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.
22 * The MemoryAccounting= unit property now defaults to on. After
23 discussions with the upstream control group maintainers we learnt
24 that the negative impact of cgroup memory accounting on current
25 kernels is finally relatively minimal, so that it should be safe to
26 enable this by default without affecting system performance. Besides
27 memory accounting only task accounting is turned on by default, all
28 other forms of resource accounting (CPU, IO, IP) remain off for now,
29 because it's not clear yet that their impact is small enough to move
30 from opt-in to opt-out. We recommend downstreams to leave memory
31 accounting on by default if kernel 4.14 or higher is primarily
32 used. On very resource constrained systems or when support for old
33 kernels is a necessity, -Dmemory-accounting-default=false can be used
34 to revert this change.
36 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
37 %udev_rules_update) and the journal catalog (%journal_catalog_update)
38 from the upgrade scriptlets of individual packages now do nothing.
39 Transfiletriggers have been added which will perform those updates
40 once at the end of the transaction.
42 Similar transfiletriggers have been added to execute any sysctl.d
43 and binfmt.d rules. Thus, it should be unnecessary to provide any
44 scriptlets to execute this configuration from package installation
47 * elogind-sysusers gained a mode where the configuration to execute is
48 specified on the command line, but this configuration is not executed
49 directly, but instead it is merged with the configuration on disk,
50 and the result is executed. This is useful for package installation
51 scripts which want to create the user before installing any files on
52 disk (in case some of those files are owned by that user), while
53 still allowing local admin overrides.
55 This functionality is exposed to rpm scriptlets through a new
56 %sysusers_create_package macro. Old %sysusers_create and
57 %sysusers_create_inline macros are deprecated.
59 A transfiletrigger for sysusers.d configuration is now installed,
60 which means that it should be unnecessary to call elogind-sysusers from
61 package installation scripts, unless the package installs any files
62 owned by those newly-created users, in which case
63 %sysusers_create_package should be used.
65 * Analogous change has been done for elogind-tmpfiles: it gained a mode
66 where the command-line configuration is merged with the configuration
67 on disk. This is exposed as the new %tmpfiles_create_package macro,
68 and %tmpfiles_create is deprecated. A transfiletrigger is installed
69 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
70 from package installation scripts.
72 * sysusers.d configuration for a user may now also specify the group
73 number, in addition to the user number ("u username 123:456"), or
74 without the user number ("u username -:456").
76 * Configution items for elogind-sysusers can now be specified as
77 positional arguments when the new --inline switch is used.
79 * The login shell of users created through sysusers.d may now be
80 specified (previously, it was always /bin/sh for root and
81 /sbin/nologin for other users).
83 * elogind-analyze gained a new --global switch to look at global user
84 configuration. It also gained a unit-paths verb to list the unit load
85 paths that are compiled into elogind (which can be used with
86 --elogind, --user, or --global).
88 * udevadm trigger gained a new --settle/-w option to wait for any
89 triggered events to finish (but just those, and not any other events
90 which are triggered meanwhile).
92 * The action that elogind takes when the lid is closed and the
93 machine is connected to external power can now be configured using
94 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
95 was determined by HandleLidSwitch=, and, for backwards compatibility,
96 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
98 * journalctl will periodically call sd_journal_process() to make it
99 resilient against inotify queue overruns when journal files are
100 rotated very quickly.
102 * Two new functions in libelogind — sd_bus_get_n_queued_read and
103 sd_bus_get_n_queued_write — may be used to check the number of
104 pending bus messages.
106 * elogind gained a new
107 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
108 which can be used to migrate foreign processes to scope and service
109 units. The primary user for this new API is elogind itself: the
110 elogind --user instance uses this call of the elogind --system
111 instance to migrate processes if it itself gets the request to
112 migrate processes and the kernel refuses this due to access
113 restrictions. Thanks to this "elogind-run --scope --user …" works
114 again in pure cgroups v2 environments when invoked from the user
117 * A new TemporaryFileSystem= setting can be used to mask out part of
118 the real file system tree with tmpfs mounts. This may be combined
119 with BindPaths= and BindReadOnlyPaths= to hide files or directories
120 not relevant to the unit, while still allowing some paths lower in
121 the tree to be accessed.
123 ProtectHome=tmpfs may now be used to hide user home and runtime
124 directories from units, in a way that is mostly equivalent to
125 "TemporaryFileSystem=/home /run/user /root".
127 * Non-service units are now started with KeyringMode=shared by default.
128 This means that mount and swapon and other mount tools have access
129 to keys in the main keyring.
131 * /sys/fs/bpf is now mounted automatically.
133 * QNX virtualization is now detected by elogind-detect-virt and may
134 be used in ConditionVirtualization=.
136 * IPAccounting= may now be enabled also for slice units.
138 * A new -Dsplit-bin= build configuration switch may be used to specify
139 whether bin and sbin directories are merged, or if they should be
140 included separately in $PATH and various listings of executable
141 directories. The build configuration scripts will try to autodetect
142 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
143 system, but distributions are encouraged to configure this
146 * A new -Dok-color= build configuration switch may be used to change
147 the colour of "OK" status messages.
149 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
150 PrivateNetwork=yes was buggy in previous versions of elogind. This
151 means that after the upgrade and daemon-reexec, any such units must
154 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
155 will not exclude read-only files owned by root from cleanup.
157 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
158 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
159 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
160 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
161 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
162 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
163 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
164 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
165 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
166 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
167 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
168 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
169 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
170 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
171 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
172 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
178 * Some keyboards come with a zoom see-saw or rocker which until now got
179 mapped to the Linux "zoomin/out" keys in hwdb. However, these
180 keycodes are not recognized by any major desktop. They now produce
181 Up/Down key events so that they can be used for scrolling.
183 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
184 slightly: previously, if an argument was specified for lines of this
185 type (i.e. the right-most column was set) this string was appended to
186 existing files each time elogind-tmpfiles was run. This behaviour was
187 different from what the documentation said, and not particularly
188 useful, as repeated elogind-tmpfiles invocations would not be
189 idempotent and grow such files without bounds. With this release
190 behaviour has been altered slightly, to match what the documentation
191 says: lines of this type only have an effect if the indicated files
192 don't exist yet, and only then the argument string is written to the
195 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
196 elogind-tmpfiles behaviour: previously, read-only files owned by root
197 were always excluded from the file "aging" algorithm (i.e. the
198 automatic clean-up of directories like /tmp based on
199 atime/mtime/ctime). We intend to drop this restriction, and age files
200 by default even when owned by root and read-only. This behaviour was
201 inherited from older tools, but there have been requests to remove
202 it, and it's not obvious why this restriction was made in the first
203 place. Please speak up now, if you are aware of software that reqires
204 this behaviour, otherwise we'll remove the restriction in v238.
206 * A new environment variable $SYSTEMD_OFFLINE is now understood by
207 systemctl. It takes a boolean argument. If on, systemctl assumes it
208 operates on an "offline" OS tree, and will not attempt to talk to the
209 service manager. Previously, this mode was implicitly enabled if a
210 chroot() environment was detected, and this new environment variable
211 now provides explicit control.
213 * .path and .socket units may now be created transiently, too.
214 Previously only service, mount, automount and timer units were
215 supported as transient units. The elogind-run tool has been updated
216 to expose this new functionality, you may hence use it now to bind
217 arbitrary commands to path or socket activation on-the-fly from the
218 command line. Moreover, almost all properties are now exposed for the
219 unit types that already supported transient operation.
221 * The elogind-mount command gained support for a new --owner= parameter
222 which takes a user name, which is then resolved and included in uid=
223 and gid= mount options string of the file system to mount.
225 * A new unit condition ConditionControlGroupController= has been added
226 that checks whether a specific cgroup controller is available.
228 * Unit files, udev's .link files, and elogind-networkd's .netdev and
229 .network files all gained support for a new condition
230 ConditionKernelVersion= for checking against specific kernel
233 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
234 support for configuring device flags in the Flags= setting. In the
235 same files, the [Tunnel] section gained support for configuring
236 AllowLocalRemote=. The [Route] section in .network files gained
237 support for configuring InitialCongestionWindow=,
238 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
239 understands RapidCommit=.
241 * elogind-networkd's DHCPv6 support gained support for Prefix
244 * sd-bus gained support for a new "watch-bind" feature. When this
245 feature is enabled, an sd_bus connection may be set up to connect to
246 an AF_UNIX socket in the file system as soon as it is created. This
247 functionality is useful for writing early-boot services that
248 automatically connect to the system bus as soon as it is started,
249 without ugly time-based polling. elogind-networkd and
250 elogind-resolved have been updated to make use of this
251 functionality. busctl exposes this functionality in a new
252 --watch-bind= command line switch.
254 * sd-bus will now optionally synthesize a local "Connected" signal as
255 soon as a D-Bus connection is set up fully. This message mirrors the
256 already existing "Disconnected" signal which is synthesized when the
257 connection is terminated. This signal is generally useful but
258 particularly handy in combination with the "watch-bind" feature
259 described above. Synthesizing of this message has to be requested
260 explicitly through the new API call sd_bus_set_connected_signal(). In
261 addition a new call sd_bus_is_ready() has been added that checks
262 whether a connection is fully set up (i.e. between the "Connected" and
263 "Disconnected" signals).
265 * sd-bus gained two new calls sd_bus_request_name_async() and
266 sd_bus_release_name_async() for asynchronously registering bus
267 names. Similar, there is now sd_bus_add_match_async() for installing
268 a signal match asynchronously. All of elogind's own services have
269 been updated to make use of these calls. Doing these operations
270 asynchronously has two benefits: it reduces the risk of deadlocks in
271 case of cyclic dependencies between bus services, and it speeds up
272 service initialization since synchronization points for bus
273 round-trips are removed.
275 * sd-bus gained two new calls sd_bus_match_signal() and
276 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
277 and sd_bus_add_match_async() but instead of taking a D-Bus match
278 string take match fields as normal function parameters.
280 * sd-bus gained two new calls sd_bus_set_sender() and
281 sd_bus_message_set_sender() for setting the sender name of outgoing
282 messages (either for all outgoing messages or for just one specific
283 one). These calls are only useful in direct connections as on
284 brokered connections the broker fills in the sender anyway,
285 overwriting whatever the client filled in.
287 * sd-event gained a new pseudo-handle that may be specified on all API
288 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
289 used this refers to the default event loop object of the calling
290 thread. Note however that this does not implicitly allocate one —
291 which has to be done prior by using sd_event_default(). Similarly
292 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
293 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
294 to the default bus of the specified type of the calling thread. Here
295 too this does not implicitly allocate bus connection objects, this
296 has to be done prior with sd_bus_default() and friends.
298 * sd-event gained a new call pair
299 sd_event_source_{get|set}_io_fd_own(). This may be used to request
300 automatic closure of the file descriptor an IO event source watches
301 when the event source is destroyed.
303 * elogind-networkd gained support for natively configuring WireGuard
306 * In previous versions elogind synthesized user records both for the
307 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
308 internally. In order to simplify distribution-wide renames of the
309 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
310 new transitional flag file has been added: if
311 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
312 user and group record within the elogind codebase is disabled.
314 * elogind-notify gained a new --uid= option for selecting the source
315 user/UID to use for notification messages sent to the service
318 * journalctl gained a new --grep= option to list only entries in which
319 the message matches a certain pattern. By default matching is case
320 insensitive if the pattern is lowercase, and case sensitive
321 otherwise. Option --case-sensitive=yes|no can be used to override
322 this an specify case sensitivity or case insensitivity.
324 * There's now a "elogind-analyze service-watchdogs" command for printing
325 the current state of the service runtime watchdog, and optionally
326 enabling or disabling the per-service watchdogs system-wide if given a
327 boolean argument (i.e. the concept you configure in WatchdogSec=), for
328 debugging purposes. There's also a kernel command line option
329 elogind.service_watchdogs= for controlling the same.
331 * Two new "log-level" and "log-target" options for elogind-analyze were
332 addded that merge the now deprecated get-log-level, set-log-level and
333 get-log-target, set-log-target pairs. The deprecated options are still
334 understood for backwards compatibility. The two new options print the
335 current value when no arguments are given, and set them when a
336 level/target is given as an argument.
338 * sysusers.d's "u" lines now optionally accept both a UID and a GID
339 specification, separated by a ":" character, in order to create users
340 where UID and GID do not match.
342 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
343 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
344 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
345 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
346 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
347 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
348 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
349 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
350 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
351 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
352 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
353 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
354 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
355 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
356 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
357 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
358 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
359 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
360 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
361 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
368 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
369 in v235 has been extended to also set the dummy.ko module option
370 numdummies=0, preventing the kernel from automatically creating
371 dummy0. All dummy interfaces must now be explicitly created.
373 * Unknown '%' specifiers in configuration files are now rejected. This
374 applies to units and tmpfiles.d configuration. Any percent characters
375 that are followed by a letter or digit that are not supposed to be
376 interpreted as the beginning of a specifier should be escaped by
377 doubling ("%%"). (So "size=5%" is still accepted, as well as
378 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
379 valid specifiers today.)
381 * systemd-resolved now maintains a new dynamic
382 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
383 recommended to make /etc/resolv.conf a symlink to it. This file
384 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
385 includes dynamically acquired search domains, achieving more correct
386 DNS resolution by software that bypasses local DNS APIs such as NSS.
388 * The "uaccess" udev tag has been dropped from /dev/kvm and
389 /dev/dri/renderD*. These devices now have the 0666 permissions by
390 default (but this may be changed at build-time). /dev/dri/renderD*
391 will now be owned by the "render" group along with /dev/kfd.
393 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
394 systemd-journal-gatewayd.service and
395 systemd-journal-upload.service. This means "nss-systemd" must be
396 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
397 services are resolved properly.
399 * In /etc/fstab two new mount options are now understood:
400 x-systemd.makefs and x-systemd.growfs. The former has the effect that
401 the configured file system is formatted before it is mounted, the
402 latter that the file system is resized to the full block device size
403 after it is mounted (i.e. if the file system is smaller than the
404 partition it resides on, it's grown). This is similar to the fsck
405 logic in /etc/fstab, and pulls in systemd-makefs@.service and
406 systemd-growfs@.service as necessary, similar to
407 systemd-fsck@.service. Resizing is currently only supported on ext4
410 * In systemd-networkd, the IPv6 RA logic now optionally may announce
411 DNS server and domain information.
413 * Support for the LUKS2 on-disk format for encrypted partitions has
414 been added. This requires libcryptsetup2 during compilation and
417 * The systemd --user instance will now signal "readiness" when its
418 basic.target unit has been reached, instead of when the run queue ran
419 empty for the first time.
421 * Tmpfiles.d with user configuration are now also supported.
422 systemd-tmpfiles gained a new --user switch, and snippets placed in
423 ~/.config/user-tmpfiles.d/ and corresponding directories will be
424 executed by systemd-tmpfiles --user running in the new
425 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
426 running in the user session.
428 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
429 %S resolves to the top-level state directory (/var/lib for the system
430 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
431 top-level cache directory (/var/cache for the system instance,
432 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
433 logs directory (/var/log for the system instance,
434 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
435 existing %t specifier, that resolves to the top-level runtime
436 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
439 * journalctl learnt a new parameter --output-fields= for limiting the
440 set of journal fields to output in verbose and JSON output modes.
442 * systemd-timesyncd's configuration file gained a new option
443 RootDistanceMaxSec= for setting the maximum root distance of servers
444 it'll use, as well as the new options PollIntervalMinSec= and
445 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
447 * bootctl gained a new command "list" for listing all available boot
448 menu items on systems that follow the boot loader specification.
450 * systemctl gained a new --dry-run switch that shows what would be done
451 instead of doing it, and is currently supported by the shutdown and
454 * ConditionSecurity= can now detect the TOMOYO security module.
456 * Unit file [Install] sections are now also respected in unit drop-in
457 files. This is intended to be used by drop-ins under /usr/lib/.
459 * systemd-firstboot may now also set the initial keyboard mapping.
461 * Udev "changed" events for devices which are exposed as systemd
462 .device units are now propagated to units specified in
463 ReloadPropagatedFrom= as reload requests.
465 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
466 unit template name (i.e. a name in the form of 'foobar@.service',
467 without the instance component between the '@' and - the '.'), then
468 the escaped sysfs path of the device is automatically used as the
471 * SystemCallFilter= in unit files has been extended so that an "errno"
472 can be specified individually for each system call. Example:
473 SystemCallFilter=~uname:EILSEQ.
475 * The cgroup delegation logic has been substantially updated. Delegate=
476 now optionally takes a list of controllers (instead of a boolean, as
477 before), which lists the controllers to delegate at least.
479 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
481 * A new LogLevelMax= setting configures the maximum log level any
482 process of the service may log at (i.e. anything with a lesser
483 priority than what is specified is automatically dropped). A new
484 LogExtraFields= setting allows configuration of additional journal
485 fields to attach to all log records generated by any of the unit's
488 * New StandardInputData= and StandardInputText= settings along with the
489 new option StandardInput=data may be used to configure textual or
490 binary data that shall be passed to the executed service process via
491 standard input, encoded in-line in the unit file.
493 * StandardInput=, StandardOutput= and StandardError= may now be used to
494 connect stdin/stdout/stderr of executed processes directly with a
495 file or AF_UNIX socket in the file system, using the new "file:" option.
497 * A new unit file option CollectMode= has been added, that allows
498 tweaking the garbage collection logic for units. It may be used to
499 tell systemd to garbage collect units that have failed automatically
500 (normally it only GCs units that exited successfully). systemd-run
501 and systemd-mount expose this new functionality with a new -G option.
503 * "machinectl bind" may now be used to bind mount non-directories
504 (i.e. regularfiles, devices, fifos, sockets).
506 * systemd-analyze gained a new verb "calendar" for validating and
507 testing calendar time specifications to use for OnCalendar= in timer
508 units. Besides validating the expression it will calculate the next
509 time the specified expression would elapse.
511 * In addition to the pre-existing FailureAction= unit file setting
512 there's now SuccessAction=, for configuring a shutdown action to
513 execute when a unit completes successfully. This is useful in
514 particular inside containers that shall terminate after some workload
515 has been completed. Also, both options are now supported for all unit
516 types, not just services.
518 * networkds's IP rule support gained two new options
519 IncomingInterface= and OutgoingInterface= for configuring the incoming
520 and outgoing interfaces of configured rules. systemd-networkd also
521 gained support for "vxcan" network devices.
523 * networkd gained a new setting RequiredForOnline=, taking a
524 boolean. If set, systemd-wait-online will take it into consideration
525 when determining that the system is up, otherwise it will ignore the
526 interface for this purpose.
528 * The sd_notify() protocol gained support for a new operation: with
529 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
530 store again, ahead of POLLHUP or POLLERR when they are removed
533 * A new document doc/UIDS-GIDS.md has been added to the source tree,
534 that documents the UID/GID range and assignment assumptions and
535 requirements of systemd.
537 * The watchdog device PID 1 will ping may now be configured through the
538 WatchdogDevice= configuration file setting, or by setting the
539 systemd.watchdog_service= kernel commandline option.
541 * systemd-resolved's gained support for registering DNS-SD services on
542 the local network using MulticastDNS. Services may either be
543 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
544 the same dir below /run, /usr/lib), or through its D-Bus API.
546 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
547 extend the effective start, runtime, and stop time. The service must
548 continue to send EXTEND_TIMEOUT_USEC within the period specified to
549 prevent the service manager from making the service as timedout.
551 * elogind-resolved's DNSSEC support gained support for RFC 8080
552 (Ed25519 keys and signatures).
554 * The elogind-resolve command line tool gained a new set of options
555 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
556 --set-nta= and --revert to configure per-interface DNS configuration
557 dynamically during runtime. It's useful for pushing DNS information
558 into elogind-resolved from DNS hook scripts that various interface
559 managing software supports (such as pppd).
561 * elogind-nspawn gained a new --network-namespace-path= command line
562 option, which may be used to make a container join an existing
563 network namespace, by specifying a path to a "netns" file.
565 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
566 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
567 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
568 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
569 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
570 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
571 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
572 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
573 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
574 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
575 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
576 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
577 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
578 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
579 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
580 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
581 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
582 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
583 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
584 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
585 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
586 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
587 Jędrzejewski-Szmek, Zeal Jagannatha
593 * INCOMPATIBILITY: systemd-logind.service and other long-running
594 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
595 communication with the outside. This generally improves security of
596 the system, and is in almost all cases a safe and good choice, as
597 these services do not and should not provide any network-facing
598 functionality. However, systemd-logind uses the glibc NSS API to
599 query the user database. This creates problems on systems where NSS
600 is set up to directly consult network services for user database
601 lookups. In particular, this creates incompatibilities with the
602 "nss-nis" module, which attempts to directly contact the NIS/YP
603 network servers it is configured for, and will now consistently
604 fail. In such cases, it is possible to turn off IP sandboxing for
605 systemd-logind.service (set IPAddressDeny= in its [Service] section
606 to the empty string, via a .d/ unit file drop-in). Downstream
607 distributions might want to update their nss-nis packaging to include
608 such a drop-in snippet, accordingly, to hide this incompatibility
609 from the user. Another option is to make use of glibc's nscd service
610 to proxy such network requests through a privilege-separated, minimal
611 local caching daemon, or to switch to more modern technologies such
612 sssd, whose NSS hook-ups generally do not involve direct network
613 access. In general, we think it's definitely time to question the
614 implementation choices of nss-nis, i.e. whether it's a good idea
615 today to embed a network-facing loadable module into all local
616 processes that need to query the user database, including the most
617 trivial and benign ones, such as "ls". For more details about
618 IPAddressDeny= see below.
620 * A new modprobe.d drop-in is now shipped by default that sets the
621 bonding module option max_bonds=0. This overrides the kernel default,
622 to avoid conflicts and ambiguity as to whether or not bond0 should be
623 managed by systemd-networkd or not. This resolves multiple issues
624 with bond0 properties not being applied, when bond0 is configured
625 with systemd-networkd. Distributors may choose to not package this,
626 however in that case users will be prevented from correctly managing
627 bond0 interface using systemd-networkd.
629 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
630 which print the logging level and target of the system manager. They
631 complement the existing "set-log-level" and "set-log-target" verbs
632 used to change those values.
634 * journald.conf gained a new boolean setting ReadKMsg= which defaults
635 to on. If turned off kernel log messages will not be read by
636 systemd-journald or included in the logs. It also gained a new
637 setting LineMax= for configuring the maximum line length in
638 STDOUT/STDERR log streams. The new default for this value is 48K, up
639 from the previous hardcoded 2048.
641 * A new unit setting RuntimeDirectoryPreserve= has been added, which
642 allows more detailed control of what to do with a runtime directory
643 configured with RuntimeDirectory= (i.e. a directory below /run or
644 $XDG_RUNTIME_DIR) after a unit is stopped.
646 * The RuntimeDirectory= setting for units gained support for creating
647 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
648 one top-level directory.
650 * Units gained new options StateDirectory=, CacheDirectory=,
651 LogsDirectory= and ConfigurationDirectory= which are closely related
652 to RuntimeDirectory= but manage per-service directories below
653 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
654 possible to write unit files which when activated automatically gain
655 properly owned service specific directories in these locations, thus
656 making unit files self-contained and increasing compatibility with
657 stateless systems and factory reset where /etc or /var are
658 unpopulated at boot. Matching these new settings there's also
659 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
660 ConfigurationDirectoryMode= for configuring the access mode of these
661 directories. These settings are particularly useful in combination
662 with DynamicUser=yes as they provide secure, properly-owned,
663 writable, and stateful locations for storage, excluded from the
664 sandbox that such services live in otherwise.
666 * Automake support has been removed from this release. systemd is now
669 * systemd-journald will now aggressively cache client metadata during
670 runtime, speeding up log write performance under pressure. This comes
671 at a small price though: as much of the metadata is read
672 asynchronously from /proc/ (and isn't implicitly attached to log
673 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
674 metadata stored alongside a log entry might be slightly
675 out-of-date. Previously it could only be slightly newer than the log
676 message. The time window is small however, and given that the kernel
677 is unlikely to be improved anytime soon in this regard, this appears
680 * nss-myhostname/systemd-resolved will now by default synthesize an
681 A/AAAA resource record for the "_gateway" hostname, pointing to the
682 current default IP gateway. Previously it did that for the "gateway"
683 name, hampering adoption, as some distributions wanted to leave that
684 host name open for local use. The old behaviour may still be
685 requested at build time.
687 * systemd-networkd's [Address] section in .network files gained a new
688 Scope= setting for configuring the IP address scope. The [Network]
689 section gained a new boolean setting ConfigureWithoutCarrier= that
690 tells systemd-networkd to ignore link sensing when configuring the
691 device. The [DHCP] section gained a new Anonymize= boolean option for
692 turning on a number of options suggested in RFC 7844. A new
693 [RoutingPolicyRule] section has been added for configuring the IP
694 routing policy. The [Route] section has gained support for a new
695 Type= setting which permits configuring
696 blackhole/unreachable/prohibit routes.
698 * The [VRF] section in .netdev files gained a new Table= setting for
699 configuring the routing table to use. The [Tunnel] section gained a
700 new Independent= boolean field for configuring tunnels independent of
701 an underlying network interface. The [Bridge] section gained a new
702 GroupForwardMask= option for configuration of propagation of link
703 local frames between bridge ports.
705 * The WakeOnLan= setting in .link files gained support for a number of
706 new modes. A new TCP6SegmentationOffload= setting has been added for
707 configuring TCP/IPv6 hardware segmentation offload.
709 * The IPv6 RA sender implementation may now optionally send out RDNSS
710 and RDNSSL records to supply DNS configuration to peers.
712 * systemd-nspawn gained support for a new --system-call-filter= command
713 line option for adding and removing entries in the default system
714 call filter it applies. Moreover systemd-nspawn has been changed to
715 implement a system call whitelist instead of a blacklist.
717 * systemd-run gained support for a new --pipe command line option. If
718 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
719 are directly passed on to the activated transient service
720 executable. This allows invoking arbitrary processes as systemd
721 services (for example to take benefit of dependency management,
722 accounting management, resource management or log management that is
723 done automatically for services) — while still allowing them to be
724 integrated in a classic UNIX shell pipeline.
726 * When a service sends RELOAD=1 via sd_notify() and reload propagation
727 using ReloadPropagationTo= is configured, a reload is now propagated
728 to configured units. (Previously this was only done on explicitly
729 requested reloads, using "systemctl reload" or an equivalent
732 * For each service unit a restart counter is now kept: it is increased
733 each time the service is restarted due to Restart=, and may be
734 queried using "systemctl show -p NRestarts …".
736 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
737 @signal and @timer have been added, for usage with SystemCallFilter=
738 in unit files and the new --system-call-filter= command line option
739 of systemd-nspawn (see above).
741 * ExecStart= lines in unit files gained two new modifiers: when a
742 command line is prefixed with "!" the command will be executed as
743 configured, except for the credentials applied by
744 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
745 "+", but does still apply namespacing options unlike "+". There's
746 also "!!" now, which is mostly identical, but becomes a NOP on
747 systems that support ambient capabilities. This is useful to write
748 unit files that work with ambient capabilities where possible but
749 automatically fall back to traditional privilege dropping mechanisms
750 on systems where this is not supported.
752 * ListenNetlink= settings in socket units now support RDMA netlink
755 * A new unit file setting LockPersonality= has been added which permits
756 locking down the chosen execution domain ("personality") of a service
759 * A new special target "getty-pre.target" has been added, which is
760 ordered before all text logins, and may be used to order services
761 before textual logins acquire access to the console.
763 * systemd will now attempt to load the virtio-rng.ko kernel module very
764 early on if a VM environment supporting this is detected. This should
765 improve entropy during early boot in virtualized environments.
767 * A _netdev option is now supported in /etc/crypttab that operates in a
768 similar way as the same option in /etc/fstab: it permits configuring
769 encrypted devices that need to be ordered after the network is up.
770 Following this logic, two new special targets
771 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
772 added that are to cryptsetup.target what remote-fs.target and
773 remote-fs-pre.target are to local-fs.target.
775 * Service units gained a new UnsetEnvironment= setting which permits
776 unsetting specific environment variables for services that are
777 normally passed to it (for example in order to mask out locale
778 settings for specific services that can't deal with it).
780 * Units acquired a new boolean option IPAccounting=. When turned on, IP
781 traffic accounting (packet count as well as byte count) is done for
782 the service, and shown as part of "systemctl status" or "systemd-run
785 * Service units acquired two new options IPAddressAllow= and
786 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
787 for configuring a simple IP access control list for all sockets of
788 the unit. These options are available also on .slice and .socket
789 units, permitting flexible access list configuration for individual
790 services as well as groups of services (as defined by a slice unit),
791 including system-wide. Note that IP ACLs configured this way are
792 enforced on every single IPv4 and IPv6 socket created by any process
793 of the service unit, and apply to ingress as well as egress traffic.
795 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
796 structured log message is generated each time the unit is stopped,
797 containing information about the consumed resources of this
800 * A new setting KeyringMode= has been added to unit files, which may be
801 used to control how the kernel keyring is set up for executed
804 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
805 "systemctl kexec" and "systemctl exit" are now always asynchronous in
806 behaviour (that is: these commands return immediately after the
807 operation was enqueued instead of waiting for the operation to
808 complete). Previously, "systemctl poweroff" and "systemctl reboot"
809 were asynchronous on systems using systemd-logind (i.e. almost
810 always, and like they were on sysvinit), and the other three commands
811 were unconditionally synchronous. With this release this is cleaned
812 up, and callers will see the same asynchronous behaviour on all
813 systems for all five operations.
815 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
818 * .timer units now accept calendar specifications in other timezones
819 than UTC or the local timezone.
821 * The tmpfiles snippet var.conf has been changed to create
822 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
823 the "utmp" group already, and it appears to be generally understood
824 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
825 databases. Previously this was implemented correctly for all these
826 databases excepts btmp, which has been opened up like this now
827 too. Note that while the other databases are world-readable
828 (i.e. 0644), btmp is not and remains more restrictive.
830 * The systemd-resolve tool gained a new --reset-server-features
831 switch. When invoked like this systemd-resolved will forget
832 everything it learnt about the features supported by the configured
833 upstream DNS servers, and restarts the feature probing logic on the
834 next resolver look-up for them at the highest feature level
837 * The status dump systemd-resolved sends to the logs upon receiving
838 SIGUSR1 now also includes information about all DNS servers it is
839 configured to use, and the features levels it probed for them.
841 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
842 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
843 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
844 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
845 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
846 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
847 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
848 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
849 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
850 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
851 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
852 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
853 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
854 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
855 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
856 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
857 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
858 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
859 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
860 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
866 * Meson is now supported as build system in addition to Automake. It is
867 our plan to remove Automake in one of our next releases, so that
868 Meson becomes our exclusive build system. Hence, please start using
869 the Meson build system in your downstream packaging. There's plenty
870 of documentation around how to use Meson, the extremely brief
873 ./autogen.sh && ./configure && make && sudo make install
877 meson build && ninja -C build && sudo ninja -C build install
879 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
880 which permits configuring a timeout on the time a job is
881 running. This is particularly useful for setting timeouts on jobs for
884 * Unit files gained two new options ConditionUser= and ConditionGroup=
885 for conditionalizing units based on the identity of the user/group
886 running a systemd user instance.
888 * systemd-networkd now understands a new FlowLabel= setting in the
889 [VXLAN] section of .network files, as well as a Priority= in
890 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
891 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
892 gained support for configuration of GENEVE links, and IPv6 address
893 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
895 * .link files now understand a new Port= setting.
897 * systemd-networkd's DHCP support gained support for DHCP option 119
898 (domain search list).
900 * systemd-networkd gained support for serving IPv6 address ranges using
901 the Router Advertisment protocol. The new .network configuration
902 section [IPv6Prefix] may be used to configure the ranges to
903 serve. This is implemented based on a new, minimal, native server
904 implementation of RA.
906 * journalctl's --output= switch gained support for a new parameter
907 "short-iso-precise" for a mode where timestamps are shown as precise
910 * systemd-udevd's "net_id" builtin may now generate stable network
911 interface names from IBM PowerVM VIO devices as well as ACPI platform
914 * MulticastDNS support in systemd-resolved may now be explicitly
915 enabled/disabled using the new MulticastDNS= configuration file
918 * systemd-resolved may now optionally use libidn2 instead of the libidn
919 for processing internationalized domain names. Support for libidn2
920 should be considered experimental and should not be enabled by
923 * "machinectl pull-tar" and related call may now do verification of
924 downloaded images using SUSE-style .sha256 checksum files in addition
925 to the already existing support for validating using Ubuntu-style
928 * sd-bus gained support for a new sd_bus_message_appendv() call which
929 is va_list equivalent of sd_bus_message_append().
931 * sd-boot gained support for validating images using SHIM/MOK.
933 * The SMACK code learnt support for "onlycap".
935 * systemd-mount --umount is now much smarter in figuring out how to
936 properly unmount a device given its mount or device path.
938 * The code to call libnss_dns as a fallback from libnss_resolve when
939 the communication with systemd-resolved fails was removed. This
940 fallback was redundant and interfered with the [!UNAVAIL=return]
941 suffix. See nss-resolve(8) for the recommended configuration.
943 * systemd-logind may now be restarted without losing state. It stores
944 the file descriptors for devices it manages in the system manager
945 using the FDSTORE= mechanism. Please note that further changes in
946 other components may be required to make use of this (for example
947 Xorg has code to listen for stops of systemd-logind and terminate
948 itself when logind is stopped or restarted, in order to avoid using
949 stale file descriptors for graphical devices, which is now
950 counterproductive and must be reverted in order for restarts of
951 systemd-logind to be safe. See
952 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
954 * All kernel install plugins are called with the environment variable
955 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
956 /etc/machine-id. If the file is missing or empty, the variable is
957 empty and BOOT_DIR_ABS is the path of a temporary directory which is
958 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
959 is empty, all plugins should not put anything in BOOT_DIR_ABS.
961 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
962 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
963 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
964 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
965 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
966 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
967 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
968 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
969 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
970 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
971 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
972 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
973 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
974 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
975 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
976 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
977 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
978 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
979 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
980 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
981 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
982 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
983 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
984 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
985 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
986 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
987 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
988 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
989 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
996 * The "hybrid" control group mode has been modified to improve
997 compatibility with "legacy" cgroups-v1 setups. Specifically, the
998 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
999 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1000 cgroups-v1 hierarchy), the only externally visible change being that
1001 the cgroups-v2 hierarchy is also mounted, to
1002 /sys/fs/cgroup/unified. This should provide a large degree of
1003 compatibility with "legacy" cgroups-v1, while taking benefit of the
1004 better management capabilities of cgroups-v2.
1006 * The default control group setup mode may be selected both a boot-time
1007 via a set of kernel command line parameters (specifically:
1008 systemd.unified_cgroup_hierarchy= and
1009 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1010 default selected on the configure command line
1011 (--with-default-hierarchy=). The upstream default is "hybrid"
1012 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1013 this will change in a future systemd version to be "unified" (pure
1014 cgroups-v2 mode). The third option for the compile time option is
1015 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1016 distributions to default to "hybrid" mode for release distributions,
1017 starting with v233. We recommend "unified" for development
1018 distributions (specifically: distributions such as Fedora's rawhide)
1019 as that's where things are headed in the long run. Use "legacy" for
1020 greatest stability and compatibility only.
1022 * Note one current limitation of "unified" and "hybrid" control group
1023 setup modes: the kernel currently does not permit the systemd --user
1024 instance (i.e. unprivileged code) to migrate processes between two
1025 disconnected cgroup subtrees, even if both are managed and owned by
1026 the user. This effectively means "systemd-run --user --scope" doesn't
1027 work when invoked from outside of any "systemd --user" service or
1028 scope. Specifically, it is not supported from session scopes. We are
1029 working on fixing this in a future systemd version. (See #3388 for
1030 further details about this.)
1032 * DBus policy files are now installed into /usr rather than /etc. Make
1033 sure your system has dbus >= 1.9.18 running before upgrading to this
1034 version, or override the install path with --with-dbuspolicydir= .
1036 * All python scripts shipped with systemd (specifically: the various
1037 tests written in Python) now require Python 3.
1039 * systemd unit tests can now run standalone (without the source or
1040 build directories), and can be installed into /usr/lib/systemd/tests/
1041 with 'make install-tests'.
1043 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1044 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1047 * Support for the %c, %r, %R specifiers in unit files has been
1048 removed. Specifiers are not supposed to be dependent on configuration
1049 in the unit file itself (so that they resolve the same regardless
1050 where used in the unit files), but these specifiers were influenced
1051 by the Slice= option.
1053 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1054 all cases. If distributions want to use a different shell for this
1055 purpose (for example Fedora's /sbin/sushell) they need to specify
1056 this explicitly at configure time using --with-debug-shell=.
1058 * The confirmation spawn prompt has been reworked to offer the
1061 (c)ontinue, proceed without asking anymore
1062 (D)ump, show the state of the unit
1063 (f)ail, don't execute the command and pretend it failed
1065 (i)nfo, show a short summary of the unit
1066 (j)obs, show jobs that are in progress
1067 (s)kip, don't execute the command and pretend it succeeded
1068 (y)es, execute the command
1070 The 'n' choice for the confirmation spawn prompt has been removed,
1071 because its meaning was confusing.
1073 The prompt may now also be redirected to an alternative console by
1074 specifying the console as parameter to systemd.confirm_spawn=.
1076 * Services of Type=notify require a READY=1 notification to be sent
1077 during startup. If no such message is sent, the service now fails,
1078 even if the main process exited with a successful exit code.
1080 * Services that fail to start up correctly now always have their
1081 ExecStopPost= commands executed. Previously, they'd enter "failed"
1082 state directly, without executing these commands.
1084 * The option MulticastDNS= of network configuration files has acquired
1085 an actual implementation. With MulticastDNS=yes a host can resolve
1086 names of remote hosts and reply to mDNS A and AAAA requests.
1088 * When units are about to be started an additional check is now done to
1089 ensure that all dependencies of type BindsTo= (when used in
1090 combination with After=) have been started.
1092 * systemd-analyze gained a new verb "syscall-filter" which shows which
1093 system call groups are defined for the SystemCallFilter= unit file
1094 setting, and which system calls they contain.
1096 * A new system call filter group "@filesystem" has been added,
1097 consisting of various file system related system calls. Group
1098 "@reboot" has been added, covering reboot, kexec and shutdown related
1099 calls. Finally, group "@swap" has been added covering swap
1100 configuration related calls.
1102 * A new unit file option RestrictNamespaces= has been added that may be
1103 used to restrict access to the various process namespace types the
1104 Linux kernel provides. Specifically, it may be used to take away the
1105 right for a service unit to create additional file system, network,
1106 user, and other namespaces. This sandboxing option is particularly
1107 relevant due to the high amount of recently discovered namespacing
1108 related vulnerabilities in the kernel.
1110 * systemd-udev's .link files gained support for a new AutoNegotiation=
1111 setting for configuring Ethernet auto-negotiation.
1113 * systemd-networkd's .network files gained support for a new
1114 ListenPort= setting in the [DHCP] section to explicitly configure the
1115 UDP client port the DHCP client shall listen on.
1117 * .network files gained a new Unmanaged= boolean setting for explicitly
1118 excluding one or more interfaces from management by systemd-networkd.
1120 * The systemd-networkd ProxyARP= option has been renamed to
1121 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1122 renamed to ReduceARPProxy=. The old names continue to be available
1125 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1126 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1128 * systemd-networkd's bonding device support gained support for two new
1129 configuration options ActiveSlave= and PrimarySlave=.
1131 * The various options in the [Match] section of .network files gained
1132 support for negative matching.
1134 * New systemd-specific mount options are now understood in /etc/fstab:
1136 x-systemd.mount-timeout= may be used to configure the maximum
1137 permitted runtime of the mount command.
1139 x-systemd.device-bound may be set to bind a mount point to its
1140 backing device unit, in order to automatically remove a mount point
1141 if its backing device is unplugged. This option may also be
1142 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1143 on the block device, which is now automatically set for all CDROM
1144 drives, so that mounted CDs are automatically unmounted when they are
1145 removed from the drive.
1147 x-systemd.after= and x-systemd.before= may be used to explicitly
1148 order a mount after or before another unit or mount point.
1150 * Enqueued start jobs for device units are now automatically garbage
1151 collected if there are no jobs waiting for them anymore.
1153 * systemctl list-jobs gained two new switches: with --after, for every
1154 queued job the jobs it's waiting for are shown; with --before the
1155 jobs which it's blocking are shown.
1157 * systemd-nspawn gained support for ephemeral boots from disk images
1158 (or in other words: --ephemeral and --image= may now be
1159 combined). Moreover, ephemeral boots are now supported for normal
1160 directories, even if the backing file system is not btrfs. Of course,
1161 if the file system does not support file system snapshots or
1162 reflinks, the initial copy operation will be relatively expensive, but
1163 this should still be suitable for many use cases.
1165 * Calendar time specifications in .timer units now support
1166 specifications relative to the end of a month by using "~" instead of
1167 "-" as separator between month and day. For example, "*-02~03" means
1168 "the third last day in February". In addition a new syntax for
1169 repeated events has been added using the "/" character. For example,
1170 "9..17/2:00" means "every two hours from 9am to 5pm".
1172 * systemd-socket-proxyd gained a new parameter --connections-max= for
1173 configuring the maximum number of concurrent connections.
1175 * sd-id128 gained a new API for generating unique IDs for the host in a
1176 way that does not leak the machine ID. Specifically,
1177 sd_id128_get_machine_app_specific() derives an ID based on the
1178 machine ID a in well-defined, non-reversible, stable way. This is
1179 useful whenever an identifier for the host is needed but where the
1180 identifier shall not be useful to identify the system beyond the
1181 scope of the application itself. (Internally this uses HMAC-SHA256 as
1182 keyed hash function using the machine ID as input.)
1184 * NotifyAccess= gained a new supported value "exec". When set
1185 notifications are accepted from all processes systemd itself invoked,
1186 including all control processes.
1188 * .nspawn files gained support for defining overlay mounts using the
1189 Overlay= and OverlayReadOnly= options. Previously this functionality
1190 was only available on the systemd-nspawn command line.
1192 * systemd-nspawn's --bind= and --overlay= options gained support for
1193 bind/overlay mounts whose source lies within the container tree by
1194 prefixing the source path with "+".
1196 * systemd-nspawn's --bind= and --overlay= options gained support for
1197 automatically allocating a temporary source directory in /var/tmp
1198 that is removed when the container dies. Specifically, if the source
1199 directory is specified as empty string this mechanism is selected. An
1200 example usage is --overlay=+/var::/var, which creates an overlay
1201 mount based on the original /var contained in the image, overlayed
1202 with a temporary directory in the host's /var/tmp. This way changes
1203 to /var are automatically flushed when the container shuts down.
1205 * systemd-nspawn --image= option does now permit raw file system block
1206 devices (in addition to images containing partition tables, as
1209 * The disk image dissection logic in systemd-nspawn gained support for
1210 automatically setting up LUKS encrypted as well as Verity protected
1211 partitions. When a container is booted from an encrypted image the
1212 passphrase is queried at start-up time. When a container with Verity
1213 data is started, the root hash is search in a ".roothash" file
1214 accompanying the disk image (alternatively, pass the root hash via
1215 the new --root-hash= command line option).
1217 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1218 be used to dissect disk images the same way as systemd-nspawn does
1219 it, following the Bootable Partition Specification. It may even be
1220 used to mount disk images with complex partition setups (including
1221 LUKS and Verity partitions) to a local host directory, in order to
1222 inspect them. This tool is not considered public API (yet), and is
1223 thus not installed into /usr/bin. Please do not rely on its
1224 existence, since it might go away or be changed in later systemd
1227 * A new generator "systemd-verity-generator" has been added, similar in
1228 style to "systemd-cryptsetup-generator", permitting automatic setup of
1229 Verity root partitions when systemd boots up. In order to make use of
1230 this your partition setup should follow the Discoverable Partitions
1231 Specification, and the GPT partition ID of the root file system
1232 partition should be identical to the upper 128bit of the Verity root
1233 hash. The GPT partition ID of the Verity partition protecting it
1234 should be the lower 128bit of the Verity root hash. If the partition
1235 image follows this model it is sufficient to specify a single
1236 "roothash=" kernel command line argument to both configure which root
1237 image and verity partition to use as well as the root hash for
1238 it. Note that systemd-nspawn's Verity support follows the same
1239 semantics, meaning that disk images with proper Verity data in place
1240 may be booted in containers with systemd-nspawn as well as on
1241 physical systems via the verity generator. Also note that the "mkosi"
1242 tool available at https://github.com/systemd/mkosi has been updated
1243 to generate Verity protected disk images following this scheme. In
1244 fact, it has been updated to generate disk images that optionally
1245 implement a complete UEFI SecureBoot trust chain, involving a signed
1246 kernel and initrd image that incorporates such a root hash as well as
1247 a Verity-enabled root partition.
1249 * The hardware database (hwdb) udev supports has been updated to carry
1250 accelerometer quirks.
1252 * All system services are now run with a fresh kernel keyring set up
1253 for them. The invocation ID is stored by default in it, thus
1254 providing a safe, non-overridable way to determine the invocation
1257 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1258 options for bind mounting arbitrary paths in a service-specific
1259 way. When these options are used, arbitrary host or service files and
1260 directories may be mounted to arbitrary locations in the service's
1263 * Documentation has been added that lists all of systemd's low-level
1264 environment variables:
1266 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1268 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1269 whether a specific socket file descriptor matches a specified socket
1272 * systemd-firstboot has been updated to check for the
1273 systemd.firstboot= kernel command line option. It accepts a boolean
1274 and when set to false the first boot questions are skipped.
1276 * systemd-fstab-generator has been updated to check for the
1277 systemd.volatile= kernel command line option, which either takes an
1278 optional boolean parameter or the special value "state". If used the
1279 system may be booted in a "volatile" boot mode. Specifically,
1280 "systemd.volatile" is used, the root directory will be mounted as
1281 tmpfs, and only /usr is mounted from the actual root file system. If
1282 "systemd.volatile=state" is used, the root directory will be mounted
1283 as usual, but /var is mounted as tmpfs. This concept provides similar
1284 functionality as systemd-nspawn's --volatile= option, but provides it
1285 on physical boots. Use this option for implementing stateless
1286 systems, or testing systems with all state and/or configuration reset
1287 to the defaults. (Note though that many distributions are not
1288 prepared to boot up without a populated /etc or /var, though.)
1290 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1291 partitions. Previously it only supported LUKS encrypted partitions
1292 for all other uses, except for the root partition itself.
1294 * Socket units gained support for listening on AF_VSOCK sockets for
1295 communication in virtualized QEMU environments.
1297 * The "configure" script gained a new option --with-fallback-hostname=
1298 for specifying the fallback hostname to use if none is configured in
1299 /etc/hostname. For example, by specifying
1300 --with-fallback-hostname=fedora it is possible to default to a
1301 hostname of "fedora" on pristine installations.
1303 * systemd-cgls gained support for a new --unit= switch for listing only
1304 the control groups of a specific unit. Similar --user-unit= has been
1305 added for listing only the control groups of a specific user unit.
1307 * systemd-mount gained a new --umount switch for unmounting a mount or
1308 automount point (and all mount/automount points below it).
1310 * systemd will now refuse full configuration reloads (via systemctl
1311 daemon-reload and related calls) unless at least 16MiB of free space
1312 are available in /run. This is a safety precaution in order to ensure
1313 that generators can safely operate after the reload completed.
1315 * A new unit file option RootImage= has been added, which has a similar
1316 effect as RootDirectory= but mounts the service's root directory from
1317 a disk image instead of plain directory. This logic reuses the same
1318 image dissection and mount logic that systemd-nspawn already uses,
1319 and hence supports any disk images systemd-nspawn supports, including
1320 those following the Discoverable Partition Specification, as well as
1321 Verity enabled images. This option enables systemd to run system
1322 services directly off disk images acting as resource bundles,
1323 possibly even including full integrity data.
1325 * A new MountAPIVFS= unit file option has been added, taking a boolean
1326 argument. If enabled /proc, /sys and /dev (collectively called the
1327 "API VFS") will be mounted for the service. This is only relevant if
1328 RootDirectory= or RootImage= is used for the service, as these mounts
1329 are of course in place in the host mount namespace anyway.
1331 * systemd-nspawn gained support for a new --pivot-root= switch. If
1332 specified the root directory within the container image is pivoted to
1333 the specified mount point, while the original root disk is moved to a
1334 different place. This option enables booting of ostree images
1335 directly with systemd-nspawn.
1337 * The systemd build scripts will no longer complain if the NTP server
1338 addresses are not changed from the defaults. Google now supports
1339 these NTP servers officially. We still recommend downstreams to
1340 properly register an NTP pool with the NTP pool project though.
1342 * coredumpctl gained a new "--reverse" option for printing the list
1343 of coredumps in reverse order.
1345 * coredumpctl will now show additional information about truncated and
1346 inaccessible coredumps, as well as coredumps that are still being
1347 processed. It also gained a new --quiet switch for suppressing
1348 additional informational message in its output.
1350 * coredumpctl gained support for only showing coredumps newer and/or
1351 older than specific timestamps, using the new --since= and --until=
1352 options, reminiscent of journalctl's options by the same name.
1354 * The systemd-coredump logic has been improved so that it may be reused
1355 to collect backtraces in non-compiled languages, for example in
1356 scripting languages such as Python.
1358 * machinectl will now show the UID shift of local containers, if user
1359 namespacing is enabled for them.
1361 * systemd will now optionally run "environment generator" binaries at
1362 configuration load time. They may be used to add environment
1363 variables to the environment block passed to services invoked. One
1364 user environment generator is shipped by default that sets up
1365 environment variables based on files dropped into /etc/environment.d
1366 and ~/.config/environment.d/.
1368 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1371 * hostnamed has been updated to report a new chassis type of
1372 "convertible" to cover "foldable" laptops that can both act as a
1373 tablet and as a laptop, such as various Lenovo Yoga devices.
1375 Contributions from: Adrián López, Alexander Galanin, Alexander
1376 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1377 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1378 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1379 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1380 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1381 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1382 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1383 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1384 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1385 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1386 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1387 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1388 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1389 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1390 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1391 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1392 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1393 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1394 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1395 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1396 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1397 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1398 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1399 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1400 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1401 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1402 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1405 — Berlin, 2017-03-01
1409 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1410 RestrictAddressFamilies= enabled. These sandboxing options should
1411 generally be compatible with the various external udev call-out
1412 binaries we are aware of, however there may be exceptions, in
1413 particular when exotic languages for these call-outs are used. In
1414 this case, consider turning off these settings locally.
1416 * The new RemoveIPC= option can be used to remove IPC objects owned by
1417 the user or group of a service when that service exits.
1419 * The new ProtectKernelModules= option can be used to disable explicit
1420 load and unload operations of kernel modules by a service. In
1421 addition access to /usr/lib/modules is removed if this option is set.
1423 * ProtectSystem= option gained a new value "strict", which causes the
1424 whole file system tree with the exception of /dev, /proc, and /sys,
1425 to be remounted read-only for a service.
1427 * The new ProtectKernelTunables= option can be used to disable
1428 modification of configuration files in /sys and /proc by a service.
1429 Various directories and files are remounted read-only, so access is
1430 restricted even if the file permissions would allow it.
1432 * The new ProtectControlGroups= option can be used to disable write
1433 access by a service to /sys/fs/cgroup.
1435 * Various systemd services have been hardened with
1436 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1437 RestrictAddressFamilies=.
1439 * Support for dynamically creating users for the lifetime of a service
1440 has been added. If DynamicUser=yes is specified, user and group IDs
1441 will be allocated from the range 61184..65519 for the lifetime of the
1442 service. They can be resolved using the new nss-systemd.so NSS
1443 module. The module must be enabled in /etc/nsswitch.conf. Services
1444 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1445 any resources allocated by the service will be cleaned up when the
1446 service exits. They also have ProtectHome=read-only and
1447 ProtectSystem=strict enabled, so they are not able to make any
1448 permanent modifications to the system.
1450 * The nss-systemd module also always resolves root and nobody, making
1451 it possible to have no /etc/passwd or /etc/group files in minimal
1452 container or chroot environments.
1454 * Services may be started with their own user namespace using the new
1455 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1456 under which the service is running are mapped. All other users are
1459 * Support for the cgroup namespace has been added to systemd-nspawn. If
1460 supported by kernel, the container system started by systemd-nspawn
1461 will have its own view of the cgroup hierarchy. This new behaviour
1462 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1464 * The new MemorySwapMax= option can be used to limit the maximum swap
1465 usage under the unified cgroup hierarchy.
1467 * Support for the CPU controller in the unified cgroup hierarchy has
1468 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1469 options. This controller requires out-of-tree patches for the kernel
1470 and the support is provisional.
1472 * Mount and automount units may now be created transiently
1473 (i.e. dynamically at runtime via the bus API, instead of requiring
1474 unit files in the file system).
1476 * systemd-mount is a new tool which may mount file systems – much like
1477 mount(8), optionally pulling in additional dependencies through
1478 transient .mount and .automount units. For example, this tool
1479 automatically runs fsck on a backing block device before mounting,
1480 and allows the automount logic to be used dynamically from the
1481 command line for establishing mount points. This tool is particularly
1482 useful when dealing with removable media, as it will ensure fsck is
1483 run – if necessary – before the first access and that the file system
1484 is quickly unmounted after each access by utilizing the automount
1485 logic. This maximizes the chance that the file system on the
1486 removable media stays in a clean state, and if it isn't in a clean
1487 state is fixed automatically.
1489 * LazyUnmount=yes option for mount units has been added to expose the
1490 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1493 * /efi will be used as the mount point of the EFI boot partition, if
1494 the directory is present, and the mount point was not configured
1495 through other means (e.g. fstab). If /efi directory does not exist,
1496 /boot will be used as before. This makes it easier to automatically
1497 mount the EFI partition on systems where /boot is used for something
1500 * When operating on GPT disk images for containers, systemd-nspawn will
1501 now mount the ESP to /boot or /efi according to the same rules as PID
1502 1 running on a host. This allows tools like "bootctl" to operate
1503 correctly within such containers, in order to make container images
1504 bootable on physical systems.
1506 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1508 * Two new user session targets have been added to support running
1509 graphical sessions under the systemd --user instance:
1510 graphical-session.target and graphical-session-pre.target. See
1511 systemd.special(7) for a description of how those targets should be
1514 * The vconsole initialization code has been significantly reworked to
1515 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1516 support unicode keymaps. Font and keymap configuration will now be
1517 copied to all allocated virtual consoles.
1519 * FreeBSD's bhyve virtualization is now detected.
1521 * Information recorded in the journal for core dumps now includes the
1522 contents of /proc/mountinfo and the command line of the process at
1523 the top of the process hierarchy (which is usually the init process
1526 * systemd-journal-gatewayd learned the --directory= option to serve
1527 files from the specified location.
1529 * journalctl --root=… can be used to peruse the journal in the
1530 /var/log/ directories inside of a container tree. This is similar to
1531 the existing --machine= option, but does not require the container to
1534 * The hardware database has been extended to support
1535 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1538 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1539 specify the click rate for mice which include a horizontal wheel with
1540 a click rate that is different than the one for the vertical wheel.
1542 * systemd-run gained a new --wait option that makes service execution
1543 synchronous. (Specifically, the command will not return until the
1544 specified service binary exited.)
1546 * systemctl gained a new --wait option that causes the start command to
1547 wait until the units being started have terminated again.
1549 * A new journal output mode "short-full" has been added which displays
1550 timestamps with abbreviated English day names and adds a timezone
1551 suffix. Those timestamps include more information than the default
1552 "short" output mode, and can be passed directly to journalctl's
1553 --since= and --until= options.
1555 * /etc/resolv.conf will be bind-mounted into containers started by
1556 systemd-nspawn, if possible, so any changes to resolv.conf contents
1557 are automatically propagated to the container.
1559 * The number of instances for socket-activated services originating
1560 from a single IP address can be limited with
1561 MaxConnectionsPerSource=, extending the existing setting of
1564 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1567 * .netdev and .network configuration can now be extended through
1570 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1571 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1572 can be enabled and disabled using the new UDPSegmentationOffload=,
1573 TCPSegmentationOffload=, GenericSegmentationOffload=,
1574 GenericReceiveOffload=, LargeReceiveOffload= options in the
1575 [Link] section of .link files.
1577 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1578 Port VLAN ID can be configured for bridge devices using the new STP=,
1579 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1580 section of .netdev files.
1582 * The route table to which routes received over DHCP or RA should be
1583 added can be configured with the new RouteTable= option in the [DHCP]
1584 and [IPv6AcceptRA] sections of .network files.
1586 * The Address Resolution Protocol can be disabled on links managed by
1587 systemd-networkd using the ARP=no setting in the [Link] section of
1590 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1591 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1592 encode information about the result and exit codes of the current
1593 service runtime cycle.
1595 * systemd-sysctl will now configure kernel parameters in the order
1596 they occur in the configuration files. This matches what sysctl
1597 has been traditionally doing.
1599 * kernel-install "plugins" that are executed to perform various
1600 tasks after a new kernel is added and before an old one is removed
1601 can now return a special value to terminate the procedure and
1602 prevent any later plugins from running.
1604 * Journald's SplitMode=login setting has been deprecated. It has been
1605 removed from documentation, and its use is discouraged. In a future
1606 release it will be completely removed, and made equivalent to current
1607 default of SplitMode=uid.
1609 * Storage=both option setting in /etc/systemd/coredump.conf has been
1610 removed. With fast LZ4 compression storing the core dump twice is not
1613 * The --share-system systemd-nspawn option has been replaced with an
1614 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1615 this functionality is discouraged. In addition the variables
1616 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1617 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1618 individual namespaces.
1620 * "machinectl list" now shows the IP address of running containers in
1621 the output, as well as OS release information.
1623 * "loginctl list" now shows the TTY of each session in the output.
1625 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1626 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1627 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1628 tracking objects in a "recursive" mode, where a single client can be
1629 counted multiple times, if it takes multiple references.
1631 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1632 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1633 process using sd-bus automatically exit if the bus connection is
1636 * Bus clients of the service manager may now "pin" loaded units into
1637 memory, by taking an explicit reference on them. This is useful to
1638 ensure the client can retrieve runtime data about the service even
1639 after the service completed execution. Taking such a reference is
1640 available only for privileged clients and should be helpful to watch
1641 running services in a race-free manner, and in particular collect
1642 information about exit statuses and results.
1644 * The nss-resolve module has been changed to strictly return UNAVAIL
1645 when communication via D-Bus with resolved failed, and NOTFOUND when
1646 a lookup completed but was negative. This means it is now possible to
1647 neatly configure fallbacks using nsswitch.conf result checking
1648 expressions. Taking benefit of this, the new recommended
1649 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1651 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1653 * A new setting CtrlAltDelBurstAction= has been added to
1654 /etc/systemd/system.conf which may be used to configure the precise
1655 behaviour if the user on the console presses Ctrl-Alt-Del more often
1656 than 7 times in 2s. Previously this would unconditionally result in
1657 an expedited, immediate reboot. With this new setting the precise
1658 operation may be configured in more detail, and also turned off
1661 * In .netdev files two new settings RemoteChecksumTx= and
1662 RemoteChecksumRx= are now understood that permit configuring the
1663 remote checksumming logic for VXLAN networks.
1665 * The service manager learnt a new "invocation ID" concept for invoked
1666 services. Each runtime cycle of a service will get a new invocation
1667 ID (a 128bit random UUID) assigned that identifies the current
1668 run of the service uniquely and globally. A new invocation ID
1669 is generated each time a service starts up. The journal will store
1670 the invocation ID of a service along with any logged messages, thus
1671 making the invocation ID useful for matching the online runtime of a
1672 service with the offline log data it generated in a safe way without
1673 relying on synchronized timestamps. In many ways this new service
1674 invocation ID concept is similar to the kernel's boot ID concept that
1675 uniquely and globally identifies the runtime of each boot. The
1676 invocation ID of a service is passed to the service itself via an
1677 environment variable ($INVOCATION_ID). A new bus call
1678 GetUnitByInvocationID() has been added that is similar to GetUnit()
1679 but instead of retrieving the bus path for a unit by its name
1680 retrieves it by its invocation ID. The returned path is valid only as
1681 long as the passed invocation ID is current.
1683 * systemd-resolved gained a new "DNSStubListener" setting in
1684 resolved.conf. It either takes a boolean value or the special values
1685 "udp" and "tcp", and configures whether to enable the stub DNS
1686 listener on 127.0.0.53:53.
1688 * IP addresses configured via networkd may now carry additional
1689 configuration settings supported by the kernel. New options include:
1690 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1691 PrefixRoute=, AutoJoin=.
1693 * The PAM configuration fragment file for "user@.service" shipped with
1694 systemd (i.e. the --user instance of systemd) has been stripped to
1695 the minimum necessary to make the system boot. Previously, it
1696 contained Fedora-specific stanzas that did not apply to other
1697 distributions. It is expected that downstream distributions add
1698 additional configuration lines, matching their needs to this file,
1699 using it only as rough template of what systemd itself needs. Note
1700 that this reduced fragment does not even include an invocation of
1701 pam_limits which most distributions probably want to add, even though
1702 systemd itself does not need it. (There's also the new build time
1703 option --with-pamconfdir=no to disable installation of the PAM
1706 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1707 capability is now also dropped from its set (in addition to
1708 CAP_SYS_MKNOD as before).
1710 * In service unit files it is now possible to connect a specific named
1711 file descriptor with stdin/stdout/stdout of an executed service. The
1712 name may be specified in matching .socket units using the
1713 FileDescriptorName= setting.
1715 * A number of journal settings may now be configured on the kernel
1716 command line. Specifically, the following options are now understood:
1717 systemd.journald.max_level_console=,
1718 systemd.journald.max_level_store=,
1719 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1720 systemd.journald.max_level_wall=.
1722 * "systemctl is-enabled --full" will now show by which symlinks a unit
1723 file is enabled in the unit dependency tree.
1725 * Support for VeraCrypt encrypted partitions has been added to the
1726 "cryptsetup" logic and /etc/crypttab.
1728 * systemd-detect-virt gained support for a new --private-users switch
1729 that checks whether the invoking processes are running inside a user
1730 namespace. Similar, a new special value "private-users" for the
1731 existing ConditionVirtualization= setting has been added, permitting
1732 skipping of specific units in user namespace environments.
1734 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1735 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1736 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1737 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1738 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1739 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1740 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1741 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1742 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1743 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1744 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1745 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1746 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1747 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1748 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1749 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1750 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1751 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1752 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1753 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1754 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1755 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1756 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1757 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1758 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1759 Jędrzejewski-Szmek, Zeal Jagannatha
1761 — Santa Fe, 2016-11-03
1765 * In service units the various ExecXYZ= settings have been extended
1766 with an additional special character as first argument of the
1767 assigned value: if the character '+' is used the specified command
1768 line it will be run with full privileges, regardless of User=,
1769 Group=, CapabilityBoundingSet= and similar options. The effect is
1770 similar to the existing PermissionsStartOnly= option, but allows
1771 configuration of this concept for each executed command line
1774 * Services may now alter the service watchdog timeout at runtime by
1775 sending a WATCHDOG_USEC= message via sd_notify().
1777 * MemoryLimit= and related unit settings now optionally take percentage
1778 specifications. The percentage is taken relative to the amount of
1779 physical memory in the system (or in case of containers, the assigned
1780 amount of memory). This allows scaling service resources neatly with
1781 the amount of RAM available on the system. Similarly, systemd-logind's
1782 RuntimeDirectorySize= option now also optionally takes percentage
1785 * In similar fashion TasksMax= takes percentage values now, too. The
1786 value is taken relative to the configured maximum number of processes
1787 on the system. The per-service task maximum has been changed to 15%
1788 using this functionality. (Effectively this is an increase of 512 →
1789 4915 for service units, given the kernel's default pid_max setting.)
1791 * Calendar time specifications in .timer units now understand a ".."
1792 syntax for time ranges. Example: "4..7:10" may now be used for
1793 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1796 * The InaccessableDirectories=, ReadOnlyDirectories= and
1797 ReadWriteDirectories= unit file settings have been renamed to
1798 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1799 applied to all kinds of file nodes, and not just directories, with
1800 the exception of symlinks. Specifically these settings may now be
1801 used on block and character device nodes, UNIX sockets and FIFOS as
1802 well as regular files. The old names of these settings remain
1803 available for compatibility.
1805 * systemd will now log about all service processes it kills forcibly
1806 (using SIGKILL) because they remained after the clean shutdown phase
1807 of the service completed. This should help identifying services that
1808 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1809 systemd-logind's configuration a similar log message is generated for
1810 processes killed at the end of each session due to this setting.
1812 * systemd will now set the $JOURNAL_STREAM environment variable for all
1813 services whose stdout/stderr are connected to the Journal (which
1814 effectively means by default: all services). The variable contains
1815 the device and inode number of the file descriptor used for
1816 stdout/stderr. This may be used by invoked programs to detect whether
1817 their stdout/stderr is connected to the Journal, in which case they
1818 can switch over to direct Journal communication, thus being able to
1819 pass extended, structured metadata along with their log messages. As
1820 one example, this is now used by glib's logging primitives.
1822 * When using systemd's default tmp.mount unit for /tmp, the mount point
1823 will now be established with the "nosuid" and "nodev" options. This
1824 avoids privilege escalation attacks that put traps and exploits into
1825 /tmp. However, this might cause problems if you e. g. put container
1826 images or overlays into /tmp; if you need this, override tmp.mount's
1827 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1830 * systemd now supports the "memory" cgroup controller also on
1833 * The systemd-cgtop tool now optionally takes a control group path as
1834 command line argument. If specified, the control group list shown is
1835 limited to subgroups of that group.
1837 * The SystemCallFilter= unit file setting gained support for
1838 pre-defined, named system call filter sets. For example
1839 SystemCallFilter=@clock is now an effective way to make all clock
1840 changing-related system calls unavailable to a service. A number of
1841 similar pre-defined groups are defined. Writing system call filters
1842 for system services is simplified substantially with this new
1843 concept. Accordingly, all of systemd's own, long-running services now
1844 enable system call filtering based on this, by default.
1846 * A new service setting MemoryDenyWriteExecute= has been added, taking
1847 a boolean value. If turned on, a service may no longer create memory
1848 mappings that are writable and executable at the same time. This
1849 enhances security for services where this is enabled as it becomes
1850 harder to dynamically write and then execute memory in exploited
1851 service processes. This option has been enabled for all of systemd's
1852 own long-running services.
1854 * A new RestrictRealtime= service setting has been added, taking a
1855 boolean argument. If set the service's processes may no longer
1856 acquire realtime scheduling. This improves security as realtime
1857 scheduling may otherwise be used to easily freeze the system.
1859 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1860 value. This may be used for requesting that the system manager inside
1861 of the container reports start-up completion to nspawn which then
1862 propagates this notification further to the service manager
1863 supervising nspawn itself. A related option NotifyReady= in .nspawn
1864 files has been added too. This functionality allows ordering of the
1865 start-up of multiple containers using the usual systemd ordering
1868 * machinectl gained a new command "stop" that is an alias for
1871 * systemd-resolved gained support for contacting DNS servers on
1872 link-local IPv6 addresses.
1874 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1875 its caches. A method call for requesting the same operation has been
1876 added to the bus API too, and is made available via "systemd-resolve
1879 * systemd-resolve gained a new --status switch. If passed a brief
1880 summary of the used DNS configuration with per-interface information
1883 * resolved.conf gained a new Cache= boolean option, defaulting to
1884 on. If turned off local DNS caching is disabled. This comes with a
1885 performance penalty in particular when DNSSEC is enabled. Note that
1886 resolved disables its internal caching implicitly anyway, when the
1887 configured DNS server is on a host-local IP address such as ::1 or
1888 127.0.0.1, thus automatically avoiding double local caching.
1890 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1891 for DNS requests. This improves compatibility with local programs
1892 that do not use the libc NSS or systemd-resolved's bus APIs for name
1893 resolution. This minimal DNS service is only available to local
1894 programs and does not implement the full DNS protocol, but enough to
1895 cover local DNS clients. A new, static resolv.conf file, listing just
1896 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1897 now recommended to make /etc/resolv.conf a symlink to this file in
1898 order to route all DNS lookups to systemd-resolved, regardless if
1899 done via NSS, the bus API or raw DNS packets. Note that this local
1900 DNS service is not as fully featured as the libc NSS or
1901 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1902 used to deliver link-local address information (as this implies
1903 sending a local interface index along), LLMNR/mDNS support via this
1904 interface is severely restricted. It is thus strongly recommended for
1905 all applications to use the libc NSS API or native systemd-resolved
1908 * systemd-networkd's bridge support learned a new setting
1909 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1910 in .network files has been added for configuring VLAN bridging in
1911 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1913 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1914 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1915 now be acquired without relying on DHCPv6. Two new options
1916 UseDomains= and UseDNS= have been added to configure this behaviour.
1918 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1919 renamed IPv6AcceptRA=, without altering its behaviour. The old
1920 setting name remains available for compatibility reasons.
1922 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1923 Key=, InputKey= and OutputKey=.
1925 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1926 interface configuration.
1928 * "systemctl edit" may now be used to create new unit files by
1929 specifying the --force switch.
1931 * sd-event gained a new function sd_event_get_iteration() for
1932 requesting the current iteration counter of the event loop. It starts
1933 at zero and is increased by one with each event loop iteration.
1935 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1936 file. It can be used in lieu of %systemd_requires in packages which
1937 don't use any systemd functionality and are intended to be installed
1938 in minimal containers without systemd present. This macro provides
1939 ordering dependencies to ensure that if the package is installed in
1940 the same rpm transaction as systemd, systemd will be installed before
1941 the scriptlets for the package are executed, allowing unit presets
1944 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1945 been added to simplify packaging of generators.
1947 * The os-release file gained VERSION_CODENAME field for the
1948 distribution nickname (e.g. VERSION_CODENAME=woody).
1950 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1951 can be set to disable parsing of metadata and the creation
1952 of persistent symlinks for that device.
1954 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1955 to make them available to logged-in users has been reverted.
1957 * Much of the common code of the various systemd components is now
1958 built into an internal shared library libsystemd-shared-231.so
1959 (incorporating the systemd version number in the name, to be updated
1960 with future releases) that the components link to. This should
1961 decrease systemd footprint both in memory during runtime and on
1962 disk. Note that the shared library is not for public use, and is
1963 neither API not ABI stable, but is likely to change with every new
1964 released update. Packagers need to make sure that binaries
1965 linking to libsystemd-shared.so are updated in step with the
1968 * Configuration for "mkosi" is now part of the systemd
1969 repository. mkosi is a tool to easily build legacy-free OS images,
1970 and is available on github: https://github.com/systemd/mkosi. If
1971 "mkosi" is invoked in the build tree a new raw OS image is generated
1972 incorporating the systemd sources currently being worked on and a
1973 clean, fresh distribution installation. The generated OS image may be
1974 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1975 UEFI PC. This functionality is particularly useful to easily test
1976 local changes made to systemd in a pristine, defined environment. See
1977 doc/HACKING for details.
1979 * configure learned the --with-support-url= option to specify the
1980 distribution's bugtracker.
1982 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1983 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1984 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1985 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1986 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1987 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1988 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1989 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1990 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1991 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1992 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1993 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1994 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1995 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1996 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1997 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1998 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1999 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2000 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2002 — Berlin, 2016-07-25
2006 * DNSSEC is now turned on by default in systemd-resolved (in
2007 "allow-downgrade" mode), but may be turned off during compile time by
2008 passing "--with-default-dnssec=no" to "configure" (and of course,
2009 during runtime with DNSSEC= in resolved.conf). We recommend
2010 downstreams to leave this on at least during development cycles and
2011 report any issues with the DNSSEC logic upstream. We are very
2012 interested in collecting feedback about the DNSSEC validator and its
2013 limitations in the wild. Note however, that DNSSEC support is
2014 probably nothing downstreams should turn on in stable distros just
2015 yet, as it might create incompatibilities with a few DNS servers and
2016 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2017 automatically whenever we detect such incompatible setups, but there
2018 might be systems we do not cover yet. Hence: please help us testing
2019 the DNSSEC code, leave this on where you can, report back, but then
2020 again don't consider turning this on in your stable, LTS or
2021 production release just yet. (Note that you have to enable
2022 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2023 and its DNSSEC mode for host name resolution from local
2026 * systemd-resolve conveniently resolves DANE records with the --tlsa
2027 option and OPENPGPKEY records with the --openpgp option. It also
2028 supports dumping raw DNS record data via the new --raw= switch.
2030 * systemd-logind will now by default terminate user processes that are
2031 part of the user session scope unit (session-XX.scope) when the user
2032 logs out. This behavior is controlled by the KillUserProcesses=
2033 setting in logind.conf, and the previous default of "no" is now
2034 changed to "yes". This means that user sessions will be properly
2035 cleaned up after, but additional steps are necessary to allow
2036 intentionally long-running processes to survive logout.
2038 While the user is logged in at least once, user@.service is running,
2039 and any service that should survive the end of any individual login
2040 session can be started at a user service or scope using systemd-run.
2041 systemd-run(1) man page has been extended with an example which shows
2042 how to run screen in a scope unit underneath user@.service. The same
2043 command works for tmux.
2045 After the user logs out of all sessions, user@.service will be
2046 terminated too, by default, unless the user has "lingering" enabled.
2047 To effectively allow users to run long-term tasks even if they are
2048 logged out, lingering must be enabled for them. See loginctl(1) for
2049 details. The default polkit policy was modified to allow users to
2050 set lingering for themselves without authentication.
2052 Previous defaults can be restored at compile time by the
2053 --without-kill-user-processes option to "configure".
2055 * systemd-logind gained new configuration settings SessionsMax= and
2056 InhibitorsMax=, both with a default of 8192. It will not register new
2057 user sessions or inhibitors above this limit.
2059 * systemd-logind will now reload configuration on SIGHUP.
2061 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2062 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2063 enable. Also, support for the "io" cgroup controller in the unified
2064 hierarchy has been added, so that the "memory", "pids" and "io" are
2065 now the controllers that are supported on the unified hierarchy.
2067 WARNING: it is not possible to use previous systemd versions with
2068 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2069 is necessary to also update systemd in the initramfs if using the
2070 unified hierarchy. An updated SELinux policy is also required.
2072 * LLDP support has been extended, and both passive (receive-only) and
2073 active (sender) modes are supported. Passive mode ("routers-only") is
2074 enabled by default in systemd-networkd. Active LLDP mode is enabled
2075 by default for containers on the internal network. The "networkctl
2076 lldp" command may be used to list information gathered. "networkctl
2077 status" will also show basic LLDP information on connected peers now.
2079 * The IAID and DUID unique identifier sent in DHCP requests may now be
2080 configured for the system and each .network file managed by
2081 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2083 * systemd-networkd gained support for configuring proxy ARP support for
2084 each interface, via the ProxyArp= setting in .network files. It also
2085 gained support for configuring the multicast querier feature of
2086 bridge devices, via the new MulticastQuerier= setting in .netdev
2087 files. Similarly, snooping on the IGMP traffic can be controlled
2088 via the new setting MulticastSnooping=.
2090 A new setting PreferredLifetime= has been added for addresses
2091 configured in .network file to configure the lifetime intended for an
2094 The systemd-networkd DHCP server gained the option EmitRouter=, which
2095 defaults to yes, to configure whether the DHCP Option 3 (Router)
2098 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2099 systemd-socket-activate and installed into /usr/bin. It is now fully
2102 * systemd-journald now uses separate threads to flush changes to disk
2103 when closing journal files, thus reducing impact of slow disk I/O on
2104 logging performance.
2106 * The sd-journal API gained two new calls
2107 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2108 can be used to open journal files using file descriptors instead of
2109 file or directory paths. sd_journal_open_container() has been
2110 deprecated, sd_journal_open_directory_fd() should be used instead
2111 with the flag SD_JOURNAL_OS_ROOT.
2113 * journalctl learned a new output mode "-o short-unix" that outputs log
2114 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2115 UTC). It also gained support for a new --no-hostname setting to
2116 suppress the hostname column in the family of "short" output modes.
2118 * systemd-ask-password now optionally skips printing of the password to
2119 stdout with --no-output which can be useful in scripts.
2121 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2122 (devices tagged with ID_MAKER_TOOL) are now tagged with
2123 "uaccess" and are available to logged in users.
2125 * The DeviceAllow= unit setting now supports specifiers (with "%").
2127 * "systemctl show" gained a new --value switch, which allows print a
2128 only the contents of a specific unit property, without also printing
2129 the property's name. Similar support was added to "show*" verbs
2130 of loginctl and machinectl that output "key=value" lists.
2132 * A new unit type "generated" was added for files dynamically generated
2133 by generator tools. Similarly, a new unit type "transient" is used
2134 for unit files created using the runtime API. "systemctl enable" will
2135 refuse to operate on such files.
2137 * A new command "systemctl revert" has been added that may be used to
2138 revert to the vendor version of a unit file, in case local changes
2139 have been made by adding drop-ins or overriding the unit file.
2141 * "machinectl clean" gained a new verb to automatically remove all or
2142 just hidden container images.
2144 * systemd-tmpfiles gained support for a new line type "e" for emptying
2145 directories, if they exist, without creating them if they don't.
2147 * systemd-nspawn gained support for automatically patching the UID/GIDs
2148 of the owners and the ACLs of all files and directories in a
2149 container tree to match the UID/GID user namespacing range selected
2150 for the container invocation. This mode is enabled via the new
2151 --private-users-chown switch. It also gained support for
2152 automatically choosing a free, previously unused UID/GID range when
2153 starting a container, via the new --private-users=pick setting (which
2154 implies --private-users-chown). Together, these options for the first
2155 time make user namespacing for nspawn containers fully automatic and
2156 thus deployable. The systemd-nspawn@.service template unit file has
2157 been changed to use this functionality by default.
2159 * systemd-nspawn gained a new --network-zone= switch, that allows
2160 creating ad-hoc virtual Ethernet links between multiple containers,
2161 that only exist as long as at least one container referencing them is
2162 running. This allows easy connecting of multiple containers with a
2163 common link that implements an Ethernet broadcast domain. Each of
2164 these network "zones" may be named relatively freely by the user, and
2165 may be referenced by any number of containers, but each container may
2166 only reference one of these "zones". On the lower level, this is
2167 implemented by an automatically managed bridge network interface for
2168 each zone, that is created when the first container referencing its
2169 zone is created and removed when the last one referencing its zone
2172 * The default start timeout may now be configured on the kernel command
2173 line via systemd.default_timeout_start_sec=. It was already
2174 configurable via the DefaultTimeoutStartSec= option in
2175 /etc/systemd/system.conf.
2177 * Socket units gained a new TriggerLimitIntervalSec= and
2178 TriggerLimitBurst= setting to configure a limit on the activation
2179 rate of the socket unit.
2181 * The LimitNICE= setting now optionally takes normal UNIX nice values
2182 in addition to the raw integer limit value. If the specified
2183 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2184 value is understood as UNIX nice value. If not prefixed like this it
2185 is understood as raw RLIMIT_NICE limit.
2187 * Note that the effect of the PrivateDevices= unit file setting changed
2188 slightly with this release: the per-device /dev file system will be
2189 mounted read-only from this version on, and will have "noexec"
2190 set. This (minor) change of behavior might cause some (exceptional)
2191 legacy software to break, when PrivateDevices=yes is set for its
2192 service. Please leave PrivateDevices= off if you run into problems
2195 * systemd-bootchart has been split out to a separate repository:
2196 https://github.com/systemd/systemd-bootchart
2198 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2199 merged into the kernel in its current form.
2201 * The compatibility libraries libsystemd-daemon.so,
2202 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2203 which have been deprecated since systemd-209 have been removed along
2204 with the corresponding pkg-config files. All symbols provided by
2205 those libraries are provided by libsystemd.so.
2207 * The Capabilities= unit file setting has been removed (it is ignored
2208 for backwards compatibility). AmbientCapabilities= and
2209 CapabilityBoundingSet= should be used instead.
2211 * A new special target has been added, initrd-root-device.target,
2212 which creates a synchronization point for dependencies of the root
2213 device in early userspace. Initramfs builders must ensure that this
2214 target is now included in early userspace.
2216 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2217 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2218 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2219 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2220 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2221 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2222 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2223 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2224 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2225 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2226 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2227 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2228 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2229 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2230 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2231 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2232 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2233 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2234 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2235 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2236 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2237 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2238 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2239 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2240 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2243 — Fairfax, 2016-05-21
2247 * The systemd-resolved DNS resolver service has gained a substantial
2248 set of new features, most prominently it may now act as a DNSSEC
2249 validating stub resolver. DNSSEC mode is currently turned off by
2250 default, but is expected to be turned on by default in one of the
2251 next releases. For now, we invite everybody to test the DNSSEC logic
2252 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2253 service also gained a full set of D-Bus interfaces, including calls
2254 to configure DNS and DNSSEC settings per link (for use by external
2255 network management software). systemd-resolved and systemd-networkd
2256 now distinguish between "search" and "routing" domains. The former
2257 are used to qualify single-label names, the latter are used purely
2258 for routing lookups within certain domains to specific links.
2259 resolved now also synthesizes RRs for all entries from /etc/hosts.
2261 * The systemd-resolve tool (which is a client utility for
2262 systemd-resolved) has been improved considerably and is now fully
2263 supported and documented. Hence it has moved from /usr/lib/systemd to
2266 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2269 * The coredump collection logic has been reworked: when a coredump is
2270 collected it is now written to disk, compressed and processed
2271 (including stacktrace extraction) from a new instantiated service
2272 systemd-coredump@.service, instead of directly from the
2273 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2274 processing large coredumps can take up a substantial amount of
2275 resources and time, and this previously happened entirely outside of
2276 systemd's service supervision. With the new logic the core_pattern
2277 hook only does minimal metadata collection before passing off control
2278 to the new instantiated service, which is configured with a time
2279 limit, a nice level and other settings to minimize negative impact on
2280 the rest of the system. Also note that the new logic will honour the
2281 RLIMIT_CORE setting of the crashed process, which now allows users
2282 and processes to turn off coredumping for their processes by setting
2285 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2286 and all forked processes by default. Previously, PID 1 would leave
2287 the setting at "0" for all processes, as set by the kernel. Note that
2288 the resource limit traditionally has no effect on the generated
2289 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2290 logic is used. Since the limit is now honoured (see above) its
2291 default has been changed so that the coredumping logic is enabled by
2292 default for all processes, while allowing specific opt-out.
2294 * When the stacktrace is extracted from processes of system users, this
2295 is now done as "systemd-coredump" user, in order to sandbox this
2296 potentially security sensitive parsing operation. (Note that when
2297 processing coredumps of normal users this is done under the user ID
2298 of process that crashed, as before.) Packagers should take notice
2299 that it is now necessary to create the "systemd-coredump" system user
2300 and group at package installation time.
2302 * The systemd-activate socket activation testing tool gained support
2303 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2304 and --seqpacket switches. It also has been extended to support both
2305 new-style and inetd-style file descriptor passing. Use the new
2306 --inetd switch to request inetd-style file descriptor passing.
2308 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2309 variable, which takes a boolean value. If set to false, ANSI color
2310 output is disabled in the tools even when run on a terminal that
2313 * The VXLAN support in networkd now supports two new settings
2314 DestinationPort= and PortRange=.
2316 * A new systemd.machine_id= kernel command line switch has been added,
2317 that may be used to set the machine ID in /etc/machine-id if it is
2318 not initialized yet. This command line option has no effect if the
2319 file is already initialized.
2321 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2322 specified command line as PID 2 rather than PID 1 in the
2323 container. In this mode PID 1 is a minimal stub init process that
2324 implements the special POSIX and Linux semantics of PID 1 regarding
2325 signal and child process management. Note that this stub init process
2326 is implemented in nspawn itself and requires no support from the
2327 container image. This new logic is useful to support running
2328 arbitrary commands in the container, as normal processes are
2329 generally not prepared to run as PID 1.
2331 * systemd-nspawn gained a new --chdir= switch for setting the current
2332 working directory for the process started in the container.
2334 * "journalctl /dev/sda" will now output all kernel log messages for
2335 specified device from the current boot, in addition to all devices
2336 that are parents of it. This should make log output about devices
2337 pretty useful, as long as kernel drivers attach enough metadata to
2338 the log messages. (The usual SATA drivers do.)
2340 * The sd-journal API gained two new calls
2341 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2342 that report whether log data from /run or /var has been found.
2344 * journalctl gained a new switch "--fields" that prints all journal
2345 record field names currently in use in the journal. This is backed
2346 by two new sd-journal API calls sd_journal_enumerate_fields() and
2347 sd_journal_restart_fields().
2349 * Most configurable timeouts in systemd now expect an argument of
2350 "infinity" to turn them off, instead of "0" as before. The semantics
2351 from now on is that a timeout of "0" means "now", and "infinity"
2352 means "never". To maintain backwards compatibility, "0" continues to
2353 turn off previously existing timeout settings.
2355 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2356 try-reload-or-restart" to clarify what it actually does: the "try"
2357 logic applies to both reloading and restarting, not just restarting.
2358 The old name continues to be accepted for compatibility.
2360 * On boot-up, when PID 1 detects that the system clock is behind the
2361 release date of the systemd version in use, the clock is now set
2362 to the latter. Previously, this was already done in timesyncd, in order
2363 to avoid running with clocks set to the various clock epochs such as
2364 1902, 1938 or 1970. With this change the logic is now done in PID 1
2365 in addition to timesyncd during early boot-up, so that it is enforced
2366 before the first process is spawned by systemd. Note that the logic
2367 in timesyncd remains, as it is more comprehensive and ensures
2368 clock monotonicity by maintaining a persistent timestamp file in
2369 /var. Since /var is generally not available in earliest boot or the
2370 initrd, this part of the logic remains in timesyncd, and is not done
2373 * Support for tweaking details in net_cls.class_id through the
2374 NetClass= configuration directive has been removed, as the kernel
2375 people have decided to deprecate that controller in cgroup v2.
2376 Userspace tools such as nftables are moving over to setting rules
2377 that are specific to the full cgroup path of a task, which obsoletes
2378 these controllers anyway. The NetClass= directive is kept around for
2379 legacy compatibility reasons. For a more in-depth description of the
2380 kernel change, please refer to the respective upstream commit:
2382 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2384 * A new service setting RuntimeMaxSec= has been added that may be used
2385 to specify a maximum runtime for a service. If the timeout is hit, the
2386 service is terminated and put into a failure state.
2388 * A new service setting AmbientCapabilities= has been added. It allows
2389 configuration of additional Linux process capabilities that are
2390 passed to the activated processes. This is only available on very
2393 * The process resource limit settings in service units may now be used
2394 to configure hard and soft limits individually.
2396 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2397 expose support for gcc's __attribute__((cleanup())) C extension.
2398 Specifically, for many object destructor functions alternative
2399 versions have been added that have names suffixed with "p" and take a
2400 pointer to a pointer to the object to destroy, instead of just a
2401 pointer to the object itself. This is useful because these destructor
2402 functions may be used directly as parameters to the cleanup
2403 construct. Internally, systemd has been a heavy user of this GCC
2404 extension for a long time, and with this change similar support is
2405 now available to consumers of the library outside of systemd. Note
2406 that by using this extension in your sources compatibility with old
2407 and strictly ANSI compatible C compilers is lost. However, all gcc or
2408 LLVM versions of recent years support this extension.
2410 * Timer units gained support for a new setting RandomizedDelaySec= that
2411 allows configuring some additional randomized delay to the configured
2412 time. This is useful to spread out timer events to avoid load peaks in
2413 clusters or larger setups.
2415 * Calendar time specifications now support sub-second accuracy.
2417 * Socket units now support listening on SCTP and UDP-lite protocol
2420 * The sd-event API now comes with a full set of man pages.
2422 * Older versions of systemd contained experimental support for
2423 compressing journal files and coredumps with the LZ4 compressor that
2424 was not compatible with the lz4 binary (due to API limitations of the
2425 lz4 library). This support has been removed; only support for files
2426 compatible with the lz4 binary remains. This LZ4 logic is now
2427 officially supported and no longer considered experimental.
2429 * The dkr image import logic has been removed again from importd. dkr's
2430 micro-services focus doesn't fit into the machine image focus of
2431 importd, and quickly got out of date with the upstream dkr API.
2433 * Creation of the /run/lock/lockdev/ directory was dropped from
2434 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2435 been available for many years. If you still need this, you need to
2436 create your own tmpfiles.d config file with:
2438 d /run/lock/lockdev 0775 root lock -
2440 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2441 and RebootArgument= have been moved from the [Service] section of
2442 unit files to [Unit], and they are now supported on all unit types,
2443 not just service units. Of course, elogind will continue to
2444 understand these settings also at the old location, in order to
2445 maintain compatibility.
2447 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2448 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2449 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2450 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2451 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2452 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2453 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2454 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2455 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2456 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2457 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2458 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2459 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2460 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2461 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2462 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2463 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2464 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2465 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2467 — Berlin, 2016-02-11
2471 * A number of properties previously only settable in unit
2472 files are now also available as properties to set when
2473 creating transient units programmatically via the bus, as it
2474 is exposed with systemd-run's --property=
2475 setting. Specifically, these are: SyslogIdentifier=,
2476 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2477 EnvironmentFile=, ReadWriteDirectories=,
2478 ReadOnlyDirectories=, InaccessibleDirectories=,
2479 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2481 * When creating transient services via the bus API it is now
2482 possible to pass in a set of file descriptors to use as
2483 STDIN/STDOUT/STDERR for the invoked process.
2485 * Slice units may now be created transiently via the bus APIs,
2486 similar to the way service and scope units may already be
2487 created transiently.
2489 * Wherever systemd expects a calendar timestamp specification
2490 (like in journalctl's --since= and --until= switches) UTC
2491 timestamps are now supported. Timestamps suffixed with "UTC"
2492 are now considered to be in Universal Time Coordinated
2493 instead of the local timezone. Also, timestamps may now
2494 optionally be specified with sub-second accuracy. Both of
2495 these additions also apply to recurring calendar event
2496 specification, such as OnCalendar= in timer units.
2498 * journalctl gained a new "--sync" switch that asks the
2499 journal daemon to write all so far unwritten log messages to
2500 disk and sync the files, before returning.
2502 * systemd-tmpfiles learned two new line types "q" and "Q" that
2503 operate like "v", but also set up a basic btrfs quota
2504 hierarchy when used on a btrfs file system with quota
2507 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2508 instead of a subvolume (even on a btrfs file system) if the
2509 root directory is a plain directory, and not a
2510 subvolume. This should simplify things with certain chroot()
2511 environments which are not aware of the concept of btrfs
2514 * systemd-detect-virt gained a new --chroot switch to detect
2515 whether execution takes place in a chroot() environment.
2517 * CPUAffinity= now takes CPU index ranges in addition to
2520 * The various memory-related resource limit settings (such as
2521 LimitAS=) now understand the usual K, M, G, ... suffixes to
2522 the base of 1024 (IEC). Similar, the time-related resource
2523 limit settings understand the usual min, h, day, ...
2526 * There's a new system.conf setting DefaultTasksMax= to
2527 control the default TasksMax= setting for services and
2528 scopes running on the system. (TasksMax= is the primary
2529 setting that exposes the "pids" cgroup controller on systemd
2530 and was introduced in the previous systemd release.) The
2531 setting now defaults to 512, which means services that are
2532 not explicitly configured otherwise will only be able to
2533 create 512 processes or threads at maximum, from this
2534 version on. Note that this means that thread- or
2535 process-heavy services might need to be reconfigured to set
2536 TasksMax= to a higher value. It is sufficient to set
2537 TasksMax= in these specific unit files to a higher value, or
2538 even "infinity". Similar, there's now a logind.conf setting
2539 UserTasksMax= that defaults to 4096 and limits the total
2540 number of processes or tasks each user may own
2541 concurrently. nspawn containers also have the TasksMax=
2542 value set by default now, to 8192. Note that all of this
2543 only has an effect if the "pids" cgroup controller is
2544 enabled in the kernel. The general benefit of these changes
2545 should be a more robust and safer system, that provides a
2546 certain amount of per-service fork() bomb protection.
2548 * systemd-nspawn gained the new --network-veth-extra= switch
2549 to define additional and arbitrarily-named virtual Ethernet
2550 links between the host and the container.
2552 * A new service execution setting PassEnvironment= has been
2553 added that allows importing select environment variables
2554 from PID1's environment block into the environment block of
2557 * Timer units gained support for a new RemainAfterElapse=
2558 setting which takes a boolean argument. It defaults to on,
2559 exposing behaviour unchanged to previous releases. If set to
2560 off, timer units are unloaded after they elapsed if they
2561 cannot elapse again. This is particularly useful for
2562 transient timer units, which shall not stay around longer
2563 than until they first elapse.
2565 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2566 default now (the kernel default is 16). This is beneficial
2567 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2568 allows substantially larger numbers of queued
2569 datagrams. This should increase the capability of systemd to
2570 parallelize boot-up, as logging and sd_notify() are unlikely
2571 to stall execution anymore. If you need to change the value
2572 from the new defaults, use the usual sysctl.d/ snippets.
2574 * The compression framing format used by the journal or
2575 coredump processing has changed to be in line with what the
2576 official LZ4 tools generate. LZ4 compression support in
2577 systemd was considered unsupported previously, as the format
2578 was not compatible with the normal tools. With this release
2579 this has changed now, and it is hence safe for downstream
2580 distributions to turn it on. While not compressing as well
2581 as the XZ, LZ4 is substantially faster, which makes
2582 it a good default choice for the compression logic in the
2583 journal and in coredump handling.
2585 * Any reference to /etc/mtab has been dropped from
2586 systemd. The file has been obsolete since a while, but
2587 systemd refused to work on systems where it was incorrectly
2588 set up (it should be a symlink or non-existent). Please make
2589 sure to update to util-linux 2.27.1 or newer in conjunction
2590 with this systemd release, which also drops any reference to
2591 /etc/mtab. If you maintain a distribution make sure that no
2592 software you package still references it, as this is a
2593 likely source of bugs. There's also a glibc bug pending,
2594 asking for removal of any reference to this obsolete file:
2596 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2598 Note that only util-linux versions built with
2599 --enable-libmount-force-mountinfo are supported.
2601 * Support for the ".snapshot" unit type has been removed. This
2602 feature turned out to be little useful and little used, and
2603 has now been removed from the core and from systemctl.
2605 * The dependency types RequiresOverridable= and
2606 RequisiteOverridable= have been removed from systemd. They
2607 have been used only very sparingly to our knowledge and
2608 other options that provide a similar effect (such as
2609 systemctl --mode=ignore-dependencies) are much more useful
2610 and commonly used. Moreover, they were only half-way
2611 implemented as the option to control behaviour regarding
2612 these dependencies was never added to systemctl. By removing
2613 these dependency types the execution engine becomes a bit
2614 simpler. Unit files that use these dependencies should be
2615 changed to use the non-Overridable dependency types
2616 instead. In fact, when parsing unit files with these
2617 options, that's what systemd will automatically convert them
2618 too, but it will also warn, asking users to fix the unit
2619 files accordingly. Removal of these dependency types should
2620 only affect a negligible number of unit files in the wild.
2622 * Behaviour of networkd's IPForward= option changed
2623 (again). It will no longer maintain a per-interface setting,
2624 but propagate one way from interfaces where this is enabled
2625 to the global kernel setting. The global setting will be
2626 enabled when requested by a network that is set up, but
2627 never be disabled again. This change was made to make sure
2628 IPv4 and IPv6 behaviour regarding packet forwarding is
2629 similar (as the Linux IPv6 stack does not support
2630 per-interface control of this setting) and to minimize
2633 * In unit files the behaviour of %u, %U, %h, %s has
2634 changed. These specifiers will now unconditionally resolve
2635 to the various user database fields of the user that the
2636 systemd instance is running as, instead of the user
2637 configured in the specific unit via User=. Note that this
2638 effectively doesn't change much, as resolving of these
2639 specifiers was already turned off in the --system instance
2640 of systemd, as we cannot do NSS lookups from PID 1. In the
2641 --user instance of systemd these specifiers where correctly
2642 resolved, but hardly made any sense, since the user instance
2643 lacks privileges to do user switches anyway, and User= is
2644 hence useless. Moreover, even in the --user instance of
2645 systemd behaviour was awkward as it would only take settings
2646 from User= assignment placed before the specifier into
2647 account. In order to unify and simplify the logic around
2648 this the specifiers will now always resolve to the
2649 credentials of the user invoking the manager (which in case
2650 of PID 1 is the root user).
2652 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2653 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2654 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2655 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2656 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2657 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2658 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2659 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2660 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2661 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2662 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2663 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2664 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2665 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2668 — Berlin, 2015-11-18
2672 * systemd now depends on util-linux v2.27. More specifically,
2673 the newly added mount monitor feature in libmount now
2674 replaces systemd's former own implementation.
2676 * libmount mandates /etc/mtab not to be regular file, and
2677 systemd now enforces this condition at early boot.
2678 /etc/mtab has been deprecated and warned about for a very
2679 long time, so systems running systemd should already have
2680 stopped having this file around as anything else than a
2681 symlink to /proc/self/mounts.
2683 * Support for the "pids" cgroup controller has been added. It
2684 allows accounting the number of tasks in a cgroup and
2685 enforcing limits on it. This adds two new setting
2686 TasksAccounting= and TasksMax= to each unit, as well as a
2687 global option DefaultTasksAccounting=.
2689 * Support for the "net_cls" cgroup controller has been added.
2690 It allows assigning a net class ID to each task in the
2691 cgroup, which can then be used in firewall rules and traffic
2692 shaping configurations. Note that the kernel netfilter net
2693 class code does not currently work reliably for ingress
2694 packets on unestablished sockets.
2696 This adds a new config directive called NetClass= to CGroup
2697 enabled units. Allowed values are positive numbers for fixed
2698 assignments and "auto" for picking a free value
2701 * 'systemctl is-system-running' now returns 'offline' if the
2702 system is not booted with systemd. This command can now be
2703 used as a substitute for 'systemd-notify --booted'.
2705 * Watchdog timeouts have been increased to 3 minutes for all
2706 in-tree service files. Apparently, disk IO issues are more
2707 frequent than we hoped, and user reported >1 minute waiting
2710 * 'machine-id-commit' functionality has been merged into
2711 'machine-id-setup --commit'. The separate binary has been
2714 * The WorkingDirectory= directive in unit files may now be set
2715 to the special value '~'. In this case, the working
2716 directory is set to the home directory of the user
2717 configured in User=.
2719 * "machinectl shell" will now open the shell in the home
2720 directory of the selected user by default.
2722 * The CrashChVT= configuration file setting is renamed to
2723 CrashChangeVT=, following our usual logic of not
2724 abbreviating unnecessarily. The old directive is still
2725 supported for compat reasons. Also, this directive now takes
2726 an integer value between 1 and 63, or a boolean value. The
2727 formerly supported '-1' value for disabling stays around for
2730 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2731 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2732 RootDirectory= properties can now be set for transient
2735 * The systemd-analyze tool gained a new "set-log-target" verb
2736 to change the logging target the system manager logs to
2737 dynamically during runtime. This is similar to how
2738 "systemd-analyze set-log-level" already changes the log
2741 * In nspawn /sys is now mounted as tmpfs, with only a selected
2742 set of subdirectories mounted in from the real sysfs. This
2743 enhances security slightly, and is useful for ensuring user
2744 namespaces work correctly.
2746 * Support for USB FunctionFS activation has been added. This
2747 allows implementation of USB gadget services that are
2748 activated as soon as they are requested, so that they don't
2749 have to run continuously, similar to classic socket
2752 * The "systemctl exit" command now optionally takes an
2753 additional parameter that sets the exit code to return from
2754 the systemd manager when exiting. This is only relevant when
2755 running the systemd user instance, or when running the
2756 system instance in a container.
2758 * sd-bus gained the new API calls sd_bus_path_encode_many()
2759 and sd_bus_path_decode_many() that allow easy encoding and
2760 decoding of multiple identifier strings inside a D-Bus
2761 object path. Another new call sd_bus_default_flush_close()
2762 has been added to flush and close per-thread default
2765 * systemd-cgtop gained support for a -M/--machine= switch to
2766 show the control groups within a certain container only.
2768 * "systemctl kill" gained support for an optional --fail
2769 switch. If specified the requested operation will fail of no
2770 processes have been killed, because the unit had no
2771 processes attached, or similar.
2773 * A new systemd.crash_reboot=1 kernel command line option has
2774 been added that triggers a reboot after crashing. This can
2775 also be set through CrashReboot= in systemd.conf.
2777 * The RuntimeDirectory= setting now understands unit
2778 specifiers like %i or %f.
2780 * A new (still internal) library API sd-ipv4acd has been added,
2781 that implements address conflict detection for IPv4. It's
2782 based on code from sd-ipv4ll, and will be useful for
2783 detecting DHCP address conflicts.
2785 * File descriptors passed during socket activation may now be
2786 named. A new API sd_listen_fds_with_names() is added to
2787 access the names. The default names may be overridden,
2788 either in the .socket file using the FileDescriptorName=
2789 parameter, or by passing FDNAME= when storing the file
2790 descriptors using sd_notify().
2792 * systemd-networkd gained support for:
2794 - Setting the IPv6 Router Advertisement settings via
2795 IPv6AcceptRouterAdvertisements= in .network files.
2797 - Configuring the HelloTimeSec=, MaxAgeSec= and
2798 ForwardDelaySec= bridge parameters in .netdev files.
2800 - Configuring PreferredSource= for static routes in
2803 * The "ask-password" framework used to query for LUKS harddisk
2804 passwords or SSL passwords during boot gained support for
2805 caching passwords in the kernel keyring, if it is
2806 available. This makes sure that the user only has to type in
2807 a passphrase once if there are multiple objects to unlock
2808 with the same one. Previously, such password caching was
2809 available only when Plymouth was used; this moves the
2810 caching logic into the systemd codebase itself. The
2811 "systemd-ask-password" utility gained a new --keyname=
2812 switch to control which kernel keyring key to use for
2813 caching a password in. This functionality is also useful for
2814 enabling display managers such as gdm to automatically
2815 unlock the user's GNOME keyring if its passphrase, the
2816 user's password and the harddisk password are the same, if
2817 gdm-autologin is used.
2819 * When downloading tar or raw images using "machinectl
2820 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2821 file is now also downloaded, if it is available and stored
2822 next to the image file.
2824 * Units of type ".socket" gained a new boolean setting
2825 Writable= which is only useful in conjunction with
2826 ListenSpecial=. If true, enables opening the specified
2827 special file in O_RDWR mode rather than O_RDONLY mode.
2829 * systemd-rfkill has been reworked to become a singleton
2830 service that is activated through /dev/rfkill on each rfkill
2831 state change and saves the settings to disk. This way,
2832 systemd-rfkill is now compatible with devices that exist
2833 only intermittendly, and even restores state if the previous
2834 system shutdown was abrupt rather than clean.
2836 * The journal daemon gained support for vacuuming old journal
2837 files controlled by the number of files that shall remain,
2838 in addition to the already existing control by size and by
2839 date. This is useful as journal interleaving performance
2840 degrades with too many separate journal files, and allows
2841 putting an effective limit on them. The new setting defaults
2842 to 100, but this may be changed by setting SystemMaxFiles=
2843 and RuntimeMaxFiles= in journald.conf. Also, the
2844 "journalctl" tool gained the new --vacuum-files= switch to
2845 manually vacuum journal files to leave only the specified
2846 number of files in place.
2848 * udev will now create /dev/disk/by-path links for ATA devices
2849 on kernels where that is supported.
2851 * Galician, Serbian, Turkish and Korean translations were added.
2853 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2854 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2855 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2856 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2857 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2858 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2859 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2860 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2861 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2862 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2863 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2864 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2865 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2866 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2867 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2868 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2869 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2870 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2872 — Berlin, 2015-10-07
2876 * The DHCP implementation of systemd-networkd gained a set of
2879 - The DHCP server now supports emitting DNS and NTP
2880 information. It may be enabled and configured via
2881 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2882 and NTP information is enabled, but no servers are
2883 configured, the corresponding uplink information (if there
2884 is any) is propagated.
2886 - Server and client now support transmission and reception
2887 of timezone information. It can be configured via the
2888 newly introduced network options UseTimezone=,
2889 EmitTimezone=, and Timezone=. Transmission of timezone
2890 information is enabled between host and containers by
2891 default now: the container will change its local timezone
2892 to what the host has set.
2894 - Lease timeouts can now be configured via
2895 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2897 - The DHCP server improved on the stability of
2898 leases. Clients are more likely to get the same lease
2899 information back, even if the server loses state.
2901 - The DHCP server supports two new configuration options to
2902 control the lease address pool metrics, PoolOffset= and
2905 * The encapsulation limit of tunnels in systemd-networkd may
2906 now be configured via 'EncapsulationLimit='. It allows
2907 modifying the maximum additional levels of encapsulation
2908 that are permitted to be prepended to a packet.
2910 * systemd now supports the concept of user buses replacing
2911 session buses, if used with dbus-1.10 (and enabled via dbus
2912 --enable-user-session). It previously only supported this on
2913 kdbus-enabled systems, and this release expands this to
2914 'dbus-daemon' systems.
2916 * systemd-networkd now supports predictable interface names
2919 * systemd now optionally supports the new Linux kernel
2920 "unified" control group hierarchy. If enabled via the kernel
2921 command-line option 'systemd.unified_cgroup_hierarchy=1',
2922 systemd will try to mount the unified cgroup hierarchy
2923 directly on /sys/fs/cgroup. If not enabled, or not
2924 available, systemd will fall back to the legacy cgroup
2925 hierarchy setup, as before. Host system and containers can
2926 mix and match legacy and unified hierarchies as they
2927 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2928 environment variable to individually select the hierarchy to
2929 use for executed containers. By default, nspawn will use the
2930 unified hierarchy for the containers if the host uses the
2931 unified hierarchy, and the legacy hierarchy otherwise.
2932 Please note that at this point the unified hierarchy is an
2933 experimental kernel feature and is likely to change in one
2934 of the next kernel releases. Therefore, it should not be
2935 enabled by default in downstream distributions yet. The
2936 minimum required kernel version for the unified hierarchy to
2937 work is 4.2. Note that when the unified hierarchy is used
2938 for the first time delegated access to controllers is
2939 safe. Because of this systemd-nspawn containers will get
2940 access to controllers now, as will systemd user
2941 sessions. This means containers and user sessions may now
2942 manage their own resources, partitioning up what the system
2945 * A new special scope unit "init.scope" has been introduced
2946 that encapsulates PID 1 of the system. It may be used to
2947 determine resource usage and enforce resource limits on PID
2948 1 itself. PID 1 hence moved out of the root of the control
2951 * The cgtop tool gained support for filtering out kernel
2952 threads when counting tasks in a control group. Also, the
2953 count of processes is now recursively summed up by
2954 default. Two options -k and --recursive= have been added to
2955 revert to old behaviour. The tool has also been updated to
2956 work correctly in containers now.
2958 * systemd-nspawn's --bind= and --bind-ro= options have been
2959 extended to allow creation of non-recursive bind mounts.
2961 * libsystemd gained two new calls sd_pid_get_cgroup() and
2962 sd_peer_get_cgroup() which return the control group path of
2963 a process or peer of a connected AF_UNIX socket. This
2964 function call is particularly useful when implementing
2965 delegated subtrees support in the control group hierarchy.
2967 * The "sd-event" event loop API of libsystemd now supports
2968 correct dequeuing of real-time signals, without losing
2971 * When systemd requests a PolicyKit decision when managing
2972 units it will now add additional fields to the request,
2973 including unit name and desired operation. This enables more
2974 powerful PolicyKit policies, that make decisions depending
2975 on these parameters.
2977 * nspawn learnt support for .nspawn settings files, that may
2978 accompany the image files or directories of containers, and
2979 may contain additional settings for the container. This is
2980 an alternative to configuring container parameters via the
2981 nspawn command line.
2983 Contributions from: Cristian Rodríguez, Daniel Mack, David
2984 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2985 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2986 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2987 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2988 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2989 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2990 Andersen, Tom Gundersen, Torstein Husebø
2992 — Berlin, 2015-09-08
2996 * machinectl gained a new verb 'shell' which opens a fresh
2997 shell on the target container or the host. It is similar to
2998 the existing 'login' command of machinectl, but spawns the
2999 shell directly without prompting for username or
3000 password. The pseudo machine '.host' now refers to the local
3001 host and is used by default. Hence, 'machinectl shell' can
3002 be used as replacement for 'su -' which spawns a session as
3003 a fresh systemd unit in a way that is fully isolated from
3004 the originating session.
3006 * systemd-networkd learned to cope with private-zone DHCP
3007 options and allows other programs to query the values.
3009 * SELinux access control when enabling/disabling units is no
3010 longer enforced with this release. The previous
3011 implementation was incorrect, and a new corrected
3012 implementation is not yet available. As unit file operations
3013 are still protected via PolicyKit and D-Bus policy this is
3014 not a security problem. Yet, distributions which care about
3015 optimal SELinux support should probably not stabilize on
3018 * sd-bus gained support for matches of type "arg0has=", that
3019 test for membership of strings in string arrays sent in bus
3022 * systemd-resolved now dumps the contents of its DNS and LLMNR
3023 caches to the logs on reception of the SIGUSR1 signal. This
3024 is useful to debug DNS behaviour.
3026 * The coredumpctl tool gained a new --directory= option to
3027 operate on journal files in a specific directory.
3029 * "systemctl reboot" and related commands gained a new
3030 "--message=" option which may be used to set a free-text
3031 wall message when shutting down or rebooting the
3032 system. This message is also logged, which is useful for
3033 figuring out the reason for a reboot or shutdown a
3036 * The "systemd-resolve-host" tool's -i switch now takes
3037 network interface numbers as alternative to interface names.
3039 * A new unit file setting for services has been introduced:
3040 UtmpMode= allows configuration of how precisely systemd
3041 handles utmp and wtmp entries for the service if this is
3042 enabled. This allows writing services that appear similar to
3043 user sessions in the output of the "w", "who", "last" and
3046 * systemd-resolved will now locally synthesize DNS resource
3047 records for the "localhost" and "gateway" domains as well as
3048 the local hostname. This should ensure that clients querying
3049 RRs via resolved will get similar results as those going via
3050 NSS, if nss-myhostname is enabled.
3052 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3053 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3054 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3055 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3056 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3057 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3058 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3059 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3060 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3061 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3062 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3063 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3065 — Berlin, 2015-08-27
3069 * The systemd-efi-boot-generator functionality was merged into
3070 systemd-gpt-auto-generator.
3072 * systemd-networkd now supports Group Policy for vxlan
3073 devices. It can be enabled via the new boolean configuration
3074 option called 'GroupPolicyExtension='.
3076 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3077 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3078 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3080 — Berlin, 2015-07-31
3084 * The python-systemd code has been removed from the systemd repository.
3085 A new repository has been created which accommodates the code from
3086 now on, and we kindly ask distributions to create a separate package
3087 for this: https://github.com/systemd/python-systemd
3089 * The systemd daemon will now reload its main configuration
3090 (/etc/systemd/system.conf) on daemon-reload.
3092 * sd-dhcp now exposes vendor specific extensions via
3093 sd_dhcp_lease_get_vendor_specific().
3095 * systemd-networkd gained a number of new configuration options.
3097 - A new boolean configuration option for TAP devices called
3098 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3099 device, thus allowing to send and receive GSO packets.
3101 - A new tunnel configuration option called 'CopyDSCP='.
3102 If enabled, the DSCP field of ip6 tunnels is copied into the
3103 decapsulated packet.
3105 - A set of boolean bridge configuration options were added.
3106 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3107 and 'UnicastFlood=' are now parsed by networkd and applied to the
3108 respective bridge link device via the respective IFLA_BRPORT_*
3111 - A new string configuration option to override the hostname sent
3112 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3113 is true, networkd will use the configured hostname instead of the
3114 system hostname when sending DHCP requests.
3116 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3117 networkd will configure the IPv6 flow-label of the tunnel device
3118 according to RFC2460.
3120 - The 'macvtap' virtual network devices are now supported, similar to
3121 the already supported 'macvlan' devices.
3123 * systemd-resolved now implements RFC5452 to improve resilience against
3124 cache poisoning. Additionally, source port randomization is enabled
3125 by default to further protect against DNS spoofing attacks.
3127 * nss-mymachines now supports translating UIDs and GIDs of running
3128 containers with user-namespaces enabled. If a container 'foo'
3129 translates a host uid 'UID' to the container uid 'TUID', then
3130 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3131 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3132 mapped as 'vg-foo-TGID'.
3134 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3135 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3136 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3137 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3138 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3139 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3140 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3141 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3142 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3143 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3145 — Berlin, 2015-07-29
3149 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3150 There are no known issues with current sysfs, and udev does not need
3151 or should be used to work around such bugs.
3153 * udev does no longer enable USB HID power management. Several reports
3154 indicate, that some devices cannot handle that setting.
3156 * The udev accelerometer helper was removed. The functionality
3157 is now fully included in iio-sensor-proxy. But this means,
3158 older iio-sensor-proxy versions will no longer provide
3159 accelerometer/orientation data with this systemd version.
3160 Please upgrade iio-sensor-proxy to version 1.0.
3162 * networkd gained a new configuration option IPv6PrivacyExtensions=
3163 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3164 for Stateless Address") on selected networks.
3166 * For the sake of fewer build-time dependencies and less code in the
3167 main repository, the python bindings are about to be removed in the
3168 next release. A new repository has been created which accommodates
3169 the code from now on, and we kindly ask distributions to create a
3170 separate package for this. The removal will take place in v223.
3172 https://github.com/systemd/python-systemd
3174 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3175 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3176 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3177 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3178 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3179 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3180 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3181 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3182 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3183 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3185 — Berlin, 2015-07-07
3189 * The sd-bus.h and sd-event.h APIs have now been declared
3190 stable and have been added to the official interface of
3191 libsystemd.so. sd-bus implements an alternative D-Bus client
3192 library, that is relatively easy to use, very efficient and
3193 supports both classic D-Bus as well as kdbus as transport
3194 backend. sd-event is a generic event loop abstraction that
3195 is built around Linux epoll, but adds features such as event
3196 prioritization or efficient timer handling. Both APIs are good
3197 choices for C programs looking for a bus and/or event loop
3198 implementation that is minimal and does not have to be
3199 portable to other kernels.
3201 * kdbus support is no longer compile-time optional. It is now
3202 always built-in. However, it can still be disabled at
3203 runtime using the kdbus=0 kernel command line setting, and
3204 that setting may be changed to default to off, by specifying
3205 --disable-kdbus at build-time. Note though that the kernel
3206 command line setting has no effect if the kdbus.ko kernel
3207 module is not installed, in which case kdbus is (obviously)
3208 also disabled. We encourage all downstream distributions to
3209 begin testing kdbus by adding it to the kernel images in the
3210 development distributions, and leaving kdbus support in
3213 * The minimal required util-linux version has been bumped to
3216 * Support for chkconfig (--enable-chkconfig) was removed in
3217 favor of calling an abstraction tool
3218 /lib/systemd/systemd-sysv-install. This needs to be
3219 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3220 in README for details.
3222 * If there's a systemd unit and a SysV init script for the
3223 same service name, and the user executes "systemctl enable"
3224 for it (or a related call), then this will now enable both
3225 (or execute the related operation on both), not just the
3228 * The libudev API documentation has been converted from gtkdoc
3231 * gudev has been removed from the systemd tree, it is now an
3234 * The systemd-cgtop tool learnt a new --raw switch to generate
3235 "raw" (machine parsable) output.
3237 * networkd's IPForwarding= .network file setting learnt the
3238 new setting "kernel", which ensures that networkd does not
3239 change the IP forwarding sysctl from the default kernel
3242 * The systemd-logind bus API now exposes a new boolean
3243 property "Docked" that reports whether logind considers the
3244 system "docked", i.e. connected to a docking station or not.
3246 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3247 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3248 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3249 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3250 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3251 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3252 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3253 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3254 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3255 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3256 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3257 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3258 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3259 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3260 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3261 Fink, Zbigniew Jędrzejewski-Szmek
3263 — Berlin, 2015-06-19
3267 * The gudev library has been extracted into a separate repository
3268 available at: https://git.gnome.org/browse/libgudev/
3269 It is now managed as part of the Gnome project. Distributions
3270 are recommended to pass --disable-gudev to systemd and use
3271 gudev from the Gnome project instead. gudev is still included
3272 in systemd, for now. It will be removed soon, though. Please
3273 also see the announcement-thread on systemd-devel:
3274 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3276 * systemd now exposes a CPUUsageNSec= property for each
3277 service unit on the bus, that contains the overall consumed
3278 CPU time of a service (the sum of what each process of the
3279 service consumed). This value is only available if
3280 CPUAccounting= is turned on for a service, and is then shown
3281 in the "systemctl status" output.
3283 * Support for configuring alternative mappings of the old SysV
3284 runlevels to systemd targets has been removed. They are now
3285 hardcoded in a way that runlevels 2, 3, 4 all map to
3286 multi-user.target and 5 to graphical.target (which
3287 previously was already the default behaviour).
3289 * The auto-mounter logic gained support for mount point
3290 expiry, using a new TimeoutIdleSec= setting in .automount
3291 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3293 * The EFI System Partition (ESP) as mounted to /boot by
3294 systemd-efi-boot-generator will now be unmounted
3295 automatically after 2 minutes of not being used. This should
3296 minimize the risk of ESP corruptions.
3298 * New /etc/fstab options x-systemd.requires= and
3299 x-systemd.requires-mounts-for= are now supported to express
3300 additional dependencies for mounts. This is useful for
3301 journalling file systems that support external journal
3302 devices or overlay file systems that require underlying file
3303 systems to be mounted.
3305 * systemd does not support direct live-upgrades (via systemctl
3306 daemon-reexec) from versions older than v44 anymore. As no
3307 distribution we are aware of shipped such old versions in a
3308 stable release this should not be problematic.
3310 * When systemd forks off a new per-connection service instance
3311 it will now set the $REMOTE_ADDR environment variable to the
3312 remote IP address, and $REMOTE_PORT environment variable to
3313 the remote IP port. This behaviour is similar to the
3314 corresponding environment variables defined by CGI.
3316 * systemd-networkd gained support for uplink failure
3317 detection. The BindCarrier= option allows binding interface
3318 configuration dynamically to the link sense of other
3319 interfaces. This is useful to achieve behaviour like in
3322 * systemd-networkd gained support for configuring the DHCP
3323 client identifier to use when requesting leases.
3325 * systemd-networkd now has a per-network UseNTP= option to
3326 configure whether NTP server information acquired via DHCP
3327 is passed on to services like systemd-timesyncd.
3329 * systemd-networkd gained support for vti6 tunnels.
3331 * Note that systemd-networkd manages the sysctl variable
3332 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3333 it is configured for since v219. The variable controls IP
3334 forwarding, and is a per-interface alternative to the global
3335 /proc/sys/net/ipv[46]/ip_forward. This setting is
3336 configurable in the IPForward= option, which defaults to
3337 "no". This means if networkd is used for an interface it is
3338 no longer sufficient to set the global sysctl option to turn
3339 on IP forwarding! Instead, the .network file option
3340 IPForward= needs to be turned on! Note that the
3341 implementation of this behaviour was broken in v219 and has
3344 * Many bonding and vxlan options are now configurable in
3347 * systemd-nspawn gained a new --property= setting to set unit
3348 properties for the container scope. This is useful for
3349 setting resource parameters (e.g. "CPUShares=500") on
3350 containers started from the command line.
3352 * systemd-nspawn gained a new --private-users= switch to make
3353 use of user namespacing available on recent Linux kernels.
3355 * systemd-nspawn may now be called as part of a shell pipeline
3356 in which case the pipes used for stdin and stdout are passed
3357 directly to the process invoked in the container, without
3358 indirection via a pseudo tty.
3360 * systemd-nspawn gained a new switch to control the UNIX
3361 signal to use when killing the init process of the container
3364 * systemd-nspawn gained a new --overlay= switch for mounting
3365 overlay file systems into the container using the new kernel
3368 * When a container image is imported via systemd-importd and
3369 the host file system is not btrfs, a loopback block device
3370 file is created in /var/lib/machines.raw with a btrfs file
3371 system inside. It is then mounted to /var/lib/machines to
3372 enable btrfs features for container management. The loopback
3373 file and btrfs file system is grown as needed when container
3374 images are imported via systemd-importd.
3376 * systemd-machined/systemd-importd gained support for btrfs
3377 quota, to enforce container disk space limits on disk. This
3378 is exposed in "machinectl set-limit".
3380 * systemd-importd now can import containers from local .tar,
3381 .raw and .qcow2 images, and export them to .tar and .raw. It
3382 can also import dkr v2 images now from the network (on top
3385 * systemd-importd gained support for verifying downloaded
3386 images with gpg2 (previously only gpg1 was supported).
3388 * systemd-machined, systemd-logind, systemd: most bus calls
3389 are now accessible to unprivileged processes via
3390 PolicyKit. Also, systemd-logind will now allow users to kill
3391 their own sessions without further privileges or
3394 * systemd-shutdownd has been removed. This service was
3395 previously responsible for implementing scheduled shutdowns
3396 as exposed in /usr/bin/shutdown's time parameter. This
3397 functionality has now been moved into systemd-logind and is
3398 accessible via a bus interface.
3400 * "systemctl reboot" gained a new switch --firmware-setup that
3401 can be used to reboot into the EFI firmware setup, if that
3402 is available. systemd-logind now exposes an API on the bus
3403 to trigger such reboots, in case graphical desktop UIs want
3404 to cover this functionality.
3406 * "systemctl enable", "systemctl disable" and "systemctl mask"
3407 now support a new "--now" switch. If specified the units
3408 that are enabled will also be started, and the ones
3409 disabled/masked also stopped.
3411 * The Gummiboot EFI boot loader tool has been merged into
3412 systemd, and renamed to "systemd-boot". The bootctl tool has been
3413 updated to support systemd-boot.
3415 * An EFI kernel stub has been added that may be used to create
3416 kernel EFI binaries that contain not only the actual kernel,
3417 but also an initrd, boot splash, command line and OS release
3418 information. This combined binary can then be signed as a
3419 single image, so that the firmware can verify it all in one
3420 step. systemd-boot has special support for EFI binaries created
3421 like this and can extract OS release information from them
3422 and show them in the boot menu. This functionality is useful
3423 to implement cryptographically verified boot schemes.
3425 * Optional support has been added to systemd-fsck to pass
3426 fsck's progress report to an AF_UNIX socket in the file
3429 * udev will no longer create device symlinks for all block
3430 devices by default. A blacklist for excluding special block
3431 devices from this logic has been turned into a whitelist
3432 that requires picking block devices explicitly that require
3435 * A new (currently still internal) API sd-device.h has been
3436 added to libsystemd. This modernized API is supposed to
3437 replace libudev eventually. In fact, already much of libudev
3438 is now just a wrapper around sd-device.h.
3440 * A new hwdb database for storing metadata about pointing
3441 stick devices has been added.
3443 * systemd-tmpfiles gained support for setting file attributes
3444 similar to the "chattr" tool with new 'h' and 'H' lines.
3446 * systemd-journald will no longer unconditionally set the
3447 btrfs NOCOW flag on new journal files. This is instead done
3448 with tmpfiles snippet using the new 'h' line type. This
3449 allows easy disabling of this logic, by masking the
3450 journal-nocow.conf tmpfiles file.
3452 * systemd-journald will now translate audit message types to
3453 human readable identifiers when writing them to the
3454 journal. This should improve readability of audit messages.
3456 * The LUKS logic gained support for the offset= and skip=
3457 options in /etc/crypttab, as previously implemented by
3460 * /usr/lib/os-release gained a new optional field VARIANT= for
3461 distributions that support multiple variants (such as a
3462 desktop edition, a server edition, ...)
3464 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3465 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3466 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3467 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3468 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3469 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3470 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3471 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3472 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3473 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3474 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3475 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3476 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3477 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3478 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3479 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3480 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3481 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3482 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3483 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3484 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3485 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3486 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3487 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3488 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3489 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3490 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3492 — Berlin, 2015-05-22
3496 * Introduce a new API "sd-hwdb.h" for querying the hardware
3497 metadata database. With this minimal interface one can query
3498 and enumerate the udev hwdb, decoupled from the old libudev
3499 library. libudev's interface for this is now only a wrapper
3500 around sd-hwdb. A new tool systemd-hwdb has been added to
3501 interface with and update the database.
3503 * When any of systemd's tools copies files (for example due to
3504 tmpfiles' C lines) a btrfs reflink will attempted first,
3505 before bytewise copying is done.
3507 * systemd-nspawn gained a new --ephemeral switch. When
3508 specified a btrfs snapshot is taken of the container's root
3509 directory, and immediately removed when the container
3510 terminates again. Thus, a container can be started whose
3511 changes never alter the container's root directory, and are
3512 lost on container termination. This switch can also be used
3513 for starting a container off the root file system of the
3514 host without affecting the host OS. This switch is only
3515 available on btrfs file systems.
3517 * systemd-nspawn gained a new --template= switch. It takes the
3518 path to a container tree to use as template for the tree
3519 specified via --directory=, should that directory be
3520 missing. This allows instantiating containers dynamically,
3521 on first run. This switch is only available on btrfs file
3524 * When a .mount unit refers to a mount point on which multiple
3525 mounts are stacked, and the .mount unit is stopped all of
3526 the stacked mount points will now be unmounted until no
3527 mount point remains.
3529 * systemd now has an explicit notion of supported and
3530 unsupported unit types. Jobs enqueued for unsupported unit
3531 types will now fail with an "unsupported" error code. More
3532 specifically .swap, .automount and .device units are not
3533 supported in containers, .busname units are not supported on
3534 non-kdbus systems. .swap and .automount are also not
3535 supported if their respective kernel compile time options
3538 * machinectl gained support for two new "copy-from" and
3539 "copy-to" commands for copying files from a running
3540 container to the host or vice versa.
3542 * machinectl gained support for a new "bind" command to bind
3543 mount host directories into local containers. This is
3544 currently only supported for nspawn containers.
3546 * networkd gained support for configuring bridge forwarding
3547 database entries (fdb) from .network files.
3549 * A new tiny daemon "systemd-importd" has been added that can
3550 download container images in tar, raw, qcow2 or dkr formats,
3551 and make them available locally in /var/lib/machines, so
3552 that they can run as nspawn containers. The daemon can GPG
3553 verify the downloads (not supported for dkr, since it has no
3554 provisions for verifying downloads). It will transparently
3555 decompress bz2, xz, gzip compressed downloads if necessary,
3556 and restore sparse files on disk. The daemon uses privilege
3557 separation to ensure the actual download logic runs with
3558 fewer privileges than the daemon itself. machinectl has
3559 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3560 make the functionality of importd available to the
3561 user. With this in place the Fedora and Ubuntu "Cloud"
3562 images can be downloaded and booted as containers unmodified
3563 (the Fedora images lack the appropriate GPG signature files
3564 currently, so they cannot be verified, but this will change
3565 soon, hopefully). Note that downloading images is currently
3566 only fully supported on btrfs.
3568 * machinectl is now able to list container images found in
3569 /var/lib/machines, along with some metadata about sizes of
3570 disk and similar. If the directory is located on btrfs and
3571 quota is enabled, this includes quota display. A new command
3572 "image-status" has been added that shows additional
3573 information about images.
3575 * machinectl is now able to clone container images
3576 efficiently, if the underlying file system (btrfs) supports
3577 it, with the new "machinectl clone" command. It also
3578 gained commands for renaming and removing images, as well as
3579 marking them read-only or read-write (supported also on
3580 legacy file systems).
3582 * networkd gained support for collecting LLDP network
3583 announcements, from hardware that supports this. This is
3584 shown in networkctl output.
3586 * systemd-run gained support for a new -t (--pty) switch for
3587 invoking a binary on a pty whose input and output is
3588 connected to the invoking terminal. This allows executing
3589 processes as system services while interactively
3590 communicating with them via the terminal. Most interestingly
3591 this is supported across container boundaries. Invoking
3592 "systemd-run -t /bin/bash" is an alternative to running a
3593 full login session, the difference being that the former
3594 will not register a session, nor go through the PAM session
3597 * tmpfiles gained support for a new "v" line type for creating
3598 btrfs subvolumes. If the underlying file system is a legacy
3599 file system, this automatically degrades to creating a
3600 normal directory. Among others /var/lib/machines is now
3601 created like this at boot, should it be missing.
3603 * The directory /var/lib/containers/ has been deprecated and
3604 been replaced by /var/lib/machines. The term "machines" has
3605 been used in the systemd context as generic term for both
3606 VMs and containers, and hence appears more appropriate for
3607 this, as the directory can also contain raw images bootable
3610 * systemd-nspawn when invoked with -M but without --directory=
3611 or --image= is now capable of searching for the container
3612 root directory, subvolume or disk image automatically, in
3613 /var/lib/machines. systemd-nspawn@.service has been updated
3614 to make use of this, thus allowing it to be used for raw
3617 * A new machines.target unit has been introduced that is
3618 supposed to group all containers/VMs invoked as services on
3619 the system. systemd-nspawn@.service has been updated to
3620 integrate with that.
3622 * machinectl gained a new "start" command, for invoking a
3623 container as a service. "machinectl start foo" is mostly
3624 equivalent to "systemctl start systemd-nspawn@foo.service",
3625 but handles escaping in a nicer way.
3627 * systemd-nspawn will now mount most of the cgroupfs tree
3628 read-only into each container, with the exception of the
3629 container's own subtree in the name=systemd hierarchy.
3631 * journald now sets the special FS_NOCOW file flag for its
3632 journal files. This should improve performance on btrfs, by
3633 avoiding heavy fragmentation when journald's write-pattern
3634 is used on COW file systems. It degrades btrfs' data
3635 integrity guarantees for the files to the same levels as for
3636 ext3/ext4 however. This should be OK though as journald does
3637 its own data integrity checks and all its objects are
3638 checksummed on disk. Also, journald should handle btrfs disk
3639 full events a lot more gracefully now, by processing SIGBUS
3640 errors, and not relying on fallocate() anymore.
3642 * When journald detects that journal files it is writing to
3643 have been deleted it will immediately start new journal
3646 * systemd now provides a way to store file descriptors
3647 per-service in PID 1. This is useful for daemons to ensure
3648 that fds they require are not lost during a daemon
3649 restart. The fds are passed to the daemon on the next
3650 invocation in the same way socket activation fds are
3651 passed. This is now used by journald to ensure that the
3652 various sockets connected to all the system's stdout/stderr
3653 are not lost when journald is restarted. File descriptors
3654 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3655 an extension to sd_notify(). Note that a limit is enforced
3656 on the number of fds a service can store in PID 1, and it
3657 defaults to 0, so that no fds may be stored, unless this is
3658 explicitly turned on.
3660 * The default TERM variable to use for units connected to a
3661 terminal, when no other value is explicitly is set is now
3662 vt220 rather than vt102. This should be fairly safe still,
3663 but allows PgUp/PgDn work.
3665 * The /etc/crypttab option header= as known from Debian is now
3668 * "loginctl user-status" and "loginctl session-status" will
3669 now show the last 10 lines of log messages of the
3670 user/session following the status output. Similar,
3671 "machinectl status" will show the last 10 log lines
3672 associated with a virtual machine or container
3673 service. (Note that this is usually not the log messages
3674 done in the VM/container itself, but simply what the
3675 container manager logs. For nspawn this includes all console
3678 * "loginctl session-status" without further argument will now
3679 show the status of the session of the caller. Similar,
3680 "lock-session", "unlock-session", "activate",
3681 "enable-linger", "disable-linger" may now be called without
3682 session/user parameter in which case they apply to the
3683 caller's session/user.
3685 * An X11 session scriptlet is now shipped that uploads
3686 $DISPLAY and $XAUTHORITY into the environment of the systemd
3687 --user daemon if a session begins. This should improve
3688 compatibility with X11 enabled applications run as systemd
3691 * Generators are now subject to masking via /etc and /run, the
3692 same way as unit files.
3694 * networkd .network files gained support for configuring
3695 per-link IPv4/IPv6 packet forwarding as well as IPv4
3696 masquerading. This is by default turned on for veth links to
3697 containers, as registered by systemd-nspawn. This means that
3698 nspawn containers run with --network-veth will now get
3699 automatic routed access to the host's networks without any
3700 further configuration or setup, as long as networkd runs on
3703 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3704 or UDP posts of a container on the host. With this in place
3705 it is possible to run containers with private veth links
3706 (--network-veth), and have their functionality exposed on
3707 the host as if their services were running directly on the
3710 * systemd-nspawn's --network-veth switch now gained a short
3711 version "-n", since with the changes above it is now truly
3712 useful out-of-the-box. The systemd-nspawn@.service has been
3713 updated to make use of it too by default.
3715 * systemd-nspawn will now maintain a per-image R/W lock, to
3716 ensure that the same image is not started more than once
3717 writable. (It's OK to run an image multiple times
3718 simultaneously in read-only mode.)
3720 * systemd-nspawn's --image= option is now capable of
3721 dissecting and booting MBR and GPT disk images that contain
3722 only a single active Linux partition. Previously it
3723 supported only GPT disk images with proper GPT type
3724 IDs. This allows running cloud images from major
3725 distributions directly with systemd-nspawn, without
3728 * In addition to collecting mouse dpi data in the udev
3729 hardware database, there's now support for collecting angle
3730 information for mouse scroll wheels. The database is
3731 supposed to guarantee similar scrolling behavior on mice
3732 that it knows about. There's also support for collecting
3733 information about Touchpad types.
3735 * udev's input_id built-in will now also collect touch screen
3736 dimension data and attach it to probed devices.
3738 * /etc/os-release gained support for a Distribution Privacy
3741 * networkd gained support for creating "ipvlan", "gretap",
3742 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3744 * systemd-tmpfiles gained support for "a" lines for setting
3747 * systemd-nspawn will now mount /tmp in the container to
3748 tmpfs, automatically.
3750 * systemd now exposes the memory.usage_in_bytes cgroup
3751 attribute and shows it for each service in the "systemctl
3752 status" output, if available.
3754 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3755 immediate reboot is triggered. This useful if shutdown is
3756 hung and is unable to complete, to expedite the
3757 operation. Note that this kind of reboot will still unmount
3758 all file systems, and hence should not result in fsck being
3761 * A .device unit for an optical block device will now be
3762 considered active only when a medium is in the drive. Also,
3763 mount units are now bound to their backing devices thus
3764 triggering automatic unmounting when devices become
3765 unavailable. With this in place systemd will now
3766 automatically unmount left-over mounts when a CD-ROM is
3767 ejected or an USB stick is yanked from the system.
3769 * networkd-wait-online now has support for waiting for
3770 specific interfaces only (with globbing), and for giving up
3771 after a configurable timeout.
3773 * networkd now exits when idle. It will be automatically
3774 restarted as soon as interfaces show up, are removed or
3775 change state. networkd will stay around as long as there is
3776 at least one DHCP state machine or similar around, that keep
3779 * networkd may now configure IPv6 link-local addressing in
3780 addition to IPv4 link-local addressing.
3782 * The IPv6 "token" for use in SLAAC may now be configured for
3783 each .network interface in networkd.
3785 * Routes configured with networkd may now be assigned a scope
3788 * networkd's [Match] sections now support globbing and lists
3789 of multiple space-separated matches per item.
3791 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3792 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3793 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3794 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3795 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3796 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3797 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3798 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3799 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3800 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3801 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3802 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3803 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3804 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3805 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3806 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3807 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3808 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3809 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3810 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3811 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3812 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3813 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3814 Hoffmann, Zbigniew Jędrzejewski-Szmek
3816 — Berlin, 2015-02-16
3820 * When querying unit file enablement status (for example via
3821 "systemctl is-enabled"), a new state "indirect" is now known
3822 which indicates that a unit might not be enabled itself, but
3823 another unit listed in its Also= setting might be.
3825 * Similar to the various existing ConditionXYZ= settings for
3826 units, there are now matching AssertXYZ= settings. While
3827 failing conditions cause a unit to be skipped, but its job
3828 to succeed, failing assertions declared like this will cause
3829 a unit start operation and its job to fail.
3831 * hostnamed now knows a new chassis type "embedded".
3833 * systemctl gained a new "edit" command. When used on a unit
3834 file, this allows extending unit files with .d/ drop-in
3835 configuration snippets or editing the full file (after
3836 copying it from /usr/lib to /etc). This will invoke the
3837 user's editor (as configured with $EDITOR), and reload the
3838 modified configuration after editing.
3840 * "systemctl status" now shows the suggested enablement state
3841 for a unit, as declared in the (usually vendor-supplied)
3842 system preset files.
3844 * nss-myhostname will now resolve the single-label host name
3845 "gateway" to the locally configured default IP routing
3846 gateways, ordered by their metrics. This assigns a stable
3847 name to the used gateways, regardless which ones are
3848 currently configured. Note that the name will only be
3849 resolved after all other name sources (if nss-myhostname is
3850 configured properly) and should hence not negatively impact
3851 systems that use the single-label host name "gateway" in
3854 * systemd-inhibit now allows filtering by mode when listing
3857 * Scope and service units gained a new "Delegate" boolean
3858 property, which, when set, allows processes running inside the
3859 unit to further partition resources. This is primarily
3860 useful for systemd user instances as well as container
3863 * journald will now pick up audit messages directly from
3864 the kernel, and log them like any other log message. The
3865 audit fields are split up and fully indexed. This means that
3866 journalctl in many ways is now a (nicer!) alternative to
3867 ausearch, the traditional audit client. Note that this
3868 implements only a minimal audit client. If you want the
3869 special audit modes like reboot-on-log-overflow, please use
3870 the traditional auditd instead, which can be used in
3871 parallel to journald.
3873 * The ConditionSecurity= unit file option now understands the
3874 special string "audit" to check whether auditing is
3877 * journalctl gained two new commands --vacuum-size= and
3878 --vacuum-time= to delete old journal files until the
3879 remaining ones take up no more than the specified size on disk,
3880 or are not older than the specified time.
3882 * A new, native PPPoE library has been added to sd-network,
3883 systemd's library of light-weight networking protocols. This
3884 library will be used in a future version of networkd to
3885 enable PPPoE communication without an external pppd daemon.
3887 * The busctl tool now understands a new "capture" verb that
3888 works similar to "monitor", but writes a packet capture
3889 trace to STDOUT that can be redirected to a file which is
3890 compatible with libcap's capture file format. This can then
3891 be loaded in Wireshark and similar tools to inspect bus
3894 * The busctl tool now understands a new "tree" verb that shows
3895 the object trees of a specific service on the bus, or of all
3898 * The busctl tool now understands a new "introspect" verb that
3899 shows all interfaces and members of objects on the bus,
3900 including their signature and values. This is particularly
3901 useful to get more information about bus objects shown by
3902 the new "busctl tree" command.
3904 * The busctl tool now understands new verbs "call",
3905 "set-property" and "get-property" for invoking bus method
3906 calls, setting and getting bus object properties in a
3909 * busctl gained a new --augment-creds= argument that controls
3910 whether the tool shall augment credential information it
3911 gets from the bus with data from /proc, in a possibly
3914 * nspawn's --link-journal= switch gained two new values
3915 "try-guest" and "try-host" that work like "guest" and
3916 "host", but do not fail if the host has no persistent
3917 journalling enabled. -j is now equivalent to
3918 --link-journal=try-guest.
3920 * macvlan network devices created by nspawn will now have
3921 stable MAC addresses.
3923 * A new SmackProcessLabel= unit setting has been added, which
3924 controls the SMACK security label processes forked off by
3925 the respective unit shall use.
3927 * If compiled with --enable-xkbcommon, systemd-localed will
3928 verify x11 keymap settings by compiling the given keymap. It
3929 will spew out warnings if the compilation fails. This
3930 requires libxkbcommon to be installed.
3932 * When a coredump is collected, a larger number of metadata
3933 fields is now collected and included in the journal records
3934 created for it. More specifically, control group membership,
3935 environment variables, memory maps, working directory,
3936 chroot directory, /proc/$PID/status, and a list of open file
3937 descriptors is now stored in the log entry.
3939 * The udev hwdb now contains DPI information for mice. For
3942 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3944 * All systemd programs that read standalone configuration
3945 files in /etc now also support a corresponding series of
3946 .conf.d configuration directories in /etc/, /run/,
3947 /usr/local/lib/, /usr/lib/, and (if configured with
3948 --enable-split-usr) /lib/. In particular, the following
3949 configuration files now have corresponding configuration
3950 directories: system.conf user.conf, logind.conf,
3951 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3952 resolved.conf, timesyncd.conf, journal-remote.conf, and
3953 journal-upload.conf. Note that distributions should use the
3954 configuration directories in /usr/lib/; the directories in
3955 /etc/ are reserved for the system administrator.
3957 * systemd-rfkill will no longer take the rfkill device name
3958 into account when storing rfkill state on disk, as the name
3959 might be dynamically assigned and not stable. Instead, the
3960 ID_PATH udev variable combined with the rfkill type (wlan,
3961 bluetooth, ...) is used.
3963 * A new service systemd-machine-id-commit.service has been
3964 added. When used on systems where /etc is read-only during
3965 boot, and /etc/machine-id is not initialized (but an empty
3966 file), this service will copy the temporary machine ID
3967 created as replacement into /etc after the system is fully
3968 booted up. This is useful for systems that are freshly
3969 installed with a non-initialized machine ID, but should get
3970 a fixed machine ID for subsequent boots.
3972 * networkd's .netdev files now provide a large set of
3973 configuration parameters for VXLAN devices. Similarly, the
3974 bridge port cost parameter is now configurable in .network
3975 files. There's also new support for configuring IP source
3976 routing. networkd .link files gained support for a new
3977 OriginalName= match that is useful to match against the
3978 original interface name the kernel assigned. .network files
3979 may include MTU= and MACAddress= fields for altering the MTU
3980 and MAC address while being connected to a specific network
3983 * The LUKS logic gained supported for configuring
3984 UUID-specific key files. There's also new support for naming
3985 LUKS device from the kernel command line, using the new
3986 luks.name= argument.
3988 * Timer units may now be transiently created via the bus API
3989 (this was previously already available for scope and service
3990 units). In addition it is now possible to create multiple
3991 transient units at the same time with a single bus call. The
3992 "systemd-run" tool has been updated to make use of this for
3993 running commands on a specified time, in at(1)-style.
3995 * tmpfiles gained support for "t" lines, for assigning
3996 extended attributes to files. Among other uses this may be
3997 used to assign SMACK labels to files.
3999 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4000 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4001 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4002 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4003 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4004 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4005 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4006 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4007 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4008 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4009 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4010 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4011 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4012 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4013 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4014 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4015 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4016 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4018 — Berlin, 2014-12-10
4022 * journalctl gained the new options -t/--identifier= to match
4023 on the syslog identifier (aka "tag"), as well as --utc to
4024 show log timestamps in the UTC timezone. journalctl now also
4025 accepts -n/--lines=all to disable line capping in a pager.
4027 * journalctl gained a new switch, --flush, that synchronously
4028 flushes logs from /run/log/journal to /var/log/journal if
4029 persistent storage is enabled. systemd-journal-flush.service
4030 now waits until the operation is complete.
4032 * Services can notify the manager before they start a reload
4033 (by sending RELOADING=1) or shutdown (by sending
4034 STOPPING=1). This allows the manager to track and show the
4035 internal state of daemons and closes a race condition when
4036 the process is still running but has closed its D-Bus
4039 * Services with Type=oneshot do not have to have any ExecStart
4042 * User units are now loaded also from
4043 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4044 /run/systemd/user directory that was already previously
4045 supported, but is under the control of the user.
4047 * Job timeouts (i.e. time-outs on the time a job that is
4048 queued stays in the run queue) can now optionally result in
4049 immediate reboot or power-off actions (JobTimeoutAction= and
4050 JobTimeoutRebootArgument=). This is useful on ".target"
4051 units, to limit the maximum time a target remains
4052 undispatched in the run queue, and to trigger an emergency
4053 operation in such a case. This is now used by default to
4054 turn off the system if boot-up (as defined by everything in
4055 basic.target) hangs and does not complete for at least
4056 15min. Also, if power-off or reboot hang for at least 30min
4057 an immediate power-off/reboot operation is triggered. This
4058 functionality is particularly useful to increase reliability
4059 on embedded devices, but also on laptops which might
4060 accidentally get powered on when carried in a backpack and
4061 whose boot stays stuck in a hard disk encryption passphrase
4064 * systemd-logind can be configured to also handle lid switch
4065 events even when the machine is docked or multiple displays
4066 are attached (HandleLidSwitchDocked= option).
4068 * A helper binary and a service have been added which can be
4069 used to resume from hibernation in the initramfs. A
4070 generator will parse the resume= option on the kernel
4071 command line to trigger resume.
4073 * A user console daemon systemd-consoled has been
4074 added. Currently, it is a preview, and will so far open a
4075 single terminal on each session of the user marked as
4076 Desktop=systemd-console.
4078 * Route metrics can be specified for DHCP routes added by
4081 * The SELinux context of socket-activated services can be set
4082 from the information provided by the networking stack
4083 (SELinuxContextFromNet= option).
4085 * Userspace firmware loading support has been removed and
4086 the minimum supported kernel version is thus bumped to 3.7.
4088 * Timeout for udev workers has been increased from 1 to 3
4089 minutes, but a warning will be printed after 1 minute to
4090 help diagnose kernel modules that take a long time to load.
4092 * Udev rules can now remove tags on devices with TAG-="foobar".
4094 * systemd's readahead implementation has been removed. In many
4095 circumstances it didn't give expected benefits even for
4096 rotational disk drives and was becoming less relevant in the
4097 age of SSDs. As none of the developers has been using
4098 rotating media anymore, and nobody stepped up to actively
4099 maintain this component of systemd it has now been removed.
4101 * Swap units can use Options= to specify discard options.
4102 Discard options specified for swaps in /etc/fstab are now
4105 * Docker containers are now detected as a separate type of
4108 * The Password Agent protocol gained support for queries where
4109 the user input is shown, useful e.g. for user names.
4110 systemd-ask-password gained a new --echo option to turn that
4113 * The default sysctl.d/ snippets will now set:
4115 net.core.default_qdisc = fq_codel
4117 This selects Fair Queuing Controlled Delay as the default
4118 queuing discipline for network interfaces. fq_codel helps
4119 fight the network bufferbloat problem. It is believed to be
4120 a good default with no tuning required for most workloads.
4121 Downstream distributions may override this choice. On 10Gbit
4122 servers that do not do forwarding, "fq" may perform better.
4123 Systems without a good clocksource should use "pfifo_fast".
4125 * If kdbus is enabled during build a new option BusPolicy= is
4126 available for service units, that allows locking all service
4127 processes into a stricter bus policy, in order to limit
4128 access to various bus services, or even hide most of them
4129 from the service's view entirely.
4131 * networkctl will now show the .network and .link file
4132 networkd has applied to a specific interface.
4134 * sd-login gained a new API call sd_session_get_desktop() to
4135 query which desktop environment has been selected for a
4138 * UNIX utmp support is now compile-time optional to support
4139 legacy-free systems.
4141 * systemctl gained two new commands "add-wants" and
4142 "add-requires" for pulling in units from specific targets
4145 * If the word "rescue" is specified on the kernel command line
4146 the system will now boot into rescue mode (aka
4147 rescue.target), which was previously available only by
4148 specifying "1" or "systemd.unit=rescue.target" on the kernel
4149 command line. This new kernel command line option nicely
4150 mirrors the already existing "emergency" kernel command line
4153 * New kernel command line options mount.usr=, mount.usrflags=,
4154 mount.usrfstype= have been added that match root=, rootflags=,
4155 rootfstype= but allow mounting a specific file system to
4158 * The $NOTIFY_SOCKET is now also passed to control processes of
4159 services, not only the main process.
4161 * This version reenables support for fsck's -l switch. This
4162 means at least version v2.25 of util-linux is required for
4163 operation, otherwise dead-locks on device nodes may
4164 occur. Again: you need to update util-linux to at least
4165 v2.25 when updating systemd to v217.
4167 * The "multi-seat-x" tool has been removed from systemd, as
4168 its functionality has been integrated into X servers 1.16,
4169 and the tool is hence redundant. It is recommended to update
4170 display managers invoking this tool to simply invoke X
4171 directly from now on, again.
4173 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4174 message flag has been added for all of systemd's PolicyKit
4175 authenticated method calls has been added. In particular
4176 this now allows optional interactive authorization via
4177 PolicyKit for many of PID1's privileged operations such as
4178 unit file enabling and disabling.
4180 * "udevadm hwdb --update" learnt a new switch "--usr" for
4181 placing the rebuilt hardware database in /usr instead of
4182 /etc. When used only hardware database entries stored in
4183 /usr will be used, and any user database entries in /etc are
4184 ignored. This functionality is useful for vendors to ship a
4185 pre-built database on systems where local configuration is
4186 unnecessary or unlikely.
4188 * Calendar time specifications in .timer units now also
4189 understand the strings "semi-annually", "quarterly" and
4190 "minutely" as shortcuts (in addition to the preexisting
4191 "anually", "hourly", ...).
4193 * systemd-tmpfiles will now correctly create files in /dev
4194 at boot which are marked for creation only at boot. It is
4195 recommended to always create static device nodes with 'c!'
4196 and 'b!', so that they are created only at boot and not
4197 overwritten at runtime.
4199 * When the watchdog logic is used for a service (WatchdogSec=)
4200 and the watchdog timeout is hit the service will now be
4201 terminated with SIGABRT (instead of just SIGTERM), in order
4202 to make sure a proper coredump and backtrace is
4203 generated. This ensures that hanging services will result in
4204 similar coredump/backtrace behaviour as services that hit a
4207 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4208 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4209 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4210 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4211 Herrmann, David Sommerseth, David Strauss, Emil Renner
4212 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4213 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4214 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4215 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4216 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4217 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4218 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4219 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4220 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4221 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4222 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4223 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4224 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4225 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4226 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4227 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4230 — Berlin, 2014-10-28
4234 * timedated no longer reads NTP implementation unit names from
4235 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4236 implementations should add a
4238 Conflicts=systemd-timesyncd.service
4240 to their unit files to take over and replace systemd's NTP
4241 default functionality.
4243 * systemd-sysusers gained a new line type "r" for configuring
4244 which UID/GID ranges to allocate system users/groups
4245 from. Lines of type "u" may now add an additional column
4246 that specifies the home directory for the system user to be
4247 created. Also, systemd-sysusers may now optionally read user
4248 information from STDIN instead of a file. This is useful for
4249 invoking it from RPM preinst scriptlets that need to create
4250 users before the first RPM file is installed since these
4251 files might need to be owned by them. A new
4252 %sysusers_create_inline RPM macro has been introduced to do
4253 just that. systemd-sysusers now updates the shadow files as
4254 well as the user/group databases, which should enhance
4255 compatibility with certain tools like grpck.
4257 * A number of bus APIs of PID 1 now optionally consult
4258 PolicyKit to permit access for otherwise unprivileged
4259 clients under certain conditions. Note that this currently
4260 doesn't support interactive authentication yet, but this is
4261 expected to be added eventually, too.
4263 * /etc/machine-info now has new fields for configuring the
4264 deployment environment of the machine, as well as the
4265 location of the machine. hostnamectl has been updated with
4266 new command to update these fields.
4268 * systemd-timesyncd has been updated to automatically acquire
4269 NTP server information from systemd-networkd, which might
4270 have been discovered via DHCP.
4272 * systemd-resolved now includes a caching DNS stub resolver
4273 and a complete LLMNR name resolution implementation. A new
4274 NSS module "nss-resolve" has been added which can be used
4275 instead of glibc's own "nss-dns" to resolve hostnames via
4276 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4277 be resolved via systemd-resolved D-Bus APIs. In contrast to
4278 the glibc internal resolver systemd-resolved is aware of
4279 multi-homed system, and keeps DNS server and caches separate
4280 and per-interface. Queries are sent simultaneously on all
4281 interfaces that have DNS servers configured, in order to
4282 properly handle VPNs and local LANs which might resolve
4283 separate sets of domain names. systemd-resolved may acquire
4284 DNS server information from systemd-networkd automatically,
4285 which in turn might have discovered them via DHCP. A tool
4286 "systemd-resolve-host" has been added that may be used to
4287 query the DNS logic in resolved. systemd-resolved implements
4288 IDNA and automatically uses IDNA or UTF-8 encoding depending
4289 on whether classic DNS or LLMNR is used as transport. In the
4290 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4291 implementation to systemd-resolved.
4293 * A new NSS module nss-mymachines has been added, that
4294 automatically resolves the names of all local registered
4295 containers to their respective IP addresses.
4297 * A new client tool "networkctl" for systemd-networkd has been
4298 added. It currently is entirely passive and will query
4299 networking configuration from udev, rtnetlink and networkd,
4300 and present it to the user in a very friendly
4301 way. Eventually, we hope to extend it to become a full
4302 control utility for networkd.
4304 * .socket units gained a new DeferAcceptSec= setting that
4305 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4306 TCP. Similarly, support for controlling TCP keep-alive
4307 settings has been added (KeepAliveTimeSec=,
4308 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4309 turning off Nagle's algorithm on TCP has been added
4312 * logind learned a new session type "web", for use in projects
4313 like Cockpit which register web clients as PAM sessions.
4315 * timer units with at least one OnCalendar= setting will now
4316 be started only after time-sync.target has been
4317 reached. This way they will not elapse before the system
4318 clock has been corrected by a local NTP client or
4319 similar. This is particular useful on RTC-less embedded
4320 machines, that come up with an invalid system clock.
4322 * systemd-nspawn's --network-veth= switch should now result in
4323 stable MAC addresses for both the outer and the inner side
4326 * systemd-nspawn gained a new --volatile= switch for running
4327 container instances with /etc or /var unpopulated.
4329 * The kdbus client code has been updated to use the new Linux
4330 3.17 memfd subsystem instead of the old kdbus-specific one.
4332 * systemd-networkd's DHCP client and server now support
4333 FORCERENEW. There are also new configuration options to
4334 configure the vendor client identifier and broadcast mode
4337 * systemd will no longer inform the kernel about the current
4338 timezone, as this is necessarily incorrect and racy as the
4339 kernel has no understanding of DST and similar
4340 concepts. This hence means FAT timestamps will be always
4341 considered UTC, similar to what Android is already
4342 doing. Also, when the RTC is configured to the local time
4343 (rather than UTC) systemd will never synchronize back to it,
4344 as this might confuse Windows at a later boot.
4346 * systemd-analyze gained a new command "verify" for offline
4347 validation of unit files.
4349 * systemd-networkd gained support for a couple of additional
4350 settings for bonding networking setups. Also, the metric for
4351 statically configured routes may now be configured. For
4352 network interfaces where this is appropriate the peer IP
4353 address may now be configured.
4355 * systemd-networkd's DHCP client will no longer request
4356 broadcasting by default, as this tripped up some networks.
4357 For hardware where broadcast is required the feature should
4358 be switched back on using RequestBroadcast=yes.
4360 * systemd-networkd will now set up IPv4LL addresses (when
4361 enabled) even if DHCP is configured successfully.
4363 * udev will now default to respect network device names given
4364 by the kernel when the kernel indicates that these are
4365 predictable. This behavior can be tweaked by changing
4366 NamePolicy= in the relevant .link file.
4368 * A new library systemd-terminal has been added that
4369 implements full TTY stream parsing and rendering. This
4370 library is supposed to be used later on for implementing a
4371 full userspace VT subsystem, replacing the current kernel
4374 * A new tool systemd-journal-upload has been added to push
4375 journal data to a remote system running
4376 systemd-journal-remote.
4378 * journald will no longer forward all local data to another
4379 running syslog daemon. This change has been made because
4380 rsyslog (which appears to be the most commonly used syslog
4381 implementation these days) no longer makes use of this, and
4382 instead pulls the data out of the journal on its own. Since
4383 forwarding the messages to a non-existent syslog server is
4384 more expensive than we assumed we have now turned this
4385 off. If you run a syslog server that is not a recent rsyslog
4386 version, you have to turn this option on again
4387 (ForwardToSyslog= in journald.conf).
4389 * journald now optionally supports the LZ4 compressor for
4390 larger journal fields. This compressor should perform much
4391 better than XZ which was the previous default.
4393 * machinectl now shows the IP addresses of local containers,
4394 if it knows them, plus the interface name of the container.
4396 * A new tool "systemd-escape" has been added that makes it
4397 easy to escape strings to build unit names and similar.
4399 * sd_notify() messages may now include a new ERRNO= field
4400 which is parsed and collected by systemd and shown among the
4401 "systemctl status" output for a service.
4403 * A new component "systemd-firstboot" has been added that
4404 queries the most basic systemd information (timezone,
4405 hostname, root password) interactively on first
4406 boot. Alternatively it may also be used to provision these
4407 things offline on OS images installed into directories.
4409 * The default sysctl.d/ snippets will now set
4411 net.ipv4.conf.default.promote_secondaries=1
4413 This has the benefit of no flushing secondary IP addresses
4414 when primary addresses are removed.
4416 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4417 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4418 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4419 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4420 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4421 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4422 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4423 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4424 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4425 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4426 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4427 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4428 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4429 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4430 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4432 — Berlin, 2014-08-19
4436 * A new tool systemd-sysusers has been added. This tool
4437 creates system users and groups in /etc/passwd and
4438 /etc/group, based on static declarative system user/group
4439 definitions in /usr/lib/sysusers.d/. This is useful to
4440 enable factory resets and volatile systems that boot up with
4441 an empty /etc directory, and thus need system users and
4442 groups created during early boot. systemd now also ships
4443 with two default sysusers.d/ files for the most basic
4444 users and groups systemd and the core operating system
4447 * A new tmpfiles snippet has been added that rebuilds the
4448 essential files in /etc on boot, should they be missing.
4450 * A directive for ensuring automatic clean-up of
4451 /var/cache/man/ has been removed from the default
4452 configuration. This line should now be shipped by the man
4453 implementation. The necessary change has been made to the
4454 man-db implementation. Note that you need to update your man
4455 implementation to one that ships this line, otherwise no
4456 automatic clean-up of /var/cache/man will take place.
4458 * A new condition ConditionNeedsUpdate= has been added that
4459 may conditionalize services to only run when /etc or /var
4460 are "older" than the vendor operating system resources in
4461 /usr. This is useful for reconstructing or updating /etc
4462 after an offline update of /usr or a factory reset, on the
4463 next reboot. Services that want to run once after such an
4464 update or reset should use this condition and order
4465 themselves before the new systemd-update-done.service, which
4466 will mark the two directories as fully updated. A number of
4467 service files have been added making use of this, to rebuild
4468 the udev hardware database, the journald message catalog and
4469 dynamic loader cache (ldconfig). The systemd-sysusers tool
4470 described above also makes use of this now. With this in
4471 place it is now possible to start up a minimal operating
4472 system with /etc empty cleanly. For more information on the
4473 concepts involved see this recent blog story:
4475 http://0pointer.de/blog/projects/stateless.html
4477 * A new system group "input" has been introduced, and all
4478 input device nodes get this group assigned. This is useful
4479 for system-level software to get access to input devices. It
4480 complements what is already done for "audio" and "video".
4482 * systemd-networkd learnt minimal DHCPv4 server support in
4483 addition to the existing DHCPv4 client support. It also
4484 learnt DHCPv6 client and IPv6 Router Solicitation client
4485 support. The DHCPv4 client gained support for static routes
4486 passed in from the server. Note that the [DHCPv4] section
4487 known in older systemd-networkd versions has been renamed to
4488 [DHCP] and is now also used by the DHCPv6 client. Existing
4489 .network files using settings of this section should be
4490 updated, though compatibility is maintained. Optionally, the
4491 client hostname may now be sent to the DHCP server.
4493 * networkd gained support for vxlan virtual networks as well
4494 as tun/tap and dummy devices.
4496 * networkd gained support for automatic allocation of address
4497 ranges for interfaces from a system-wide pool of
4498 addresses. This is useful for dynamically managing a large
4499 number of interfaces with a single network configuration
4500 file. In particular this is useful to easily assign
4501 appropriate IP addresses to the veth links of a large number
4502 of nspawn instances.
4504 * RPM macros for processing sysusers, sysctl and binfmt
4505 drop-in snippets at package installation time have been
4508 * The /etc/os-release file should now be placed in
4509 /usr/lib/os-release. The old location is automatically
4510 created as symlink. /usr/lib is the more appropriate
4511 location of this file, since it shall actually describe the
4512 vendor operating system shipped in /usr, and not the
4513 configuration stored in /etc.
4515 * .mount units gained a new boolean SloppyOptions= setting
4516 that maps to mount(8)'s -s option which enables permissive
4517 parsing of unknown mount options.
4519 * tmpfiles learnt a new "L+" directive which creates a symlink
4520 but (unlike "L") deletes a pre-existing file first, should
4521 it already exist and not already be the correct
4522 symlink. Similarly, "b+", "c+" and "p+" directives have been
4523 added as well, which create block and character devices, as
4524 well as fifos in the filesystem, possibly removing any
4525 pre-existing files of different types.
4527 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4528 'argument' field (which so far specified the source to
4529 symlink/copy the files from) is now optional. If omitted the
4530 same file os copied from /usr/share/factory/ suffixed by the
4531 full destination path. This is useful for populating /etc
4532 with essential files, by copying them from vendor defaults
4533 shipped in /usr/share/factory/etc.
4535 * A new command "systemctl preset-all" has been added that
4536 applies the service preset settings to all installed unit
4537 files. A new switch --preset-mode= has been added that
4538 controls whether only enable or only disable operations
4541 * A new command "systemctl is-system-running" has been added
4542 that allows checking the overall state of the system, for
4543 example whether it is fully up and running.
4545 * When the system boots up with an empty /etc, the equivalent
4546 to "systemctl preset-all" is executed during early boot, to
4547 make sure all default services are enabled after a factory
4550 * systemd now contains a minimal preset file that enables the
4551 most basic services systemd ships by default.
4553 * Unit files' [Install] section gained a new DefaultInstance=
4554 field for defining the default instance to create if a
4555 template unit is enabled with no instance specified.
4557 * A new passive target cryptsetup-pre.target has been added
4558 that may be used by services that need to make they run and
4559 finish before the first LUKS cryptographic device is set up.
4561 * The /dev/loop-control and /dev/btrfs-control device nodes
4562 are now owned by the "disk" group by default, opening up
4563 access to this group.
4565 * systemd-coredump will now automatically generate a
4566 stack trace of all core dumps taking place on the system,
4567 based on elfutils' libdw library. This stack trace is logged
4570 * systemd-coredump may now optionally store coredumps directly
4571 on disk (in /var/lib/systemd/coredump, possibly compressed),
4572 instead of storing them unconditionally in the journal. This
4573 mode is the new default. A new configuration file
4574 /etc/systemd/coredump.conf has been added to configure this
4575 and other parameters of systemd-coredump.
4577 * coredumpctl gained a new "info" verb to show details about a
4578 specific coredump. A new switch "-1" has also been added
4579 that makes sure to only show information about the most
4580 recent entry instead of all entries. Also, as the tool is
4581 generally useful now the "systemd-" prefix of the binary
4582 name has been removed. Distributions that want to maintain
4583 compatibility with the old name should add a symlink from
4584 the old name to the new name.
4586 * journald's SplitMode= now defaults to "uid". This makes sure
4587 that unprivileged users can access their own coredumps with
4588 coredumpctl without restrictions.
4590 * New kernel command line options "systemd.wants=" (for
4591 pulling an additional unit during boot), "systemd.mask="
4592 (for masking a specific unit for the boot), and
4593 "systemd.debug-shell" (for enabling the debug shell on tty9)
4594 have been added. This is implemented in the new generator
4595 "systemd-debug-generator".
4597 * systemd-nspawn will now by default filter a couple of
4598 syscalls for containers, among them those required for
4599 kernel module loading, direct x86 IO port access, swap
4600 management, and kexec. Most importantly though
4601 open_by_handle_at() is now prohibited for containers,
4602 closing a hole similar to a recently discussed vulnerability
4603 in docker regarding access to files on file hierarchies the
4604 container should normally not have access to. Note that, for
4605 nspawn, we generally make no security claims anyway (and
4606 this is explicitly documented in the man page), so this is
4607 just a fix for one of the most obvious problems.
4609 * A new man page file-hierarchy(7) has been added that
4610 contains a minimized, modernized version of the file system
4611 layout systemd expects, similar in style to the FHS
4612 specification or hier(5). A new tool systemd-path(1) has
4613 been added to query many of these paths for the local
4616 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4617 longer done. Since the directory now has a per-user size
4618 limit, and is cleaned on logout this appears unnecessary,
4619 in particular since this now brings the lifecycle of this
4620 directory closer in line with how IPC objects are handled.
4622 * systemd.pc now exports a number of additional directories,
4623 including $libdir (which is useful to identify the library
4624 path for the primary architecture of the system), and a
4625 couple of drop-in directories.
4627 * udev's predictable network interface names now use the dev_port
4628 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4629 distinguish between ports of the same PCI function. dev_id should
4630 only be used for ports using the same HW address, hence the need
4633 * machined has been updated to export the OS version of a
4634 container (read from /etc/os-release and
4635 /usr/lib/os-release) on the bus. This is now shown in
4636 "machinectl status" for a machine.
4638 * A new service setting RestartForceExitStatus= has been
4639 added. If configured to a set of exit signals or process
4640 return values, the service will be restarted when the main
4641 daemon process exits with any of them, regardless of the
4644 * systemctl's -H switch for connecting to remote systemd
4645 machines has been extended so that it may be used to
4646 directly connect to a specific container on the
4647 host. "systemctl -H root@foobar:waldi" will now connect as
4648 user "root" to host "foobar", and then proceed directly to
4649 the container named "waldi". Note that currently you have to
4650 authenticate as user "root" for this to work, as entering
4651 containers is a privileged operation.
4653 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4654 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4655 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4656 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4657 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4658 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4659 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4660 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4661 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4662 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4663 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4664 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4666 — Berlin, 2014-07-03
4670 * As an experimental feature, udev now tries to lock the
4671 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4672 executes events for the disk or any of its partitions.
4673 Applications like partitioning programs can lock the
4674 disk device node (flock(LOCK_EX)) and claim temporary
4675 device ownership that way; udev will entirely skip all event
4676 handling for this disk and its partitions. If the disk
4677 was opened for writing, the close will trigger a partition
4678 table rescan in udev's "watch" facility, and if needed
4679 synthesize "change" events for the disk and all its partitions.
4680 This is now unconditionally enabled, and if it turns out to
4681 cause major problems, we might turn it on only for specific
4682 devices, or might need to disable it entirely. Device Mapper
4683 devices are excluded from this logic.
4685 * We temporarily dropped the "-l" switch for fsck invocations,
4686 since they collide with the flock() logic above. util-linux
4687 upstream has been changed already to avoid this conflict,
4688 and we will readd "-l" as soon as util-linux with this
4689 change has been released.
4691 * The dependency on libattr has been removed. Since a long
4692 time, the extended attribute calls have moved to glibc, and
4693 libattr is thus unnecessary.
4695 * Virtualization detection works without privileges now. This
4696 means the systemd-detect-virt binary no longer requires
4697 CAP_SYS_PTRACE file capabilities, and our daemons can run
4698 with fewer privileges.
4700 * systemd-networkd now runs under its own "systemd-network"
4701 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4702 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4703 loses the ability to write to files owned by root this way.
4705 * Similarly, systemd-resolved now runs under its own
4706 "systemd-resolve" user with no capabilities remaining.
4708 * Similarly, systemd-bus-proxyd now runs under its own
4709 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4711 * systemd-networkd gained support for setting up "veth"
4712 virtual Ethernet devices for container connectivity, as well
4713 as GRE and VTI tunnels.
4715 * systemd-networkd will no longer automatically attempt to
4716 manually load kernel modules necessary for certain tunnel
4717 transports. Instead, it is assumed the kernel loads them
4718 automatically when required. This only works correctly on
4719 very new kernels. On older kernels, please consider adding
4720 the kernel modules to /etc/modules-load.d/ as a work-around.
4722 * The resolv.conf file systemd-resolved generates has been
4723 moved to /run/systemd/resolve/. If you have a symlink from
4724 /etc/resolv.conf, it might be necessary to correct it.
4726 * Two new service settings, ProtectHome= and ProtectSystem=,
4727 have been added. When enabled, they will make the user data
4728 (such as /home) inaccessible or read-only and the system
4729 (such as /usr) read-only, for specific services. This allows
4730 very light-weight per-service sandboxing to avoid
4731 modifications of user data or system files from
4732 services. These two new switches have been enabled for all
4733 of systemd's long-running services, where appropriate.
4735 * Socket units gained new SocketUser= and SocketGroup=
4736 settings to set the owner user and group of AF_UNIX sockets
4737 and FIFOs in the file system.
4739 * Socket units gained a new RemoveOnStop= setting. If enabled,
4740 all FIFOS and sockets in the file system will be removed
4741 when the specific socket unit is stopped.
4743 * Socket units gained a new Symlinks= setting. It takes a list
4744 of symlinks to create to file system sockets or FIFOs
4745 created by the specific Unix sockets. This is useful to
4746 manage symlinks to socket nodes with the same life-cycle as
4749 * The /dev/log socket and /dev/initctl FIFO have been moved to
4750 /run, and have been replaced by symlinks. This allows
4751 connecting to these facilities even if PrivateDevices=yes is
4752 used for a service (which makes /dev/log itself unavailable,
4753 but /run is left). This also has the benefit of ensuring
4754 that /dev only contains device nodes, directories and
4755 symlinks, and nothing else.
4757 * sd-daemon gained two new calls sd_pid_notify() and
4758 sd_pid_notifyf(). They are similar to sd_notify() and
4759 sd_notifyf(), but allow overriding of the source PID of
4760 notification messages if permissions permit this. This is
4761 useful to send notify messages on behalf of a different
4762 process (for example, the parent process). The
4763 systemd-notify tool has been updated to make use of this
4764 when sending messages (so that notification messages now
4765 originate from the shell script invoking systemd-notify and
4766 not the systemd-notify process itself. This should minimize
4767 a race where systemd fails to associate notification
4768 messages to services when the originating process already
4771 * A new "on-abnormal" setting for Restart= has been added. If
4772 set, it will result in automatic restarts on all "abnormal"
4773 reasons for a process to exit, which includes unclean
4774 signals, core dumps, timeouts and watchdog timeouts, but
4775 does not include clean and unclean exit codes or clean
4776 signals. Restart=on-abnormal is an alternative for
4777 Restart=on-failure for services that shall be able to
4778 terminate and avoid restarts on certain errors, by
4779 indicating so with an unclean exit code. Restart=on-failure
4780 or Restart=on-abnormal is now the recommended setting for
4781 all long-running services.
4783 * If the InaccessibleDirectories= service setting points to a
4784 mount point (or if there are any submounts contained within
4785 it), it is now attempted to completely unmount it, to make
4786 the file systems truly unavailable for the respective
4789 * The ReadOnlyDirectories= service setting and
4790 systemd-nspawn's --read-only parameter are now recursively
4791 applied to all submounts, too.
4793 * Mount units may now be created transiently via the bus APIs.
4795 * The support for SysV and LSB init scripts has been removed
4796 from the systemd daemon itself. Instead, it is now
4797 implemented as a generator that creates native systemd units
4798 from these scripts when needed. This enables us to remove a
4799 substantial amount of legacy code from PID 1, following the
4800 fact that many distributions only ship a very small number
4801 of LSB/SysV init scripts nowadays.
4803 * Privileged Xen (dom0) domains are not considered
4804 virtualization anymore by the virtualization detection
4805 logic. After all, they generally have unrestricted access to
4806 the hardware and usually are used to manage the unprivileged
4809 * systemd-tmpfiles gained a new "C" line type, for copying
4810 files or entire directories.
4812 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4813 lines. So far, they have been non-globbing versions of the
4814 latter, and have thus been redundant. In future, it is
4815 recommended to only use "z". "m" has hence been removed
4816 from the documentation, even though it stays supported.
4818 * A tmpfiles snippet to recreate the most basic structure in
4819 /var has been added. This is enough to create the /var/run →
4820 /run symlink and create a couple of structural
4821 directories. This allows systems to boot up with an empty or
4822 volatile /var. Of course, while with this change, the core OS
4823 now is capable with dealing with a volatile /var, not all
4824 user services are ready for it. However, we hope that sooner
4825 or later, many service daemons will be changed upstream so
4826 that they are able to automatically create their necessary
4827 directories in /var at boot, should they be missing. This is
4828 the first step to allow state-less systems that only require
4829 the vendor image for /usr to boot.
4831 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4832 empty tmpfs instance to a specific directory. This is
4833 particularly useful for making use of the automatic
4834 reconstruction of /var (see above), by passing --tmpfs=/var.
4836 * Access modes specified in tmpfiles snippets may now be
4837 prefixed with "~", which indicates that they shall be masked
4838 by whether the existing file or directory is currently
4839 writable, readable or executable at all. Also, if specified,
4840 the sgid/suid/sticky bits will be masked for all
4843 * A new passive target unit "network-pre.target" has been
4844 added which is useful for services that shall run before any
4845 network is configured, for example firewall scripts.
4847 * The "floppy" group that previously owned the /dev/fd*
4848 devices is no longer used. The "disk" group is now used
4849 instead. Distributions should probably deprecate usage of
4852 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4853 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4854 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4855 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4856 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4857 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4860 — Berlin, 2014-06-11
4864 * A new "systemd-timesyncd" daemon has been added for
4865 synchronizing the system clock across the network. It
4866 implements an SNTP client. In contrast to NTP
4867 implementations such as chrony or the NTP reference server,
4868 this only implements a client side, and does not bother with
4869 the full NTP complexity, focusing only on querying time from
4870 one remote server and synchronizing the local clock to
4871 it. Unless you intend to serve NTP to networked clients or
4872 want to connect to local hardware clocks, this simple NTP
4873 client should be more than appropriate for most
4874 installations. The daemon runs with minimal privileges, and
4875 has been hooked up with networkd to only operate when
4876 network connectivity is available. The daemon saves the
4877 current clock to disk every time a new NTP sync has been
4878 acquired, and uses this to possibly correct the system clock
4879 early at bootup, in order to accommodate for systems that
4880 lack an RTC such as the Raspberry Pi and embedded devices,
4881 and to make sure that time monotonically progresses on these
4882 systems, even if it is not always correct. To make use of
4883 this daemon, a new system user and group "systemd-timesync"
4884 needs to be created on installation of systemd.
4886 * The queue "seqnum" interface of libudev has been disabled, as
4887 it was generally incompatible with device namespacing as
4888 sequence numbers of devices go "missing" if the devices are
4889 part of a different namespace.
4891 * "systemctl list-timers" and "systemctl list-sockets" gained
4892 a --recursive switch for showing units of these types also
4893 for all local containers, similar in style to the already
4894 supported --recursive switch for "systemctl list-units".
4896 * A new RebootArgument= setting has been added for service
4897 units, which may be used to specify a kernel reboot argument
4898 to use when triggering reboots with StartLimitAction=.
4900 * A new FailureAction= setting has been added for service
4901 units which may be used to specify an operation to trigger
4902 when a service fails. This works similarly to
4903 StartLimitAction=, but unlike it, controls what is done
4904 immediately rather than only after several attempts to
4905 restart the service in question.
4907 * hostnamed got updated to also expose the kernel name,
4908 release, and version on the bus. This is useful for
4909 executing commands like hostnamectl with the -H switch.
4910 systemd-analyze makes use of this to properly display
4911 details when running non-locally.
4913 * The bootchart tool can now show cgroup information in the
4914 graphs it generates.
4916 * The CFS CPU quota cgroup attribute is now exposed for
4917 services. The new CPUQuota= switch has been added for this
4918 which takes a percentage value. Setting this will have the
4919 result that a service may never get more CPU time than the
4920 specified percentage, even if the machine is otherwise idle.
4922 * systemd-networkd learned IPIP and SIT tunnel support.
4924 * LSB init scripts exposing a dependency on $network will now
4925 get a dependency on network-online.target rather than simply
4926 network.target. This should bring LSB handling closer to
4927 what it was on SysV systems.
4929 * A new fsck.repair= kernel option has been added to control
4930 how fsck shall deal with unclean file systems at boot.
4932 * The (.ini) configuration file parser will now silently
4933 ignore sections whose name begins with "X-". This may be
4934 used to maintain application-specific extension sections in unit
4937 * machined gained a new API to query the IP addresses of
4938 registered containers. "machinectl status" has been updated
4939 to show these addresses in its output.
4941 * A new call sd_uid_get_display() has been added to the
4942 sd-login APIs for querying the "primary" session of a
4943 user. The "primary" session of the user is elected from the
4944 user's sessions and generally a graphical session is
4945 preferred over a text one.
4947 * A minimal systemd-resolved daemon has been added. It
4948 currently simply acts as a companion to systemd-networkd and
4949 manages resolv.conf based on per-interface DNS
4950 configuration, possibly supplied via DHCP. In the long run
4951 we hope to extend this into a local DNSSEC enabled DNS and
4954 * The systemd-networkd-wait-online tool is now enabled by
4955 default. It will delay network-online.target until a network
4956 connection has been configured. The tool primarily integrates
4957 with networkd, but will also make a best effort to make sense
4958 of network configuration performed in some other way.
4960 * Two new service options StartupCPUShares= and
4961 StartupBlockIOWeight= have been added that work similarly to
4962 CPUShares= and BlockIOWeight= however only apply during
4963 system startup. This is useful to prioritize certain services
4964 differently during bootup than during normal runtime.
4966 * hostnamed has been changed to prefer the statically
4967 configured hostname in /etc/hostname (unless set to
4968 'localhost' or empty) over any dynamic one supplied by
4969 dhcp. With this change, the rules for picking the hostname
4970 match more closely the rules of other configuration settings
4971 where the local administrator's configuration in /etc always
4972 overrides any other settings.
4974 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4975 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4976 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4977 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4978 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4979 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4980 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4981 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4982 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4983 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4984 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4985 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4986 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4987 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4988 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4989 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4992 — Beijing, 2014-05-28
4996 * When restoring the screen brightness at boot, stay away from
4997 the darkest setting or from the lowest 5% of the available
4998 range, depending on which is the larger value of both. This
4999 should effectively protect the user from rebooting into a
5000 black screen, should the brightness have been set to minimum
5003 * sd-login gained a new sd_machine_get_class() call to
5004 determine the class ("vm" or "container") of a machine
5005 registered with machined.
5007 * sd-login gained new calls
5008 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5009 to query the identity of the peer of a local AF_UNIX
5010 connection. They operate similarly to their sd_pid_get_xyz()
5013 * PID 1 will now maintain a system-wide system state engine
5014 with the states "starting", "running", "degraded",
5015 "maintenance", "stopping". These states are bound to system
5016 startup, normal runtime, runtime with at least one failed
5017 service, rescue/emergency mode and system shutdown. This
5018 state is shown in the "systemctl status" output when no unit
5019 name is passed. It is useful to determine system state, in
5020 particularly when doing so for many systems or containers at
5023 * A new command "list-machines" has been added to "systemctl"
5024 that lists all local OS containers and shows their system
5025 state (see above), if systemd runs inside of them.
5027 * systemctl gained a new "-r" switch to recursively enumerate
5028 units on all local containers, when used with the
5029 "list-unit" command (which is the default one that is
5030 executed when no parameters are specified).
5032 * The GPT automatic partition discovery logic will now honour
5033 two GPT partition flags: one may be set on a partition to
5034 cause it to be mounted read-only, and the other may be set
5035 on a partition to ignore it during automatic discovery.
5037 * Two new GPT type UUIDs have been added for automatic root
5038 partition discovery, for 32-bit and 64-bit ARM. This is not
5039 particularly useful for discovering the root directory on
5040 these architectures during bare-metal boots (since UEFI is
5041 not common there), but still very useful to allow booting of
5042 ARM disk images in nspawn with the -i option.
5044 * MAC addresses of interfaces created with nspawn's
5045 --network-interface= switch will now be generated from the
5046 machine name, and thus be stable between multiple invocations
5049 * logind will now automatically remove all IPC objects owned
5050 by a user if she or he fully logs out. This makes sure that
5051 users who are logged out cannot continue to consume IPC
5052 resources. This covers SysV memory, semaphores and message
5053 queues as well as POSIX shared memory and message
5054 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5055 limits. With this functionality, that is corrected. This may
5056 be turned off by using the RemoveIPC= switch of logind.conf.
5058 * The systemd-machine-id-setup and tmpfiles tools gained a
5059 --root= switch to operate on a specific root directory,
5062 * journald can now forward logged messages to the TTYs of all
5063 logged in users ("wall"). This is the default for all
5064 emergency messages now.
5066 * A new tool systemd-journal-remote has been added to stream
5067 journal log messages across the network.
5069 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5070 controller trees are mounted into it. Note that the
5071 directories mounted beneath it are not read-only. This is a
5072 security measure and is particularly useful because glibc
5073 actually includes a search logic to pick any tmpfs it can
5074 find to implement shm_open() if /dev/shm is not available
5075 (which it might very well be in namespaced setups).
5077 * machinectl gained a new "poweroff" command to cleanly power
5078 down a local OS container.
5080 * The PrivateDevices= unit file setting will now also drop the
5081 CAP_MKNOD capability from the capability bound set, and
5082 imply DevicePolicy=closed.
5084 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5085 comprehensively on all long-running systemd services where
5086 this is appropriate.
5088 * systemd-udevd will now run in a disassociated mount
5089 namespace. To mount directories from udev rules, make sure to
5090 pull in mount units via SYSTEMD_WANTS properties.
5092 * The kdbus support gained support for uploading policy into
5093 the kernel. sd-bus gained support for creating "monitoring"
5094 connections that can eavesdrop into all bus communication
5095 for debugging purposes.
5097 * Timestamps may now be specified in seconds since the UNIX
5098 epoch Jan 1st, 1970 by specifying "@" followed by the value
5101 * Native tcpwrap support in systemd has been removed. tcpwrap
5102 is old code, not really maintained anymore and has serious
5103 shortcomings, and better options such as firewalls
5104 exist. For setups that require tcpwrap usage, please
5105 consider invoking your socket-activated service via tcpd,
5106 like on traditional inetd.
5108 * A new system.conf configuration option
5109 DefaultTimerAccuracySec= has been added that controls the
5110 default AccuracySec= setting of .timer units.
5112 * Timer units gained a new WakeSystem= switch. If enabled,
5113 timers configured this way will cause the system to resume
5114 from system suspend (if the system supports that, which most
5117 * Timer units gained a new Persistent= switch. If enabled,
5118 timers configured this way will save to disk when they have
5119 been last triggered. This information is then used on next
5120 reboot to possible execute overdue timer events, that
5121 could not take place because the system was powered off.
5122 This enables simple anacron-like behaviour for timer units.
5124 * systemctl's "list-timers" will now also list the time a
5125 timer unit was last triggered in addition to the next time
5126 it will be triggered.
5128 * systemd-networkd will now assign predictable IPv4LL
5129 addresses to its local interfaces.
5131 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5132 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5133 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5134 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5135 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5136 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5137 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5138 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5141 — Berlin, 2014-03-25
5145 * A new unit file setting RestrictAddressFamilies= has been
5146 added to restrict which socket address families unit
5147 processes gain access to. This takes address family names
5148 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5149 attack surface of services via exotic protocol stacks. This
5150 is built on seccomp system call filters.
5152 * Two new unit file settings RuntimeDirectory= and
5153 RuntimeDirectoryMode= have been added that may be used to
5154 manage a per-daemon runtime directories below /run. This is
5155 an alternative for setting up directory permissions with
5156 tmpfiles snippets, and has the advantage that the runtime
5157 directory's lifetime is bound to the daemon runtime and that
5158 the daemon starts up with an empty directory each time. This
5159 is particularly useful when writing services that drop
5160 privileges using the User= or Group= setting.
5162 * The DeviceAllow= unit setting now supports globbing for
5163 matching against device group names.
5165 * The systemd configuration file system.conf gained new
5166 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5167 DefaultMemoryAccounting= to globally turn on/off accounting
5168 for specific resources (cgroups) for all units. These
5169 settings may still be overridden individually in each unit
5172 * systemd-gpt-auto-generator is now able to discover /srv and
5173 root partitions in addition to /home and swap partitions. It
5174 also supports LUKS-encrypted partitions now. With this in
5175 place, automatic discovery of partitions to mount following
5176 the Discoverable Partitions Specification
5177 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5178 is now a lot more complete. This allows booting without
5179 /etc/fstab and without root= on the kernel command line on
5180 systems prepared appropriately.
5182 * systemd-nspawn gained a new --image= switch which allows
5183 booting up disk images and Linux installations on any block
5184 device that follow the Discoverable Partitions Specification
5185 (see above). This means that installations made with
5186 appropriately updated installers may now be started and
5187 deployed using container managers, completely
5188 unmodified. (We hope that libvirt-lxc will add support for
5189 this feature soon, too.)
5191 * systemd-nspawn gained a new --network-macvlan= setting to
5192 set up a private macvlan interface for the
5193 container. Similarly, systemd-networkd gained a new
5194 Kind=macvlan setting in .netdev files.
5196 * systemd-networkd now supports configuring local addresses
5199 * A new tool systemd-network-wait-online has been added to
5200 synchronously wait for network connectivity using
5203 * The sd-bus.h bus API gained a new sd_bus_track object for
5204 tracking the life-cycle of bus peers. Note that sd-bus.h is
5205 still not a public API though (unless you specify
5206 --enable-kdbus on the configure command line, which however
5207 voids your warranty and you get no API stability guarantee).
5209 * The $XDG_RUNTIME_DIR runtime directories for each user are
5210 now individual tmpfs instances, which has the benefit of
5211 introducing separate pools for each user, with individual
5212 size limits, and thus making sure that unprivileged clients
5213 can no longer negatively impact the system or other users by
5214 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5215 RuntimeDirectorySize= has been introduced that allows
5216 controlling the default size limit for all users. It
5217 defaults to 10% of the available physical memory. This is no
5218 replacement for quotas on tmpfs though (which the kernel
5219 still does not support), as /dev/shm and /tmp are still
5220 shared resources used by both the system and unprivileged
5223 * logind will now automatically turn off automatic suspending
5224 on laptop lid close when more than one display is
5225 connected. This was previously expected to be implemented
5226 individually in desktop environments (such as GNOME),
5227 however has been added to logind now, in order to fix a
5228 boot-time race where a desktop environment might not have
5229 been started yet and thus not been able to take an inhibitor
5230 lock at the time where logind already suspends the system
5231 due to a closed lid.
5233 * logind will now wait at least 30s after each system
5234 suspend/resume cycle, and 3min after system boot before
5235 suspending the system due to a closed laptop lid. This
5236 should give USB docking stations and similar enough time to
5237 be probed and configured after system resume and boot in
5238 order to then act as suspend blocker.
5240 * systemd-run gained a new --property= setting which allows
5241 initialization of resource control properties (and others)
5242 for the created scope or service unit. Example: "systemd-run
5243 --property=BlockIOWeight=10 updatedb" may be used to run
5244 updatedb at a low block IO scheduling weight.
5246 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5247 now also work in --scope mode.
5249 * When systemd is compiled with kdbus support, basic support
5250 for enforced policies is now in place. (Note that enabling
5251 kdbus still voids your warranty and no API compatibility
5254 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5255 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5256 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5257 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5258 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5259 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5260 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5261 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5262 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5263 Zbigniew Jędrzejewski-Szmek
5265 — Berlin, 2014-03-12
5269 * systemd will now relabel /dev after loading the SMACK policy
5270 according to SMACK rules.
5272 * A new unit file option AppArmorProfile= has been added to
5273 set the AppArmor profile for the processes of a unit.
5275 * A new condition check ConditionArchitecture= has been added
5276 to conditionalize units based on the system architecture, as
5277 reported by uname()'s "machine" field.
5279 * systemd-networkd now supports matching on the system
5280 virtualization, architecture, kernel command line, host name
5283 * logind is now a lot more aggressive when suspending the
5284 machine due to a closed laptop lid. Instead of acting only
5285 on the lid close action, it will continuously watch the lid
5286 status and act on it. This is useful for laptops where the
5287 power button is on the outside of the chassis so that it can
5288 be reached without opening the lid (such as the Lenovo
5289 Yoga). On those machines, logind will now immediately
5290 re-suspend the machine if the power button has been
5291 accidentally pressed while the laptop was suspended and in a
5292 backpack or similar.
5294 * logind will now watch SW_DOCK switches and inhibit reaction
5295 to the lid switch if it is pressed. This means that logind
5296 will not suspend the machine anymore if the lid is closed
5297 and the system is docked, if the laptop supports SW_DOCK
5298 notifications via the input layer. Note that ACPI docking
5299 stations do not generate this currently. Also note that this
5300 logic is usually not fully sufficient and Desktop
5301 Environments should take a lid switch inhibitor lock when an
5302 external display is connected, as systemd will not watch
5305 * nspawn will now make use of the devices cgroup controller by
5306 default, and only permit creation of and access to the usual
5307 API device nodes like /dev/null or /dev/random, as well as
5308 access to (but not creation of) the pty devices.
5310 * We will now ship a default .network file for
5311 systemd-networkd that automatically configures DHCP for
5312 network interfaces created by nspawn's --network-veth or
5313 --network-bridge= switches.
5315 * systemd will now understand the usual M, K, G, T suffixes
5316 according to SI conventions (i.e. to the base 1000) when
5317 referring to throughput and hardware metrics. It will stay
5318 with IEC conventions (i.e. to the base 1024) for software
5319 metrics, according to what is customary according to
5320 Wikipedia. We explicitly document which base applies for
5321 each configuration option.
5323 * The DeviceAllow= setting in unit files now supports a syntax
5324 to whitelist an entire group of devices node majors at once,
5325 based on the /proc/devices listing. For example, with the
5326 string "char-pts", it is now possible to whitelist all
5327 current and future pseudo-TTYs at once.
5329 * sd-event learned a new "post" event source. Event sources of
5330 this type are triggered by the dispatching of any event
5331 source of a type that is not "post". This is useful for
5332 implementing clean-up and check event sources that are
5333 triggered by other work being done in the program.
5335 * systemd-networkd is no longer statically enabled, but uses
5336 the usual [Install] sections so that it can be
5337 enabled/disabled using systemctl. It still is enabled by
5340 * When creating a veth interface pair with systemd-nspawn, the
5341 host side will now be prefixed with "vb-" if
5342 --network-bridge= is used, and with "ve-" if --network-veth
5343 is used. This way, it is easy to distinguish these cases on
5344 the host, for example to apply different configuration to
5345 them with systemd-networkd.
5347 * The compatibility libraries for libsystemd-journal.so,
5348 libsystem-id128.so, libsystemd-login.so and
5349 libsystemd-daemon.so do not make use of IFUNC
5350 anymore. Instead, we now build libsystemd.so multiple times
5351 under these alternative names. This means that the footprint
5352 is drastically increased, but given that these are
5353 transitional compatibility libraries, this should not matter
5354 much. This change has been made necessary to support the ARM
5355 platform for these compatibility libraries, as the ARM
5356 toolchain is not really at the same level as the toolchain
5357 for other architectures like x86 and does not support
5358 IFUNC. Please make sure to use --enable-compat-libs only
5359 during a transitional period!
5361 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5362 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5363 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5364 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5365 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5366 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5367 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5368 Zbigniew Jędrzejewski-Szmek
5370 — Berlin, 2014-02-24
5374 * A new component "systemd-networkd" has been added that can
5375 be used to configure local network interfaces statically or
5376 via DHCP. It is capable of bringing up bridges, VLANs, and
5377 bonding. Currently, no hook-ups for interactive network
5378 configuration are provided. Use this for your initrd,
5379 container, embedded, or server setup if you need a simple,
5380 yet powerful, network configuration solution. This
5381 configuration subsystem is quite nifty, as it allows wildcard
5382 hotplug matching in interfaces. For example, with a single
5383 configuration snippet, you can configure that all Ethernet
5384 interfaces showing up are automatically added to a bridge,
5385 or similar. It supports link-sensing and more.
5387 * A new tool "systemd-socket-proxyd" has been added which can
5388 act as a bidirectional proxy for TCP sockets. This is
5389 useful for adding socket activation support to services that
5390 do not actually support socket activation, including virtual
5391 machines and the like.
5393 * Add a new tool to save/restore rfkill state on
5396 * Save/restore state of keyboard backlights in addition to
5397 display backlights on shutdown/boot.
5399 * udev learned a new SECLABEL{} construct to label device
5400 nodes with a specific security label when they appear. For
5401 now, only SECLABEL{selinux} is supported, but the syntax is
5402 prepared for additional security frameworks.
5404 * udev gained a new scheme to configure link-level attributes
5405 from files in /etc/systemd/network/*.link. These files can
5406 match against MAC address, device path, driver name and type,
5407 and will apply attributes like the naming policy, link speed,
5408 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5409 address assignment policy (randomized, ...).
5411 * The configuration of network interface naming rules for
5412 "permanent interface names" has changed: a new NamePolicy=
5413 setting in the [Link] section of .link files determines the
5414 priority of possible naming schemes (onboard, slot, MAC,
5415 path). The default value of this setting is determined by
5416 /usr/lib/net/links/99-default.link. Old
5417 80-net-name-slot.rules udev configuration file has been
5418 removed, so local configuration overriding this file should
5419 be adapted to override 99-default.link instead.
5421 * When the User= switch is used in a unit file, also
5422 initialize $SHELL= based on the user database entry.
5424 * systemd no longer depends on libdbus. All communication is
5425 now done with sd-bus, systemd's low-level bus library
5428 * kdbus support has been added to PID 1 itself. When kdbus is
5429 enabled, this causes PID 1 to set up the system bus and
5430 enable support for a new ".busname" unit type that
5431 encapsulates bus name activation on kdbus. It works a little
5432 bit like ".socket" units, except for bus names. A new
5433 generator has been added that converts classic dbus1 service
5434 activation files automatically into native systemd .busname
5437 * sd-bus: add a light-weight vtable implementation that allows
5438 defining objects on the bus with a simple static const
5439 vtable array of its methods, signals and properties.
5441 * systemd will not generate or install static dbus
5442 introspection data anymore to /usr/share/dbus-1/interfaces,
5443 as the precise format of these files is unclear, and
5444 nothing makes use of it.
5446 * A proxy daemon is now provided to proxy clients connecting
5447 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5448 compatibility with classic D-Bus.
5450 * A bus driver implementation has been added that supports the
5451 classic D-Bus bus driver calls on kdbus, also for
5452 compatibility purposes.
5454 * A new API "sd-event.h" has been added that implements a
5455 minimal event loop API built around epoll. It provides a
5456 couple of features that direct epoll usage is lacking:
5457 prioritization of events, scales to large numbers of timer
5458 events, per-event timer slack (accuracy), system-wide
5459 coalescing of timer events, exit handlers, watchdog
5460 supervision support using systemd's sd_notify() API, child
5463 * A new API "sd-rntl.h" has been added that provides an API
5464 around the route netlink interface of the kernel, similar in
5465 style to "sd-bus.h".
5467 * A new API "sd-dhcp-client.h" has been added that provides a
5468 small DHCPv4 client-side implementation. This is used by
5471 * There is a new kernel command line option
5472 "systemd.restore_state=0|1". When set to "0", none of the
5473 systemd tools will restore saved runtime state to hardware
5474 devices. More specifically, the rfkill and backlight states
5477 * The FsckPassNo= compatibility option in mount/service units
5478 has been removed. The fstab generator will now add the
5479 necessary dependencies automatically, and does not require
5480 PID1's support for that anymore.
5482 * journalctl gained a new switch, --list-boots, that lists
5483 recent boots with their times and boot IDs.
5485 * The various tools like systemctl, loginctl, timedatectl,
5486 busctl, systemd-run, ... have gained a new switch "-M" to
5487 connect to a specific, local OS container (as direct
5488 connection, without requiring SSH). This works on any
5489 container that is registered with machined, such as those
5490 created by libvirt-lxc or nspawn.
5492 * systemd-run and systemd-analyze also gained support for "-H"
5493 to connect to remote hosts via SSH. This is particularly
5494 useful for systemd-run because it enables queuing of jobs
5495 onto remote systems.
5497 * machinectl gained a new command "login" to open a getty
5498 login in any local container. This works with any container
5499 that is registered with machined (such as those created by
5500 libvirt-lxc or nspawn), and which runs systemd inside.
5502 * machinectl gained a new "reboot" command that may be used to
5503 trigger a reboot on a specific container that is registered
5504 with machined. This works on any container that runs an init
5505 system of some kind.
5507 * systemctl gained a new "list-timers" command to print a nice
5508 listing of installed timer units with the times they elapse
5511 * Alternative reboot() parameters may now be specified on the
5512 "systemctl reboot" command line and are passed to the
5513 reboot() system call.
5515 * systemctl gained a new --job-mode= switch to configure the
5516 mode to queue a job with. This is a more generic version of
5517 --fail, --irreversible, and --ignore-dependencies, which are
5518 still available but not advertised anymore.
5520 * /etc/systemd/system.conf gained new settings to configure
5521 various default timeouts of units, as well as the default
5522 start limit interval and burst. These may still be overridden
5525 * PID1 will now export on the bus profile data of the security
5526 policy upload process (such as the SELinux policy upload to
5529 * journald: when forwarding logs to the console, include
5530 timestamps (following the setting in
5531 /sys/module/printk/parameters/time).
5533 * OnCalendar= in timer units now understands the special
5534 strings "yearly" and "annually". (Both are equivalent)
5536 * The accuracy of timer units is now configurable with the new
5537 AccuracySec= setting. It defaults to 1min.
5539 * A new dependency type JoinsNamespaceOf= has been added that
5540 allows running two services within the same /tmp and network
5541 namespace, if PrivateNetwork= or PrivateTmp= are used.
5543 * A new command "cat" has been added to systemctl. It outputs
5544 the original unit file of a unit, and concatenates the
5545 contents of additional "drop-in" unit file snippets, so that
5546 the full configuration is shown.
5548 * systemctl now supports globbing on the various "list-xyz"
5549 commands, like "list-units" or "list-sockets", as well as on
5550 those commands which take multiple unit names.
5552 * journalctl's --unit= switch gained support for globbing.
5554 * All systemd daemons now make use of the watchdog logic so
5555 that systemd automatically notices when they hang.
5557 * If the $container_ttys environment variable is set,
5558 getty-generator will automatically spawn a getty for each
5559 listed tty. This is useful for container managers to request
5560 login gettys to be spawned on as many ttys as needed.
5562 * %h, %s, %U specifier support is not available anymore when
5563 used in unit files for PID 1. This is because NSS calls are
5564 not safe from PID 1. They stay available for --user
5565 instances of systemd, and as special case for the root user.
5567 * loginctl gained a new "--no-legend" switch to turn off output
5570 * The "sd-login.h" API gained three new calls:
5571 sd_session_is_remote(), sd_session_get_remote_user(),
5572 sd_session_get_remote_host() to query information about
5575 * The udev hardware database now also carries vendor/product
5576 information of SDIO devices.
5578 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5579 determine whether watchdog notifications are requested by
5582 * Socket-activated per-connection services now include a
5583 short description of the connection parameters in the
5586 * tmpfiles gained a new "--boot" option. When this is not used,
5587 only lines where the command character is not suffixed with
5588 "!" are executed. When this option is specified, those
5589 options are executed too. This partitions tmpfiles
5590 directives into those that can be safely executed at any
5591 time, and those which should be run only at boot (for
5592 example, a line that creates /run/nologin).
5594 * A new API "sd-resolve.h" has been added which provides a simple
5595 asynchronous wrapper around glibc NSS host name resolution
5596 calls, such as getaddrinfo(). In contrast to glibc's
5597 getaddrinfo_a(), it does not use signals. In contrast to most
5598 other asynchronous name resolution libraries, this one does
5599 not reimplement DNS, but reuses NSS, so that alternate
5600 host name resolution systems continue to work, such as mDNS,
5601 LDAP, etc. This API is based on libasyncns, but it has been
5602 cleaned up for inclusion in systemd.
5604 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5605 "sd-daemon.h" are no longer found in individual libraries
5606 libsystemd-journal.so, libsystemd-login.so,
5607 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5608 merged them into a single library, libsystemd.so, which
5609 provides all symbols. The reason for this is cyclic
5610 dependencies, as these libraries tend to use each other's
5611 symbols. So far, we have managed to workaround that by linking
5612 a copy of a good part of our code into each of these
5613 libraries again and again, which, however, makes certain
5614 things hard to do, like sharing static variables. Also, it
5615 substantially increases footprint. With this change, there
5616 is only one library for the basic APIs systemd
5617 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5618 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5619 library as well, however are subject to the --enable-kdbus
5620 switch (see below). Note that "sd-dhcp-client.h" is not part
5621 of this library (this is because it only consumes, never
5622 provides, services of/to other APIs). To make the transition
5623 easy from the separate libraries to the unified one, we
5624 provide the --enable-compat-libs compile-time switch which
5625 will generate stub libraries that are compatible with the
5626 old ones but redirect all calls to the new one.
5628 * All of the kdbus logic and the new APIs "sd-bus.h",
5629 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5630 and "sd-utf8.h" are compile-time optional via the
5631 "--enable-kdbus" switch, and they are not compiled in by
5632 default. To make use of kdbus, you have to explicitly enable
5633 the switch. Note however, that neither the kernel nor the
5634 userspace API for all of this is considered stable yet. We
5635 want to maintain the freedom to still change the APIs for
5636 now. By specifying this build-time switch, you acknowledge
5637 that you are aware of the instability of the current
5640 * Also, note that while kdbus is pretty much complete,
5641 it lacks one thing: proper policy support. This means you
5642 can build a fully working system with all features; however,
5643 it will be highly insecure. Policy support will be added in
5644 one of the next releases, at the same time that we will
5645 declare the APIs stable.
5647 * When the kernel command line argument "kdbus" is specified,
5648 systemd will automatically load the kdbus.ko kernel module. At
5649 this stage of development, it is only useful for testing kdbus
5650 and should not be used in production. Note: if "--enable-kdbus"
5651 is specified, and the kdbus.ko kernel module is available, and
5652 "kdbus" is added to the kernel command line, the entire system
5653 runs with kdbus instead of dbus-daemon, with the above mentioned
5654 problem of missing the system policy enforcement. Also a future
5655 version of kdbus.ko or a newer systemd will not be compatible with
5656 each other, and will unlikely be able to boot the machine if only
5657 one of them is updated.
5659 * systemctl gained a new "import-environment" command which
5660 uploads the caller's environment (or parts thereof) into the
5661 service manager so that it is inherited by services started
5662 by the manager. This is useful to upload variables like
5663 $DISPLAY into the user service manager.
5665 * A new PrivateDevices= switch has been added to service units
5666 which allows running a service with a namespaced /dev
5667 directory that does not contain any device nodes for
5668 physical devices. More specifically, it only includes devices
5669 such as /dev/null, /dev/urandom, and /dev/zero which are API
5672 * logind has been extended to support behaviour like VT
5673 switching on seats that do not support a VT. This makes
5674 multi-session available on seats that are not the first seat
5675 (seat0), and on systems where kernel support for VTs has
5676 been disabled at compile-time.
5678 * If a process holds a delay lock for system sleep or shutdown
5679 and fails to release it in time, we will now log its
5680 identity. This makes it easier to identify processes that
5681 cause slow suspends or power-offs.
5683 * When parsing /etc/crypttab, support for a new key-slot=
5684 option as supported by Debian is added. It allows indicating
5685 which LUKS slot to use on disk, speeding up key loading.
5687 * The sd_journal_sendv() API call has been checked and
5688 officially declared to be async-signal-safe so that it may
5689 be invoked from signal handlers for logging purposes.
5691 * Boot-time status output is now enabled automatically after a
5692 short timeout if boot does not progress, in order to give
5693 the user an indication what she or he is waiting for.
5695 * The boot-time output has been improved to show how much time
5696 remains until jobs expire.
5698 * The KillMode= switch in service units gained a new possible
5699 value "mixed". If set, and the unit is shut down, then the
5700 initial SIGTERM signal is sent only to the main daemon
5701 process, while the following SIGKILL signal is sent to
5702 all remaining processes of the service.
5704 * When a scope unit is registered, a new property "Controller"
5705 may be set. If set to a valid bus name, systemd will send a
5706 RequestStop() signal to this name when it would like to shut
5707 down the scope. This may be used to hook manager logic into
5708 the shutdown logic of scope units. Also, scope units may now
5709 be put in a special "abandoned" state, in which case the
5710 manager process which created them takes no further
5711 responsibilities for it.
5713 * When reading unit files, systemd will now verify
5714 the access mode of these files, and warn about certain
5715 suspicious combinations. This has been added to make it
5716 easier to track down packaging bugs where unit files are
5717 marked executable or world-writable.
5719 * systemd-nspawn gained a new "--setenv=" switch to set
5720 container-wide environment variables. The similar option in
5721 systemd-activate was renamed from "--environment=" to
5722 "--setenv=" for consistency.
5724 * systemd-nspawn has been updated to create a new kdbus domain
5725 for each container that is invoked, thus allowing each
5726 container to have its own set of system and user buses,
5727 independent of the host.
5729 * systemd-nspawn gained a new --drop-capability= switch to run
5730 the container with less capabilities than the default. Both
5731 --drop-capability= and --capability= now take the special
5732 string "all" for dropping or keeping all capabilities.
5734 * systemd-nspawn gained new switches for executing containers
5735 with specific SELinux labels set.
5737 * systemd-nspawn gained a new --quiet switch to not generate
5738 any additional output but the container's own console
5741 * systemd-nspawn gained a new --share-system switch to run a
5742 container without PID namespacing enabled.
5744 * systemd-nspawn gained a new --register= switch to control
5745 whether the container is registered with systemd-machined or
5746 not. This is useful for containers that do not run full
5747 OS images, but only specific apps.
5749 * systemd-nspawn gained a new --keep-unit which may be used
5750 when invoked as the only program from a service unit, and
5751 results in registration of the unit service itself in
5752 systemd-machined, instead of a newly opened scope unit.
5754 * systemd-nspawn gained a new --network-interface= switch for
5755 moving arbitrary interfaces to the container. The new
5756 --network-veth switch creates a virtual Ethernet connection
5757 between host and container. The new --network-bridge=
5758 switch then allows assigning the host side of this virtual
5759 Ethernet connection to a bridge device.
5761 * systemd-nspawn gained a new --personality= switch for
5762 setting the kernel personality for the container. This is
5763 useful when running a 32-bit container on a 64-bit host. A
5764 similar option Personality= is now also available for service
5767 * logind will now also track a "Desktop" identifier for each
5768 session which encodes the desktop environment of it. This is
5769 useful for desktop environments that want to identify
5770 multiple running sessions of itself easily.
5772 * A new SELinuxContext= setting for service units has been
5773 added that allows setting a specific SELinux execution
5774 context for a service.
5776 * Most systemd client tools will now honour $SYSTEMD_LESS for
5777 settings of the "less" pager. By default, these tools will
5778 override $LESS to allow certain operations to work, such as
5779 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5780 influence this logic.
5782 * systemd's "seccomp" hook-up has been changed to make use of
5783 the libseccomp library instead of using its own
5784 implementation. This has benefits for portability among
5787 * For usage together with SystemCallFilter=, a new
5788 SystemCallErrorNumber= setting has been introduced that
5789 allows configuration of a system error number to be returned
5790 on filtered system calls, instead of immediately killing the
5791 process. Also, SystemCallArchitectures= has been added to
5792 limit access to system calls of a particular architecture
5793 (in order to turn off support for unused secondary
5794 architectures). There is also a global
5795 SystemCallArchitectures= setting in system.conf now to turn
5796 off support for non-native system calls system-wide.
5798 * systemd requires a kernel with a working name_to_handle_at(),
5799 please see the kernel config requirements in the README file.
5801 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5802 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5803 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5804 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5805 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5806 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5807 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5808 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5809 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5810 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5811 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5812 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5813 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5814 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5815 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5816 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5817 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5818 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5819 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5820 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5821 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5822 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5823 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5824 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5826 — Berlin, 2014-02-20
5830 * logind has gained support for facilitating privileged input
5831 and drm device access for unprivileged clients. This work is
5832 useful to allow Wayland display servers (and similar
5833 programs, such as kmscon) to run under the user's ID and
5834 access input and drm devices which are normally
5835 protected. When this is used (and the kernel is new enough)
5836 logind will "mute" IO on the file descriptors passed to
5837 Wayland as long as it is in the background and "unmute" it
5838 if it returns into the foreground. This allows secure
5839 session switching without allowing background sessions to
5840 eavesdrop on input and display data. This also introduces
5841 session switching support if VT support is turned off in the
5842 kernel, and on seats that are not seat0.
5844 * A new kernel command line option luks.options= is understood
5845 now which allows specifying LUKS options for usage for LUKS
5846 encrypted partitions specified with luks.uuid=.
5848 * tmpfiles.d(5) snippets may now use specifier expansion in
5849 path names. More specifically %m, %b, %H, %v, are now
5850 replaced by the local machine id, boot id, hostname, and
5851 kernel version number.
5853 * A new tmpfiles.d(5) command "m" has been introduced which
5854 may be used to change the owner/group/access mode of a file
5855 or directory if it exists, but do nothing if it does not.
5857 * This release removes high-level support for the
5858 MemorySoftLimit= cgroup setting. The underlying kernel
5859 cgroup attribute memory.soft_limit= is currently badly
5860 designed and likely to be removed from the kernel API in its
5861 current form, hence we should not expose it for now.
5863 * The memory.use_hierarchy cgroup attribute is now enabled for
5864 all cgroups systemd creates in the memory cgroup
5865 hierarchy. This option is likely to be come the built-in
5866 default in the kernel anyway, and the non-hierarchical mode
5867 never made much sense in the intrinsically hierarchical
5870 * A new field _SYSTEMD_SLICE= is logged along with all journal
5871 messages containing the slice a message was generated
5872 from. This is useful to allow easy per-customer filtering of
5873 logs among other things.
5875 * systemd-journald will no longer adjust the group of journal
5876 files it creates to the "systemd-journal" group. Instead we
5877 rely on the journal directory to be owned by the
5878 "systemd-journal" group, and its setgid bit set, so that the
5879 kernel file system layer will automatically enforce that
5880 journal files inherit this group assignment. The reason for
5881 this change is that we cannot allow NSS look-ups from
5882 journald which would be necessary to resolve
5883 "systemd-journal" to a numeric GID, because this might
5884 create deadlocks if NSS involves synchronous queries to
5885 other daemons (such as nscd, or sssd) which in turn are
5886 logging clients of journald and might block on it, which
5887 would then dead lock. A tmpfiles.d(5) snippet included in
5888 systemd will make sure the setgid bit and group are
5889 properly set on the journal directory if it exists on every
5890 boot. However, we recommend adjusting it manually after
5891 upgrades too (or from RPM scriptlets), so that the change is
5892 not delayed until next reboot.
5894 * Backlight and random seed files in /var/lib/ have moved into
5895 the /var/lib/systemd/ directory, in order to centralize all
5896 systemd generated files in one directory.
5898 * Boot time performance measurements (as displayed by
5899 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5900 performance information if that's available to determine how
5901 much time BIOS and boot loader initialization required. With
5902 a sufficiently new BIOS you hence no longer need to boot
5903 with Gummiboot to get access to such information.
5905 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5906 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5907 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5908 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5909 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5910 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5911 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5913 — Berlin, 2013-10-02
5917 * The Restart= option for services now understands a new
5918 on-watchdog setting, which will restart the service
5919 automatically if the service stops sending out watchdog keep
5920 alive messages (as configured with WatchdogSec=).
5922 * The getty generator (which is responsible for bringing up a
5923 getty on configured serial consoles) will no longer only
5924 start a getty on the primary kernel console but on all
5925 others, too. This makes the order in which console= is
5926 specified on the kernel command line less important.
5928 * libsystemd-logind gained a new sd_session_get_vt() call to
5929 retrieve the VT number of a session.
5931 * If the option "tries=0" is set for an entry of /etc/crypttab
5932 its passphrase is queried indefinitely instead of any
5933 maximum number of tries.
5935 * If a service with a configure PID file terminates its PID
5936 file will now be removed automatically if it still exists
5937 afterwards. This should put an end to stale PID files.
5939 * systemd-run will now also take relative binary path names
5940 for execution and no longer insists on absolute paths.
5942 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5943 paths that are optionally prefixed with "-" to indicate that
5944 it should not be considered a failure if they do not exist.
5946 * journalctl -o (and similar commands) now understands a new
5947 output mode "short-precise", it is similar to "short" but
5948 shows timestamps with usec accuracy.
5950 * The option "discard" (as known from Debian) is now
5951 synonymous to "allow-discards" in /etc/crypttab. In fact,
5952 "discard" is preferred now (since it is easier to remember
5955 * Some licensing clean-ups were made, so that more code is now
5956 LGPL-2.1 licensed than before.
5958 * A minimal tool to save/restore the display backlight
5959 brightness across reboots has been added. It will store the
5960 backlight setting as late as possible at shutdown, and
5961 restore it as early as possible during reboot.
5963 * A logic to automatically discover and enable home and swap
5964 partitions on GPT disks has been added. With this in place
5965 /etc/fstab becomes optional for many setups as systemd can
5966 discover certain partitions located on the root disk
5967 automatically. Home partitions are recognized under their
5968 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5969 partitions are recognized under their GPT type ID
5970 0657fd6da4ab43c484e50933c84b4f4f.
5972 * systemd will no longer pass any environment from the kernel
5973 or initrd to system services. If you want to set an
5974 environment for all services, do so via the kernel command
5975 line systemd.setenv= assignment.
5977 * The systemd-sysctl tool no longer natively reads the file
5978 /etc/sysctl.conf. If desired, the file should be symlinked
5979 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5980 legacy support by a symlink rather than built-in code, it
5981 also makes the otherwise hidden order of application of the
5982 different files visible. (Note that this partly reverts to a
5983 pre-198 application order of sysctl knobs!)
5985 * The "systemctl set-log-level" and "systemctl dump" commands
5986 have been moved to systemd-analyze.
5988 * systemd-run learned the new --remain-after-exit switch,
5989 which causes the scope unit not to be cleaned up
5990 automatically after the process terminated.
5992 * tmpfiles learned a new --exclude-prefix= switch to exclude
5993 certain paths from operation.
5995 * journald will now automatically flush all messages to disk
5996 as soon as a message at the log level CRIT, ALERT or EMERG
5999 Contributions from: Andrew Cook, Brandon Philips, Christian
6000 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6001 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6002 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6003 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6004 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6005 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6006 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6007 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6008 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6009 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6010 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6011 William Giokas, Zbigniew Jędrzejewski-Szmek
6013 — Berlin, 2013-09-13
6017 * The documentation has been updated to cover the various new
6018 concepts introduced with 205.
6020 * Unit files now understand the new %v specifier which
6021 resolves to the kernel version string as returned by "uname
6024 * systemctl now supports filtering the unit list output by
6025 load state, active state and sub state, using the new
6028 * "systemctl status" will now show the results of the
6029 condition checks (like ConditionPathExists= and similar) of
6030 the last start attempts of the unit. They are also logged to
6033 * "journalctl -b" may now be used to look for boot output of a
6034 specific boot. Try "journalctl -b -1" for the previous boot,
6035 but the syntax is substantially more powerful.
6037 * "journalctl --show-cursor" has been added which prints the
6038 cursor string the last shown log line. This may then be used
6039 with the new "journalctl --after-cursor=" switch to continue
6040 browsing logs from that point on.
6042 * "journalctl --force" may now be used to force regeneration
6045 * Creation of "dead" device nodes has been moved from udev
6046 into kmod and tmpfiles. Previously, udev would read the kmod
6047 databases to pre-generate dead device nodes based on meta
6048 information contained in kernel modules, so that these would
6049 be auto-loaded on access rather then at boot. As this
6050 does not really have much to do with the exposing actual
6051 kernel devices to userspace this has always been slightly
6052 alien in the udev codebase. Following the new scheme kmod
6053 will now generate a runtime snippet for tmpfiles from the
6054 module meta information and it now is tmpfiles' job to the
6055 create the nodes. This also allows overriding access and
6056 other parameters for the nodes using the usual tmpfiles
6057 facilities. As side effect this allows us to remove the
6058 CAP_SYS_MKNOD capability bit from udevd entirely.
6060 * logind's device ACLs may now be applied to these "dead"
6061 devices nodes too, thus finally allowing managed access to
6062 devices such as /dev/snd/sequencer without loading the
6063 backing module right-away.
6065 * A new RPM macro has been added that may be used to apply
6066 tmpfiles configuration during package installation.
6068 * systemd-detect-virt and ConditionVirtualization= now can
6069 detect User-Mode-Linux machines (UML).
6071 * journald will now implicitly log the effective capabilities
6072 set of processes in the message metadata.
6074 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6076 * The initrd interface has been simplified (more specifically,
6077 support for passing performance data via environment
6078 variables and fsck results via files in /run has been
6079 removed). These features were non-essential, and are
6080 nowadays available in a much nicer way by having systemd in
6081 the initrd serialize its state and have the hosts systemd
6082 deserialize it again.
6084 * The udev "keymap" data files and tools to apply keyboard
6085 specific mappings of scan to key codes, and force-release
6086 scan code lists have been entirely replaced by a udev
6087 "keyboard" builtin and a hwdb data file.
6089 * systemd will now honour the kernel's "quiet" command line
6090 argument also during late shutdown, resulting in a
6091 completely silent shutdown when used.
6093 * There's now an option to control the SO_REUSEPORT socket
6094 option in .socket units.
6096 * Instance units will now automatically get a per-template
6097 subslice of system.slice unless something else is explicitly
6098 configured. For example, instances of sshd@.service will now
6099 implicitly be placed in system-sshd.slice rather than
6100 system.slice as before.
6102 * Test coverage support may now be enabled at build time.
6104 Contributions from: Dave Reisner, Frederic Crozat, Harald
6105 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6106 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6107 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6108 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6109 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6110 Giokas, Zbigniew Jędrzejewski-Szmek
6112 — Berlin, 2013-07-23
6116 * Two new unit types have been introduced:
6118 Scope units are very similar to service units, however, are
6119 created out of pre-existing processes — instead of PID 1
6120 forking off the processes. By using scope units it is
6121 possible for system services and applications to group their
6122 own child processes (worker processes) in a powerful way
6123 which then maybe used to organize them, or kill them
6124 together, or apply resource limits on them.
6126 Slice units may be used to partition system resources in an
6127 hierarchical fashion and then assign other units to them. By
6128 default there are now three slices: system.slice (for all
6129 system services), user.slice (for all user sessions),
6130 machine.slice (for VMs and containers).
6132 Slices and scopes have been introduced primarily in
6133 context of the work to move cgroup handling to a
6134 single-writer scheme, where only PID 1
6135 creates/removes/manages cgroups.
6137 * There's a new concept of "transient" units. In contrast to
6138 normal units these units are created via an API at runtime,
6139 not from configuration from disk. More specifically this
6140 means it is now possible to run arbitrary programs as
6141 independent services, with all execution parameters passed
6142 in via bus APIs rather than read from disk. Transient units
6143 make systemd substantially more dynamic then it ever was,
6144 and useful as a general batch manager.
6146 * logind has been updated to make use of scope and slice units
6147 for managing user sessions. As a user logs in he will get
6148 his own private slice unit, to which all sessions are added
6149 as scope units. We also added support for automatically
6150 adding an instance of user@.service for the user into the
6151 slice. Effectively logind will no longer create cgroup
6152 hierarchies on its own now, it will defer entirely to PID 1
6153 for this by means of scope, service and slice units. Since
6154 user sessions this way become entities managed by PID 1
6155 the output of "systemctl" is now a lot more comprehensive.
6157 * A new mini-daemon "systemd-machined" has been added which
6158 may be used by virtualization managers to register local
6159 VMs/containers. nspawn has been updated accordingly, and
6160 libvirt will be updated shortly. machined will collect a bit
6161 of meta information about the VMs/containers, and assign
6162 them their own scope unit (see above). The collected
6163 meta-data is then made available via the "machinectl" tool,
6164 and exposed in "ps" and similar tools. machined/machinectl
6165 is compile-time optional.
6167 * As discussed earlier, the low-level cgroup configuration
6168 options ControlGroup=, ControlGroupModify=,
6169 ControlGroupPersistent=, ControlGroupAttribute= have been
6170 removed. Please use high-level attribute settings instead as
6171 well as slice units.
6173 * A new bus call SetUnitProperties() has been added to alter
6174 various runtime parameters of a unit. This is primarily
6175 useful to alter cgroup parameters dynamically in a nice way,
6176 but will be extended later on to make more properties
6177 modifiable at runtime. systemctl gained a new set-properties
6178 command that wraps this call.
6180 * A new tool "systemd-run" has been added which can be used to
6181 run arbitrary command lines as transient services or scopes,
6182 while configuring a number of settings via the command
6183 line. This tool is currently very basic, however already
6184 very useful. We plan to extend this tool to even allow
6185 queuing of execution jobs with time triggers from the
6186 command line, similar in fashion to "at".
6188 * nspawn will now inform the user explicitly that kernels with
6189 audit enabled break containers, and suggest the user to turn
6192 * Support for detecting the IMA and AppArmor security
6193 frameworks with ConditionSecurity= has been added.
6195 * journalctl gained a new "-k" switch for showing only kernel
6196 messages, mimicking dmesg output; in addition to "--user"
6197 and "--system" switches for showing only user's own logs
6200 * systemd-delta can now show information about drop-in
6201 snippets extending unit files.
6203 * libsystemd-bus has been substantially updated but is still
6204 not available as public API.
6206 * systemd will now look for the "debug" argument on the kernel
6207 command line and enable debug logging, similar to what
6208 "systemd.log_level=debug" already did before.
6210 * "systemctl set-default", "systemctl get-default" has been
6211 added to configure the default.target symlink, which
6212 controls what to boot into by default.
6214 * "systemctl set-log-level" has been added as a convenient
6215 way to raise and lower systemd logging threshold.
6217 * "systemd-analyze plot" will now show the time the various
6218 generators needed for execution, as well as information
6219 about the unit file loading.
6221 * libsystemd-journal gained a new sd_journal_open_files() call
6222 for opening specific journal files. journactl also gained a
6223 new switch to expose this new functionality. Previously we
6224 only supported opening all files from a directory, or all
6225 files from the system, as opening individual files only is
6226 racy due to journal file rotation.
6228 * systemd gained the new DefaultEnvironment= setting in
6229 /etc/systemd/system.conf to set environment variables for
6232 * If a privileged process logs a journal message with the
6233 OBJECT_PID= field set, then journald will automatically
6234 augment this with additional OBJECT_UID=, OBJECT_GID=,
6235 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6236 system services want to log events about specific client
6237 processes. journactl/systemctl has been updated to make use
6238 of this information if all log messages regarding a specific
6241 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6242 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6243 Reisner, David Coppa, David King, David Strauss, Eelco
6244 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6245 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6246 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6247 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6248 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6249 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6250 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6251 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6252 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6253 Łukasz Stelmach, 장동준
6257 * The Python bindings gained some minimal support for the APIs
6258 exposed by libsystemd-logind.
6260 * ConditionSecurity= gained support for detecting SMACK. Since
6261 this condition already supports SELinux and AppArmor we only
6262 miss IMA for this. Patches welcome!
6264 Contributions from: Karol Lewandowski, Lennart Poettering,
6265 Zbigniew Jędrzejewski-Szmek
6269 * systemd-nspawn will now create /etc/resolv.conf if
6270 necessary, before bind-mounting the host's file onto it.
6272 * systemd-nspawn will now store meta information about a
6273 container on the container's cgroup as extended attribute
6274 fields, including the root directory.
6276 * The cgroup hierarchy has been reworked in many ways. All
6277 objects any of the components systemd creates in the cgroup
6278 tree are now suffixed. More specifically, user sessions are
6279 now placed in cgroups suffixed with ".session", users in
6280 cgroups suffixed with ".user", and nspawn containers in
6281 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6282 names are now escaped in a simple scheme to avoid collision
6283 of userspace object names with kernel filenames. This work
6284 is preparation for making these objects relocatable in the
6285 cgroup tree, in order to allow easy resource partitioning of
6286 these objects without causing naming conflicts.
6288 * systemctl list-dependencies gained the new switches
6289 --plain, --reverse, --after and --before.
6291 * systemd-inhibit now shows the process name of processes that
6292 have taken an inhibitor lock.
6294 * nss-myhostname will now also resolve "localhost"
6295 implicitly. This makes /etc/hosts an optional file and
6296 nicely handles that on IPv6 ::1 maps to both "localhost" and
6299 * libsystemd-logind.so gained a new call
6300 sd_get_machine_names() to enumerate running containers and
6301 VMs (currently only supported by very new libvirt and
6302 nspawn). sd_login_monitor can now be used to watch
6303 VMs/containers coming and going.
6305 * .include is not allowed recursively anymore, and only in
6306 unit files. Usually it is better to use drop-in snippets in
6307 .d/*.conf anyway, as introduced with systemd 198.
6309 * systemd-analyze gained a new "critical-chain" command that
6310 determines the slowest chain of units run during system
6311 boot-up. It is very useful for tracking down where
6312 optimizing boot time is the most beneficial.
6314 * systemd will no longer allow manipulating service paths in
6315 the name=systemd:/system cgroup tree using ControlGroup= in
6316 units. (But is still fine with it in all other dirs.)
6318 * There's a new systemd-nspawn@.service service file that may
6319 be used to easily run nspawn containers as system
6320 services. With the container's root directory in
6321 /var/lib/container/foobar it is now sufficient to run
6322 "systemctl start systemd-nspawn@foobar.service" to boot it.
6324 * systemd-cgls gained a new parameter "--machine" to list only
6325 the processes within a certain container.
6327 * ConditionSecurity= now can check for "apparmor". We still
6328 are lacking checks for SMACK and IMA for this condition
6329 check though. Patches welcome!
6331 * A new configuration file /etc/systemd/sleep.conf has been
6332 added that may be used to configure which kernel operation
6333 systemd is supposed to execute when "suspend", "hibernate"
6334 or "hybrid-sleep" is requested. This makes the new kernel
6335 "freeze" state accessible to the user.
6337 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6338 the passed argument if applicable.
6340 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6341 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6342 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6343 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6344 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6345 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6346 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6351 * The output of 'systemctl list-jobs' got some polishing. The
6352 '--type=' argument may now be passed more than once. A new
6353 command 'systemctl list-sockets' has been added which shows
6354 a list of kernel sockets systemd is listening on with the
6355 socket units they belong to, plus the units these socket
6358 * The experimental libsystemd-bus library got substantial
6359 updates to work in conjunction with the (also experimental)
6360 kdbus kernel project. It works well enough to exchange
6361 messages with some sophistication. Note that kdbus is not
6362 ready yet, and the library is mostly an elaborate test case
6363 for now, and not installable.
6365 * systemd gained a new unit 'systemd-static-nodes.service'
6366 that generates static device nodes earlier during boot, and
6367 can run in conjunction with udev.
6369 * libsystemd-login gained a new call sd_pid_get_user_unit()
6370 to retrieve the user systemd unit a process is running
6371 in. This is useful for systems where systemd is used as
6374 * systemd-nspawn now places all containers in the new /machine
6375 top-level cgroup directory in the name=systemd
6376 hierarchy. libvirt will soon do the same, so that we get a
6377 uniform separation of /system, /user and /machine for system
6378 services, user processes and containers/virtual
6379 machines. This new cgroup hierarchy is also useful to stick
6380 stable names to specific container instances, which can be
6381 recognized later this way (this name may be controlled
6382 via systemd-nspawn's new -M switch). libsystemd-login also
6383 gained a new call sd_pid_get_machine_name() to retrieve the
6384 name of the container/VM a specific process belongs to.
6386 * bootchart can now store its data in the journal.
6388 * libsystemd-journal gained a new call
6389 sd_journal_add_conjunction() for AND expressions to the
6390 matching logic. This can be used to express more complex
6391 logical expressions.
6393 * journactl can now take multiple --unit= and --user-unit=
6396 * The cryptsetup logic now understands the "luks.key=" kernel
6397 command line switch for specifying a file to read the
6398 decryption key from. Also, if a configured key file is not
6399 found the tool will now automatically fall back to prompting
6402 * Python systemd.journal module was updated to wrap recently
6403 added functions from libsystemd-journal. The interface was
6404 changed to bring the low level interface in s.j._Reader
6405 closer to the C API, and the high level interface in
6406 s.j.Reader was updated to wrap and convert all data about
6409 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6410 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6411 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6412 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6413 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6414 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6418 * journalctl --update-catalog now understands a new --root=
6419 option to operate on catalogs found in a different root
6422 * During shutdown after systemd has terminated all running
6423 services a final killing loop kills all remaining left-over
6424 processes. We will now print the name of these processes
6425 when we send SIGKILL to them, since this usually indicates a
6428 * If /etc/crypttab refers to password files stored on
6429 configured mount points automatic dependencies will now be
6430 generated to ensure the specific mount is established first
6431 before the key file is attempted to be read.
6433 * 'systemctl status' will now show information about the
6434 network sockets a socket unit is listening on.
6436 * 'systemctl status' will also shown information about any
6437 drop-in configuration file for units. (Drop-In configuration
6438 files in this context are files such as
6439 /etc/systemd/systemd/foobar.service.d/*.conf)
6441 * systemd-cgtop now optionally shows summed up CPU times of
6442 cgroups. Press '%' while running cgtop to switch between
6443 percentage and absolute mode. This is useful to determine
6444 which cgroups use up the most CPU time over the entire
6445 runtime of the system. systemd-cgtop has also been updated
6446 to be 'pipeable' for processing with further shell tools.
6448 * 'hostnamectl set-hostname' will now allow setting of FQDN
6451 * The formatting and parsing of time span values has been
6452 changed. The parser now understands fractional expressions
6453 such as "5.5h". The formatter will now output fractional
6454 expressions for all time spans under 1min, i.e. "5.123456s"
6455 rather than "5s 123ms 456us". For time spans under 1s
6456 millisecond values are shown, for those under 1ms
6457 microsecond values are shown. This should greatly improve
6458 all time-related output of systemd.
6460 * libsystemd-login and libsystemd-journal gained new
6461 functions for querying the poll() events mask and poll()
6462 timeout value for integration into arbitrary event
6465 * localectl gained the ability to list available X11 keymaps
6466 (models, layouts, variants, options).
6468 * 'systemd-analyze dot' gained the ability to filter for
6469 specific units via shell-style globs, to create smaller,
6470 more useful graphs. I.e. it is now possible to create simple
6471 graphs of all the dependencies between only target units, or
6472 of all units that Avahi has dependencies with.
6474 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6475 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6476 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6477 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6478 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6479 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6480 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6484 * The boot-time readahead implementation for rotating media
6485 will now read the read-ahead data in multiple passes which
6486 consist of all read requests made in equidistant time
6487 intervals. This means instead of strictly reading read-ahead
6488 data in its physical order on disk we now try to find a
6489 middle ground between physical and access time order.
6491 * /etc/os-release files gained a new BUILD_ID= field for usage
6492 on operating systems that provide continuous builds of OS
6495 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6496 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6497 William Douglas, Zbigniew Jędrzejewski-Szmek
6501 * systemd-python gained an API exposing libsystemd-daemon.
6503 * The SMACK setup logic gained support for uploading CIPSO
6506 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6507 ReadOnlyDirectories= and InaccessibleDirectories= has
6508 changed. The private /tmp and /var/tmp directories are now
6509 shared by all processes of a service (which means
6510 ExecStartPre= may now leave data in /tmp that ExecStart= of
6511 the same service can still access). When a service is
6512 stopped its temporary directories are immediately deleted
6513 (normal clean-up with tmpfiles is still done in addition to
6516 * By default, systemd will now set a couple of sysctl
6517 variables in the kernel: the safe sysrq options are turned
6518 on, IP route verification is turned on, and source routing
6519 disabled. The recently added hardlink and softlink
6520 protection of the kernel is turned on. These settings should
6521 be reasonably safe, and good defaults for all new systems.
6523 * The predictable network naming logic may now be turned off
6524 with a new kernel command line switch: net.ifnames=0.
6526 * A new libsystemd-bus module has been added that implements a
6527 pretty complete D-Bus client library. For details see:
6529 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6531 * journald will now explicitly flush the journal files to disk
6532 at the latest 5min after each write. The file will then also
6533 be marked offline until the next write. This should increase
6534 reliability in case of a crash. The synchronization delay
6535 can be configured via SyncIntervalSec= in journald.conf.
6537 * There's a new remote-fs-setup.target unit that can be used
6538 to pull in specific services when at least one remote file
6539 system is to be mounted.
6541 * There are new targets timers.target and paths.target as
6542 canonical targets to pull user timer and path units in
6543 from. This complements sockets.target with a similar
6544 purpose for socket units.
6546 * libudev gained a new call udev_device_set_attribute_value()
6547 to set sysfs attributes of a device.
6549 * The udev daemon now sets the default number of worker
6550 processes executed in parallel based on the number of available
6551 CPUs instead of the amount of available RAM. This is supposed
6552 to provide a more reliable default and limit a too aggressive
6553 parallelism for setups with 1000s of devices connected.
6555 Contributions from: Auke Kok, Colin Walters, Cristian
6556 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6557 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6558 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6559 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6560 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6561 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6562 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6563 Zbigniew Jędrzejewski-Szmek
6567 * Configuration of unit files may now be extended via drop-in
6568 files without having to edit/override the unit files
6569 themselves. More specifically, if the administrator wants to
6570 change one value for a service file foobar.service he can
6571 now do so by dropping in a configuration snippet into
6572 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6573 will load all these snippets and apply them on top of the
6574 main unit configuration file, possibly extending or
6575 overriding its settings. Using these drop-in snippets is
6576 generally nicer than the two earlier options for changing
6577 unit files locally: copying the files from
6578 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6579 them there; or creating a new file in /etc/systemd/system/
6580 that incorporates the original one via ".include". Drop-in
6581 snippets into these .d/ directories can be placed in any
6582 directory systemd looks for units in, and the usual
6583 overriding semantics between /usr/lib, /etc and /run apply
6586 * Most unit file settings which take lists of items can now be
6587 reset by assigning the empty string to them. For example,
6588 normally, settings such as Environment=FOO=BAR append a new
6589 environment variable assignment to the environment block,
6590 each time they are used. By assigning Environment= the empty
6591 string the environment block can be reset to empty. This is
6592 particularly useful with the .d/*.conf drop-in snippets
6593 mentioned above, since this adds the ability to reset list
6594 settings from vendor unit files via these drop-ins.
6596 * systemctl gained a new "list-dependencies" command for
6597 listing the dependencies of a unit recursively.
6599 * Inhibitors are now honored and listed by "systemctl
6600 suspend", "systemctl poweroff" (and similar) too, not only
6601 GNOME. These commands will also list active sessions by
6604 * Resource limits (as exposed by the various control group
6605 controllers) can now be controlled dynamically at runtime
6606 for all units. More specifically, you can now use a command
6607 like "systemctl set-cgroup-attr foobar.service cpu.shares
6608 2000" to alter the CPU shares a specific service gets. These
6609 settings are stored persistently on disk, and thus allow the
6610 administrator to easily adjust the resource usage of
6611 services with a few simple commands. This dynamic resource
6612 management logic is also available to other programs via the
6613 bus. Almost any kernel cgroup attribute and controller is
6616 * systemd-vconsole-setup will now copy all font settings to
6617 all allocated VTs, where it previously applied them only to
6620 * libsystemd-login gained the new sd_session_get_tty() API
6623 * This release drops support for a few legacy or
6624 distribution-specific LSB facility names when parsing init
6625 scripts: $x-display-manager, $mail-transfer-agent,
6626 $mail-transport-agent, $mail-transfer-agent, $smtp,
6627 $null. Also, the mail-transfer-agent.target unit backing
6628 this has been removed. Distributions which want to retain
6629 compatibility with this should carry the burden for
6630 supporting this themselves and patch support for these back
6631 in, if they really need to. Also, the facilities $syslog and
6632 $local_fs are now ignored, since systemd does not support
6633 early-boot LSB init scripts anymore, and these facilities
6634 are implied anyway for normal services. syslog.target has
6637 * There are new bus calls on PID1's Manager object for
6638 cancelling jobs, and removing snapshot units. Previously,
6639 both calls were only available on the Job and Snapshot
6642 * systemd-journal-gatewayd gained SSL support.
6644 * The various "environment" files, such as /etc/locale.conf
6645 now support continuation lines with a backslash ("\") as
6646 last character in the line, similarly in style (but different)
6647 to how this is supported in shells.
6649 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6650 now implicitly appended to every log entry logged. systemctl
6651 has been updated to filter by this field when operating on a
6652 user systemd instance.
6654 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6655 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6656 the container. This makes it easier to boot unmodified
6657 Fedora systems in a container, which however still requires
6658 audit=0 to be passed on the kernel command line. Auditing in
6659 kernel and userspace is unfortunately still too broken in
6660 context of containers, hence we recommend compiling it out
6661 of the kernel or using audit=0. Hopefully this will be fixed
6662 one day for good in the kernel.
6664 * nspawn gained the new --bind= and --bind-ro= parameters to
6665 bind mount specific directories from the host into the
6668 * nspawn will now mount its own devpts file system instance
6669 into the container, in order not to leak pty devices from
6670 the host into the container.
6672 * systemd will now read the firmware boot time performance
6673 information from the EFI variables, if the used boot loader
6674 supports this, and takes it into account for boot performance
6675 analysis via "systemd-analyze". This is currently supported
6676 only in conjunction with Gummiboot, but could be supported
6677 by other boot loaders too. For details see:
6679 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6681 * A new generator has been added that automatically mounts the
6682 EFI System Partition (ESP) to /boot, if that directory
6683 exists, is empty, and no other file system has been
6684 configured to be mounted there.
6686 * logind will now send out PrepareForSleep(false) out
6687 unconditionally, after coming back from suspend. This may be
6688 used by applications as asynchronous notification for
6689 system resume events.
6691 * "systemctl unlock-sessions" has been added, that allows
6692 unlocking the screens of all user sessions at once, similar
6693 to how "systemctl lock-sessions" already locked all users
6694 sessions. This is backed by a new D-Bus call UnlockSessions().
6696 * "loginctl seat-status" will now show the master device of a
6697 seat. (i.e. the device of a seat that needs to be around for
6698 the seat to be considered available, usually the graphics
6701 * tmpfiles gained a new "X" line type, that allows
6702 configuration of files and directories (with wildcards) that
6703 shall be excluded from automatic cleanup ("aging").
6705 * udev default rules set the device node permissions now only
6706 at "add" events, and do not change them any longer with a
6707 later "change" event.
6709 * The log messages for lid events and power/sleep keypresses
6710 now carry a message ID.
6712 * We now have a substantially larger unit test suite, but this
6713 continues to be work in progress.
6715 * udevadm hwdb gained a new --root= parameter to change the
6716 root directory to operate relative to.
6718 * logind will now issue a background sync() request to the kernel
6719 early at shutdown, so that dirty buffers are flushed to disk early
6720 instead of at the last moment, in order to optimize shutdown
6723 * A new bootctl tool has been added that is an interface for
6724 certain boot loader operations. This is currently a preview
6725 and is likely to be extended into a small mechanism daemon
6726 like timedated, localed, hostnamed, and can be used by
6727 graphical UIs to enumerate available boot options, and
6728 request boot into firmware operations.
6730 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6731 the rest of the package. It also has been updated to work
6732 correctly in initrds.
6734 * Policykit previously has been runtime optional, and is now
6735 also compile time optional via a configure switch.
6737 * systemd-analyze has been reimplemented in C. Also "systemctl
6738 dot" has moved into systemd-analyze.
6740 * "systemctl status" with no further parameters will now print
6741 the status of all active or failed units.
6743 * Operations such as "systemctl start" can now be executed
6744 with a new mode "--irreversible" which may be used to queue
6745 operations that cannot accidentally be reversed by a later
6746 job queuing. This is by default used to make shutdown
6747 requests more robust.
6749 * The Python API of systemd now gained a new module for
6750 reading journal files.
6752 * A new tool kernel-install has been added that can install
6753 kernel images according to the Boot Loader Specification:
6755 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6757 * Boot time console output has been improved to provide
6758 animated boot time output for hanging jobs.
6760 * A new tool systemd-activate has been added which can be used
6761 to test socket activation with, directly from the command
6762 line. This should make it much easier to test and debug
6763 socket activation in daemons.
6765 * journalctl gained a new "--reverse" (or -r) option to show
6766 journal output in reverse order (i.e. newest line first).
6768 * journalctl gained a new "--pager-end" (or -e) option to jump
6769 to immediately jump to the end of the journal in the
6770 pager. This is only supported in conjunction with "less".
6772 * journalctl gained a new "--user-unit=" option, that works
6773 similarly to "--unit=" but filters for user units rather than
6776 * A number of unit files to ease adoption of systemd in
6777 initrds has been added. This moves some minimal logic from
6778 the various initrd implementations into systemd proper.
6780 * The journal files are now owned by a new group
6781 "systemd-journal", which exists specifically to allow access
6782 to the journal, and nothing else. Previously, we used the
6783 "adm" group for that, which however possibly covers more
6784 than just journal/log file access. This new group is now
6785 already used by systemd-journal-gatewayd to ensure this
6786 daemon gets access to the journal files and as little else
6787 as possible. Note that "make install" will also set FS ACLs
6788 up for /var/log/journal to give "adm" and "wheel" read
6789 access to it, in addition to "systemd-journal" which owns
6790 the journal files. We recommend that packaging scripts also
6791 add read access to "adm" + "wheel" to /var/log/journal, and
6792 all existing/future journal files. To normal users and
6793 administrators little changes, however packagers need to
6794 ensure to create the "systemd-journal" system group at
6795 package installation time.
6797 * The systemd-journal-gatewayd now runs as unprivileged user
6798 systemd-journal-gateway:systemd-journal-gateway. Packaging
6799 scripts need to create these system user/group at
6802 * timedated now exposes a new boolean property CanNTP that
6803 indicates whether a local NTP service is available or not.
6805 * systemd-detect-virt will now also detect xen PVs
6807 * The pstore file system is now mounted by default, if it is
6810 * In addition to the SELinux and IMA policies we will now also
6811 load SMACK policies at early boot.
6813 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6814 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6815 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6816 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6817 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6818 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6819 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6820 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6821 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6822 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6823 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6824 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6825 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6826 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6830 * Timer units now support calendar time events in addition to
6831 monotonic time events. That means you can now trigger a unit
6832 based on a calendar time specification such as "Thu,Fri
6833 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6834 or fifth day of any month of the year 2013, given that it is
6835 a thursday or friday. This brings timer event support
6836 considerably closer to cron's capabilities. For details on
6837 the supported calendar time specification language see
6840 * udev now supports a number of different naming policies for
6841 network interfaces for predictable names, and a combination
6842 of these policies is now the default. Please see this wiki
6843 document for details:
6845 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6847 * Auke Kok's bootchart implementation has been added to the
6848 systemd tree. It is an optional component that can graph the
6849 boot in quite some detail. It is one of the best bootchart
6850 implementations around and minimal in its code and
6853 * nss-myhostname has been integrated into the systemd source
6854 tree. nss-myhostname guarantees that the local hostname
6855 always stays resolvable via NSS. It has been a weak
6856 requirement of systemd-hostnamed since a long time, and
6857 since its code is actually trivial we decided to just
6858 include it in systemd's source tree. It can be turned off
6859 with a configure switch.
6861 * The read-ahead logic is now capable of properly detecting
6862 whether a btrfs file system is on SSD or rotating media, in
6863 order to optimize the read-ahead scheme. Previously, it was
6864 only capable of detecting this on traditional file systems
6867 * In udev, additional device properties are now read from the
6868 IAB in addition to the OUI database. Also, Bluetooth company
6869 identities are attached to the devices as well.
6871 * In service files %U may be used as specifier that is
6872 replaced by the configured user name of the service.
6874 * nspawn may now be invoked without a controlling TTY. This
6875 makes it suitable for invocation as its own service. This
6876 may be used to set up a simple containerized server system
6877 using only core OS tools.
6879 * systemd and nspawn can now accept socket file descriptors
6880 when they are started for socket activation. This enables
6881 implementation of socket activated nspawn
6882 containers. i.e. think about autospawning an entire OS image
6883 when the first SSH or HTTP connection is received. We expect
6884 that similar functionality will also be added to libvirt-lxc
6887 * journalctl will now suppress ANSI color codes when
6888 presenting log data.
6890 * systemctl will no longer show control group information for
6891 a unit if the control group is empty anyway.
6893 * logind can now automatically suspend/hibernate/shutdown the
6896 * /etc/machine-info and hostnamed now also expose the chassis
6897 type of the system. This can be used to determine whether
6898 the local system is a laptop, desktop, handset or
6899 tablet. This information may either be configured by the
6900 user/vendor or is automatically determined from ACPI and DMI
6901 information if possible.
6903 * A number of PolicyKit actions are now bound together with
6904 "imply" rules. This should simplify creating UIs because
6905 many actions will now authenticate similar ones as well.
6907 * Unit files learnt a new condition ConditionACPower= which
6908 may be used to conditionalize a unit depending on whether an
6909 AC power source is connected or not, of whether the system
6910 is running on battery power.
6912 * systemctl gained a new "is-failed" verb that may be used in
6913 shell scripts and suchlike to check whether a specific unit
6914 is in the "failed" state.
6916 * The EnvironmentFile= setting in unit files now supports file
6917 globbing, and can hence be used to easily read a number of
6918 environment files at once.
6920 * systemd will no longer detect and recognize specific
6921 distributions. All distribution-specific #ifdeffery has been
6922 removed, systemd is now fully generic and
6923 distribution-agnostic. Effectively, not too much is lost as
6924 a lot of the code is still accessible via explicit configure
6925 switches. However, support for some distribution specific
6926 legacy configuration file formats has been dropped. We
6927 recommend distributions to simply adopt the configuration
6928 files everybody else uses now and convert the old
6929 configuration from packaging scripts. Most distributions
6930 already did that. If that's not possible or desirable,
6931 distributions are welcome to forward port the specific
6932 pieces of code locally from the git history.
6934 * When logging a message about a unit systemd will now always
6935 log the unit name in the message meta data.
6937 * localectl will now also discover system locale data that is
6938 not stored in locale archives, but directly unpacked.
6940 * logind will no longer unconditionally use framebuffer
6941 devices as seat masters, i.e. as devices that are required
6942 to be existing before a seat is considered preset. Instead,
6943 it will now look for all devices that are tagged as
6944 "seat-master" in udev. By default, framebuffer devices will
6945 be marked as such, but depending on local systems, other
6946 devices might be marked as well. This may be used to
6947 integrate graphics cards using closed source drivers (such
6948 as NVidia ones) more nicely into logind. Note however, that
6949 we recommend using the open source NVidia drivers instead,
6950 and no udev rules for the closed-source drivers will be
6951 shipped from us upstream.
6953 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6954 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6955 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6956 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6957 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6958 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6959 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6960 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6961 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6962 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6963 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6964 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6969 * udev gained support for loading additional device properties
6970 from an indexed database that is keyed by vendor/product IDs
6971 and similar device identifiers. For the beginning this
6972 "hwdb" is populated with data from the well-known PCI and
6973 USB database, but also includes PNP, ACPI and OID data. In
6974 the longer run this indexed database shall grow into
6975 becoming the one central database for non-essential
6976 userspace device metadata. Previously, data from the PCI/USB
6977 database was only attached to select devices, since the
6978 lookup was a relatively expensive operation due to O(n) time
6979 complexity (with n being the number of entries in the
6980 database). Since this is now O(1), we decided to add in this
6981 data for all devices where this is available, by
6982 default. Note that the indexed database needs to be rebuilt
6983 when new data files are installed. To achieve this you need
6984 to update your packaging scripts to invoke "udevadm hwdb
6985 --update" after installation of hwdb data files. For
6986 RPM-based distributions we introduced the new
6987 %udev_hwdb_update macro for this purpose.
6989 * The Journal gained support for the "Message Catalog", an
6990 indexed database to link up additional information with
6991 journal entries. For further details please check:
6993 https://www.freedesktop.org/wiki/Software/systemd/catalog
6995 The indexed message catalog database also needs to be
6996 rebuilt after installation of message catalog files. Use
6997 "journalctl --update-catalog" for this. For RPM-based
6998 distributions we introduced the %journal_catalog_update
6999 macro for this purpose.
7001 * The Python Journal bindings gained support for the standard
7002 Python logging framework.
7004 * The Journal API gained new functions for checking whether
7005 the underlying file system of a journal file is capable of
7006 properly reporting file change notifications, or whether
7007 applications that want to reflect journal changes "live"
7008 need to recheck journal files continuously in appropriate
7011 * It is now possible to set the "age" field for tmpfiles
7012 entries to 0, indicating that files matching this entry
7013 shall always be removed when the directories are cleaned up.
7015 * coredumpctl gained a new "gdb" verb which invokes gdb
7016 right-away on the selected coredump.
7018 * There's now support for "hybrid sleep" on kernels that
7019 support this, in addition to "suspend" and "hibernate". Use
7020 "systemctl hybrid-sleep" to make use of this.
7022 * logind's HandleSuspendKey= setting (and related settings)
7023 now gained support for a new "lock" setting to simply
7024 request the screen lock on all local sessions, instead of
7025 actually executing a suspend or hibernation.
7027 * systemd will now mount the EFI variables file system by
7030 * Socket units now gained support for configuration of the
7031 SMACK security label.
7033 * timedatectl will now output the time of the last and next
7034 daylight saving change.
7036 * We dropped support for various legacy and distro-specific
7037 concepts, such as insserv, early-boot SysV services
7038 (i.e. those for non-standard runlevels such as 'b' or 'S')
7039 or ArchLinux /etc/rc.conf support. We recommend the
7040 distributions who still need support this to either continue
7041 to maintain the necessary patches downstream, or find a
7042 different solution. (Talk to us if you have questions!)
7044 * Various systemd components will now bypass PolicyKit checks
7045 for root and otherwise handle properly if PolicyKit is not
7046 found to be around. This should fix most issues for
7047 PolicyKit-less systems. Quite frankly this should have been
7048 this way since day one. It is absolutely our intention to
7049 make systemd work fine on PolicyKit-less systems, and we
7050 consider it a bug if something does not work as it should if
7051 PolicyKit is not around.
7053 * For embedded systems it is now possible to build udev and
7054 systemd without blkid and/or kmod support.
7056 * "systemctl switch-root" is now capable of switching root
7057 more than once. I.e. in addition to transitions from the
7058 initrd to the host OS it is now possible to transition to
7059 further OS images from the host. This is useful to implement
7060 offline updating tools.
7062 * Various other additions have been made to the RPM macros
7063 shipped with systemd. Use %udev_rules_update() after
7064 installing new udev rules files. %_udevhwdbdir,
7065 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7066 %_sysctldir are now available which resolve to the right
7067 directories for packages to place various data files in.
7069 * journalctl gained the new --full switch (in addition to
7070 --all, to disable ellipsation for long messages.
7072 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7073 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7074 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7075 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7076 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7077 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7078 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7079 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7080 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7084 * journalctl gained new --since= and --until= switches to
7085 filter by time. It also now supports nice filtering for
7086 units via --unit=/-u.
7088 * Type=oneshot services may use ExecReload= and do the
7091 * The journal daemon now supports time-based rotation and
7092 vacuuming, in addition to the usual disk-space based
7095 * The journal will now index the available field values for
7096 each field name. This enables clients to show pretty drop
7097 downs of available match values when filtering. The bash
7098 completion of journalctl has been updated
7099 accordingly. journalctl gained a new switch -F to list all
7100 values a certain field takes in the journal database.
7102 * More service events are now written as structured messages
7103 to the journal, and made recognizable via message IDs.
7105 * The timedated, localed and hostnamed mini-services which
7106 previously only provided support for changing time, locale
7107 and hostname settings from graphical DEs such as GNOME now
7108 also have a minimal (but very useful) text-based client
7109 utility each. This is probably the nicest way to changing
7110 these settings from the command line now, especially since
7111 it lists available options and is fully integrated with bash
7114 * There's now a new tool "systemd-coredumpctl" to list and
7115 extract coredumps from the journal.
7117 * We now install a README each in /var/log/ and
7118 /etc/rc.d/init.d explaining where the system logs and init
7119 scripts went. This hopefully should help folks who go to
7120 that dirs and look into the otherwise now empty void and
7121 scratch their heads.
7123 * When user-services are invoked (by systemd --user) the
7124 $MANAGERPID env var is set to the PID of systemd.
7126 * SIGRTMIN+24 when sent to a --user instance will now result
7127 in immediate termination of systemd.
7129 * gatewayd received numerous feature additions such as a
7130 "follow" mode, for live syncing and filtering.
7132 * browse.html now allows filtering and showing detailed
7133 information on specific entries. Keyboard navigation and
7134 mouse screen support has been added.
7136 * gatewayd/journalctl now supports HTML5/JSON
7137 Server-Sent-Events as output.
7139 * The SysV init script compatibility logic will now
7140 heuristically determine whether a script supports the
7141 "reload" verb, and only then make this available as
7144 * "systemctl status --follow" has been removed, use "journalctl
7147 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7148 have been removed since they are hardly useful to be
7151 * And I'd like to take the opportunity to specifically mention
7152 Zbigniew for his great contributions. Zbigniew, you rock!
7154 Contributions from: Andrew Eikum, Christian Hesse, Colin
7155 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7156 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7157 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7158 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7159 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7160 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7164 * If /etc/vconsole.conf is non-existent or empty we will no
7165 longer load any console font or key map at boot by
7166 default. Instead the kernel defaults will be left
7167 intact. This is definitely the right thing to do, as no
7168 configuration should mean no configuration, and hard-coding
7169 font names that are different on all archs is probably a bad
7170 idea. Also, the kernel default key map and font should be
7171 good enough for most cases anyway, and mostly identical to
7172 the userspace fonts/key maps we previously overloaded them
7173 with. If distributions want to continue to default to a
7174 non-kernel font or key map they should ship a default
7175 /etc/vconsole.conf with the appropriate contents.
7177 Contributions from: Colin Walters, Daniel J Walsh, Dave
7178 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7179 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7183 * journalctl gained a new --cursor= switch to show entries
7184 starting from the specified location in the journal.
7186 * We now enforce a size limit on journal entry fields exported
7187 with "-o json" in journalctl. Fields larger than 4K will be
7188 assigned null. This can be turned off with --all.
7190 * An (optional) journal gateway daemon is now available as
7191 "systemd-journal-gatewayd.service". This service provides
7192 access to the journal via HTTP and JSON. This functionality
7193 will be used to implement live log synchronization in both
7194 pull and push modes, but has various other users too, such
7195 as easy log access for debugging of embedded devices. Right
7196 now it is already useful to retrieve the journal via HTTP:
7198 # systemctl start systemd-journal-gatewayd.service
7199 # wget http://localhost:19531/entries
7201 This will download the journal contents in a
7202 /var/log/messages compatible format. The same as JSON:
7204 # curl -H"Accept: application/json" http://localhost:19531/entries
7206 This service is also accessible via a web browser where a
7207 single static HTML5 app is served that uses the JSON logic
7208 to enable the user to do some basic browsing of the
7209 journal. This will be extended later on. Here's an example
7210 screenshot of this app in its current state:
7212 http://0pointer.de/public/journal-gatewayd
7214 Contributions from: Kay Sievers, Lennart Poettering, Robert
7215 Milasan, Tom Gundersen
7219 * The bash completion logic is now available for journalctl
7222 * We do not mount the "cpuset" controller anymore together with
7223 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7224 started if no parameters are assigned to it. "cpuset" hence
7225 broke code that assumed it could create "cpu" groups and
7228 * journalctl -f will now subscribe to terminal size changes,
7229 and line break accordingly.
7231 Contributions from: Dave Reisner, Kay Sievers, Lennart
7232 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7236 * nspawn will now create a symlink /etc/localtime in the
7237 container environment, copying the host's timezone
7238 setting. Previously this has been done via a bind mount, but
7239 since symlinks cannot be bind mounted this has now been
7240 changed to create/update the appropriate symlink.
7242 * journalctl -n's line number argument is now optional, and
7243 will default to 10 if omitted.
7245 * journald will now log the maximum size the journal files may
7246 take up on disk. This is particularly useful if the default
7247 built-in logic of determining this parameter from the file
7248 system size is used. Use "systemctl status
7249 systemd-journald.service" to see this information.
7251 * The multi-seat X wrapper tool has been stripped down. As X
7252 is now capable of enumerating graphics devices via udev in a
7253 seat-aware way the wrapper is not strictly necessary
7254 anymore. A stripped down temporary stop-gap is still shipped
7255 until the upstream display managers have been updated to
7256 fully support the new X logic. Expect this wrapper to be
7257 removed entirely in one of the next releases.
7259 * HandleSleepKey= in logind.conf has been split up into
7260 HandleSuspendKey= and HandleHibernateKey=. The old setting
7261 is not available anymore. X11 and the kernel are
7262 distinguishing between these keys and we should too. This
7263 also means the inhibition lock for these keys has been split
7266 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7267 Poettering, Lukas Nykryn, Václav Pavlín
7271 * Whenever a unit changes state we will now log this to the
7272 journal and show along the unit's own log output in
7275 * ConditionPathIsMountPoint= can now properly detect bind
7276 mount points too. (Previously, a bind mount of one file
7277 system to another place in the same file system could not be
7278 detected as mount, since they shared struct stat's st_dev
7281 * We will now mount the cgroup controllers cpu, cpuacct,
7282 cpuset and the controllers net_cls, net_prio together by
7285 * nspawn containers will now have a virtualized boot
7286 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7287 over with a randomized ID at container initialization). This
7288 has the effect of making "journalctl -b" do the right thing
7291 * The JSON output journal serialization has been updated not
7292 to generate "endless" list objects anymore, but rather one
7293 JSON object per line. This is more in line how most JSON
7294 parsers expect JSON objects. The new output mode
7295 "json-pretty" has been added to provide similar output, but
7296 neatly aligned for readability by humans.
7298 * We dropped all explicit sync() invocations in the shutdown
7299 code. The kernel does this implicitly anyway in the kernel
7300 reboot() syscall. halt(8)'s -n option is now a compatibility
7303 * We now support virtualized reboot() in containers, as
7304 supported by newer kernels. We will fall back to exit() if
7305 CAP_SYS_REBOOT is not available to the container. Also,
7306 nspawn makes use of this now and will actually reboot the
7307 container if the containerized OS asks for that.
7309 * journalctl will only show local log output by default
7310 now. Use --merge (-m) to show remote log output, too.
7312 * libsystemd-journal gained the new sd_journal_get_usage()
7313 call to determine the current disk usage of all journal
7314 files. This is exposed in the new "journalctl --disk-usage"
7317 * journald gained a new configuration setting SplitMode= in
7318 journald.conf which may be used to control how user journals
7319 are split off. See journald.conf(5) for details.
7321 * A new condition type ConditionFileNotEmpty= has been added.
7323 * tmpfiles' "w" lines now support file globbing, to write
7324 multiple files at once.
7326 * We added Python bindings for the journal submission
7327 APIs. More Python APIs for a number of selected APIs will
7328 likely follow. Note that we intend to add native bindings
7329 only for the Python language, as we consider it common
7330 enough to deserve bindings shipped within systemd. There are
7331 various projects outside of systemd that provide bindings
7332 for languages such as PHP or Lua.
7334 * Many conditions will now resolve specifiers such as %i. In
7335 addition, PathChanged= and related directives of .path units
7336 now support specifiers as well.
7338 * There's now a new RPM macro definition for the system preset
7341 * journald will now warn if it ca not forward a message to the
7342 syslog daemon because its socket is full.
7344 * timedated will no longer write or process /etc/timezone,
7345 except on Debian. As we do not support late mounted /usr
7346 anymore /etc/localtime always being a symlink is now safe,
7347 and hence the information in /etc/timezone is not necessary
7350 * logind will now always reserve one VT for a text getty (VT6
7351 by default). Previously if more than 6 X sessions where
7352 started they took up all the VTs with auto-spawned gettys,
7353 so that no text gettys were available anymore.
7355 * udev will now automatically inform the btrfs kernel logic
7356 about btrfs RAID components showing up. This should make
7357 simple hotplug based btrfs RAID assembly work.
7359 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7360 (but not for its children which will stay at the kernel
7361 default). This should allow setups with a lot more listening
7364 * systemd will now always pass the configured timezone to the
7365 kernel at boot. timedated will do the same when the timezone
7368 * logind's inhibition logic has been updated. By default,
7369 logind will now handle the lid switch, the power and sleep
7370 keys all the time, even in graphical sessions. If DEs want
7371 to handle these events on their own they should take the new
7372 handle-power-key, handle-sleep-key and handle-lid-switch
7373 inhibitors during their runtime. A simple way to achieve
7374 that is to invoke the DE wrapped in an invocation of:
7376 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7378 * Access to unit operations is now checked via SELinux taking
7379 the unit file label and client process label into account.
7381 * systemd will now notify the administrator in the journal
7382 when he over-mounts a non-empty directory.
7384 * There are new specifiers that are resolved in unit files,
7385 for the host name (%H), the machine ID (%m) and the boot ID
7388 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7389 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7390 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7391 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7392 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7393 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7394 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7398 * Support for reading structured kernel messages from
7399 /dev/kmsg has now been added and is enabled by default.
7401 * Support for reading kernel messages from /proc/kmsg has now
7402 been removed. If you want kernel messages in the journal
7403 make sure to run a recent kernel (>= 3.5) that supports
7404 reading structured messages from /dev/kmsg (see
7405 above). /proc/kmsg is now exclusive property of classic
7406 syslog daemons again.
7408 * The libudev API gained the new
7409 udev_device_new_from_device_id() call.
7411 * The logic for file system namespace (ReadOnlyDirectory=,
7412 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7413 require pivot_root() anymore. This means fewer temporary
7414 directories are created below /tmp for this feature.
7416 * nspawn containers will now see and receive all submounts
7417 made on the host OS below the root file system of the
7420 * Forward Secure Sealing is now supported for Journal files,
7421 which provide cryptographical sealing of journal files so
7422 that attackers cannot alter log history anymore without this
7423 being detectable. Lennart will soon post a blog story about
7424 this explaining it in more detail.
7426 * There are two new service settings RestartPreventExitStatus=
7427 and SuccessExitStatus= which allow configuration of exit
7428 status (exit code or signal) which will be excepted from the
7429 restart logic, resp. consider successful.
7431 * journalctl gained the new --verify switch that can be used
7432 to check the integrity of the structure of journal files and
7433 (if Forward Secure Sealing is enabled) the contents of
7436 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7437 and similar symlinks pre-created. This makes running shells
7438 as container init process a lot more fun.
7440 * The fstab support can now handle PARTUUID= and PARTLABEL=
7443 * A new ConditionHost= condition has been added to match
7444 against the hostname (with globs) and machine ID. This is
7445 useful for clusters where a single OS image is used to
7446 provision a large number of hosts which shall run slightly
7447 different sets of services.
7449 * Services which hit the restart limit will now be placed in a
7452 Contributions from: Bertram Poettering, Dave Reisner, Huang
7453 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7454 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7458 * When running in --user mode systemd will now become a
7459 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7460 tree a lot more organized.
7462 * A new PartOf= unit dependency type has been introduced that
7463 may be used to group services in a natural way.
7465 * "systemctl enable" may now be used to enable instances of
7468 * journalctl now prints error log levels in red, and
7469 warning/notice log levels in bright white. It also supports
7470 filtering by log level now.
7472 * cgtop gained a new -n switch (similar to top), to configure
7473 the maximum number of iterations to run for. It also gained
7474 -b, to run in batch mode (accepting no input).
7476 * The suffix ".service" may now be omitted on most systemctl
7477 command lines involving service unit names.
7479 * There's a new bus call in logind to lock all sessions, as
7480 well as a loginctl verb for it "lock-sessions".
7482 * libsystemd-logind.so gained a new call sd_journal_perror()
7483 that works similar to libc perror() but logs to the journal
7484 and encodes structured information about the error number.
7486 * /etc/crypttab entries now understand the new keyfile-size=
7489 * shutdown(8) now can send a (configurable) wall message when
7490 a shutdown is cancelled.
7492 * The mount propagation mode for the root file system will now
7493 default to "shared", which is useful to make containers work
7494 nicely out-of-the-box so that they receive new mounts from
7495 the host. This can be undone locally by running "mount
7496 --make-rprivate /" if needed.
7498 * The prefdm.service file has been removed. Distributions
7499 should maintain this unit downstream if they intend to keep
7500 it around. However, we recommend writing normal unit files
7501 for display managers instead.
7503 * Since systemd is a crucial part of the OS we will now
7504 default to a number of compiler switches that improve
7505 security (hardening) such as read-only relocations, stack
7506 protection, and suchlike.
7508 * The TimeoutSec= setting for services is now split into
7509 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7510 of individual time outs for the start and the stop phase of
7513 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7514 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7515 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7516 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7517 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7518 Gundersen, Zbigniew Jędrzejewski-Szmek
7522 * The journal and id128 C APIs are now fully documented as man
7525 * Extra safety checks have been added when transitioning from
7526 the initial RAM disk to the main system to avoid accidental
7529 * /etc/crypttab entries now understand the new keyfile-offset=
7532 * systemctl -t can now be used to filter by unit load state.
7534 * The journal C API gained the new sd_journal_wait() call to
7535 make writing synchronous journal clients easier.
7537 * journalctl gained the new -D switch to show journals from a
7540 * journalctl now displays a special marker between log
7541 messages of two different boots.
7543 * The journal is now explicitly flushed to /var via a service
7544 systemd-journal-flush.service, rather than implicitly simply
7545 by seeing /var/log/journal to be writable.
7547 * journalctl (and the journal C APIs) can now match for much
7548 more complex expressions, with alternatives and
7551 * When transitioning from the initial RAM disk to the main
7552 system we will now kill all processes in a killing spree to
7553 ensure no processes stay around by accident.
7555 * Three new specifiers may be used in unit files: %u, %h, %s
7556 resolve to the user name, user home directory resp. user
7557 shell. This is useful for running systemd user instances.
7559 * We now automatically rotate journal files if their data
7560 object hash table gets a fill level > 75%. We also size the
7561 hash table based on the configured maximum file size. This
7562 together should lower hash collisions drastically and thus
7563 speed things up a bit.
7565 * journalctl gained the new "--header" switch to introspect
7566 header data of journal files.
7568 * A new setting SystemCallFilters= has been added to services
7569 which may be used to apply blacklists or whitelists to
7570 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7572 * nspawn gained a new --link-journal= switch (and quicker: -j)
7573 to link the container journal with the host. This makes it
7574 very easy to centralize log viewing on the host for all
7575 guests while still keeping the journal files separated.
7577 * Many bugfixes and optimizations
7579 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7580 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7581 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7586 * Several tools now understand kernel command line arguments,
7587 which are only read when run in an initial RAM disk. They
7588 usually follow closely their normal counterparts, but are
7591 * There's a new tool to analyze the readahead files that are
7592 automatically generated at boot. Use:
7594 /usr/lib/systemd/systemd-readahead analyze /.readahead
7596 * We now provide an early debug shell on tty9 if this enabled. Use:
7598 systemctl enable debug-shell.service
7600 * All plymouth related units have been moved into the Plymouth
7601 package. Please make sure to upgrade your Plymouth version
7604 * systemd-tmpfiles now supports getting passed the basename of
7605 a configuration file only, in which case it will look for it
7606 in all appropriate directories automatically.
7608 * udevadm info now takes a /dev or /sys path as argument, and
7609 does the right thing. Example:
7611 udevadm info /dev/sda
7612 udevadm info /sys/class/block/sda
7614 * systemctl now prints a warning if a unit is stopped but a
7615 unit that might trigger it continues to run. Example: a
7616 service is stopped but the socket that activates it is left
7619 * "systemctl status" will now mention if the log output was
7620 shortened due to rotation since a service has been started.
7622 * The journal API now exposes functions to determine the
7623 "cutoff" times due to rotation.
7625 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7626 immediately flushing of runtime logs to /var if possible,
7627 resp. for triggering immediate rotation of the journal
7630 * It is now considered an error if a service is attempted to
7631 be stopped that is not loaded.
7633 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7635 * systemd-analyze now supports Python 3
7637 * tmpfiles now supports cleaning up directories via aging
7638 where the first level dirs are always kept around but
7639 directories beneath it automatically aged. This is enabled
7640 by prefixing the age field with '~'.
7642 * Seat objects now expose CanGraphical, CanTTY properties
7643 which is required to deal with very fast bootups where the
7644 display manager might be running before the graphics drivers
7645 completed initialization.
7647 * Seat objects now expose a State property.
7649 * We now include RPM macros for service enabling/disabling
7650 based on the preset logic. We recommend RPM based
7651 distributions to make use of these macros if possible. This
7652 makes it simpler to reuse RPM spec files across
7655 * We now make sure that the collected systemd unit name is
7656 always valid when services log to the journal via
7659 * There's a new man page kernel-command-line(7) detailing all
7660 command line options we understand.
7662 * The fstab generator may now be disabled at boot by passing
7663 fstab=0 on the kernel command line.
7665 * A new kernel command line option modules-load= is now understood
7666 to load a specific kernel module statically, early at boot.
7668 * Unit names specified on the systemctl command line are now
7669 automatically escaped as needed. Also, if file system or
7670 device paths are specified they are automatically turned
7671 into the appropriate mount or device unit names. Example:
7673 systemctl status /home
7674 systemctl status /dev/sda
7676 * The SysVConsole= configuration option has been removed from
7677 system.conf parsing.
7679 * The SysV search path is no longer exported on the D-Bus
7682 * The Names= option has been removed from unit file parsing.
7684 * There's a new man page bootup(7) detailing the boot process.
7686 * Every unit and every generator we ship with systemd now
7687 comes with full documentation. The self-explanatory boot is
7690 * A couple of services gained "systemd-" prefixes in their
7691 name if they wrap systemd code, rather than only external
7692 code. Among them fsck@.service which is now
7693 systemd-fsck@.service.
7695 * The HaveWatchdog property has been removed from the D-Bus
7698 * systemd.confirm_spawn= on the kernel command line should now
7701 * There's a new man page crypttab(5) which details all options
7702 we actually understand.
7704 * systemd-nspawn gained a new --capability= switch to pass
7705 additional capabilities to the container.
7707 * timedated will now read known NTP implementation unit names
7708 from /usr/lib/systemd/ntp-units.d/*.list,
7709 systemd-timedated-ntp.target has been removed.
7711 * journalctl gained a new switch "-b" that lists log data of
7712 the current boot only.
7714 * The notify socket is in the abstract namespace again, in
7715 order to support daemons which chroot() at start-up.
7717 * There is a new Storage= configuration option for journald
7718 which allows configuration of where log data should go. This
7719 also provides a way to disable journal logging entirely, so
7720 that data collected is only forwarded to the console, the
7721 kernel log buffer or another syslog implementation.
7723 * Many bugfixes and optimizations
7725 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7726 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7727 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7728 Shawn Landden, Tom Gundersen
7732 * "systemctl help <unit>" now shows the man page if one is
7735 * Several new man pages have been added.
7737 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7738 MaxLevelConsole= can now be specified in
7739 journald.conf. These options allow reducing the amount of
7740 data stored on disk or forwarded by the log level.
7742 * TimerSlackNSec= can now be specified in system.conf for
7743 PID1. This allows system-wide power savings.
7745 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7746 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7751 * logind is now capable of (optionally) handling power and
7752 sleep keys as well as the lid switch.
7754 * journalctl now understands the syntax "journalctl
7755 /usr/bin/avahi-daemon" to get all log output of a specific
7758 * CapabilityBoundingSet= in system.conf now also influences
7759 the capability bound set of usermode helpers of the kernel.
7761 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7762 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7763 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7764 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7768 * Note that we skipped 139 releases here in order to set the
7769 new version to something that is greater than both udev's
7770 and systemd's most recent version number.
7772 * udev: all udev sources are merged into the systemd source tree now.
7773 All future udev development will happen in the systemd tree. It
7774 is still fully supported to use the udev daemon and tools without
7775 systemd running, like in initramfs or other init systems. Building
7776 udev though, will require the *build* of the systemd tree, but
7777 udev can be properly *run* without systemd.
7779 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7780 should be used to create dead device nodes as workarounds for broken
7783 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7784 no longer supported. udev_monitor_new_from_netlink() needs to be
7785 used to subscribe to events.
7787 * udev: when udevd is started by systemd, processes which are left
7788 behind by forking them off of udev rules, are unconditionally cleaned
7789 up and killed now after the event handling has finished. Services or
7790 daemons must be started as systemd services. Services can be
7791 pulled-in by udev to get started, but they can no longer be directly
7792 forked by udev rules.
7794 * udev: the daemon binary is called systemd-udevd now and installed
7795 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7796 to adapt to that, create symlink, or rename the binary after building
7799 * libudev no longer provides these symbols:
7800 udev_monitor_from_socket()
7801 udev_queue_get_failed_list_entry()
7802 udev_get_{dev,sys,run}_path()
7803 The versions number was bumped and symbol versioning introduced.
7805 * systemd-loginctl and systemd-journalctl have been renamed
7806 to loginctl and journalctl to match systemctl.
7808 * The config files: /etc/systemd/systemd-logind.conf and
7809 /etc/systemd/systemd-journald.conf have been renamed to
7810 logind.conf and journald.conf. Package updates should rename
7811 the files to the new names on upgrade.
7813 * For almost all files the license is now LGPL2.1+, changed
7814 from the previous GPL2.0+. Exceptions are some minor stuff
7815 of udev (which will be changed to LGPL2.1 eventually, too),
7816 and the MIT licensed sd-daemon.[ch] library that is suitable
7817 to be used as drop-in files.
7819 * systemd and logind now handle system sleep states, in
7820 particular suspending and hibernating.
7822 * logind now implements a sleep/shutdown/idle inhibiting logic
7823 suitable for a variety of uses. Soonishly Lennart will blog
7824 about this in more detail.
7826 * var-run.mount and var-lock.mount are no longer provided
7827 (which previously bind mounted these directories to their new
7828 places). Distributions which have not converted these
7829 directories to symlinks should consider stealing these files
7830 from git history and add them downstream.
7832 * We introduced the Documentation= field for units and added
7833 this to all our shipped units. This is useful to make it
7834 easier to explore the boot and the purpose of the various
7837 * All smaller setup units (such as
7838 systemd-vconsole-setup.service) now detect properly if they
7839 are run in a container and are skipped when
7840 appropriate. This guarantees an entirely noise-free boot in
7841 Linux container environments such as systemd-nspawn.
7843 * A framework for implementing offline system updates is now
7844 integrated, for details see:
7845 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7847 * A new service type Type=idle is available now which helps us
7848 avoiding ugly interleaving of getty output and boot status
7851 * There's now a system-wide CapabilityBoundingSet= option to
7852 globally reduce the set of capabilities for the
7853 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7854 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7855 even CAP_NET_ADMIN system-wide for secure systems.
7857 * There are now system-wide DefaultLimitXXX= options to
7858 globally change the defaults of the various resource limits
7859 for all units started by PID 1.
7861 * Harald Hoyer's systemd test suite has been integrated into
7862 systemd which allows easy testing of systemd builds in qemu
7863 and nspawn. (This is really awesome! Ask us for details!)
7865 * The fstab parser is now implemented as generator, not inside
7868 * systemctl will now warn you if .mount units generated from
7869 /etc/fstab are out of date due to changes in fstab that
7870 have not been read by systemd yet.
7872 * systemd is now suitable for usage in initrds. Dracut has
7873 already been updated to make use of this. With this in place
7874 initrds get a slight bit faster but primarily are much
7875 easier to introspect and debug since "systemctl status" in
7876 the host system can be used to introspect initrd services,
7877 and the journal from the initrd is kept around too.
7879 * systemd-delta has been added, a tool to explore differences
7880 between user/admin configuration and vendor defaults.
7882 * PrivateTmp= now affects both /tmp and /var/tmp.
7884 * Boot time status messages are now much prettier and feature
7885 proper english language. Booting up systemd has never been
7888 * Read-ahead pack files now include the inode number of all
7889 files to pre-cache. When the inode changes the pre-caching
7890 is not attempted. This should be nicer to deal with updated
7891 packages which might result in changes of read-ahead
7894 * We now temporaritly lower the kernel's read_ahead_kb variable
7895 when collecting read-ahead data to ensure the kernel's
7896 built-in read-ahead does not add noise to our measurements
7897 of necessary blocks to pre-cache.
7899 * There's now RequiresMountsFor= to add automatic dependencies
7900 for all mounts necessary for a specific file system path.
7902 * MountAuto= and SwapAuto= have been removed from
7903 system.conf. Mounting file systems at boot has to take place
7906 * nspawn now learned a new switch --uuid= to set the machine
7907 ID on the command line.
7909 * nspawn now learned the -b switch to automatically search
7912 * vt102 is now the default TERM for serial TTYs, upgraded from
7915 * systemd-logind now works on VT-less systems.
7917 * The build tree has been reorganized. The individual
7918 components now have directories of their own.
7920 * A new condition type ConditionPathIsReadWrite= is now available.
7922 * nspawn learned the new -C switch to create cgroups for the
7923 container in other hierarchies.
7925 * We now have support for hardware watchdogs, configurable in
7928 * The scheduled shutdown logic now has a public API.
7930 * We now mount /tmp as tmpfs by default, but this can be
7931 masked and /etc/fstab can override it.
7933 * Since udisks does not make use of /media anymore we are not
7934 mounting a tmpfs on it anymore.
7936 * journalctl gained a new --local switch to only interleave
7937 locally generated journal files.
7939 * We can now load the IMA policy at boot automatically.
7941 * The GTK tools have been split off into a systemd-ui.
7943 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7944 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7945 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7946 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7947 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7948 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7949 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7950 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7951 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7956 * This is mostly a bugfix release
7958 * Support optional initialization of the machine ID from the
7959 KVM or container configured UUID.
7961 * Support immediate reboots with "systemctl reboot -ff"
7963 * Show /etc/os-release data in systemd-analyze output
7965 * Many bugfixes for the journal, including endianness fixes and
7966 ensuring that disk space enforcement works
7968 * sd-login.h is C++ compatible again
7970 * Extend the /etc/os-release format on request of the Debian
7973 * We now refuse non-UTF8 strings used in various configuration
7974 and unit files. This is done to ensure we do not pass invalid
7975 data over D-Bus or expose it elsewhere.
7977 * Register Mimo USB Screens as suitable for automatic seat
7980 * Read SELinux client context from journal clients in a race
7983 * Reorder configuration file lookup order. /etc now always
7984 overrides /run in order to allow the administrator to always
7985 and unconditionally override vendor-supplied or
7986 automatically generated data.
7988 * The various user visible bits of the journal now have man
7989 pages. We still lack man pages for the journal API calls
7992 * We now ship all man pages in HTML format again in the
7995 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7996 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7997 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8002 * This is mostly a bugfix release
8004 * systems lacking /etc/os-release are no longer supported.
8006 * Various functionality updates to libsystemd-login.so
8008 * Track class of PAM logins to distinguish greeters from
8011 Contributions from: Kay Sievers, Lennart Poettering, Michael
8016 * This is an important bugfix release for v41.
8018 * Building man pages is now optional which should be useful
8019 for those building systemd from git but unwilling to install
8022 * Watchdog support for supervising services is now usable. In
8023 a future release support for hardware watchdogs
8024 (i.e. /dev/watchdog) will be added building on this.
8026 * Service start rate limiting is now configurable and can be
8027 turned off per service. When a start rate limit is hit a
8028 reboot can automatically be triggered.
8030 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8032 Contributions from: Benjamin Franzke, Bill Nottingham,
8033 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8034 Schmidt, Michał Górny, Piotr Drąg
8038 * The systemd binary is installed /usr/lib/systemd/systemd now;
8039 An existing /sbin/init symlink needs to be adapted with the
8042 * The code that loads kernel modules has been ported to invoke
8043 libkmod directly, instead of modprobe. This means we do not
8044 support systems with module-init-tools anymore.
8046 * Watchdog support is now already useful, but still not
8049 * A new kernel command line option systemd.setenv= is
8050 understood to set system wide environment variables
8051 dynamically at boot.
8053 * We now limit the set of capabilities of systemd-journald.
8055 * We now set SIGPIPE to ignore by default, since it only is
8056 useful in shell pipelines, and has little use in general
8057 code. This can be disabled with IgnoreSIPIPE=no in unit
8060 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8061 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8066 * This is mostly a bugfix release
8068 * We now expose the reason why a service failed in the
8069 "Result" D-Bus property.
8071 * Rudimentary service watchdog support (will be completed over
8072 the next few releases.)
8074 * When systemd forks off in order execute some service we will
8075 now immediately changes its argv[0] to reflect which process
8076 it will execute. This is useful to minimize the time window
8077 with a generic argv[0], which makes bootcharts more useful
8079 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8080 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8081 Mike Kazantsev, Ray Strode
8085 * This is mostly a test release, but incorporates many
8088 * New systemd-cgtop tool to show control groups by their
8091 * Linking against libacl for ACLs is optional again. If
8092 disabled, support tracking device access for active logins
8093 goes becomes unavailable, and so does access to the user
8094 journals by the respective users.
8096 * If a group "adm" exists, journal files are automatically
8097 owned by them, thus allow members of this group full access
8098 to the system journal as well as all user journals.
8100 * The journal now stores the SELinux context of the logging
8101 client for all entries.
8103 * Add C++ inclusion guards to all public headers
8105 * New output mode "cat" in the journal to print only text
8106 messages, without any meta data like date or time.
8108 * Include tiny X server wrapper as a temporary stop-gap to
8109 teach XOrg udev display enumeration. This is used by display
8110 managers such as gdm, and will go away as soon as XOrg
8111 learned native udev hotplugging for display devices.
8113 * Add new systemd-cat tool for executing arbitrary programs
8114 with STDERR/STDOUT connected to the journal. Can also act as
8115 BSD logger replacement, and does so by default.
8117 * Optionally store all locally generated coredumps in the
8118 journal along with meta data.
8120 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8121 writing short strings to files (for usage for /sys), and for
8122 creating symlinks, character and block device nodes.
8124 * New unit file option ControlGroupPersistent= to make cgroups
8125 persistent, following the mechanisms outlined in
8126 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8128 * Support multiple local RTCs in a sane way
8130 * No longer monopolize IO when replaying readahead data on
8131 rotating disks, since we might starve non-file-system IO to
8132 death, since fanotify() will not see accesses done by blkid,
8135 * Do not show kernel threads in systemd-cgls anymore, unless
8136 requested with new -k switch.
8138 Contributions from: Dan Horák, Kay Sievers, Lennart
8139 Poettering, Michal Schmidt
8143 * This is mostly a test release, but incorporates many
8146 * The git repository moved to:
8147 git://anongit.freedesktop.org/systemd/systemd
8148 ssh://git.freedesktop.org/git/systemd/systemd
8150 * First release with the journal
8151 http://0pointer.de/blog/projects/the-journal.html
8153 * The journal replaces both systemd-kmsg-syslogd and
8154 systemd-stdout-bridge.
8156 * New sd_pid_get_unit() API call in libsystemd-logind
8158 * Many systemadm clean-ups
8160 * Introduce remote-fs-pre.target which is ordered before all
8161 remote mounts and may be used to start services before all
8164 * Added Mageia support
8166 * Add bash completion for systemd-loginctl
8168 * Actively monitor PID file creation for daemons which exit in
8169 the parent process before having finished writing the PID
8170 file in the daemon process. Daemons which do this need to be
8171 fixed (i.e. PID file creation must have finished before the
8172 parent exits), but we now react a bit more gracefully to them.
8174 * Add colourful boot output, mimicking the well-known output
8175 of existing distributions.
8177 * New option PassCredentials= for socket units, for
8178 compatibility with a recent kernel ABI breakage.
8180 * /etc/rc.local is now hooked in via a generator binary, and
8181 thus will no longer act as synchronization point during
8184 * systemctl list-unit-files now supports --root=.
8186 * systemd-tmpfiles now understands two new commands: z, Z for
8187 relabelling files according to the SELinux database. This is
8188 useful to apply SELinux labels to specific files in /sys,
8191 * Output of SysV services is now forwarded to both the console
8192 and the journal by default, not only just the console.
8194 * New man pages for all APIs from libsystemd-login.
8196 * The build tree got reorganized and the build system is a
8197 lot more modular allowing embedded setups to specifically
8198 select the components of systemd they are interested in.
8200 * Support for Linux systems lacking the kernel VT subsystem is
8203 * configure's --with-rootdir= got renamed to
8204 --with-rootprefix= to follow the naming used by udev and
8207 * Unless specified otherwise we will now install to /usr instead
8208 of /usr/local by default.
8210 * Processes with '@' in argv[0][0] are now excluded from the
8211 final shut-down killing spree, following the logic explained
8213 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8215 * All processes remaining in a service cgroup when we enter
8216 the START or START_PRE states are now killed with
8217 SIGKILL. That means it is no longer possible to spawn
8218 background processes from ExecStart= lines (which was never
8219 supported anyway, and bad style).
8221 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8222 reloading of units together.
8224 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8225 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8226 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8227 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8228 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob