_request_name_async() and
272 sd_bus_release_name_async() for asynchronously registering bus
273 names. Similar, there is now sd_bus_add_match_async() for installing
274 a signal match asynchronously. All of elogind's own services have
275 been updated to make use of these calls. Doing these operations
276 asynchronously has two benefits: it reduces the risk of deadlocks in
277 case of cyclic dependencies between bus services, and it speeds up
278 service initialization since synchronization points for bus
279 round-trips are removed.
281 * sd-bus gained two new calls sd_bus_match_signal() and
282 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
283 and sd_bus_add_match_async() but instead of taking a D-Bus match
284 string take match fields as normal function parameters.
286 * sd-bus gained two new calls sd_bus_set_sender() and
287 sd_bus_message_set_sender() for setting the sender name of outgoing
288 messages (either for all outgoing messages or for just one specific
289 one). These calls are only useful in direct connections as on
290 brokered connections the broker fills in the sender anyway,
291 overwriting whatever the client filled in.
293 * sd-event gained a new pseudo-handle that may be specified on all API
294 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
295 used this refers to the default event loop object of the calling
296 thread. Note however that this does not implicitly allocate one —
297 which has to be done prior by using sd_event_default(). Similarly
298 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
299 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
300 to the default bus of the specified type of the calling thread. Here
301 too this does not implicitly allocate bus connection objects, this
302 has to be done prior with sd_bus_default() and friends.
304 * sd-event gained a new call pair
305 sd_event_source_{get|set}_io_fd_own(). This may be used to request
306 automatic closure of the file descriptor an IO event source watches
307 when the event source is destroyed.
309 * elogind-networkd gained support for natively configuring WireGuard
312 * In previous versions elogind synthesized user records both for the
313 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
314 internally. In order to simplify distribution-wide renames of the
315 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
316 new transitional flag file has been added: if
317 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
318 user and group record within the elogind codebase is disabled.
320 * elogind-notify gained a new --uid= option for selecting the source
321 user/UID to use for notification messages sent to the service
324 * journalctl gained a new --grep= option to list only entries in which
325 the message matches a certain pattern. By default matching is case
326 insensitive if the pattern is lowercase, and case sensitive
327 otherwise. Option --case-sensitive=yes|no can be used to override
328 this an specify case sensitivity or case insensitivity.
330 * There's now a "elogind-analyze service-watchdogs" command for printing
331 the current state of the service runtime watchdog, and optionally
332 enabling or disabling the per-service watchdogs system-wide if given a
333 boolean argument (i.e. the concept you configure in WatchdogSec=), for
334 debugging purposes. There's also a kernel command line option
335 elogind.service_watchdogs= for controlling the same.
337 * Two new "log-level" and "log-target" options for elogind-analyze were
338 addded that merge the now deprecated get-log-level, set-log-level and
339 get-log-target, set-log-target pairs. The deprecated options are still
340 understood for backwards compatibility. The two new options print the
341 current value when no arguments are given, and set them when a
342 level/target is given as an argument.
344 * sysusers.d's "u" lines now optionally accept both a UID and a GID
345 specification, separated by a ":" character, in order to create users
346 where UID and GID do not match.
348 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
349 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
350 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
351 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
352 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
353 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
354 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
355 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
356 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
357 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
358 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
359 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
360 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
361 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
362 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
363 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
364 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
365 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
366 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
367 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
374 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
375 in v235 has been extended to also set the dummy.ko module option
376 numdummies=0, preventing the kernel from automatically creating
377 dummy0. All dummy interfaces must now be explicitly created.
379 * Unknown '%' specifiers in configuration files are now rejected. This
380 applies to units and tmpfiles.d configuration. Any percent characters
381 that are followed by a letter or digit that are not supposed to be
382 interpreted as the beginning of a specifier should be escaped by
383 doubling ("%%"). (So "size=5%" is still accepted, as well as
384 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
385 valid specifiers today.)
387 * systemd-resolved now maintains a new dynamic
388 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
389 recommended to make /etc/resolv.conf a symlink to it. This file
390 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
391 includes dynamically acquired search domains, achieving more correct
392 DNS resolution by software that bypasses local DNS APIs such as NSS.
394 * The "uaccess" udev tag has been dropped from /dev/kvm and
395 /dev/dri/renderD*. These devices now have the 0666 permissions by
396 default (but this may be changed at build-time). /dev/dri/renderD*
397 will now be owned by the "render" group along with /dev/kfd.
399 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
400 systemd-journal-gatewayd.service and
401 systemd-journal-upload.service. This means "nss-systemd" must be
402 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
403 services are resolved properly.
405 * In /etc/fstab two new mount options are now understood:
406 x-systemd.makefs and x-systemd.growfs. The former has the effect that
407 the configured file system is formatted before it is mounted, the
408 latter that the file system is resized to the full block device size
409 after it is mounted (i.e. if the file system is smaller than the
410 partition it resides on, it's grown). This is similar to the fsck
411 logic in /etc/fstab, and pulls in systemd-makefs@.service and
412 systemd-growfs@.service as necessary, similar to
413 systemd-fsck@.service. Resizing is currently only supported on ext4
416 * In systemd-networkd, the IPv6 RA logic now optionally may announce
417 DNS server and domain information.
419 * Support for the LUKS2 on-disk format for encrypted partitions has
420 been added. This requires libcryptsetup2 during compilation and
423 * The systemd --user instance will now signal "readiness" when its
424 basic.target unit has been reached, instead of when the run queue ran
425 empty for the first time.
427 * Tmpfiles.d with user configuration are now also supported.
428 systemd-tmpfiles gained a new --user switch, and snippets placed in
429 ~/.config/user-tmpfiles.d/ and corresponding directories will be
430 executed by systemd-tmpfiles --user running in the new
431 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
432 running in the user session.
434 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
435 %S resolves to the top-level state directory (/var/lib for the system
436 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
437 top-level cache directory (/var/cache for the system instance,
438 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
439 logs directory (/var/log for the system instance,
440 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
441 existing %t specifier, that resolves to the top-level runtime
442 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
445 * journalctl learnt a new parameter --output-fields= for limiting the
446 set of journal fields to output in verbose and JSON output modes.
448 * systemd-timesyncd's configuration file gained a new option
449 RootDistanceMaxSec= for setting the maximum root distance of servers
450 it'll use, as well as the new options PollIntervalMinSec= and
451 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
453 * bootctl gained a new command "list" for listing all available boot
454 menu items on systems that follow the boot loader specification.
456 * systemctl gained a new --dry-run switch that shows what would be done
457 instead of doing it, and is currently supported by the shutdown and
460 * ConditionSecurity= can now detect the TOMOYO security module.
462 * Unit file [Install] sections are now also respected in unit drop-in
463 files. This is intended to be used by drop-ins under /usr/lib/.
465 * systemd-firstboot may now also set the initial keyboard mapping.
467 * Udev "changed" events for devices which are exposed as systemd
468 .device units are now propagated to units specified in
469 ReloadPropagatedFrom= as reload requests.
471 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
472 unit template name (i.e. a name in the form of 'foobar@.service',
473 without the instance component between the '@' and - the '.'), then
474 the escaped sysfs path of the device is automatically used as the
477 * SystemCallFilter= in unit files has been extended so that an "errno"
478 can be specified individually for each system call. Example:
479 SystemCallFilter=~uname:EILSEQ.
481 * The cgroup delegation logic has been substantially updated. Delegate=
482 now optionally takes a list of controllers (instead of a boolean, as
483 before), which lists the controllers to delegate at least.
485 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
487 * A new LogLevelMax= setting configures the maximum log level any
488 process of the service may log at (i.e. anything with a lesser
489 priority than what is specified is automatically dropped). A new
490 LogExtraFields= setting allows configuration of additional journal
491 fields to attach to all log records generated by any of the unit's
494 * New StandardInputData= and StandardInputText= settings along with the
495 new option StandardInput=data may be used to configure textual or
496 binary data that shall be passed to the executed service process via
497 standard input, encoded in-line in the unit file.
499 * StandardInput=, StandardOutput= and StandardError= may now be used to
500 connect stdin/stdout/stderr of executed processes directly with a
501 file or AF_UNIX socket in the file system, using the new "file:" option.
503 * A new unit file option CollectMode= has been added, that allows
504 tweaking the garbage collection logic for units. It may be used to
505 tell systemd to garbage collect units that have failed automatically
506 (normally it only GCs units that exited successfully). systemd-run
507 and systemd-mount expose this new functionality with a new -G option.
509 * "machinectl bind" may now be used to bind mount non-directories
510 (i.e. regularfiles, devices, fifos, sockets).
512 * systemd-analyze gained a new verb "calendar" for validating and
513 testing calendar time specifications to use for OnCalendar= in timer
514 units. Besides validating the expression it will calculate the next
515 time the specified expression would elapse.
517 * In addition to the pre-existing FailureAction= unit file setting
518 there's now SuccessAction=, for configuring a shutdown action to
519 execute when a unit completes successfully. This is useful in
520 particular inside containers that shall terminate after some workload
521 has been completed. Also, both options are now supported for all unit
522 types, not just services.
524 * networkds's IP rule support gained two new options
525 IncomingInterface= and OutgoingInterface= for configuring the incoming
526 and outgoing interfaces of configured rules. systemd-networkd also
527 gained support for "vxcan" network devices.
529 * networkd gained a new setting RequiredForOnline=, taking a
530 boolean. If set, systemd-wait-online will take it into consideration
531 when determining that the system is up, otherwise it will ignore the
532 interface for this purpose.
534 * The sd_notify() protocol gained support for a new operation: with
535 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
536 store again, ahead of POLLHUP or POLLERR when they are removed
539 * A new document doc/UIDS-GIDS.md has been added to the source tree,
540 that documents the UID/GID range and assignment assumptions and
541 requirements of systemd.
543 * The watchdog device PID 1 will ping may now be configured through the
544 WatchdogDevice= configuration file setting, or by setting the
545 systemd.watchdog_service= kernel commandline option.
547 * systemd-resolved's gained support for registering DNS-SD services on
548 the local network using MulticastDNS. Services may either be
549 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
550 the same dir below /run, /usr/lib), or through its D-Bus API.
552 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
553 extend the effective start, runtime, and stop time. The service must
554 continue to send EXTEND_TIMEOUT_USEC within the period specified to
555 prevent the service manager from making the service as timedout.
557 * elogind-resolved's DNSSEC support gained support for RFC 8080
558 (Ed25519 keys and signatures).
560 * The elogind-resolve command line tool gained a new set of options
561 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
562 --set-nta= and --revert to configure per-interface DNS configuration
563 dynamically during runtime. It's useful for pushing DNS information
564 into elogind-resolved from DNS hook scripts that various interface
565 managing software supports (such as pppd).
567 * elogind-nspawn gained a new --network-namespace-path= command line
568 option, which may be used to make a container join an existing
569 network namespace, by specifying a path to a "netns" file.
571 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
572 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
573 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
574 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
575 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
576 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
577 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
578 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
579 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
580 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
581 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
582 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
583 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
584 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
585 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
586 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
587 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
588 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
589 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
590 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
591 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
592 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
593 Jędrzejewski-Szmek, Zeal Jagannatha
599 * INCOMPATIBILITY: systemd-logind.service and other long-running
600 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
601 communication with the outside. This generally improves security of
602 the system, and is in almost all cases a safe and good choice, as
603 these services do not and should not provide any network-facing
604 functionality. However, systemd-logind uses the glibc NSS API to
605 query the user database. This creates problems on systems where NSS
606 is set up to directly consult network services for user database
607 lookups. In particular, this creates incompatibilities with the
608 "nss-nis" module, which attempts to directly contact the NIS/YP
609 network servers it is configured for, and will now consistently
610 fail. In such cases, it is possible to turn off IP sandboxing for
611 systemd-logind.service (set IPAddressDeny= in its [Service] section
612 to the empty string, via a .d/ unit file drop-in). Downstream
613 distributions might want to update their nss-nis packaging to include
614 such a drop-in snippet, accordingly, to hide this incompatibility
615 from the user. Another option is to make use of glibc's nscd service
616 to proxy such network requests through a privilege-separated, minimal
617 local caching daemon, or to switch to more modern technologies such
618 sssd, whose NSS hook-ups generally do not involve direct network
619 access. In general, we think it's definitely time to question the
620 implementation choices of nss-nis, i.e. whether it's a good idea
621 today to embed a network-facing loadable module into all local
622 processes that need to query the user database, including the most
623 trivial and benign ones, such as "ls". For more details about
624 IPAddressDeny= see below.
626 * A new modprobe.d drop-in is now shipped by default that sets the
627 bonding module option max_bonds=0. This overrides the kernel default,
628 to avoid conflicts and ambiguity as to whether or not bond0 should be
629 managed by systemd-networkd or not. This resolves multiple issues
630 with bond0 properties not being applied, when bond0 is configured
631 with systemd-networkd. Distributors may choose to not package this,
632 however in that case users will be prevented from correctly managing
633 bond0 interface using systemd-networkd.
635 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
636 which print the logging level and target of the system manager. They
637 complement the existing "set-log-level" and "set-log-target" verbs
638 used to change those values.
640 * journald.conf gained a new boolean setting ReadKMsg= which defaults
641 to on. If turned off kernel log messages will not be read by
642 systemd-journald or included in the logs. It also gained a new
643 setting LineMax= for configuring the maximum line length in
644 STDOUT/STDERR log streams. The new default for this value is 48K, up
645 from the previous hardcoded 2048.
647 * A new unit setting RuntimeDirectoryPreserve= has been added, which
648 allows more detailed control of what to do with a runtime directory
649 configured with RuntimeDirectory= (i.e. a directory below /run or
650 $XDG_RUNTIME_DIR) after a unit is stopped.
652 * The RuntimeDirectory= setting for units gained support for creating
653 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
654 one top-level directory.
656 * Units gained new options StateDirectory=, CacheDirectory=,
657 LogsDirectory= and ConfigurationDirectory= which are closely related
658 to RuntimeDirectory= but manage per-service directories below
659 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
660 possible to write unit files which when activated automatically gain
661 properly owned service specific directories in these locations, thus
662 making unit files self-contained and increasing compatibility with
663 stateless systems and factory reset where /etc or /var are
664 unpopulated at boot. Matching these new settings there's also
665 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
666 ConfigurationDirectoryMode= for configuring the access mode of these
667 directories. These settings are particularly useful in combination
668 with DynamicUser=yes as they provide secure, properly-owned,
669 writable, and stateful locations for storage, excluded from the
670 sandbox that such services live in otherwise.
672 * Automake support has been removed from this release. systemd is now
675 * systemd-journald will now aggressively cache client metadata during
676 runtime, speeding up log write performance under pressure. This comes
677 at a small price though: as much of the metadata is read
678 asynchronously from /proc/ (and isn't implicitly attached to log
679 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
680 metadata stored alongside a log entry might be slightly
681 out-of-date. Previously it could only be slightly newer than the log
682 message. The time window is small however, and given that the kernel
683 is unlikely to be improved anytime soon in this regard, this appears
686 * nss-myhostname/systemd-resolved will now by default synthesize an
687 A/AAAA resource record for the "_gateway" hostname, pointing to the
688 current default IP gateway. Previously it did that for the "gateway"
689 name, hampering adoption, as some distributions wanted to leave that
690 host name open for local use. The old behaviour may still be
691 requested at build time.
693 * systemd-networkd's [Address] section in .network files gained a new
694 Scope= setting for configuring the IP address scope. The [Network]
695 section gained a new boolean setting ConfigureWithoutCarrier= that
696 tells systemd-networkd to ignore link sensing when configuring the
697 device. The [DHCP] section gained a new Anonymize= boolean option for
698 turning on a number of options suggested in RFC 7844. A new
699 [RoutingPolicyRule] section has been added for configuring the IP
700 routing policy. The [Route] section has gained support for a new
701 Type= setting which permits configuring
702 blackhole/unreachable/prohibit routes.
704 * The [VRF] section in .netdev files gained a new Table= setting for
705 configuring the routing table to use. The [Tunnel] section gained a
706 new Independent= boolean field for configuring tunnels independent of
707 an underlying network interface. The [Bridge] section gained a new
708 GroupForwardMask= option for configuration of propagation of link
709 local frames between bridge ports.
711 * The WakeOnLan= setting in .link files gained support for a number of
712 new modes. A new TCP6SegmentationOffload= setting has been added for
713 configuring TCP/IPv6 hardware segmentation offload.
715 * The IPv6 RA sender implementation may now optionally send out RDNSS
716 and RDNSSL records to supply DNS configuration to peers.
718 * systemd-nspawn gained support for a new --system-call-filter= command
719 line option for adding and removing entries in the default system
720 call filter it applies. Moreover systemd-nspawn has been changed to
721 implement a system call whitelist instead of a blacklist.
723 * systemd-run gained support for a new --pipe command line option. If
724 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
725 are directly passed on to the activated transient service
726 executable. This allows invoking arbitrary processes as systemd
727 services (for example to take benefit of dependency management,
728 accounting management, resource management or log management that is
729 done automatically for services) — while still allowing them to be
730 integrated in a classic UNIX shell pipeline.
732 * When a service sends RELOAD=1 via sd_notify() and reload propagation
733 using ReloadPropagationTo= is configured, a reload is now propagated
734 to configured units. (Previously this was only done on explicitly
735 requested reloads, using "systemctl reload" or an equivalent
738 * For each service unit a restart counter is now kept: it is increased
739 each time the service is restarted due to Restart=, and may be
740 queried using "systemctl show -p NRestarts …".
742 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
743 @signal and @timer have been added, for usage with SystemCallFilter=
744 in unit files and the new --system-call-filter= command line option
745 of systemd-nspawn (see above).
747 * ExecStart= lines in unit files gained two new modifiers: when a
748 command line is prefixed with "!" the command will be executed as
749 configured, except for the credentials applied by
750 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
751 "+", but does still apply namespacing options unlike "+". There's
752 also "!!" now, which is mostly identical, but becomes a NOP on
753 systems that support ambient capabilities. This is useful to write
754 unit files that work with ambient capabilities where possible but
755 automatically fall back to traditional privilege dropping mechanisms
756 on systems where this is not supported.
758 * ListenNetlink= settings in socket units now support RDMA netlink
761 * A new unit file setting LockPersonality= has been added which permits
762 locking down the chosen execution domain ("personality") of a service
765 * A new special target "getty-pre.target" has been added, which is
766 ordered before all text logins, and may be used to order services
767 before textual logins acquire access to the console.
769 * systemd will now attempt to load the virtio-rng.ko kernel module very
770 early on if a VM environment supporting this is detected. This should
771 improve entropy during early boot in virtualized environments.
773 * A _netdev option is now supported in /etc/crypttab that operates in a
774 similar way as the same option in /etc/fstab: it permits configuring
775 encrypted devices that need to be ordered after the network is up.
776 Following this logic, two new special targets
777 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
778 added that are to cryptsetup.target what remote-fs.target and
779 remote-fs-pre.target are to local-fs.target.
781 * Service units gained a new UnsetEnvironment= setting which permits
782 unsetting specific environment variables for services that are
783 normally passed to it (for example in order to mask out locale
784 settings for specific services that can't deal with it).
786 * Units acquired a new boolean option IPAccounting=. When turned on, IP
787 traffic accounting (packet count as well as byte count) is done for
788 the service, and shown as part of "systemctl status" or "systemd-run
791 * Service units acquired two new options IPAddressAllow= and
792 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
793 for configuring a simple IP access control list for all sockets of
794 the unit. These options are available also on .slice and .socket
795 units, permitting flexible access list configuration for individual
796 services as well as groups of services (as defined by a slice unit),
797 including system-wide. Note that IP ACLs configured this way are
798 enforced on every single IPv4 and IPv6 socket created by any process
799 of the service unit, and apply to ingress as well as egress traffic.
801 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
802 structured log message is generated each time the unit is stopped,
803 containing information about the consumed resources of this
806 * A new setting KeyringMode= has been added to unit files, which may be
807 used to control how the kernel keyring is set up for executed
810 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
811 "systemctl kexec" and "systemctl exit" are now always asynchronous in
812 behaviour (that is: these commands return immediately after the
813 operation was enqueued instead of waiting for the operation to
814 complete). Previously, "systemctl poweroff" and "systemctl reboot"
815 were asynchronous on systems using systemd-logind (i.e. almost
816 always, and like they were on sysvinit), and the other three commands
817 were unconditionally synchronous. With this release this is cleaned
818 up, and callers will see the same asynchronous behaviour on all
819 systems for all five operations.
821 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
824 * .timer units now accept calendar specifications in other timezones
825 than UTC or the local timezone.
827 * The tmpfiles snippet var.conf has been changed to create
828 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
829 the "utmp" group already, and it appears to be generally understood
830 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
831 databases. Previously this was implemented correctly for all these
832 databases excepts btmp, which has been opened up like this now
833 too. Note that while the other databases are world-readable
834 (i.e. 0644), btmp is not and remains more restrictive.
836 * The systemd-resolve tool gained a new --reset-server-features
837 switch. When invoked like this systemd-resolved will forget
838 everything it learnt about the features supported by the configured
839 upstream DNS servers, and restarts the feature probing logic on the
840 next resolver look-up for them at the highest feature level
843 * The status dump systemd-resolved sends to the logs upon receiving
844 SIGUSR1 now also includes information about all DNS servers it is
845 configured to use, and the features levels it probed for them.
847 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
848 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
849 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
850 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
851 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
852 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
853 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
854 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
855 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
856 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
857 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
858 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
859 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
860 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
861 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
862 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
863 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
864 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
865 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
866 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
872 * Meson is now supported as build system in addition to Automake. It is
873 our plan to remove Automake in one of our next releases, so that
874 Meson becomes our exclusive build system. Hence, please start using
875 the Meson build system in your downstream packaging. There's plenty
876 of documentation around how to use Meson, the extremely brief
879 ./autogen.sh && ./configure && make && sudo make install
883 meson build && ninja -C build && sudo ninja -C build install
885 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
886 which permits configuring a timeout on the time a job is
887 running. This is particularly useful for setting timeouts on jobs for
890 * Unit files gained two new options ConditionUser= and ConditionGroup=
891 for conditionalizing units based on the identity of the user/group
892 running a systemd user instance.
894 * systemd-networkd now understands a new FlowLabel= setting in the
895 [VXLAN] section of .network files, as well as a Priority= in
896 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
897 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
898 gained support for configuration of GENEVE links, and IPv6 address
899 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
901 * .link files now understand a new Port= setting.
903 * systemd-networkd's DHCP support gained support for DHCP option 119
904 (domain search list).
906 * systemd-networkd gained support for serving IPv6 address ranges using
907 the Router Advertisment protocol. The new .network configuration
908 section [IPv6Prefix] may be used to configure the ranges to
909 serve. This is implemented based on a new, minimal, native server
910 implementation of RA.
912 * journalctl's --output= switch gained support for a new parameter
913 "short-iso-precise" for a mode where timestamps are shown as precise
916 * systemd-udevd's "net_id" builtin may now generate stable network
917 interface names from IBM PowerVM VIO devices as well as ACPI platform
920 * MulticastDNS support in systemd-resolved may now be explicitly
921 enabled/disabled using the new MulticastDNS= configuration file
924 * systemd-resolved may now optionally use libidn2 instead of the libidn
925 for processing internationalized domain names. Support for libidn2
926 should be considered experimental and should not be enabled by
929 * "machinectl pull-tar" and related call may now do verification of
930 downloaded images using SUSE-style .sha256 checksum files in addition
931 to the already existing support for validating using Ubuntu-style
934 * sd-bus gained support for a new sd_bus_message_appendv() call which
935 is va_list equivalent of sd_bus_message_append().
937 * sd-boot gained support for validating images using SHIM/MOK.
939 * The SMACK code learnt support for "onlycap".
941 * systemd-mount --umount is now much smarter in figuring out how to
942 properly unmount a device given its mount or device path.
944 * The code to call libnss_dns as a fallback from libnss_resolve when
945 the communication with systemd-resolved fails was removed. This
946 fallback was redundant and interfered with the [!UNAVAIL=return]
947 suffix. See nss-resolve(8) for the recommended configuration.
949 * systemd-logind may now be restarted without losing state. It stores
950 the file descriptors for devices it manages in the system manager
951 using the FDSTORE= mechanism. Please note that further changes in
952 other components may be required to make use of this (for example
953 Xorg has code to listen for stops of systemd-logind and terminate
954 itself when logind is stopped or restarted, in order to avoid using
955 stale file descriptors for graphical devices, which is now
956 counterproductive and must be reverted in order for restarts of
957 systemd-logind to be safe. See
958 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
960 * All kernel install plugins are called with the environment variable
961 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
962 /etc/machine-id. If the file is missing or empty, the variable is
963 empty and BOOT_DIR_ABS is the path of a temporary directory which is
964 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
965 is empty, all plugins should not put anything in BOOT_DIR_ABS.
967 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
968 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
969 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
970 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
971 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
972 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
973 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
974 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
975 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
976 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
977 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
978 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
979 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
980 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
981 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
982 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
983 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
984 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
985 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
986 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
987 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
988 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
989 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
990 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
991 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
992 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
993 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
994 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
995 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1002 * The "hybrid" control group mode has been modified to improve
1003 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1004 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1005 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1006 cgroups-v1 hierarchy), the only externally visible change being that
1007 the cgroups-v2 hierarchy is also mounted, to
1008 /sys/fs/cgroup/unified. This should provide a large degree of
1009 compatibility with "legacy" cgroups-v1, while taking benefit of the
1010 better management capabilities of cgroups-v2.
1012 * The default control group setup mode may be selected both a boot-time
1013 via a set of kernel command line parameters (specifically:
1014 systemd.unified_cgroup_hierarchy= and
1015 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1016 default selected on the configure command line
1017 (--with-default-hierarchy=). The upstream default is "hybrid"
1018 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1019 this will change in a future systemd version to be "unified" (pure
1020 cgroups-v2 mode). The third option for the compile time option is
1021 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1022 distributions to default to "hybrid" mode for release distributions,
1023 starting with v233. We recommend "unified" for development
1024 distributions (specifically: distributions such as Fedora's rawhide)
1025 as that's where things are headed in the long run. Use "legacy" for
1026 greatest stability and compatibility only.
1028 * Note one current limitation of "unified" and "hybrid" control group
1029 setup modes: the kernel currently does not permit the systemd --user
1030 instance (i.e. unprivileged code) to migrate processes between two
1031 disconnected cgroup subtrees, even if both are managed and owned by
1032 the user. This effectively means "systemd-run --user --scope" doesn't
1033 work when invoked from outside of any "systemd --user" service or
1034 scope. Specifically, it is not supported from session scopes. We are
1035 working on fixing this in a future systemd version. (See #3388 for
1036 further details about this.)
1038 * DBus policy files are now installed into /usr rather than /etc. Make
1039 sure your system has dbus >= 1.9.18 running before upgrading to this
1040 version, or override the install path with --with-dbuspolicydir= .
1042 * All python scripts shipped with systemd (specifically: the various
1043 tests written in Python) now require Python 3.
1045 * systemd unit tests can now run standalone (without the source or
1046 build directories), and can be installed into /usr/lib/systemd/tests/
1047 with 'make install-tests'.
1049 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1050 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1053 * Support for the %c, %r, %R specifiers in unit files has been
1054 removed. Specifiers are not supposed to be dependent on configuration
1055 in the unit file itself (so that they resolve the same regardless
1056 where used in the unit files), but these specifiers were influenced
1057 by the Slice= option.
1059 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1060 all cases. If distributions want to use a different shell for this
1061 purpose (for example Fedora's /sbin/sushell) they need to specify
1062 this explicitly at configure time using --with-debug-shell=.
1064 * The confirmation spawn prompt has been reworked to offer the
1067 (c)ontinue, proceed without asking anymore
1068 (D)ump, show the state of the unit
1069 (f)ail, don't execute the command and pretend it failed
1071 (i)nfo, show a short summary of the unit
1072 (j)obs, show jobs that are in progress
1073 (s)kip, don't execute the command and pretend it succeeded
1074 (y)es, execute the command
1076 The 'n' choice for the confirmation spawn prompt has been removed,
1077 because its meaning was confusing.
1079 The prompt may now also be redirected to an alternative console by
1080 specifying the console as parameter to systemd.confirm_spawn=.
1082 * Services of Type=notify require a READY=1 notification to be sent
1083 during startup. If no such message is sent, the service now fails,
1084 even if the main process exited with a successful exit code.
1086 * Services that fail to start up correctly now always have their
1087 ExecStopPost= commands executed. Previously, they'd enter "failed"
1088 state directly, without executing these commands.
1090 * The option MulticastDNS= of network configuration files has acquired
1091 an actual implementation. With MulticastDNS=yes a host can resolve
1092 names of remote hosts and reply to mDNS A and AAAA requests.
1094 * When units are about to be started an additional check is now done to
1095 ensure that all dependencies of type BindsTo= (when used in
1096 combination with After=) have been started.
1098 * systemd-analyze gained a new verb "syscall-filter" which shows which
1099 system call groups are defined for the SystemCallFilter= unit file
1100 setting, and which system calls they contain.
1102 * A new system call filter group "@filesystem" has been added,
1103 consisting of various file system related system calls. Group
1104 "@reboot" has been added, covering reboot, kexec and shutdown related
1105 calls. Finally, group "@swap" has been added covering swap
1106 configuration related calls.
1108 * A new unit file option RestrictNamespaces= has been added that may be
1109 used to restrict access to the various process namespace types the
1110 Linux kernel provides. Specifically, it may be used to take away the
1111 right for a service unit to create additional file system, network,
1112 user, and other namespaces. This sandboxing option is particularly
1113 relevant due to the high amount of recently discovered namespacing
1114 related vulnerabilities in the kernel.
1116 * systemd-udev's .link files gained support for a new AutoNegotiation=
1117 setting for configuring Ethernet auto-negotiation.
1119 * systemd-networkd's .network files gained support for a new
1120 ListenPort= setting in the [DHCP] section to explicitly configure the
1121 UDP client port the DHCP client shall listen on.
1123 * .network files gained a new Unmanaged= boolean setting for explicitly
1124 excluding one or more interfaces from management by systemd-networkd.
1126 * The systemd-networkd ProxyARP= option has been renamed to
1127 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1128 renamed to ReduceARPProxy=. The old names continue to be available
1131 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1132 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1134 * systemd-networkd's bonding device support gained support for two new
1135 configuration options ActiveSlave= and PrimarySlave=.
1137 * The various options in the [Match] section of .network files gained
1138 support for negative matching.
1140 * New systemd-specific mount options are now understood in /etc/fstab:
1142 x-systemd.mount-timeout= may be used to configure the maximum
1143 permitted runtime of the mount command.
1145 x-systemd.device-bound may be set to bind a mount point to its
1146 backing device unit, in order to automatically remove a mount point
1147 if its backing device is unplugged. This option may also be
1148 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1149 on the block device, which is now automatically set for all CDROM
1150 drives, so that mounted CDs are automatically unmounted when they are
1151 removed from the drive.
1153 x-systemd.after= and x-systemd.before= may be used to explicitly
1154 order a mount after or before another unit or mount point.
1156 * Enqueued start jobs for device units are now automatically garbage
1157 collected if there are no jobs waiting for them anymore.
1159 * systemctl list-jobs gained two new switches: with --after, for every
1160 queued job the jobs it's waiting for are shown; with --before the
1161 jobs which it's blocking are shown.
1163 * systemd-nspawn gained support for ephemeral boots from disk images
1164 (or in other words: --ephemeral and --image= may now be
1165 combined). Moreover, ephemeral boots are now supported for normal
1166 directories, even if the backing file system is not btrfs. Of course,
1167 if the file system does not support file system snapshots or
1168 reflinks, the initial copy operation will be relatively expensive, but
1169 this should still be suitable for many use cases.
1171 * Calendar time specifications in .timer units now support
1172 specifications relative to the end of a month by using "~" instead of
1173 "-" as separator between month and day. For example, "*-02~03" means
1174 "the third last day in February". In addition a new syntax for
1175 repeated events has been added using the "/" character. For example,
1176 "9..17/2:00" means "every two hours from 9am to 5pm".
1178 * systemd-socket-proxyd gained a new parameter --connections-max= for
1179 configuring the maximum number of concurrent connections.
1181 * sd-id128 gained a new API for generating unique IDs for the host in a
1182 way that does not leak the machine ID. Specifically,
1183 sd_id128_get_machine_app_specific() derives an ID based on the
1184 machine ID a in well-defined, non-reversible, stable way. This is
1185 useful whenever an identifier for the host is needed but where the
1186 identifier shall not be useful to identify the system beyond the
1187 scope of the application itself. (Internally this uses HMAC-SHA256 as
1188 keyed hash function using the machine ID as input.)
1190 * NotifyAccess= gained a new supported value "exec". When set
1191 notifications are accepted from all processes systemd itself invoked,
1192 including all control processes.
1194 * .nspawn files gained support for defining overlay mounts using the
1195 Overlay= and OverlayReadOnly= options. Previously this functionality
1196 was only available on the systemd-nspawn command line.
1198 * systemd-nspawn's --bind= and --overlay= options gained support for
1199 bind/overlay mounts whose source lies within the container tree by
1200 prefixing the source path with "+".
1202 * systemd-nspawn's --bind= and --overlay= options gained support for
1203 automatically allocating a temporary source directory in /var/tmp
1204 that is removed when the container dies. Specifically, if the source
1205 directory is specified as empty string this mechanism is selected. An
1206 example usage is --overlay=+/var::/var, which creates an overlay
1207 mount based on the original /var contained in the image, overlayed
1208 with a temporary directory in the host's /var/tmp. This way changes
1209 to /var are automatically flushed when the container shuts down.
1211 * systemd-nspawn --image= option does now permit raw file system block
1212 devices (in addition to images containing partition tables, as
1215 * The disk image dissection logic in systemd-nspawn gained support for
1216 automatically setting up LUKS encrypted as well as Verity protected
1217 partitions. When a container is booted from an encrypted image the
1218 passphrase is queried at start-up time. When a container with Verity
1219 data is started, the root hash is search in a ".roothash" file
1220 accompanying the disk image (alternatively, pass the root hash via
1221 the new --root-hash= command line option).
1223 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1224 be used to dissect disk images the same way as systemd-nspawn does
1225 it, following the Bootable Partition Specification. It may even be
1226 used to mount disk images with complex partition setups (including
1227 LUKS and Verity partitions) to a local host directory, in order to
1228 inspect them. This tool is not considered public API (yet), and is
1229 thus not installed into /usr/bin. Please do not rely on its
1230 existence, since it might go away or be changed in later systemd
1233 * A new generator "systemd-verity-generator" has been added, similar in
1234 style to "systemd-cryptsetup-generator", permitting automatic setup of
1235 Verity root partitions when systemd boots up. In order to make use of
1236 this your partition setup should follow the Discoverable Partitions
1237 Specification, and the GPT partition ID of the root file system
1238 partition should be identical to the upper 128bit of the Verity root
1239 hash. The GPT partition ID of the Verity partition protecting it
1240 should be the lower 128bit of the Verity root hash. If the partition
1241 image follows this model it is sufficient to specify a single
1242 "roothash=" kernel command line argument to both configure which root
1243 image and verity partition to use as well as the root hash for
1244 it. Note that systemd-nspawn's Verity support follows the same
1245 semantics, meaning that disk images with proper Verity data in place
1246 may be booted in containers with systemd-nspawn as well as on
1247 physical systems via the verity generator. Also note that the "mkosi"
1248 tool available at https://github.com/systemd/mkosi has been updated
1249 to generate Verity protected disk images following this scheme. In
1250 fact, it has been updated to generate disk images that optionally
1251 implement a complete UEFI SecureBoot trust chain, involving a signed
1252 kernel and initrd image that incorporates such a root hash as well as
1253 a Verity-enabled root partition.
1255 * The hardware database (hwdb) udev supports has been updated to carry
1256 accelerometer quirks.
1258 * All system services are now run with a fresh kernel keyring set up
1259 for them. The invocation ID is stored by default in it, thus
1260 providing a safe, non-overridable way to determine the invocation
1263 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1264 options for bind mounting arbitrary paths in a service-specific
1265 way. When these options are used, arbitrary host or service files and
1266 directories may be mounted to arbitrary locations in the service's
1269 * Documentation has been added that lists all of systemd's low-level
1270 environment variables:
1272 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1274 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1275 whether a specific socket file descriptor matches a specified socket
1278 * systemd-firstboot has been updated to check for the
1279 systemd.firstboot= kernel command line option. It accepts a boolean
1280 and when set to false the first boot questions are skipped.
1282 * systemd-fstab-generator has been updated to check for the
1283 systemd.volatile= kernel command line option, which either takes an
1284 optional boolean parameter or the special value "state". If used the
1285 system may be booted in a "volatile" boot mode. Specifically,
1286 "systemd.volatile" is used, the root directory will be mounted as
1287 tmpfs, and only /usr is mounted from the actual root file system. If
1288 "systemd.volatile=state" is used, the root directory will be mounted
1289 as usual, but /var is mounted as tmpfs. This concept provides similar
1290 functionality as systemd-nspawn's --volatile= option, but provides it
1291 on physical boots. Use this option for implementing stateless
1292 systems, or testing systems with all state and/or configuration reset
1293 to the defaults. (Note though that many distributions are not
1294 prepared to boot up without a populated /etc or /var, though.)
1296 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1297 partitions. Previously it only supported LUKS encrypted partitions
1298 for all other uses, except for the root partition itself.
1300 * Socket units gained support for listening on AF_VSOCK sockets for
1301 communication in virtualized QEMU environments.
1303 * The "configure" script gained a new option --with-fallback-hostname=
1304 for specifying the fallback hostname to use if none is configured in
1305 /etc/hostname. For example, by specifying
1306 --with-fallback-hostname=fedora it is possible to default to a
1307 hostname of "fedora" on pristine installations.
1309 * systemd-cgls gained support for a new --unit= switch for listing only
1310 the control groups of a specific unit. Similar --user-unit= has been
1311 added for listing only the control groups of a specific user unit.
1313 * systemd-mount gained a new --umount switch for unmounting a mount or
1314 automount point (and all mount/automount points below it).
1316 * systemd will now refuse full configuration reloads (via systemctl
1317 daemon-reload and related calls) unless at least 16MiB of free space
1318 are available in /run. This is a safety precaution in order to ensure
1319 that generators can safely operate after the reload completed.
1321 * A new unit file option RootImage= has been added, which has a similar
1322 effect as RootDirectory= but mounts the service's root directory from
1323 a disk image instead of plain directory. This logic reuses the same
1324 image dissection and mount logic that systemd-nspawn already uses,
1325 and hence supports any disk images systemd-nspawn supports, including
1326 those following the Discoverable Partition Specification, as well as
1327 Verity enabled images. This option enables systemd to run system
1328 services directly off disk images acting as resource bundles,
1329 possibly even including full integrity data.
1331 * A new MountAPIVFS= unit file option has been added, taking a boolean
1332 argument. If enabled /proc, /sys and /dev (collectively called the
1333 "API VFS") will be mounted for the service. This is only relevant if
1334 RootDirectory= or RootImage= is used for the service, as these mounts
1335 are of course in place in the host mount namespace anyway.
1337 * systemd-nspawn gained support for a new --pivot-root= switch. If
1338 specified the root directory within the container image is pivoted to
1339 the specified mount point, while the original root disk is moved to a
1340 different place. This option enables booting of ostree images
1341 directly with systemd-nspawn.
1343 * The systemd build scripts will no longer complain if the NTP server
1344 addresses are not changed from the defaults. Google now supports
1345 these NTP servers officially. We still recommend downstreams to
1346 properly register an NTP pool with the NTP pool project though.
1348 * coredumpctl gained a new "--reverse" option for printing the list
1349 of coredumps in reverse order.
1351 * coredumpctl will now show additional information about truncated and
1352 inaccessible coredumps, as well as coredumps that are still being
1353 processed. It also gained a new --quiet switch for suppressing
1354 additional informational message in its output.
1356 * coredumpctl gained support for only showing coredumps newer and/or
1357 older than specific timestamps, using the new --since= and --until=
1358 options, reminiscent of journalctl's options by the same name.
1360 * The systemd-coredump logic has been improved so that it may be reused
1361 to collect backtraces in non-compiled languages, for example in
1362 scripting languages such as Python.
1364 * machinectl will now show the UID shift of local containers, if user
1365 namespacing is enabled for them.
1367 * systemd will now optionally run "environment generator" binaries at
1368 configuration load time. They may be used to add environment
1369 variables to the environment block passed to services invoked. One
1370 user environment generator is shipped by default that sets up
1371 environment variables based on files dropped into /etc/environment.d
1372 and ~/.config/environment.d/.
1374 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1377 * hostnamed has been updated to report a new chassis type of
1378 "convertible" to cover "foldable" laptops that can both act as a
1379 tablet and as a laptop, such as various Lenovo Yoga devices.
1381 Contributions from: Adrián López, Alexander Galanin, Alexander
1382 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1383 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1384 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1385 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1386 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1387 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1388 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1389 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1390 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1391 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1392 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1393 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1394 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1395 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1396 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1397 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1398 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1399 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1400 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1401 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1402 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1403 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1404 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1405 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1406 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1407 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1408 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1411 — Berlin, 2017-03-01
1415 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1416 RestrictAddressFamilies= enabled. These sandboxing options should
1417 generally be compatible with the various external udev call-out
1418 binaries we are aware of, however there may be exceptions, in
1419 particular when exotic languages for these call-outs are used. In
1420 this case, consider turning off these settings locally.
1422 * The new RemoveIPC= option can be used to remove IPC objects owned by
1423 the user or group of a service when that service exits.
1425 * The new ProtectKernelModules= option can be used to disable explicit
1426 load and unload operations of kernel modules by a service. In
1427 addition access to /usr/lib/modules is removed if this option is set.
1429 * ProtectSystem= option gained a new value "strict", which causes the
1430 whole file system tree with the exception of /dev, /proc, and /sys,
1431 to be remounted read-only for a service.
1433 * The new ProtectKernelTunables= option can be used to disable
1434 modification of configuration files in /sys and /proc by a service.
1435 Various directories and files are remounted read-only, so access is
1436 restricted even if the file permissions would allow it.
1438 * The new ProtectControlGroups= option can be used to disable write
1439 access by a service to /sys/fs/cgroup.
1441 * Various systemd services have been hardened with
1442 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1443 RestrictAddressFamilies=.
1445 * Support for dynamically creating users for the lifetime of a service
1446 has been added. If DynamicUser=yes is specified, user and group IDs
1447 will be allocated from the range 61184..65519 for the lifetime of the
1448 service. They can be resolved using the new nss-systemd.so NSS
1449 module. The module must be enabled in /etc/nsswitch.conf. Services
1450 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1451 any resources allocated by the service will be cleaned up when the
1452 service exits. They also have ProtectHome=read-only and
1453 ProtectSystem=strict enabled, so they are not able to make any
1454 permanent modifications to the system.
1456 * The nss-systemd module also always resolves root and nobody, making
1457 it possible to have no /etc/passwd or /etc/group files in minimal
1458 container or chroot environments.
1460 * Services may be started with their own user namespace using the new
1461 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1462 under which the service is running are mapped. All other users are
1465 * Support for the cgroup namespace has been added to systemd-nspawn. If
1466 supported by kernel, the container system started by systemd-nspawn
1467 will have its own view of the cgroup hierarchy. This new behaviour
1468 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1470 * The new MemorySwapMax= option can be used to limit the maximum swap
1471 usage under the unified cgroup hierarchy.
1473 * Support for the CPU controller in the unified cgroup hierarchy has
1474 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1475 options. This controller requires out-of-tree patches for the kernel
1476 and the support is provisional.
1478 * Mount and automount units may now be created transiently
1479 (i.e. dynamically at runtime via the bus API, instead of requiring
1480 unit files in the file system).
1482 * systemd-mount is a new tool which may mount file systems – much like
1483 mount(8), optionally pulling in additional dependencies through
1484 transient .mount and .automount units. For example, this tool
1485 automatically runs fsck on a backing block device before mounting,
1486 and allows the automount logic to be used dynamically from the
1487 command line for establishing mount points. This tool is particularly
1488 useful when dealing with removable media, as it will ensure fsck is
1489 run – if necessary – before the first access and that the file system
1490 is quickly unmounted after each access by utilizing the automount
1491 logic. This maximizes the chance that the file system on the
1492 removable media stays in a clean state, and if it isn't in a clean
1493 state is fixed automatically.
1495 * LazyUnmount=yes option for mount units has been added to expose the
1496 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1499 * /efi will be used as the mount point of the EFI boot partition, if
1500 the directory is present, and the mount point was not configured
1501 through other means (e.g. fstab). If /efi directory does not exist,
1502 /boot will be used as before. This makes it easier to automatically
1503 mount the EFI partition on systems where /boot is used for something
1506 * When operating on GPT disk images for containers, systemd-nspawn will
1507 now mount the ESP to /boot or /efi according to the same rules as PID
1508 1 running on a host. This allows tools like "bootctl" to operate
1509 correctly within such containers, in order to make container images
1510 bootable on physical systems.
1512 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1514 * Two new user session targets have been added to support running
1515 graphical sessions under the systemd --user instance:
1516 graphical-session.target and graphical-session-pre.target. See
1517 systemd.special(7) for a description of how those targets should be
1520 * The vconsole initialization code has been significantly reworked to
1521 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1522 support unicode keymaps. Font and keymap configuration will now be
1523 copied to all allocated virtual consoles.
1525 * FreeBSD's bhyve virtualization is now detected.
1527 * Information recorded in the journal for core dumps now includes the
1528 contents of /proc/mountinfo and the command line of the process at
1529 the top of the process hierarchy (which is usually the init process
1532 * systemd-journal-gatewayd learned the --directory= option to serve
1533 files from the specified location.
1535 * journalctl --root=… can be used to peruse the journal in the
1536 /var/log/ directories inside of a container tree. This is similar to
1537 the existing --machine= option, but does not require the container to
1540 * The hardware database has been extended to support
1541 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1544 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1545 specify the click rate for mice which include a horizontal wheel with
1546 a click rate that is different than the one for the vertical wheel.
1548 * systemd-run gained a new --wait option that makes service execution
1549 synchronous. (Specifically, the command will not return until the
1550 specified service binary exited.)
1552 * systemctl gained a new --wait option that causes the start command to
1553 wait until the units being started have terminated again.
1555 * A new journal output mode "short-full" has been added which displays
1556 timestamps with abbreviated English day names and adds a timezone
1557 suffix. Those timestamps include more information than the default
1558 "short" output mode, and can be passed directly to journalctl's
1559 --since= and --until= options.
1561 * /etc/resolv.conf will be bind-mounted into containers started by
1562 systemd-nspawn, if possible, so any changes to resolv.conf contents
1563 are automatically propagated to the container.
1565 * The number of instances for socket-activated services originating
1566 from a single IP address can be limited with
1567 MaxConnectionsPerSource=, extending the existing setting of
1570 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1573 * .netdev and .network configuration can now be extended through
1576 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1577 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1578 can be enabled and disabled using the new UDPSegmentationOffload=,
1579 TCPSegmentationOffload=, GenericSegmentationOffload=,
1580 GenericReceiveOffload=, LargeReceiveOffload= options in the
1581 [Link] section of .link files.
1583 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1584 Port VLAN ID can be configured for bridge devices using the new STP=,
1585 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1586 section of .netdev files.
1588 * The route table to which routes received over DHCP or RA should be
1589 added can be configured with the new RouteTable= option in the [DHCP]
1590 and [IPv6AcceptRA] sections of .network files.
1592 * The Address Resolution Protocol can be disabled on links managed by
1593 systemd-networkd using the ARP=no setting in the [Link] section of
1596 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1597 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1598 encode information about the result and exit codes of the current
1599 service runtime cycle.
1601 * systemd-sysctl will now configure kernel parameters in the order
1602 they occur in the configuration files. This matches what sysctl
1603 has been traditionally doing.
1605 * kernel-install "plugins" that are executed to perform various
1606 tasks after a new kernel is added and before an old one is removed
1607 can now return a special value to terminate the procedure and
1608 prevent any later plugins from running.
1610 * Journald's SplitMode=login setting has been deprecated. It has been
1611 removed from documentation, and its use is discouraged. In a future
1612 release it will be completely removed, and made equivalent to current
1613 default of SplitMode=uid.
1615 * Storage=both option setting in /etc/systemd/coredump.conf has been
1616 removed. With fast LZ4 compression storing the core dump twice is not
1619 * The --share-system systemd-nspawn option has been replaced with an
1620 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1621 this functionality is discouraged. In addition the variables
1622 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1623 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1624 individual namespaces.
1626 * "machinectl list" now shows the IP address of running containers in
1627 the output, as well as OS release information.
1629 * "loginctl list" now shows the TTY of each session in the output.
1631 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1632 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1633 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1634 tracking objects in a "recursive" mode, where a single client can be
1635 counted multiple times, if it takes multiple references.
1637 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1638 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1639 process using sd-bus automatically exit if the bus connection is
1642 * Bus clients of the service manager may now "pin" loaded units into
1643 memory, by taking an explicit reference on them. This is useful to
1644 ensure the client can retrieve runtime data about the service even
1645 after the service completed execution. Taking such a reference is
1646 available only for privileged clients and should be helpful to watch
1647 running services in a race-free manner, and in particular collect
1648 information about exit statuses and results.
1650 * The nss-resolve module has been changed to strictly return UNAVAIL
1651 when communication via D-Bus with resolved failed, and NOTFOUND when
1652 a lookup completed but was negative. This means it is now possible to
1653 neatly configure fallbacks using nsswitch.conf result checking
1654 expressions. Taking benefit of this, the new recommended
1655 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1657 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1659 * A new setting CtrlAltDelBurstAction= has been added to
1660 /etc/systemd/system.conf which may be used to configure the precise
1661 behaviour if the user on the console presses Ctrl-Alt-Del more often
1662 than 7 times in 2s. Previously this would unconditionally result in
1663 an expedited, immediate reboot. With this new setting the precise
1664 operation may be configured in more detail, and also turned off
1667 * In .netdev files two new settings RemoteChecksumTx= and
1668 RemoteChecksumRx= are now understood that permit configuring the
1669 remote checksumming logic for VXLAN networks.
1671 * The service manager learnt a new "invocation ID" concept for invoked
1672 services. Each runtime cycle of a service will get a new invocation
1673 ID (a 128bit random UUID) assigned that identifies the current
1674 run of the service uniquely and globally. A new invocation ID
1675 is generated each time a service starts up. The journal will store
1676 the invocation ID of a service along with any logged messages, thus
1677 making the invocation ID useful for matching the online runtime of a
1678 service with the offline log data it generated in a safe way without
1679 relying on synchronized timestamps. In many ways this new service
1680 invocation ID concept is similar to the kernel's boot ID concept that
1681 uniquely and globally identifies the runtime of each boot. The
1682 invocation ID of a service is passed to the service itself via an
1683 environment variable ($INVOCATION_ID). A new bus call
1684 GetUnitByInvocationID() has been added that is similar to GetUnit()
1685 but instead of retrieving the bus path for a unit by its name
1686 retrieves it by its invocation ID. The returned path is valid only as
1687 long as the passed invocation ID is current.
1689 * systemd-resolved gained a new "DNSStubListener" setting in
1690 resolved.conf. It either takes a boolean value or the special values
1691 "udp" and "tcp", and configures whether to enable the stub DNS
1692 listener on 127.0.0.53:53.
1694 * IP addresses configured via networkd may now carry additional
1695 configuration settings supported by the kernel. New options include:
1696 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1697 PrefixRoute=, AutoJoin=.
1699 * The PAM configuration fragment file for "user@.service" shipped with
1700 systemd (i.e. the --user instance of systemd) has been stripped to
1701 the minimum necessary to make the system boot. Previously, it
1702 contained Fedora-specific stanzas that did not apply to other
1703 distributions. It is expected that downstream distributions add
1704 additional configuration lines, matching their needs to this file,
1705 using it only as rough template of what systemd itself needs. Note
1706 that this reduced fragment does not even include an invocation of
1707 pam_limits which most distributions probably want to add, even though
1708 systemd itself does not need it. (There's also the new build time
1709 option --with-pamconfdir=no to disable installation of the PAM
1712 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1713 capability is now also dropped from its set (in addition to
1714 CAP_SYS_MKNOD as before).
1716 * In service unit files it is now possible to connect a specific named
1717 file descriptor with stdin/stdout/stdout of an executed service. The
1718 name may be specified in matching .socket units using the
1719 FileDescriptorName= setting.
1721 * A number of journal settings may now be configured on the kernel
1722 command line. Specifically, the following options are now understood:
1723 systemd.journald.max_level_console=,
1724 systemd.journald.max_level_store=,
1725 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1726 systemd.journald.max_level_wall=.
1728 * "systemctl is-enabled --full" will now show by which symlinks a unit
1729 file is enabled in the unit dependency tree.
1731 * Support for VeraCrypt encrypted partitions has been added to the
1732 "cryptsetup" logic and /etc/crypttab.
1734 * systemd-detect-virt gained support for a new --private-users switch
1735 that checks whether the invoking processes are running inside a user
1736 namespace. Similar, a new special value "private-users" for the
1737 existing ConditionVirtualization= setting has been added, permitting
1738 skipping of specific units in user namespace environments.
1740 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1741 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1742 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1743 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1744 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1745 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1746 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1747 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1748 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1749 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1750 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1751 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1752 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1753 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1754 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1755 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1756 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1757 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1758 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1759 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1760 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1761 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1762 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1763 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1764 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1765 Jędrzejewski-Szmek, Zeal Jagannatha
1767 — Santa Fe, 2016-11-03
1771 * In service units the various ExecXYZ= settings have been extended
1772 with an additional special character as first argument of the
1773 assigned value: if the character '+' is used the specified command
1774 line it will be run with full privileges, regardless of User=,
1775 Group=, CapabilityBoundingSet= and similar options. The effect is
1776 similar to the existing PermissionsStartOnly= option, but allows
1777 configuration of this concept for each executed command line
1780 * Services may now alter the service watchdog timeout at runtime by
1781 sending a WATCHDOG_USEC= message via sd_notify().
1783 * MemoryLimit= and related unit settings now optionally take percentage
1784 specifications. The percentage is taken relative to the amount of
1785 physical memory in the system (or in case of containers, the assigned
1786 amount of memory). This allows scaling service resources neatly with
1787 the amount of RAM available on the system. Similarly, systemd-logind's
1788 RuntimeDirectorySize= option now also optionally takes percentage
1791 * In similar fashion TasksMax= takes percentage values now, too. The
1792 value is taken relative to the configured maximum number of processes
1793 on the system. The per-service task maximum has been changed to 15%
1794 using this functionality. (Effectively this is an increase of 512 →
1795 4915 for service units, given the kernel's default pid_max setting.)
1797 * Calendar time specifications in .timer units now understand a ".."
1798 syntax for time ranges. Example: "4..7:10" may now be used for
1799 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1802 * The InaccessableDirectories=, ReadOnlyDirectories= and
1803 ReadWriteDirectories= unit file settings have been renamed to
1804 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1805 applied to all kinds of file nodes, and not just directories, with
1806 the exception of symlinks. Specifically these settings may now be
1807 used on block and character device nodes, UNIX sockets and FIFOS as
1808 well as regular files. The old names of these settings remain
1809 available for compatibility.
1811 * systemd will now log about all service processes it kills forcibly
1812 (using SIGKILL) because they remained after the clean shutdown phase
1813 of the service completed. This should help identifying services that
1814 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1815 systemd-logind's configuration a similar log message is generated for
1816 processes killed at the end of each session due to this setting.
1818 * systemd will now set the $JOURNAL_STREAM environment variable for all
1819 services whose stdout/stderr are connected to the Journal (which
1820 effectively means by default: all services). The variable contains
1821 the device and inode number of the file descriptor used for
1822 stdout/stderr. This may be used by invoked programs to detect whether
1823 their stdout/stderr is connected to the Journal, in which case they
1824 can switch over to direct Journal communication, thus being able to
1825 pass extended, structured metadata along with their log messages. As
1826 one example, this is now used by glib's logging primitives.
1828 * When using systemd's default tmp.mount unit for /tmp, the mount point
1829 will now be established with the "nosuid" and "nodev" options. This
1830 avoids privilege escalation attacks that put traps and exploits into
1831 /tmp. However, this might cause problems if you e. g. put container
1832 images or overlays into /tmp; if you need this, override tmp.mount's
1833 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1836 * systemd now supports the "memory" cgroup controller also on
1839 * The systemd-cgtop tool now optionally takes a control group path as
1840 command line argument. If specified, the control group list shown is
1841 limited to subgroups of that group.
1843 * The SystemCallFilter= unit file setting gained support for
1844 pre-defined, named system call filter sets. For example
1845 SystemCallFilter=@clock is now an effective way to make all clock
1846 changing-related system calls unavailable to a service. A number of
1847 similar pre-defined groups are defined. Writing system call filters
1848 for system services is simplified substantially with this new
1849 concept. Accordingly, all of systemd's own, long-running services now
1850 enable system call filtering based on this, by default.
1852 * A new service setting MemoryDenyWriteExecute= has been added, taking
1853 a boolean value. If turned on, a service may no longer create memory
1854 mappings that are writable and executable at the same time. This
1855 enhances security for services where this is enabled as it becomes
1856 harder to dynamically write and then execute memory in exploited
1857 service processes. This option has been enabled for all of systemd's
1858 own long-running services.
1860 * A new RestrictRealtime= service setting has been added, taking a
1861 boolean argument. If set the service's processes may no longer
1862 acquire realtime scheduling. This improves security as realtime
1863 scheduling may otherwise be used to easily freeze the system.
1865 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1866 value. This may be used for requesting that the system manager inside
1867 of the container reports start-up completion to nspawn which then
1868 propagates this notification further to the service manager
1869 supervising nspawn itself. A related option NotifyReady= in .nspawn
1870 files has been added too. This functionality allows ordering of the
1871 start-up of multiple containers using the usual systemd ordering
1874 * machinectl gained a new command "stop" that is an alias for
1877 * systemd-resolved gained support for contacting DNS servers on
1878 link-local IPv6 addresses.
1880 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1881 its caches. A method call for requesting the same operation has been
1882 added to the bus API too, and is made available via "systemd-resolve
1885 * systemd-resolve gained a new --status switch. If passed a brief
1886 summary of the used DNS configuration with per-interface information
1889 * resolved.conf gained a new Cache= boolean option, defaulting to
1890 on. If turned off local DNS caching is disabled. This comes with a
1891 performance penalty in particular when DNSSEC is enabled. Note that
1892 resolved disables its internal caching implicitly anyway, when the
1893 configured DNS server is on a host-local IP address such as ::1 or
1894 127.0.0.1, thus automatically avoiding double local caching.
1896 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1897 for DNS requests. This improves compatibility with local programs
1898 that do not use the libc NSS or systemd-resolved's bus APIs for name
1899 resolution. This minimal DNS service is only available to local
1900 programs and does not implement the full DNS protocol, but enough to
1901 cover local DNS clients. A new, static resolv.conf file, listing just
1902 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1903 now recommended to make /etc/resolv.conf a symlink to this file in
1904 order to route all DNS lookups to systemd-resolved, regardless if
1905 done via NSS, the bus API or raw DNS packets. Note that this local
1906 DNS service is not as fully featured as the libc NSS or
1907 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1908 used to deliver link-local address information (as this implies
1909 sending a local interface index along), LLMNR/mDNS support via this
1910 interface is severely restricted. It is thus strongly recommended for
1911 all applications to use the libc NSS API or native systemd-resolved
1914 * systemd-networkd's bridge support learned a new setting
1915 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1916 in .network files has been added for configuring VLAN bridging in
1917 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1919 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1920 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1921 now be acquired without relying on DHCPv6. Two new options
1922 UseDomains= and UseDNS= have been added to configure this behaviour.
1924 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1925 renamed IPv6AcceptRA=, without altering its behaviour. The old
1926 setting name remains available for compatibility reasons.
1928 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1929 Key=, InputKey= and OutputKey=.
1931 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1932 interface configuration.
1934 * "systemctl edit" may now be used to create new unit files by
1935 specifying the --force switch.
1937 * sd-event gained a new function sd_event_get_iteration() for
1938 requesting the current iteration counter of the event loop. It starts
1939 at zero and is increased by one with each event loop iteration.
1941 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1942 file. It can be used in lieu of %systemd_requires in packages which
1943 don't use any systemd functionality and are intended to be installed
1944 in minimal containers without systemd present. This macro provides
1945 ordering dependencies to ensure that if the package is installed in
1946 the same rpm transaction as systemd, systemd will be installed before
1947 the scriptlets for the package are executed, allowing unit presets
1950 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1951 been added to simplify packaging of generators.
1953 * The os-release file gained VERSION_CODENAME field for the
1954 distribution nickname (e.g. VERSION_CODENAME=woody).
1956 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1957 can be set to disable parsing of metadata and the creation
1958 of persistent symlinks for that device.
1960 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1961 to make them available to logged-in users has been reverted.
1963 * Much of the common code of the various systemd components is now
1964 built into an internal shared library libsystemd-shared-231.so
1965 (incorporating the systemd version number in the name, to be updated
1966 with future releases) that the components link to. This should
1967 decrease systemd footprint both in memory during runtime and on
1968 disk. Note that the shared library is not for public use, and is
1969 neither API not ABI stable, but is likely to change with every new
1970 released update. Packagers need to make sure that binaries
1971 linking to libsystemd-shared.so are updated in step with the
1974 * Configuration for "mkosi" is now part of the systemd
1975 repository. mkosi is a tool to easily build legacy-free OS images,
1976 and is available on github: https://github.com/systemd/mkosi. If
1977 "mkosi" is invoked in the build tree a new raw OS image is generated
1978 incorporating the systemd sources currently being worked on and a
1979 clean, fresh distribution installation. The generated OS image may be
1980 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1981 UEFI PC. This functionality is particularly useful to easily test
1982 local changes made to systemd in a pristine, defined environment. See
1983 doc/HACKING for details.
1985 * configure learned the --with-support-url= option to specify the
1986 distribution's bugtracker.
1988 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1989 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1990 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1991 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1992 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1993 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1994 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1995 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1996 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1997 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1998 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1999 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2000 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2001 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2002 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2003 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2004 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2005 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2006 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2008 — Berlin, 2016-07-25
2012 * DNSSEC is now turned on by default in systemd-resolved (in
2013 "allow-downgrade" mode), but may be turned off during compile time by
2014 passing "--with-default-dnssec=no" to "configure" (and of course,
2015 during runtime with DNSSEC= in resolved.conf). We recommend
2016 downstreams to leave this on at least during development cycles and
2017 report any issues with the DNSSEC logic upstream. We are very
2018 interested in collecting feedback about the DNSSEC validator and its
2019 limitations in the wild. Note however, that DNSSEC support is
2020 probably nothing downstreams should turn on in stable distros just
2021 yet, as it might create incompatibilities with a few DNS servers and
2022 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2023 automatically whenever we detect such incompatible setups, but there
2024 might be systems we do not cover yet. Hence: please help us testing
2025 the DNSSEC code, leave this on where you can, report back, but then
2026 again don't consider turning this on in your stable, LTS or
2027 production release just yet. (Note that you have to enable
2028 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2029 and its DNSSEC mode for host name resolution from local
2032 * systemd-resolve conveniently resolves DANE records with the --tlsa
2033 option and OPENPGPKEY records with the --openpgp option. It also
2034 supports dumping raw DNS record data via the new --raw= switch.
2036 * systemd-logind will now by default terminate user processes that are
2037 part of the user session scope unit (session-XX.scope) when the user
2038 logs out. This behavior is controlled by the KillUserProcesses=
2039 setting in logind.conf, and the previous default of "no" is now
2040 changed to "yes". This means that user sessions will be properly
2041 cleaned up after, but additional steps are necessary to allow
2042 intentionally long-running processes to survive logout.
2044 While the user is logged in at least once, user@.service is running,
2045 and any service that should survive the end of any individual login
2046 session can be started at a user service or scope using systemd-run.
2047 systemd-run(1) man page has been extended with an example which shows
2048 how to run screen in a scope unit underneath user@.service. The same
2049 command works for tmux.
2051 After the user logs out of all sessions, user@.service will be
2052 terminated too, by default, unless the user has "lingering" enabled.
2053 To effectively allow users to run long-term tasks even if they are
2054 logged out, lingering must be enabled for them. See loginctl(1) for
2055 details. The default polkit policy was modified to allow users to
2056 set lingering for themselves without authentication.
2058 Previous defaults can be restored at compile time by the
2059 --without-kill-user-processes option to "configure".
2061 * systemd-logind gained new configuration settings SessionsMax= and
2062 InhibitorsMax=, both with a default of 8192. It will not register new
2063 user sessions or inhibitors above this limit.
2065 * systemd-logind will now reload configuration on SIGHUP.
2067 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2068 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2069 enable. Also, support for the "io" cgroup controller in the unified
2070 hierarchy has been added, so that the "memory", "pids" and "io" are
2071 now the controllers that are supported on the unified hierarchy.
2073 WARNING: it is not possible to use previous systemd versions with
2074 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2075 is necessary to also update systemd in the initramfs if using the
2076 unified hierarchy. An updated SELinux policy is also required.
2078 * LLDP support has been extended, and both passive (receive-only) and
2079 active (sender) modes are supported. Passive mode ("routers-only") is
2080 enabled by default in systemd-networkd. Active LLDP mode is enabled
2081 by default for containers on the internal network. The "networkctl
2082 lldp" command may be used to list information gathered. "networkctl
2083 status" will also show basic LLDP information on connected peers now.
2085 * The IAID and DUID unique identifier sent in DHCP requests may now be
2086 configured for the system and each .network file managed by
2087 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2089 * systemd-networkd gained support for configuring proxy ARP support for
2090 each interface, via the ProxyArp= setting in .network files. It also
2091 gained support for configuring the multicast querier feature of
2092 bridge devices, via the new MulticastQuerier= setting in .netdev
2093 files. Similarly, snooping on the IGMP traffic can be controlled
2094 via the new setting MulticastSnooping=.
2096 A new setting PreferredLifetime= has been added for addresses
2097 configured in .network file to configure the lifetime intended for an
2100 The systemd-networkd DHCP server gained the option EmitRouter=, which
2101 defaults to yes, to configure whether the DHCP Option 3 (Router)
2104 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2105 systemd-socket-activate and installed into /usr/bin. It is now fully
2108 * systemd-journald now uses separate threads to flush changes to disk
2109 when closing journal files, thus reducing impact of slow disk I/O on
2110 logging performance.
2112 * The sd-journal API gained two new calls
2113 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2114 can be used to open journal files using file descriptors instead of
2115 file or directory paths. sd_journal_open_container() has been
2116 deprecated, sd_journal_open_directory_fd() should be used instead
2117 with the flag SD_JOURNAL_OS_ROOT.
2119 * journalctl learned a new output mode "-o short-unix" that outputs log
2120 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2121 UTC). It also gained support for a new --no-hostname setting to
2122 suppress the hostname column in the family of "short" output modes.
2124 * systemd-ask-password now optionally skips printing of the password to
2125 stdout with --no-output which can be useful in scripts.
2127 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2128 (devices tagged with ID_MAKER_TOOL) are now tagged with
2129 "uaccess" and are available to logged in users.
2131 * The DeviceAllow= unit setting now supports specifiers (with "%").
2133 * "systemctl show" gained a new --value switch, which allows print a
2134 only the contents of a specific unit property, without also printing
2135 the property's name. Similar support was added to "show*" verbs
2136 of loginctl and machinectl that output "key=value" lists.
2138 * A new unit type "generated" was added for files dynamically generated
2139 by generator tools. Similarly, a new unit type "transient" is used
2140 for unit files created using the runtime API. "systemctl enable" will
2141 refuse to operate on such files.
2143 * A new command "systemctl revert" has been added that may be used to
2144 revert to the vendor version of a unit file, in case local changes
2145 have been made by adding drop-ins or overriding the unit file.
2147 * "machinectl clean" gained a new verb to automatically remove all or
2148 just hidden container images.
2150 * systemd-tmpfiles gained support for a new line type "e" for emptying
2151 directories, if they exist, without creating them if they don't.
2153 * systemd-nspawn gained support for automatically patching the UID/GIDs
2154 of the owners and the ACLs of all files and directories in a
2155 container tree to match the UID/GID user namespacing range selected
2156 for the container invocation. This mode is enabled via the new
2157 --private-users-chown switch. It also gained support for
2158 automatically choosing a free, previously unused UID/GID range when
2159 starting a container, via the new --private-users=pick setting (which
2160 implies --private-users-chown). Together, these options for the first
2161 time make user namespacing for nspawn containers fully automatic and
2162 thus deployable. The systemd-nspawn@.service template unit file has
2163 been changed to use this functionality by default.
2165 * systemd-nspawn gained a new --network-zone= switch, that allows
2166 creating ad-hoc virtual Ethernet links between multiple containers,
2167 that only exist as long as at least one container referencing them is
2168 running. This allows easy connecting of multiple containers with a
2169 common link that implements an Ethernet broadcast domain. Each of
2170 these network "zones" may be named relatively freely by the user, and
2171 may be referenced by any number of containers, but each container may
2172 only reference one of these "zones". On the lower level, this is
2173 implemented by an automatically managed bridge network interface for
2174 each zone, that is created when the first container referencing its
2175 zone is created and removed when the last one referencing its zone
2178 * The default start timeout may now be configured on the kernel command
2179 line via systemd.default_timeout_start_sec=. It was already
2180 configurable via the DefaultTimeoutStartSec= option in
2181 /etc/systemd/system.conf.
2183 * Socket units gained a new TriggerLimitIntervalSec= and
2184 TriggerLimitBurst= setting to configure a limit on the activation
2185 rate of the socket unit.
2187 * The LimitNICE= setting now optionally takes normal UNIX nice values
2188 in addition to the raw integer limit value. If the specified
2189 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2190 value is understood as UNIX nice value. If not prefixed like this it
2191 is understood as raw RLIMIT_NICE limit.
2193 * Note that the effect of the PrivateDevices= unit file setting changed
2194 slightly with this release: the per-device /dev file system will be
2195 mounted read-only from this version on, and will have "noexec"
2196 set. This (minor) change of behavior might cause some (exceptional)
2197 legacy software to break, when PrivateDevices=yes is set for its
2198 service. Please leave PrivateDevices= off if you run into problems
2201 * systemd-bootchart has been split out to a separate repository:
2202 https://github.com/systemd/systemd-bootchart
2204 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2205 merged into the kernel in its current form.
2207 * The compatibility libraries libsystemd-daemon.so,
2208 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2209 which have been deprecated since systemd-209 have been removed along
2210 with the corresponding pkg-config files. All symbols provided by
2211 those libraries are provided by libsystemd.so.
2213 * The Capabilities= unit file setting has been removed (it is ignored
2214 for backwards compatibility). AmbientCapabilities= and
2215 CapabilityBoundingSet= should be used instead.
2217 * A new special target has been added, initrd-root-device.target,
2218 which creates a synchronization point for dependencies of the root
2219 device in early userspace. Initramfs builders must ensure that this
2220 target is now included in early userspace.
2222 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2223 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2224 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2225 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2226 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2227 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2228 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2229 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2230 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2231 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2232 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2233 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2234 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2235 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2236 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2237 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2238 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2239 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2240 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2241 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2242 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2243 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2244 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2245 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2246 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2249 — Fairfax, 2016-05-21
2253 * The systemd-resolved DNS resolver service has gained a substantial
2254 set of new features, most prominently it may now act as a DNSSEC
2255 validating stub resolver. DNSSEC mode is currently turned off by
2256 default, but is expected to be turned on by default in one of the
2257 next releases. For now, we invite everybody to test the DNSSEC logic
2258 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2259 service also gained a full set of D-Bus interfaces, including calls
2260 to configure DNS and DNSSEC settings per link (for use by external
2261 network management software). systemd-resolved and systemd-networkd
2262 now distinguish between "search" and "routing" domains. The former
2263 are used to qualify single-label names, the latter are used purely
2264 for routing lookups within certain domains to specific links.
2265 resolved now also synthesizes RRs for all entries from /etc/hosts.
2267 * The systemd-resolve tool (which is a client utility for
2268 systemd-resolved) has been improved considerably and is now fully
2269 supported and documented. Hence it has moved from /usr/lib/systemd to
2272 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2275 * The coredump collection logic has been reworked: when a coredump is
2276 collected it is now written to disk, compressed and processed
2277 (including stacktrace extraction) from a new instantiated service
2278 systemd-coredump@.service, instead of directly from the
2279 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2280 processing large coredumps can take up a substantial amount of
2281 resources and time, and this previously happened entirely outside of
2282 systemd's service supervision. With the new logic the core_pattern
2283 hook only does minimal metadata collection before passing off control
2284 to the new instantiated service, which is configured with a time
2285 limit, a nice level and other settings to minimize negative impact on
2286 the rest of the system. Also note that the new logic will honour the
2287 RLIMIT_CORE setting of the crashed process, which now allows users
2288 and processes to turn off coredumping for their processes by setting
2291 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2292 and all forked processes by default. Previously, PID 1 would leave
2293 the setting at "0" for all processes, as set by the kernel. Note that
2294 the resource limit traditionally has no effect on the generated
2295 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2296 logic is used. Since the limit is now honoured (see above) its
2297 default has been changed so that the coredumping logic is enabled by
2298 default for all processes, while allowing specific opt-out.
2300 * When the stacktrace is extracted from processes of system users, this
2301 is now done as "systemd-coredump" user, in order to sandbox this
2302 potentially security sensitive parsing operation. (Note that when
2303 processing coredumps of normal users this is done under the user ID
2304 of process that crashed, as before.) Packagers should take notice
2305 that it is now necessary to create the "systemd-coredump" system user
2306 and group at package installation time.
2308 * The systemd-activate socket activation testing tool gained support
2309 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2310 and --seqpacket switches. It also has been extended to support both
2311 new-style and inetd-style file descriptor passing. Use the new
2312 --inetd switch to request inetd-style file descriptor passing.
2314 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2315 variable, which takes a boolean value. If set to false, ANSI color
2316 output is disabled in the tools even when run on a terminal that
2319 * The VXLAN support in networkd now supports two new settings
2320 DestinationPort= and PortRange=.
2322 * A new systemd.machine_id= kernel command line switch has been added,
2323 that may be used to set the machine ID in /etc/machine-id if it is
2324 not initialized yet. This command line option has no effect if the
2325 file is already initialized.
2327 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2328 specified command line as PID 2 rather than PID 1 in the
2329 container. In this mode PID 1 is a minimal stub init process that
2330 implements the special POSIX and Linux semantics of PID 1 regarding
2331 signal and child process management. Note that this stub init process
2332 is implemented in nspawn itself and requires no support from the
2333 container image. This new logic is useful to support running
2334 arbitrary commands in the container, as normal processes are
2335 generally not prepared to run as PID 1.
2337 * systemd-nspawn gained a new --chdir= switch for setting the current
2338 working directory for the process started in the container.
2340 * "journalctl /dev/sda" will now output all kernel log messages for
2341 specified device from the current boot, in addition to all devices
2342 that are parents of it. This should make log output about devices
2343 pretty useful, as long as kernel drivers attach enough metadata to
2344 the log messages. (The usual SATA drivers do.)
2346 * The sd-journal API gained two new calls
2347 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2348 that report whether log data from /run or /var has been found.
2350 * journalctl gained a new switch "--fields" that prints all journal
2351 record field names currently in use in the journal. This is backed
2352 by two new sd-journal API calls sd_journal_enumerate_fields() and
2353 sd_journal_restart_fields().
2355 * Most configurable timeouts in systemd now expect an argument of
2356 "infinity" to turn them off, instead of "0" as before. The semantics
2357 from now on is that a timeout of "0" means "now", and "infinity"
2358 means "never". To maintain backwards compatibility, "0" continues to
2359 turn off previously existing timeout settings.
2361 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2362 try-reload-or-restart" to clarify what it actually does: the "try"
2363 logic applies to both reloading and restarting, not just restarting.
2364 The old name continues to be accepted for compatibility.
2366 * On boot-up, when PID 1 detects that the system clock is behind the
2367 release date of the systemd version in use, the clock is now set
2368 to the latter. Previously, this was already done in timesyncd, in order
2369 to avoid running with clocks set to the various clock epochs such as
2370 1902, 1938 or 1970. With this change the logic is now done in PID 1
2371 in addition to timesyncd during early boot-up, so that it is enforced
2372 before the first process is spawned by systemd. Note that the logic
2373 in timesyncd remains, as it is more comprehensive and ensures
2374 clock monotonicity by maintaining a persistent timestamp file in
2375 /var. Since /var is generally not available in earliest boot or the
2376 initrd, this part of the logic remains in timesyncd, and is not done
2379 * Support for tweaking details in net_cls.class_id through the
2380 NetClass= configuration directive has been removed, as the kernel
2381 people have decided to deprecate that controller in cgroup v2.
2382 Userspace tools such as nftables are moving over to setting rules
2383 that are specific to the full cgroup path of a task, which obsoletes
2384 these controllers anyway. The NetClass= directive is kept around for
2385 legacy compatibility reasons. For a more in-depth description of the
2386 kernel change, please refer to the respective upstream commit:
2388 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2390 * A new service setting RuntimeMaxSec= has been added that may be used
2391 to specify a maximum runtime for a service. If the timeout is hit, the
2392 service is terminated and put into a failure state.
2394 * A new service setting AmbientCapabilities= has been added. It allows
2395 configuration of additional Linux process capabilities that are
2396 passed to the activated processes. This is only available on very
2399 * The process resource limit settings in service units may now be used
2400 to configure hard and soft limits individually.
2402 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2403 expose support for gcc's __attribute__((cleanup())) C extension.
2404 Specifically, for many object destructor functions alternative
2405 versions have been added that have names suffixed with "p" and take a
2406 pointer to a pointer to the object to destroy, instead of just a
2407 pointer to the object itself. This is useful because these destructor
2408 functions may be used directly as parameters to the cleanup
2409 construct. Internally, systemd has been a heavy user of this GCC
2410 extension for a long time, and with this change similar support is
2411 now available to consumers of the library outside of systemd. Note
2412 that by using this extension in your sources compatibility with old
2413 and strictly ANSI compatible C compilers is lost. However, all gcc or
2414 LLVM versions of recent years support this extension.
2416 * Timer units gained support for a new setting RandomizedDelaySec= that
2417 allows configuring some additional randomized delay to the configured
2418 time. This is useful to spread out timer events to avoid load peaks in
2419 clusters or larger setups.
2421 * Calendar time specifications now support sub-second accuracy.
2423 * Socket units now support listening on SCTP and UDP-lite protocol
2426 * The sd-event API now comes with a full set of man pages.
2428 * Older versions of systemd contained experimental support for
2429 compressing journal files and coredumps with the LZ4 compressor that
2430 was not compatible with the lz4 binary (due to API limitations of the
2431 lz4 library). This support has been removed; only support for files
2432 compatible with the lz4 binary remains. This LZ4 logic is now
2433 officially supported and no longer considered experimental.
2435 * The dkr image import logic has been removed again from importd. dkr's
2436 micro-services focus doesn't fit into the machine image focus of
2437 importd, and quickly got out of date with the upstream dkr API.
2439 * Creation of the /run/lock/lockdev/ directory was dropped from
2440 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2441 been available for many years. If you still need this, you need to
2442 create your own tmpfiles.d config file with:
2444 d /run/lock/lockdev 0775 root lock -
2446 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2447 and RebootArgument= have been moved from the [Service] section of
2448 unit files to [Unit], and they are now supported on all unit types,
2449 not just service units. Of course, elogind will continue to
2450 understand these settings also at the old location, in order to
2451 maintain compatibility.
2453 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2454 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2455 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2456 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2457 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2458 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2459 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2460 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2461 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2462 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2463 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2464 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2465 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2466 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2467 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2468 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2469 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2470 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2471 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2473 — Berlin, 2016-02-11
2477 * A number of properties previously only settable in unit
2478 files are now also available as properties to set when
2479 creating transient units programmatically via the bus, as it
2480 is exposed with systemd-run's --property=
2481 setting. Specifically, these are: SyslogIdentifier=,
2482 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2483 EnvironmentFile=, ReadWriteDirectories=,
2484 ReadOnlyDirectories=, InaccessibleDirectories=,
2485 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2487 * When creating transient services via the bus API it is now
2488 possible to pass in a set of file descriptors to use as
2489 STDIN/STDOUT/STDERR for the invoked process.
2491 * Slice units may now be created transiently via the bus APIs,
2492 similar to the way service and scope units may already be
2493 created transiently.
2495 * Wherever systemd expects a calendar timestamp specification
2496 (like in journalctl's --since= and --until= switches) UTC
2497 timestamps are now supported. Timestamps suffixed with "UTC"
2498 are now considered to be in Universal Time Coordinated
2499 instead of the local timezone. Also, timestamps may now
2500 optionally be specified with sub-second accuracy. Both of
2501 these additions also apply to recurring calendar event
2502 specification, such as OnCalendar= in timer units.
2504 * journalctl gained a new "--sync" switch that asks the
2505 journal daemon to write all so far unwritten log messages to
2506 disk and sync the files, before returning.
2508 * systemd-tmpfiles learned two new line types "q" and "Q" that
2509 operate like "v", but also set up a basic btrfs quota
2510 hierarchy when used on a btrfs file system with quota
2513 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2514 instead of a subvolume (even on a btrfs file system) if the
2515 root directory is a plain directory, and not a
2516 subvolume. This should simplify things with certain chroot()
2517 environments which are not aware of the concept of btrfs
2520 * systemd-detect-virt gained a new --chroot switch to detect
2521 whether execution takes place in a chroot() environment.
2523 * CPUAffinity= now takes CPU index ranges in addition to
2526 * The various memory-related resource limit settings (such as
2527 LimitAS=) now understand the usual K, M, G, ... suffixes to
2528 the base of 1024 (IEC). Similar, the time-related resource
2529 limit settings understand the usual min, h, day, ...
2532 * There's a new system.conf setting DefaultTasksMax= to
2533 control the default TasksMax= setting for services and
2534 scopes running on the system. (TasksMax= is the primary
2535 setting that exposes the "pids" cgroup controller on systemd
2536 and was introduced in the previous systemd release.) The
2537 setting now defaults to 512, which means services that are
2538 not explicitly configured otherwise will only be able to
2539 create 512 processes or threads at maximum, from this
2540 version on. Note that this means that thread- or
2541 process-heavy services might need to be reconfigured to set
2542 TasksMax= to a higher value. It is sufficient to set
2543 TasksMax= in these specific unit files to a higher value, or
2544 even "infinity". Similar, there's now a logind.conf setting
2545 UserTasksMax= that defaults to 4096 and limits the total
2546 number of processes or tasks each user may own
2547 concurrently. nspawn containers also have the TasksMax=
2548 value set by default now, to 8192. Note that all of this
2549 only has an effect if the "pids" cgroup controller is
2550 enabled in the kernel. The general benefit of these changes
2551 should be a more robust and safer system, that provides a
2552 certain amount of per-service fork() bomb protection.
2554 * systemd-nspawn gained the new --network-veth-extra= switch
2555 to define additional and arbitrarily-named virtual Ethernet
2556 links between the host and the container.
2558 * A new service execution setting PassEnvironment= has been
2559 added that allows importing select environment variables
2560 from PID1's environment block into the environment block of
2563 * Timer units gained support for a new RemainAfterElapse=
2564 setting which takes a boolean argument. It defaults to on,
2565 exposing behaviour unchanged to previous releases. If set to
2566 off, timer units are unloaded after they elapsed if they
2567 cannot elapse again. This is particularly useful for
2568 transient timer units, which shall not stay around longer
2569 than until they first elapse.
2571 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2572 default now (the kernel default is 16). This is beneficial
2573 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2574 allows substantially larger numbers of queued
2575 datagrams. This should increase the capability of systemd to
2576 parallelize boot-up, as logging and sd_notify() are unlikely
2577 to stall execution anymore. If you need to change the value
2578 from the new defaults, use the usual sysctl.d/ snippets.
2580 * The compression framing format used by the journal or
2581 coredump processing has changed to be in line with what the
2582 official LZ4 tools generate. LZ4 compression support in
2583 systemd was considered unsupported previously, as the format
2584 was not compatible with the normal tools. With this release
2585 this has changed now, and it is hence safe for downstream
2586 distributions to turn it on. While not compressing as well
2587 as the XZ, LZ4 is substantially faster, which makes
2588 it a good default choice for the compression logic in the
2589 journal and in coredump handling.
2591 * Any reference to /etc/mtab has been dropped from
2592 systemd. The file has been obsolete since a while, but
2593 systemd refused to work on systems where it was incorrectly
2594 set up (it should be a symlink or non-existent). Please make
2595 sure to update to util-linux 2.27.1 or newer in conjunction
2596 with this systemd release, which also drops any reference to
2597 /etc/mtab. If you maintain a distribution make sure that no
2598 software you package still references it, as this is a
2599 likely source of bugs. There's also a glibc bug pending,
2600 asking for removal of any reference to this obsolete file:
2602 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2604 Note that only util-linux versions built with
2605 --enable-libmount-force-mountinfo are supported.
2607 * Support for the ".snapshot" unit type has been removed. This
2608 feature turned out to be little useful and little used, and
2609 has now been removed from the core and from systemctl.
2611 * The dependency types RequiresOverridable= and
2612 RequisiteOverridable= have been removed from systemd. They
2613 have been used only very sparingly to our knowledge and
2614 other options that provide a similar effect (such as
2615 systemctl --mode=ignore-dependencies) are much more useful
2616 and commonly used. Moreover, they were only half-way
2617 implemented as the option to control behaviour regarding
2618 these dependencies was never added to systemctl. By removing
2619 these dependency types the execution engine becomes a bit
2620 simpler. Unit files that use these dependencies should be
2621 changed to use the non-Overridable dependency types
2622 instead. In fact, when parsing unit files with these
2623 options, that's what systemd will automatically convert them
2624 too, but it will also warn, asking users to fix the unit
2625 files accordingly. Removal of these dependency types should
2626 only affect a negligible number of unit files in the wild.
2628 * Behaviour of networkd's IPForward= option changed
2629 (again). It will no longer maintain a per-interface setting,
2630 but propagate one way from interfaces where this is enabled
2631 to the global kernel setting. The global setting will be
2632 enabled when requested by a network that is set up, but
2633 never be disabled again. This change was made to make sure
2634 IPv4 and IPv6 behaviour regarding packet forwarding is
2635 similar (as the Linux IPv6 stack does not support
2636 per-interface control of this setting) and to minimize
2639 * In unit files the behaviour of %u, %U, %h, %s has
2640 changed. These specifiers will now unconditionally resolve
2641 to the various user database fields of the user that the
2642 systemd instance is running as, instead of the user
2643 configured in the specific unit via User=. Note that this
2644 effectively doesn't change much, as resolving of these
2645 specifiers was already turned off in the --system instance
2646 of systemd, as we cannot do NSS lookups from PID 1. In the
2647 --user instance of systemd these specifiers where correctly
2648 resolved, but hardly made any sense, since the user instance
2649 lacks privileges to do user switches anyway, and User= is
2650 hence useless. Moreover, even in the --user instance of
2651 systemd behaviour was awkward as it would only take settings
2652 from User= assignment placed before the specifier into
2653 account. In order to unify and simplify the logic around
2654 this the specifiers will now always resolve to the
2655 credentials of the user invoking the manager (which in case
2656 of PID 1 is the root user).
2658 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2659 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2660 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2661 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2662 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2663 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2664 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2665 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2666 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2667 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2668 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2669 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2670 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2671 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2674 — Berlin, 2015-11-18
2678 * systemd now depends on util-linux v2.27. More specifically,
2679 the newly added mount monitor feature in libmount now
2680 replaces systemd's former own implementation.
2682 * libmount mandates /etc/mtab not to be regular file, and
2683 systemd now enforces this condition at early boot.
2684 /etc/mtab has been deprecated and warned about for a very
2685 long time, so systems running systemd should already have
2686 stopped having this file around as anything else than a
2687 symlink to /proc/self/mounts.
2689 * Support for the "pids" cgroup controller has been added. It
2690 allows accounting the number of tasks in a cgroup and
2691 enforcing limits on it. This adds two new setting
2692 TasksAccounting= and TasksMax= to each unit, as well as a
2693 global option DefaultTasksAccounting=.
2695 * Support for the "net_cls" cgroup controller has been added.
2696 It allows assigning a net class ID to each task in the
2697 cgroup, which can then be used in firewall rules and traffic
2698 shaping configurations. Note that the kernel netfilter net
2699 class code does not currently work reliably for ingress
2700 packets on unestablished sockets.
2702 This adds a new config directive called NetClass= to CGroup
2703 enabled units. Allowed values are positive numbers for fixed
2704 assignments and "auto" for picking a free value
2707 * 'systemctl is-system-running' now returns 'offline' if the
2708 system is not booted with systemd. This command can now be
2709 used as a substitute for 'systemd-notify --booted'.
2711 * Watchdog timeouts have been increased to 3 minutes for all
2712 in-tree service files. Apparently, disk IO issues are more
2713 frequent than we hoped, and user reported >1 minute waiting
2716 * 'machine-id-commit' functionality has been merged into
2717 'machine-id-setup --commit'. The separate binary has been
2720 * The WorkingDirectory= directive in unit files may now be set
2721 to the special value '~'. In this case, the working
2722 directory is set to the home directory of the user
2723 configured in User=.
2725 * "machinectl shell" will now open the shell in the home
2726 directory of the selected user by default.
2728 * The CrashChVT= configuration file setting is renamed to
2729 CrashChangeVT=, following our usual logic of not
2730 abbreviating unnecessarily. The old directive is still
2731 supported for compat reasons. Also, this directive now takes
2732 an integer value between 1 and 63, or a boolean value. The
2733 formerly supported '-1' value for disabling stays around for
2736 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2737 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2738 RootDirectory= properties can now be set for transient
2741 * The systemd-analyze tool gained a new "set-log-target" verb
2742 to change the logging target the system manager logs to
2743 dynamically during runtime. This is similar to how
2744 "systemd-analyze set-log-level" already changes the log
2747 * In nspawn /sys is now mounted as tmpfs, with only a selected
2748 set of subdirectories mounted in from the real sysfs. This
2749 enhances security slightly, and is useful for ensuring user
2750 namespaces work correctly.
2752 * Support for USB FunctionFS activation has been added. This
2753 allows implementation of USB gadget services that are
2754 activated as soon as they are requested, so that they don't
2755 have to run continuously, similar to classic socket
2758 * The "systemctl exit" command now optionally takes an
2759 additional parameter that sets the exit code to return from
2760 the systemd manager when exiting. This is only relevant when
2761 running the systemd user instance, or when running the
2762 system instance in a container.
2764 * sd-bus gained the new API calls sd_bus_path_encode_many()
2765 and sd_bus_path_decode_many() that allow easy encoding and
2766 decoding of multiple identifier strings inside a D-Bus
2767 object path. Another new call sd_bus_default_flush_close()
2768 has been added to flush and close per-thread default
2771 * systemd-cgtop gained support for a -M/--machine= switch to
2772 show the control groups within a certain container only.
2774 * "systemctl kill" gained support for an optional --fail
2775 switch. If specified the requested operation will fail of no
2776 processes have been killed, because the unit had no
2777 processes attached, or similar.
2779 * A new systemd.crash_reboot=1 kernel command line option has
2780 been added that triggers a reboot after crashing. This can
2781 also be set through CrashReboot= in systemd.conf.
2783 * The RuntimeDirectory= setting now understands unit
2784 specifiers like %i or %f.
2786 * A new (still internal) library API sd-ipv4acd has been added,
2787 that implements address conflict detection for IPv4. It's
2788 based on code from sd-ipv4ll, and will be useful for
2789 detecting DHCP address conflicts.
2791 * File descriptors passed during socket activation may now be
2792 named. A new API sd_listen_fds_with_names() is added to
2793 access the names. The default names may be overridden,
2794 either in the .socket file using the FileDescriptorName=
2795 parameter, or by passing FDNAME= when storing the file
2796 descriptors using sd_notify().
2798 * systemd-networkd gained support for:
2800 - Setting the IPv6 Router Advertisement settings via
2801 IPv6AcceptRouterAdvertisements= in .network files.
2803 - Configuring the HelloTimeSec=, MaxAgeSec= and
2804 ForwardDelaySec= bridge parameters in .netdev files.
2806 - Configuring PreferredSource= for static routes in
2809 * The "ask-password" framework used to query for LUKS harddisk
2810 passwords or SSL passwords during boot gained support for
2811 caching passwords in the kernel keyring, if it is
2812 available. This makes sure that the user only has to type in
2813 a passphrase once if there are multiple objects to unlock
2814 with the same one. Previously, such password caching was
2815 available only when Plymouth was used; this moves the
2816 caching logic into the systemd codebase itself. The
2817 "systemd-ask-password" utility gained a new --keyname=
2818 switch to control which kernel keyring key to use for
2819 caching a password in. This functionality is also useful for
2820 enabling display managers such as gdm to automatically
2821 unlock the user's GNOME keyring if its passphrase, the
2822 user's password and the harddisk password are the same, if
2823 gdm-autologin is used.
2825 * When downloading tar or raw images using "machinectl
2826 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2827 file is now also downloaded, if it is available and stored
2828 next to the image file.
2830 * Units of type ".socket" gained a new boolean setting
2831 Writable= which is only useful in conjunction with
2832 ListenSpecial=. If true, enables opening the specified
2833 special file in O_RDWR mode rather than O_RDONLY mode.
2835 * systemd-rfkill has been reworked to become a singleton
2836 service that is activated through /dev/rfkill on each rfkill
2837 state change and saves the settings to disk. This way,
2838 systemd-rfkill is now compatible with devices that exist
2839 only intermittendly, and even restores state if the previous
2840 system shutdown was abrupt rather than clean.
2842 * The journal daemon gained support for vacuuming old journal
2843 files controlled by the number of files that shall remain,
2844 in addition to the already existing control by size and by
2845 date. This is useful as journal interleaving performance
2846 degrades with too many separate journal files, and allows
2847 putting an effective limit on them. The new setting defaults
2848 to 100, but this may be changed by setting SystemMaxFiles=
2849 and RuntimeMaxFiles= in journald.conf. Also, the
2850 "journalctl" tool gained the new --vacuum-files= switch to
2851 manually vacuum journal files to leave only the specified
2852 number of files in place.
2854 * udev will now create /dev/disk/by-path links for ATA devices
2855 on kernels where that is supported.
2857 * Galician, Serbian, Turkish and Korean translations were added.
2859 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2860 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2861 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2862 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2863 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2864 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2865 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2866 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2867 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2868 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2869 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2870 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2871 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2872 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2873 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2874 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2875 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2876 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2878 — Berlin, 2015-10-07
2882 * The DHCP implementation of systemd-networkd gained a set of
2885 - The DHCP server now supports emitting DNS and NTP
2886 information. It may be enabled and configured via
2887 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2888 and NTP information is enabled, but no servers are
2889 configured, the corresponding uplink information (if there
2890 is any) is propagated.
2892 - Server and client now support transmission and reception
2893 of timezone information. It can be configured via the
2894 newly introduced network options UseTimezone=,
2895 EmitTimezone=, and Timezone=. Transmission of timezone
2896 information is enabled between host and containers by
2897 default now: the container will change its local timezone
2898 to what the host has set.
2900 - Lease timeouts can now be configured via
2901 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2903 - The DHCP server improved on the stability of
2904 leases. Clients are more likely to get the same lease
2905 information back, even if the server loses state.
2907 - The DHCP server supports two new configuration options to
2908 control the lease address pool metrics, PoolOffset= and
2911 * The encapsulation limit of tunnels in systemd-networkd may
2912 now be configured via 'EncapsulationLimit='. It allows
2913 modifying the maximum additional levels of encapsulation
2914 that are permitted to be prepended to a packet.
2916 * systemd now supports the concept of user buses replacing
2917 session buses, if used with dbus-1.10 (and enabled via dbus
2918 --enable-user-session). It previously only supported this on
2919 kdbus-enabled systems, and this release expands this to
2920 'dbus-daemon' systems.
2922 * systemd-networkd now supports predictable interface names
2925 * systemd now optionally supports the new Linux kernel
2926 "unified" control group hierarchy. If enabled via the kernel
2927 command-line option 'systemd.unified_cgroup_hierarchy=1',
2928 systemd will try to mount the unified cgroup hierarchy
2929 directly on /sys/fs/cgroup. If not enabled, or not
2930 available, systemd will fall back to the legacy cgroup
2931 hierarchy setup, as before. Host system and containers can
2932 mix and match legacy and unified hierarchies as they
2933 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2934 environment variable to individually select the hierarchy to
2935 use for executed containers. By default, nspawn will use the
2936 unified hierarchy for the containers if the host uses the
2937 unified hierarchy, and the legacy hierarchy otherwise.
2938 Please note that at this point the unified hierarchy is an
2939 experimental kernel feature and is likely to change in one
2940 of the next kernel releases. Therefore, it should not be
2941 enabled by default in downstream distributions yet. The
2942 minimum required kernel version for the unified hierarchy to
2943 work is 4.2. Note that when the unified hierarchy is used
2944 for the first time delegated access to controllers is
2945 safe. Because of this systemd-nspawn containers will get
2946 access to controllers now, as will systemd user
2947 sessions. This means containers and user sessions may now
2948 manage their own resources, partitioning up what the system
2951 * A new special scope unit "init.scope" has been introduced
2952 that encapsulates PID 1 of the system. It may be used to
2953 determine resource usage and enforce resource limits on PID
2954 1 itself. PID 1 hence moved out of the root of the control
2957 * The cgtop tool gained support for filtering out kernel
2958 threads when counting tasks in a control group. Also, the
2959 count of processes is now recursively summed up by
2960 default. Two options -k and --recursive= have been added to
2961 revert to old behaviour. The tool has also been updated to
2962 work correctly in containers now.
2964 * systemd-nspawn's --bind= and --bind-ro= options have been
2965 extended to allow creation of non-recursive bind mounts.
2967 * libsystemd gained two new calls sd_pid_get_cgroup() and
2968 sd_peer_get_cgroup() which return the control group path of
2969 a process or peer of a connected AF_UNIX socket. This
2970 function call is particularly useful when implementing
2971 delegated subtrees support in the control group hierarchy.
2973 * The "sd-event" event loop API of libsystemd now supports
2974 correct dequeuing of real-time signals, without losing
2977 * When systemd requests a PolicyKit decision when managing
2978 units it will now add additional fields to the request,
2979 including unit name and desired operation. This enables more
2980 powerful PolicyKit policies, that make decisions depending
2981 on these parameters.
2983 * nspawn learnt support for .nspawn settings files, that may
2984 accompany the image files or directories of containers, and
2985 may contain additional settings for the container. This is
2986 an alternative to configuring container parameters via the
2987 nspawn command line.
2989 Contributions from: Cristian Rodríguez, Daniel Mack, David
2990 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2991 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2992 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2993 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2994 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2995 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2996 Andersen, Tom Gundersen, Torstein Husebø
2998 — Berlin, 2015-09-08
3002 * machinectl gained a new verb 'shell' which opens a fresh
3003 shell on the target container or the host. It is similar to
3004 the existing 'login' command of machinectl, but spawns the
3005 shell directly without prompting for username or
3006 password. The pseudo machine '.host' now refers to the local
3007 host and is used by default. Hence, 'machinectl shell' can
3008 be used as replacement for 'su -' which spawns a session as
3009 a fresh systemd unit in a way that is fully isolated from
3010 the originating session.
3012 * systemd-networkd learned to cope with private-zone DHCP
3013 options and allows other programs to query the values.
3015 * SELinux access control when enabling/disabling units is no
3016 longer enforced with this release. The previous
3017 implementation was incorrect, and a new corrected
3018 implementation is not yet available. As unit file operations
3019 are still protected via PolicyKit and D-Bus policy this is
3020 not a security problem. Yet, distributions which care about
3021 optimal SELinux support should probably not stabilize on
3024 * sd-bus gained support for matches of type "arg0has=", that
3025 test for membership of strings in string arrays sent in bus
3028 * systemd-resolved now dumps the contents of its DNS and LLMNR
3029 caches to the logs on reception of the SIGUSR1 signal. This
3030 is useful to debug DNS behaviour.
3032 * The coredumpctl tool gained a new --directory= option to
3033 operate on journal files in a specific directory.
3035 * "systemctl reboot" and related commands gained a new
3036 "--message=" option which may be used to set a free-text
3037 wall message when shutting down or rebooting the
3038 system. This message is also logged, which is useful for
3039 figuring out the reason for a reboot or shutdown a
3042 * The "systemd-resolve-host" tool's -i switch now takes
3043 network interface numbers as alternative to interface names.
3045 * A new unit file setting for services has been introduced:
3046 UtmpMode= allows configuration of how precisely systemd
3047 handles utmp and wtmp entries for the service if this is
3048 enabled. This allows writing services that appear similar to
3049 user sessions in the output of the "w", "who", "last" and
3052 * systemd-resolved will now locally synthesize DNS resource
3053 records for the "localhost" and "gateway" domains as well as
3054 the local hostname. This should ensure that clients querying
3055 RRs via resolved will get similar results as those going via
3056 NSS, if nss-myhostname is enabled.
3058 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3059 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3060 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3061 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3062 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3063 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3064 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3065 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3066 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3067 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3068 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3069 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3071 — Berlin, 2015-08-27
3075 * The systemd-efi-boot-generator functionality was merged into
3076 systemd-gpt-auto-generator.
3078 * systemd-networkd now supports Group Policy for vxlan
3079 devices. It can be enabled via the new boolean configuration
3080 option called 'GroupPolicyExtension='.
3082 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3083 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3084 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3086 — Berlin, 2015-07-31
3090 * The python-systemd code has been removed from the systemd repository.
3091 A new repository has been created which accommodates the code from
3092 now on, and we kindly ask distributions to create a separate package
3093 for this: https://github.com/systemd/python-systemd
3095 * The systemd daemon will now reload its main configuration
3096 (/etc/systemd/system.conf) on daemon-reload.
3098 * sd-dhcp now exposes vendor specific extensions via
3099 sd_dhcp_lease_get_vendor_specific().
3101 * systemd-networkd gained a number of new configuration options.
3103 - A new boolean configuration option for TAP devices called
3104 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3105 device, thus allowing to send and receive GSO packets.
3107 - A new tunnel configuration option called 'CopyDSCP='.
3108 If enabled, the DSCP field of ip6 tunnels is copied into the
3109 decapsulated packet.
3111 - A set of boolean bridge configuration options were added.
3112 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3113 and 'UnicastFlood=' are now parsed by networkd and applied to the
3114 respective bridge link device via the respective IFLA_BRPORT_*
3117 - A new string configuration option to override the hostname sent
3118 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3119 is true, networkd will use the configured hostname instead of the
3120 system hostname when sending DHCP requests.
3122 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3123 networkd will configure the IPv6 flow-label of the tunnel device
3124 according to RFC2460.
3126 - The 'macvtap' virtual network devices are now supported, similar to
3127 the already supported 'macvlan' devices.
3129 * systemd-resolved now implements RFC5452 to improve resilience against
3130 cache poisoning. Additionally, source port randomization is enabled
3131 by default to further protect against DNS spoofing attacks.
3133 * nss-mymachines now supports translating UIDs and GIDs of running
3134 containers with user-namespaces enabled. If a container 'foo'
3135 translates a host uid 'UID' to the container uid 'TUID', then
3136 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3137 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3138 mapped as 'vg-foo-TGID'.
3140 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3141 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3142 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3143 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3144 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3145 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3146 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3147 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3148 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3149 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3151 — Berlin, 2015-07-29
3155 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3156 There are no known issues with current sysfs, and udev does not need
3157 or should be used to work around such bugs.
3159 * udev does no longer enable USB HID power management. Several reports
3160 indicate, that some devices cannot handle that setting.
3162 * The udev accelerometer helper was removed. The functionality
3163 is now fully included in iio-sensor-proxy. But this means,
3164 older iio-sensor-proxy versions will no longer provide
3165 accelerometer/orientation data with this systemd version.
3166 Please upgrade iio-sensor-proxy to version 1.0.
3168 * networkd gained a new configuration option IPv6PrivacyExtensions=
3169 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3170 for Stateless Address") on selected networks.
3172 * For the sake of fewer build-time dependencies and less code in the
3173 main repository, the python bindings are about to be removed in the
3174 next release. A new repository has been created which accommodates
3175 the code from now on, and we kindly ask distributions to create a
3176 separate package for this. The removal will take place in v223.
3178 https://github.com/systemd/python-systemd
3180 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3181 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3182 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3183 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3184 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3185 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3186 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3187 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3188 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3189 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3191 — Berlin, 2015-07-07
3195 * The sd-bus.h and sd-event.h APIs have now been declared
3196 stable and have been added to the official interface of
3197 libsystemd.so. sd-bus implements an alternative D-Bus client
3198 library, that is relatively easy to use, very efficient and
3199 supports both classic D-Bus as well as kdbus as transport
3200 backend. sd-event is a generic event loop abstraction that
3201 is built around Linux epoll, but adds features such as event
3202 prioritization or efficient timer handling. Both APIs are good
3203 choices for C programs looking for a bus and/or event loop
3204 implementation that is minimal and does not have to be
3205 portable to other kernels.
3207 * kdbus support is no longer compile-time optional. It is now
3208 always built-in. However, it can still be disabled at
3209 runtime using the kdbus=0 kernel command line setting, and
3210 that setting may be changed to default to off, by specifying
3211 --disable-kdbus at build-time. Note though that the kernel
3212 command line setting has no effect if the kdbus.ko kernel
3213 module is not installed, in which case kdbus is (obviously)
3214 also disabled. We encourage all downstream distributions to
3215 begin testing kdbus by adding it to the kernel images in the
3216 development distributions, and leaving kdbus support in
3219 * The minimal required util-linux version has been bumped to
3222 * Support for chkconfig (--enable-chkconfig) was removed in
3223 favor of calling an abstraction tool
3224 /lib/systemd/systemd-sysv-install. This needs to be
3225 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3226 in README for details.
3228 * If there's a systemd unit and a SysV init script for the
3229 same service name, and the user executes "systemctl enable"
3230 for it (or a related call), then this will now enable both
3231 (or execute the related operation on both), not just the
3234 * The libudev API documentation has been converted from gtkdoc
3237 * gudev has been removed from the systemd tree, it is now an
3240 * The systemd-cgtop tool learnt a new --raw switch to generate
3241 "raw" (machine parsable) output.
3243 * networkd's IPForwarding= .network file setting learnt the
3244 new setting "kernel", which ensures that networkd does not
3245 change the IP forwarding sysctl from the default kernel
3248 * The systemd-logind bus API now exposes a new boolean
3249 property "Docked" that reports whether logind considers the
3250 system "docked", i.e. connected to a docking station or not.
3252 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3253 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3254 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3255 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3256 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3257 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3258 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3259 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3260 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3261 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3262 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3263 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3264 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3265 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3266 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3267 Fink, Zbigniew Jędrzejewski-Szmek
3269 — Berlin, 2015-06-19
3273 * The gudev library has been extracted into a separate repository
3274 available at: https://git.gnome.org/browse/libgudev/
3275 It is now managed as part of the Gnome project. Distributions
3276 are recommended to pass --disable-gudev to systemd and use
3277 gudev from the Gnome project instead. gudev is still included
3278 in systemd, for now. It will be removed soon, though. Please
3279 also see the announcement-thread on systemd-devel:
3280 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3282 * systemd now exposes a CPUUsageNSec= property for each
3283 service unit on the bus, that contains the overall consumed
3284 CPU time of a service (the sum of what each process of the
3285 service consumed). This value is only available if
3286 CPUAccounting= is turned on for a service, and is then shown
3287 in the "systemctl status" output.
3289 * Support for configuring alternative mappings of the old SysV
3290 runlevels to systemd targets has been removed. They are now
3291 hardcoded in a way that runlevels 2, 3, 4 all map to
3292 multi-user.target and 5 to graphical.target (which
3293 previously was already the default behaviour).
3295 * The auto-mounter logic gained support for mount point
3296 expiry, using a new TimeoutIdleSec= setting in .automount
3297 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3299 * The EFI System Partition (ESP) as mounted to /boot by
3300 systemd-efi-boot-generator will now be unmounted
3301 automatically after 2 minutes of not being used. This should
3302 minimize the risk of ESP corruptions.
3304 * New /etc/fstab options x-systemd.requires= and
3305 x-systemd.requires-mounts-for= are now supported to express
3306 additional dependencies for mounts. This is useful for
3307 journalling file systems that support external journal
3308 devices or overlay file systems that require underlying file
3309 systems to be mounted.
3311 * systemd does not support direct live-upgrades (via systemctl
3312 daemon-reexec) from versions older than v44 anymore. As no
3313 distribution we are aware of shipped such old versions in a
3314 stable release this should not be problematic.
3316 * When systemd forks off a new per-connection service instance
3317 it will now set the $REMOTE_ADDR environment variable to the
3318 remote IP address, and $REMOTE_PORT environment variable to
3319 the remote IP port. This behaviour is similar to the
3320 corresponding environment variables defined by CGI.
3322 * systemd-networkd gained support for uplink failure
3323 detection. The BindCarrier= option allows binding interface
3324 configuration dynamically to the link sense of other
3325 interfaces. This is useful to achieve behaviour like in
3328 * systemd-networkd gained support for configuring the DHCP
3329 client identifier to use when requesting leases.
3331 * systemd-networkd now has a per-network UseNTP= option to
3332 configure whether NTP server information acquired via DHCP
3333 is passed on to services like systemd-timesyncd.
3335 * systemd-networkd gained support for vti6 tunnels.
3337 * Note that systemd-networkd manages the sysctl variable
3338 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3339 it is configured for since v219. The variable controls IP
3340 forwarding, and is a per-interface alternative to the global
3341 /proc/sys/net/ipv[46]/ip_forward. This setting is
3342 configurable in the IPForward= option, which defaults to
3343 "no". This means if networkd is used for an interface it is
3344 no longer sufficient to set the global sysctl option to turn
3345 on IP forwarding! Instead, the .network file option
3346 IPForward= needs to be turned on! Note that the
3347 implementation of this behaviour was broken in v219 and has
3350 * Many bonding and vxlan options are now configurable in
3353 * systemd-nspawn gained a new --property= setting to set unit
3354 properties for the container scope. This is useful for
3355 setting resource parameters (e.g. "CPUShares=500") on
3356 containers started from the command line.
3358 * systemd-nspawn gained a new --private-users= switch to make
3359 use of user namespacing available on recent Linux kernels.
3361 * systemd-nspawn may now be called as part of a shell pipeline
3362 in which case the pipes used for stdin and stdout are passed
3363 directly to the process invoked in the container, without
3364 indirection via a pseudo tty.
3366 * systemd-nspawn gained a new switch to control the UNIX
3367 signal to use when killing the init process of the container
3370 * systemd-nspawn gained a new --overlay= switch for mounting
3371 overlay file systems into the container using the new kernel
3374 * When a container image is imported via systemd-importd and
3375 the host file system is not btrfs, a loopback block device
3376 file is created in /var/lib/machines.raw with a btrfs file
3377 system inside. It is then mounted to /var/lib/machines to
3378 enable btrfs features for container management. The loopback
3379 file and btrfs file system is grown as needed when container
3380 images are imported via systemd-importd.
3382 * systemd-machined/systemd-importd gained support for btrfs
3383 quota, to enforce container disk space limits on disk. This
3384 is exposed in "machinectl set-limit".
3386 * systemd-importd now can import containers from local .tar,
3387 .raw and .qcow2 images, and export them to .tar and .raw. It
3388 can also import dkr v2 images now from the network (on top
3391 * systemd-importd gained support for verifying downloaded
3392 images with gpg2 (previously only gpg1 was supported).
3394 * systemd-machined, systemd-logind, systemd: most bus calls
3395 are now accessible to unprivileged processes via
3396 PolicyKit. Also, systemd-logind will now allow users to kill
3397 their own sessions without further privileges or
3400 * systemd-shutdownd has been removed. This service was
3401 previously responsible for implementing scheduled shutdowns
3402 as exposed in /usr/bin/shutdown's time parameter. This
3403 functionality has now been moved into systemd-logind and is
3404 accessible via a bus interface.
3406 * "systemctl reboot" gained a new switch --firmware-setup that
3407 can be used to reboot into the EFI firmware setup, if that
3408 is available. systemd-logind now exposes an API on the bus
3409 to trigger such reboots, in case graphical desktop UIs want
3410 to cover this functionality.
3412 * "systemctl enable", "systemctl disable" and "systemctl mask"
3413 now support a new "--now" switch. If specified the units
3414 that are enabled will also be started, and the ones
3415 disabled/masked also stopped.
3417 * The Gummiboot EFI boot loader tool has been merged into
3418 systemd, and renamed to "systemd-boot". The bootctl tool has been
3419 updated to support systemd-boot.
3421 * An EFI kernel stub has been added that may be used to create
3422 kernel EFI binaries that contain not only the actual kernel,
3423 but also an initrd, boot splash, command line and OS release
3424 information. This combined binary can then be signed as a
3425 single image, so that the firmware can verify it all in one
3426 step. systemd-boot has special support for EFI binaries created
3427 like this and can extract OS release information from them
3428 and show them in the boot menu. This functionality is useful
3429 to implement cryptographically verified boot schemes.
3431 * Optional support has been added to systemd-fsck to pass
3432 fsck's progress report to an AF_UNIX socket in the file
3435 * udev will no longer create device symlinks for all block
3436 devices by default. A blacklist for excluding special block
3437 devices from this logic has been turned into a whitelist
3438 that requires picking block devices explicitly that require
3441 * A new (currently still internal) API sd-device.h has been
3442 added to libsystemd. This modernized API is supposed to
3443 replace libudev eventually. In fact, already much of libudev
3444 is now just a wrapper around sd-device.h.
3446 * A new hwdb database for storing metadata about pointing
3447 stick devices has been added.
3449 * systemd-tmpfiles gained support for setting file attributes
3450 similar to the "chattr" tool with new 'h' and 'H' lines.
3452 * systemd-journald will no longer unconditionally set the
3453 btrfs NOCOW flag on new journal files. This is instead done
3454 with tmpfiles snippet using the new 'h' line type. This
3455 allows easy disabling of this logic, by masking the
3456 journal-nocow.conf tmpfiles file.
3458 * systemd-journald will now translate audit message types to
3459 human readable identifiers when writing them to the
3460 journal. This should improve readability of audit messages.
3462 * The LUKS logic gained support for the offset= and skip=
3463 options in /etc/crypttab, as previously implemented by
3466 * /usr/lib/os-release gained a new optional field VARIANT= for
3467 distributions that support multiple variants (such as a
3468 desktop edition, a server edition, ...)
3470 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3471 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3472 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3473 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3474 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3475 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3476 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3477 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3478 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3479 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3480 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3481 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3482 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3483 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3484 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3485 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3486 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3487 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3488 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3489 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3490 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3491 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3492 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3493 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3494 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3495 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3496 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3498 — Berlin, 2015-05-22
3502 * Introduce a new API "sd-hwdb.h" for querying the hardware
3503 metadata database. With this minimal interface one can query
3504 and enumerate the udev hwdb, decoupled from the old libudev
3505 library. libudev's interface for this is now only a wrapper
3506 around sd-hwdb. A new tool systemd-hwdb has been added to
3507 interface with and update the database.
3509 * When any of systemd's tools copies files (for example due to
3510 tmpfiles' C lines) a btrfs reflink will attempted first,
3511 before bytewise copying is done.
3513 * systemd-nspawn gained a new --ephemeral switch. When
3514 specified a btrfs snapshot is taken of the container's root
3515 directory, and immediately removed when the container
3516 terminates again. Thus, a container can be started whose
3517 changes never alter the container's root directory, and are
3518 lost on container termination. This switch can also be used
3519 for starting a container off the root file system of the
3520 host without affecting the host OS. This switch is only
3521 available on btrfs file systems.
3523 * systemd-nspawn gained a new --template= switch. It takes the
3524 path to a container tree to use as template for the tree
3525 specified via --directory=, should that directory be
3526 missing. This allows instantiating containers dynamically,
3527 on first run. This switch is only available on btrfs file
3530 * When a .mount unit refers to a mount point on which multiple
3531 mounts are stacked, and the .mount unit is stopped all of
3532 the stacked mount points will now be unmounted until no
3533 mount point remains.
3535 * systemd now has an explicit notion of supported and
3536 unsupported unit types. Jobs enqueued for unsupported unit
3537 types will now fail with an "unsupported" error code. More
3538 specifically .swap, .automount and .device units are not
3539 supported in containers, .busname units are not supported on
3540 non-kdbus systems. .swap and .automount are also not
3541 supported if their respective kernel compile time options
3544 * machinectl gained support for two new "copy-from" and
3545 "copy-to" commands for copying files from a running
3546 container to the host or vice versa.
3548 * machinectl gained support for a new "bind" command to bind
3549 mount host directories into local containers. This is
3550 currently only supported for nspawn containers.
3552 * networkd gained support for configuring bridge forwarding
3553 database entries (fdb) from .network files.
3555 * A new tiny daemon "systemd-importd" has been added that can
3556 download container images in tar, raw, qcow2 or dkr formats,
3557 and make them available locally in /var/lib/machines, so
3558 that they can run as nspawn containers. The daemon can GPG
3559 verify the downloads (not supported for dkr, since it has no
3560 provisions for verifying downloads). It will transparently
3561 decompress bz2, xz, gzip compressed downloads if necessary,
3562 and restore sparse files on disk. The daemon uses privilege
3563 separation to ensure the actual download logic runs with
3564 fewer privileges than the daemon itself. machinectl has
3565 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3566 make the functionality of importd available to the
3567 user. With this in place the Fedora and Ubuntu "Cloud"
3568 images can be downloaded and booted as containers unmodified
3569 (the Fedora images lack the appropriate GPG signature files
3570 currently, so they cannot be verified, but this will change
3571 soon, hopefully). Note that downloading images is currently
3572 only fully supported on btrfs.
3574 * machinectl is now able to list container images found in
3575 /var/lib/machines, along with some metadata about sizes of
3576 disk and similar. If the directory is located on btrfs and
3577 quota is enabled, this includes quota display. A new command
3578 "image-status" has been added that shows additional
3579 information about images.
3581 * machinectl is now able to clone container images
3582 efficiently, if the underlying file system (btrfs) supports
3583 it, with the new "machinectl clone" command. It also
3584 gained commands for renaming and removing images, as well as
3585 marking them read-only or read-write (supported also on
3586 legacy file systems).
3588 * networkd gained support for collecting LLDP network
3589 announcements, from hardware that supports this. This is
3590 shown in networkctl output.
3592 * systemd-run gained support for a new -t (--pty) switch for
3593 invoking a binary on a pty whose input and output is
3594 connected to the invoking terminal. This allows executing
3595 processes as system services while interactively
3596 communicating with them via the terminal. Most interestingly
3597 this is supported across container boundaries. Invoking
3598 "systemd-run -t /bin/bash" is an alternative to running a
3599 full login session, the difference being that the former
3600 will not register a session, nor go through the PAM session
3603 * tmpfiles gained support for a new "v" line type for creating
3604 btrfs subvolumes. If the underlying file system is a legacy
3605 file system, this automatically degrades to creating a
3606 normal directory. Among others /var/lib/machines is now
3607 created like this at boot, should it be missing.
3609 * The directory /var/lib/containers/ has been deprecated and
3610 been replaced by /var/lib/machines. The term "machines" has
3611 been used in the systemd context as generic term for both
3612 VMs and containers, and hence appears more appropriate for
3613 this, as the directory can also contain raw images bootable
3616 * systemd-nspawn when invoked with -M but without --directory=
3617 or --image= is now capable of searching for the container
3618 root directory, subvolume or disk image automatically, in
3619 /var/lib/machines. systemd-nspawn@.service has been updated
3620 to make use of this, thus allowing it to be used for raw
3623 * A new machines.target unit has been introduced that is
3624 supposed to group all containers/VMs invoked as services on
3625 the system. systemd-nspawn@.service has been updated to
3626 integrate with that.
3628 * machinectl gained a new "start" command, for invoking a
3629 container as a service. "machinectl start foo" is mostly
3630 equivalent to "systemctl start systemd-nspawn@foo.service",
3631 but handles escaping in a nicer way.
3633 * systemd-nspawn will now mount most of the cgroupfs tree
3634 read-only into each container, with the exception of the
3635 container's own subtree in the name=systemd hierarchy.
3637 * journald now sets the special FS_NOCOW file flag for its
3638 journal files. This should improve performance on btrfs, by
3639 avoiding heavy fragmentation when journald's write-pattern
3640 is used on COW file systems. It degrades btrfs' data
3641 integrity guarantees for the files to the same levels as for
3642 ext3/ext4 however. This should be OK though as journald does
3643 its own data integrity checks and all its objects are
3644 checksummed on disk. Also, journald should handle btrfs disk
3645 full events a lot more gracefully now, by processing SIGBUS
3646 errors, and not relying on fallocate() anymore.
3648 * When journald detects that journal files it is writing to
3649 have been deleted it will immediately start new journal
3652 * systemd now provides a way to store file descriptors
3653 per-service in PID 1. This is useful for daemons to ensure
3654 that fds they require are not lost during a daemon
3655 restart. The fds are passed to the daemon on the next
3656 invocation in the same way socket activation fds are
3657 passed. This is now used by journald to ensure that the
3658 various sockets connected to all the system's stdout/stderr
3659 are not lost when journald is restarted. File descriptors
3660 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3661 an extension to sd_notify(). Note that a limit is enforced
3662 on the number of fds a service can store in PID 1, and it
3663 defaults to 0, so that no fds may be stored, unless this is
3664 explicitly turned on.
3666 * The default TERM variable to use for units connected to a
3667 terminal, when no other value is explicitly is set is now
3668 vt220 rather than vt102. This should be fairly safe still,
3669 but allows PgUp/PgDn work.
3671 * The /etc/crypttab option header= as known from Debian is now
3674 * "loginctl user-status" and "loginctl session-status" will
3675 now show the last 10 lines of log messages of the
3676 user/session following the status output. Similar,
3677 "machinectl status" will show the last 10 log lines
3678 associated with a virtual machine or container
3679 service. (Note that this is usually not the log messages
3680 done in the VM/container itself, but simply what the
3681 container manager logs. For nspawn this includes all console
3684 * "loginctl session-status" without further argument will now
3685 show the status of the session of the caller. Similar,
3686 "lock-session", "unlock-session", "activate",
3687 "enable-linger", "disable-linger" may now be called without
3688 session/user parameter in which case they apply to the
3689 caller's session/user.
3691 * An X11 session scriptlet is now shipped that uploads
3692 $DISPLAY and $XAUTHORITY into the environment of the systemd
3693 --user daemon if a session begins. This should improve
3694 compatibility with X11 enabled applications run as systemd
3697 * Generators are now subject to masking via /etc and /run, the
3698 same way as unit files.
3700 * networkd .network files gained support for configuring
3701 per-link IPv4/IPv6 packet forwarding as well as IPv4
3702 masquerading. This is by default turned on for veth links to
3703 containers, as registered by systemd-nspawn. This means that
3704 nspawn containers run with --network-veth will now get
3705 automatic routed access to the host's networks without any
3706 further configuration or setup, as long as networkd runs on
3709 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3710 or UDP posts of a container on the host. With this in place
3711 it is possible to run containers with private veth links
3712 (--network-veth), and have their functionality exposed on
3713 the host as if their services were running directly on the
3716 * systemd-nspawn's --network-veth switch now gained a short
3717 version "-n", since with the changes above it is now truly
3718 useful out-of-the-box. The systemd-nspawn@.service has been
3719 updated to make use of it too by default.
3721 * systemd-nspawn will now maintain a per-image R/W lock, to
3722 ensure that the same image is not started more than once
3723 writable. (It's OK to run an image multiple times
3724 simultaneously in read-only mode.)
3726 * systemd-nspawn's --image= option is now capable of
3727 dissecting and booting MBR and GPT disk images that contain
3728 only a single active Linux partition. Previously it
3729 supported only GPT disk images with proper GPT type
3730 IDs. This allows running cloud images from major
3731 distributions directly with systemd-nspawn, without
3734 * In addition to collecting mouse dpi data in the udev
3735 hardware database, there's now support for collecting angle
3736 information for mouse scroll wheels. The database is
3737 supposed to guarantee similar scrolling behavior on mice
3738 that it knows about. There's also support for collecting
3739 information about Touchpad types.
3741 * udev's input_id built-in will now also collect touch screen
3742 dimension data and attach it to probed devices.
3744 * /etc/os-release gained support for a Distribution Privacy
3747 * networkd gained support for creating "ipvlan", "gretap",
3748 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3750 * systemd-tmpfiles gained support for "a" lines for setting
3753 * systemd-nspawn will now mount /tmp in the container to
3754 tmpfs, automatically.
3756 * systemd now exposes the memory.usage_in_bytes cgroup
3757 attribute and shows it for each service in the "systemctl
3758 status" output, if available.
3760 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3761 immediate reboot is triggered. This useful if shutdown is
3762 hung and is unable to complete, to expedite the
3763 operation. Note that this kind of reboot will still unmount
3764 all file systems, and hence should not result in fsck being
3767 * A .device unit for an optical block device will now be
3768 considered active only when a medium is in the drive. Also,
3769 mount units are now bound to their backing devices thus
3770 triggering automatic unmounting when devices become
3771 unavailable. With this in place systemd will now
3772 automatically unmount left-over mounts when a CD-ROM is
3773 ejected or an USB stick is yanked from the system.
3775 * networkd-wait-online now has support for waiting for
3776 specific interfaces only (with globbing), and for giving up
3777 after a configurable timeout.
3779 * networkd now exits when idle. It will be automatically
3780 restarted as soon as interfaces show up, are removed or
3781 change state. networkd will stay around as long as there is
3782 at least one DHCP state machine or similar around, that keep
3785 * networkd may now configure IPv6 link-local addressing in
3786 addition to IPv4 link-local addressing.
3788 * The IPv6 "token" for use in SLAAC may now be configured for
3789 each .network interface in networkd.
3791 * Routes configured with networkd may now be assigned a scope
3794 * networkd's [Match] sections now support globbing and lists
3795 of multiple space-separated matches per item.
3797 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3798 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3799 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3800 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3801 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3802 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3803 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3804 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3805 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3806 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3807 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3808 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3809 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3810 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3811 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3812 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3813 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3814 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3815 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3816 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3817 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3818 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3819 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3820 Hoffmann, Zbigniew Jędrzejewski-Szmek
3822 — Berlin, 2015-02-16
3826 * When querying unit file enablement status (for example via
3827 "systemctl is-enabled"), a new state "indirect" is now known
3828 which indicates that a unit might not be enabled itself, but
3829 another unit listed in its Also= setting might be.
3831 * Similar to the various existing ConditionXYZ= settings for
3832 units, there are now matching AssertXYZ= settings. While
3833 failing conditions cause a unit to be skipped, but its job
3834 to succeed, failing assertions declared like this will cause
3835 a unit start operation and its job to fail.
3837 * hostnamed now knows a new chassis type "embedded".
3839 * systemctl gained a new "edit" command. When used on a unit
3840 file, this allows extending unit files with .d/ drop-in
3841 configuration snippets or editing the full file (after
3842 copying it from /usr/lib to /etc). This will invoke the
3843 user's editor (as configured with $EDITOR), and reload the
3844 modified configuration after editing.
3846 * "systemctl status" now shows the suggested enablement state
3847 for a unit, as declared in the (usually vendor-supplied)
3848 system preset files.
3850 * nss-myhostname will now resolve the single-label host name
3851 "gateway" to the locally configured default IP routing
3852 gateways, ordered by their metrics. This assigns a stable
3853 name to the used gateways, regardless which ones are
3854 currently configured. Note that the name will only be
3855 resolved after all other name sources (if nss-myhostname is
3856 configured properly) and should hence not negatively impact
3857 systems that use the single-label host name "gateway" in
3860 * systemd-inhibit now allows filtering by mode when listing
3863 * Scope and service units gained a new "Delegate" boolean
3864 property, which, when set, allows processes running inside the
3865 unit to further partition resources. This is primarily
3866 useful for systemd user instances as well as container
3869 * journald will now pick up audit messages directly from
3870 the kernel, and log them like any other log message. The
3871 audit fields are split up and fully indexed. This means that
3872 journalctl in many ways is now a (nicer!) alternative to
3873 ausearch, the traditional audit client. Note that this
3874 implements only a minimal audit client. If you want the
3875 special audit modes like reboot-on-log-overflow, please use
3876 the traditional auditd instead, which can be used in
3877 parallel to journald.
3879 * The ConditionSecurity= unit file option now understands the
3880 special string "audit" to check whether auditing is
3883 * journalctl gained two new commands --vacuum-size= and
3884 --vacuum-time= to delete old journal files until the
3885 remaining ones take up no more than the specified size on disk,
3886 or are not older than the specified time.
3888 * A new, native PPPoE library has been added to sd-network,
3889 systemd's library of light-weight networking protocols. This
3890 library will be used in a future version of networkd to
3891 enable PPPoE communication without an external pppd daemon.
3893 * The busctl tool now understands a new "capture" verb that
3894 works similar to "monitor", but writes a packet capture
3895 trace to STDOUT that can be redirected to a file which is
3896 compatible with libcap's capture file format. This can then
3897 be loaded in Wireshark and similar tools to inspect bus
3900 * The busctl tool now understands a new "tree" verb that shows
3901 the object trees of a specific service on the bus, or of all
3904 * The busctl tool now understands a new "introspect" verb that
3905 shows all interfaces and members of objects on the bus,
3906 including their signature and values. This is particularly
3907 useful to get more information about bus objects shown by
3908 the new "busctl tree" command.
3910 * The busctl tool now understands new verbs "call",
3911 "set-property" and "get-property" for invoking bus method
3912 calls, setting and getting bus object properties in a
3915 * busctl gained a new --augment-creds= argument that controls
3916 whether the tool shall augment credential information it
3917 gets from the bus with data from /proc, in a possibly
3920 * nspawn's --link-journal= switch gained two new values
3921 "try-guest" and "try-host" that work like "guest" and
3922 "host", but do not fail if the host has no persistent
3923 journalling enabled. -j is now equivalent to
3924 --link-journal=try-guest.
3926 * macvlan network devices created by nspawn will now have
3927 stable MAC addresses.
3929 * A new SmackProcessLabel= unit setting has been added, which
3930 controls the SMACK security label processes forked off by
3931 the respective unit shall use.
3933 * If compiled with --enable-xkbcommon, systemd-localed will
3934 verify x11 keymap settings by compiling the given keymap. It
3935 will spew out warnings if the compilation fails. This
3936 requires libxkbcommon to be installed.
3938 * When a coredump is collected, a larger number of metadata
3939 fields is now collected and included in the journal records
3940 created for it. More specifically, control group membership,
3941 environment variables, memory maps, working directory,
3942 chroot directory, /proc/$PID/status, and a list of open file
3943 descriptors is now stored in the log entry.
3945 * The udev hwdb now contains DPI information for mice. For
3948 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3950 * All systemd programs that read standalone configuration
3951 files in /etc now also support a corresponding series of
3952 .conf.d configuration directories in /etc/, /run/,
3953 /usr/local/lib/, /usr/lib/, and (if configured with
3954 --enable-split-usr) /lib/. In particular, the following
3955 configuration files now have corresponding configuration
3956 directories: system.conf user.conf, logind.conf,
3957 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3958 resolved.conf, timesyncd.conf, journal-remote.conf, and
3959 journal-upload.conf. Note that distributions should use the
3960 configuration directories in /usr/lib/; the directories in
3961 /etc/ are reserved for the system administrator.
3963 * systemd-rfkill will no longer take the rfkill device name
3964 into account when storing rfkill state on disk, as the name
3965 might be dynamically assigned and not stable. Instead, the
3966 ID_PATH udev variable combined with the rfkill type (wlan,
3967 bluetooth, ...) is used.
3969 * A new service systemd-machine-id-commit.service has been
3970 added. When used on systems where /etc is read-only during
3971 boot, and /etc/machine-id is not initialized (but an empty
3972 file), this service will copy the temporary machine ID
3973 created as replacement into /etc after the system is fully
3974 booted up. This is useful for systems that are freshly
3975 installed with a non-initialized machine ID, but should get
3976 a fixed machine ID for subsequent boots.
3978 * networkd's .netdev files now provide a large set of
3979 configuration parameters for VXLAN devices. Similarly, the
3980 bridge port cost parameter is now configurable in .network
3981 files. There's also new support for configuring IP source
3982 routing. networkd .link files gained support for a new
3983 OriginalName= match that is useful to match against the
3984 original interface name the kernel assigned. .network files
3985 may include MTU= and MACAddress= fields for altering the MTU
3986 and MAC address while being connected to a specific network
3989 * The LUKS logic gained supported for configuring
3990 UUID-specific key files. There's also new support for naming
3991 LUKS device from the kernel command line, using the new
3992 luks.name= argument.
3994 * Timer units may now be transiently created via the bus API
3995 (this was previously already available for scope and service
3996 units). In addition it is now possible to create multiple
3997 transient units at the same time with a single bus call. The
3998 "systemd-run" tool has been updated to make use of this for
3999 running commands on a specified time, in at(1)-style.
4001 * tmpfiles gained support for "t" lines, for assigning
4002 extended attributes to files. Among other uses this may be
4003 used to assign SMACK labels to files.
4005 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4006 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4007 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4008 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4009 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4010 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4011 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4012 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4013 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4014 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4015 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4016 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4017 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4018 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4019 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4020 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4021 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4022 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4024 — Berlin, 2014-12-10
4028 * journalctl gained the new options -t/--identifier= to match
4029 on the syslog identifier (aka "tag"), as well as --utc to
4030 show log timestamps in the UTC timezone. journalctl now also
4031 accepts -n/--lines=all to disable line capping in a pager.
4033 * journalctl gained a new switch, --flush, that synchronously
4034 flushes logs from /run/log/journal to /var/log/journal if
4035 persistent storage is enabled. systemd-journal-flush.service
4036 now waits until the operation is complete.
4038 * Services can notify the manager before they start a reload
4039 (by sending RELOADING=1) or shutdown (by sending
4040 STOPPING=1). This allows the manager to track and show the
4041 internal state of daemons and closes a race condition when
4042 the process is still running but has closed its D-Bus
4045 * Services with Type=oneshot do not have to have any ExecStart
4048 * User units are now loaded also from
4049 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4050 /run/systemd/user directory that was already previously
4051 supported, but is under the control of the user.
4053 * Job timeouts (i.e. time-outs on the time a job that is
4054 queued stays in the run queue) can now optionally result in
4055 immediate reboot or power-off actions (JobTimeoutAction= and
4056 JobTimeoutRebootArgument=). This is useful on ".target"
4057 units, to limit the maximum time a target remains
4058 undispatched in the run queue, and to trigger an emergency
4059 operation in such a case. This is now used by default to
4060 turn off the system if boot-up (as defined by everything in
4061 basic.target) hangs and does not complete for at least
4062 15min. Also, if power-off or reboot hang for at least 30min
4063 an immediate power-off/reboot operation is triggered. This
4064 functionality is particularly useful to increase reliability
4065 on embedded devices, but also on laptops which might
4066 accidentally get powered on when carried in a backpack and
4067 whose boot stays stuck in a hard disk encryption passphrase
4070 * systemd-logind can be configured to also handle lid switch
4071 events even when the machine is docked or multiple displays
4072 are attached (HandleLidSwitchDocked= option).
4074 * A helper binary and a service have been added which can be
4075 used to resume from hibernation in the initramfs. A
4076 generator will parse the resume= option on the kernel
4077 command line to trigger resume.
4079 * A user console daemon systemd-consoled has been
4080 added. Currently, it is a preview, and will so far open a
4081 single terminal on each session of the user marked as
4082 Desktop=systemd-console.
4084 * Route metrics can be specified for DHCP routes added by
4087 * The SELinux context of socket-activated services can be set
4088 from the information provided by the networking stack
4089 (SELinuxContextFromNet= option).
4091 * Userspace firmware loading support has been removed and
4092 the minimum supported kernel version is thus bumped to 3.7.
4094 * Timeout for udev workers has been increased from 1 to 3
4095 minutes, but a warning will be printed after 1 minute to
4096 help diagnose kernel modules that take a long time to load.
4098 * Udev rules can now remove tags on devices with TAG-="foobar".
4100 * systemd's readahead implementation has been removed. In many
4101 circumstances it didn't give expected benefits even for
4102 rotational disk drives and was becoming less relevant in the
4103 age of SSDs. As none of the developers has been using
4104 rotating media anymore, and nobody stepped up to actively
4105 maintain this component of systemd it has now been removed.
4107 * Swap units can use Options= to specify discard options.
4108 Discard options specified for swaps in /etc/fstab are now
4111 * Docker containers are now detected as a separate type of
4114 * The Password Agent protocol gained support for queries where
4115 the user input is shown, useful e.g. for user names.
4116 systemd-ask-password gained a new --echo option to turn that
4119 * The default sysctl.d/ snippets will now set:
4121 net.core.default_qdisc = fq_codel
4123 This selects Fair Queuing Controlled Delay as the default
4124 queuing discipline for network interfaces. fq_codel helps
4125 fight the network bufferbloat problem. It is believed to be
4126 a good default with no tuning required for most workloads.
4127 Downstream distributions may override this choice. On 10Gbit
4128 servers that do not do forwarding, "fq" may perform better.
4129 Systems without a good clocksource should use "pfifo_fast".
4131 * If kdbus is enabled during build a new option BusPolicy= is
4132 available for service units, that allows locking all service
4133 processes into a stricter bus policy, in order to limit
4134 access to various bus services, or even hide most of them
4135 from the service's view entirely.
4137 * networkctl will now show the .network and .link file
4138 networkd has applied to a specific interface.
4140 * sd-login gained a new API call sd_session_get_desktop() to
4141 query which desktop environment has been selected for a
4144 * UNIX utmp support is now compile-time optional to support
4145 legacy-free systems.
4147 * systemctl gained two new commands "add-wants" and
4148 "add-requires" for pulling in units from specific targets
4151 * If the word "rescue" is specified on the kernel command line
4152 the system will now boot into rescue mode (aka
4153 rescue.target), which was previously available only by
4154 specifying "1" or "systemd.unit=rescue.target" on the kernel
4155 command line. This new kernel command line option nicely
4156 mirrors the already existing "emergency" kernel command line
4159 * New kernel command line options mount.usr=, mount.usrflags=,
4160 mount.usrfstype= have been added that match root=, rootflags=,
4161 rootfstype= but allow mounting a specific file system to
4164 * The $NOTIFY_SOCKET is now also passed to control processes of
4165 services, not only the main process.
4167 * This version reenables support for fsck's -l switch. This
4168 means at least version v2.25 of util-linux is required for
4169 operation, otherwise dead-locks on device nodes may
4170 occur. Again: you need to update util-linux to at least
4171 v2.25 when updating systemd to v217.
4173 * The "multi-seat-x" tool has been removed from systemd, as
4174 its functionality has been integrated into X servers 1.16,
4175 and the tool is hence redundant. It is recommended to update
4176 display managers invoking this tool to simply invoke X
4177 directly from now on, again.
4179 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4180 message flag has been added for all of systemd's PolicyKit
4181 authenticated method calls has been added. In particular
4182 this now allows optional interactive authorization via
4183 PolicyKit for many of PID1's privileged operations such as
4184 unit file enabling and disabling.
4186 * "udevadm hwdb --update" learnt a new switch "--usr" for
4187 placing the rebuilt hardware database in /usr instead of
4188 /etc. When used only hardware database entries stored in
4189 /usr will be used, and any user database entries in /etc are
4190 ignored. This functionality is useful for vendors to ship a
4191 pre-built database on systems where local configuration is
4192 unnecessary or unlikely.
4194 * Calendar time specifications in .timer units now also
4195 understand the strings "semi-annually", "quarterly" and
4196 "minutely" as shortcuts (in addition to the preexisting
4197 "anually", "hourly", ...).
4199 * systemd-tmpfiles will now correctly create files in /dev
4200 at boot which are marked for creation only at boot. It is
4201 recommended to always create static device nodes with 'c!'
4202 and 'b!', so that they are created only at boot and not
4203 overwritten at runtime.
4205 * When the watchdog logic is used for a service (WatchdogSec=)
4206 and the watchdog timeout is hit the service will now be
4207 terminated with SIGABRT (instead of just SIGTERM), in order
4208 to make sure a proper coredump and backtrace is
4209 generated. This ensures that hanging services will result in
4210 similar coredump/backtrace behaviour as services that hit a
4213 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4214 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4215 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4216 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4217 Herrmann, David Sommerseth, David Strauss, Emil Renner
4218 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4219 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4220 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4221 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4222 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4223 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4224 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4225 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4226 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4227 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4228 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4229 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4230 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4231 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4232 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4233 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4236 — Berlin, 2014-10-28
4240 * timedated no longer reads NTP implementation unit names from
4241 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4242 implementations should add a
4244 Conflicts=systemd-timesyncd.service
4246 to their unit files to take over and replace systemd's NTP
4247 default functionality.
4249 * systemd-sysusers gained a new line type "r" for configuring
4250 which UID/GID ranges to allocate system users/groups
4251 from. Lines of type "u" may now add an additional column
4252 that specifies the home directory for the system user to be
4253 created. Also, systemd-sysusers may now optionally read user
4254 information from STDIN instead of a file. This is useful for
4255 invoking it from RPM preinst scriptlets that need to create
4256 users before the first RPM file is installed since these
4257 files might need to be owned by them. A new
4258 %sysusers_create_inline RPM macro has been introduced to do
4259 just that. systemd-sysusers now updates the shadow files as
4260 well as the user/group databases, which should enhance
4261 compatibility with certain tools like grpck.
4263 * A number of bus APIs of PID 1 now optionally consult
4264 PolicyKit to permit access for otherwise unprivileged
4265 clients under certain conditions. Note that this currently
4266 doesn't support interactive authentication yet, but this is
4267 expected to be added eventually, too.
4269 * /etc/machine-info now has new fields for configuring the
4270 deployment environment of the machine, as well as the
4271 location of the machine. hostnamectl has been updated with
4272 new command to update these fields.
4274 * systemd-timesyncd has been updated to automatically acquire
4275 NTP server information from systemd-networkd, which might
4276 have been discovered via DHCP.
4278 * systemd-resolved now includes a caching DNS stub resolver
4279 and a complete LLMNR name resolution implementation. A new
4280 NSS module "nss-resolve" has been added which can be used
4281 instead of glibc's own "nss-dns" to resolve hostnames via
4282 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4283 be resolved via systemd-resolved D-Bus APIs. In contrast to
4284 the glibc internal resolver systemd-resolved is aware of
4285 multi-homed system, and keeps DNS server and caches separate
4286 and per-interface. Queries are sent simultaneously on all
4287 interfaces that have DNS servers configured, in order to
4288 properly handle VPNs and local LANs which might resolve
4289 separate sets of domain names. systemd-resolved may acquire
4290 DNS server information from systemd-networkd automatically,
4291 which in turn might have discovered them via DHCP. A tool
4292 "systemd-resolve-host" has been added that may be used to
4293 query the DNS logic in resolved. systemd-resolved implements
4294 IDNA and automatically uses IDNA or UTF-8 encoding depending
4295 on whether classic DNS or LLMNR is used as transport. In the
4296 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4297 implementation to systemd-resolved.
4299 * A new NSS module nss-mymachines has been added, that
4300 automatically resolves the names of all local registered
4301 containers to their respective IP addresses.
4303 * A new client tool "networkctl" for systemd-networkd has been
4304 added. It currently is entirely passive and will query
4305 networking configuration from udev, rtnetlink and networkd,
4306 and present it to the user in a very friendly
4307 way. Eventually, we hope to extend it to become a full
4308 control utility for networkd.
4310 * .socket units gained a new DeferAcceptSec= setting that
4311 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4312 TCP. Similarly, support for controlling TCP keep-alive
4313 settings has been added (KeepAliveTimeSec=,
4314 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4315 turning off Nagle's algorithm on TCP has been added
4318 * logind learned a new session type "web", for use in projects
4319 like Cockpit which register web clients as PAM sessions.
4321 * timer units with at least one OnCalendar= setting will now
4322 be started only after time-sync.target has been
4323 reached. This way they will not elapse before the system
4324 clock has been corrected by a local NTP client or
4325 similar. This is particular useful on RTC-less embedded
4326 machines, that come up with an invalid system clock.
4328 * systemd-nspawn's --network-veth= switch should now result in
4329 stable MAC addresses for both the outer and the inner side
4332 * systemd-nspawn gained a new --volatile= switch for running
4333 container instances with /etc or /var unpopulated.
4335 * The kdbus client code has been updated to use the new Linux
4336 3.17 memfd subsystem instead of the old kdbus-specific one.
4338 * systemd-networkd's DHCP client and server now support
4339 FORCERENEW. There are also new configuration options to
4340 configure the vendor client identifier and broadcast mode
4343 * systemd will no longer inform the kernel about the current
4344 timezone, as this is necessarily incorrect and racy as the
4345 kernel has no understanding of DST and similar
4346 concepts. This hence means FAT timestamps will be always
4347 considered UTC, similar to what Android is already
4348 doing. Also, when the RTC is configured to the local time
4349 (rather than UTC) systemd will never synchronize back to it,
4350 as this might confuse Windows at a later boot.
4352 * systemd-analyze gained a new command "verify" for offline
4353 validation of unit files.
4355 * systemd-networkd gained support for a couple of additional
4356 settings for bonding networking setups. Also, the metric for
4357 statically configured routes may now be configured. For
4358 network interfaces where this is appropriate the peer IP
4359 address may now be configured.
4361 * systemd-networkd's DHCP client will no longer request
4362 broadcasting by default, as this tripped up some networks.
4363 For hardware where broadcast is required the feature should
4364 be switched back on using RequestBroadcast=yes.
4366 * systemd-networkd will now set up IPv4LL addresses (when
4367 enabled) even if DHCP is configured successfully.
4369 * udev will now default to respect network device names given
4370 by the kernel when the kernel indicates that these are
4371 predictable. This behavior can be tweaked by changing
4372 NamePolicy= in the relevant .link file.
4374 * A new library systemd-terminal has been added that
4375 implements full TTY stream parsing and rendering. This
4376 library is supposed to be used later on for implementing a
4377 full userspace VT subsystem, replacing the current kernel
4380 * A new tool systemd-journal-upload has been added to push
4381 journal data to a remote system running
4382 systemd-journal-remote.
4384 * journald will no longer forward all local data to another
4385 running syslog daemon. This change has been made because
4386 rsyslog (which appears to be the most commonly used syslog
4387 implementation these days) no longer makes use of this, and
4388 instead pulls the data out of the journal on its own. Since
4389 forwarding the messages to a non-existent syslog server is
4390 more expensive than we assumed we have now turned this
4391 off. If you run a syslog server that is not a recent rsyslog
4392 version, you have to turn this option on again
4393 (ForwardToSyslog= in journald.conf).
4395 * journald now optionally supports the LZ4 compressor for
4396 larger journal fields. This compressor should perform much
4397 better than XZ which was the previous default.
4399 * machinectl now shows the IP addresses of local containers,
4400 if it knows them, plus the interface name of the container.
4402 * A new tool "systemd-escape" has been added that makes it
4403 easy to escape strings to build unit names and similar.
4405 * sd_notify() messages may now include a new ERRNO= field
4406 which is parsed and collected by systemd and shown among the
4407 "systemctl status" output for a service.
4409 * A new component "systemd-firstboot" has been added that
4410 queries the most basic systemd information (timezone,
4411 hostname, root password) interactively on first
4412 boot. Alternatively it may also be used to provision these
4413 things offline on OS images installed into directories.
4415 * The default sysctl.d/ snippets will now set
4417 net.ipv4.conf.default.promote_secondaries=1
4419 This has the benefit of no flushing secondary IP addresses
4420 when primary addresses are removed.
4422 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4423 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4424 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4425 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4426 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4427 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4428 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4429 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4430 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4431 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4432 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4433 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4434 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4435 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4436 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4438 — Berlin, 2014-08-19
4442 * A new tool systemd-sysusers has been added. This tool
4443 creates system users and groups in /etc/passwd and
4444 /etc/group, based on static declarative system user/group
4445 definitions in /usr/lib/sysusers.d/. This is useful to
4446 enable factory resets and volatile systems that boot up with
4447 an empty /etc directory, and thus need system users and
4448 groups created during early boot. systemd now also ships
4449 with two default sysusers.d/ files for the most basic
4450 users and groups systemd and the core operating system
4453 * A new tmpfiles snippet has been added that rebuilds the
4454 essential files in /etc on boot, should they be missing.
4456 * A directive for ensuring automatic clean-up of
4457 /var/cache/man/ has been removed from the default
4458 configuration. This line should now be shipped by the man
4459 implementation. The necessary change has been made to the
4460 man-db implementation. Note that you need to update your man
4461 implementation to one that ships this line, otherwise no
4462 automatic clean-up of /var/cache/man will take place.
4464 * A new condition ConditionNeedsUpdate= has been added that
4465 may conditionalize services to only run when /etc or /var
4466 are "older" than the vendor operating system resources in
4467 /usr. This is useful for reconstructing or updating /etc
4468 after an offline update of /usr or a factory reset, on the
4469 next reboot. Services that want to run once after such an
4470 update or reset should use this condition and order
4471 themselves before the new systemd-update-done.service, which
4472 will mark the two directories as fully updated. A number of
4473 service files have been added making use of this, to rebuild
4474 the udev hardware database, the journald message catalog and
4475 dynamic loader cache (ldconfig). The systemd-sysusers tool
4476 described above also makes use of this now. With this in
4477 place it is now possible to start up a minimal operating
4478 system with /etc empty cleanly. For more information on the
4479 concepts involved see this recent blog story:
4481 http://0pointer.de/blog/projects/stateless.html
4483 * A new system group "input" has been introduced, and all
4484 input device nodes get this group assigned. This is useful
4485 for system-level software to get access to input devices. It
4486 complements what is already done for "audio" and "video".
4488 * systemd-networkd learnt minimal DHCPv4 server support in
4489 addition to the existing DHCPv4 client support. It also
4490 learnt DHCPv6 client and IPv6 Router Solicitation client
4491 support. The DHCPv4 client gained support for static routes
4492 passed in from the server. Note that the [DHCPv4] section
4493 known in older systemd-networkd versions has been renamed to
4494 [DHCP] and is now also used by the DHCPv6 client. Existing
4495 .network files using settings of this section should be
4496 updated, though compatibility is maintained. Optionally, the
4497 client hostname may now be sent to the DHCP server.
4499 * networkd gained support for vxlan virtual networks as well
4500 as tun/tap and dummy devices.
4502 * networkd gained support for automatic allocation of address
4503 ranges for interfaces from a system-wide pool of
4504 addresses. This is useful for dynamically managing a large
4505 number of interfaces with a single network configuration
4506 file. In particular this is useful to easily assign
4507 appropriate IP addresses to the veth links of a large number
4508 of nspawn instances.
4510 * RPM macros for processing sysusers, sysctl and binfmt
4511 drop-in snippets at package installation time have been
4514 * The /etc/os-release file should now be placed in
4515 /usr/lib/os-release. The old location is automatically
4516 created as symlink. /usr/lib is the more appropriate
4517 location of this file, since it shall actually describe the
4518 vendor operating system shipped in /usr, and not the
4519 configuration stored in /etc.
4521 * .mount units gained a new boolean SloppyOptions= setting
4522 that maps to mount(8)'s -s option which enables permissive
4523 parsing of unknown mount options.
4525 * tmpfiles learnt a new "L+" directive which creates a symlink
4526 but (unlike "L") deletes a pre-existing file first, should
4527 it already exist and not already be the correct
4528 symlink. Similarly, "b+", "c+" and "p+" directives have been
4529 added as well, which create block and character devices, as
4530 well as fifos in the filesystem, possibly removing any
4531 pre-existing files of different types.
4533 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4534 'argument' field (which so far specified the source to
4535 symlink/copy the files from) is now optional. If omitted the
4536 same file os copied from /usr/share/factory/ suffixed by the
4537 full destination path. This is useful for populating /etc
4538 with essential files, by copying them from vendor defaults
4539 shipped in /usr/share/factory/etc.
4541 * A new command "systemctl preset-all" has been added that
4542 applies the service preset settings to all installed unit
4543 files. A new switch --preset-mode= has been added that
4544 controls whether only enable or only disable operations
4547 * A new command "systemctl is-system-running" has been added
4548 that allows checking the overall state of the system, for
4549 example whether it is fully up and running.
4551 * When the system boots up with an empty /etc, the equivalent
4552 to "systemctl preset-all" is executed during early boot, to
4553 make sure all default services are enabled after a factory
4556 * systemd now contains a minimal preset file that enables the
4557 most basic services systemd ships by default.
4559 * Unit files' [Install] section gained a new DefaultInstance=
4560 field for defining the default instance to create if a
4561 template unit is enabled with no instance specified.
4563 * A new passive target cryptsetup-pre.target has been added
4564 that may be used by services that need to make they run and
4565 finish before the first LUKS cryptographic device is set up.
4567 * The /dev/loop-control and /dev/btrfs-control device nodes
4568 are now owned by the "disk" group by default, opening up
4569 access to this group.
4571 * systemd-coredump will now automatically generate a
4572 stack trace of all core dumps taking place on the system,
4573 based on elfutils' libdw library. This stack trace is logged
4576 * systemd-coredump may now optionally store coredumps directly
4577 on disk (in /var/lib/systemd/coredump, possibly compressed),
4578 instead of storing them unconditionally in the journal. This
4579 mode is the new default. A new configuration file
4580 /etc/systemd/coredump.conf has been added to configure this
4581 and other parameters of systemd-coredump.
4583 * coredumpctl gained a new "info" verb to show details about a
4584 specific coredump. A new switch "-1" has also been added
4585 that makes sure to only show information about the most
4586 recent entry instead of all entries. Also, as the tool is
4587 generally useful now the "systemd-" prefix of the binary
4588 name has been removed. Distributions that want to maintain
4589 compatibility with the old name should add a symlink from
4590 the old name to the new name.
4592 * journald's SplitMode= now defaults to "uid". This makes sure
4593 that unprivileged users can access their own coredumps with
4594 coredumpctl without restrictions.
4596 * New kernel command line options "systemd.wants=" (for
4597 pulling an additional unit during boot), "systemd.mask="
4598 (for masking a specific unit for the boot), and
4599 "systemd.debug-shell" (for enabling the debug shell on tty9)
4600 have been added. This is implemented in the new generator
4601 "systemd-debug-generator".
4603 * systemd-nspawn will now by default filter a couple of
4604 syscalls for containers, among them those required for
4605 kernel module loading, direct x86 IO port access, swap
4606 management, and kexec. Most importantly though
4607 open_by_handle_at() is now prohibited for containers,
4608 closing a hole similar to a recently discussed vulnerability
4609 in docker regarding access to files on file hierarchies the
4610 container should normally not have access to. Note that, for
4611 nspawn, we generally make no security claims anyway (and
4612 this is explicitly documented in the man page), so this is
4613 just a fix for one of the most obvious problems.
4615 * A new man page file-hierarchy(7) has been added that
4616 contains a minimized, modernized version of the file system
4617 layout systemd expects, similar in style to the FHS
4618 specification or hier(5). A new tool systemd-path(1) has
4619 been added to query many of these paths for the local
4622 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4623 longer done. Since the directory now has a per-user size
4624 limit, and is cleaned on logout this appears unnecessary,
4625 in particular since this now brings the lifecycle of this
4626 directory closer in line with how IPC objects are handled.
4628 * systemd.pc now exports a number of additional directories,
4629 including $libdir (which is useful to identify the library
4630 path for the primary architecture of the system), and a
4631 couple of drop-in directories.
4633 * udev's predictable network interface names now use the dev_port
4634 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4635 distinguish between ports of the same PCI function. dev_id should
4636 only be used for ports using the same HW address, hence the need
4639 * machined has been updated to export the OS version of a
4640 container (read from /etc/os-release and
4641 /usr/lib/os-release) on the bus. This is now shown in
4642 "machinectl status" for a machine.
4644 * A new service setting RestartForceExitStatus= has been
4645 added. If configured to a set of exit signals or process
4646 return values, the service will be restarted when the main
4647 daemon process exits with any of them, regardless of the
4650 * systemctl's -H switch for connecting to remote systemd
4651 machines has been extended so that it may be used to
4652 directly connect to a specific container on the
4653 host. "systemctl -H root@foobar:waldi" will now connect as
4654 user "root" to host "foobar", and then proceed directly to
4655 the container named "waldi". Note that currently you have to
4656 authenticate as user "root" for this to work, as entering
4657 containers is a privileged operation.
4659 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4660 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4661 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4662 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4663 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4664 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4665 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4666 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4667 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4668 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4669 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4670 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4672 — Berlin, 2014-07-03
4676 * As an experimental feature, udev now tries to lock the
4677 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4678 executes events for the disk or any of its partitions.
4679 Applications like partitioning programs can lock the
4680 disk device node (flock(LOCK_EX)) and claim temporary
4681 device ownership that way; udev will entirely skip all event
4682 handling for this disk and its partitions. If the disk
4683 was opened for writing, the close will trigger a partition
4684 table rescan in udev's "watch" facility, and if needed
4685 synthesize "change" events for the disk and all its partitions.
4686 This is now unconditionally enabled, and if it turns out to
4687 cause major problems, we might turn it on only for specific
4688 devices, or might need to disable it entirely. Device Mapper
4689 devices are excluded from this logic.
4691 * We temporarily dropped the "-l" switch for fsck invocations,
4692 since they collide with the flock() logic above. util-linux
4693 upstream has been changed already to avoid this conflict,
4694 and we will readd "-l" as soon as util-linux with this
4695 change has been released.
4697 * The dependency on libattr has been removed. Since a long
4698 time, the extended attribute calls have moved to glibc, and
4699 libattr is thus unnecessary.
4701 * Virtualization detection works without privileges now. This
4702 means the systemd-detect-virt binary no longer requires
4703 CAP_SYS_PTRACE file capabilities, and our daemons can run
4704 with fewer privileges.
4706 * systemd-networkd now runs under its own "systemd-network"
4707 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4708 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4709 loses the ability to write to files owned by root this way.
4711 * Similarly, systemd-resolved now runs under its own
4712 "systemd-resolve" user with no capabilities remaining.
4714 * Similarly, systemd-bus-proxyd now runs under its own
4715 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4717 * systemd-networkd gained support for setting up "veth"
4718 virtual Ethernet devices for container connectivity, as well
4719 as GRE and VTI tunnels.
4721 * systemd-networkd will no longer automatically attempt to
4722 manually load kernel modules necessary for certain tunnel
4723 transports. Instead, it is assumed the kernel loads them
4724 automatically when required. This only works correctly on
4725 very new kernels. On older kernels, please consider adding
4726 the kernel modules to /etc/modules-load.d/ as a work-around.
4728 * The resolv.conf file systemd-resolved generates has been
4729 moved to /run/systemd/resolve/. If you have a symlink from
4730 /etc/resolv.conf, it might be necessary to correct it.
4732 * Two new service settings, ProtectHome= and ProtectSystem=,
4733 have been added. When enabled, they will make the user data
4734 (such as /home) inaccessible or read-only and the system
4735 (such as /usr) read-only, for specific services. This allows
4736 very light-weight per-service sandboxing to avoid
4737 modifications of user data or system files from
4738 services. These two new switches have been enabled for all
4739 of systemd's long-running services, where appropriate.
4741 * Socket units gained new SocketUser= and SocketGroup=
4742 settings to set the owner user and group of AF_UNIX sockets
4743 and FIFOs in the file system.
4745 * Socket units gained a new RemoveOnStop= setting. If enabled,
4746 all FIFOS and sockets in the file system will be removed
4747 when the specific socket unit is stopped.
4749 * Socket units gained a new Symlinks= setting. It takes a list
4750 of symlinks to create to file system sockets or FIFOs
4751 created by the specific Unix sockets. This is useful to
4752 manage symlinks to socket nodes with the same life-cycle as
4755 * The /dev/log socket and /dev/initctl FIFO have been moved to
4756 /run, and have been replaced by symlinks. This allows
4757 connecting to these facilities even if PrivateDevices=yes is
4758 used for a service (which makes /dev/log itself unavailable,
4759 but /run is left). This also has the benefit of ensuring
4760 that /dev only contains device nodes, directories and
4761 symlinks, and nothing else.
4763 * sd-daemon gained two new calls sd_pid_notify() and
4764 sd_pid_notifyf(). They are similar to sd_notify() and
4765 sd_notifyf(), but allow overriding of the source PID of
4766 notification messages if permissions permit this. This is
4767 useful to send notify messages on behalf of a different
4768 process (for example, the parent process). The
4769 systemd-notify tool has been updated to make use of this
4770 when sending messages (so that notification messages now
4771 originate from the shell script invoking systemd-notify and
4772 not the systemd-notify process itself. This should minimize
4773 a race where systemd fails to associate notification
4774 messages to services when the originating process already
4777 * A new "on-abnormal" setting for Restart= has been added. If
4778 set, it will result in automatic restarts on all "abnormal"
4779 reasons for a process to exit, which includes unclean
4780 signals, core dumps, timeouts and watchdog timeouts, but
4781 does not include clean and unclean exit codes or clean
4782 signals. Restart=on-abnormal is an alternative for
4783 Restart=on-failure for services that shall be able to
4784 terminate and avoid restarts on certain errors, by
4785 indicating so with an unclean exit code. Restart=on-failure
4786 or Restart=on-abnormal is now the recommended setting for
4787 all long-running services.
4789 * If the InaccessibleDirectories= service setting points to a
4790 mount point (or if there are any submounts contained within
4791 it), it is now attempted to completely unmount it, to make
4792 the file systems truly unavailable for the respective
4795 * The ReadOnlyDirectories= service setting and
4796 systemd-nspawn's --read-only parameter are now recursively
4797 applied to all submounts, too.
4799 * Mount units may now be created transiently via the bus APIs.
4801 * The support for SysV and LSB init scripts has been removed
4802 from the systemd daemon itself. Instead, it is now
4803 implemented as a generator that creates native systemd units
4804 from these scripts when needed. This enables us to remove a
4805 substantial amount of legacy code from PID 1, following the
4806 fact that many distributions only ship a very small number
4807 of LSB/SysV init scripts nowadays.
4809 * Privileged Xen (dom0) domains are not considered
4810 virtualization anymore by the virtualization detection
4811 logic. After all, they generally have unrestricted access to
4812 the hardware and usually are used to manage the unprivileged
4815 * systemd-tmpfiles gained a new "C" line type, for copying
4816 files or entire directories.
4818 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4819 lines. So far, they have been non-globbing versions of the
4820 latter, and have thus been redundant. In future, it is
4821 recommended to only use "z". "m" has hence been removed
4822 from the documentation, even though it stays supported.
4824 * A tmpfiles snippet to recreate the most basic structure in
4825 /var has been added. This is enough to create the /var/run →
4826 /run symlink and create a couple of structural
4827 directories. This allows systems to boot up with an empty or
4828 volatile /var. Of course, while with this change, the core OS
4829 now is capable with dealing with a volatile /var, not all
4830 user services are ready for it. However, we hope that sooner
4831 or later, many service daemons will be changed upstream so
4832 that they are able to automatically create their necessary
4833 directories in /var at boot, should they be missing. This is
4834 the first step to allow state-less systems that only require
4835 the vendor image for /usr to boot.
4837 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4838 empty tmpfs instance to a specific directory. This is
4839 particularly useful for making use of the automatic
4840 reconstruction of /var (see above), by passing --tmpfs=/var.
4842 * Access modes specified in tmpfiles snippets may now be
4843 prefixed with "~", which indicates that they shall be masked
4844 by whether the existing file or directory is currently
4845 writable, readable or executable at all. Also, if specified,
4846 the sgid/suid/sticky bits will be masked for all
4849 * A new passive target unit "network-pre.target" has been
4850 added which is useful for services that shall run before any
4851 network is configured, for example firewall scripts.
4853 * The "floppy" group that previously owned the /dev/fd*
4854 devices is no longer used. The "disk" group is now used
4855 instead. Distributions should probably deprecate usage of
4858 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4859 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4860 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4861 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4862 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4863 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4866 — Berlin, 2014-06-11
4870 * A new "systemd-timesyncd" daemon has been added for
4871 synchronizing the system clock across the network. It
4872 implements an SNTP client. In contrast to NTP
4873 implementations such as chrony or the NTP reference server,
4874 this only implements a client side, and does not bother with
4875 the full NTP complexity, focusing only on querying time from
4876 one remote server and synchronizing the local clock to
4877 it. Unless you intend to serve NTP to networked clients or
4878 want to connect to local hardware clocks, this simple NTP
4879 client should be more than appropriate for most
4880 installations. The daemon runs with minimal privileges, and
4881 has been hooked up with networkd to only operate when
4882 network connectivity is available. The daemon saves the
4883 current clock to disk every time a new NTP sync has been
4884 acquired, and uses this to possibly correct the system clock
4885 early at bootup, in order to accommodate for systems that
4886 lack an RTC such as the Raspberry Pi and embedded devices,
4887 and to make sure that time monotonically progresses on these
4888 systems, even if it is not always correct. To make use of
4889 this daemon, a new system user and group "systemd-timesync"
4890 needs to be created on installation of systemd.
4892 * The queue "seqnum" interface of libudev has been disabled, as
4893 it was generally incompatible with device namespacing as
4894 sequence numbers of devices go "missing" if the devices are
4895 part of a different namespace.
4897 * "systemctl list-timers" and "systemctl list-sockets" gained
4898 a --recursive switch for showing units of these types also
4899 for all local containers, similar in style to the already
4900 supported --recursive switch for "systemctl list-units".
4902 * A new RebootArgument= setting has been added for service
4903 units, which may be used to specify a kernel reboot argument
4904 to use when triggering reboots with StartLimitAction=.
4906 * A new FailureAction= setting has been added for service
4907 units which may be used to specify an operation to trigger
4908 when a service fails. This works similarly to
4909 StartLimitAction=, but unlike it, controls what is done
4910 immediately rather than only after several attempts to
4911 restart the service in question.
4913 * hostnamed got updated to also expose the kernel name,
4914 release, and version on the bus. This is useful for
4915 executing commands like hostnamectl with the -H switch.
4916 systemd-analyze makes use of this to properly display
4917 details when running non-locally.
4919 * The bootchart tool can now show cgroup information in the
4920 graphs it generates.
4922 * The CFS CPU quota cgroup attribute is now exposed for
4923 services. The new CPUQuota= switch has been added for this
4924 which takes a percentage value. Setting this will have the
4925 result that a service may never get more CPU time than the
4926 specified percentage, even if the machine is otherwise idle.
4928 * systemd-networkd learned IPIP and SIT tunnel support.
4930 * LSB init scripts exposing a dependency on $network will now
4931 get a dependency on network-online.target rather than simply
4932 network.target. This should bring LSB handling closer to
4933 what it was on SysV systems.
4935 * A new fsck.repair= kernel option has been added to control
4936 how fsck shall deal with unclean file systems at boot.
4938 * The (.ini) configuration file parser will now silently
4939 ignore sections whose name begins with "X-". This may be
4940 used to maintain application-specific extension sections in unit
4943 * machined gained a new API to query the IP addresses of
4944 registered containers. "machinectl status" has been updated
4945 to show these addresses in its output.
4947 * A new call sd_uid_get_display() has been added to the
4948 sd-login APIs for querying the "primary" session of a
4949 user. The "primary" session of the user is elected from the
4950 user's sessions and generally a graphical session is
4951 preferred over a text one.
4953 * A minimal systemd-resolved daemon has been added. It
4954 currently simply acts as a companion to systemd-networkd and
4955 manages resolv.conf based on per-interface DNS
4956 configuration, possibly supplied via DHCP. In the long run
4957 we hope to extend this into a local DNSSEC enabled DNS and
4960 * The systemd-networkd-wait-online tool is now enabled by
4961 default. It will delay network-online.target until a network
4962 connection has been configured. The tool primarily integrates
4963 with networkd, but will also make a best effort to make sense
4964 of network configuration performed in some other way.
4966 * Two new service options StartupCPUShares= and
4967 StartupBlockIOWeight= have been added that work similarly to
4968 CPUShares= and BlockIOWeight= however only apply during
4969 system startup. This is useful to prioritize certain services
4970 differently during bootup than during normal runtime.
4972 * hostnamed has been changed to prefer the statically
4973 configured hostname in /etc/hostname (unless set to
4974 'localhost' or empty) over any dynamic one supplied by
4975 dhcp. With this change, the rules for picking the hostname
4976 match more closely the rules of other configuration settings
4977 where the local administrator's configuration in /etc always
4978 overrides any other settings.
4980 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4981 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4982 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4983 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4984 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4985 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4986 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4987 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4988 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4989 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4990 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4991 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4992 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4993 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4994 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4995 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4998 — Beijing, 2014-05-28
5002 * When restoring the screen brightness at boot, stay away from
5003 the darkest setting or from the lowest 5% of the available
5004 range, depending on which is the larger value of both. This
5005 should effectively protect the user from rebooting into a
5006 black screen, should the brightness have been set to minimum
5009 * sd-login gained a new sd_machine_get_class() call to
5010 determine the class ("vm" or "container") of a machine
5011 registered with machined.
5013 * sd-login gained new calls
5014 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5015 to query the identity of the peer of a local AF_UNIX
5016 connection. They operate similarly to their sd_pid_get_xyz()
5019 * PID 1 will now maintain a system-wide system state engine
5020 with the states "starting", "running", "degraded",
5021 "maintenance", "stopping". These states are bound to system
5022 startup, normal runtime, runtime with at least one failed
5023 service, rescue/emergency mode and system shutdown. This
5024 state is shown in the "systemctl status" output when no unit
5025 name is passed. It is useful to determine system state, in
5026 particularly when doing so for many systems or containers at
5029 * A new command "list-machines" has been added to "systemctl"
5030 that lists all local OS containers and shows their system
5031 state (see above), if systemd runs inside of them.
5033 * systemctl gained a new "-r" switch to recursively enumerate
5034 units on all local containers, when used with the
5035 "list-unit" command (which is the default one that is
5036 executed when no parameters are specified).
5038 * The GPT automatic partition discovery logic will now honour
5039 two GPT partition flags: one may be set on a partition to
5040 cause it to be mounted read-only, and the other may be set
5041 on a partition to ignore it during automatic discovery.
5043 * Two new GPT type UUIDs have been added for automatic root
5044 partition discovery, for 32-bit and 64-bit ARM. This is not
5045 particularly useful for discovering the root directory on
5046 these architectures during bare-metal boots (since UEFI is
5047 not common there), but still very useful to allow booting of
5048 ARM disk images in nspawn with the -i option.
5050 * MAC addresses of interfaces created with nspawn's
5051 --network-interface= switch will now be generated from the
5052 machine name, and thus be stable between multiple invocations
5055 * logind will now automatically remove all IPC objects owned
5056 by a user if she or he fully logs out. This makes sure that
5057 users who are logged out cannot continue to consume IPC
5058 resources. This covers SysV memory, semaphores and message
5059 queues as well as POSIX shared memory and message
5060 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5061 limits. With this functionality, that is corrected. This may
5062 be turned off by using the RemoveIPC= switch of logind.conf.
5064 * The systemd-machine-id-setup and tmpfiles tools gained a
5065 --root= switch to operate on a specific root directory,
5068 * journald can now forward logged messages to the TTYs of all
5069 logged in users ("wall"). This is the default for all
5070 emergency messages now.
5072 * A new tool systemd-journal-remote has been added to stream
5073 journal log messages across the network.
5075 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5076 controller trees are mounted into it. Note that the
5077 directories mounted beneath it are not read-only. This is a
5078 security measure and is particularly useful because glibc
5079 actually includes a search logic to pick any tmpfs it can
5080 find to implement shm_open() if /dev/shm is not available
5081 (which it might very well be in namespaced setups).
5083 * machinectl gained a new "poweroff" command to cleanly power
5084 down a local OS container.
5086 * The PrivateDevices= unit file setting will now also drop the
5087 CAP_MKNOD capability from the capability bound set, and
5088 imply DevicePolicy=closed.
5090 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5091 comprehensively on all long-running systemd services where
5092 this is appropriate.
5094 * systemd-udevd will now run in a disassociated mount
5095 namespace. To mount directories from udev rules, make sure to
5096 pull in mount units via SYSTEMD_WANTS properties.
5098 * The kdbus support gained support for uploading policy into
5099 the kernel. sd-bus gained support for creating "monitoring"
5100 connections that can eavesdrop into all bus communication
5101 for debugging purposes.
5103 * Timestamps may now be specified in seconds since the UNIX
5104 epoch Jan 1st, 1970 by specifying "@" followed by the value
5107 * Native tcpwrap support in systemd has been removed. tcpwrap
5108 is old code, not really maintained anymore and has serious
5109 shortcomings, and better options such as firewalls
5110 exist. For setups that require tcpwrap usage, please
5111 consider invoking your socket-activated service via tcpd,
5112 like on traditional inetd.
5114 * A new system.conf configuration option
5115 DefaultTimerAccuracySec= has been added that controls the
5116 default AccuracySec= setting of .timer units.
5118 * Timer units gained a new WakeSystem= switch. If enabled,
5119 timers configured this way will cause the system to resume
5120 from system suspend (if the system supports that, which most
5123 * Timer units gained a new Persistent= switch. If enabled,
5124 timers configured this way will save to disk when they have
5125 been last triggered. This information is then used on next
5126 reboot to possible execute overdue timer events, that
5127 could not take place because the system was powered off.
5128 This enables simple anacron-like behaviour for timer units.
5130 * systemctl's "list-timers" will now also list the time a
5131 timer unit was last triggered in addition to the next time
5132 it will be triggered.
5134 * systemd-networkd will now assign predictable IPv4LL
5135 addresses to its local interfaces.
5137 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5138 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5139 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5140 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5141 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5142 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5143 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5144 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5147 — Berlin, 2014-03-25
5151 * A new unit file setting RestrictAddressFamilies= has been
5152 added to restrict which socket address families unit
5153 processes gain access to. This takes address family names
5154 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5155 attack surface of services via exotic protocol stacks. This
5156 is built on seccomp system call filters.
5158 * Two new unit file settings RuntimeDirectory= and
5159 RuntimeDirectoryMode= have been added that may be used to
5160 manage a per-daemon runtime directories below /run. This is
5161 an alternative for setting up directory permissions with
5162 tmpfiles snippets, and has the advantage that the runtime
5163 directory's lifetime is bound to the daemon runtime and that
5164 the daemon starts up with an empty directory each time. This
5165 is particularly useful when writing services that drop
5166 privileges using the User= or Group= setting.
5168 * The DeviceAllow= unit setting now supports globbing for
5169 matching against device group names.
5171 * The systemd configuration file system.conf gained new
5172 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5173 DefaultMemoryAccounting= to globally turn on/off accounting
5174 for specific resources (cgroups) for all units. These
5175 settings may still be overridden individually in each unit
5178 * systemd-gpt-auto-generator is now able to discover /srv and
5179 root partitions in addition to /home and swap partitions. It
5180 also supports LUKS-encrypted partitions now. With this in
5181 place, automatic discovery of partitions to mount following
5182 the Discoverable Partitions Specification
5183 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5184 is now a lot more complete. This allows booting without
5185 /etc/fstab and without root= on the kernel command line on
5186 systems prepared appropriately.
5188 * systemd-nspawn gained a new --image= switch which allows
5189 booting up disk images and Linux installations on any block
5190 device that follow the Discoverable Partitions Specification
5191 (see above). This means that installations made with
5192 appropriately updated installers may now be started and
5193 deployed using container managers, completely
5194 unmodified. (We hope that libvirt-lxc will add support for
5195 this feature soon, too.)
5197 * systemd-nspawn gained a new --network-macvlan= setting to
5198 set up a private macvlan interface for the
5199 container. Similarly, systemd-networkd gained a new
5200 Kind=macvlan setting in .netdev files.
5202 * systemd-networkd now supports configuring local addresses
5205 * A new tool systemd-network-wait-online has been added to
5206 synchronously wait for network connectivity using
5209 * The sd-bus.h bus API gained a new sd_bus_track object for
5210 tracking the life-cycle of bus peers. Note that sd-bus.h is
5211 still not a public API though (unless you specify
5212 --enable-kdbus on the configure command line, which however
5213 voids your warranty and you get no API stability guarantee).
5215 * The $XDG_RUNTIME_DIR runtime directories for each user are
5216 now individual tmpfs instances, which has the benefit of
5217 introducing separate pools for each user, with individual
5218 size limits, and thus making sure that unprivileged clients
5219 can no longer negatively impact the system or other users by
5220 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5221 RuntimeDirectorySize= has been introduced that allows
5222 controlling the default size limit for all users. It
5223 defaults to 10% of the available physical memory. This is no
5224 replacement for quotas on tmpfs though (which the kernel
5225 still does not support), as /dev/shm and /tmp are still
5226 shared resources used by both the system and unprivileged
5229 * logind will now automatically turn off automatic suspending
5230 on laptop lid close when more than one display is
5231 connected. This was previously expected to be implemented
5232 individually in desktop environments (such as GNOME),
5233 however has been added to logind now, in order to fix a
5234 boot-time race where a desktop environment might not have
5235 been started yet and thus not been able to take an inhibitor
5236 lock at the time where logind already suspends the system
5237 due to a closed lid.
5239 * logind will now wait at least 30s after each system
5240 suspend/resume cycle, and 3min after system boot before
5241 suspending the system due to a closed laptop lid. This
5242 should give USB docking stations and similar enough time to
5243 be probed and configured after system resume and boot in
5244 order to then act as suspend blocker.
5246 * systemd-run gained a new --property= setting which allows
5247 initialization of resource control properties (and others)
5248 for the created scope or service unit. Example: "systemd-run
5249 --property=BlockIOWeight=10 updatedb" may be used to run
5250 updatedb at a low block IO scheduling weight.
5252 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5253 now also work in --scope mode.
5255 * When systemd is compiled with kdbus support, basic support
5256 for enforced policies is now in place. (Note that enabling
5257 kdbus still voids your warranty and no API compatibility
5260 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5261 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5262 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5263 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5264 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5265 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5266 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5267 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5268 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5269 Zbigniew Jędrzejewski-Szmek
5271 — Berlin, 2014-03-12
5275 * systemd will now relabel /dev after loading the SMACK policy
5276 according to SMACK rules.
5278 * A new unit file option AppArmorProfile= has been added to
5279 set the AppArmor profile for the processes of a unit.
5281 * A new condition check ConditionArchitecture= has been added
5282 to conditionalize units based on the system architecture, as
5283 reported by uname()'s "machine" field.
5285 * systemd-networkd now supports matching on the system
5286 virtualization, architecture, kernel command line, host name
5289 * logind is now a lot more aggressive when suspending the
5290 machine due to a closed laptop lid. Instead of acting only
5291 on the lid close action, it will continuously watch the lid
5292 status and act on it. This is useful for laptops where the
5293 power button is on the outside of the chassis so that it can
5294 be reached without opening the lid (such as the Lenovo
5295 Yoga). On those machines, logind will now immediately
5296 re-suspend the machine if the power button has been
5297 accidentally pressed while the laptop was suspended and in a
5298 backpack or similar.
5300 * logind will now watch SW_DOCK switches and inhibit reaction
5301 to the lid switch if it is pressed. This means that logind
5302 will not suspend the machine anymore if the lid is closed
5303 and the system is docked, if the laptop supports SW_DOCK
5304 notifications via the input layer. Note that ACPI docking
5305 stations do not generate this currently. Also note that this
5306 logic is usually not fully sufficient and Desktop
5307 Environments should take a lid switch inhibitor lock when an
5308 external display is connected, as systemd will not watch
5311 * nspawn will now make use of the devices cgroup controller by
5312 default, and only permit creation of and access to the usual
5313 API device nodes like /dev/null or /dev/random, as well as
5314 access to (but not creation of) the pty devices.
5316 * We will now ship a default .network file for
5317 systemd-networkd that automatically configures DHCP for
5318 network interfaces created by nspawn's --network-veth or
5319 --network-bridge= switches.
5321 * systemd will now understand the usual M, K, G, T suffixes
5322 according to SI conventions (i.e. to the base 1000) when
5323 referring to throughput and hardware metrics. It will stay
5324 with IEC conventions (i.e. to the base 1024) for software
5325 metrics, according to what is customary according to
5326 Wikipedia. We explicitly document which base applies for
5327 each configuration option.
5329 * The DeviceAllow= setting in unit files now supports a syntax
5330 to whitelist an entire group of devices node majors at once,
5331 based on the /proc/devices listing. For example, with the
5332 string "char-pts", it is now possible to whitelist all
5333 current and future pseudo-TTYs at once.
5335 * sd-event learned a new "post" event source. Event sources of
5336 this type are triggered by the dispatching of any event
5337 source of a type that is not "post". This is useful for
5338 implementing clean-up and check event sources that are
5339 triggered by other work being done in the program.
5341 * systemd-networkd is no longer statically enabled, but uses
5342 the usual [Install] sections so that it can be
5343 enabled/disabled using systemctl. It still is enabled by
5346 * When creating a veth interface pair with systemd-nspawn, the
5347 host side will now be prefixed with "vb-" if
5348 --network-bridge= is used, and with "ve-" if --network-veth
5349 is used. This way, it is easy to distinguish these cases on
5350 the host, for example to apply different configuration to
5351 them with systemd-networkd.
5353 * The compatibility libraries for libsystemd-journal.so,
5354 libsystem-id128.so, libsystemd-login.so and
5355 libsystemd-daemon.so do not make use of IFUNC
5356 anymore. Instead, we now build libsystemd.so multiple times
5357 under these alternative names. This means that the footprint
5358 is drastically increased, but given that these are
5359 transitional compatibility libraries, this should not matter
5360 much. This change has been made necessary to support the ARM
5361 platform for these compatibility libraries, as the ARM
5362 toolchain is not really at the same level as the toolchain
5363 for other architectures like x86 and does not support
5364 IFUNC. Please make sure to use --enable-compat-libs only
5365 during a transitional period!
5367 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5368 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5369 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5370 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5371 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5372 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5373 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5374 Zbigniew Jędrzejewski-Szmek
5376 — Berlin, 2014-02-24
5380 * A new component "systemd-networkd" has been added that can
5381 be used to configure local network interfaces statically or
5382 via DHCP. It is capable of bringing up bridges, VLANs, and
5383 bonding. Currently, no hook-ups for interactive network
5384 configuration are provided. Use this for your initrd,
5385 container, embedded, or server setup if you need a simple,
5386 yet powerful, network configuration solution. This
5387 configuration subsystem is quite nifty, as it allows wildcard
5388 hotplug matching in interfaces. For example, with a single
5389 configuration snippet, you can configure that all Ethernet
5390 interfaces showing up are automatically added to a bridge,
5391 or similar. It supports link-sensing and more.
5393 * A new tool "systemd-socket-proxyd" has been added which can
5394 act as a bidirectional proxy for TCP sockets. This is
5395 useful for adding socket activation support to services that
5396 do not actually support socket activation, including virtual
5397 machines and the like.
5399 * Add a new tool to save/restore rfkill state on
5402 * Save/restore state of keyboard backlights in addition to
5403 display backlights on shutdown/boot.
5405 * udev learned a new SECLABEL{} construct to label device
5406 nodes with a specific security label when they appear. For
5407 now, only SECLABEL{selinux} is supported, but the syntax is
5408 prepared for additional security frameworks.
5410 * udev gained a new scheme to configure link-level attributes
5411 from files in /etc/systemd/network/*.link. These files can
5412 match against MAC address, device path, driver name and type,
5413 and will apply attributes like the naming policy, link speed,
5414 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5415 address assignment policy (randomized, ...).
5417 * The configuration of network interface naming rules for
5418 "permanent interface names" has changed: a new NamePolicy=
5419 setting in the [Link] section of .link files determines the
5420 priority of possible naming schemes (onboard, slot, MAC,
5421 path). The default value of this setting is determined by
5422 /usr/lib/net/links/99-default.link. Old
5423 80-net-name-slot.rules udev configuration file has been
5424 removed, so local configuration overriding this file should
5425 be adapted to override 99-default.link instead.
5427 * When the User= switch is used in a unit file, also
5428 initialize $SHELL= based on the user database entry.
5430 * systemd no longer depends on libdbus. All communication is
5431 now done with sd-bus, systemd's low-level bus library
5434 * kdbus support has been added to PID 1 itself. When kdbus is
5435 enabled, this causes PID 1 to set up the system bus and
5436 enable support for a new ".busname" unit type that
5437 encapsulates bus name activation on kdbus. It works a little
5438 bit like ".socket" units, except for bus names. A new
5439 generator has been added that converts classic dbus1 service
5440 activation files automatically into native systemd .busname
5443 * sd-bus: add a light-weight vtable implementation that allows
5444 defining objects on the bus with a simple static const
5445 vtable array of its methods, signals and properties.
5447 * systemd will not generate or install static dbus
5448 introspection data anymore to /usr/share/dbus-1/interfaces,
5449 as the precise format of these files is unclear, and
5450 nothing makes use of it.
5452 * A proxy daemon is now provided to proxy clients connecting
5453 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5454 compatibility with classic D-Bus.
5456 * A bus driver implementation has been added that supports the
5457 classic D-Bus bus driver calls on kdbus, also for
5458 compatibility purposes.
5460 * A new API "sd-event.h" has been added that implements a
5461 minimal event loop API built around epoll. It provides a
5462 couple of features that direct epoll usage is lacking:
5463 prioritization of events, scales to large numbers of timer
5464 events, per-event timer slack (accuracy), system-wide
5465 coalescing of timer events, exit handlers, watchdog
5466 supervision support using systemd's sd_notify() API, child
5469 * A new API "sd-rntl.h" has been added that provides an API
5470 around the route netlink interface of the kernel, similar in
5471 style to "sd-bus.h".
5473 * A new API "sd-dhcp-client.h" has been added that provides a
5474 small DHCPv4 client-side implementation. This is used by
5477 * There is a new kernel command line option
5478 "systemd.restore_state=0|1". When set to "0", none of the
5479 systemd tools will restore saved runtime state to hardware
5480 devices. More specifically, the rfkill and backlight states
5483 * The FsckPassNo= compatibility option in mount/service units
5484 has been removed. The fstab generator will now add the
5485 necessary dependencies automatically, and does not require
5486 PID1's support for that anymore.
5488 * journalctl gained a new switch, --list-boots, that lists
5489 recent boots with their times and boot IDs.
5491 * The various tools like systemctl, loginctl, timedatectl,
5492 busctl, systemd-run, ... have gained a new switch "-M" to
5493 connect to a specific, local OS container (as direct
5494 connection, without requiring SSH). This works on any
5495 container that is registered with machined, such as those
5496 created by libvirt-lxc or nspawn.
5498 * systemd-run and systemd-analyze also gained support for "-H"
5499 to connect to remote hosts via SSH. This is particularly
5500 useful for systemd-run because it enables queuing of jobs
5501 onto remote systems.
5503 * machinectl gained a new command "login" to open a getty
5504 login in any local container. This works with any container
5505 that is registered with machined (such as those created by
5506 libvirt-lxc or nspawn), and which runs systemd inside.
5508 * machinectl gained a new "reboot" command that may be used to
5509 trigger a reboot on a specific container that is registered
5510 with machined. This works on any container that runs an init
5511 system of some kind.
5513 * systemctl gained a new "list-timers" command to print a nice
5514 listing of installed timer units with the times they elapse
5517 * Alternative reboot() parameters may now be specified on the
5518 "systemctl reboot" command line and are passed to the
5519 reboot() system call.
5521 * systemctl gained a new --job-mode= switch to configure the
5522 mode to queue a job with. This is a more generic version of
5523 --fail, --irreversible, and --ignore-dependencies, which are
5524 still available but not advertised anymore.
5526 * /etc/systemd/system.conf gained new settings to configure
5527 various default timeouts of units, as well as the default
5528 start limit interval and burst. These may still be overridden
5531 * PID1 will now export on the bus profile data of the security
5532 policy upload process (such as the SELinux policy upload to
5535 * journald: when forwarding logs to the console, include
5536 timestamps (following the setting in
5537 /sys/module/printk/parameters/time).
5539 * OnCalendar= in timer units now understands the special
5540 strings "yearly" and "annually". (Both are equivalent)
5542 * The accuracy of timer units is now configurable with the new
5543 AccuracySec= setting. It defaults to 1min.
5545 * A new dependency type JoinsNamespaceOf= has been added that
5546 allows running two services within the same /tmp and network
5547 namespace, if PrivateNetwork= or PrivateTmp= are used.
5549 * A new command "cat" has been added to systemctl. It outputs
5550 the original unit file of a unit, and concatenates the
5551 contents of additional "drop-in" unit file snippets, so that
5552 the full configuration is shown.
5554 * systemctl now supports globbing on the various "list-xyz"
5555 commands, like "list-units" or "list-sockets", as well as on
5556 those commands which take multiple unit names.
5558 * journalctl's --unit= switch gained support for globbing.
5560 * All systemd daemons now make use of the watchdog logic so
5561 that systemd automatically notices when they hang.
5563 * If the $container_ttys environment variable is set,
5564 getty-generator will automatically spawn a getty for each
5565 listed tty. This is useful for container managers to request
5566 login gettys to be spawned on as many ttys as needed.
5568 * %h, %s, %U specifier support is not available anymore when
5569 used in unit files for PID 1. This is because NSS calls are
5570 not safe from PID 1. They stay available for --user
5571 instances of systemd, and as special case for the root user.
5573 * loginctl gained a new "--no-legend" switch to turn off output
5576 * The "sd-login.h" API gained three new calls:
5577 sd_session_is_remote(), sd_session_get_remote_user(),
5578 sd_session_get_remote_host() to query information about
5581 * The udev hardware database now also carries vendor/product
5582 information of SDIO devices.
5584 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5585 determine whether watchdog notifications are requested by
5588 * Socket-activated per-connection services now include a
5589 short description of the connection parameters in the
5592 * tmpfiles gained a new "--boot" option. When this is not used,
5593 only lines where the command character is not suffixed with
5594 "!" are executed. When this option is specified, those
5595 options are executed too. This partitions tmpfiles
5596 directives into those that can be safely executed at any
5597 time, and those which should be run only at boot (for
5598 example, a line that creates /run/nologin).
5600 * A new API "sd-resolve.h" has been added which provides a simple
5601 asynchronous wrapper around glibc NSS host name resolution
5602 calls, such as getaddrinfo(). In contrast to glibc's
5603 getaddrinfo_a(), it does not use signals. In contrast to most
5604 other asynchronous name resolution libraries, this one does
5605 not reimplement DNS, but reuses NSS, so that alternate
5606 host name resolution systems continue to work, such as mDNS,
5607 LDAP, etc. This API is based on libasyncns, but it has been
5608 cleaned up for inclusion in systemd.
5610 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5611 "sd-daemon.h" are no longer found in individual libraries
5612 libsystemd-journal.so, libsystemd-login.so,
5613 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5614 merged them into a single library, libsystemd.so, which
5615 provides all symbols. The reason for this is cyclic
5616 dependencies, as these libraries tend to use each other's
5617 symbols. So far, we have managed to workaround that by linking
5618 a copy of a good part of our code into each of these
5619 libraries again and again, which, however, makes certain
5620 things hard to do, like sharing static variables. Also, it
5621 substantially increases footprint. With this change, there
5622 is only one library for the basic APIs systemd
5623 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5624 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5625 library as well, however are subject to the --enable-kdbus
5626 switch (see below). Note that "sd-dhcp-client.h" is not part
5627 of this library (this is because it only consumes, never
5628 provides, services of/to other APIs). To make the transition
5629 easy from the separate libraries to the unified one, we
5630 provide the --enable-compat-libs compile-time switch which
5631 will generate stub libraries that are compatible with the
5632 old ones but redirect all calls to the new one.
5634 * All of the kdbus logic and the new APIs "sd-bus.h",
5635 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5636 and "sd-utf8.h" are compile-time optional via the
5637 "--enable-kdbus" switch, and they are not compiled in by
5638 default. To make use of kdbus, you have to explicitly enable
5639 the switch. Note however, that neither the kernel nor the
5640 userspace API for all of this is considered stable yet. We
5641 want to maintain the freedom to still change the APIs for
5642 now. By specifying this build-time switch, you acknowledge
5643 that you are aware of the instability of the current
5646 * Also, note that while kdbus is pretty much complete,
5647 it lacks one thing: proper policy support. This means you
5648 can build a fully working system with all features; however,
5649 it will be highly insecure. Policy support will be added in
5650 one of the next releases, at the same time that we will
5651 declare the APIs stable.
5653 * When the kernel command line argument "kdbus" is specified,
5654 systemd will automatically load the kdbus.ko kernel module. At
5655 this stage of development, it is only useful for testing kdbus
5656 and should not be used in production. Note: if "--enable-kdbus"
5657 is specified, and the kdbus.ko kernel module is available, and
5658 "kdbus" is added to the kernel command line, the entire system
5659 runs with kdbus instead of dbus-daemon, with the above mentioned
5660 problem of missing the system policy enforcement. Also a future
5661 version of kdbus.ko or a newer systemd will not be compatible with
5662 each other, and will unlikely be able to boot the machine if only
5663 one of them is updated.
5665 * systemctl gained a new "import-environment" command which
5666 uploads the caller's environment (or parts thereof) into the
5667 service manager so that it is inherited by services started
5668 by the manager. This is useful to upload variables like
5669 $DISPLAY into the user service manager.
5671 * A new PrivateDevices= switch has been added to service units
5672 which allows running a service with a namespaced /dev
5673 directory that does not contain any device nodes for
5674 physical devices. More specifically, it only includes devices
5675 such as /dev/null, /dev/urandom, and /dev/zero which are API
5678 * logind has been extended to support behaviour like VT
5679 switching on seats that do not support a VT. This makes
5680 multi-session available on seats that are not the first seat
5681 (seat0), and on systems where kernel support for VTs has
5682 been disabled at compile-time.
5684 * If a process holds a delay lock for system sleep or shutdown
5685 and fails to release it in time, we will now log its
5686 identity. This makes it easier to identify processes that
5687 cause slow suspends or power-offs.
5689 * When parsing /etc/crypttab, support for a new key-slot=
5690 option as supported by Debian is added. It allows indicating
5691 which LUKS slot to use on disk, speeding up key loading.
5693 * The sd_journal_sendv() API call has been checked and
5694 officially declared to be async-signal-safe so that it may
5695 be invoked from signal handlers for logging purposes.
5697 * Boot-time status output is now enabled automatically after a
5698 short timeout if boot does not progress, in order to give
5699 the user an indication what she or he is waiting for.
5701 * The boot-time output has been improved to show how much time
5702 remains until jobs expire.
5704 * The KillMode= switch in service units gained a new possible
5705 value "mixed". If set, and the unit is shut down, then the
5706 initial SIGTERM signal is sent only to the main daemon
5707 process, while the following SIGKILL signal is sent to
5708 all remaining processes of the service.
5710 * When a scope unit is registered, a new property "Controller"
5711 may be set. If set to a valid bus name, systemd will send a
5712 RequestStop() signal to this name when it would like to shut
5713 down the scope. This may be used to hook manager logic into
5714 the shutdown logic of scope units. Also, scope units may now
5715 be put in a special "abandoned" state, in which case the
5716 manager process which created them takes no further
5717 responsibilities for it.
5719 * When reading unit files, systemd will now verify
5720 the access mode of these files, and warn about certain
5721 suspicious combinations. This has been added to make it
5722 easier to track down packaging bugs where unit files are
5723 marked executable or world-writable.
5725 * systemd-nspawn gained a new "--setenv=" switch to set
5726 container-wide environment variables. The similar option in
5727 systemd-activate was renamed from "--environment=" to
5728 "--setenv=" for consistency.
5730 * systemd-nspawn has been updated to create a new kdbus domain
5731 for each container that is invoked, thus allowing each
5732 container to have its own set of system and user buses,
5733 independent of the host.
5735 * systemd-nspawn gained a new --drop-capability= switch to run
5736 the container with less capabilities than the default. Both
5737 --drop-capability= and --capability= now take the special
5738 string "all" for dropping or keeping all capabilities.
5740 * systemd-nspawn gained new switches for executing containers
5741 with specific SELinux labels set.
5743 * systemd-nspawn gained a new --quiet switch to not generate
5744 any additional output but the container's own console
5747 * systemd-nspawn gained a new --share-system switch to run a
5748 container without PID namespacing enabled.
5750 * systemd-nspawn gained a new --register= switch to control
5751 whether the container is registered with systemd-machined or
5752 not. This is useful for containers that do not run full
5753 OS images, but only specific apps.
5755 * systemd-nspawn gained a new --keep-unit which may be used
5756 when invoked as the only program from a service unit, and
5757 results in registration of the unit service itself in
5758 systemd-machined, instead of a newly opened scope unit.
5760 * systemd-nspawn gained a new --network-interface= switch for
5761 moving arbitrary interfaces to the container. The new
5762 --network-veth switch creates a virtual Ethernet connection
5763 between host and container. The new --network-bridge=
5764 switch then allows assigning the host side of this virtual
5765 Ethernet connection to a bridge device.
5767 * systemd-nspawn gained a new --personality= switch for
5768 setting the kernel personality for the container. This is
5769 useful when running a 32-bit container on a 64-bit host. A
5770 similar option Personality= is now also available for service
5773 * logind will now also track a "Desktop" identifier for each
5774 session which encodes the desktop environment of it. This is
5775 useful for desktop environments that want to identify
5776 multiple running sessions of itself easily.
5778 * A new SELinuxContext= setting for service units has been
5779 added that allows setting a specific SELinux execution
5780 context for a service.
5782 * Most systemd client tools will now honour $SYSTEMD_LESS for
5783 settings of the "less" pager. By default, these tools will
5784 override $LESS to allow certain operations to work, such as
5785 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5786 influence this logic.
5788 * systemd's "seccomp" hook-up has been changed to make use of
5789 the libseccomp library instead of using its own
5790 implementation. This has benefits for portability among
5793 * For usage together with SystemCallFilter=, a new
5794 SystemCallErrorNumber= setting has been introduced that
5795 allows configuration of a system error number to be returned
5796 on filtered system calls, instead of immediately killing the
5797 process. Also, SystemCallArchitectures= has been added to
5798 limit access to system calls of a particular architecture
5799 (in order to turn off support for unused secondary
5800 architectures). There is also a global
5801 SystemCallArchitectures= setting in system.conf now to turn
5802 off support for non-native system calls system-wide.
5804 * systemd requires a kernel with a working name_to_handle_at(),
5805 please see the kernel config requirements in the README file.
5807 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5808 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5809 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5810 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5811 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5812 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5813 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5814 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5815 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5816 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5817 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5818 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5819 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5820 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5821 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5822 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5823 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5824 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5825 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5826 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5827 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5828 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5829 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5830 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5832 — Berlin, 2014-02-20
5836 * logind has gained support for facilitating privileged input
5837 and drm device access for unprivileged clients. This work is
5838 useful to allow Wayland display servers (and similar
5839 programs, such as kmscon) to run under the user's ID and
5840 access input and drm devices which are normally
5841 protected. When this is used (and the kernel is new enough)
5842 logind will "mute" IO on the file descriptors passed to
5843 Wayland as long as it is in the background and "unmute" it
5844 if it returns into the foreground. This allows secure
5845 session switching without allowing background sessions to
5846 eavesdrop on input and display data. This also introduces
5847 session switching support if VT support is turned off in the
5848 kernel, and on seats that are not seat0.
5850 * A new kernel command line option luks.options= is understood
5851 now which allows specifying LUKS options for usage for LUKS
5852 encrypted partitions specified with luks.uuid=.
5854 * tmpfiles.d(5) snippets may now use specifier expansion in
5855 path names. More specifically %m, %b, %H, %v, are now
5856 replaced by the local machine id, boot id, hostname, and
5857 kernel version number.
5859 * A new tmpfiles.d(5) command "m" has been introduced which
5860 may be used to change the owner/group/access mode of a file
5861 or directory if it exists, but do nothing if it does not.
5863 * This release removes high-level support for the
5864 MemorySoftLimit= cgroup setting. The underlying kernel
5865 cgroup attribute memory.soft_limit= is currently badly
5866 designed and likely to be removed from the kernel API in its
5867 current form, hence we should not expose it for now.
5869 * The memory.use_hierarchy cgroup attribute is now enabled for
5870 all cgroups systemd creates in the memory cgroup
5871 hierarchy. This option is likely to be come the built-in
5872 default in the kernel anyway, and the non-hierarchical mode
5873 never made much sense in the intrinsically hierarchical
5876 * A new field _SYSTEMD_SLICE= is logged along with all journal
5877 messages containing the slice a message was generated
5878 from. This is useful to allow easy per-customer filtering of
5879 logs among other things.
5881 * systemd-journald will no longer adjust the group of journal
5882 files it creates to the "systemd-journal" group. Instead we
5883 rely on the journal directory to be owned by the
5884 "systemd-journal" group, and its setgid bit set, so that the
5885 kernel file system layer will automatically enforce that
5886 journal files inherit this group assignment. The reason for
5887 this change is that we cannot allow NSS look-ups from
5888 journald which would be necessary to resolve
5889 "systemd-journal" to a numeric GID, because this might
5890 create deadlocks if NSS involves synchronous queries to
5891 other daemons (such as nscd, or sssd) which in turn are
5892 logging clients of journald and might block on it, which
5893 would then dead lock. A tmpfiles.d(5) snippet included in
5894 systemd will make sure the setgid bit and group are
5895 properly set on the journal directory if it exists on every
5896 boot. However, we recommend adjusting it manually after
5897 upgrades too (or from RPM scriptlets), so that the change is
5898 not delayed until next reboot.
5900 * Backlight and random seed files in /var/lib/ have moved into
5901 the /var/lib/systemd/ directory, in order to centralize all
5902 systemd generated files in one directory.
5904 * Boot time performance measurements (as displayed by
5905 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5906 performance information if that's available to determine how
5907 much time BIOS and boot loader initialization required. With
5908 a sufficiently new BIOS you hence no longer need to boot
5909 with Gummiboot to get access to such information.
5911 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5912 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5913 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5914 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5915 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5916 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5917 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5919 — Berlin, 2013-10-02
5923 * The Restart= option for services now understands a new
5924 on-watchdog setting, which will restart the service
5925 automatically if the service stops sending out watchdog keep
5926 alive messages (as configured with WatchdogSec=).
5928 * The getty generator (which is responsible for bringing up a
5929 getty on configured serial consoles) will no longer only
5930 start a getty on the primary kernel console but on all
5931 others, too. This makes the order in which console= is
5932 specified on the kernel command line less important.
5934 * libsystemd-logind gained a new sd_session_get_vt() call to
5935 retrieve the VT number of a session.
5937 * If the option "tries=0" is set for an entry of /etc/crypttab
5938 its passphrase is queried indefinitely instead of any
5939 maximum number of tries.
5941 * If a service with a configure PID file terminates its PID
5942 file will now be removed automatically if it still exists
5943 afterwards. This should put an end to stale PID files.
5945 * systemd-run will now also take relative binary path names
5946 for execution and no longer insists on absolute paths.
5948 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5949 paths that are optionally prefixed with "-" to indicate that
5950 it should not be considered a failure if they do not exist.
5952 * journalctl -o (and similar commands) now understands a new
5953 output mode "short-precise", it is similar to "short" but
5954 shows timestamps with usec accuracy.
5956 * The option "discard" (as known from Debian) is now
5957 synonymous to "allow-discards" in /etc/crypttab. In fact,
5958 "discard" is preferred now (since it is easier to remember
5961 * Some licensing clean-ups were made, so that more code is now
5962 LGPL-2.1 licensed than before.
5964 * A minimal tool to save/restore the display backlight
5965 brightness across reboots has been added. It will store the
5966 backlight setting as late as possible at shutdown, and
5967 restore it as early as possible during reboot.
5969 * A logic to automatically discover and enable home and swap
5970 partitions on GPT disks has been added. With this in place
5971 /etc/fstab becomes optional for many setups as systemd can
5972 discover certain partitions located on the root disk
5973 automatically. Home partitions are recognized under their
5974 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5975 partitions are recognized under their GPT type ID
5976 0657fd6da4ab43c484e50933c84b4f4f.
5978 * systemd will no longer pass any environment from the kernel
5979 or initrd to system services. If you want to set an
5980 environment for all services, do so via the kernel command
5981 line systemd.setenv= assignment.
5983 * The systemd-sysctl tool no longer natively reads the file
5984 /etc/sysctl.conf. If desired, the file should be symlinked
5985 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5986 legacy support by a symlink rather than built-in code, it
5987 also makes the otherwise hidden order of application of the
5988 different files visible. (Note that this partly reverts to a
5989 pre-198 application order of sysctl knobs!)
5991 * The "systemctl set-log-level" and "systemctl dump" commands
5992 have been moved to systemd-analyze.
5994 * systemd-run learned the new --remain-after-exit switch,
5995 which causes the scope unit not to be cleaned up
5996 automatically after the process terminated.
5998 * tmpfiles learned a new --exclude-prefix= switch to exclude
5999 certain paths from operation.
6001 * journald will now automatically flush all messages to disk
6002 as soon as a message at the log level CRIT, ALERT or EMERG
6005 Contributions from: Andrew Cook, Brandon Philips, Christian
6006 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6007 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6008 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6009 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6010 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6011 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6012 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6013 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6014 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6015 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6016 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6017 William Giokas, Zbigniew Jędrzejewski-Szmek
6019 — Berlin, 2013-09-13
6023 * The documentation has been updated to cover the various new
6024 concepts introduced with 205.
6026 * Unit files now understand the new %v specifier which
6027 resolves to the kernel version string as returned by "uname
6030 * systemctl now supports filtering the unit list output by
6031 load state, active state and sub state, using the new
6034 * "systemctl status" will now show the results of the
6035 condition checks (like ConditionPathExists= and similar) of
6036 the last start attempts of the unit. They are also logged to
6039 * "journalctl -b" may now be used to look for boot output of a
6040 specific boot. Try "journalctl -b -1" for the previous boot,
6041 but the syntax is substantially more powerful.
6043 * "journalctl --show-cursor" has been added which prints the
6044 cursor string the last shown log line. This may then be used
6045 with the new "journalctl --after-cursor=" switch to continue
6046 browsing logs from that point on.
6048 * "journalctl --force" may now be used to force regeneration
6051 * Creation of "dead" device nodes has been moved from udev
6052 into kmod and tmpfiles. Previously, udev would read the kmod
6053 databases to pre-generate dead device nodes based on meta
6054 information contained in kernel modules, so that these would
6055 be auto-loaded on access rather then at boot. As this
6056 does not really have much to do with the exposing actual
6057 kernel devices to userspace this has always been slightly
6058 alien in the udev codebase. Following the new scheme kmod
6059 will now generate a runtime snippet for tmpfiles from the
6060 module meta information and it now is tmpfiles' job to the
6061 create the nodes. This also allows overriding access and
6062 other parameters for the nodes using the usual tmpfiles
6063 facilities. As side effect this allows us to remove the
6064 CAP_SYS_MKNOD capability bit from udevd entirely.
6066 * logind's device ACLs may now be applied to these "dead"
6067 devices nodes too, thus finally allowing managed access to
6068 devices such as /dev/snd/sequencer without loading the
6069 backing module right-away.
6071 * A new RPM macro has been added that may be used to apply
6072 tmpfiles configuration during package installation.
6074 * systemd-detect-virt and ConditionVirtualization= now can
6075 detect User-Mode-Linux machines (UML).
6077 * journald will now implicitly log the effective capabilities
6078 set of processes in the message metadata.
6080 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6082 * The initrd interface has been simplified (more specifically,
6083 support for passing performance data via environment
6084 variables and fsck results via files in /run has been
6085 removed). These features were non-essential, and are
6086 nowadays available in a much nicer way by having systemd in
6087 the initrd serialize its state and have the hosts systemd
6088 deserialize it again.
6090 * The udev "keymap" data files and tools to apply keyboard
6091 specific mappings of scan to key codes, and force-release
6092 scan code lists have been entirely replaced by a udev
6093 "keyboard" builtin and a hwdb data file.
6095 * systemd will now honour the kernel's "quiet" command line
6096 argument also during late shutdown, resulting in a
6097 completely silent shutdown when used.
6099 * There's now an option to control the SO_REUSEPORT socket
6100 option in .socket units.
6102 * Instance units will now automatically get a per-template
6103 subslice of system.slice unless something else is explicitly
6104 configured. For example, instances of sshd@.service will now
6105 implicitly be placed in system-sshd.slice rather than
6106 system.slice as before.
6108 * Test coverage support may now be enabled at build time.
6110 Contributions from: Dave Reisner, Frederic Crozat, Harald
6111 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6112 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6113 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6114 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6115 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6116 Giokas, Zbigniew Jędrzejewski-Szmek
6118 — Berlin, 2013-07-23
6122 * Two new unit types have been introduced:
6124 Scope units are very similar to service units, however, are
6125 created out of pre-existing processes — instead of PID 1
6126 forking off the processes. By using scope units it is
6127 possible for system services and applications to group their
6128 own child processes (worker processes) in a powerful way
6129 which then maybe used to organize them, or kill them
6130 together, or apply resource limits on them.
6132 Slice units may be used to partition system resources in an
6133 hierarchical fashion and then assign other units to them. By
6134 default there are now three slices: system.slice (for all
6135 system services), user.slice (for all user sessions),
6136 machine.slice (for VMs and containers).
6138 Slices and scopes have been introduced primarily in
6139 context of the work to move cgroup handling to a
6140 single-writer scheme, where only PID 1
6141 creates/removes/manages cgroups.
6143 * There's a new concept of "transient" units. In contrast to
6144 normal units these units are created via an API at runtime,
6145 not from configuration from disk. More specifically this
6146 means it is now possible to run arbitrary programs as
6147 independent services, with all execution parameters passed
6148 in via bus APIs rather than read from disk. Transient units
6149 make systemd substantially more dynamic then it ever was,
6150 and useful as a general batch manager.
6152 * logind has been updated to make use of scope and slice units
6153 for managing user sessions. As a user logs in he will get
6154 his own private slice unit, to which all sessions are added
6155 as scope units. We also added support for automatically
6156 adding an instance of user@.service for the user into the
6157 slice. Effectively logind will no longer create cgroup
6158 hierarchies on its own now, it will defer entirely to PID 1
6159 for this by means of scope, service and slice units. Since
6160 user sessions this way become entities managed by PID 1
6161 the output of "systemctl" is now a lot more comprehensive.
6163 * A new mini-daemon "systemd-machined" has been added which
6164 may be used by virtualization managers to register local
6165 VMs/containers. nspawn has been updated accordingly, and
6166 libvirt will be updated shortly. machined will collect a bit
6167 of meta information about the VMs/containers, and assign
6168 them their own scope unit (see above). The collected
6169 meta-data is then made available via the "machinectl" tool,
6170 and exposed in "ps" and similar tools. machined/machinectl
6171 is compile-time optional.
6173 * As discussed earlier, the low-level cgroup configuration
6174 options ControlGroup=, ControlGroupModify=,
6175 ControlGroupPersistent=, ControlGroupAttribute= have been
6176 removed. Please use high-level attribute settings instead as
6177 well as slice units.
6179 * A new bus call SetUnitProperties() has been added to alter
6180 various runtime parameters of a unit. This is primarily
6181 useful to alter cgroup parameters dynamically in a nice way,
6182 but will be extended later on to make more properties
6183 modifiable at runtime. systemctl gained a new set-properties
6184 command that wraps this call.
6186 * A new tool "systemd-run" has been added which can be used to
6187 run arbitrary command lines as transient services or scopes,
6188 while configuring a number of settings via the command
6189 line. This tool is currently very basic, however already
6190 very useful. We plan to extend this tool to even allow
6191 queuing of execution jobs with time triggers from the
6192 command line, similar in fashion to "at".
6194 * nspawn will now inform the user explicitly that kernels with
6195 audit enabled break containers, and suggest the user to turn
6198 * Support for detecting the IMA and AppArmor security
6199 frameworks with ConditionSecurity= has been added.
6201 * journalctl gained a new "-k" switch for showing only kernel
6202 messages, mimicking dmesg output; in addition to "--user"
6203 and "--system" switches for showing only user's own logs
6206 * systemd-delta can now show information about drop-in
6207 snippets extending unit files.
6209 * libsystemd-bus has been substantially updated but is still
6210 not available as public API.
6212 * systemd will now look for the "debug" argument on the kernel
6213 command line and enable debug logging, similar to what
6214 "systemd.log_level=debug" already did before.
6216 * "systemctl set-default", "systemctl get-default" has been
6217 added to configure the default.target symlink, which
6218 controls what to boot into by default.
6220 * "systemctl set-log-level" has been added as a convenient
6221 way to raise and lower systemd logging threshold.
6223 * "systemd-analyze plot" will now show the time the various
6224 generators needed for execution, as well as information
6225 about the unit file loading.
6227 * libsystemd-journal gained a new sd_journal_open_files() call
6228 for opening specific journal files. journactl also gained a
6229 new switch to expose this new functionality. Previously we
6230 only supported opening all files from a directory, or all
6231 files from the system, as opening individual files only is
6232 racy due to journal file rotation.
6234 * systemd gained the new DefaultEnvironment= setting in
6235 /etc/systemd/system.conf to set environment variables for
6238 * If a privileged process logs a journal message with the
6239 OBJECT_PID= field set, then journald will automatically
6240 augment this with additional OBJECT_UID=, OBJECT_GID=,
6241 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6242 system services want to log events about specific client
6243 processes. journactl/systemctl has been updated to make use
6244 of this information if all log messages regarding a specific
6247 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6248 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6249 Reisner, David Coppa, David King, David Strauss, Eelco
6250 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6251 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6252 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6253 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6254 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6255 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6256 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6257 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6258 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6259 Łukasz Stelmach, 장동준
6263 * The Python bindings gained some minimal support for the APIs
6264 exposed by libsystemd-logind.
6266 * ConditionSecurity= gained support for detecting SMACK. Since
6267 this condition already supports SELinux and AppArmor we only
6268 miss IMA for this. Patches welcome!
6270 Contributions from: Karol Lewandowski, Lennart Poettering,
6271 Zbigniew Jędrzejewski-Szmek
6275 * systemd-nspawn will now create /etc/resolv.conf if
6276 necessary, before bind-mounting the host's file onto it.
6278 * systemd-nspawn will now store meta information about a
6279 container on the container's cgroup as extended attribute
6280 fields, including the root directory.
6282 * The cgroup hierarchy has been reworked in many ways. All
6283 objects any of the components systemd creates in the cgroup
6284 tree are now suffixed. More specifically, user sessions are
6285 now placed in cgroups suffixed with ".session", users in
6286 cgroups suffixed with ".user", and nspawn containers in
6287 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6288 names are now escaped in a simple scheme to avoid collision
6289 of userspace object names with kernel filenames. This work
6290 is preparation for making these objects relocatable in the
6291 cgroup tree, in order to allow easy resource partitioning of
6292 these objects without causing naming conflicts.
6294 * systemctl list-dependencies gained the new switches
6295 --plain, --reverse, --after and --before.
6297 * systemd-inhibit now shows the process name of processes that
6298 have taken an inhibitor lock.
6300 * nss-myhostname will now also resolve "localhost"
6301 implicitly. This makes /etc/hosts an optional file and
6302 nicely handles that on IPv6 ::1 maps to both "localhost" and
6305 * libsystemd-logind.so gained a new call
6306 sd_get_machine_names() to enumerate running containers and
6307 VMs (currently only supported by very new libvirt and
6308 nspawn). sd_login_monitor can now be used to watch
6309 VMs/containers coming and going.
6311 * .include is not allowed recursively anymore, and only in
6312 unit files. Usually it is better to use drop-in snippets in
6313 .d/*.conf anyway, as introduced with systemd 198.
6315 * systemd-analyze gained a new "critical-chain" command that
6316 determines the slowest chain of units run during system
6317 boot-up. It is very useful for tracking down where
6318 optimizing boot time is the most beneficial.
6320 * systemd will no longer allow manipulating service paths in
6321 the name=systemd:/system cgroup tree using ControlGroup= in
6322 units. (But is still fine with it in all other dirs.)
6324 * There's a new systemd-nspawn@.service service file that may
6325 be used to easily run nspawn containers as system
6326 services. With the container's root directory in
6327 /var/lib/container/foobar it is now sufficient to run
6328 "systemctl start systemd-nspawn@foobar.service" to boot it.
6330 * systemd-cgls gained a new parameter "--machine" to list only
6331 the processes within a certain container.
6333 * ConditionSecurity= now can check for "apparmor". We still
6334 are lacking checks for SMACK and IMA for this condition
6335 check though. Patches welcome!
6337 * A new configuration file /etc/systemd/sleep.conf has been
6338 added that may be used to configure which kernel operation
6339 systemd is supposed to execute when "suspend", "hibernate"
6340 or "hybrid-sleep" is requested. This makes the new kernel
6341 "freeze" state accessible to the user.
6343 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6344 the passed argument if applicable.
6346 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6347 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6348 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6349 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6350 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6351 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6352 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6357 * The output of 'systemctl list-jobs' got some polishing. The
6358 '--type=' argument may now be passed more than once. A new
6359 command 'systemctl list-sockets' has been added which shows
6360 a list of kernel sockets systemd is listening on with the
6361 socket units they belong to, plus the units these socket
6364 * The experimental libsystemd-bus library got substantial
6365 updates to work in conjunction with the (also experimental)
6366 kdbus kernel project. It works well enough to exchange
6367 messages with some sophistication. Note that kdbus is not
6368 ready yet, and the library is mostly an elaborate test case
6369 for now, and not installable.
6371 * systemd gained a new unit 'systemd-static-nodes.service'
6372 that generates static device nodes earlier during boot, and
6373 can run in conjunction with udev.
6375 * libsystemd-login gained a new call sd_pid_get_user_unit()
6376 to retrieve the user systemd unit a process is running
6377 in. This is useful for systems where systemd is used as
6380 * systemd-nspawn now places all containers in the new /machine
6381 top-level cgroup directory in the name=systemd
6382 hierarchy. libvirt will soon do the same, so that we get a
6383 uniform separation of /system, /user and /machine for system
6384 services, user processes and containers/virtual
6385 machines. This new cgroup hierarchy is also useful to stick
6386 stable names to specific container instances, which can be
6387 recognized later this way (this name may be controlled
6388 via systemd-nspawn's new -M switch). libsystemd-login also
6389 gained a new call sd_pid_get_machine_name() to retrieve the
6390 name of the container/VM a specific process belongs to.
6392 * bootchart can now store its data in the journal.
6394 * libsystemd-journal gained a new call
6395 sd_journal_add_conjunction() for AND expressions to the
6396 matching logic. This can be used to express more complex
6397 logical expressions.
6399 * journactl can now take multiple --unit= and --user-unit=
6402 * The cryptsetup logic now understands the "luks.key=" kernel
6403 command line switch for specifying a file to read the
6404 decryption key from. Also, if a configured key file is not
6405 found the tool will now automatically fall back to prompting
6408 * Python systemd.journal module was updated to wrap recently
6409 added functions from libsystemd-journal. The interface was
6410 changed to bring the low level interface in s.j._Reader
6411 closer to the C API, and the high level interface in
6412 s.j.Reader was updated to wrap and convert all data about
6415 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6416 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6417 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6418 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6419 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6420 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6424 * journalctl --update-catalog now understands a new --root=
6425 option to operate on catalogs found in a different root
6428 * During shutdown after systemd has terminated all running
6429 services a final killing loop kills all remaining left-over
6430 processes. We will now print the name of these processes
6431 when we send SIGKILL to them, since this usually indicates a
6434 * If /etc/crypttab refers to password files stored on
6435 configured mount points automatic dependencies will now be
6436 generated to ensure the specific mount is established first
6437 before the key file is attempted to be read.
6439 * 'systemctl status' will now show information about the
6440 network sockets a socket unit is listening on.
6442 * 'systemctl status' will also shown information about any
6443 drop-in configuration file for units. (Drop-In configuration
6444 files in this context are files such as
6445 /etc/systemd/systemd/foobar.service.d/*.conf)
6447 * systemd-cgtop now optionally shows summed up CPU times of
6448 cgroups. Press '%' while running cgtop to switch between
6449 percentage and absolute mode. This is useful to determine
6450 which cgroups use up the most CPU time over the entire
6451 runtime of the system. systemd-cgtop has also been updated
6452 to be 'pipeable' for processing with further shell tools.
6454 * 'hostnamectl set-hostname' will now allow setting of FQDN
6457 * The formatting and parsing of time span values has been
6458 changed. The parser now understands fractional expressions
6459 such as "5.5h". The formatter will now output fractional
6460 expressions for all time spans under 1min, i.e. "5.123456s"
6461 rather than "5s 123ms 456us". For time spans under 1s
6462 millisecond values are shown, for those under 1ms
6463 microsecond values are shown. This should greatly improve
6464 all time-related output of systemd.
6466 * libsystemd-login and libsystemd-journal gained new
6467 functions for querying the poll() events mask and poll()
6468 timeout value for integration into arbitrary event
6471 * localectl gained the ability to list available X11 keymaps
6472 (models, layouts, variants, options).
6474 * 'systemd-analyze dot' gained the ability to filter for
6475 specific units via shell-style globs, to create smaller,
6476 more useful graphs. I.e. it is now possible to create simple
6477 graphs of all the dependencies between only target units, or
6478 of all units that Avahi has dependencies with.
6480 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6481 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6482 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6483 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6484 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6485 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6486 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6490 * The boot-time readahead implementation for rotating media
6491 will now read the read-ahead data in multiple passes which
6492 consist of all read requests made in equidistant time
6493 intervals. This means instead of strictly reading read-ahead
6494 data in its physical order on disk we now try to find a
6495 middle ground between physical and access time order.
6497 * /etc/os-release files gained a new BUILD_ID= field for usage
6498 on operating systems that provide continuous builds of OS
6501 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6502 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6503 William Douglas, Zbigniew Jędrzejewski-Szmek
6507 * systemd-python gained an API exposing libsystemd-daemon.
6509 * The SMACK setup logic gained support for uploading CIPSO
6512 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6513 ReadOnlyDirectories= and InaccessibleDirectories= has
6514 changed. The private /tmp and /var/tmp directories are now
6515 shared by all processes of a service (which means
6516 ExecStartPre= may now leave data in /tmp that ExecStart= of
6517 the same service can still access). When a service is
6518 stopped its temporary directories are immediately deleted
6519 (normal clean-up with tmpfiles is still done in addition to
6522 * By default, systemd will now set a couple of sysctl
6523 variables in the kernel: the safe sysrq options are turned
6524 on, IP route verification is turned on, and source routing
6525 disabled. The recently added hardlink and softlink
6526 protection of the kernel is turned on. These settings should
6527 be reasonably safe, and good defaults for all new systems.
6529 * The predictable network naming logic may now be turned off
6530 with a new kernel command line switch: net.ifnames=0.
6532 * A new libsystemd-bus module has been added that implements a
6533 pretty complete D-Bus client library. For details see:
6535 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6537 * journald will now explicitly flush the journal files to disk
6538 at the latest 5min after each write. The file will then also
6539 be marked offline until the next write. This should increase
6540 reliability in case of a crash. The synchronization delay
6541 can be configured via SyncIntervalSec= in journald.conf.
6543 * There's a new remote-fs-setup.target unit that can be used
6544 to pull in specific services when at least one remote file
6545 system is to be mounted.
6547 * There are new targets timers.target and paths.target as
6548 canonical targets to pull user timer and path units in
6549 from. This complements sockets.target with a similar
6550 purpose for socket units.
6552 * libudev gained a new call udev_device_set_attribute_value()
6553 to set sysfs attributes of a device.
6555 * The udev daemon now sets the default number of worker
6556 processes executed in parallel based on the number of available
6557 CPUs instead of the amount of available RAM. This is supposed
6558 to provide a more reliable default and limit a too aggressive
6559 parallelism for setups with 1000s of devices connected.
6561 Contributions from: Auke Kok, Colin Walters, Cristian
6562 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6563 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6564 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6565 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6566 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6567 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6568 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6569 Zbigniew Jędrzejewski-Szmek
6573 * Configuration of unit files may now be extended via drop-in
6574 files without having to edit/override the unit files
6575 themselves. More specifically, if the administrator wants to
6576 change one value for a service file foobar.service he can
6577 now do so by dropping in a configuration snippet into
6578 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6579 will load all these snippets and apply them on top of the
6580 main unit configuration file, possibly extending or
6581 overriding its settings. Using these drop-in snippets is
6582 generally nicer than the two earlier options for changing
6583 unit files locally: copying the files from
6584 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6585 them there; or creating a new file in /etc/systemd/system/
6586 that incorporates the original one via ".include". Drop-in
6587 snippets into these .d/ directories can be placed in any
6588 directory systemd looks for units in, and the usual
6589 overriding semantics between /usr/lib, /etc and /run apply
6592 * Most unit file settings which take lists of items can now be
6593 reset by assigning the empty string to them. For example,
6594 normally, settings such as Environment=FOO=BAR append a new
6595 environment variable assignment to the environment block,
6596 each time they are used. By assigning Environment= the empty
6597 string the environment block can be reset to empty. This is
6598 particularly useful with the .d/*.conf drop-in snippets
6599 mentioned above, since this adds the ability to reset list
6600 settings from vendor unit files via these drop-ins.
6602 * systemctl gained a new "list-dependencies" command for
6603 listing the dependencies of a unit recursively.
6605 * Inhibitors are now honored and listed by "systemctl
6606 suspend", "systemctl poweroff" (and similar) too, not only
6607 GNOME. These commands will also list active sessions by
6610 * Resource limits (as exposed by the various control group
6611 controllers) can now be controlled dynamically at runtime
6612 for all units. More specifically, you can now use a command
6613 like "systemctl set-cgroup-attr foobar.service cpu.shares
6614 2000" to alter the CPU shares a specific service gets. These
6615 settings are stored persistently on disk, and thus allow the
6616 administrator to easily adjust the resource usage of
6617 services with a few simple commands. This dynamic resource
6618 management logic is also available to other programs via the
6619 bus. Almost any kernel cgroup attribute and controller is
6622 * systemd-vconsole-setup will now copy all font settings to
6623 all allocated VTs, where it previously applied them only to
6626 * libsystemd-login gained the new sd_session_get_tty() API
6629 * This release drops support for a few legacy or
6630 distribution-specific LSB facility names when parsing init
6631 scripts: $x-display-manager, $mail-transfer-agent,
6632 $mail-transport-agent, $mail-transfer-agent, $smtp,
6633 $null. Also, the mail-transfer-agent.target unit backing
6634 this has been removed. Distributions which want to retain
6635 compatibility with this should carry the burden for
6636 supporting this themselves and patch support for these back
6637 in, if they really need to. Also, the facilities $syslog and
6638 $local_fs are now ignored, since systemd does not support
6639 early-boot LSB init scripts anymore, and these facilities
6640 are implied anyway for normal services. syslog.target has
6643 * There are new bus calls on PID1's Manager object for
6644 cancelling jobs, and removing snapshot units. Previously,
6645 both calls were only available on the Job and Snapshot
6648 * systemd-journal-gatewayd gained SSL support.
6650 * The various "environment" files, such as /etc/locale.conf
6651 now support continuation lines with a backslash ("\") as
6652 last character in the line, similarly in style (but different)
6653 to how this is supported in shells.
6655 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6656 now implicitly appended to every log entry logged. systemctl
6657 has been updated to filter by this field when operating on a
6658 user systemd instance.
6660 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6661 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6662 the container. This makes it easier to boot unmodified
6663 Fedora systems in a container, which however still requires
6664 audit=0 to be passed on the kernel command line. Auditing in
6665 kernel and userspace is unfortunately still too broken in
6666 context of containers, hence we recommend compiling it out
6667 of the kernel or using audit=0. Hopefully this will be fixed
6668 one day for good in the kernel.
6670 * nspawn gained the new --bind= and --bind-ro= parameters to
6671 bind mount specific directories from the host into the
6674 * nspawn will now mount its own devpts file system instance
6675 into the container, in order not to leak pty devices from
6676 the host into the container.
6678 * systemd will now read the firmware boot time performance
6679 information from the EFI variables, if the used boot loader
6680 supports this, and takes it into account for boot performance
6681 analysis via "systemd-analyze". This is currently supported
6682 only in conjunction with Gummiboot, but could be supported
6683 by other boot loaders too. For details see:
6685 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6687 * A new generator has been added that automatically mounts the
6688 EFI System Partition (ESP) to /boot, if that directory
6689 exists, is empty, and no other file system has been
6690 configured to be mounted there.
6692 * logind will now send out PrepareForSleep(false) out
6693 unconditionally, after coming back from suspend. This may be
6694 used by applications as asynchronous notification for
6695 system resume events.
6697 * "systemctl unlock-sessions" has been added, that allows
6698 unlocking the screens of all user sessions at once, similar
6699 to how "systemctl lock-sessions" already locked all users
6700 sessions. This is backed by a new D-Bus call UnlockSessions().
6702 * "loginctl seat-status" will now show the master device of a
6703 seat. (i.e. the device of a seat that needs to be around for
6704 the seat to be considered available, usually the graphics
6707 * tmpfiles gained a new "X" line type, that allows
6708 configuration of files and directories (with wildcards) that
6709 shall be excluded from automatic cleanup ("aging").
6711 * udev default rules set the device node permissions now only
6712 at "add" events, and do not change them any longer with a
6713 later "change" event.
6715 * The log messages for lid events and power/sleep keypresses
6716 now carry a message ID.
6718 * We now have a substantially larger unit test suite, but this
6719 continues to be work in progress.
6721 * udevadm hwdb gained a new --root= parameter to change the
6722 root directory to operate relative to.
6724 * logind will now issue a background sync() request to the kernel
6725 early at shutdown, so that dirty buffers are flushed to disk early
6726 instead of at the last moment, in order to optimize shutdown
6729 * A new bootctl tool has been added that is an interface for
6730 certain boot loader operations. This is currently a preview
6731 and is likely to be extended into a small mechanism daemon
6732 like timedated, localed, hostnamed, and can be used by
6733 graphical UIs to enumerate available boot options, and
6734 request boot into firmware operations.
6736 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6737 the rest of the package. It also has been updated to work
6738 correctly in initrds.
6740 * Policykit previously has been runtime optional, and is now
6741 also compile time optional via a configure switch.
6743 * systemd-analyze has been reimplemented in C. Also "systemctl
6744 dot" has moved into systemd-analyze.
6746 * "systemctl status" with no further parameters will now print
6747 the status of all active or failed units.
6749 * Operations such as "systemctl start" can now be executed
6750 with a new mode "--irreversible" which may be used to queue
6751 operations that cannot accidentally be reversed by a later
6752 job queuing. This is by default used to make shutdown
6753 requests more robust.
6755 * The Python API of systemd now gained a new module for
6756 reading journal files.
6758 * A new tool kernel-install has been added that can install
6759 kernel images according to the Boot Loader Specification:
6761 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6763 * Boot time console output has been improved to provide
6764 animated boot time output for hanging jobs.
6766 * A new tool systemd-activate has been added which can be used
6767 to test socket activation with, directly from the command
6768 line. This should make it much easier to test and debug
6769 socket activation in daemons.
6771 * journalctl gained a new "--reverse" (or -r) option to show
6772 journal output in reverse order (i.e. newest line first).
6774 * journalctl gained a new "--pager-end" (or -e) option to jump
6775 to immediately jump to the end of the journal in the
6776 pager. This is only supported in conjunction with "less".
6778 * journalctl gained a new "--user-unit=" option, that works
6779 similarly to "--unit=" but filters for user units rather than
6782 * A number of unit files to ease adoption of systemd in
6783 initrds has been added. This moves some minimal logic from
6784 the various initrd implementations into systemd proper.
6786 * The journal files are now owned by a new group
6787 "systemd-journal", which exists specifically to allow access
6788 to the journal, and nothing else. Previously, we used the
6789 "adm" group for that, which however possibly covers more
6790 than just journal/log file access. This new group is now
6791 already used by systemd-journal-gatewayd to ensure this
6792 daemon gets access to the journal files and as little else
6793 as possible. Note that "make install" will also set FS ACLs
6794 up for /var/log/journal to give "adm" and "wheel" read
6795 access to it, in addition to "systemd-journal" which owns
6796 the journal files. We recommend that packaging scripts also
6797 add read access to "adm" + "wheel" to /var/log/journal, and
6798 all existing/future journal files. To normal users and
6799 administrators little changes, however packagers need to
6800 ensure to create the "systemd-journal" system group at
6801 package installation time.
6803 * The systemd-journal-gatewayd now runs as unprivileged user
6804 systemd-journal-gateway:systemd-journal-gateway. Packaging
6805 scripts need to create these system user/group at
6808 * timedated now exposes a new boolean property CanNTP that
6809 indicates whether a local NTP service is available or not.
6811 * systemd-detect-virt will now also detect xen PVs
6813 * The pstore file system is now mounted by default, if it is
6816 * In addition to the SELinux and IMA policies we will now also
6817 load SMACK policies at early boot.
6819 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6820 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6821 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6822 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6823 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6824 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6825 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6826 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6827 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6828 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6829 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6830 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6831 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6832 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6836 * Timer units now support calendar time events in addition to
6837 monotonic time events. That means you can now trigger a unit
6838 based on a calendar time specification such as "Thu,Fri
6839 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6840 or fifth day of any month of the year 2013, given that it is
6841 a thursday or friday. This brings timer event support
6842 considerably closer to cron's capabilities. For details on
6843 the supported calendar time specification language see
6846 * udev now supports a number of different naming policies for
6847 network interfaces for predictable names, and a combination
6848 of these policies is now the default. Please see this wiki
6849 document for details:
6851 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6853 * Auke Kok's bootchart implementation has been added to the
6854 systemd tree. It is an optional component that can graph the
6855 boot in quite some detail. It is one of the best bootchart
6856 implementations around and minimal in its code and
6859 * nss-myhostname has been integrated into the systemd source
6860 tree. nss-myhostname guarantees that the local hostname
6861 always stays resolvable via NSS. It has been a weak
6862 requirement of systemd-hostnamed since a long time, and
6863 since its code is actually trivial we decided to just
6864 include it in systemd's source tree. It can be turned off
6865 with a configure switch.
6867 * The read-ahead logic is now capable of properly detecting
6868 whether a btrfs file system is on SSD or rotating media, in
6869 order to optimize the read-ahead scheme. Previously, it was
6870 only capable of detecting this on traditional file systems
6873 * In udev, additional device properties are now read from the
6874 IAB in addition to the OUI database. Also, Bluetooth company
6875 identities are attached to the devices as well.
6877 * In service files %U may be used as specifier that is
6878 replaced by the configured user name of the service.
6880 * nspawn may now be invoked without a controlling TTY. This
6881 makes it suitable for invocation as its own service. This
6882 may be used to set up a simple containerized server system
6883 using only core OS tools.
6885 * systemd and nspawn can now accept socket file descriptors
6886 when they are started for socket activation. This enables
6887 implementation of socket activated nspawn
6888 containers. i.e. think about autospawning an entire OS image
6889 when the first SSH or HTTP connection is received. We expect
6890 that similar functionality will also be added to libvirt-lxc
6893 * journalctl will now suppress ANSI color codes when
6894 presenting log data.
6896 * systemctl will no longer show control group information for
6897 a unit if the control group is empty anyway.
6899 * logind can now automatically suspend/hibernate/shutdown the
6902 * /etc/machine-info and hostnamed now also expose the chassis
6903 type of the system. This can be used to determine whether
6904 the local system is a laptop, desktop, handset or
6905 tablet. This information may either be configured by the
6906 user/vendor or is automatically determined from ACPI and DMI
6907 information if possible.
6909 * A number of PolicyKit actions are now bound together with
6910 "imply" rules. This should simplify creating UIs because
6911 many actions will now authenticate similar ones as well.
6913 * Unit files learnt a new condition ConditionACPower= which
6914 may be used to conditionalize a unit depending on whether an
6915 AC power source is connected or not, of whether the system
6916 is running on battery power.
6918 * systemctl gained a new "is-failed" verb that may be used in
6919 shell scripts and suchlike to check whether a specific unit
6920 is in the "failed" state.
6922 * The EnvironmentFile= setting in unit files now supports file
6923 globbing, and can hence be used to easily read a number of
6924 environment files at once.
6926 * systemd will no longer detect and recognize specific
6927 distributions. All distribution-specific #ifdeffery has been
6928 removed, systemd is now fully generic and
6929 distribution-agnostic. Effectively, not too much is lost as
6930 a lot of the code is still accessible via explicit configure
6931 switches. However, support for some distribution specific
6932 legacy configuration file formats has been dropped. We
6933 recommend distributions to simply adopt the configuration
6934 files everybody else uses now and convert the old
6935 configuration from packaging scripts. Most distributions
6936 already did that. If that's not possible or desirable,
6937 distributions are welcome to forward port the specific
6938 pieces of code locally from the git history.
6940 * When logging a message about a unit systemd will now always
6941 log the unit name in the message meta data.
6943 * localectl will now also discover system locale data that is
6944 not stored in locale archives, but directly unpacked.
6946 * logind will no longer unconditionally use framebuffer
6947 devices as seat masters, i.e. as devices that are required
6948 to be existing before a seat is considered preset. Instead,
6949 it will now look for all devices that are tagged as
6950 "seat-master" in udev. By default, framebuffer devices will
6951 be marked as such, but depending on local systems, other
6952 devices might be marked as well. This may be used to
6953 integrate graphics cards using closed source drivers (such
6954 as NVidia ones) more nicely into logind. Note however, that
6955 we recommend using the open source NVidia drivers instead,
6956 and no udev rules for the closed-source drivers will be
6957 shipped from us upstream.
6959 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6960 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6961 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6962 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6963 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6964 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6965 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6966 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6967 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6968 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6969 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6970 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6975 * udev gained support for loading additional device properties
6976 from an indexed database that is keyed by vendor/product IDs
6977 and similar device identifiers. For the beginning this
6978 "hwdb" is populated with data from the well-known PCI and
6979 USB database, but also includes PNP, ACPI and OID data. In
6980 the longer run this indexed database shall grow into
6981 becoming the one central database for non-essential
6982 userspace device metadata. Previously, data from the PCI/USB
6983 database was only attached to select devices, since the
6984 lookup was a relatively expensive operation due to O(n) time
6985 complexity (with n being the number of entries in the
6986 database). Since this is now O(1), we decided to add in this
6987 data for all devices where this is available, by
6988 default. Note that the indexed database needs to be rebuilt
6989 when new data files are installed. To achieve this you need
6990 to update your packaging scripts to invoke "udevadm hwdb
6991 --update" after installation of hwdb data files. For
6992 RPM-based distributions we introduced the new
6993 %udev_hwdb_update macro for this purpose.
6995 * The Journal gained support for the "Message Catalog", an
6996 indexed database to link up additional information with
6997 journal entries. For further details please check:
6999 https://www.freedesktop.org/wiki/Software/systemd/catalog
7001 The indexed message catalog database also needs to be
7002 rebuilt after installation of message catalog files. Use
7003 "journalctl --update-catalog" for this. For RPM-based
7004 distributions we introduced the %journal_catalog_update
7005 macro for this purpose.
7007 * The Python Journal bindings gained support for the standard
7008 Python logging framework.
7010 * The Journal API gained new functions for checking whether
7011 the underlying file system of a journal file is capable of
7012 properly reporting file change notifications, or whether
7013 applications that want to reflect journal changes "live"
7014 need to recheck journal files continuously in appropriate
7017 * It is now possible to set the "age" field for tmpfiles
7018 entries to 0, indicating that files matching this entry
7019 shall always be removed when the directories are cleaned up.
7021 * coredumpctl gained a new "gdb" verb which invokes gdb
7022 right-away on the selected coredump.
7024 * There's now support for "hybrid sleep" on kernels that
7025 support this, in addition to "suspend" and "hibernate". Use
7026 "systemctl hybrid-sleep" to make use of this.
7028 * logind's HandleSuspendKey= setting (and related settings)
7029 now gained support for a new "lock" setting to simply
7030 request the screen lock on all local sessions, instead of
7031 actually executing a suspend or hibernation.
7033 * systemd will now mount the EFI variables file system by
7036 * Socket units now gained support for configuration of the
7037 SMACK security label.
7039 * timedatectl will now output the time of the last and next
7040 daylight saving change.
7042 * We dropped support for various legacy and distro-specific
7043 concepts, such as insserv, early-boot SysV services
7044 (i.e. those for non-standard runlevels such as 'b' or 'S')
7045 or ArchLinux /etc/rc.conf support. We recommend the
7046 distributions who still need support this to either continue
7047 to maintain the necessary patches downstream, or find a
7048 different solution. (Talk to us if you have questions!)
7050 * Various systemd components will now bypass PolicyKit checks
7051 for root and otherwise handle properly if PolicyKit is not
7052 found to be around. This should fix most issues for
7053 PolicyKit-less systems. Quite frankly this should have been
7054 this way since day one. It is absolutely our intention to
7055 make systemd work fine on PolicyKit-less systems, and we
7056 consider it a bug if something does not work as it should if
7057 PolicyKit is not around.
7059 * For embedded systems it is now possible to build udev and
7060 systemd without blkid and/or kmod support.
7062 * "systemctl switch-root" is now capable of switching root
7063 more than once. I.e. in addition to transitions from the
7064 initrd to the host OS it is now possible to transition to
7065 further OS images from the host. This is useful to implement
7066 offline updating tools.
7068 * Various other additions have been made to the RPM macros
7069 shipped with systemd. Use %udev_rules_update() after
7070 installing new udev rules files. %_udevhwdbdir,
7071 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7072 %_sysctldir are now available which resolve to the right
7073 directories for packages to place various data files in.
7075 * journalctl gained the new --full switch (in addition to
7076 --all, to disable ellipsation for long messages.
7078 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7079 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7080 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7081 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7082 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7083 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7084 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7085 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7086 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7090 * journalctl gained new --since= and --until= switches to
7091 filter by time. It also now supports nice filtering for
7092 units via --unit=/-u.
7094 * Type=oneshot services may use ExecReload= and do the
7097 * The journal daemon now supports time-based rotation and
7098 vacuuming, in addition to the usual disk-space based
7101 * The journal will now index the available field values for
7102 each field name. This enables clients to show pretty drop
7103 downs of available match values when filtering. The bash
7104 completion of journalctl has been updated
7105 accordingly. journalctl gained a new switch -F to list all
7106 values a certain field takes in the journal database.
7108 * More service events are now written as structured messages
7109 to the journal, and made recognizable via message IDs.
7111 * The timedated, localed and hostnamed mini-services which
7112 previously only provided support for changing time, locale
7113 and hostname settings from graphical DEs such as GNOME now
7114 also have a minimal (but very useful) text-based client
7115 utility each. This is probably the nicest way to changing
7116 these settings from the command line now, especially since
7117 it lists available options and is fully integrated with bash
7120 * There's now a new tool "systemd-coredumpctl" to list and
7121 extract coredumps from the journal.
7123 * We now install a README each in /var/log/ and
7124 /etc/rc.d/init.d explaining where the system logs and init
7125 scripts went. This hopefully should help folks who go to
7126 that dirs and look into the otherwise now empty void and
7127 scratch their heads.
7129 * When user-services are invoked (by systemd --user) the
7130 $MANAGERPID env var is set to the PID of systemd.
7132 * SIGRTMIN+24 when sent to a --user instance will now result
7133 in immediate termination of systemd.
7135 * gatewayd received numerous feature additions such as a
7136 "follow" mode, for live syncing and filtering.
7138 * browse.html now allows filtering and showing detailed
7139 information on specific entries. Keyboard navigation and
7140 mouse screen support has been added.
7142 * gatewayd/journalctl now supports HTML5/JSON
7143 Server-Sent-Events as output.
7145 * The SysV init script compatibility logic will now
7146 heuristically determine whether a script supports the
7147 "reload" verb, and only then make this available as
7150 * "systemctl status --follow" has been removed, use "journalctl
7153 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7154 have been removed since they are hardly useful to be
7157 * And I'd like to take the opportunity to specifically mention
7158 Zbigniew for his great contributions. Zbigniew, you rock!
7160 Contributions from: Andrew Eikum, Christian Hesse, Colin
7161 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7162 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7163 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7164 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7165 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7166 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7170 * If /etc/vconsole.conf is non-existent or empty we will no
7171 longer load any console font or key map at boot by
7172 default. Instead the kernel defaults will be left
7173 intact. This is definitely the right thing to do, as no
7174 configuration should mean no configuration, and hard-coding
7175 font names that are different on all archs is probably a bad
7176 idea. Also, the kernel default key map and font should be
7177 good enough for most cases anyway, and mostly identical to
7178 the userspace fonts/key maps we previously overloaded them
7179 with. If distributions want to continue to default to a
7180 non-kernel font or key map they should ship a default
7181 /etc/vconsole.conf with the appropriate contents.
7183 Contributions from: Colin Walters, Daniel J Walsh, Dave
7184 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7185 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7189 * journalctl gained a new --cursor= switch to show entries
7190 starting from the specified location in the journal.
7192 * We now enforce a size limit on journal entry fields exported
7193 with "-o json" in journalctl. Fields larger than 4K will be
7194 assigned null. This can be turned off with --all.
7196 * An (optional) journal gateway daemon is now available as
7197 "systemd-journal-gatewayd.service". This service provides
7198 access to the journal via HTTP and JSON. This functionality
7199 will be used to implement live log synchronization in both
7200 pull and push modes, but has various other users too, such
7201 as easy log access for debugging of embedded devices. Right
7202 now it is already useful to retrieve the journal via HTTP:
7204 # systemctl start systemd-journal-gatewayd.service
7205 # wget http://localhost:19531/entries
7207 This will download the journal contents in a
7208 /var/log/messages compatible format. The same as JSON:
7210 # curl -H"Accept: application/json" http://localhost:19531/entries
7212 This service is also accessible via a web browser where a
7213 single static HTML5 app is served that uses the JSON logic
7214 to enable the user to do some basic browsing of the
7215 journal. This will be extended later on. Here's an example
7216 screenshot of this app in its current state:
7218 http://0pointer.de/public/journal-gatewayd
7220 Contributions from: Kay Sievers, Lennart Poettering, Robert
7221 Milasan, Tom Gundersen
7225 * The bash completion logic is now available for journalctl
7228 * We do not mount the "cpuset" controller anymore together with
7229 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7230 started if no parameters are assigned to it. "cpuset" hence
7231 broke code that assumed it could create "cpu" groups and
7234 * journalctl -f will now subscribe to terminal size changes,
7235 and line break accordingly.
7237 Contributions from: Dave Reisner, Kay Sievers, Lennart
7238 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7242 * nspawn will now create a symlink /etc/localtime in the
7243 container environment, copying the host's timezone
7244 setting. Previously this has been done via a bind mount, but
7245 since symlinks cannot be bind mounted this has now been
7246 changed to create/update the appropriate symlink.
7248 * journalctl -n's line number argument is now optional, and
7249 will default to 10 if omitted.
7251 * journald will now log the maximum size the journal files may
7252 take up on disk. This is particularly useful if the default
7253 built-in logic of determining this parameter from the file
7254 system size is used. Use "systemctl status
7255 systemd-journald.service" to see this information.
7257 * The multi-seat X wrapper tool has been stripped down. As X
7258 is now capable of enumerating graphics devices via udev in a
7259 seat-aware way the wrapper is not strictly necessary
7260 anymore. A stripped down temporary stop-gap is still shipped
7261 until the upstream display managers have been updated to
7262 fully support the new X logic. Expect this wrapper to be
7263 removed entirely in one of the next releases.
7265 * HandleSleepKey= in logind.conf has been split up into
7266 HandleSuspendKey= and HandleHibernateKey=. The old setting
7267 is not available anymore. X11 and the kernel are
7268 distinguishing between these keys and we should too. This
7269 also means the inhibition lock for these keys has been split
7272 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7273 Poettering, Lukas Nykryn, Václav Pavlín
7277 * Whenever a unit changes state we will now log this to the
7278 journal and show along the unit's own log output in
7281 * ConditionPathIsMountPoint= can now properly detect bind
7282 mount points too. (Previously, a bind mount of one file
7283 system to another place in the same file system could not be
7284 detected as mount, since they shared struct stat's st_dev
7287 * We will now mount the cgroup controllers cpu, cpuacct,
7288 cpuset and the controllers net_cls, net_prio together by
7291 * nspawn containers will now have a virtualized boot
7292 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7293 over with a randomized ID at container initialization). This
7294 has the effect of making "journalctl -b" do the right thing
7297 * The JSON output journal serialization has been updated not
7298 to generate "endless" list objects anymore, but rather one
7299 JSON object per line. This is more in line how most JSON
7300 parsers expect JSON objects. The new output mode
7301 "json-pretty" has been added to provide similar output, but
7302 neatly aligned for readability by humans.
7304 * We dropped all explicit sync() invocations in the shutdown
7305 code. The kernel does this implicitly anyway in the kernel
7306 reboot() syscall. halt(8)'s -n option is now a compatibility
7309 * We now support virtualized reboot() in containers, as
7310 supported by newer kernels. We will fall back to exit() if
7311 CAP_SYS_REBOOT is not available to the container. Also,
7312 nspawn makes use of this now and will actually reboot the
7313 container if the containerized OS asks for that.
7315 * journalctl will only show local log output by default
7316 now. Use --merge (-m) to show remote log output, too.
7318 * libsystemd-journal gained the new sd_journal_get_usage()
7319 call to determine the current disk usage of all journal
7320 files. This is exposed in the new "journalctl --disk-usage"
7323 * journald gained a new configuration setting SplitMode= in
7324 journald.conf which may be used to control how user journals
7325 are split off. See journald.conf(5) for details.
7327 * A new condition type ConditionFileNotEmpty= has been added.
7329 * tmpfiles' "w" lines now support file globbing, to write
7330 multiple files at once.
7332 * We added Python bindings for the journal submission
7333 APIs. More Python APIs for a number of selected APIs will
7334 likely follow. Note that we intend to add native bindings
7335 only for the Python language, as we consider it common
7336 enough to deserve bindings shipped within systemd. There are
7337 various projects outside of systemd that provide bindings
7338 for languages such as PHP or Lua.
7340 * Many conditions will now resolve specifiers such as %i. In
7341 addition, PathChanged= and related directives of .path units
7342 now support specifiers as well.
7344 * There's now a new RPM macro definition for the system preset
7347 * journald will now warn if it ca not forward a message to the
7348 syslog daemon because its socket is full.
7350 * timedated will no longer write or process /etc/timezone,
7351 except on Debian. As we do not support late mounted /usr
7352 anymore /etc/localtime always being a symlink is now safe,
7353 and hence the information in /etc/timezone is not necessary
7356 * logind will now always reserve one VT for a text getty (VT6
7357 by default). Previously if more than 6 X sessions where
7358 started they took up all the VTs with auto-spawned gettys,
7359 so that no text gettys were available anymore.
7361 * udev will now automatically inform the btrfs kernel logic
7362 about btrfs RAID components showing up. This should make
7363 simple hotplug based btrfs RAID assembly work.
7365 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7366 (but not for its children which will stay at the kernel
7367 default). This should allow setups with a lot more listening
7370 * systemd will now always pass the configured timezone to the
7371 kernel at boot. timedated will do the same when the timezone
7374 * logind's inhibition logic has been updated. By default,
7375 logind will now handle the lid switch, the power and sleep
7376 keys all the time, even in graphical sessions. If DEs want
7377 to handle these events on their own they should take the new
7378 handle-power-key, handle-sleep-key and handle-lid-switch
7379 inhibitors during their runtime. A simple way to achieve
7380 that is to invoke the DE wrapped in an invocation of:
7382 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7384 * Access to unit operations is now checked via SELinux taking
7385 the unit file label and client process label into account.
7387 * systemd will now notify the administrator in the journal
7388 when he over-mounts a non-empty directory.
7390 * There are new specifiers that are resolved in unit files,
7391 for the host name (%H), the machine ID (%m) and the boot ID
7394 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7395 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7396 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7397 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7398 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7399 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7400 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7404 * Support for reading structured kernel messages from
7405 /dev/kmsg has now been added and is enabled by default.
7407 * Support for reading kernel messages from /proc/kmsg has now
7408 been removed. If you want kernel messages in the journal
7409 make sure to run a recent kernel (>= 3.5) that supports
7410 reading structured messages from /dev/kmsg (see
7411 above). /proc/kmsg is now exclusive property of classic
7412 syslog daemons again.
7414 * The libudev API gained the new
7415 udev_device_new_from_device_id() call.
7417 * The logic for file system namespace (ReadOnlyDirectory=,
7418 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7419 require pivot_root() anymore. This means fewer temporary
7420 directories are created below /tmp for this feature.
7422 * nspawn containers will now see and receive all submounts
7423 made on the host OS below the root file system of the
7426 * Forward Secure Sealing is now supported for Journal files,
7427 which provide cryptographical sealing of journal files so
7428 that attackers cannot alter log history anymore without this
7429 being detectable. Lennart will soon post a blog story about
7430 this explaining it in more detail.
7432 * There are two new service settings RestartPreventExitStatus=
7433 and SuccessExitStatus= which allow configuration of exit
7434 status (exit code or signal) which will be excepted from the
7435 restart logic, resp. consider successful.
7437 * journalctl gained the new --verify switch that can be used
7438 to check the integrity of the structure of journal files and
7439 (if Forward Secure Sealing is enabled) the contents of
7442 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7443 and similar symlinks pre-created. This makes running shells
7444 as container init process a lot more fun.
7446 * The fstab support can now handle PARTUUID= and PARTLABEL=
7449 * A new ConditionHost= condition has been added to match
7450 against the hostname (with globs) and machine ID. This is
7451 useful for clusters where a single OS image is used to
7452 provision a large number of hosts which shall run slightly
7453 different sets of services.
7455 * Services which hit the restart limit will now be placed in a
7458 Contributions from: Bertram Poettering, Dave Reisner, Huang
7459 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7460 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7464 * When running in --user mode systemd will now become a
7465 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7466 tree a lot more organized.
7468 * A new PartOf= unit dependency type has been introduced that
7469 may be used to group services in a natural way.
7471 * "systemctl enable" may now be used to enable instances of
7474 * journalctl now prints error log levels in red, and
7475 warning/notice log levels in bright white. It also supports
7476 filtering by log level now.
7478 * cgtop gained a new -n switch (similar to top), to configure
7479 the maximum number of iterations to run for. It also gained
7480 -b, to run in batch mode (accepting no input).
7482 * The suffix ".service" may now be omitted on most systemctl
7483 command lines involving service unit names.
7485 * There's a new bus call in logind to lock all sessions, as
7486 well as a loginctl verb for it "lock-sessions".
7488 * libsystemd-logind.so gained a new call sd_journal_perror()
7489 that works similar to libc perror() but logs to the journal
7490 and encodes structured information about the error number.
7492 * /etc/crypttab entries now understand the new keyfile-size=
7495 * shutdown(8) now can send a (configurable) wall message when
7496 a shutdown is cancelled.
7498 * The mount propagation mode for the root file system will now
7499 default to "shared", which is useful to make containers work
7500 nicely out-of-the-box so that they receive new mounts from
7501 the host. This can be undone locally by running "mount
7502 --make-rprivate /" if needed.
7504 * The prefdm.service file has been removed. Distributions
7505 should maintain this unit downstream if they intend to keep
7506 it around. However, we recommend writing normal unit files
7507 for display managers instead.
7509 * Since systemd is a crucial part of the OS we will now
7510 default to a number of compiler switches that improve
7511 security (hardening) such as read-only relocations, stack
7512 protection, and suchlike.
7514 * The TimeoutSec= setting for services is now split into
7515 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7516 of individual time outs for the start and the stop phase of
7519 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7520 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7521 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7522 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7523 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7524 Gundersen, Zbigniew Jędrzejewski-Szmek
7528 * The journal and id128 C APIs are now fully documented as man
7531 * Extra safety checks have been added when transitioning from
7532 the initial RAM disk to the main system to avoid accidental
7535 * /etc/crypttab entries now understand the new keyfile-offset=
7538 * systemctl -t can now be used to filter by unit load state.
7540 * The journal C API gained the new sd_journal_wait() call to
7541 make writing synchronous journal clients easier.
7543 * journalctl gained the new -D switch to show journals from a
7546 * journalctl now displays a special marker between log
7547 messages of two different boots.
7549 * The journal is now explicitly flushed to /var via a service
7550 systemd-journal-flush.service, rather than implicitly simply
7551 by seeing /var/log/journal to be writable.
7553 * journalctl (and the journal C APIs) can now match for much
7554 more complex expressions, with alternatives and
7557 * When transitioning from the initial RAM disk to the main
7558 system we will now kill all processes in a killing spree to
7559 ensure no processes stay around by accident.
7561 * Three new specifiers may be used in unit files: %u, %h, %s
7562 resolve to the user name, user home directory resp. user
7563 shell. This is useful for running systemd user instances.
7565 * We now automatically rotate journal files if their data
7566 object hash table gets a fill level > 75%. We also size the
7567 hash table based on the configured maximum file size. This
7568 together should lower hash collisions drastically and thus
7569 speed things up a bit.
7571 * journalctl gained the new "--header" switch to introspect
7572 header data of journal files.
7574 * A new setting SystemCallFilters= has been added to services
7575 which may be used to apply blacklists or whitelists to
7576 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7578 * nspawn gained a new --link-journal= switch (and quicker: -j)
7579 to link the container journal with the host. This makes it
7580 very easy to centralize log viewing on the host for all
7581 guests while still keeping the journal files separated.
7583 * Many bugfixes and optimizations
7585 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7586 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7587 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7592 * Several tools now understand kernel command line arguments,
7593 which are only read when run in an initial RAM disk. They
7594 usually follow closely their normal counterparts, but are
7597 * There's a new tool to analyze the readahead files that are
7598 automatically generated at boot. Use:
7600 /usr/lib/systemd/systemd-readahead analyze /.readahead
7602 * We now provide an early debug shell on tty9 if this enabled. Use:
7604 systemctl enable debug-shell.service
7606 * All plymouth related units have been moved into the Plymouth
7607 package. Please make sure to upgrade your Plymouth version
7610 * systemd-tmpfiles now supports getting passed the basename of
7611 a configuration file only, in which case it will look for it
7612 in all appropriate directories automatically.
7614 * udevadm info now takes a /dev or /sys path as argument, and
7615 does the right thing. Example:
7617 udevadm info /dev/sda
7618 udevadm info /sys/class/block/sda
7620 * systemctl now prints a warning if a unit is stopped but a
7621 unit that might trigger it continues to run. Example: a
7622 service is stopped but the socket that activates it is left
7625 * "systemctl status" will now mention if the log output was
7626 shortened due to rotation since a service has been started.
7628 * The journal API now exposes functions to determine the
7629 "cutoff" times due to rotation.
7631 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7632 immediately flushing of runtime logs to /var if possible,
7633 resp. for triggering immediate rotation of the journal
7636 * It is now considered an error if a service is attempted to
7637 be stopped that is not loaded.
7639 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7641 * systemd-analyze now supports Python 3
7643 * tmpfiles now supports cleaning up directories via aging
7644 where the first level dirs are always kept around but
7645 directories beneath it automatically aged. This is enabled
7646 by prefixing the age field with '~'.
7648 * Seat objects now expose CanGraphical, CanTTY properties
7649 which is required to deal with very fast bootups where the
7650 display manager might be running before the graphics drivers
7651 completed initialization.
7653 * Seat objects now expose a State property.
7655 * We now include RPM macros for service enabling/disabling
7656 based on the preset logic. We recommend RPM based
7657 distributions to make use of these macros if possible. This
7658 makes it simpler to reuse RPM spec files across
7661 * We now make sure that the collected systemd unit name is
7662 always valid when services log to the journal via
7665 * There's a new man page kernel-command-line(7) detailing all
7666 command line options we understand.
7668 * The fstab generator may now be disabled at boot by passing
7669 fstab=0 on the kernel command line.
7671 * A new kernel command line option modules-load= is now understood
7672 to load a specific kernel module statically, early at boot.
7674 * Unit names specified on the systemctl command line are now
7675 automatically escaped as needed. Also, if file system or
7676 device paths are specified they are automatically turned
7677 into the appropriate mount or device unit names. Example:
7679 systemctl status /home
7680 systemctl status /dev/sda
7682 * The SysVConsole= configuration option has been removed from
7683 system.conf parsing.
7685 * The SysV search path is no longer exported on the D-Bus
7688 * The Names= option has been removed from unit file parsing.
7690 * There's a new man page bootup(7) detailing the boot process.
7692 * Every unit and every generator we ship with systemd now
7693 comes with full documentation. The self-explanatory boot is
7696 * A couple of services gained "systemd-" prefixes in their
7697 name if they wrap systemd code, rather than only external
7698 code. Among them fsck@.service which is now
7699 systemd-fsck@.service.
7701 * The HaveWatchdog property has been removed from the D-Bus
7704 * systemd.confirm_spawn= on the kernel command line should now
7707 * There's a new man page crypttab(5) which details all options
7708 we actually understand.
7710 * systemd-nspawn gained a new --capability= switch to pass
7711 additional capabilities to the container.
7713 * timedated will now read known NTP implementation unit names
7714 from /usr/lib/systemd/ntp-units.d/*.list,
7715 systemd-timedated-ntp.target has been removed.
7717 * journalctl gained a new switch "-b" that lists log data of
7718 the current boot only.
7720 * The notify socket is in the abstract namespace again, in
7721 order to support daemons which chroot() at start-up.
7723 * There is a new Storage= configuration option for journald
7724 which allows configuration of where log data should go. This
7725 also provides a way to disable journal logging entirely, so
7726 that data collected is only forwarded to the console, the
7727 kernel log buffer or another syslog implementation.
7729 * Many bugfixes and optimizations
7731 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7732 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7733 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7734 Shawn Landden, Tom Gundersen
7738 * "systemctl help <unit>" now shows the man page if one is
7741 * Several new man pages have been added.
7743 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7744 MaxLevelConsole= can now be specified in
7745 journald.conf. These options allow reducing the amount of
7746 data stored on disk or forwarded by the log level.
7748 * TimerSlackNSec= can now be specified in system.conf for
7749 PID1. This allows system-wide power savings.
7751 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7752 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7757 * logind is now capable of (optionally) handling power and
7758 sleep keys as well as the lid switch.
7760 * journalctl now understands the syntax "journalctl
7761 /usr/bin/avahi-daemon" to get all log output of a specific
7764 * CapabilityBoundingSet= in system.conf now also influences
7765 the capability bound set of usermode helpers of the kernel.
7767 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7768 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7769 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7770 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7774 * Note that we skipped 139 releases here in order to set the
7775 new version to something that is greater than both udev's
7776 and systemd's most recent version number.
7778 * udev: all udev sources are merged into the systemd source tree now.
7779 All future udev development will happen in the systemd tree. It
7780 is still fully supported to use the udev daemon and tools without
7781 systemd running, like in initramfs or other init systems. Building
7782 udev though, will require the *build* of the systemd tree, but
7783 udev can be properly *run* without systemd.
7785 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7786 should be used to create dead device nodes as workarounds for broken
7789 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7790 no longer supported. udev_monitor_new_from_netlink() needs to be
7791 used to subscribe to events.
7793 * udev: when udevd is started by systemd, processes which are left
7794 behind by forking them off of udev rules, are unconditionally cleaned
7795 up and killed now after the event handling has finished. Services or
7796 daemons must be started as systemd services. Services can be
7797 pulled-in by udev to get started, but they can no longer be directly
7798 forked by udev rules.
7800 * udev: the daemon binary is called systemd-udevd now and installed
7801 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7802 to adapt to that, create symlink, or rename the binary after building
7805 * libudev no longer provides these symbols:
7806 udev_monitor_from_socket()
7807 udev_queue_get_failed_list_entry()
7808 udev_get_{dev,sys,run}_path()
7809 The versions number was bumped and symbol versioning introduced.
7811 * systemd-loginctl and systemd-journalctl have been renamed
7812 to loginctl and journalctl to match systemctl.
7814 * The config files: /etc/systemd/systemd-logind.conf and
7815 /etc/systemd/systemd-journald.conf have been renamed to
7816 logind.conf and journald.conf. Package updates should rename
7817 the files to the new names on upgrade.
7819 * For almost all files the license is now LGPL2.1+, changed
7820 from the previous GPL2.0+. Exceptions are some minor stuff
7821 of udev (which will be changed to LGPL2.1 eventually, too),
7822 and the MIT licensed sd-daemon.[ch] library that is suitable
7823 to be used as drop-in files.
7825 * systemd and logind now handle system sleep states, in
7826 particular suspending and hibernating.
7828 * logind now implements a sleep/shutdown/idle inhibiting logic
7829 suitable for a variety of uses. Soonishly Lennart will blog
7830 about this in more detail.
7832 * var-run.mount and var-lock.mount are no longer provided
7833 (which previously bind mounted these directories to their new
7834 places). Distributions which have not converted these
7835 directories to symlinks should consider stealing these files
7836 from git history and add them downstream.
7838 * We introduced the Documentation= field for units and added
7839 this to all our shipped units. This is useful to make it
7840 easier to explore the boot and the purpose of the various
7843 * All smaller setup units (such as
7844 systemd-vconsole-setup.service) now detect properly if they
7845 are run in a container and are skipped when
7846 appropriate. This guarantees an entirely noise-free boot in
7847 Linux container environments such as systemd-nspawn.
7849 * A framework for implementing offline system updates is now
7850 integrated, for details see:
7851 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7853 * A new service type Type=idle is available now which helps us
7854 avoiding ugly interleaving of getty output and boot status
7857 * There's now a system-wide CapabilityBoundingSet= option to
7858 globally reduce the set of capabilities for the
7859 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7860 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7861 even CAP_NET_ADMIN system-wide for secure systems.
7863 * There are now system-wide DefaultLimitXXX= options to
7864 globally change the defaults of the various resource limits
7865 for all units started by PID 1.
7867 * Harald Hoyer's systemd test suite has been integrated into
7868 systemd which allows easy testing of systemd builds in qemu
7869 and nspawn. (This is really awesome! Ask us for details!)
7871 * The fstab parser is now implemented as generator, not inside
7874 * systemctl will now warn you if .mount units generated from
7875 /etc/fstab are out of date due to changes in fstab that
7876 have not been read by systemd yet.
7878 * systemd is now suitable for usage in initrds. Dracut has
7879 already been updated to make use of this. With this in place
7880 initrds get a slight bit faster but primarily are much
7881 easier to introspect and debug since "systemctl status" in
7882 the host system can be used to introspect initrd services,
7883 and the journal from the initrd is kept around too.
7885 * systemd-delta has been added, a tool to explore differences
7886 between user/admin configuration and vendor defaults.
7888 * PrivateTmp= now affects both /tmp and /var/tmp.
7890 * Boot time status messages are now much prettier and feature
7891 proper english language. Booting up systemd has never been
7894 * Read-ahead pack files now include the inode number of all
7895 files to pre-cache. When the inode changes the pre-caching
7896 is not attempted. This should be nicer to deal with updated
7897 packages which might result in changes of read-ahead
7900 * We now temporaritly lower the kernel's read_ahead_kb variable
7901 when collecting read-ahead data to ensure the kernel's
7902 built-in read-ahead does not add noise to our measurements
7903 of necessary blocks to pre-cache.
7905 * There's now RequiresMountsFor= to add automatic dependencies
7906 for all mounts necessary for a specific file system path.
7908 * MountAuto= and SwapAuto= have been removed from
7909 system.conf. Mounting file systems at boot has to take place
7912 * nspawn now learned a new switch --uuid= to set the machine
7913 ID on the command line.
7915 * nspawn now learned the -b switch to automatically search
7918 * vt102 is now the default TERM for serial TTYs, upgraded from
7921 * systemd-logind now works on VT-less systems.
7923 * The build tree has been reorganized. The individual
7924 components now have directories of their own.
7926 * A new condition type ConditionPathIsReadWrite= is now available.
7928 * nspawn learned the new -C switch to create cgroups for the
7929 container in other hierarchies.
7931 * We now have support for hardware watchdogs, configurable in
7934 * The scheduled shutdown logic now has a public API.
7936 * We now mount /tmp as tmpfs by default, but this can be
7937 masked and /etc/fstab can override it.
7939 * Since udisks does not make use of /media anymore we are not
7940 mounting a tmpfs on it anymore.
7942 * journalctl gained a new --local switch to only interleave
7943 locally generated journal files.
7945 * We can now load the IMA policy at boot automatically.
7947 * The GTK tools have been split off into a systemd-ui.
7949 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7950 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7951 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7952 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7953 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7954 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7955 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7956 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7957 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7962 * This is mostly a bugfix release
7964 * Support optional initialization of the machine ID from the
7965 KVM or container configured UUID.
7967 * Support immediate reboots with "systemctl reboot -ff"
7969 * Show /etc/os-release data in systemd-analyze output
7971 * Many bugfixes for the journal, including endianness fixes and
7972 ensuring that disk space enforcement works
7974 * sd-login.h is C++ compatible again
7976 * Extend the /etc/os-release format on request of the Debian
7979 * We now refuse non-UTF8 strings used in various configuration
7980 and unit files. This is done to ensure we do not pass invalid
7981 data over D-Bus or expose it elsewhere.
7983 * Register Mimo USB Screens as suitable for automatic seat
7986 * Read SELinux client context from journal clients in a race
7989 * Reorder configuration file lookup order. /etc now always
7990 overrides /run in order to allow the administrator to always
7991 and unconditionally override vendor-supplied or
7992 automatically generated data.
7994 * The various user visible bits of the journal now have man
7995 pages. We still lack man pages for the journal API calls
7998 * We now ship all man pages in HTML format again in the
8001 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8002 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8003 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8008 * This is mostly a bugfix release
8010 * systems lacking /etc/os-release are no longer supported.
8012 * Various functionality updates to libsystemd-login.so
8014 * Track class of PAM logins to distinguish greeters from
8017 Contributions from: Kay Sievers, Lennart Poettering, Michael
8022 * This is an important bugfix release for v41.
8024 * Building man pages is now optional which should be useful
8025 for those building systemd from git but unwilling to install
8028 * Watchdog support for supervising services is now usable. In
8029 a future release support for hardware watchdogs
8030 (i.e. /dev/watchdog) will be added building on this.
8032 * Service start rate limiting is now configurable and can be
8033 turned off per service. When a start rate limit is hit a
8034 reboot can automatically be triggered.
8036 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8038 Contributions from: Benjamin Franzke, Bill Nottingham,
8039 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8040 Schmidt, Michał Górny, Piotr Drąg
8044 * The systemd binary is installed /usr/lib/systemd/systemd now;
8045 An existing /sbin/init symlink needs to be adapted with the
8048 * The code that loads kernel modules has been ported to invoke
8049 libkmod directly, instead of modprobe. This means we do not
8050 support systems with module-init-tools anymore.
8052 * Watchdog support is now already useful, but still not
8055 * A new kernel command line option systemd.setenv= is
8056 understood to set system wide environment variables
8057 dynamically at boot.
8059 * We now limit the set of capabilities of systemd-journald.
8061 * We now set SIGPIPE to ignore by default, since it only is
8062 useful in shell pipelines, and has little use in general
8063 code. This can be disabled with IgnoreSIPIPE=no in unit
8066 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8067 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8072 * This is mostly a bugfix release
8074 * We now expose the reason why a service failed in the
8075 "Result" D-Bus property.
8077 * Rudimentary service watchdog support (will be completed over
8078 the next few releases.)
8080 * When systemd forks off in order execute some service we will
8081 now immediately changes its argv[0] to reflect which process
8082 it will execute. This is useful to minimize the time window
8083 with a generic argv[0], which makes bootcharts more useful
8085 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8086 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8087 Mike Kazantsev, Ray Strode
8091 * This is mostly a test release, but incorporates many
8094 * New systemd-cgtop tool to show control groups by their
8097 * Linking against libacl for ACLs is optional again. If
8098 disabled, support tracking device access for active logins
8099 goes becomes unavailable, and so does access to the user
8100 journals by the respective users.
8102 * If a group "adm" exists, journal files are automatically
8103 owned by them, thus allow members of this group full access
8104 to the system journal as well as all user journals.
8106 * The journal now stores the SELinux context of the logging
8107 client for all entries.
8109 * Add C++ inclusion guards to all public headers
8111 * New output mode "cat" in the journal to print only text
8112 messages, without any meta data like date or time.
8114 * Include tiny X server wrapper as a temporary stop-gap to
8115 teach XOrg udev display enumeration. This is used by display
8116 managers such as gdm, and will go away as soon as XOrg
8117 learned native udev hotplugging for display devices.
8119 * Add new systemd-cat tool for executing arbitrary programs
8120 with STDERR/STDOUT connected to the journal. Can also act as
8121 BSD logger replacement, and does so by default.
8123 * Optionally store all locally generated coredumps in the
8124 journal along with meta data.
8126 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8127 writing short strings to files (for usage for /sys), and for
8128 creating symlinks, character and block device nodes.
8130 * New unit file option ControlGroupPersistent= to make cgroups
8131 persistent, following the mechanisms outlined in
8132 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8134 * Support multiple local RTCs in a sane way
8136 * No longer monopolize IO when replaying readahead data on
8137 rotating disks, since we might starve non-file-system IO to
8138 death, since fanotify() will not see accesses done by blkid,
8141 * Do not show kernel threads in systemd-cgls anymore, unless
8142 requested with new -k switch.
8144 Contributions from: Dan Horák, Kay Sievers, Lennart
8145 Poettering, Michal Schmidt
8149 * This is mostly a test release, but incorporates many
8152 * The git repository moved to:
8153 git://anongit.freedesktop.org/systemd/systemd
8154 ssh://git.freedesktop.org/git/systemd/systemd
8156 * First release with the journal
8157 http://0pointer.de/blog/projects/the-journal.html
8159 * The journal replaces both systemd-kmsg-syslogd and
8160 systemd-stdout-bridge.
8162 * New sd_pid_get_unit() API call in libsystemd-logind
8164 * Many systemadm clean-ups
8166 * Introduce remote-fs-pre.target which is ordered before all
8167 remote mounts and may be used to start services before all
8170 * Added Mageia support
8172 * Add bash completion for systemd-loginctl
8174 * Actively monitor PID file creation for daemons which exit in
8175 the parent process before having finished writing the PID
8176 file in the daemon process. Daemons which do this need to be
8177 fixed (i.e. PID file creation must have finished before the
8178 parent exits), but we now react a bit more gracefully to them.
8180 * Add colourful boot output, mimicking the well-known output
8181 of existing distributions.
8183 * New option PassCredentials= for socket units, for
8184 compatibility with a recent kernel ABI breakage.
8186 * /etc/rc.local is now hooked in via a generator binary, and
8187 thus will no longer act as synchronization point during
8190 * systemctl list-unit-files now supports --root=.
8192 * systemd-tmpfiles now understands two new commands: z, Z for
8193 relabelling files according to the SELinux database. This is
8194 useful to apply SELinux labels to specific files in /sys,
8197 * Output of SysV services is now forwarded to both the console
8198 and the journal by default, not only just the console.
8200 * New man pages for all APIs from libsystemd-login.
8202 * The build tree got reorganized and the build system is a
8203 lot more modular allowing embedded setups to specifically
8204 select the components of systemd they are interested in.
8206 * Support for Linux systems lacking the kernel VT subsystem is
8209 * configure's --with-rootdir= got renamed to
8210 --with-rootprefix= to follow the naming used by udev and
8213 * Unless specified otherwise we will now install to /usr instead
8214 of /usr/local by default.
8216 * Processes with '@' in argv[0][0] are now excluded from the
8217 final shut-down killing spree, following the logic explained
8219 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8221 * All processes remaining in a service cgroup when we enter
8222 the START or START_PRE states are now killed with
8223 SIGKILL. That means it is no longer possible to spawn
8224 background processes from ExecStart= lines (which was never
8225 supported anyway, and bad style).
8227 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8228 reloading of units together.
8230 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8231 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8232 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8233 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8234 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob