1 systemd System and Service Manager
3 CHANGES WITH 237 in spe:
5 * Some keyboards come with a zoom see-saw or rocker which until now got
6 mapped to the Linux "zoomin/out" keys in hwdb. However, these
7 keycodes are not recognized by any major desktop. They now produce
8 Up/Down key events so that they can be used for scrolling.
10 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
11 slightly: previously, if an argument was specified for lines of this
12 type (i.e. the right-most column was set) this string was appended to
13 existing files each time elogind-tmpfiles was run. This behaviour was
14 different from what the documentation said, and not particularly
15 useful, as repeated elogind-tmpfiles invocations would not be
16 idempotent and grow such files without bounds. With this release
17 behaviour has been altered slightly, to match what the documentation
18 says: lines of this type only have an effect if the indicated files
19 don't exist yet, and only then the argument string is written to the
22 * A new environment variable $SYSTEMD_OFFLINE is now understood by
23 systemctl. It takes a boolean argument. If on, systemctl assumes it
24 operates on an "offline" OS tree, and will not attempt to talk to the
25 service manager. Previously, this mode was implicitly enabled if a
26 chroot() environment was detected, and this new environment variable
27 now provides explicit control.
29 * .path and .socket units may now be created transiently,
30 too. Previously only service, mount, automount and timer units were
31 supported as transient units. The elogind-run tool has been updated
32 to expose this new functionality, you may hence use it now to bind
33 arbitrary commands to path or socket activation on-the-fly from the
34 command line. Moreover, for the unit types already supporting
35 transient operation most properties are now hooked up.
37 * The elogind-mount command gained support for a new --owner= parameter
38 which takes a user name, which is then resolved and included in uid=
39 and gid= mount options string of the file system to mount.
41 * A new unit condition ConditionControlGroupController= has been added
42 that checks whether a specific cgroup controller is available.
44 * Unit files, udev's .link files, and elogind-networkd's .netdev and
45 .network files all gained support for a new condition
46 ConditionKernelVersion= for checking against specific kernel
49 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
50 support for configuring device flags in the Flags= setting In the
51 same files, the [Tunnel] section gained support for configuring
52 AllowLocalRemote=. The [Route] section in .network files gained
53 support for configuring InitialCongestionWindow=,
54 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
55 understands RapidCommit=.
57 * elogind-networkd's DHCPv6 support gained support for Prefix
60 * sd-bus gained support for a new "watch-bind" feature. When this
61 feature is enabled, an sd_bus connection may be set up to connect to
62 an AF_UNIX socket in the file system as soon as it is created. This
63 functionality is useful for writing early-boot services that
64 automatically connect to the system bus as soon as it is started,
65 without ugly time-based polling. elogind-networkd and
66 elogind-resolved have been updated to make use of this
67 functionality. busctl exposes this functionality in a new
68 --watch-bind= command line switch.
70 * sd-bus will now optionally synthesize a local "Connected" signal as
71 soon as a D-Bus connection is set up fully. This message mirrors the
72 already existing "Disconnected" signal which is synthesized when the
73 connection is terminated. This signal is generally useful but
74 particularly handy in combination with the "watch-bind" feature
75 described above. Synthesizing of this message has to be requested
76 explicitly through the new API call sd_bus_set_connected_signal(). In
77 addition a new call sd_bus_is_ready() has been added that checks
78 wether a connection is fully set up (i.e. between the "Connected" and
79 "Disconnected" signals).
81 * sd-bus gained two new calls sd_bus_request_name_async() and
82 sd_bus_release_name_async() for asynchronously registering bus
83 names. Similar, there is now sd_bus_add_match_async() for installing
84 a signal match asynchronously. All of elogind's own services have
85 been updated to make use of these calls. Doing these operations
86 asynchronously has two benefits: it reduces the risk of deadlocks in
87 case of cyclic dependencies between bus services, and it speeds up
88 service initialization since synchronization points for bus
89 round-trips are removed.
91 * sd-bus gained two new calls sd_bus_match_signal() and
92 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
93 and sd_bus_add_match_async() but instead of taking a D-Bus match
94 string take match fields as normal function parameters.
96 * sd-bus gained two new calls sd_bus_set_sender() and
97 sd_bus_message_set_sender() for setting the sender name of outgoing
98 messages (either for all outgoing messages or for just one specific
99 one). These calls are only useful in direct connections as on
100 brokered connections the broker fills in the sender anyway,
101 overwriting whatever the client filled in.
103 * sd-event gained a new pseudo-handle that may be specified on all API
104 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
105 used this refers to the default event loop object of the calling
106 thread. Note however that this does not implicitly allocate one —
107 which has to be done prior by using sd_event_default(). Similar
108 sd-bus gained a three new pseudo-handles SD_BUS_DEFAULT,
109 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
110 to the default bus of the specified type of the calling thread. Here
111 too this does not implicitly allocate bus connection objects, this
112 has to be done prior with sd_bus_default() and friends.
114 * sd-event gained a new call pair
115 sd_event_source_{get|set}_io_fd_own(). This may be used to optionally
116 automatically close the file descriptor an IO event source watches
117 when the event source is destroyed.
119 * elogind-networkd gained support for natively configuring WireGuard
122 * Already in previous versions elogind synthesized user records both
123 for the "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind
124 and internally. In order to simplify distribution-wide renames of the
125 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
126 new transitional flag file has been added: if
127 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
128 user and group record within the elogind codebase is disabled.
130 * elogind-notify gained a new --uid= option for selecting the source
131 user/UID to use for notification messages sent to the service
134 * There's now a "elogind-analyze service-watchdogs" command for turning
135 on or off the per-service watchdogs system-wide (i.e. the concept you
136 configure in WatchdogSec=), for debugging purposes. There's also a
137 kernel command line option elogind.service_watchdogs= for controlling
140 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
141 Alexis Deruelle, Armin Widegreen, bleep_blop, Bruce A. Johnson, Chris
142 Down, Colin Walters, Dimitri John Ledkov, Dmitry Rozhkov, Ewout van
143 Mansom, Felipe Sateler, Franck Bui, Frantisek Sumsal, George Gaydarov,
144 Gianluca Boiano, Hans-Christian Noren Egtvedt, Hans de Goede, Henrik
145 Grindal Bakken, Jan Alexander Steffens, Jan Klötzke, jdkbx, Jérémy
146 Rosen, Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg,
147 Jörg Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart
148 Poettering, Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák,
149 Martin Pitt, Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott,
150 Max Harmathy, Michael Biebl, Michal Koutný, Michal Sekletar, Michał
151 Szczepański, Mike Gilbert, Nathaniel McCallum, Olaf Hering, Olivier
152 Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer, Piotr Drąg,
153 Reverend Homer, Robert Kolchmeyer, Samuel Dionne-Riel, Sergey
154 Ptashnick, Shawn Landden, Susant Sahani, Sylvain Plantefève, Thomas
155 Huth, Tomasz Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu
156 Watanabe, Zachary Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан
157 Георгиевски, Дилян Палаузов
163 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
164 in v235 has been extended to also set the dummy.ko module option
165 numdummies=0, preventing the kernel from automatically creating
166 dummy0. All dummy interfaces must now be explicitly created.
168 * Unknown '%' specifiers in configuration files are now rejected. This
169 applies to units and tmpfiles.d configuration. Any percent characters
170 that are followed by a letter or digit that are not supposed to be
171 interpreted as the beginning of a specifier should be escaped by
172 doubling ("%%"). (So "size=5%" is still accepted, as well as
173 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
174 valid specifiers today.)
176 * systemd-resolved now maintains a new dynamic
177 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
178 recommended to make /etc/resolv.conf a symlink to it. This file
179 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
180 includes dynamically acquired search domains, achieving more correct
181 DNS resolution by software that bypasses local DNS APIs such as NSS.
183 * The "uaccess" udev tag has been dropped from /dev/kvm and
184 /dev/dri/renderD*. These devices now have the 0666 permissions by
185 default (but this may be changed at build-time). /dev/dri/renderD*
186 will now be owned by the "render" group along with /dev/kfd.
188 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
189 systemd-journal-gatewayd.service and
190 systemd-journal-upload.service. This means "nss-systemd" must be
191 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
192 services are resolved properly.
194 * In /etc/fstab two new mount options are now understood:
195 x-systemd.makefs and x-systemd.growfs. The former has the effect that
196 the configured file system is formatted before it is mounted, the
197 latter that the file system is resized to the full block device size
198 after it is mounted (i.e. if the file system is smaller than the
199 partition it resides on, it's grown). This is similar to the fsck
200 logic in /etc/fstab, and pulls in systemd-makefs@.service and
201 systemd-growfs@.service as necessary, similar to
202 systemd-fsck@.service. Resizing is currently only supported on ext4
205 * In systemd-networkd, the IPv6 RA logic now optionally may announce
206 DNS server and domain information.
208 * Support for the LUKS2 on-disk format for encrypted partitions has
209 been added. This requires libcryptsetup2 during compilation and
212 * The systemd --user instance will now signal "readiness" when its
213 basic.target unit has been reached, instead of when the run queue ran
214 empty for the first time.
216 * Tmpfiles.d with user configuration are now also supported.
217 systemd-tmpfiles gained a new --user switch, and snippets placed in
218 ~/.config/user-tmpfiles.d/ and corresponding directories will be
219 executed by systemd-tmpfiles --user running in the new
220 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
221 running in the user session.
223 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
224 %S resolves to the top-level state directory (/var/lib for the system
225 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
226 top-level cache directory (/var/cache for the system instance,
227 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
228 logs directory (/var/log for the system instance,
229 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
230 existing %t specifier, that resolves to the top-level runtime
231 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
234 * journalctl learnt a new parameter --output-fields= for limiting the
235 set of journal fields to output in verbose and JSON output modes.
237 * systemd-timesyncd's configuration file gained a new option
238 RootDistanceMaxSec= for setting the maximum root distance of servers
239 it'll use, as well as the new options PollIntervalMinSec= and
240 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
242 * bootctl gained a new command "list" for listing all available boot
243 menu items on systems that follow the boot loader specification.
245 * systemctl gained a new --dry-run switch that shows what would be done
246 instead of doing it, and is currently supported by the shutdown and
249 * ConditionSecurity= can now detect the TOMOYO security module.
251 * Unit file [Install] sections are now also respected in unit drop-in
252 files. This is intended to be used by drop-ins under /usr/lib/.
254 * systemd-firstboot may now also set the initial keyboard mapping.
256 * Udev "changed" events for devices which are exposed as systemd
257 .device units are now propagated to units specified in
258 ReloadPropagatedFrom= as reload requests.
260 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
261 unit template name (i.e. a name in the form of 'foobar@.service',
262 without the instance component between the '@' and - the '.'), then
263 the escaped sysfs path of the device is automatically used as the
266 * SystemCallFilter= in unit files has been extended so that an "errno"
267 can be specified individually for each system call. Example:
268 SystemCallFilter=~uname:EILSEQ.
270 * The cgroup delegation logic has been substantially updated. Delegate=
271 now optionally takes a list of controllers (instead of a boolean, as
272 before), which lists the controllers to delegate at least.
274 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
276 * A new LogLevelMax= setting configures the maximum log level any
277 process of the service may log at (i.e. anything with a lesser
278 priority than what is specified is automatically dropped). A new
279 LogExtraFields= setting allows configuration of additional journal
280 fields to attach to all log records generated by any of the unit's
283 * New StandardInputData= and StandardInputText= settings along with the
284 new option StandardInput=data may be used to configure textual or
285 binary data that shall be passed to the executed service process via
286 standard input, encoded in-line in the unit file.
288 * StandardInput=, StandardOutput= and StandardError= may now be used to
289 connect stdin/stdout/stderr of executed processes directly with a
290 file or AF_UNIX socket in the file system, using the new "file:" option.
292 * A new unit file option CollectMode= has been added, that allows
293 tweaking the garbage collection logic for units. It may be used to
294 tell systemd to garbage collect units that have failed automatically
295 (normally it only GCs units that exited successfully). systemd-run
296 and systemd-mount expose this new functionality with a new -G option.
298 * "machinectl bind" may now be used to bind mount non-directories
299 (i.e. regularfiles, devices, fifos, sockets).
301 * systemd-analyze gained a new verb "calendar" for validating and
302 testing calendar time specifications to use for OnCalendar= in timer
303 units. Besides validating the expression it will calculate the next
304 time the specified expression would elapse.
306 * In addition to the pre-existing FailureAction= unit file setting
307 there's now SuccessAction=, for configuring a shutdown action to
308 execute when a unit completes successfully. This is useful in
309 particular inside containers that shall terminate after some workload
310 has been completed. Also, both options are now supported for all unit
311 types, not just services.
313 * networkds's IP rule support gained two new options
314 IncomingInterface= and OutgoingInterface= for configuring the incoming
315 and outgoing interfaces of configured rules. systemd-networkd also
316 gained support for "vxcan" network devices.
318 * networkd gained a new setting RequiredForOnline=, taking a
319 boolean. If set, systemd-wait-online will take it into consideration
320 when determining that the system is up, otherwise it will ignore the
321 interface for this purpose.
323 * The sd_notify() protocol gained support for a new operation: with
324 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
325 store again, ahead of POLLHUP or POLLERR when they are removed
328 * A new document UIDS-GIDS.md has been added to the source tree, that
329 documents the UID/GID range and assignment assumptions and
330 requirements of systemd.
332 * The watchdog device PID 1 will ping may now be configured through the
333 WatchdogDevice= configuration file setting, or by setting the
334 systemd.watchdog_service= kernel commandline option.
336 * systemd-resolved's gained support for registering DNS-SD services on
337 the local network using MulticastDNS. Services may either be
338 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
339 the same dir below /run, /usr/lib), or through its D-Bus API.
341 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
342 extend the effective start, runtime, and stop time. The service must
343 continue to send EXTEND_TIMEOUT_USEC within the period specified to
344 prevent the service manager from making the service as timedout.
346 * elogind-resolved's DNSSEC support gained support for RFC 8080
347 (Ed25519 keys and signatures).
349 * The elogind-resolve command line tool gained a new set of options
350 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
351 --set-nta= and --revert to configure per-interface DNS configuration
352 dynamically during runtime. It's useful for pushing DNS information
353 into elogind-resolved from DNS hook scripts that various interface
354 managing software supports (such as pppd).
356 * elogind-nspawn gained a new --network-namespace-path= command line
357 option, which may be used to make a container join an existing
358 network namespace, by specifying a path to a "netns" file.
360 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
361 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
362 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
363 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
364 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
365 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
366 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
367 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
368 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
369 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
370 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
371 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
372 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
373 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
374 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
375 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
376 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
377 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
378 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
379 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
380 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
381 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
382 Jędrzejewski-Szmek, Zeal Jagannatha
388 * INCOMPATIBILITY: systemd-logind.service and other long-running
389 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
390 communication with the outside. This generally improves security of
391 the system, and is in almost all cases a safe and good choice, as
392 these services do not and should not provide any network-facing
393 functionality. However, systemd-logind uses the glibc NSS API to
394 query the user database. This creates problems on systems where NSS
395 is set up to directly consult network services for user database
396 lookups. In particular, this creates incompatibilities with the
397 "nss-nis" module, which attempts to directly contact the NIS/YP
398 network servers it is configured for, and will now consistently
399 fail. In such cases, it is possible to turn off IP sandboxing for
400 systemd-logind.service (set IPAddressDeny= in its [Service] section
401 to the empty string, via a .d/ unit file drop-in). Downstream
402 distributions might want to update their nss-nis packaging to include
403 such a drop-in snippet, accordingly, to hide this incompatibility
404 from the user. Another option is to make use of glibc's nscd service
405 to proxy such network requests through a privilege-separated, minimal
406 local caching daemon, or to switch to more modern technologies such
407 sssd, whose NSS hook-ups generally do not involve direct network
408 access. In general, we think it's definitely time to question the
409 implementation choices of nss-nis, i.e. whether it's a good idea
410 today to embed a network-facing loadable module into all local
411 processes that need to query the user database, including the most
412 trivial and benign ones, such as "ls". For more details about
413 IPAddressDeny= see below.
415 * A new modprobe.d drop-in is now shipped by default that sets the
416 bonding module option max_bonds=0. This overrides the kernel default,
417 to avoid conflicts and ambiguity as to whether or not bond0 should be
418 managed by systemd-networkd or not. This resolves multiple issues
419 with bond0 properties not being applied, when bond0 is configured
420 with systemd-networkd. Distributors may choose to not package this,
421 however in that case users will be prevented from correctly managing
422 bond0 interface using systemd-networkd.
424 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
425 which print the logging level and target of the system manager. They
426 complement the existing "set-log-level" and "set-log-target" verbs
427 used to change those values.
429 * journald.conf gained a new boolean setting ReadKMsg= which defaults
430 to on. If turned off kernel log messages will not be read by
431 systemd-journald or included in the logs. It also gained a new
432 setting LineMax= for configuring the maximum line length in
433 STDOUT/STDERR log streams. The new default for this value is 48K, up
434 from the previous hardcoded 2048.
436 * A new unit setting RuntimeDirectoryPreserve= has been added, which
437 allows more detailed control of what to do with a runtime directory
438 configured with RuntimeDirectory= (i.e. a directory below /run or
439 $XDG_RUNTIME_DIR) after a unit is stopped.
441 * The RuntimeDirectory= setting for units gained support for creating
442 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
443 one top-level directory.
445 * Units gained new options StateDirectory=, CacheDirectory=,
446 LogsDirectory= and ConfigurationDirectory= which are closely related
447 to RuntimeDirectory= but manage per-service directories below
448 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
449 possible to write unit files which when activated automatically gain
450 properly owned service specific directories in these locations, thus
451 making unit files self-contained and increasing compatibility with
452 stateless systems and factory reset where /etc or /var are
453 unpopulated at boot. Matching these new settings there's also
454 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
455 ConfigurationDirectoryMode= for configuring the access mode of these
456 directories. These settings are particularly useful in combination
457 with DynamicUser=yes as they provide secure, properly-owned,
458 writable, and stateful locations for storage, excluded from the
459 sandbox that such services live in otherwise.
461 * Automake support has been removed from this release. systemd is now
464 * systemd-journald will now aggressively cache client metadata during
465 runtime, speeding up log write performance under pressure. This comes
466 at a small price though: as much of the metadata is read
467 asynchronously from /proc/ (and isn't implicitly attached to log
468 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
469 metadata stored alongside a log entry might be slightly
470 out-of-date. Previously it could only be slightly newer than the log
471 message. The time window is small however, and given that the kernel
472 is unlikely to be improved anytime soon in this regard, this appears
475 * nss-myhostname/systemd-resolved will now by default synthesize an
476 A/AAAA resource record for the "_gateway" hostname, pointing to the
477 current default IP gateway. Previously it did that for the "gateway"
478 name, hampering adoption, as some distributions wanted to leave that
479 host name open for local use. The old behaviour may still be
480 requested at build time.
482 * systemd-networkd's [Address] section in .network files gained a new
483 Scope= setting for configuring the IP address scope. The [Network]
484 section gained a new boolean setting ConfigureWithoutCarrier= that
485 tells systemd-networkd to ignore link sensing when configuring the
486 device. The [DHCP] section gained a new Anonymize= boolean option for
487 turning on a number of options suggested in RFC 7844. A new
488 [RoutingPolicyRule] section has been added for configuring the IP
489 routing policy. The [Route] section has gained support for a new
490 Type= setting which permits configuring
491 blackhole/unreachable/prohibit routes.
493 * The [VRF] section in .netdev files gained a new Table= setting for
494 configuring the routing table to use. The [Tunnel] section gained a
495 new Independent= boolean field for configuring tunnels independent of
496 an underlying network interface. The [Bridge] section gained a new
497 GroupForwardMask= option for configuration of propagation of link
498 local frames between bridge ports.
500 * The WakeOnLan= setting in .link files gained support for a number of
501 new modes. A new TCP6SegmentationOffload= setting has been added for
502 configuring TCP/IPv6 hardware segmentation offload.
504 * The IPv6 RA sender implementation may now optionally send out RDNSS
505 and RDNSSL records to supply DNS configuration to peers.
507 * systemd-nspawn gained support for a new --system-call-filter= command
508 line option for adding and removing entries in the default system
509 call filter it applies. Moreover systemd-nspawn has been changed to
510 implement a system call whitelist instead of a blacklist.
512 * systemd-run gained support for a new --pipe command line option. If
513 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
514 are directly passed on to the activated transient service
515 executable. This allows invoking arbitrary processes as systemd
516 services (for example to take benefit of dependency management,
517 accounting management, resource management or log management that is
518 done automatically for services) — while still allowing them to be
519 integrated in a classic UNIX shell pipeline.
521 * When a service sends RELOAD=1 via sd_notify() and reload propagation
522 using ReloadPropagationTo= is configured, a reload is now propagated
523 to configured units. (Previously this was only done on explicitly
524 requested reloads, using "systemctl reload" or an equivalent
527 * For each service unit a restart counter is now kept: it is increased
528 each time the service is restarted due to Restart=, and may be
529 queried using "systemctl show -p NRestarts …".
531 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
532 @signal and @timer have been added, for usage with SystemCallFilter=
533 in unit files and the new --system-call-filter= command line option
534 of systemd-nspawn (see above).
536 * ExecStart= lines in unit files gained two new modifiers: when a
537 command line is prefixed with "!" the command will be executed as
538 configured, except for the credentials applied by
539 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
540 "+", but does still apply namespacing options unlike "+". There's
541 also "!!" now, which is mostly identical, but becomes a NOP on
542 systems that support ambient capabilities. This is useful to write
543 unit files that work with ambient capabilities where possible but
544 automatically fall back to traditional privilege dropping mechanisms
545 on systems where this is not supported.
547 * ListenNetlink= settings in socket units now support RDMA netlink
550 * A new unit file setting LockPersonality= has been added which permits
551 locking down the chosen execution domain ("personality") of a service
554 * A new special target "getty-pre.target" has been added, which is
555 ordered before all text logins, and may be used to order services
556 before textual logins acquire access to the console.
558 * systemd will now attempt to load the virtio-rng.ko kernel module very
559 early on if a VM environment supporting this is detected. This should
560 improve entropy during early boot in virtualized environments.
562 * A _netdev option is now supported in /etc/crypttab that operates in a
563 similar way as the same option in /etc/fstab: it permits configuring
564 encrypted devices that need to be ordered after the network is up.
565 Following this logic, two new special targets
566 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
567 added that are to cryptsetup.target what remote-fs.target and
568 remote-fs-pre.target are to local-fs.target.
570 * Service units gained a new UnsetEnvironment= setting which permits
571 unsetting specific environment variables for services that are
572 normally passed to it (for example in order to mask out locale
573 settings for specific services that can't deal with it).
575 * Units acquired a new boolean option IPAccounting=. When turned on, IP
576 traffic accounting (packet count as well as byte count) is done for
577 the service, and shown as part of "systemctl status" or "systemd-run
580 * Service units acquired two new options IPAddressAllow= and
581 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
582 for configuring a simple IP access control list for all sockets of
583 the unit. These options are available also on .slice and .socket
584 units, permitting flexible access list configuration for individual
585 services as well as groups of services (as defined by a slice unit),
586 including system-wide. Note that IP ACLs configured this way are
587 enforced on every single IPv4 and IPv6 socket created by any process
588 of the service unit, and apply to ingress as well as egress traffic.
590 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
591 structured log message is generated each time the unit is stopped,
592 containing information about the consumed resources of this
595 * A new setting KeyringMode= has been added to unit files, which may be
596 used to control how the kernel keyring is set up for executed
599 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
600 "systemctl kexec" and "systemctl exit" are now always asynchronous in
601 behaviour (that is: these commands return immediately after the
602 operation was enqueued instead of waiting for the operation to
603 complete). Previously, "systemctl poweroff" and "systemctl reboot"
604 were asynchronous on systems using systemd-logind (i.e. almost
605 always, and like they were on sysvinit), and the other three commands
606 were unconditionally synchronous. With this release this is cleaned
607 up, and callers will see the same asynchronous behaviour on all
608 systems for all five operations.
610 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
613 * .timer units now accept calendar specifications in other timezones
614 than UTC or the local timezone.
616 * The tmpfiles snippet var.conf has been changed to create
617 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
618 the "utmp" group already, and it appears to be generally understood
619 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
620 databases. Previously this was implemented correctly for all these
621 databases excepts btmp, which has been opened up like this now
622 too. Note that while the other databases are world-readable
623 (i.e. 0644), btmp is not and remains more restrictive.
625 * The systemd-resolve tool gained a new --reset-server-features
626 switch. When invoked like this systemd-resolved will forget
627 everything it learnt about the features supported by the configured
628 upstream DNS servers, and restarts the feature probing logic on the
629 next resolver look-up for them at the highest feature level
632 * The status dump systemd-resolved sends to the logs upon receiving
633 SIGUSR1 now also includes information about all DNS servers it is
634 configured to use, and the features levels it probed for them.
636 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
637 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
638 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
639 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
640 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
641 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
642 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
643 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
644 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
645 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
646 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
647 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
648 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
649 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
650 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
651 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
652 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
653 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
654 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
655 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
661 * Meson is now supported as build system in addition to Automake. It is
662 our plan to remove Automake in one of our next releases, so that
663 Meson becomes our exclusive build system. Hence, please start using
664 the Meson build system in your downstream packaging. There's plenty
665 of documentation around how to use Meson, the extremely brief
668 ./autogen.sh && ./configure && make && sudo make install
672 meson build && ninja -C build && sudo ninja -C build install
674 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
675 which permits configuring a timeout on the time a job is
676 running. This is particularly useful for setting timeouts on jobs for
679 * Unit files gained two new options ConditionUser= and ConditionGroup=
680 for conditionalizing units based on the identity of the user/group
681 running a systemd user instance.
683 * systemd-networkd now understands a new FlowLabel= setting in the
684 [VXLAN] section of .network files, as well as a Priority= in
685 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
686 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
687 gained support for configuration of GENEVE links, and IPv6 address
688 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
690 * .link files now understand a new Port= setting.
692 * systemd-networkd's DHCP support gained support for DHCP option 119
693 (domain search list).
695 * systemd-networkd gained support for serving IPv6 address ranges using
696 the Router Advertisment protocol. The new .network configuration
697 section [IPv6Prefix] may be used to configure the ranges to
698 serve. This is implemented based on a new, minimal, native server
699 implementation of RA.
701 * journalctl's --output= switch gained support for a new parameter
702 "short-iso-precise" for a mode where timestamps are shown as precise
705 * systemd-udevd's "net_id" builtin may now generate stable network
706 interface names from IBM PowerVM VIO devices as well as ACPI platform
709 * MulticastDNS support in systemd-resolved may now be explicitly
710 enabled/disabled using the new MulticastDNS= configuration file
713 * systemd-resolved may now optionally use libidn2 instead of the libidn
714 for processing internationalized domain names. Support for libidn2
715 should be considered experimental and should not be enabled by
718 * "machinectl pull-tar" and related call may now do verification of
719 downloaded images using SUSE-style .sha256 checksum files in addition
720 to the already existing support for validating using Ubuntu-style
723 * sd-bus gained support for a new sd_bus_message_appendv() call which
724 is va_list equivalent of sd_bus_message_append().
726 * sd-boot gained support for validating images using SHIM/MOK.
728 * The SMACK code learnt support for "onlycap".
730 * systemd-mount --umount is now much smarter in figuring out how to
731 properly unmount a device given its mount or device path.
733 * The code to call libnss_dns as a fallback from libnss_resolve when
734 the communication with systemd-resolved fails was removed. This
735 fallback was redundant and interfered with the [!UNAVAIL=return]
736 suffix. See nss-resolve(8) for the recommended configuration.
738 * systemd-logind may now be restarted without losing state. It stores
739 the file descriptors for devices it manages in the system manager
740 using the FDSTORE= mechanism. Please note that further changes in
741 other components may be required to make use of this (for example
742 Xorg has code to listen for stops of systemd-logind and terminate
743 itself when logind is stopped or restarted, in order to avoid using
744 stale file descriptors for graphical devices, which is now
745 counterproductive and must be reverted in order for restarts of
746 systemd-logind to be safe. See
747 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
749 * All kernel install plugins are called with the environment variable
750 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
751 /etc/machine-id. If the file is missing or empty, the variable is
752 empty and BOOT_DIR_ABS is the path of a temporary directory which is
753 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
754 is empty, all plugins should not put anything in BOOT_DIR_ABS.
756 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
757 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
758 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
759 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
760 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
761 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
762 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
763 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
764 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
765 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
766 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
767 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
768 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
769 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
770 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
771 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
772 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
773 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
774 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
775 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
776 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
777 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
778 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
779 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
780 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
781 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
782 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
783 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
784 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
791 * The "hybrid" control group mode has been modified to improve
792 compatibility with "legacy" cgroups-v1 setups. Specifically, the
793 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
794 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
795 cgroups-v1 hierarchy), the only externally visible change being that
796 the cgroups-v2 hierarchy is also mounted, to
797 /sys/fs/cgroup/unified. This should provide a large degree of
798 compatibility with "legacy" cgroups-v1, while taking benefit of the
799 better management capabilities of cgroups-v2.
801 * The default control group setup mode may be selected both a boot-time
802 via a set of kernel command line parameters (specifically:
803 systemd.unified_cgroup_hierarchy= and
804 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
805 default selected on the configure command line
806 (--with-default-hierarchy=). The upstream default is "hybrid"
807 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
808 this will change in a future systemd version to be "unified" (pure
809 cgroups-v2 mode). The third option for the compile time option is
810 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
811 distributions to default to "hybrid" mode for release distributions,
812 starting with v233. We recommend "unified" for development
813 distributions (specifically: distributions such as Fedora's rawhide)
814 as that's where things are headed in the long run. Use "legacy" for
815 greatest stability and compatibility only.
817 * Note one current limitation of "unified" and "hybrid" control group
818 setup modes: the kernel currently does not permit the systemd --user
819 instance (i.e. unprivileged code) to migrate processes between two
820 disconnected cgroup subtrees, even if both are managed and owned by
821 the user. This effectively means "systemd-run --user --scope" doesn't
822 work when invoked from outside of any "systemd --user" service or
823 scope. Specifically, it is not supported from session scopes. We are
824 working on fixing this in a future systemd version. (See #3388 for
825 further details about this.)
827 * DBus policy files are now installed into /usr rather than /etc. Make
828 sure your system has dbus >= 1.9.18 running before upgrading to this
829 version, or override the install path with --with-dbuspolicydir= .
831 * All python scripts shipped with systemd (specifically: the various
832 tests written in Python) now require Python 3.
834 * systemd unit tests can now run standalone (without the source or
835 build directories), and can be installed into /usr/lib/systemd/tests/
836 with 'make install-tests'.
838 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
839 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
842 * Support for the %c, %r, %R specifiers in unit files has been
843 removed. Specifiers are not supposed to be dependent on configuration
844 in the unit file itself (so that they resolve the same regardless
845 where used in the unit files), but these specifiers were influenced
846 by the Slice= option.
848 * The shell invoked by debug-shell.service now defaults to /bin/sh in
849 all cases. If distributions want to use a different shell for this
850 purpose (for example Fedora's /sbin/sushell) they need to specify
851 this explicitly at configure time using --with-debug-shell=.
853 * The confirmation spawn prompt has been reworked to offer the
856 (c)ontinue, proceed without asking anymore
857 (D)ump, show the state of the unit
858 (f)ail, don't execute the command and pretend it failed
860 (i)nfo, show a short summary of the unit
861 (j)obs, show jobs that are in progress
862 (s)kip, don't execute the command and pretend it succeeded
863 (y)es, execute the command
865 The 'n' choice for the confirmation spawn prompt has been removed,
866 because its meaning was confusing.
868 The prompt may now also be redirected to an alternative console by
869 specifying the console as parameter to systemd.confirm_spawn=.
871 * Services of Type=notify require a READY=1 notification to be sent
872 during startup. If no such message is sent, the service now fails,
873 even if the main process exited with a successful exit code.
875 * Services that fail to start up correctly now always have their
876 ExecStopPost= commands executed. Previously, they'd enter "failed"
877 state directly, without executing these commands.
879 * The option MulticastDNS= of network configuration files has acquired
880 an actual implementation. With MulticastDNS=yes a host can resolve
881 names of remote hosts and reply to mDNS A and AAAA requests.
883 * When units are about to be started an additional check is now done to
884 ensure that all dependencies of type BindsTo= (when used in
885 combination with After=) have been started.
887 * systemd-analyze gained a new verb "syscall-filter" which shows which
888 system call groups are defined for the SystemCallFilter= unit file
889 setting, and which system calls they contain.
891 * A new system call filter group "@filesystem" has been added,
892 consisting of various file system related system calls. Group
893 "@reboot" has been added, covering reboot, kexec and shutdown related
894 calls. Finally, group "@swap" has been added covering swap
895 configuration related calls.
897 * A new unit file option RestrictNamespaces= has been added that may be
898 used to restrict access to the various process namespace types the
899 Linux kernel provides. Specifically, it may be used to take away the
900 right for a service unit to create additional file system, network,
901 user, and other namespaces. This sandboxing option is particularly
902 relevant due to the high amount of recently discovered namespacing
903 related vulnerabilities in the kernel.
905 * systemd-udev's .link files gained support for a new AutoNegotiation=
906 setting for configuring Ethernet auto-negotiation.
908 * systemd-networkd's .network files gained support for a new
909 ListenPort= setting in the [DHCP] section to explicitly configure the
910 UDP client port the DHCP client shall listen on.
912 * .network files gained a new Unmanaged= boolean setting for explicitly
913 excluding one or more interfaces from management by systemd-networkd.
915 * The systemd-networkd ProxyARP= option has been renamed to
916 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
917 renamed to ReduceARPProxy=. The old names continue to be available
920 * systemd-networkd gained support for configuring IPv6 Proxy NDP
921 addresses via the new IPv6ProxyNDPAddress= .network file setting.
923 * systemd-networkd's bonding device support gained support for two new
924 configuration options ActiveSlave= and PrimarySlave=.
926 * The various options in the [Match] section of .network files gained
927 support for negative matching.
929 * New systemd-specific mount options are now understood in /etc/fstab:
931 x-systemd.mount-timeout= may be used to configure the maximum
932 permitted runtime of the mount command.
934 x-systemd.device-bound may be set to bind a mount point to its
935 backing device unit, in order to automatically remove a mount point
936 if its backing device is unplugged. This option may also be
937 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
938 on the block device, which is now automatically set for all CDROM
939 drives, so that mounted CDs are automatically unmounted when they are
940 removed from the drive.
942 x-systemd.after= and x-systemd.before= may be used to explicitly
943 order a mount after or before another unit or mount point.
945 * Enqueued start jobs for device units are now automatically garbage
946 collected if there are no jobs waiting for them anymore.
948 * systemctl list-jobs gained two new switches: with --after, for every
949 queued job the jobs it's waiting for are shown; with --before the
950 jobs which it's blocking are shown.
952 * systemd-nspawn gained support for ephemeral boots from disk images
953 (or in other words: --ephemeral and --image= may now be
954 combined). Moreover, ephemeral boots are now supported for normal
955 directories, even if the backing file system is not btrfs. Of course,
956 if the file system does not support file system snapshots or
957 reflinks, the initial copy operation will be relatively expensive, but
958 this should still be suitable for many use cases.
960 * Calendar time specifications in .timer units now support
961 specifications relative to the end of a month by using "~" instead of
962 "-" as separator between month and day. For example, "*-02~03" means
963 "the third last day in February". In addition a new syntax for
964 repeated events has been added using the "/" character. For example,
965 "9..17/2:00" means "every two hours from 9am to 5pm".
967 * systemd-socket-proxyd gained a new parameter --connections-max= for
968 configuring the maximum number of concurrent connections.
970 * sd-id128 gained a new API for generating unique IDs for the host in a
971 way that does not leak the machine ID. Specifically,
972 sd_id128_get_machine_app_specific() derives an ID based on the
973 machine ID a in well-defined, non-reversible, stable way. This is
974 useful whenever an identifier for the host is needed but where the
975 identifier shall not be useful to identify the system beyond the
976 scope of the application itself. (Internally this uses HMAC-SHA256 as
977 keyed hash function using the machine ID as input.)
979 * NotifyAccess= gained a new supported value "exec". When set
980 notifications are accepted from all processes systemd itself invoked,
981 including all control processes.
983 * .nspawn files gained support for defining overlay mounts using the
984 Overlay= and OverlayReadOnly= options. Previously this functionality
985 was only available on the systemd-nspawn command line.
987 * systemd-nspawn's --bind= and --overlay= options gained support for
988 bind/overlay mounts whose source lies within the container tree by
989 prefixing the source path with "+".
991 * systemd-nspawn's --bind= and --overlay= options gained support for
992 automatically allocating a temporary source directory in /var/tmp
993 that is removed when the container dies. Specifically, if the source
994 directory is specified as empty string this mechanism is selected. An
995 example usage is --overlay=+/var::/var, which creates an overlay
996 mount based on the original /var contained in the image, overlayed
997 with a temporary directory in the host's /var/tmp. This way changes
998 to /var are automatically flushed when the container shuts down.
1000 * systemd-nspawn --image= option does now permit raw file system block
1001 devices (in addition to images containing partition tables, as
1004 * The disk image dissection logic in systemd-nspawn gained support for
1005 automatically setting up LUKS encrypted as well as Verity protected
1006 partitions. When a container is booted from an encrypted image the
1007 passphrase is queried at start-up time. When a container with Verity
1008 data is started, the root hash is search in a ".roothash" file
1009 accompanying the disk image (alternatively, pass the root hash via
1010 the new --root-hash= command line option).
1012 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1013 be used to dissect disk images the same way as systemd-nspawn does
1014 it, following the Bootable Partition Specification. It may even be
1015 used to mount disk images with complex partition setups (including
1016 LUKS and Verity partitions) to a local host directory, in order to
1017 inspect them. This tool is not considered public API (yet), and is
1018 thus not installed into /usr/bin. Please do not rely on its
1019 existence, since it might go away or be changed in later systemd
1022 * A new generator "systemd-verity-generator" has been added, similar in
1023 style to "systemd-cryptsetup-generator", permitting automatic setup of
1024 Verity root partitions when systemd boots up. In order to make use of
1025 this your partition setup should follow the Discoverable Partitions
1026 Specification, and the GPT partition ID of the root file system
1027 partition should be identical to the upper 128bit of the Verity root
1028 hash. The GPT partition ID of the Verity partition protecting it
1029 should be the lower 128bit of the Verity root hash. If the partition
1030 image follows this model it is sufficient to specify a single
1031 "roothash=" kernel command line argument to both configure which root
1032 image and verity partition to use as well as the root hash for
1033 it. Note that systemd-nspawn's Verity support follows the same
1034 semantics, meaning that disk images with proper Verity data in place
1035 may be booted in containers with systemd-nspawn as well as on
1036 physical systems via the verity generator. Also note that the "mkosi"
1037 tool available at https://github.com/systemd/mkosi has been updated
1038 to generate Verity protected disk images following this scheme. In
1039 fact, it has been updated to generate disk images that optionally
1040 implement a complete UEFI SecureBoot trust chain, involving a signed
1041 kernel and initrd image that incorporates such a root hash as well as
1042 a Verity-enabled root partition.
1044 * The hardware database (hwdb) udev supports has been updated to carry
1045 accelerometer quirks.
1047 * All system services are now run with a fresh kernel keyring set up
1048 for them. The invocation ID is stored by default in it, thus
1049 providing a safe, non-overridable way to determine the invocation
1052 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1053 options for bind mounting arbitrary paths in a service-specific
1054 way. When these options are used, arbitrary host or service files and
1055 directories may be mounted to arbitrary locations in the service's
1058 * Documentation has been added that lists all of systemd's low-level
1059 environment variables:
1061 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
1063 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1064 whether a specific socket file descriptor matches a specified socket
1067 * systemd-firstboot has been updated to check for the
1068 systemd.firstboot= kernel command line option. It accepts a boolean
1069 and when set to false the first boot questions are skipped.
1071 * systemd-fstab-generator has been updated to check for the
1072 systemd.volatile= kernel command line option, which either takes an
1073 optional boolean parameter or the special value "state". If used the
1074 system may be booted in a "volatile" boot mode. Specifically,
1075 "systemd.volatile" is used, the root directory will be mounted as
1076 tmpfs, and only /usr is mounted from the actual root file system. If
1077 "systemd.volatile=state" is used, the root directory will be mounted
1078 as usual, but /var is mounted as tmpfs. This concept provides similar
1079 functionality as systemd-nspawn's --volatile= option, but provides it
1080 on physical boots. Use this option for implementing stateless
1081 systems, or testing systems with all state and/or configuration reset
1082 to the defaults. (Note though that many distributions are not
1083 prepared to boot up without a populated /etc or /var, though.)
1085 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1086 partitions. Previously it only supported LUKS encrypted partitions
1087 for all other uses, except for the root partition itself.
1089 * Socket units gained support for listening on AF_VSOCK sockets for
1090 communication in virtualized QEMU environments.
1092 * The "configure" script gained a new option --with-fallback-hostname=
1093 for specifying the fallback hostname to use if none is configured in
1094 /etc/hostname. For example, by specifying
1095 --with-fallback-hostname=fedora it is possible to default to a
1096 hostname of "fedora" on pristine installations.
1098 * systemd-cgls gained support for a new --unit= switch for listing only
1099 the control groups of a specific unit. Similar --user-unit= has been
1100 added for listing only the control groups of a specific user unit.
1102 * systemd-mount gained a new --umount switch for unmounting a mount or
1103 automount point (and all mount/automount points below it).
1105 * systemd will now refuse full configuration reloads (via systemctl
1106 daemon-reload and related calls) unless at least 16MiB of free space
1107 are available in /run. This is a safety precaution in order to ensure
1108 that generators can safely operate after the reload completed.
1110 * A new unit file option RootImage= has been added, which has a similar
1111 effect as RootDirectory= but mounts the service's root directory from
1112 a disk image instead of plain directory. This logic reuses the same
1113 image dissection and mount logic that systemd-nspawn already uses,
1114 and hence supports any disk images systemd-nspawn supports, including
1115 those following the Discoverable Partition Specification, as well as
1116 Verity enabled images. This option enables systemd to run system
1117 services directly off disk images acting as resource bundles,
1118 possibly even including full integrity data.
1120 * A new MountAPIVFS= unit file option has been added, taking a boolean
1121 argument. If enabled /proc, /sys and /dev (collectively called the
1122 "API VFS") will be mounted for the service. This is only relevant if
1123 RootDirectory= or RootImage= is used for the service, as these mounts
1124 are of course in place in the host mount namespace anyway.
1126 * systemd-nspawn gained support for a new --pivot-root= switch. If
1127 specified the root directory within the container image is pivoted to
1128 the specified mount point, while the original root disk is moved to a
1129 different place. This option enables booting of ostree images
1130 directly with systemd-nspawn.
1132 * The systemd build scripts will no longer complain if the NTP server
1133 addresses are not changed from the defaults. Google now supports
1134 these NTP servers officially. We still recommend downstreams to
1135 properly register an NTP pool with the NTP pool project though.
1137 * coredumpctl gained a new "--reverse" option for printing the list
1138 of coredumps in reverse order.
1140 * coredumpctl will now show additional information about truncated and
1141 inaccessible coredumps, as well as coredumps that are still being
1142 processed. It also gained a new --quiet switch for suppressing
1143 additional informational message in its output.
1145 * coredumpctl gained support for only showing coredumps newer and/or
1146 older than specific timestamps, using the new --since= and --until=
1147 options, reminiscent of journalctl's options by the same name.
1149 * The systemd-coredump logic has been improved so that it may be reused
1150 to collect backtraces in non-compiled languages, for example in
1151 scripting languages such as Python.
1153 * machinectl will now show the UID shift of local containers, if user
1154 namespacing is enabled for them.
1156 * systemd will now optionally run "environment generator" binaries at
1157 configuration load time. They may be used to add environment
1158 variables to the environment block passed to services invoked. One
1159 user environment generator is shipped by default that sets up
1160 environment variables based on files dropped into /etc/environment.d
1161 and ~/.config/environment.d/.
1163 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1166 * hostnamed has been updated to report a new chassis type of
1167 "convertible" to cover "foldable" laptops that can both act as a
1168 tablet and as a laptop, such as various Lenovo Yoga devices.
1170 Contributions from: Adrián López, Alexander Galanin, Alexander
1171 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1172 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1173 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1174 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1175 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1176 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1177 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1178 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1179 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1180 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1181 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1182 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1183 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1184 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1185 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1186 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1187 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1188 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1189 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1190 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1191 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1192 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1193 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1194 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1195 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1196 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1197 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1200 — Berlin, 2017-03-01
1204 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1205 RestrictAddressFamilies= enabled. These sandboxing options should
1206 generally be compatible with the various external udev call-out
1207 binaries we are aware of, however there may be exceptions, in
1208 particular when exotic languages for these call-outs are used. In
1209 this case, consider turning off these settings locally.
1211 * The new RemoveIPC= option can be used to remove IPC objects owned by
1212 the user or group of a service when that service exits.
1214 * The new ProtectKernelModules= option can be used to disable explicit
1215 load and unload operations of kernel modules by a service. In
1216 addition access to /usr/lib/modules is removed if this option is set.
1218 * ProtectSystem= option gained a new value "strict", which causes the
1219 whole file system tree with the exception of /dev, /proc, and /sys,
1220 to be remounted read-only for a service.
1222 * The new ProtectKernelTunables= option can be used to disable
1223 modification of configuration files in /sys and /proc by a service.
1224 Various directories and files are remounted read-only, so access is
1225 restricted even if the file permissions would allow it.
1227 * The new ProtectControlGroups= option can be used to disable write
1228 access by a service to /sys/fs/cgroup.
1230 * Various systemd services have been hardened with
1231 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1232 RestrictAddressFamilies=.
1234 * Support for dynamically creating users for the lifetime of a service
1235 has been added. If DynamicUser=yes is specified, user and group IDs
1236 will be allocated from the range 61184..65519 for the lifetime of the
1237 service. They can be resolved using the new nss-systemd.so NSS
1238 module. The module must be enabled in /etc/nsswitch.conf. Services
1239 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1240 any resources allocated by the service will be cleaned up when the
1241 service exits. They also have ProtectHome=read-only and
1242 ProtectSystem=strict enabled, so they are not able to make any
1243 permanent modifications to the system.
1245 * The nss-systemd module also always resolves root and nobody, making
1246 it possible to have no /etc/passwd or /etc/group files in minimal
1247 container or chroot environments.
1249 * Services may be started with their own user namespace using the new
1250 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1251 under which the service is running are mapped. All other users are
1254 * Support for the cgroup namespace has been added to systemd-nspawn. If
1255 supported by kernel, the container system started by systemd-nspawn
1256 will have its own view of the cgroup hierarchy. This new behaviour
1257 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1259 * The new MemorySwapMax= option can be used to limit the maximum swap
1260 usage under the unified cgroup hierarchy.
1262 * Support for the CPU controller in the unified cgroup hierarchy has
1263 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1264 options. This controller requires out-of-tree patches for the kernel
1265 and the support is provisional.
1267 * Mount and automount units may now be created transiently
1268 (i.e. dynamically at runtime via the bus API, instead of requiring
1269 unit files in the file system).
1271 * systemd-mount is a new tool which may mount file systems – much like
1272 mount(8), optionally pulling in additional dependencies through
1273 transient .mount and .automount units. For example, this tool
1274 automatically runs fsck on a backing block device before mounting,
1275 and allows the automount logic to be used dynamically from the
1276 command line for establishing mount points. This tool is particularly
1277 useful when dealing with removable media, as it will ensure fsck is
1278 run – if necessary – before the first access and that the file system
1279 is quickly unmounted after each access by utilizing the automount
1280 logic. This maximizes the chance that the file system on the
1281 removable media stays in a clean state, and if it isn't in a clean
1282 state is fixed automatically.
1284 * LazyUnmount=yes option for mount units has been added to expose the
1285 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1288 * /efi will be used as the mount point of the EFI boot partition, if
1289 the directory is present, and the mount point was not configured
1290 through other means (e.g. fstab). If /efi directory does not exist,
1291 /boot will be used as before. This makes it easier to automatically
1292 mount the EFI partition on systems where /boot is used for something
1295 * When operating on GPT disk images for containers, systemd-nspawn will
1296 now mount the ESP to /boot or /efi according to the same rules as PID
1297 1 running on a host. This allows tools like "bootctl" to operate
1298 correctly within such containers, in order to make container images
1299 bootable on physical systems.
1301 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1303 * Two new user session targets have been added to support running
1304 graphical sessions under the systemd --user instance:
1305 graphical-session.target and graphical-session-pre.target. See
1306 systemd.special(7) for a description of how those targets should be
1309 * The vconsole initialization code has been significantly reworked to
1310 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1311 support unicode keymaps. Font and keymap configuration will now be
1312 copied to all allocated virtual consoles.
1314 * FreeBSD's bhyve virtualization is now detected.
1316 * Information recorded in the journal for core dumps now includes the
1317 contents of /proc/mountinfo and the command line of the process at
1318 the top of the process hierarchy (which is usually the init process
1321 * systemd-journal-gatewayd learned the --directory= option to serve
1322 files from the specified location.
1324 * journalctl --root=… can be used to peruse the journal in the
1325 /var/log/ directories inside of a container tree. This is similar to
1326 the existing --machine= option, but does not require the container to
1329 * The hardware database has been extended to support
1330 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1333 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1334 specify the click rate for mice which include a horizontal wheel with
1335 a click rate that is different than the one for the vertical wheel.
1337 * systemd-run gained a new --wait option that makes service execution
1338 synchronous. (Specifically, the command will not return until the
1339 specified service binary exited.)
1341 * systemctl gained a new --wait option that causes the start command to
1342 wait until the units being started have terminated again.
1344 * A new journal output mode "short-full" has been added which displays
1345 timestamps with abbreviated English day names and adds a timezone
1346 suffix. Those timestamps include more information than the default
1347 "short" output mode, and can be passed directly to journalctl's
1348 --since= and --until= options.
1350 * /etc/resolv.conf will be bind-mounted into containers started by
1351 systemd-nspawn, if possible, so any changes to resolv.conf contents
1352 are automatically propagated to the container.
1354 * The number of instances for socket-activated services originating
1355 from a single IP address can be limited with
1356 MaxConnectionsPerSource=, extending the existing setting of
1359 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1362 * .netdev and .network configuration can now be extended through
1365 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1366 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1367 can be enabled and disabled using the new UDPSegmentationOffload=,
1368 TCPSegmentationOffload=, GenericSegmentationOffload=,
1369 GenericReceiveOffload=, LargeReceiveOffload= options in the
1370 [Link] section of .link files.
1372 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1373 Port VLAN ID can be configured for bridge devices using the new STP=,
1374 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1375 section of .netdev files.
1377 * The route table to which routes received over DHCP or RA should be
1378 added can be configured with the new RouteTable= option in the [DHCP]
1379 and [IPv6AcceptRA] sections of .network files.
1381 * The Address Resolution Protocol can be disabled on links managed by
1382 systemd-networkd using the ARP=no setting in the [Link] section of
1385 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1386 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1387 encode information about the result and exit codes of the current
1388 service runtime cycle.
1390 * systemd-sysctl will now configure kernel parameters in the order
1391 they occur in the configuration files. This matches what sysctl
1392 has been traditionally doing.
1394 * kernel-install "plugins" that are executed to perform various
1395 tasks after a new kernel is added and before an old one is removed
1396 can now return a special value to terminate the procedure and
1397 prevent any later plugins from running.
1399 * Journald's SplitMode=login setting has been deprecated. It has been
1400 removed from documentation, and its use is discouraged. In a future
1401 release it will be completely removed, and made equivalent to current
1402 default of SplitMode=uid.
1404 * Storage=both option setting in /etc/systemd/coredump.conf has been
1405 removed. With fast LZ4 compression storing the core dump twice is not
1408 * The --share-system systemd-nspawn option has been replaced with an
1409 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1410 this functionality is discouraged. In addition the variables
1411 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1412 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1413 individual namespaces.
1415 * "machinectl list" now shows the IP address of running containers in
1416 the output, as well as OS release information.
1418 * "loginctl list" now shows the TTY of each session in the output.
1420 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1421 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1422 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1423 tracking objects in a "recursive" mode, where a single client can be
1424 counted multiple times, if it takes multiple references.
1426 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1427 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1428 process using sd-bus automatically exit if the bus connection is
1431 * Bus clients of the service manager may now "pin" loaded units into
1432 memory, by taking an explicit reference on them. This is useful to
1433 ensure the client can retrieve runtime data about the service even
1434 after the service completed execution. Taking such a reference is
1435 available only for privileged clients and should be helpful to watch
1436 running services in a race-free manner, and in particular collect
1437 information about exit statuses and results.
1439 * The nss-resolve module has been changed to strictly return UNAVAIL
1440 when communication via D-Bus with resolved failed, and NOTFOUND when
1441 a lookup completed but was negative. This means it is now possible to
1442 neatly configure fallbacks using nsswitch.conf result checking
1443 expressions. Taking benefit of this, the new recommended
1444 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1446 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1448 * A new setting CtrlAltDelBurstAction= has been added to
1449 /etc/systemd/system.conf which may be used to configure the precise
1450 behaviour if the user on the console presses Ctrl-Alt-Del more often
1451 than 7 times in 2s. Previously this would unconditionally result in
1452 an expedited, immediate reboot. With this new setting the precise
1453 operation may be configured in more detail, and also turned off
1456 * In .netdev files two new settings RemoteChecksumTx= and
1457 RemoteChecksumRx= are now understood that permit configuring the
1458 remote checksumming logic for VXLAN networks.
1460 * The service manager learnt a new "invocation ID" concept for invoked
1461 services. Each runtime cycle of a service will get a new invocation
1462 ID (a 128bit random UUID) assigned that identifies the current
1463 run of the service uniquely and globally. A new invocation ID
1464 is generated each time a service starts up. The journal will store
1465 the invocation ID of a service along with any logged messages, thus
1466 making the invocation ID useful for matching the online runtime of a
1467 service with the offline log data it generated in a safe way without
1468 relying on synchronized timestamps. In many ways this new service
1469 invocation ID concept is similar to the kernel's boot ID concept that
1470 uniquely and globally identifies the runtime of each boot. The
1471 invocation ID of a service is passed to the service itself via an
1472 environment variable ($INVOCATION_ID). A new bus call
1473 GetUnitByInvocationID() has been added that is similar to GetUnit()
1474 but instead of retrieving the bus path for a unit by its name
1475 retrieves it by its invocation ID. The returned path is valid only as
1476 long as the passed invocation ID is current.
1478 * systemd-resolved gained a new "DNSStubListener" setting in
1479 resolved.conf. It either takes a boolean value or the special values
1480 "udp" and "tcp", and configures whether to enable the stub DNS
1481 listener on 127.0.0.53:53.
1483 * IP addresses configured via networkd may now carry additional
1484 configuration settings supported by the kernel. New options include:
1485 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1486 PrefixRoute=, AutoJoin=.
1488 * The PAM configuration fragment file for "user@.service" shipped with
1489 systemd (i.e. the --user instance of systemd) has been stripped to
1490 the minimum necessary to make the system boot. Previously, it
1491 contained Fedora-specific stanzas that did not apply to other
1492 distributions. It is expected that downstream distributions add
1493 additional configuration lines, matching their needs to this file,
1494 using it only as rough template of what systemd itself needs. Note
1495 that this reduced fragment does not even include an invocation of
1496 pam_limits which most distributions probably want to add, even though
1497 systemd itself does not need it. (There's also the new build time
1498 option --with-pamconfdir=no to disable installation of the PAM
1501 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1502 capability is now also dropped from its set (in addition to
1503 CAP_SYS_MKNOD as before).
1505 * In service unit files it is now possible to connect a specific named
1506 file descriptor with stdin/stdout/stdout of an executed service. The
1507 name may be specified in matching .socket units using the
1508 FileDescriptorName= setting.
1510 * A number of journal settings may now be configured on the kernel
1511 command line. Specifically, the following options are now understood:
1512 systemd.journald.max_level_console=,
1513 systemd.journald.max_level_store=,
1514 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1515 systemd.journald.max_level_wall=.
1517 * "systemctl is-enabled --full" will now show by which symlinks a unit
1518 file is enabled in the unit dependency tree.
1520 * Support for VeraCrypt encrypted partitions has been added to the
1521 "cryptsetup" logic and /etc/crypttab.
1523 * systemd-detect-virt gained support for a new --private-users switch
1524 that checks whether the invoking processes are running inside a user
1525 namespace. Similar, a new special value "private-users" for the
1526 existing ConditionVirtualization= setting has been added, permitting
1527 skipping of specific units in user namespace environments.
1529 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1530 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1531 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1532 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1533 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1534 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1535 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1536 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1537 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1538 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1539 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1540 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1541 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1542 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1543 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1544 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1545 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1546 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1547 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1548 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1549 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1550 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1551 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1552 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1553 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1554 Jędrzejewski-Szmek, Zeal Jagannatha
1556 — Santa Fe, 2016-11-03
1560 * In service units the various ExecXYZ= settings have been extended
1561 with an additional special character as first argument of the
1562 assigned value: if the character '+' is used the specified command
1563 line it will be run with full privileges, regardless of User=,
1564 Group=, CapabilityBoundingSet= and similar options. The effect is
1565 similar to the existing PermissionsStartOnly= option, but allows
1566 configuration of this concept for each executed command line
1569 * Services may now alter the service watchdog timeout at runtime by
1570 sending a WATCHDOG_USEC= message via sd_notify().
1572 * MemoryLimit= and related unit settings now optionally take percentage
1573 specifications. The percentage is taken relative to the amount of
1574 physical memory in the system (or in case of containers, the assigned
1575 amount of memory). This allows scaling service resources neatly with
1576 the amount of RAM available on the system. Similarly, systemd-logind's
1577 RuntimeDirectorySize= option now also optionally takes percentage
1580 * In similar fashion TasksMax= takes percentage values now, too. The
1581 value is taken relative to the configured maximum number of processes
1582 on the system. The per-service task maximum has been changed to 15%
1583 using this functionality. (Effectively this is an increase of 512 →
1584 4915 for service units, given the kernel's default pid_max setting.)
1586 * Calendar time specifications in .timer units now understand a ".."
1587 syntax for time ranges. Example: "4..7:10" may now be used for
1588 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1591 * The InaccessableDirectories=, ReadOnlyDirectories= and
1592 ReadWriteDirectories= unit file settings have been renamed to
1593 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1594 applied to all kinds of file nodes, and not just directories, with
1595 the exception of symlinks. Specifically these settings may now be
1596 used on block and character device nodes, UNIX sockets and FIFOS as
1597 well as regular files. The old names of these settings remain
1598 available for compatibility.
1600 * systemd will now log about all service processes it kills forcibly
1601 (using SIGKILL) because they remained after the clean shutdown phase
1602 of the service completed. This should help identifying services that
1603 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1604 systemd-logind's configuration a similar log message is generated for
1605 processes killed at the end of each session due to this setting.
1607 * systemd will now set the $JOURNAL_STREAM environment variable for all
1608 services whose stdout/stderr are connected to the Journal (which
1609 effectively means by default: all services). The variable contains
1610 the device and inode number of the file descriptor used for
1611 stdout/stderr. This may be used by invoked programs to detect whether
1612 their stdout/stderr is connected to the Journal, in which case they
1613 can switch over to direct Journal communication, thus being able to
1614 pass extended, structured metadata along with their log messages. As
1615 one example, this is now used by glib's logging primitives.
1617 * When using systemd's default tmp.mount unit for /tmp, the mount point
1618 will now be established with the "nosuid" and "nodev" options. This
1619 avoids privilege escalation attacks that put traps and exploits into
1620 /tmp. However, this might cause problems if you e. g. put container
1621 images or overlays into /tmp; if you need this, override tmp.mount's
1622 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1625 * systemd now supports the "memory" cgroup controller also on
1628 * The systemd-cgtop tool now optionally takes a control group path as
1629 command line argument. If specified, the control group list shown is
1630 limited to subgroups of that group.
1632 * The SystemCallFilter= unit file setting gained support for
1633 pre-defined, named system call filter sets. For example
1634 SystemCallFilter=@clock is now an effective way to make all clock
1635 changing-related system calls unavailable to a service. A number of
1636 similar pre-defined groups are defined. Writing system call filters
1637 for system services is simplified substantially with this new
1638 concept. Accordingly, all of systemd's own, long-running services now
1639 enable system call filtering based on this, by default.
1641 * A new service setting MemoryDenyWriteExecute= has been added, taking
1642 a boolean value. If turned on, a service may no longer create memory
1643 mappings that are writable and executable at the same time. This
1644 enhances security for services where this is enabled as it becomes
1645 harder to dynamically write and then execute memory in exploited
1646 service processes. This option has been enabled for all of systemd's
1647 own long-running services.
1649 * A new RestrictRealtime= service setting has been added, taking a
1650 boolean argument. If set the service's processes may no longer
1651 acquire realtime scheduling. This improves security as realtime
1652 scheduling may otherwise be used to easily freeze the system.
1654 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1655 value. This may be used for requesting that the system manager inside
1656 of the container reports start-up completion to nspawn which then
1657 propagates this notification further to the service manager
1658 supervising nspawn itself. A related option NotifyReady= in .nspawn
1659 files has been added too. This functionality allows ordering of the
1660 start-up of multiple containers using the usual systemd ordering
1663 * machinectl gained a new command "stop" that is an alias for
1666 * systemd-resolved gained support for contacting DNS servers on
1667 link-local IPv6 addresses.
1669 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1670 its caches. A method call for requesting the same operation has been
1671 added to the bus API too, and is made available via "systemd-resolve
1674 * systemd-resolve gained a new --status switch. If passed a brief
1675 summary of the used DNS configuration with per-interface information
1678 * resolved.conf gained a new Cache= boolean option, defaulting to
1679 on. If turned off local DNS caching is disabled. This comes with a
1680 performance penalty in particular when DNSSEC is enabled. Note that
1681 resolved disables its internal caching implicitly anyway, when the
1682 configured DNS server is on a host-local IP address such as ::1 or
1683 127.0.0.1, thus automatically avoiding double local caching.
1685 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1686 for DNS requests. This improves compatibility with local programs
1687 that do not use the libc NSS or systemd-resolved's bus APIs for name
1688 resolution. This minimal DNS service is only available to local
1689 programs and does not implement the full DNS protocol, but enough to
1690 cover local DNS clients. A new, static resolv.conf file, listing just
1691 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1692 now recommended to make /etc/resolv.conf a symlink to this file in
1693 order to route all DNS lookups to systemd-resolved, regardless if
1694 done via NSS, the bus API or raw DNS packets. Note that this local
1695 DNS service is not as fully featured as the libc NSS or
1696 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1697 used to deliver link-local address information (as this implies
1698 sending a local interface index along), LLMNR/mDNS support via this
1699 interface is severely restricted. It is thus strongly recommended for
1700 all applications to use the libc NSS API or native systemd-resolved
1703 * systemd-networkd's bridge support learned a new setting
1704 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1705 in .network files has been added for configuring VLAN bridging in
1706 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1708 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1709 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1710 now be acquired without relying on DHCPv6. Two new options
1711 UseDomains= and UseDNS= have been added to configure this behaviour.
1713 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1714 renamed IPv6AcceptRA=, without altering its behaviour. The old
1715 setting name remains available for compatibility reasons.
1717 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1718 Key=, InputKey= and OutputKey=.
1720 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1721 interface configuration.
1723 * "systemctl edit" may now be used to create new unit files by
1724 specifying the --force switch.
1726 * sd-event gained a new function sd_event_get_iteration() for
1727 requesting the current iteration counter of the event loop. It starts
1728 at zero and is increased by one with each event loop iteration.
1730 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1731 file. It can be used in lieu of %systemd_requires in packages which
1732 don't use any systemd functionality and are intended to be installed
1733 in minimal containers without systemd present. This macro provides
1734 ordering dependencies to ensure that if the package is installed in
1735 the same rpm transaction as systemd, systemd will be installed before
1736 the scriptlets for the package are executed, allowing unit presets
1739 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1740 been added to simplify packaging of generators.
1742 * The os-release file gained VERSION_CODENAME field for the
1743 distribution nickname (e.g. VERSION_CODENAME=woody).
1745 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1746 can be set to disable parsing of metadata and the creation
1747 of persistent symlinks for that device.
1749 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1750 to make them available to logged-in users has been reverted.
1752 * Much of the common code of the various systemd components is now
1753 built into an internal shared library libsystemd-shared-231.so
1754 (incorporating the systemd version number in the name, to be updated
1755 with future releases) that the components link to. This should
1756 decrease systemd footprint both in memory during runtime and on
1757 disk. Note that the shared library is not for public use, and is
1758 neither API not ABI stable, but is likely to change with every new
1759 released update. Packagers need to make sure that binaries
1760 linking to libsystemd-shared.so are updated in step with the
1763 * Configuration for "mkosi" is now part of the systemd
1764 repository. mkosi is a tool to easily build legacy-free OS images,
1765 and is available on github: https://github.com/systemd/mkosi. If
1766 "mkosi" is invoked in the build tree a new raw OS image is generated
1767 incorporating the systemd sources currently being worked on and a
1768 clean, fresh distribution installation. The generated OS image may be
1769 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1770 UEFI PC. This functionality is particularly useful to easily test
1771 local changes made to systemd in a pristine, defined environment. See
1772 HACKING for details.
1774 * configure learned the --with-support-url= option to specify the
1775 distribution's bugtracker.
1777 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1778 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1779 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1780 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1781 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1782 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1783 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1784 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1785 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1786 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1787 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1788 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1789 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1790 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1791 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1792 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1793 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1794 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1795 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1797 — Berlin, 2016-07-25
1801 * DNSSEC is now turned on by default in systemd-resolved (in
1802 "allow-downgrade" mode), but may be turned off during compile time by
1803 passing "--with-default-dnssec=no" to "configure" (and of course,
1804 during runtime with DNSSEC= in resolved.conf). We recommend
1805 downstreams to leave this on at least during development cycles and
1806 report any issues with the DNSSEC logic upstream. We are very
1807 interested in collecting feedback about the DNSSEC validator and its
1808 limitations in the wild. Note however, that DNSSEC support is
1809 probably nothing downstreams should turn on in stable distros just
1810 yet, as it might create incompatibilities with a few DNS servers and
1811 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1812 automatically whenever we detect such incompatible setups, but there
1813 might be systems we do not cover yet. Hence: please help us testing
1814 the DNSSEC code, leave this on where you can, report back, but then
1815 again don't consider turning this on in your stable, LTS or
1816 production release just yet. (Note that you have to enable
1817 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1818 and its DNSSEC mode for host name resolution from local
1821 * systemd-resolve conveniently resolves DANE records with the --tlsa
1822 option and OPENPGPKEY records with the --openpgp option. It also
1823 supports dumping raw DNS record data via the new --raw= switch.
1825 * systemd-logind will now by default terminate user processes that are
1826 part of the user session scope unit (session-XX.scope) when the user
1827 logs out. This behavior is controlled by the KillUserProcesses=
1828 setting in logind.conf, and the previous default of "no" is now
1829 changed to "yes". This means that user sessions will be properly
1830 cleaned up after, but additional steps are necessary to allow
1831 intentionally long-running processes to survive logout.
1833 While the user is logged in at least once, user@.service is running,
1834 and any service that should survive the end of any individual login
1835 session can be started at a user service or scope using systemd-run.
1836 systemd-run(1) man page has been extended with an example which shows
1837 how to run screen in a scope unit underneath user@.service. The same
1838 command works for tmux.
1840 After the user logs out of all sessions, user@.service will be
1841 terminated too, by default, unless the user has "lingering" enabled.
1842 To effectively allow users to run long-term tasks even if they are
1843 logged out, lingering must be enabled for them. See loginctl(1) for
1844 details. The default polkit policy was modified to allow users to
1845 set lingering for themselves without authentication.
1847 Previous defaults can be restored at compile time by the
1848 --without-kill-user-processes option to "configure".
1850 * systemd-logind gained new configuration settings SessionsMax= and
1851 InhibitorsMax=, both with a default of 8192. It will not register new
1852 user sessions or inhibitors above this limit.
1854 * systemd-logind will now reload configuration on SIGHUP.
1856 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1857 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1858 enable. Also, support for the "io" cgroup controller in the unified
1859 hierarchy has been added, so that the "memory", "pids" and "io" are
1860 now the controllers that are supported on the unified hierarchy.
1862 WARNING: it is not possible to use previous systemd versions with
1863 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1864 is necessary to also update systemd in the initramfs if using the
1865 unified hierarchy. An updated SELinux policy is also required.
1867 * LLDP support has been extended, and both passive (receive-only) and
1868 active (sender) modes are supported. Passive mode ("routers-only") is
1869 enabled by default in systemd-networkd. Active LLDP mode is enabled
1870 by default for containers on the internal network. The "networkctl
1871 lldp" command may be used to list information gathered. "networkctl
1872 status" will also show basic LLDP information on connected peers now.
1874 * The IAID and DUID unique identifier sent in DHCP requests may now be
1875 configured for the system and each .network file managed by
1876 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1878 * systemd-networkd gained support for configuring proxy ARP support for
1879 each interface, via the ProxyArp= setting in .network files. It also
1880 gained support for configuring the multicast querier feature of
1881 bridge devices, via the new MulticastQuerier= setting in .netdev
1882 files. Similarly, snooping on the IGMP traffic can be controlled
1883 via the new setting MulticastSnooping=.
1885 A new setting PreferredLifetime= has been added for addresses
1886 configured in .network file to configure the lifetime intended for an
1889 The systemd-networkd DHCP server gained the option EmitRouter=, which
1890 defaults to yes, to configure whether the DHCP Option 3 (Router)
1893 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1894 systemd-socket-activate and installed into /usr/bin. It is now fully
1897 * systemd-journald now uses separate threads to flush changes to disk
1898 when closing journal files, thus reducing impact of slow disk I/O on
1899 logging performance.
1901 * The sd-journal API gained two new calls
1902 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1903 can be used to open journal files using file descriptors instead of
1904 file or directory paths. sd_journal_open_container() has been
1905 deprecated, sd_journal_open_directory_fd() should be used instead
1906 with the flag SD_JOURNAL_OS_ROOT.
1908 * journalctl learned a new output mode "-o short-unix" that outputs log
1909 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1910 UTC). It also gained support for a new --no-hostname setting to
1911 suppress the hostname column in the family of "short" output modes.
1913 * systemd-ask-password now optionally skips printing of the password to
1914 stdout with --no-output which can be useful in scripts.
1916 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1917 (devices tagged with ID_MAKER_TOOL) are now tagged with
1918 "uaccess" and are available to logged in users.
1920 * The DeviceAllow= unit setting now supports specifiers (with "%").
1922 * "systemctl show" gained a new --value switch, which allows print a
1923 only the contents of a specific unit property, without also printing
1924 the property's name. Similar support was added to "show*" verbs
1925 of loginctl and machinectl that output "key=value" lists.
1927 * A new unit type "generated" was added for files dynamically generated
1928 by generator tools. Similarly, a new unit type "transient" is used
1929 for unit files created using the runtime API. "systemctl enable" will
1930 refuse to operate on such files.
1932 * A new command "systemctl revert" has been added that may be used to
1933 revert to the vendor version of a unit file, in case local changes
1934 have been made by adding drop-ins or overriding the unit file.
1936 * "machinectl clean" gained a new verb to automatically remove all or
1937 just hidden container images.
1939 * systemd-tmpfiles gained support for a new line type "e" for emptying
1940 directories, if they exist, without creating them if they don't.
1942 * systemd-nspawn gained support for automatically patching the UID/GIDs
1943 of the owners and the ACLs of all files and directories in a
1944 container tree to match the UID/GID user namespacing range selected
1945 for the container invocation. This mode is enabled via the new
1946 --private-users-chown switch. It also gained support for
1947 automatically choosing a free, previously unused UID/GID range when
1948 starting a container, via the new --private-users=pick setting (which
1949 implies --private-users-chown). Together, these options for the first
1950 time make user namespacing for nspawn containers fully automatic and
1951 thus deployable. The systemd-nspawn@.service template unit file has
1952 been changed to use this functionality by default.
1954 * systemd-nspawn gained a new --network-zone= switch, that allows
1955 creating ad-hoc virtual Ethernet links between multiple containers,
1956 that only exist as long as at least one container referencing them is
1957 running. This allows easy connecting of multiple containers with a
1958 common link that implements an Ethernet broadcast domain. Each of
1959 these network "zones" may be named relatively freely by the user, and
1960 may be referenced by any number of containers, but each container may
1961 only reference one of these "zones". On the lower level, this is
1962 implemented by an automatically managed bridge network interface for
1963 each zone, that is created when the first container referencing its
1964 zone is created and removed when the last one referencing its zone
1967 * The default start timeout may now be configured on the kernel command
1968 line via systemd.default_timeout_start_sec=. It was already
1969 configurable via the DefaultTimeoutStartSec= option in
1970 /etc/systemd/system.conf.
1972 * Socket units gained a new TriggerLimitIntervalSec= and
1973 TriggerLimitBurst= setting to configure a limit on the activation
1974 rate of the socket unit.
1976 * The LimitNICE= setting now optionally takes normal UNIX nice values
1977 in addition to the raw integer limit value. If the specified
1978 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1979 value is understood as UNIX nice value. If not prefixed like this it
1980 is understood as raw RLIMIT_NICE limit.
1982 * Note that the effect of the PrivateDevices= unit file setting changed
1983 slightly with this release: the per-device /dev file system will be
1984 mounted read-only from this version on, and will have "noexec"
1985 set. This (minor) change of behavior might cause some (exceptional)
1986 legacy software to break, when PrivateDevices=yes is set for its
1987 service. Please leave PrivateDevices= off if you run into problems
1990 * systemd-bootchart has been split out to a separate repository:
1991 https://github.com/systemd/systemd-bootchart
1993 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1994 merged into the kernel in its current form.
1996 * The compatibility libraries libsystemd-daemon.so,
1997 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1998 which have been deprecated since systemd-209 have been removed along
1999 with the corresponding pkg-config files. All symbols provided by
2000 those libraries are provided by libsystemd.so.
2002 * The Capabilities= unit file setting has been removed (it is ignored
2003 for backwards compatibility). AmbientCapabilities= and
2004 CapabilityBoundingSet= should be used instead.
2006 * A new special target has been added, initrd-root-device.target,
2007 which creates a synchronization point for dependencies of the root
2008 device in early userspace. Initramfs builders must ensure that this
2009 target is now included in early userspace.
2011 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2012 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2013 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2014 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2015 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2016 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2017 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2018 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2019 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2020 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2021 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2022 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2023 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2024 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2025 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2026 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2027 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2028 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2029 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2030 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2031 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2032 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2033 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2034 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2035 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2038 — Fairfax, 2016-05-21
2042 * The systemd-resolved DNS resolver service has gained a substantial
2043 set of new features, most prominently it may now act as a DNSSEC
2044 validating stub resolver. DNSSEC mode is currently turned off by
2045 default, but is expected to be turned on by default in one of the
2046 next releases. For now, we invite everybody to test the DNSSEC logic
2047 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2048 service also gained a full set of D-Bus interfaces, including calls
2049 to configure DNS and DNSSEC settings per link (for use by external
2050 network management software). systemd-resolved and systemd-networkd
2051 now distinguish between "search" and "routing" domains. The former
2052 are used to qualify single-label names, the latter are used purely
2053 for routing lookups within certain domains to specific links.
2054 resolved now also synthesizes RRs for all entries from /etc/hosts.
2056 * The systemd-resolve tool (which is a client utility for
2057 systemd-resolved) has been improved considerably and is now fully
2058 supported and documented. Hence it has moved from /usr/lib/systemd to
2061 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2064 * The coredump collection logic has been reworked: when a coredump is
2065 collected it is now written to disk, compressed and processed
2066 (including stacktrace extraction) from a new instantiated service
2067 systemd-coredump@.service, instead of directly from the
2068 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2069 processing large coredumps can take up a substantial amount of
2070 resources and time, and this previously happened entirely outside of
2071 systemd's service supervision. With the new logic the core_pattern
2072 hook only does minimal metadata collection before passing off control
2073 to the new instantiated service, which is configured with a time
2074 limit, a nice level and other settings to minimize negative impact on
2075 the rest of the system. Also note that the new logic will honour the
2076 RLIMIT_CORE setting of the crashed process, which now allows users
2077 and processes to turn off coredumping for their processes by setting
2080 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2081 and all forked processes by default. Previously, PID 1 would leave
2082 the setting at "0" for all processes, as set by the kernel. Note that
2083 the resource limit traditionally has no effect on the generated
2084 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2085 logic is used. Since the limit is now honoured (see above) its
2086 default has been changed so that the coredumping logic is enabled by
2087 default for all processes, while allowing specific opt-out.
2089 * When the stacktrace is extracted from processes of system users, this
2090 is now done as "systemd-coredump" user, in order to sandbox this
2091 potentially security sensitive parsing operation. (Note that when
2092 processing coredumps of normal users this is done under the user ID
2093 of process that crashed, as before.) Packagers should take notice
2094 that it is now necessary to create the "systemd-coredump" system user
2095 and group at package installation time.
2097 * The systemd-activate socket activation testing tool gained support
2098 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2099 and --seqpacket switches. It also has been extended to support both
2100 new-style and inetd-style file descriptor passing. Use the new
2101 --inetd switch to request inetd-style file descriptor passing.
2103 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2104 variable, which takes a boolean value. If set to false, ANSI color
2105 output is disabled in the tools even when run on a terminal that
2108 * The VXLAN support in networkd now supports two new settings
2109 DestinationPort= and PortRange=.
2111 * A new systemd.machine_id= kernel command line switch has been added,
2112 that may be used to set the machine ID in /etc/machine-id if it is
2113 not initialized yet. This command line option has no effect if the
2114 file is already initialized.
2116 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2117 specified command line as PID 2 rather than PID 1 in the
2118 container. In this mode PID 1 is a minimal stub init process that
2119 implements the special POSIX and Linux semantics of PID 1 regarding
2120 signal and child process management. Note that this stub init process
2121 is implemented in nspawn itself and requires no support from the
2122 container image. This new logic is useful to support running
2123 arbitrary commands in the container, as normal processes are
2124 generally not prepared to run as PID 1.
2126 * systemd-nspawn gained a new --chdir= switch for setting the current
2127 working directory for the process started in the container.
2129 * "journalctl /dev/sda" will now output all kernel log messages for
2130 specified device from the current boot, in addition to all devices
2131 that are parents of it. This should make log output about devices
2132 pretty useful, as long as kernel drivers attach enough metadata to
2133 the log messages. (The usual SATA drivers do.)
2135 * The sd-journal API gained two new calls
2136 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2137 that report whether log data from /run or /var has been found.
2139 * journalctl gained a new switch "--fields" that prints all journal
2140 record field names currently in use in the journal. This is backed
2141 by two new sd-journal API calls sd_journal_enumerate_fields() and
2142 sd_journal_restart_fields().
2144 * Most configurable timeouts in systemd now expect an argument of
2145 "infinity" to turn them off, instead of "0" as before. The semantics
2146 from now on is that a timeout of "0" means "now", and "infinity"
2147 means "never". To maintain backwards compatibility, "0" continues to
2148 turn off previously existing timeout settings.
2150 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2151 try-reload-or-restart" to clarify what it actually does: the "try"
2152 logic applies to both reloading and restarting, not just restarting.
2153 The old name continues to be accepted for compatibility.
2155 * On boot-up, when PID 1 detects that the system clock is behind the
2156 release date of the systemd version in use, the clock is now set
2157 to the latter. Previously, this was already done in timesyncd, in order
2158 to avoid running with clocks set to the various clock epochs such as
2159 1902, 1938 or 1970. With this change the logic is now done in PID 1
2160 in addition to timesyncd during early boot-up, so that it is enforced
2161 before the first process is spawned by systemd. Note that the logic
2162 in timesyncd remains, as it is more comprehensive and ensures
2163 clock monotonicity by maintaining a persistent timestamp file in
2164 /var. Since /var is generally not available in earliest boot or the
2165 initrd, this part of the logic remains in timesyncd, and is not done
2168 * Support for tweaking details in net_cls.class_id through the
2169 NetClass= configuration directive has been removed, as the kernel
2170 people have decided to deprecate that controller in cgroup v2.
2171 Userspace tools such as nftables are moving over to setting rules
2172 that are specific to the full cgroup path of a task, which obsoletes
2173 these controllers anyway. The NetClass= directive is kept around for
2174 legacy compatibility reasons. For a more in-depth description of the
2175 kernel change, please refer to the respective upstream commit:
2177 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2179 * A new service setting RuntimeMaxSec= has been added that may be used
2180 to specify a maximum runtime for a service. If the timeout is hit, the
2181 service is terminated and put into a failure state.
2183 * A new service setting AmbientCapabilities= has been added. It allows
2184 configuration of additional Linux process capabilities that are
2185 passed to the activated processes. This is only available on very
2188 * The process resource limit settings in service units may now be used
2189 to configure hard and soft limits individually.
2191 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2192 expose support for gcc's __attribute__((cleanup())) C extension.
2193 Specifically, for many object destructor functions alternative
2194 versions have been added that have names suffixed with "p" and take a
2195 pointer to a pointer to the object to destroy, instead of just a
2196 pointer to the object itself. This is useful because these destructor
2197 functions may be used directly as parameters to the cleanup
2198 construct. Internally, systemd has been a heavy user of this GCC
2199 extension for a long time, and with this change similar support is
2200 now available to consumers of the library outside of systemd. Note
2201 that by using this extension in your sources compatibility with old
2202 and strictly ANSI compatible C compilers is lost. However, all gcc or
2203 LLVM versions of recent years support this extension.
2205 * Timer units gained support for a new setting RandomizedDelaySec= that
2206 allows configuring some additional randomized delay to the configured
2207 time. This is useful to spread out timer events to avoid load peaks in
2208 clusters or larger setups.
2210 * Calendar time specifications now support sub-second accuracy.
2212 * Socket units now support listening on SCTP and UDP-lite protocol
2215 * The sd-event API now comes with a full set of man pages.
2217 * Older versions of systemd contained experimental support for
2218 compressing journal files and coredumps with the LZ4 compressor that
2219 was not compatible with the lz4 binary (due to API limitations of the
2220 lz4 library). This support has been removed; only support for files
2221 compatible with the lz4 binary remains. This LZ4 logic is now
2222 officially supported and no longer considered experimental.
2224 * The dkr image import logic has been removed again from importd. dkr's
2225 micro-services focus doesn't fit into the machine image focus of
2226 importd, and quickly got out of date with the upstream dkr API.
2228 * Creation of the /run/lock/lockdev/ directory was dropped from
2229 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2230 been available for many years. If you still need this, you need to
2231 create your own tmpfiles.d config file with:
2233 d /run/lock/lockdev 0775 root lock -
2235 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2236 and RebootArgument= have been moved from the [Service] section of
2237 unit files to [Unit], and they are now supported on all unit types,
2238 not just service units. Of course, elogind will continue to
2239 understand these settings also at the old location, in order to
2240 maintain compatibility.
2242 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2243 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2244 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2245 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2246 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2247 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2248 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2249 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2250 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2251 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2252 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2253 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2254 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2255 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2256 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2257 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2258 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2259 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2260 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2262 — Berlin, 2016-02-11
2266 * A number of properties previously only settable in unit
2267 files are now also available as properties to set when
2268 creating transient units programmatically via the bus, as it
2269 is exposed with systemd-run's --property=
2270 setting. Specifically, these are: SyslogIdentifier=,
2271 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2272 EnvironmentFile=, ReadWriteDirectories=,
2273 ReadOnlyDirectories=, InaccessibleDirectories=,
2274 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2276 * When creating transient services via the bus API it is now
2277 possible to pass in a set of file descriptors to use as
2278 STDIN/STDOUT/STDERR for the invoked process.
2280 * Slice units may now be created transiently via the bus APIs,
2281 similar to the way service and scope units may already be
2282 created transiently.
2284 * Wherever systemd expects a calendar timestamp specification
2285 (like in journalctl's --since= and --until= switches) UTC
2286 timestamps are now supported. Timestamps suffixed with "UTC"
2287 are now considered to be in Universal Time Coordinated
2288 instead of the local timezone. Also, timestamps may now
2289 optionally be specified with sub-second accuracy. Both of
2290 these additions also apply to recurring calendar event
2291 specification, such as OnCalendar= in timer units.
2293 * journalctl gained a new "--sync" switch that asks the
2294 journal daemon to write all so far unwritten log messages to
2295 disk and sync the files, before returning.
2297 * systemd-tmpfiles learned two new line types "q" and "Q" that
2298 operate like "v", but also set up a basic btrfs quota
2299 hierarchy when used on a btrfs file system with quota
2302 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2303 instead of a subvolume (even on a btrfs file system) if the
2304 root directory is a plain directory, and not a
2305 subvolume. This should simplify things with certain chroot()
2306 environments which are not aware of the concept of btrfs
2309 * systemd-detect-virt gained a new --chroot switch to detect
2310 whether execution takes place in a chroot() environment.
2312 * CPUAffinity= now takes CPU index ranges in addition to
2315 * The various memory-related resource limit settings (such as
2316 LimitAS=) now understand the usual K, M, G, ... suffixes to
2317 the base of 1024 (IEC). Similar, the time-related resource
2318 limit settings understand the usual min, h, day, ...
2321 * There's a new system.conf setting DefaultTasksMax= to
2322 control the default TasksMax= setting for services and
2323 scopes running on the system. (TasksMax= is the primary
2324 setting that exposes the "pids" cgroup controller on systemd
2325 and was introduced in the previous systemd release.) The
2326 setting now defaults to 512, which means services that are
2327 not explicitly configured otherwise will only be able to
2328 create 512 processes or threads at maximum, from this
2329 version on. Note that this means that thread- or
2330 process-heavy services might need to be reconfigured to set
2331 TasksMax= to a higher value. It is sufficient to set
2332 TasksMax= in these specific unit files to a higher value, or
2333 even "infinity". Similar, there's now a logind.conf setting
2334 UserTasksMax= that defaults to 4096 and limits the total
2335 number of processes or tasks each user may own
2336 concurrently. nspawn containers also have the TasksMax=
2337 value set by default now, to 8192. Note that all of this
2338 only has an effect if the "pids" cgroup controller is
2339 enabled in the kernel. The general benefit of these changes
2340 should be a more robust and safer system, that provides a
2341 certain amount of per-service fork() bomb protection.
2343 * systemd-nspawn gained the new --network-veth-extra= switch
2344 to define additional and arbitrarily-named virtual Ethernet
2345 links between the host and the container.
2347 * A new service execution setting PassEnvironment= has been
2348 added that allows importing select environment variables
2349 from PID1's environment block into the environment block of
2352 * Timer units gained support for a new RemainAfterElapse=
2353 setting which takes a boolean argument. It defaults to on,
2354 exposing behaviour unchanged to previous releases. If set to
2355 off, timer units are unloaded after they elapsed if they
2356 cannot elapse again. This is particularly useful for
2357 transient timer units, which shall not stay around longer
2358 than until they first elapse.
2360 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2361 default now (the kernel default is 16). This is beneficial
2362 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2363 allows substantially larger numbers of queued
2364 datagrams. This should increase the capability of systemd to
2365 parallelize boot-up, as logging and sd_notify() are unlikely
2366 to stall execution anymore. If you need to change the value
2367 from the new defaults, use the usual sysctl.d/ snippets.
2369 * The compression framing format used by the journal or
2370 coredump processing has changed to be in line with what the
2371 official LZ4 tools generate. LZ4 compression support in
2372 systemd was considered unsupported previously, as the format
2373 was not compatible with the normal tools. With this release
2374 this has changed now, and it is hence safe for downstream
2375 distributions to turn it on. While not compressing as well
2376 as the XZ, LZ4 is substantially faster, which makes
2377 it a good default choice for the compression logic in the
2378 journal and in coredump handling.
2380 * Any reference to /etc/mtab has been dropped from
2381 systemd. The file has been obsolete since a while, but
2382 systemd refused to work on systems where it was incorrectly
2383 set up (it should be a symlink or non-existent). Please make
2384 sure to update to util-linux 2.27.1 or newer in conjunction
2385 with this systemd release, which also drops any reference to
2386 /etc/mtab. If you maintain a distribution make sure that no
2387 software you package still references it, as this is a
2388 likely source of bugs. There's also a glibc bug pending,
2389 asking for removal of any reference to this obsolete file:
2391 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2393 Note that only util-linux versions built with
2394 --enable-libmount-force-mountinfo are supported.
2396 * Support for the ".snapshot" unit type has been removed. This
2397 feature turned out to be little useful and little used, and
2398 has now been removed from the core and from systemctl.
2400 * The dependency types RequiresOverridable= and
2401 RequisiteOverridable= have been removed from systemd. They
2402 have been used only very sparingly to our knowledge and
2403 other options that provide a similar effect (such as
2404 systemctl --mode=ignore-dependencies) are much more useful
2405 and commonly used. Moreover, they were only half-way
2406 implemented as the option to control behaviour regarding
2407 these dependencies was never added to systemctl. By removing
2408 these dependency types the execution engine becomes a bit
2409 simpler. Unit files that use these dependencies should be
2410 changed to use the non-Overridable dependency types
2411 instead. In fact, when parsing unit files with these
2412 options, that's what systemd will automatically convert them
2413 too, but it will also warn, asking users to fix the unit
2414 files accordingly. Removal of these dependency types should
2415 only affect a negligible number of unit files in the wild.
2417 * Behaviour of networkd's IPForward= option changed
2418 (again). It will no longer maintain a per-interface setting,
2419 but propagate one way from interfaces where this is enabled
2420 to the global kernel setting. The global setting will be
2421 enabled when requested by a network that is set up, but
2422 never be disabled again. This change was made to make sure
2423 IPv4 and IPv6 behaviour regarding packet forwarding is
2424 similar (as the Linux IPv6 stack does not support
2425 per-interface control of this setting) and to minimize
2428 * In unit files the behaviour of %u, %U, %h, %s has
2429 changed. These specifiers will now unconditionally resolve
2430 to the various user database fields of the user that the
2431 systemd instance is running as, instead of the user
2432 configured in the specific unit via User=. Note that this
2433 effectively doesn't change much, as resolving of these
2434 specifiers was already turned off in the --system instance
2435 of systemd, as we cannot do NSS lookups from PID 1. In the
2436 --user instance of systemd these specifiers where correctly
2437 resolved, but hardly made any sense, since the user instance
2438 lacks privileges to do user switches anyway, and User= is
2439 hence useless. Moreover, even in the --user instance of
2440 systemd behaviour was awkward as it would only take settings
2441 from User= assignment placed before the specifier into
2442 account. In order to unify and simplify the logic around
2443 this the specifiers will now always resolve to the
2444 credentials of the user invoking the manager (which in case
2445 of PID 1 is the root user).
2447 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2448 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2449 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2450 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2451 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2452 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2453 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2454 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2455 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2456 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2457 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2458 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2459 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2460 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2463 — Berlin, 2015-11-18
2467 * systemd now depends on util-linux v2.27. More specifically,
2468 the newly added mount monitor feature in libmount now
2469 replaces systemd's former own implementation.
2471 * libmount mandates /etc/mtab not to be regular file, and
2472 systemd now enforces this condition at early boot.
2473 /etc/mtab has been deprecated and warned about for a very
2474 long time, so systems running systemd should already have
2475 stopped having this file around as anything else than a
2476 symlink to /proc/self/mounts.
2478 * Support for the "pids" cgroup controller has been added. It
2479 allows accounting the number of tasks in a cgroup and
2480 enforcing limits on it. This adds two new setting
2481 TasksAccounting= and TasksMax= to each unit, as well as a
2482 global option DefaultTasksAccounting=.
2484 * Support for the "net_cls" cgroup controller has been added.
2485 It allows assigning a net class ID to each task in the
2486 cgroup, which can then be used in firewall rules and traffic
2487 shaping configurations. Note that the kernel netfilter net
2488 class code does not currently work reliably for ingress
2489 packets on unestablished sockets.
2491 This adds a new config directive called NetClass= to CGroup
2492 enabled units. Allowed values are positive numbers for fixed
2493 assignments and "auto" for picking a free value
2496 * 'systemctl is-system-running' now returns 'offline' if the
2497 system is not booted with systemd. This command can now be
2498 used as a substitute for 'systemd-notify --booted'.
2500 * Watchdog timeouts have been increased to 3 minutes for all
2501 in-tree service files. Apparently, disk IO issues are more
2502 frequent than we hoped, and user reported >1 minute waiting
2505 * 'machine-id-commit' functionality has been merged into
2506 'machine-id-setup --commit'. The separate binary has been
2509 * The WorkingDirectory= directive in unit files may now be set
2510 to the special value '~'. In this case, the working
2511 directory is set to the home directory of the user
2512 configured in User=.
2514 * "machinectl shell" will now open the shell in the home
2515 directory of the selected user by default.
2517 * The CrashChVT= configuration file setting is renamed to
2518 CrashChangeVT=, following our usual logic of not
2519 abbreviating unnecessarily. The old directive is still
2520 supported for compat reasons. Also, this directive now takes
2521 an integer value between 1 and 63, or a boolean value. The
2522 formerly supported '-1' value for disabling stays around for
2525 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2526 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2527 RootDirectory= properties can now be set for transient
2530 * The systemd-analyze tool gained a new "set-log-target" verb
2531 to change the logging target the system manager logs to
2532 dynamically during runtime. This is similar to how
2533 "systemd-analyze set-log-level" already changes the log
2536 * In nspawn /sys is now mounted as tmpfs, with only a selected
2537 set of subdirectories mounted in from the real sysfs. This
2538 enhances security slightly, and is useful for ensuring user
2539 namespaces work correctly.
2541 * Support for USB FunctionFS activation has been added. This
2542 allows implementation of USB gadget services that are
2543 activated as soon as they are requested, so that they don't
2544 have to run continuously, similar to classic socket
2547 * The "systemctl exit" command now optionally takes an
2548 additional parameter that sets the exit code to return from
2549 the systemd manager when exiting. This is only relevant when
2550 running the systemd user instance, or when running the
2551 system instance in a container.
2553 * sd-bus gained the new API calls sd_bus_path_encode_many()
2554 and sd_bus_path_decode_many() that allow easy encoding and
2555 decoding of multiple identifier strings inside a D-Bus
2556 object path. Another new call sd_bus_default_flush_close()
2557 has been added to flush and close per-thread default
2560 * systemd-cgtop gained support for a -M/--machine= switch to
2561 show the control groups within a certain container only.
2563 * "systemctl kill" gained support for an optional --fail
2564 switch. If specified the requested operation will fail of no
2565 processes have been killed, because the unit had no
2566 processes attached, or similar.
2568 * A new systemd.crash_reboot=1 kernel command line option has
2569 been added that triggers a reboot after crashing. This can
2570 also be set through CrashReboot= in systemd.conf.
2572 * The RuntimeDirectory= setting now understands unit
2573 specifiers like %i or %f.
2575 * A new (still internal) library API sd-ipv4acd has been added,
2576 that implements address conflict detection for IPv4. It's
2577 based on code from sd-ipv4ll, and will be useful for
2578 detecting DHCP address conflicts.
2580 * File descriptors passed during socket activation may now be
2581 named. A new API sd_listen_fds_with_names() is added to
2582 access the names. The default names may be overridden,
2583 either in the .socket file using the FileDescriptorName=
2584 parameter, or by passing FDNAME= when storing the file
2585 descriptors using sd_notify().
2587 * systemd-networkd gained support for:
2589 - Setting the IPv6 Router Advertisement settings via
2590 IPv6AcceptRouterAdvertisements= in .network files.
2592 - Configuring the HelloTimeSec=, MaxAgeSec= and
2593 ForwardDelaySec= bridge parameters in .netdev files.
2595 - Configuring PreferredSource= for static routes in
2598 * The "ask-password" framework used to query for LUKS harddisk
2599 passwords or SSL passwords during boot gained support for
2600 caching passwords in the kernel keyring, if it is
2601 available. This makes sure that the user only has to type in
2602 a passphrase once if there are multiple objects to unlock
2603 with the same one. Previously, such password caching was
2604 available only when Plymouth was used; this moves the
2605 caching logic into the systemd codebase itself. The
2606 "systemd-ask-password" utility gained a new --keyname=
2607 switch to control which kernel keyring key to use for
2608 caching a password in. This functionality is also useful for
2609 enabling display managers such as gdm to automatically
2610 unlock the user's GNOME keyring if its passphrase, the
2611 user's password and the harddisk password are the same, if
2612 gdm-autologin is used.
2614 * When downloading tar or raw images using "machinectl
2615 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2616 file is now also downloaded, if it is available and stored
2617 next to the image file.
2619 * Units of type ".socket" gained a new boolean setting
2620 Writable= which is only useful in conjunction with
2621 ListenSpecial=. If true, enables opening the specified
2622 special file in O_RDWR mode rather than O_RDONLY mode.
2624 * systemd-rfkill has been reworked to become a singleton
2625 service that is activated through /dev/rfkill on each rfkill
2626 state change and saves the settings to disk. This way,
2627 systemd-rfkill is now compatible with devices that exist
2628 only intermittendly, and even restores state if the previous
2629 system shutdown was abrupt rather than clean.
2631 * The journal daemon gained support for vacuuming old journal
2632 files controlled by the number of files that shall remain,
2633 in addition to the already existing control by size and by
2634 date. This is useful as journal interleaving performance
2635 degrades with too many separate journal files, and allows
2636 putting an effective limit on them. The new setting defaults
2637 to 100, but this may be changed by setting SystemMaxFiles=
2638 and RuntimeMaxFiles= in journald.conf. Also, the
2639 "journalctl" tool gained the new --vacuum-files= switch to
2640 manually vacuum journal files to leave only the specified
2641 number of files in place.
2643 * udev will now create /dev/disk/by-path links for ATA devices
2644 on kernels where that is supported.
2646 * Galician, Serbian, Turkish and Korean translations were added.
2648 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2649 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2650 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2651 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2652 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2653 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2654 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2655 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2656 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2657 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2658 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2659 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2660 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2661 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2662 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2663 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2664 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2665 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2667 — Berlin, 2015-10-07
2671 * The DHCP implementation of systemd-networkd gained a set of
2674 - The DHCP server now supports emitting DNS and NTP
2675 information. It may be enabled and configured via
2676 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2677 and NTP information is enabled, but no servers are
2678 configured, the corresponding uplink information (if there
2679 is any) is propagated.
2681 - Server and client now support transmission and reception
2682 of timezone information. It can be configured via the
2683 newly introduced network options UseTimezone=,
2684 EmitTimezone=, and Timezone=. Transmission of timezone
2685 information is enabled between host and containers by
2686 default now: the container will change its local timezone
2687 to what the host has set.
2689 - Lease timeouts can now be configured via
2690 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2692 - The DHCP server improved on the stability of
2693 leases. Clients are more likely to get the same lease
2694 information back, even if the server loses state.
2696 - The DHCP server supports two new configuration options to
2697 control the lease address pool metrics, PoolOffset= and
2700 * The encapsulation limit of tunnels in systemd-networkd may
2701 now be configured via 'EncapsulationLimit='. It allows
2702 modifying the maximum additional levels of encapsulation
2703 that are permitted to be prepended to a packet.
2705 * systemd now supports the concept of user buses replacing
2706 session buses, if used with dbus-1.10 (and enabled via dbus
2707 --enable-user-session). It previously only supported this on
2708 kdbus-enabled systems, and this release expands this to
2709 'dbus-daemon' systems.
2711 * systemd-networkd now supports predictable interface names
2714 * systemd now optionally supports the new Linux kernel
2715 "unified" control group hierarchy. If enabled via the kernel
2716 command-line option 'systemd.unified_cgroup_hierarchy=1',
2717 systemd will try to mount the unified cgroup hierarchy
2718 directly on /sys/fs/cgroup. If not enabled, or not
2719 available, systemd will fall back to the legacy cgroup
2720 hierarchy setup, as before. Host system and containers can
2721 mix and match legacy and unified hierarchies as they
2722 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2723 environment variable to individually select the hierarchy to
2724 use for executed containers. By default, nspawn will use the
2725 unified hierarchy for the containers if the host uses the
2726 unified hierarchy, and the legacy hierarchy otherwise.
2727 Please note that at this point the unified hierarchy is an
2728 experimental kernel feature and is likely to change in one
2729 of the next kernel releases. Therefore, it should not be
2730 enabled by default in downstream distributions yet. The
2731 minimum required kernel version for the unified hierarchy to
2732 work is 4.2. Note that when the unified hierarchy is used
2733 for the first time delegated access to controllers is
2734 safe. Because of this systemd-nspawn containers will get
2735 access to controllers now, as will systemd user
2736 sessions. This means containers and user sessions may now
2737 manage their own resources, partitioning up what the system
2740 * A new special scope unit "init.scope" has been introduced
2741 that encapsulates PID 1 of the system. It may be used to
2742 determine resource usage and enforce resource limits on PID
2743 1 itself. PID 1 hence moved out of the root of the control
2746 * The cgtop tool gained support for filtering out kernel
2747 threads when counting tasks in a control group. Also, the
2748 count of processes is now recursively summed up by
2749 default. Two options -k and --recursive= have been added to
2750 revert to old behaviour. The tool has also been updated to
2751 work correctly in containers now.
2753 * systemd-nspawn's --bind= and --bind-ro= options have been
2754 extended to allow creation of non-recursive bind mounts.
2756 * libsystemd gained two new calls sd_pid_get_cgroup() and
2757 sd_peer_get_cgroup() which return the control group path of
2758 a process or peer of a connected AF_UNIX socket. This
2759 function call is particularly useful when implementing
2760 delegated subtrees support in the control group hierarchy.
2762 * The "sd-event" event loop API of libsystemd now supports
2763 correct dequeuing of real-time signals, without losing
2766 * When systemd requests a PolicyKit decision when managing
2767 units it will now add additional fields to the request,
2768 including unit name and desired operation. This enables more
2769 powerful PolicyKit policies, that make decisions depending
2770 on these parameters.
2772 * nspawn learnt support for .nspawn settings files, that may
2773 accompany the image files or directories of containers, and
2774 may contain additional settings for the container. This is
2775 an alternative to configuring container parameters via the
2776 nspawn command line.
2778 Contributions from: Cristian Rodríguez, Daniel Mack, David
2779 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2780 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2781 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2782 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2783 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2784 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2785 Andersen, Tom Gundersen, Torstein Husebø
2787 — Berlin, 2015-09-08
2791 * machinectl gained a new verb 'shell' which opens a fresh
2792 shell on the target container or the host. It is similar to
2793 the existing 'login' command of machinectl, but spawns the
2794 shell directly without prompting for username or
2795 password. The pseudo machine '.host' now refers to the local
2796 host and is used by default. Hence, 'machinectl shell' can
2797 be used as replacement for 'su -' which spawns a session as
2798 a fresh systemd unit in a way that is fully isolated from
2799 the originating session.
2801 * systemd-networkd learned to cope with private-zone DHCP
2802 options and allows other programs to query the values.
2804 * SELinux access control when enabling/disabling units is no
2805 longer enforced with this release. The previous
2806 implementation was incorrect, and a new corrected
2807 implementation is not yet available. As unit file operations
2808 are still protected via PolicyKit and D-Bus policy this is
2809 not a security problem. Yet, distributions which care about
2810 optimal SELinux support should probably not stabilize on
2813 * sd-bus gained support for matches of type "arg0has=", that
2814 test for membership of strings in string arrays sent in bus
2817 * systemd-resolved now dumps the contents of its DNS and LLMNR
2818 caches to the logs on reception of the SIGUSR1 signal. This
2819 is useful to debug DNS behaviour.
2821 * The coredumpctl tool gained a new --directory= option to
2822 operate on journal files in a specific directory.
2824 * "systemctl reboot" and related commands gained a new
2825 "--message=" option which may be used to set a free-text
2826 wall message when shutting down or rebooting the
2827 system. This message is also logged, which is useful for
2828 figuring out the reason for a reboot or shutdown a
2831 * The "systemd-resolve-host" tool's -i switch now takes
2832 network interface numbers as alternative to interface names.
2834 * A new unit file setting for services has been introduced:
2835 UtmpMode= allows configuration of how precisely systemd
2836 handles utmp and wtmp entries for the service if this is
2837 enabled. This allows writing services that appear similar to
2838 user sessions in the output of the "w", "who", "last" and
2841 * systemd-resolved will now locally synthesize DNS resource
2842 records for the "localhost" and "gateway" domains as well as
2843 the local hostname. This should ensure that clients querying
2844 RRs via resolved will get similar results as those going via
2845 NSS, if nss-myhostname is enabled.
2847 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2848 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2849 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2850 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2851 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2852 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2853 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2854 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2855 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2856 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2857 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2858 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2860 — Berlin, 2015-08-27
2864 * The systemd-efi-boot-generator functionality was merged into
2865 systemd-gpt-auto-generator.
2867 * systemd-networkd now supports Group Policy for vxlan
2868 devices. It can be enabled via the new boolean configuration
2869 option called 'GroupPolicyExtension='.
2871 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2872 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2873 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2875 — Berlin, 2015-07-31
2879 * The python-systemd code has been removed from the systemd repository.
2880 A new repository has been created which accommodates the code from
2881 now on, and we kindly ask distributions to create a separate package
2882 for this: https://github.com/systemd/python-systemd
2884 * The systemd daemon will now reload its main configuration
2885 (/etc/systemd/system.conf) on daemon-reload.
2887 * sd-dhcp now exposes vendor specific extensions via
2888 sd_dhcp_lease_get_vendor_specific().
2890 * systemd-networkd gained a number of new configuration options.
2892 - A new boolean configuration option for TAP devices called
2893 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2894 device, thus allowing to send and receive GSO packets.
2896 - A new tunnel configuration option called 'CopyDSCP='.
2897 If enabled, the DSCP field of ip6 tunnels is copied into the
2898 decapsulated packet.
2900 - A set of boolean bridge configuration options were added.
2901 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2902 and 'UnicastFlood=' are now parsed by networkd and applied to the
2903 respective bridge link device via the respective IFLA_BRPORT_*
2906 - A new string configuration option to override the hostname sent
2907 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2908 is true, networkd will use the configured hostname instead of the
2909 system hostname when sending DHCP requests.
2911 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2912 networkd will configure the IPv6 flow-label of the tunnel device
2913 according to RFC2460.
2915 - The 'macvtap' virtual network devices are now supported, similar to
2916 the already supported 'macvlan' devices.
2918 * systemd-resolved now implements RFC5452 to improve resilience against
2919 cache poisoning. Additionally, source port randomization is enabled
2920 by default to further protect against DNS spoofing attacks.
2922 * nss-mymachines now supports translating UIDs and GIDs of running
2923 containers with user-namespaces enabled. If a container 'foo'
2924 translates a host uid 'UID' to the container uid 'TUID', then
2925 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2926 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2927 mapped as 'vg-foo-TGID'.
2929 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2930 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2931 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2932 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2933 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2934 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2935 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2936 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2937 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2938 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2940 — Berlin, 2015-07-29
2944 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2945 There are no known issues with current sysfs, and udev does not need
2946 or should be used to work around such bugs.
2948 * udev does no longer enable USB HID power management. Several reports
2949 indicate, that some devices cannot handle that setting.
2951 * The udev accelerometer helper was removed. The functionality
2952 is now fully included in iio-sensor-proxy. But this means,
2953 older iio-sensor-proxy versions will no longer provide
2954 accelerometer/orientation data with this systemd version.
2955 Please upgrade iio-sensor-proxy to version 1.0.
2957 * networkd gained a new configuration option IPv6PrivacyExtensions=
2958 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2959 for Stateless Address") on selected networks.
2961 * For the sake of fewer build-time dependencies and less code in the
2962 main repository, the python bindings are about to be removed in the
2963 next release. A new repository has been created which accommodates
2964 the code from now on, and we kindly ask distributions to create a
2965 separate package for this. The removal will take place in v223.
2967 https://github.com/systemd/python-systemd
2969 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2970 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2971 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2972 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2973 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2974 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2975 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2976 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2977 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2978 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2980 — Berlin, 2015-07-07
2984 * The sd-bus.h and sd-event.h APIs have now been declared
2985 stable and have been added to the official interface of
2986 libsystemd.so. sd-bus implements an alternative D-Bus client
2987 library, that is relatively easy to use, very efficient and
2988 supports both classic D-Bus as well as kdbus as transport
2989 backend. sd-event is a generic event loop abstraction that
2990 is built around Linux epoll, but adds features such as event
2991 prioritization or efficient timer handling. Both APIs are good
2992 choices for C programs looking for a bus and/or event loop
2993 implementation that is minimal and does not have to be
2994 portable to other kernels.
2996 * kdbus support is no longer compile-time optional. It is now
2997 always built-in. However, it can still be disabled at
2998 runtime using the kdbus=0 kernel command line setting, and
2999 that setting may be changed to default to off, by specifying
3000 --disable-kdbus at build-time. Note though that the kernel
3001 command line setting has no effect if the kdbus.ko kernel
3002 module is not installed, in which case kdbus is (obviously)
3003 also disabled. We encourage all downstream distributions to
3004 begin testing kdbus by adding it to the kernel images in the
3005 development distributions, and leaving kdbus support in
3008 * The minimal required util-linux version has been bumped to
3011 * Support for chkconfig (--enable-chkconfig) was removed in
3012 favor of calling an abstraction tool
3013 /lib/systemd/systemd-sysv-install. This needs to be
3014 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3015 in README for details.
3017 * If there's a systemd unit and a SysV init script for the
3018 same service name, and the user executes "systemctl enable"
3019 for it (or a related call), then this will now enable both
3020 (or execute the related operation on both), not just the
3023 * The libudev API documentation has been converted from gtkdoc
3026 * gudev has been removed from the systemd tree, it is now an
3029 * The systemd-cgtop tool learnt a new --raw switch to generate
3030 "raw" (machine parsable) output.
3032 * networkd's IPForwarding= .network file setting learnt the
3033 new setting "kernel", which ensures that networkd does not
3034 change the IP forwarding sysctl from the default kernel
3037 * The systemd-logind bus API now exposes a new boolean
3038 property "Docked" that reports whether logind considers the
3039 system "docked", i.e. connected to a docking station or not.
3041 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3042 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3043 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3044 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3045 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3046 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3047 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3048 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3049 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3050 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3051 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3052 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3053 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3054 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3055 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3056 Fink, Zbigniew Jędrzejewski-Szmek
3058 — Berlin, 2015-06-19
3062 * The gudev library has been extracted into a separate repository
3063 available at: https://git.gnome.org/browse/libgudev/
3064 It is now managed as part of the Gnome project. Distributions
3065 are recommended to pass --disable-gudev to systemd and use
3066 gudev from the Gnome project instead. gudev is still included
3067 in systemd, for now. It will be removed soon, though. Please
3068 also see the announcement-thread on systemd-devel:
3069 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3071 * systemd now exposes a CPUUsageNSec= property for each
3072 service unit on the bus, that contains the overall consumed
3073 CPU time of a service (the sum of what each process of the
3074 service consumed). This value is only available if
3075 CPUAccounting= is turned on for a service, and is then shown
3076 in the "systemctl status" output.
3078 * Support for configuring alternative mappings of the old SysV
3079 runlevels to systemd targets has been removed. They are now
3080 hardcoded in a way that runlevels 2, 3, 4 all map to
3081 multi-user.target and 5 to graphical.target (which
3082 previously was already the default behaviour).
3084 * The auto-mounter logic gained support for mount point
3085 expiry, using a new TimeoutIdleSec= setting in .automount
3086 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3088 * The EFI System Partition (ESP) as mounted to /boot by
3089 systemd-efi-boot-generator will now be unmounted
3090 automatically after 2 minutes of not being used. This should
3091 minimize the risk of ESP corruptions.
3093 * New /etc/fstab options x-systemd.requires= and
3094 x-systemd.requires-mounts-for= are now supported to express
3095 additional dependencies for mounts. This is useful for
3096 journalling file systems that support external journal
3097 devices or overlay file systems that require underlying file
3098 systems to be mounted.
3100 * systemd does not support direct live-upgrades (via systemctl
3101 daemon-reexec) from versions older than v44 anymore. As no
3102 distribution we are aware of shipped such old versions in a
3103 stable release this should not be problematic.
3105 * When systemd forks off a new per-connection service instance
3106 it will now set the $REMOTE_ADDR environment variable to the
3107 remote IP address, and $REMOTE_PORT environment variable to
3108 the remote IP port. This behaviour is similar to the
3109 corresponding environment variables defined by CGI.
3111 * systemd-networkd gained support for uplink failure
3112 detection. The BindCarrier= option allows binding interface
3113 configuration dynamically to the link sense of other
3114 interfaces. This is useful to achieve behaviour like in
3117 * systemd-networkd gained support for configuring the DHCP
3118 client identifier to use when requesting leases.
3120 * systemd-networkd now has a per-network UseNTP= option to
3121 configure whether NTP server information acquired via DHCP
3122 is passed on to services like systemd-timesyncd.
3124 * systemd-networkd gained support for vti6 tunnels.
3126 * Note that systemd-networkd manages the sysctl variable
3127 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3128 it is configured for since v219. The variable controls IP
3129 forwarding, and is a per-interface alternative to the global
3130 /proc/sys/net/ipv[46]/ip_forward. This setting is
3131 configurable in the IPForward= option, which defaults to
3132 "no". This means if networkd is used for an interface it is
3133 no longer sufficient to set the global sysctl option to turn
3134 on IP forwarding! Instead, the .network file option
3135 IPForward= needs to be turned on! Note that the
3136 implementation of this behaviour was broken in v219 and has
3139 * Many bonding and vxlan options are now configurable in
3142 * systemd-nspawn gained a new --property= setting to set unit
3143 properties for the container scope. This is useful for
3144 setting resource parameters (e.g. "CPUShares=500") on
3145 containers started from the command line.
3147 * systemd-nspawn gained a new --private-users= switch to make
3148 use of user namespacing available on recent Linux kernels.
3150 * systemd-nspawn may now be called as part of a shell pipeline
3151 in which case the pipes used for stdin and stdout are passed
3152 directly to the process invoked in the container, without
3153 indirection via a pseudo tty.
3155 * systemd-nspawn gained a new switch to control the UNIX
3156 signal to use when killing the init process of the container
3159 * systemd-nspawn gained a new --overlay= switch for mounting
3160 overlay file systems into the container using the new kernel
3163 * When a container image is imported via systemd-importd and
3164 the host file system is not btrfs, a loopback block device
3165 file is created in /var/lib/machines.raw with a btrfs file
3166 system inside. It is then mounted to /var/lib/machines to
3167 enable btrfs features for container management. The loopback
3168 file and btrfs file system is grown as needed when container
3169 images are imported via systemd-importd.
3171 * systemd-machined/systemd-importd gained support for btrfs
3172 quota, to enforce container disk space limits on disk. This
3173 is exposed in "machinectl set-limit".
3175 * systemd-importd now can import containers from local .tar,
3176 .raw and .qcow2 images, and export them to .tar and .raw. It
3177 can also import dkr v2 images now from the network (on top
3180 * systemd-importd gained support for verifying downloaded
3181 images with gpg2 (previously only gpg1 was supported).
3183 * systemd-machined, systemd-logind, systemd: most bus calls
3184 are now accessible to unprivileged processes via
3185 PolicyKit. Also, systemd-logind will now allow users to kill
3186 their own sessions without further privileges or
3189 * systemd-shutdownd has been removed. This service was
3190 previously responsible for implementing scheduled shutdowns
3191 as exposed in /usr/bin/shutdown's time parameter. This
3192 functionality has now been moved into systemd-logind and is
3193 accessible via a bus interface.
3195 * "systemctl reboot" gained a new switch --firmware-setup that
3196 can be used to reboot into the EFI firmware setup, if that
3197 is available. systemd-logind now exposes an API on the bus
3198 to trigger such reboots, in case graphical desktop UIs want
3199 to cover this functionality.
3201 * "systemctl enable", "systemctl disable" and "systemctl mask"
3202 now support a new "--now" switch. If specified the units
3203 that are enabled will also be started, and the ones
3204 disabled/masked also stopped.
3206 * The Gummiboot EFI boot loader tool has been merged into
3207 systemd, and renamed to "systemd-boot". The bootctl tool has been
3208 updated to support systemd-boot.
3210 * An EFI kernel stub has been added that may be used to create
3211 kernel EFI binaries that contain not only the actual kernel,
3212 but also an initrd, boot splash, command line and OS release
3213 information. This combined binary can then be signed as a
3214 single image, so that the firmware can verify it all in one
3215 step. systemd-boot has special support for EFI binaries created
3216 like this and can extract OS release information from them
3217 and show them in the boot menu. This functionality is useful
3218 to implement cryptographically verified boot schemes.
3220 * Optional support has been added to systemd-fsck to pass
3221 fsck's progress report to an AF_UNIX socket in the file
3224 * udev will no longer create device symlinks for all block
3225 devices by default. A blacklist for excluding special block
3226 devices from this logic has been turned into a whitelist
3227 that requires picking block devices explicitly that require
3230 * A new (currently still internal) API sd-device.h has been
3231 added to libsystemd. This modernized API is supposed to
3232 replace libudev eventually. In fact, already much of libudev
3233 is now just a wrapper around sd-device.h.
3235 * A new hwdb database for storing metadata about pointing
3236 stick devices has been added.
3238 * systemd-tmpfiles gained support for setting file attributes
3239 similar to the "chattr" tool with new 'h' and 'H' lines.
3241 * systemd-journald will no longer unconditionally set the
3242 btrfs NOCOW flag on new journal files. This is instead done
3243 with tmpfiles snippet using the new 'h' line type. This
3244 allows easy disabling of this logic, by masking the
3245 journal-nocow.conf tmpfiles file.
3247 * systemd-journald will now translate audit message types to
3248 human readable identifiers when writing them to the
3249 journal. This should improve readability of audit messages.
3251 * The LUKS logic gained support for the offset= and skip=
3252 options in /etc/crypttab, as previously implemented by
3255 * /usr/lib/os-release gained a new optional field VARIANT= for
3256 distributions that support multiple variants (such as a
3257 desktop edition, a server edition, ...)
3259 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3260 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3261 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3262 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3263 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3264 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3265 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3266 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3267 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3268 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3269 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3270 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3271 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3272 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3273 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3274 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3275 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3276 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3277 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3278 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3279 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3280 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3281 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3282 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3283 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3284 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3285 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3287 — Berlin, 2015-05-22
3291 * Introduce a new API "sd-hwdb.h" for querying the hardware
3292 metadata database. With this minimal interface one can query
3293 and enumerate the udev hwdb, decoupled from the old libudev
3294 library. libudev's interface for this is now only a wrapper
3295 around sd-hwdb. A new tool systemd-hwdb has been added to
3296 interface with and update the database.
3298 * When any of systemd's tools copies files (for example due to
3299 tmpfiles' C lines) a btrfs reflink will attempted first,
3300 before bytewise copying is done.
3302 * systemd-nspawn gained a new --ephemeral switch. When
3303 specified a btrfs snapshot is taken of the container's root
3304 directory, and immediately removed when the container
3305 terminates again. Thus, a container can be started whose
3306 changes never alter the container's root directory, and are
3307 lost on container termination. This switch can also be used
3308 for starting a container off the root file system of the
3309 host without affecting the host OS. This switch is only
3310 available on btrfs file systems.
3312 * systemd-nspawn gained a new --template= switch. It takes the
3313 path to a container tree to use as template for the tree
3314 specified via --directory=, should that directory be
3315 missing. This allows instantiating containers dynamically,
3316 on first run. This switch is only available on btrfs file
3319 * When a .mount unit refers to a mount point on which multiple
3320 mounts are stacked, and the .mount unit is stopped all of
3321 the stacked mount points will now be unmounted until no
3322 mount point remains.
3324 * systemd now has an explicit notion of supported and
3325 unsupported unit types. Jobs enqueued for unsupported unit
3326 types will now fail with an "unsupported" error code. More
3327 specifically .swap, .automount and .device units are not
3328 supported in containers, .busname units are not supported on
3329 non-kdbus systems. .swap and .automount are also not
3330 supported if their respective kernel compile time options
3333 * machinectl gained support for two new "copy-from" and
3334 "copy-to" commands for copying files from a running
3335 container to the host or vice versa.
3337 * machinectl gained support for a new "bind" command to bind
3338 mount host directories into local containers. This is
3339 currently only supported for nspawn containers.
3341 * networkd gained support for configuring bridge forwarding
3342 database entries (fdb) from .network files.
3344 * A new tiny daemon "systemd-importd" has been added that can
3345 download container images in tar, raw, qcow2 or dkr formats,
3346 and make them available locally in /var/lib/machines, so
3347 that they can run as nspawn containers. The daemon can GPG
3348 verify the downloads (not supported for dkr, since it has no
3349 provisions for verifying downloads). It will transparently
3350 decompress bz2, xz, gzip compressed downloads if necessary,
3351 and restore sparse files on disk. The daemon uses privilege
3352 separation to ensure the actual download logic runs with
3353 fewer privileges than the daemon itself. machinectl has
3354 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3355 make the functionality of importd available to the
3356 user. With this in place the Fedora and Ubuntu "Cloud"
3357 images can be downloaded and booted as containers unmodified
3358 (the Fedora images lack the appropriate GPG signature files
3359 currently, so they cannot be verified, but this will change
3360 soon, hopefully). Note that downloading images is currently
3361 only fully supported on btrfs.
3363 * machinectl is now able to list container images found in
3364 /var/lib/machines, along with some metadata about sizes of
3365 disk and similar. If the directory is located on btrfs and
3366 quota is enabled, this includes quota display. A new command
3367 "image-status" has been added that shows additional
3368 information about images.
3370 * machinectl is now able to clone container images
3371 efficiently, if the underlying file system (btrfs) supports
3372 it, with the new "machinectl clone" command. It also
3373 gained commands for renaming and removing images, as well as
3374 marking them read-only or read-write (supported also on
3375 legacy file systems).
3377 * networkd gained support for collecting LLDP network
3378 announcements, from hardware that supports this. This is
3379 shown in networkctl output.
3381 * systemd-run gained support for a new -t (--pty) switch for
3382 invoking a binary on a pty whose input and output is
3383 connected to the invoking terminal. This allows executing
3384 processes as system services while interactively
3385 communicating with them via the terminal. Most interestingly
3386 this is supported across container boundaries. Invoking
3387 "systemd-run -t /bin/bash" is an alternative to running a
3388 full login session, the difference being that the former
3389 will not register a session, nor go through the PAM session
3392 * tmpfiles gained support for a new "v" line type for creating
3393 btrfs subvolumes. If the underlying file system is a legacy
3394 file system, this automatically degrades to creating a
3395 normal directory. Among others /var/lib/machines is now
3396 created like this at boot, should it be missing.
3398 * The directory /var/lib/containers/ has been deprecated and
3399 been replaced by /var/lib/machines. The term "machines" has
3400 been used in the systemd context as generic term for both
3401 VMs and containers, and hence appears more appropriate for
3402 this, as the directory can also contain raw images bootable
3405 * systemd-nspawn when invoked with -M but without --directory=
3406 or --image= is now capable of searching for the container
3407 root directory, subvolume or disk image automatically, in
3408 /var/lib/machines. systemd-nspawn@.service has been updated
3409 to make use of this, thus allowing it to be used for raw
3412 * A new machines.target unit has been introduced that is
3413 supposed to group all containers/VMs invoked as services on
3414 the system. systemd-nspawn@.service has been updated to
3415 integrate with that.
3417 * machinectl gained a new "start" command, for invoking a
3418 container as a service. "machinectl start foo" is mostly
3419 equivalent to "systemctl start systemd-nspawn@foo.service",
3420 but handles escaping in a nicer way.
3422 * systemd-nspawn will now mount most of the cgroupfs tree
3423 read-only into each container, with the exception of the
3424 container's own subtree in the name=systemd hierarchy.
3426 * journald now sets the special FS_NOCOW file flag for its
3427 journal files. This should improve performance on btrfs, by
3428 avoiding heavy fragmentation when journald's write-pattern
3429 is used on COW file systems. It degrades btrfs' data
3430 integrity guarantees for the files to the same levels as for
3431 ext3/ext4 however. This should be OK though as journald does
3432 its own data integrity checks and all its objects are
3433 checksummed on disk. Also, journald should handle btrfs disk
3434 full events a lot more gracefully now, by processing SIGBUS
3435 errors, and not relying on fallocate() anymore.
3437 * When journald detects that journal files it is writing to
3438 have been deleted it will immediately start new journal
3441 * systemd now provides a way to store file descriptors
3442 per-service in PID 1. This is useful for daemons to ensure
3443 that fds they require are not lost during a daemon
3444 restart. The fds are passed to the daemon on the next
3445 invocation in the same way socket activation fds are
3446 passed. This is now used by journald to ensure that the
3447 various sockets connected to all the system's stdout/stderr
3448 are not lost when journald is restarted. File descriptors
3449 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3450 an extension to sd_notify(). Note that a limit is enforced
3451 on the number of fds a service can store in PID 1, and it
3452 defaults to 0, so that no fds may be stored, unless this is
3453 explicitly turned on.
3455 * The default TERM variable to use for units connected to a
3456 terminal, when no other value is explicitly is set is now
3457 vt220 rather than vt102. This should be fairly safe still,
3458 but allows PgUp/PgDn work.
3460 * The /etc/crypttab option header= as known from Debian is now
3463 * "loginctl user-status" and "loginctl session-status" will
3464 now show the last 10 lines of log messages of the
3465 user/session following the status output. Similar,
3466 "machinectl status" will show the last 10 log lines
3467 associated with a virtual machine or container
3468 service. (Note that this is usually not the log messages
3469 done in the VM/container itself, but simply what the
3470 container manager logs. For nspawn this includes all console
3473 * "loginctl session-status" without further argument will now
3474 show the status of the session of the caller. Similar,
3475 "lock-session", "unlock-session", "activate",
3476 "enable-linger", "disable-linger" may now be called without
3477 session/user parameter in which case they apply to the
3478 caller's session/user.
3480 * An X11 session scriptlet is now shipped that uploads
3481 $DISPLAY and $XAUTHORITY into the environment of the systemd
3482 --user daemon if a session begins. This should improve
3483 compatibility with X11 enabled applications run as systemd
3486 * Generators are now subject to masking via /etc and /run, the
3487 same way as unit files.
3489 * networkd .network files gained support for configuring
3490 per-link IPv4/IPv6 packet forwarding as well as IPv4
3491 masquerading. This is by default turned on for veth links to
3492 containers, as registered by systemd-nspawn. This means that
3493 nspawn containers run with --network-veth will now get
3494 automatic routed access to the host's networks without any
3495 further configuration or setup, as long as networkd runs on
3498 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3499 or UDP posts of a container on the host. With this in place
3500 it is possible to run containers with private veth links
3501 (--network-veth), and have their functionality exposed on
3502 the host as if their services were running directly on the
3505 * systemd-nspawn's --network-veth switch now gained a short
3506 version "-n", since with the changes above it is now truly
3507 useful out-of-the-box. The systemd-nspawn@.service has been
3508 updated to make use of it too by default.
3510 * systemd-nspawn will now maintain a per-image R/W lock, to
3511 ensure that the same image is not started more than once
3512 writable. (It's OK to run an image multiple times
3513 simultaneously in read-only mode.)
3515 * systemd-nspawn's --image= option is now capable of
3516 dissecting and booting MBR and GPT disk images that contain
3517 only a single active Linux partition. Previously it
3518 supported only GPT disk images with proper GPT type
3519 IDs. This allows running cloud images from major
3520 distributions directly with systemd-nspawn, without
3523 * In addition to collecting mouse dpi data in the udev
3524 hardware database, there's now support for collecting angle
3525 information for mouse scroll wheels. The database is
3526 supposed to guarantee similar scrolling behavior on mice
3527 that it knows about. There's also support for collecting
3528 information about Touchpad types.
3530 * udev's input_id built-in will now also collect touch screen
3531 dimension data and attach it to probed devices.
3533 * /etc/os-release gained support for a Distribution Privacy
3536 * networkd gained support for creating "ipvlan", "gretap",
3537 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3539 * systemd-tmpfiles gained support for "a" lines for setting
3542 * systemd-nspawn will now mount /tmp in the container to
3543 tmpfs, automatically.
3545 * systemd now exposes the memory.usage_in_bytes cgroup
3546 attribute and shows it for each service in the "systemctl
3547 status" output, if available.
3549 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3550 immediate reboot is triggered. This useful if shutdown is
3551 hung and is unable to complete, to expedite the
3552 operation. Note that this kind of reboot will still unmount
3553 all file systems, and hence should not result in fsck being
3556 * A .device unit for an optical block device will now be
3557 considered active only when a medium is in the drive. Also,
3558 mount units are now bound to their backing devices thus
3559 triggering automatic unmounting when devices become
3560 unavailable. With this in place systemd will now
3561 automatically unmount left-over mounts when a CD-ROM is
3562 ejected or an USB stick is yanked from the system.
3564 * networkd-wait-online now has support for waiting for
3565 specific interfaces only (with globbing), and for giving up
3566 after a configurable timeout.
3568 * networkd now exits when idle. It will be automatically
3569 restarted as soon as interfaces show up, are removed or
3570 change state. networkd will stay around as long as there is
3571 at least one DHCP state machine or similar around, that keep
3574 * networkd may now configure IPv6 link-local addressing in
3575 addition to IPv4 link-local addressing.
3577 * The IPv6 "token" for use in SLAAC may now be configured for
3578 each .network interface in networkd.
3580 * Routes configured with networkd may now be assigned a scope
3583 * networkd's [Match] sections now support globbing and lists
3584 of multiple space-separated matches per item.
3586 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3587 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3588 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3589 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3590 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3591 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3592 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3593 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3594 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3595 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3596 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3597 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3598 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3599 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3600 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3601 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3602 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3603 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3604 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3605 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3606 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3607 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3608 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3609 Hoffmann, Zbigniew Jędrzejewski-Szmek
3611 — Berlin, 2015-02-16
3615 * When querying unit file enablement status (for example via
3616 "systemctl is-enabled"), a new state "indirect" is now known
3617 which indicates that a unit might not be enabled itself, but
3618 another unit listed in its Also= setting might be.
3620 * Similar to the various existing ConditionXYZ= settings for
3621 units, there are now matching AssertXYZ= settings. While
3622 failing conditions cause a unit to be skipped, but its job
3623 to succeed, failing assertions declared like this will cause
3624 a unit start operation and its job to fail.
3626 * hostnamed now knows a new chassis type "embedded".
3628 * systemctl gained a new "edit" command. When used on a unit
3629 file, this allows extending unit files with .d/ drop-in
3630 configuration snippets or editing the full file (after
3631 copying it from /usr/lib to /etc). This will invoke the
3632 user's editor (as configured with $EDITOR), and reload the
3633 modified configuration after editing.
3635 * "systemctl status" now shows the suggested enablement state
3636 for a unit, as declared in the (usually vendor-supplied)
3637 system preset files.
3639 * nss-myhostname will now resolve the single-label host name
3640 "gateway" to the locally configured default IP routing
3641 gateways, ordered by their metrics. This assigns a stable
3642 name to the used gateways, regardless which ones are
3643 currently configured. Note that the name will only be
3644 resolved after all other name sources (if nss-myhostname is
3645 configured properly) and should hence not negatively impact
3646 systems that use the single-label host name "gateway" in
3649 * systemd-inhibit now allows filtering by mode when listing
3652 * Scope and service units gained a new "Delegate" boolean
3653 property, which, when set, allows processes running inside the
3654 unit to further partition resources. This is primarily
3655 useful for systemd user instances as well as container
3658 * journald will now pick up audit messages directly from
3659 the kernel, and log them like any other log message. The
3660 audit fields are split up and fully indexed. This means that
3661 journalctl in many ways is now a (nicer!) alternative to
3662 ausearch, the traditional audit client. Note that this
3663 implements only a minimal audit client. If you want the
3664 special audit modes like reboot-on-log-overflow, please use
3665 the traditional auditd instead, which can be used in
3666 parallel to journald.
3668 * The ConditionSecurity= unit file option now understands the
3669 special string "audit" to check whether auditing is
3672 * journalctl gained two new commands --vacuum-size= and
3673 --vacuum-time= to delete old journal files until the
3674 remaining ones take up no more than the specified size on disk,
3675 or are not older than the specified time.
3677 * A new, native PPPoE library has been added to sd-network,
3678 systemd's library of light-weight networking protocols. This
3679 library will be used in a future version of networkd to
3680 enable PPPoE communication without an external pppd daemon.
3682 * The busctl tool now understands a new "capture" verb that
3683 works similar to "monitor", but writes a packet capture
3684 trace to STDOUT that can be redirected to a file which is
3685 compatible with libcap's capture file format. This can then
3686 be loaded in Wireshark and similar tools to inspect bus
3689 * The busctl tool now understands a new "tree" verb that shows
3690 the object trees of a specific service on the bus, or of all
3693 * The busctl tool now understands a new "introspect" verb that
3694 shows all interfaces and members of objects on the bus,
3695 including their signature and values. This is particularly
3696 useful to get more information about bus objects shown by
3697 the new "busctl tree" command.
3699 * The busctl tool now understands new verbs "call",
3700 "set-property" and "get-property" for invoking bus method
3701 calls, setting and getting bus object properties in a
3704 * busctl gained a new --augment-creds= argument that controls
3705 whether the tool shall augment credential information it
3706 gets from the bus with data from /proc, in a possibly
3709 * nspawn's --link-journal= switch gained two new values
3710 "try-guest" and "try-host" that work like "guest" and
3711 "host", but do not fail if the host has no persistent
3712 journalling enabled. -j is now equivalent to
3713 --link-journal=try-guest.
3715 * macvlan network devices created by nspawn will now have
3716 stable MAC addresses.
3718 * A new SmackProcessLabel= unit setting has been added, which
3719 controls the SMACK security label processes forked off by
3720 the respective unit shall use.
3722 * If compiled with --enable-xkbcommon, systemd-localed will
3723 verify x11 keymap settings by compiling the given keymap. It
3724 will spew out warnings if the compilation fails. This
3725 requires libxkbcommon to be installed.
3727 * When a coredump is collected, a larger number of metadata
3728 fields is now collected and included in the journal records
3729 created for it. More specifically, control group membership,
3730 environment variables, memory maps, working directory,
3731 chroot directory, /proc/$PID/status, and a list of open file
3732 descriptors is now stored in the log entry.
3734 * The udev hwdb now contains DPI information for mice. For
3737 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3739 * All systemd programs that read standalone configuration
3740 files in /etc now also support a corresponding series of
3741 .conf.d configuration directories in /etc/, /run/,
3742 /usr/local/lib/, /usr/lib/, and (if configured with
3743 --enable-split-usr) /lib/. In particular, the following
3744 configuration files now have corresponding configuration
3745 directories: system.conf user.conf, logind.conf,
3746 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3747 resolved.conf, timesyncd.conf, journal-remote.conf, and
3748 journal-upload.conf. Note that distributions should use the
3749 configuration directories in /usr/lib/; the directories in
3750 /etc/ are reserved for the system administrator.
3752 * systemd-rfkill will no longer take the rfkill device name
3753 into account when storing rfkill state on disk, as the name
3754 might be dynamically assigned and not stable. Instead, the
3755 ID_PATH udev variable combined with the rfkill type (wlan,
3756 bluetooth, ...) is used.
3758 * A new service systemd-machine-id-commit.service has been
3759 added. When used on systems where /etc is read-only during
3760 boot, and /etc/machine-id is not initialized (but an empty
3761 file), this service will copy the temporary machine ID
3762 created as replacement into /etc after the system is fully
3763 booted up. This is useful for systems that are freshly
3764 installed with a non-initialized machine ID, but should get
3765 a fixed machine ID for subsequent boots.
3767 * networkd's .netdev files now provide a large set of
3768 configuration parameters for VXLAN devices. Similarly, the
3769 bridge port cost parameter is now configurable in .network
3770 files. There's also new support for configuring IP source
3771 routing. networkd .link files gained support for a new
3772 OriginalName= match that is useful to match against the
3773 original interface name the kernel assigned. .network files
3774 may include MTU= and MACAddress= fields for altering the MTU
3775 and MAC address while being connected to a specific network
3778 * The LUKS logic gained supported for configuring
3779 UUID-specific key files. There's also new support for naming
3780 LUKS device from the kernel command line, using the new
3781 luks.name= argument.
3783 * Timer units may now be transiently created via the bus API
3784 (this was previously already available for scope and service
3785 units). In addition it is now possible to create multiple
3786 transient units at the same time with a single bus call. The
3787 "systemd-run" tool has been updated to make use of this for
3788 running commands on a specified time, in at(1)-style.
3790 * tmpfiles gained support for "t" lines, for assigning
3791 extended attributes to files. Among other uses this may be
3792 used to assign SMACK labels to files.
3794 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3795 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3796 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3797 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3798 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3799 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3800 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3801 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3802 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3803 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3804 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3805 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3806 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3807 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3808 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3809 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3810 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3811 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3813 — Berlin, 2014-12-10
3817 * journalctl gained the new options -t/--identifier= to match
3818 on the syslog identifier (aka "tag"), as well as --utc to
3819 show log timestamps in the UTC timezone. journalctl now also
3820 accepts -n/--lines=all to disable line capping in a pager.
3822 * journalctl gained a new switch, --flush, that synchronously
3823 flushes logs from /run/log/journal to /var/log/journal if
3824 persistent storage is enabled. systemd-journal-flush.service
3825 now waits until the operation is complete.
3827 * Services can notify the manager before they start a reload
3828 (by sending RELOADING=1) or shutdown (by sending
3829 STOPPING=1). This allows the manager to track and show the
3830 internal state of daemons and closes a race condition when
3831 the process is still running but has closed its D-Bus
3834 * Services with Type=oneshot do not have to have any ExecStart
3837 * User units are now loaded also from
3838 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3839 /run/systemd/user directory that was already previously
3840 supported, but is under the control of the user.
3842 * Job timeouts (i.e. time-outs on the time a job that is
3843 queued stays in the run queue) can now optionally result in
3844 immediate reboot or power-off actions (JobTimeoutAction= and
3845 JobTimeoutRebootArgument=). This is useful on ".target"
3846 units, to limit the maximum time a target remains
3847 undispatched in the run queue, and to trigger an emergency
3848 operation in such a case. This is now used by default to
3849 turn off the system if boot-up (as defined by everything in
3850 basic.target) hangs and does not complete for at least
3851 15min. Also, if power-off or reboot hang for at least 30min
3852 an immediate power-off/reboot operation is triggered. This
3853 functionality is particularly useful to increase reliability
3854 on embedded devices, but also on laptops which might
3855 accidentally get powered on when carried in a backpack and
3856 whose boot stays stuck in a hard disk encryption passphrase
3859 * systemd-logind can be configured to also handle lid switch
3860 events even when the machine is docked or multiple displays
3861 are attached (HandleLidSwitchDocked= option).
3863 * A helper binary and a service have been added which can be
3864 used to resume from hibernation in the initramfs. A
3865 generator will parse the resume= option on the kernel
3866 command line to trigger resume.
3868 * A user console daemon systemd-consoled has been
3869 added. Currently, it is a preview, and will so far open a
3870 single terminal on each session of the user marked as
3871 Desktop=systemd-console.
3873 * Route metrics can be specified for DHCP routes added by
3876 * The SELinux context of socket-activated services can be set
3877 from the information provided by the networking stack
3878 (SELinuxContextFromNet= option).
3880 * Userspace firmware loading support has been removed and
3881 the minimum supported kernel version is thus bumped to 3.7.
3883 * Timeout for udev workers has been increased from 1 to 3
3884 minutes, but a warning will be printed after 1 minute to
3885 help diagnose kernel modules that take a long time to load.
3887 * Udev rules can now remove tags on devices with TAG-="foobar".
3889 * systemd's readahead implementation has been removed. In many
3890 circumstances it didn't give expected benefits even for
3891 rotational disk drives and was becoming less relevant in the
3892 age of SSDs. As none of the developers has been using
3893 rotating media anymore, and nobody stepped up to actively
3894 maintain this component of systemd it has now been removed.
3896 * Swap units can use Options= to specify discard options.
3897 Discard options specified for swaps in /etc/fstab are now
3900 * Docker containers are now detected as a separate type of
3903 * The Password Agent protocol gained support for queries where
3904 the user input is shown, useful e.g. for user names.
3905 systemd-ask-password gained a new --echo option to turn that
3908 * The default sysctl.d/ snippets will now set:
3910 net.core.default_qdisc = fq_codel
3912 This selects Fair Queuing Controlled Delay as the default
3913 queuing discipline for network interfaces. fq_codel helps
3914 fight the network bufferbloat problem. It is believed to be
3915 a good default with no tuning required for most workloads.
3916 Downstream distributions may override this choice. On 10Gbit
3917 servers that do not do forwarding, "fq" may perform better.
3918 Systems without a good clocksource should use "pfifo_fast".
3920 * If kdbus is enabled during build a new option BusPolicy= is
3921 available for service units, that allows locking all service
3922 processes into a stricter bus policy, in order to limit
3923 access to various bus services, or even hide most of them
3924 from the service's view entirely.
3926 * networkctl will now show the .network and .link file
3927 networkd has applied to a specific interface.
3929 * sd-login gained a new API call sd_session_get_desktop() to
3930 query which desktop environment has been selected for a
3933 * UNIX utmp support is now compile-time optional to support
3934 legacy-free systems.
3936 * systemctl gained two new commands "add-wants" and
3937 "add-requires" for pulling in units from specific targets
3940 * If the word "rescue" is specified on the kernel command line
3941 the system will now boot into rescue mode (aka
3942 rescue.target), which was previously available only by
3943 specifying "1" or "systemd.unit=rescue.target" on the kernel
3944 command line. This new kernel command line option nicely
3945 mirrors the already existing "emergency" kernel command line
3948 * New kernel command line options mount.usr=, mount.usrflags=,
3949 mount.usrfstype= have been added that match root=, rootflags=,
3950 rootfstype= but allow mounting a specific file system to
3953 * The $NOTIFY_SOCKET is now also passed to control processes of
3954 services, not only the main process.
3956 * This version reenables support for fsck's -l switch. This
3957 means at least version v2.25 of util-linux is required for
3958 operation, otherwise dead-locks on device nodes may
3959 occur. Again: you need to update util-linux to at least
3960 v2.25 when updating systemd to v217.
3962 * The "multi-seat-x" tool has been removed from systemd, as
3963 its functionality has been integrated into X servers 1.16,
3964 and the tool is hence redundant. It is recommended to update
3965 display managers invoking this tool to simply invoke X
3966 directly from now on, again.
3968 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3969 message flag has been added for all of systemd's PolicyKit
3970 authenticated method calls has been added. In particular
3971 this now allows optional interactive authorization via
3972 PolicyKit for many of PID1's privileged operations such as
3973 unit file enabling and disabling.
3975 * "udevadm hwdb --update" learnt a new switch "--usr" for
3976 placing the rebuilt hardware database in /usr instead of
3977 /etc. When used only hardware database entries stored in
3978 /usr will be used, and any user database entries in /etc are
3979 ignored. This functionality is useful for vendors to ship a
3980 pre-built database on systems where local configuration is
3981 unnecessary or unlikely.
3983 * Calendar time specifications in .timer units now also
3984 understand the strings "semi-annually", "quarterly" and
3985 "minutely" as shortcuts (in addition to the preexisting
3986 "anually", "hourly", ...).
3988 * systemd-tmpfiles will now correctly create files in /dev
3989 at boot which are marked for creation only at boot. It is
3990 recommended to always create static device nodes with 'c!'
3991 and 'b!', so that they are created only at boot and not
3992 overwritten at runtime.
3994 * When the watchdog logic is used for a service (WatchdogSec=)
3995 and the watchdog timeout is hit the service will now be
3996 terminated with SIGABRT (instead of just SIGTERM), in order
3997 to make sure a proper coredump and backtrace is
3998 generated. This ensures that hanging services will result in
3999 similar coredump/backtrace behaviour as services that hit a
4002 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4003 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4004 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4005 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4006 Herrmann, David Sommerseth, David Strauss, Emil Renner
4007 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4008 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4009 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4010 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4011 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4012 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4013 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4014 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4015 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4016 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4017 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4018 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4019 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4020 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4021 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4022 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4025 — Berlin, 2014-10-28
4029 * timedated no longer reads NTP implementation unit names from
4030 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4031 implementations should add a
4033 Conflicts=systemd-timesyncd.service
4035 to their unit files to take over and replace systemd's NTP
4036 default functionality.
4038 * systemd-sysusers gained a new line type "r" for configuring
4039 which UID/GID ranges to allocate system users/groups
4040 from. Lines of type "u" may now add an additional column
4041 that specifies the home directory for the system user to be
4042 created. Also, systemd-sysusers may now optionally read user
4043 information from STDIN instead of a file. This is useful for
4044 invoking it from RPM preinst scriptlets that need to create
4045 users before the first RPM file is installed since these
4046 files might need to be owned by them. A new
4047 %sysusers_create_inline RPM macro has been introduced to do
4048 just that. systemd-sysusers now updates the shadow files as
4049 well as the user/group databases, which should enhance
4050 compatibility with certain tools like grpck.
4052 * A number of bus APIs of PID 1 now optionally consult
4053 PolicyKit to permit access for otherwise unprivileged
4054 clients under certain conditions. Note that this currently
4055 doesn't support interactive authentication yet, but this is
4056 expected to be added eventually, too.
4058 * /etc/machine-info now has new fields for configuring the
4059 deployment environment of the machine, as well as the
4060 location of the machine. hostnamectl has been updated with
4061 new command to update these fields.
4063 * systemd-timesyncd has been updated to automatically acquire
4064 NTP server information from systemd-networkd, which might
4065 have been discovered via DHCP.
4067 * systemd-resolved now includes a caching DNS stub resolver
4068 and a complete LLMNR name resolution implementation. A new
4069 NSS module "nss-resolve" has been added which can be used
4070 instead of glibc's own "nss-dns" to resolve hostnames via
4071 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4072 be resolved via systemd-resolved D-Bus APIs. In contrast to
4073 the glibc internal resolver systemd-resolved is aware of
4074 multi-homed system, and keeps DNS server and caches separate
4075 and per-interface. Queries are sent simultaneously on all
4076 interfaces that have DNS servers configured, in order to
4077 properly handle VPNs and local LANs which might resolve
4078 separate sets of domain names. systemd-resolved may acquire
4079 DNS server information from systemd-networkd automatically,
4080 which in turn might have discovered them via DHCP. A tool
4081 "systemd-resolve-host" has been added that may be used to
4082 query the DNS logic in resolved. systemd-resolved implements
4083 IDNA and automatically uses IDNA or UTF-8 encoding depending
4084 on whether classic DNS or LLMNR is used as transport. In the
4085 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4086 implementation to systemd-resolved.
4088 * A new NSS module nss-mymachines has been added, that
4089 automatically resolves the names of all local registered
4090 containers to their respective IP addresses.
4092 * A new client tool "networkctl" for systemd-networkd has been
4093 added. It currently is entirely passive and will query
4094 networking configuration from udev, rtnetlink and networkd,
4095 and present it to the user in a very friendly
4096 way. Eventually, we hope to extend it to become a full
4097 control utility for networkd.
4099 * .socket units gained a new DeferAcceptSec= setting that
4100 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4101 TCP. Similarly, support for controlling TCP keep-alive
4102 settings has been added (KeepAliveTimeSec=,
4103 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4104 turning off Nagle's algorithm on TCP has been added
4107 * logind learned a new session type "web", for use in projects
4108 like Cockpit which register web clients as PAM sessions.
4110 * timer units with at least one OnCalendar= setting will now
4111 be started only after time-sync.target has been
4112 reached. This way they will not elapse before the system
4113 clock has been corrected by a local NTP client or
4114 similar. This is particular useful on RTC-less embedded
4115 machines, that come up with an invalid system clock.
4117 * systemd-nspawn's --network-veth= switch should now result in
4118 stable MAC addresses for both the outer and the inner side
4121 * systemd-nspawn gained a new --volatile= switch for running
4122 container instances with /etc or /var unpopulated.
4124 * The kdbus client code has been updated to use the new Linux
4125 3.17 memfd subsystem instead of the old kdbus-specific one.
4127 * systemd-networkd's DHCP client and server now support
4128 FORCERENEW. There are also new configuration options to
4129 configure the vendor client identifier and broadcast mode
4132 * systemd will no longer inform the kernel about the current
4133 timezone, as this is necessarily incorrect and racy as the
4134 kernel has no understanding of DST and similar
4135 concepts. This hence means FAT timestamps will be always
4136 considered UTC, similar to what Android is already
4137 doing. Also, when the RTC is configured to the local time
4138 (rather than UTC) systemd will never synchronize back to it,
4139 as this might confuse Windows at a later boot.
4141 * systemd-analyze gained a new command "verify" for offline
4142 validation of unit files.
4144 * systemd-networkd gained support for a couple of additional
4145 settings for bonding networking setups. Also, the metric for
4146 statically configured routes may now be configured. For
4147 network interfaces where this is appropriate the peer IP
4148 address may now be configured.
4150 * systemd-networkd's DHCP client will no longer request
4151 broadcasting by default, as this tripped up some networks.
4152 For hardware where broadcast is required the feature should
4153 be switched back on using RequestBroadcast=yes.
4155 * systemd-networkd will now set up IPv4LL addresses (when
4156 enabled) even if DHCP is configured successfully.
4158 * udev will now default to respect network device names given
4159 by the kernel when the kernel indicates that these are
4160 predictable. This behavior can be tweaked by changing
4161 NamePolicy= in the relevant .link file.
4163 * A new library systemd-terminal has been added that
4164 implements full TTY stream parsing and rendering. This
4165 library is supposed to be used later on for implementing a
4166 full userspace VT subsystem, replacing the current kernel
4169 * A new tool systemd-journal-upload has been added to push
4170 journal data to a remote system running
4171 systemd-journal-remote.
4173 * journald will no longer forward all local data to another
4174 running syslog daemon. This change has been made because
4175 rsyslog (which appears to be the most commonly used syslog
4176 implementation these days) no longer makes use of this, and
4177 instead pulls the data out of the journal on its own. Since
4178 forwarding the messages to a non-existent syslog server is
4179 more expensive than we assumed we have now turned this
4180 off. If you run a syslog server that is not a recent rsyslog
4181 version, you have to turn this option on again
4182 (ForwardToSyslog= in journald.conf).
4184 * journald now optionally supports the LZ4 compressor for
4185 larger journal fields. This compressor should perform much
4186 better than XZ which was the previous default.
4188 * machinectl now shows the IP addresses of local containers,
4189 if it knows them, plus the interface name of the container.
4191 * A new tool "systemd-escape" has been added that makes it
4192 easy to escape strings to build unit names and similar.
4194 * sd_notify() messages may now include a new ERRNO= field
4195 which is parsed and collected by systemd and shown among the
4196 "systemctl status" output for a service.
4198 * A new component "systemd-firstboot" has been added that
4199 queries the most basic systemd information (timezone,
4200 hostname, root password) interactively on first
4201 boot. Alternatively it may also be used to provision these
4202 things offline on OS images installed into directories.
4204 * The default sysctl.d/ snippets will now set
4206 net.ipv4.conf.default.promote_secondaries=1
4208 This has the benefit of no flushing secondary IP addresses
4209 when primary addresses are removed.
4211 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4212 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4213 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4214 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4215 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4216 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4217 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4218 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4219 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4220 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4221 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4222 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4223 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4224 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4225 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4227 — Berlin, 2014-08-19
4231 * A new tool systemd-sysusers has been added. This tool
4232 creates system users and groups in /etc/passwd and
4233 /etc/group, based on static declarative system user/group
4234 definitions in /usr/lib/sysusers.d/. This is useful to
4235 enable factory resets and volatile systems that boot up with
4236 an empty /etc directory, and thus need system users and
4237 groups created during early boot. systemd now also ships
4238 with two default sysusers.d/ files for the most basic
4239 users and groups systemd and the core operating system
4242 * A new tmpfiles snippet has been added that rebuilds the
4243 essential files in /etc on boot, should they be missing.
4245 * A directive for ensuring automatic clean-up of
4246 /var/cache/man/ has been removed from the default
4247 configuration. This line should now be shipped by the man
4248 implementation. The necessary change has been made to the
4249 man-db implementation. Note that you need to update your man
4250 implementation to one that ships this line, otherwise no
4251 automatic clean-up of /var/cache/man will take place.
4253 * A new condition ConditionNeedsUpdate= has been added that
4254 may conditionalize services to only run when /etc or /var
4255 are "older" than the vendor operating system resources in
4256 /usr. This is useful for reconstructing or updating /etc
4257 after an offline update of /usr or a factory reset, on the
4258 next reboot. Services that want to run once after such an
4259 update or reset should use this condition and order
4260 themselves before the new systemd-update-done.service, which
4261 will mark the two directories as fully updated. A number of
4262 service files have been added making use of this, to rebuild
4263 the udev hardware database, the journald message catalog and
4264 dynamic loader cache (ldconfig). The systemd-sysusers tool
4265 described above also makes use of this now. With this in
4266 place it is now possible to start up a minimal operating
4267 system with /etc empty cleanly. For more information on the
4268 concepts involved see this recent blog story:
4270 http://0pointer.de/blog/projects/stateless.html
4272 * A new system group "input" has been introduced, and all
4273 input device nodes get this group assigned. This is useful
4274 for system-level software to get access to input devices. It
4275 complements what is already done for "audio" and "video".
4277 * systemd-networkd learnt minimal DHCPv4 server support in
4278 addition to the existing DHCPv4 client support. It also
4279 learnt DHCPv6 client and IPv6 Router Solicitation client
4280 support. The DHCPv4 client gained support for static routes
4281 passed in from the server. Note that the [DHCPv4] section
4282 known in older systemd-networkd versions has been renamed to
4283 [DHCP] and is now also used by the DHCPv6 client. Existing
4284 .network files using settings of this section should be
4285 updated, though compatibility is maintained. Optionally, the
4286 client hostname may now be sent to the DHCP server.
4288 * networkd gained support for vxlan virtual networks as well
4289 as tun/tap and dummy devices.
4291 * networkd gained support for automatic allocation of address
4292 ranges for interfaces from a system-wide pool of
4293 addresses. This is useful for dynamically managing a large
4294 number of interfaces with a single network configuration
4295 file. In particular this is useful to easily assign
4296 appropriate IP addresses to the veth links of a large number
4297 of nspawn instances.
4299 * RPM macros for processing sysusers, sysctl and binfmt
4300 drop-in snippets at package installation time have been
4303 * The /etc/os-release file should now be placed in
4304 /usr/lib/os-release. The old location is automatically
4305 created as symlink. /usr/lib is the more appropriate
4306 location of this file, since it shall actually describe the
4307 vendor operating system shipped in /usr, and not the
4308 configuration stored in /etc.
4310 * .mount units gained a new boolean SloppyOptions= setting
4311 that maps to mount(8)'s -s option which enables permissive
4312 parsing of unknown mount options.
4314 * tmpfiles learnt a new "L+" directive which creates a symlink
4315 but (unlike "L") deletes a pre-existing file first, should
4316 it already exist and not already be the correct
4317 symlink. Similarly, "b+", "c+" and "p+" directives have been
4318 added as well, which create block and character devices, as
4319 well as fifos in the filesystem, possibly removing any
4320 pre-existing files of different types.
4322 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4323 'argument' field (which so far specified the source to
4324 symlink/copy the files from) is now optional. If omitted the
4325 same file os copied from /usr/share/factory/ suffixed by the
4326 full destination path. This is useful for populating /etc
4327 with essential files, by copying them from vendor defaults
4328 shipped in /usr/share/factory/etc.
4330 * A new command "systemctl preset-all" has been added that
4331 applies the service preset settings to all installed unit
4332 files. A new switch --preset-mode= has been added that
4333 controls whether only enable or only disable operations
4336 * A new command "systemctl is-system-running" has been added
4337 that allows checking the overall state of the system, for
4338 example whether it is fully up and running.
4340 * When the system boots up with an empty /etc, the equivalent
4341 to "systemctl preset-all" is executed during early boot, to
4342 make sure all default services are enabled after a factory
4345 * systemd now contains a minimal preset file that enables the
4346 most basic services systemd ships by default.
4348 * Unit files' [Install] section gained a new DefaultInstance=
4349 field for defining the default instance to create if a
4350 template unit is enabled with no instance specified.
4352 * A new passive target cryptsetup-pre.target has been added
4353 that may be used by services that need to make they run and
4354 finish before the first LUKS cryptographic device is set up.
4356 * The /dev/loop-control and /dev/btrfs-control device nodes
4357 are now owned by the "disk" group by default, opening up
4358 access to this group.
4360 * systemd-coredump will now automatically generate a
4361 stack trace of all core dumps taking place on the system,
4362 based on elfutils' libdw library. This stack trace is logged
4365 * systemd-coredump may now optionally store coredumps directly
4366 on disk (in /var/lib/systemd/coredump, possibly compressed),
4367 instead of storing them unconditionally in the journal. This
4368 mode is the new default. A new configuration file
4369 /etc/systemd/coredump.conf has been added to configure this
4370 and other parameters of systemd-coredump.
4372 * coredumpctl gained a new "info" verb to show details about a
4373 specific coredump. A new switch "-1" has also been added
4374 that makes sure to only show information about the most
4375 recent entry instead of all entries. Also, as the tool is
4376 generally useful now the "systemd-" prefix of the binary
4377 name has been removed. Distributions that want to maintain
4378 compatibility with the old name should add a symlink from
4379 the old name to the new name.
4381 * journald's SplitMode= now defaults to "uid". This makes sure
4382 that unprivileged users can access their own coredumps with
4383 coredumpctl without restrictions.
4385 * New kernel command line options "systemd.wants=" (for
4386 pulling an additional unit during boot), "systemd.mask="
4387 (for masking a specific unit for the boot), and
4388 "systemd.debug-shell" (for enabling the debug shell on tty9)
4389 have been added. This is implemented in the new generator
4390 "systemd-debug-generator".
4392 * systemd-nspawn will now by default filter a couple of
4393 syscalls for containers, among them those required for
4394 kernel module loading, direct x86 IO port access, swap
4395 management, and kexec. Most importantly though
4396 open_by_handle_at() is now prohibited for containers,
4397 closing a hole similar to a recently discussed vulnerability
4398 in docker regarding access to files on file hierarchies the
4399 container should normally not have access to. Note that, for
4400 nspawn, we generally make no security claims anyway (and
4401 this is explicitly documented in the man page), so this is
4402 just a fix for one of the most obvious problems.
4404 * A new man page file-hierarchy(7) has been added that
4405 contains a minimized, modernized version of the file system
4406 layout systemd expects, similar in style to the FHS
4407 specification or hier(5). A new tool systemd-path(1) has
4408 been added to query many of these paths for the local
4411 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4412 longer done. Since the directory now has a per-user size
4413 limit, and is cleaned on logout this appears unnecessary,
4414 in particular since this now brings the lifecycle of this
4415 directory closer in line with how IPC objects are handled.
4417 * systemd.pc now exports a number of additional directories,
4418 including $libdir (which is useful to identify the library
4419 path for the primary architecture of the system), and a
4420 couple of drop-in directories.
4422 * udev's predictable network interface names now use the dev_port
4423 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4424 distinguish between ports of the same PCI function. dev_id should
4425 only be used for ports using the same HW address, hence the need
4428 * machined has been updated to export the OS version of a
4429 container (read from /etc/os-release and
4430 /usr/lib/os-release) on the bus. This is now shown in
4431 "machinectl status" for a machine.
4433 * A new service setting RestartForceExitStatus= has been
4434 added. If configured to a set of exit signals or process
4435 return values, the service will be restarted when the main
4436 daemon process exits with any of them, regardless of the
4439 * systemctl's -H switch for connecting to remote systemd
4440 machines has been extended so that it may be used to
4441 directly connect to a specific container on the
4442 host. "systemctl -H root@foobar:waldi" will now connect as
4443 user "root" to host "foobar", and then proceed directly to
4444 the container named "waldi". Note that currently you have to
4445 authenticate as user "root" for this to work, as entering
4446 containers is a privileged operation.
4448 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4449 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4450 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4451 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4452 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4453 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4454 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4455 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4456 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4457 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4458 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4459 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4461 — Berlin, 2014-07-03
4465 * As an experimental feature, udev now tries to lock the
4466 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4467 executes events for the disk or any of its partitions.
4468 Applications like partitioning programs can lock the
4469 disk device node (flock(LOCK_EX)) and claim temporary
4470 device ownership that way; udev will entirely skip all event
4471 handling for this disk and its partitions. If the disk
4472 was opened for writing, the close will trigger a partition
4473 table rescan in udev's "watch" facility, and if needed
4474 synthesize "change" events for the disk and all its partitions.
4475 This is now unconditionally enabled, and if it turns out to
4476 cause major problems, we might turn it on only for specific
4477 devices, or might need to disable it entirely. Device Mapper
4478 devices are excluded from this logic.
4480 * We temporarily dropped the "-l" switch for fsck invocations,
4481 since they collide with the flock() logic above. util-linux
4482 upstream has been changed already to avoid this conflict,
4483 and we will readd "-l" as soon as util-linux with this
4484 change has been released.
4486 * The dependency on libattr has been removed. Since a long
4487 time, the extended attribute calls have moved to glibc, and
4488 libattr is thus unnecessary.
4490 * Virtualization detection works without privileges now. This
4491 means the systemd-detect-virt binary no longer requires
4492 CAP_SYS_PTRACE file capabilities, and our daemons can run
4493 with fewer privileges.
4495 * systemd-networkd now runs under its own "systemd-network"
4496 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4497 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4498 loses the ability to write to files owned by root this way.
4500 * Similarly, systemd-resolved now runs under its own
4501 "systemd-resolve" user with no capabilities remaining.
4503 * Similarly, systemd-bus-proxyd now runs under its own
4504 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4506 * systemd-networkd gained support for setting up "veth"
4507 virtual Ethernet devices for container connectivity, as well
4508 as GRE and VTI tunnels.
4510 * systemd-networkd will no longer automatically attempt to
4511 manually load kernel modules necessary for certain tunnel
4512 transports. Instead, it is assumed the kernel loads them
4513 automatically when required. This only works correctly on
4514 very new kernels. On older kernels, please consider adding
4515 the kernel modules to /etc/modules-load.d/ as a work-around.
4517 * The resolv.conf file systemd-resolved generates has been
4518 moved to /run/systemd/resolve/. If you have a symlink from
4519 /etc/resolv.conf, it might be necessary to correct it.
4521 * Two new service settings, ProtectHome= and ProtectSystem=,
4522 have been added. When enabled, they will make the user data
4523 (such as /home) inaccessible or read-only and the system
4524 (such as /usr) read-only, for specific services. This allows
4525 very light-weight per-service sandboxing to avoid
4526 modifications of user data or system files from
4527 services. These two new switches have been enabled for all
4528 of systemd's long-running services, where appropriate.
4530 * Socket units gained new SocketUser= and SocketGroup=
4531 settings to set the owner user and group of AF_UNIX sockets
4532 and FIFOs in the file system.
4534 * Socket units gained a new RemoveOnStop= setting. If enabled,
4535 all FIFOS and sockets in the file system will be removed
4536 when the specific socket unit is stopped.
4538 * Socket units gained a new Symlinks= setting. It takes a list
4539 of symlinks to create to file system sockets or FIFOs
4540 created by the specific Unix sockets. This is useful to
4541 manage symlinks to socket nodes with the same life-cycle as
4544 * The /dev/log socket and /dev/initctl FIFO have been moved to
4545 /run, and have been replaced by symlinks. This allows
4546 connecting to these facilities even if PrivateDevices=yes is
4547 used for a service (which makes /dev/log itself unavailable,
4548 but /run is left). This also has the benefit of ensuring
4549 that /dev only contains device nodes, directories and
4550 symlinks, and nothing else.
4552 * sd-daemon gained two new calls sd_pid_notify() and
4553 sd_pid_notifyf(). They are similar to sd_notify() and
4554 sd_notifyf(), but allow overriding of the source PID of
4555 notification messages if permissions permit this. This is
4556 useful to send notify messages on behalf of a different
4557 process (for example, the parent process). The
4558 systemd-notify tool has been updated to make use of this
4559 when sending messages (so that notification messages now
4560 originate from the shell script invoking systemd-notify and
4561 not the systemd-notify process itself. This should minimize
4562 a race where systemd fails to associate notification
4563 messages to services when the originating process already
4566 * A new "on-abnormal" setting for Restart= has been added. If
4567 set, it will result in automatic restarts on all "abnormal"
4568 reasons for a process to exit, which includes unclean
4569 signals, core dumps, timeouts and watchdog timeouts, but
4570 does not include clean and unclean exit codes or clean
4571 signals. Restart=on-abnormal is an alternative for
4572 Restart=on-failure for services that shall be able to
4573 terminate and avoid restarts on certain errors, by
4574 indicating so with an unclean exit code. Restart=on-failure
4575 or Restart=on-abnormal is now the recommended setting for
4576 all long-running services.
4578 * If the InaccessibleDirectories= service setting points to a
4579 mount point (or if there are any submounts contained within
4580 it), it is now attempted to completely unmount it, to make
4581 the file systems truly unavailable for the respective
4584 * The ReadOnlyDirectories= service setting and
4585 systemd-nspawn's --read-only parameter are now recursively
4586 applied to all submounts, too.
4588 * Mount units may now be created transiently via the bus APIs.
4590 * The support for SysV and LSB init scripts has been removed
4591 from the systemd daemon itself. Instead, it is now
4592 implemented as a generator that creates native systemd units
4593 from these scripts when needed. This enables us to remove a
4594 substantial amount of legacy code from PID 1, following the
4595 fact that many distributions only ship a very small number
4596 of LSB/SysV init scripts nowadays.
4598 * Privileged Xen (dom0) domains are not considered
4599 virtualization anymore by the virtualization detection
4600 logic. After all, they generally have unrestricted access to
4601 the hardware and usually are used to manage the unprivileged
4604 * systemd-tmpfiles gained a new "C" line type, for copying
4605 files or entire directories.
4607 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4608 lines. So far, they have been non-globbing versions of the
4609 latter, and have thus been redundant. In future, it is
4610 recommended to only use "z". "m" has hence been removed
4611 from the documentation, even though it stays supported.
4613 * A tmpfiles snippet to recreate the most basic structure in
4614 /var has been added. This is enough to create the /var/run →
4615 /run symlink and create a couple of structural
4616 directories. This allows systems to boot up with an empty or
4617 volatile /var. Of course, while with this change, the core OS
4618 now is capable with dealing with a volatile /var, not all
4619 user services are ready for it. However, we hope that sooner
4620 or later, many service daemons will be changed upstream so
4621 that they are able to automatically create their necessary
4622 directories in /var at boot, should they be missing. This is
4623 the first step to allow state-less systems that only require
4624 the vendor image for /usr to boot.
4626 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4627 empty tmpfs instance to a specific directory. This is
4628 particularly useful for making use of the automatic
4629 reconstruction of /var (see above), by passing --tmpfs=/var.
4631 * Access modes specified in tmpfiles snippets may now be
4632 prefixed with "~", which indicates that they shall be masked
4633 by whether the existing file or directory is currently
4634 writable, readable or executable at all. Also, if specified,
4635 the sgid/suid/sticky bits will be masked for all
4638 * A new passive target unit "network-pre.target" has been
4639 added which is useful for services that shall run before any
4640 network is configured, for example firewall scripts.
4642 * The "floppy" group that previously owned the /dev/fd*
4643 devices is no longer used. The "disk" group is now used
4644 instead. Distributions should probably deprecate usage of
4647 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4648 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4649 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4650 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4651 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4652 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4655 — Berlin, 2014-06-11
4659 * A new "systemd-timesyncd" daemon has been added for
4660 synchronizing the system clock across the network. It
4661 implements an SNTP client. In contrast to NTP
4662 implementations such as chrony or the NTP reference server,
4663 this only implements a client side, and does not bother with
4664 the full NTP complexity, focusing only on querying time from
4665 one remote server and synchronizing the local clock to
4666 it. Unless you intend to serve NTP to networked clients or
4667 want to connect to local hardware clocks, this simple NTP
4668 client should be more than appropriate for most
4669 installations. The daemon runs with minimal privileges, and
4670 has been hooked up with networkd to only operate when
4671 network connectivity is available. The daemon saves the
4672 current clock to disk every time a new NTP sync has been
4673 acquired, and uses this to possibly correct the system clock
4674 early at bootup, in order to accommodate for systems that
4675 lack an RTC such as the Raspberry Pi and embedded devices,
4676 and to make sure that time monotonically progresses on these
4677 systems, even if it is not always correct. To make use of
4678 this daemon, a new system user and group "systemd-timesync"
4679 needs to be created on installation of systemd.
4681 * The queue "seqnum" interface of libudev has been disabled, as
4682 it was generally incompatible with device namespacing as
4683 sequence numbers of devices go "missing" if the devices are
4684 part of a different namespace.
4686 * "systemctl list-timers" and "systemctl list-sockets" gained
4687 a --recursive switch for showing units of these types also
4688 for all local containers, similar in style to the already
4689 supported --recursive switch for "systemctl list-units".
4691 * A new RebootArgument= setting has been added for service
4692 units, which may be used to specify a kernel reboot argument
4693 to use when triggering reboots with StartLimitAction=.
4695 * A new FailureAction= setting has been added for service
4696 units which may be used to specify an operation to trigger
4697 when a service fails. This works similarly to
4698 StartLimitAction=, but unlike it, controls what is done
4699 immediately rather than only after several attempts to
4700 restart the service in question.
4702 * hostnamed got updated to also expose the kernel name,
4703 release, and version on the bus. This is useful for
4704 executing commands like hostnamectl with the -H switch.
4705 systemd-analyze makes use of this to properly display
4706 details when running non-locally.
4708 * The bootchart tool can now show cgroup information in the
4709 graphs it generates.
4711 * The CFS CPU quota cgroup attribute is now exposed for
4712 services. The new CPUQuota= switch has been added for this
4713 which takes a percentage value. Setting this will have the
4714 result that a service may never get more CPU time than the
4715 specified percentage, even if the machine is otherwise idle.
4717 * systemd-networkd learned IPIP and SIT tunnel support.
4719 * LSB init scripts exposing a dependency on $network will now
4720 get a dependency on network-online.target rather than simply
4721 network.target. This should bring LSB handling closer to
4722 what it was on SysV systems.
4724 * A new fsck.repair= kernel option has been added to control
4725 how fsck shall deal with unclean file systems at boot.
4727 * The (.ini) configuration file parser will now silently
4728 ignore sections whose name begins with "X-". This may be
4729 used to maintain application-specific extension sections in unit
4732 * machined gained a new API to query the IP addresses of
4733 registered containers. "machinectl status" has been updated
4734 to show these addresses in its output.
4736 * A new call sd_uid_get_display() has been added to the
4737 sd-login APIs for querying the "primary" session of a
4738 user. The "primary" session of the user is elected from the
4739 user's sessions and generally a graphical session is
4740 preferred over a text one.
4742 * A minimal systemd-resolved daemon has been added. It
4743 currently simply acts as a companion to systemd-networkd and
4744 manages resolv.conf based on per-interface DNS
4745 configuration, possibly supplied via DHCP. In the long run
4746 we hope to extend this into a local DNSSEC enabled DNS and
4749 * The systemd-networkd-wait-online tool is now enabled by
4750 default. It will delay network-online.target until a network
4751 connection has been configured. The tool primarily integrates
4752 with networkd, but will also make a best effort to make sense
4753 of network configuration performed in some other way.
4755 * Two new service options StartupCPUShares= and
4756 StartupBlockIOWeight= have been added that work similarly to
4757 CPUShares= and BlockIOWeight= however only apply during
4758 system startup. This is useful to prioritize certain services
4759 differently during bootup than during normal runtime.
4761 * hostnamed has been changed to prefer the statically
4762 configured hostname in /etc/hostname (unless set to
4763 'localhost' or empty) over any dynamic one supplied by
4764 dhcp. With this change, the rules for picking the hostname
4765 match more closely the rules of other configuration settings
4766 where the local administrator's configuration in /etc always
4767 overrides any other settings.
4769 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4770 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4771 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4772 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4773 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4774 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4775 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4776 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4777 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4778 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4779 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4780 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4781 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4782 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4783 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4784 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4787 — Beijing, 2014-05-28
4791 * When restoring the screen brightness at boot, stay away from
4792 the darkest setting or from the lowest 5% of the available
4793 range, depending on which is the larger value of both. This
4794 should effectively protect the user from rebooting into a
4795 black screen, should the brightness have been set to minimum
4798 * sd-login gained a new sd_machine_get_class() call to
4799 determine the class ("vm" or "container") of a machine
4800 registered with machined.
4802 * sd-login gained new calls
4803 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4804 to query the identity of the peer of a local AF_UNIX
4805 connection. They operate similarly to their sd_pid_get_xyz()
4808 * PID 1 will now maintain a system-wide system state engine
4809 with the states "starting", "running", "degraded",
4810 "maintenance", "stopping". These states are bound to system
4811 startup, normal runtime, runtime with at least one failed
4812 service, rescue/emergency mode and system shutdown. This
4813 state is shown in the "systemctl status" output when no unit
4814 name is passed. It is useful to determine system state, in
4815 particularly when doing so for many systems or containers at
4818 * A new command "list-machines" has been added to "systemctl"
4819 that lists all local OS containers and shows their system
4820 state (see above), if systemd runs inside of them.
4822 * systemctl gained a new "-r" switch to recursively enumerate
4823 units on all local containers, when used with the
4824 "list-unit" command (which is the default one that is
4825 executed when no parameters are specified).
4827 * The GPT automatic partition discovery logic will now honour
4828 two GPT partition flags: one may be set on a partition to
4829 cause it to be mounted read-only, and the other may be set
4830 on a partition to ignore it during automatic discovery.
4832 * Two new GPT type UUIDs have been added for automatic root
4833 partition discovery, for 32-bit and 64-bit ARM. This is not
4834 particularly useful for discovering the root directory on
4835 these architectures during bare-metal boots (since UEFI is
4836 not common there), but still very useful to allow booting of
4837 ARM disk images in nspawn with the -i option.
4839 * MAC addresses of interfaces created with nspawn's
4840 --network-interface= switch will now be generated from the
4841 machine name, and thus be stable between multiple invocations
4844 * logind will now automatically remove all IPC objects owned
4845 by a user if she or he fully logs out. This makes sure that
4846 users who are logged out cannot continue to consume IPC
4847 resources. This covers SysV memory, semaphores and message
4848 queues as well as POSIX shared memory and message
4849 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4850 limits. With this functionality, that is corrected. This may
4851 be turned off by using the RemoveIPC= switch of logind.conf.
4853 * The systemd-machine-id-setup and tmpfiles tools gained a
4854 --root= switch to operate on a specific root directory,
4857 * journald can now forward logged messages to the TTYs of all
4858 logged in users ("wall"). This is the default for all
4859 emergency messages now.
4861 * A new tool systemd-journal-remote has been added to stream
4862 journal log messages across the network.
4864 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4865 controller trees are mounted into it. Note that the
4866 directories mounted beneath it are not read-only. This is a
4867 security measure and is particularly useful because glibc
4868 actually includes a search logic to pick any tmpfs it can
4869 find to implement shm_open() if /dev/shm is not available
4870 (which it might very well be in namespaced setups).
4872 * machinectl gained a new "poweroff" command to cleanly power
4873 down a local OS container.
4875 * The PrivateDevices= unit file setting will now also drop the
4876 CAP_MKNOD capability from the capability bound set, and
4877 imply DevicePolicy=closed.
4879 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4880 comprehensively on all long-running systemd services where
4881 this is appropriate.
4883 * systemd-udevd will now run in a disassociated mount
4884 namespace. To mount directories from udev rules, make sure to
4885 pull in mount units via SYSTEMD_WANTS properties.
4887 * The kdbus support gained support for uploading policy into
4888 the kernel. sd-bus gained support for creating "monitoring"
4889 connections that can eavesdrop into all bus communication
4890 for debugging purposes.
4892 * Timestamps may now be specified in seconds since the UNIX
4893 epoch Jan 1st, 1970 by specifying "@" followed by the value
4896 * Native tcpwrap support in systemd has been removed. tcpwrap
4897 is old code, not really maintained anymore and has serious
4898 shortcomings, and better options such as firewalls
4899 exist. For setups that require tcpwrap usage, please
4900 consider invoking your socket-activated service via tcpd,
4901 like on traditional inetd.
4903 * A new system.conf configuration option
4904 DefaultTimerAccuracySec= has been added that controls the
4905 default AccuracySec= setting of .timer units.
4907 * Timer units gained a new WakeSystem= switch. If enabled,
4908 timers configured this way will cause the system to resume
4909 from system suspend (if the system supports that, which most
4912 * Timer units gained a new Persistent= switch. If enabled,
4913 timers configured this way will save to disk when they have
4914 been last triggered. This information is then used on next
4915 reboot to possible execute overdue timer events, that
4916 could not take place because the system was powered off.
4917 This enables simple anacron-like behaviour for timer units.
4919 * systemctl's "list-timers" will now also list the time a
4920 timer unit was last triggered in addition to the next time
4921 it will be triggered.
4923 * systemd-networkd will now assign predictable IPv4LL
4924 addresses to its local interfaces.
4926 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4927 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4928 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4929 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4930 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4931 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4932 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4933 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4936 — Berlin, 2014-03-25
4940 * A new unit file setting RestrictAddressFamilies= has been
4941 added to restrict which socket address families unit
4942 processes gain access to. This takes address family names
4943 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4944 attack surface of services via exotic protocol stacks. This
4945 is built on seccomp system call filters.
4947 * Two new unit file settings RuntimeDirectory= and
4948 RuntimeDirectoryMode= have been added that may be used to
4949 manage a per-daemon runtime directories below /run. This is
4950 an alternative for setting up directory permissions with
4951 tmpfiles snippets, and has the advantage that the runtime
4952 directory's lifetime is bound to the daemon runtime and that
4953 the daemon starts up with an empty directory each time. This
4954 is particularly useful when writing services that drop
4955 privileges using the User= or Group= setting.
4957 * The DeviceAllow= unit setting now supports globbing for
4958 matching against device group names.
4960 * The systemd configuration file system.conf gained new
4961 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4962 DefaultMemoryAccounting= to globally turn on/off accounting
4963 for specific resources (cgroups) for all units. These
4964 settings may still be overridden individually in each unit
4967 * systemd-gpt-auto-generator is now able to discover /srv and
4968 root partitions in addition to /home and swap partitions. It
4969 also supports LUKS-encrypted partitions now. With this in
4970 place, automatic discovery of partitions to mount following
4971 the Discoverable Partitions Specification
4972 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4973 is now a lot more complete. This allows booting without
4974 /etc/fstab and without root= on the kernel command line on
4975 systems prepared appropriately.
4977 * systemd-nspawn gained a new --image= switch which allows
4978 booting up disk images and Linux installations on any block
4979 device that follow the Discoverable Partitions Specification
4980 (see above). This means that installations made with
4981 appropriately updated installers may now be started and
4982 deployed using container managers, completely
4983 unmodified. (We hope that libvirt-lxc will add support for
4984 this feature soon, too.)
4986 * systemd-nspawn gained a new --network-macvlan= setting to
4987 set up a private macvlan interface for the
4988 container. Similarly, systemd-networkd gained a new
4989 Kind=macvlan setting in .netdev files.
4991 * systemd-networkd now supports configuring local addresses
4994 * A new tool systemd-network-wait-online has been added to
4995 synchronously wait for network connectivity using
4998 * The sd-bus.h bus API gained a new sd_bus_track object for
4999 tracking the life-cycle of bus peers. Note that sd-bus.h is
5000 still not a public API though (unless you specify
5001 --enable-kdbus on the configure command line, which however
5002 voids your warranty and you get no API stability guarantee).
5004 * The $XDG_RUNTIME_DIR runtime directories for each user are
5005 now individual tmpfs instances, which has the benefit of
5006 introducing separate pools for each user, with individual
5007 size limits, and thus making sure that unprivileged clients
5008 can no longer negatively impact the system or other users by
5009 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5010 RuntimeDirectorySize= has been introduced that allows
5011 controlling the default size limit for all users. It
5012 defaults to 10% of the available physical memory. This is no
5013 replacement for quotas on tmpfs though (which the kernel
5014 still does not support), as /dev/shm and /tmp are still
5015 shared resources used by both the system and unprivileged
5018 * logind will now automatically turn off automatic suspending
5019 on laptop lid close when more than one display is
5020 connected. This was previously expected to be implemented
5021 individually in desktop environments (such as GNOME),
5022 however has been added to logind now, in order to fix a
5023 boot-time race where a desktop environment might not have
5024 been started yet and thus not been able to take an inhibitor
5025 lock at the time where logind already suspends the system
5026 due to a closed lid.
5028 * logind will now wait at least 30s after each system
5029 suspend/resume cycle, and 3min after system boot before
5030 suspending the system due to a closed laptop lid. This
5031 should give USB docking stations and similar enough time to
5032 be probed and configured after system resume and boot in
5033 order to then act as suspend blocker.
5035 * systemd-run gained a new --property= setting which allows
5036 initialization of resource control properties (and others)
5037 for the created scope or service unit. Example: "systemd-run
5038 --property=BlockIOWeight=10 updatedb" may be used to run
5039 updatedb at a low block IO scheduling weight.
5041 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5042 now also work in --scope mode.
5044 * When systemd is compiled with kdbus support, basic support
5045 for enforced policies is now in place. (Note that enabling
5046 kdbus still voids your warranty and no API compatibility
5049 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5050 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5051 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5052 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5053 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5054 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5055 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5056 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5057 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5058 Zbigniew Jędrzejewski-Szmek
5060 — Berlin, 2014-03-12
5064 * systemd will now relabel /dev after loading the SMACK policy
5065 according to SMACK rules.
5067 * A new unit file option AppArmorProfile= has been added to
5068 set the AppArmor profile for the processes of a unit.
5070 * A new condition check ConditionArchitecture= has been added
5071 to conditionalize units based on the system architecture, as
5072 reported by uname()'s "machine" field.
5074 * systemd-networkd now supports matching on the system
5075 virtualization, architecture, kernel command line, host name
5078 * logind is now a lot more aggressive when suspending the
5079 machine due to a closed laptop lid. Instead of acting only
5080 on the lid close action, it will continuously watch the lid
5081 status and act on it. This is useful for laptops where the
5082 power button is on the outside of the chassis so that it can
5083 be reached without opening the lid (such as the Lenovo
5084 Yoga). On those machines, logind will now immediately
5085 re-suspend the machine if the power button has been
5086 accidentally pressed while the laptop was suspended and in a
5087 backpack or similar.
5089 * logind will now watch SW_DOCK switches and inhibit reaction
5090 to the lid switch if it is pressed. This means that logind
5091 will not suspend the machine anymore if the lid is closed
5092 and the system is docked, if the laptop supports SW_DOCK
5093 notifications via the input layer. Note that ACPI docking
5094 stations do not generate this currently. Also note that this
5095 logic is usually not fully sufficient and Desktop
5096 Environments should take a lid switch inhibitor lock when an
5097 external display is connected, as systemd will not watch
5100 * nspawn will now make use of the devices cgroup controller by
5101 default, and only permit creation of and access to the usual
5102 API device nodes like /dev/null or /dev/random, as well as
5103 access to (but not creation of) the pty devices.
5105 * We will now ship a default .network file for
5106 systemd-networkd that automatically configures DHCP for
5107 network interfaces created by nspawn's --network-veth or
5108 --network-bridge= switches.
5110 * systemd will now understand the usual M, K, G, T suffixes
5111 according to SI conventions (i.e. to the base 1000) when
5112 referring to throughput and hardware metrics. It will stay
5113 with IEC conventions (i.e. to the base 1024) for software
5114 metrics, according to what is customary according to
5115 Wikipedia. We explicitly document which base applies for
5116 each configuration option.
5118 * The DeviceAllow= setting in unit files now supports a syntax
5119 to whitelist an entire group of devices node majors at once,
5120 based on the /proc/devices listing. For example, with the
5121 string "char-pts", it is now possible to whitelist all
5122 current and future pseudo-TTYs at once.
5124 * sd-event learned a new "post" event source. Event sources of
5125 this type are triggered by the dispatching of any event
5126 source of a type that is not "post". This is useful for
5127 implementing clean-up and check event sources that are
5128 triggered by other work being done in the program.
5130 * systemd-networkd is no longer statically enabled, but uses
5131 the usual [Install] sections so that it can be
5132 enabled/disabled using systemctl. It still is enabled by
5135 * When creating a veth interface pair with systemd-nspawn, the
5136 host side will now be prefixed with "vb-" if
5137 --network-bridge= is used, and with "ve-" if --network-veth
5138 is used. This way, it is easy to distinguish these cases on
5139 the host, for example to apply different configuration to
5140 them with systemd-networkd.
5142 * The compatibility libraries for libsystemd-journal.so,
5143 libsystem-id128.so, libsystemd-login.so and
5144 libsystemd-daemon.so do not make use of IFUNC
5145 anymore. Instead, we now build libsystemd.so multiple times
5146 under these alternative names. This means that the footprint
5147 is drastically increased, but given that these are
5148 transitional compatibility libraries, this should not matter
5149 much. This change has been made necessary to support the ARM
5150 platform for these compatibility libraries, as the ARM
5151 toolchain is not really at the same level as the toolchain
5152 for other architectures like x86 and does not support
5153 IFUNC. Please make sure to use --enable-compat-libs only
5154 during a transitional period!
5156 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5157 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5158 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5159 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5160 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5161 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5162 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5163 Zbigniew Jędrzejewski-Szmek
5165 — Berlin, 2014-02-24
5169 * A new component "systemd-networkd" has been added that can
5170 be used to configure local network interfaces statically or
5171 via DHCP. It is capable of bringing up bridges, VLANs, and
5172 bonding. Currently, no hook-ups for interactive network
5173 configuration are provided. Use this for your initrd,
5174 container, embedded, or server setup if you need a simple,
5175 yet powerful, network configuration solution. This
5176 configuration subsystem is quite nifty, as it allows wildcard
5177 hotplug matching in interfaces. For example, with a single
5178 configuration snippet, you can configure that all Ethernet
5179 interfaces showing up are automatically added to a bridge,
5180 or similar. It supports link-sensing and more.
5182 * A new tool "systemd-socket-proxyd" has been added which can
5183 act as a bidirectional proxy for TCP sockets. This is
5184 useful for adding socket activation support to services that
5185 do not actually support socket activation, including virtual
5186 machines and the like.
5188 * Add a new tool to save/restore rfkill state on
5191 * Save/restore state of keyboard backlights in addition to
5192 display backlights on shutdown/boot.
5194 * udev learned a new SECLABEL{} construct to label device
5195 nodes with a specific security label when they appear. For
5196 now, only SECLABEL{selinux} is supported, but the syntax is
5197 prepared for additional security frameworks.
5199 * udev gained a new scheme to configure link-level attributes
5200 from files in /etc/systemd/network/*.link. These files can
5201 match against MAC address, device path, driver name and type,
5202 and will apply attributes like the naming policy, link speed,
5203 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5204 address assignment policy (randomized, ...).
5206 * The configuration of network interface naming rules for
5207 "permanent interface names" has changed: a new NamePolicy=
5208 setting in the [Link] section of .link files determines the
5209 priority of possible naming schemes (onboard, slot, MAC,
5210 path). The default value of this setting is determined by
5211 /usr/lib/net/links/99-default.link. Old
5212 80-net-name-slot.rules udev configuration file has been
5213 removed, so local configuration overriding this file should
5214 be adapted to override 99-default.link instead.
5216 * When the User= switch is used in a unit file, also
5217 initialize $SHELL= based on the user database entry.
5219 * systemd no longer depends on libdbus. All communication is
5220 now done with sd-bus, systemd's low-level bus library
5223 * kdbus support has been added to PID 1 itself. When kdbus is
5224 enabled, this causes PID 1 to set up the system bus and
5225 enable support for a new ".busname" unit type that
5226 encapsulates bus name activation on kdbus. It works a little
5227 bit like ".socket" units, except for bus names. A new
5228 generator has been added that converts classic dbus1 service
5229 activation files automatically into native systemd .busname
5232 * sd-bus: add a light-weight vtable implementation that allows
5233 defining objects on the bus with a simple static const
5234 vtable array of its methods, signals and properties.
5236 * systemd will not generate or install static dbus
5237 introspection data anymore to /usr/share/dbus-1/interfaces,
5238 as the precise format of these files is unclear, and
5239 nothing makes use of it.
5241 * A proxy daemon is now provided to proxy clients connecting
5242 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5243 compatibility with classic D-Bus.
5245 * A bus driver implementation has been added that supports the
5246 classic D-Bus bus driver calls on kdbus, also for
5247 compatibility purposes.
5249 * A new API "sd-event.h" has been added that implements a
5250 minimal event loop API built around epoll. It provides a
5251 couple of features that direct epoll usage is lacking:
5252 prioritization of events, scales to large numbers of timer
5253 events, per-event timer slack (accuracy), system-wide
5254 coalescing of timer events, exit handlers, watchdog
5255 supervision support using systemd's sd_notify() API, child
5258 * A new API "sd-rntl.h" has been added that provides an API
5259 around the route netlink interface of the kernel, similar in
5260 style to "sd-bus.h".
5262 * A new API "sd-dhcp-client.h" has been added that provides a
5263 small DHCPv4 client-side implementation. This is used by
5266 * There is a new kernel command line option
5267 "systemd.restore_state=0|1". When set to "0", none of the
5268 systemd tools will restore saved runtime state to hardware
5269 devices. More specifically, the rfkill and backlight states
5272 * The FsckPassNo= compatibility option in mount/service units
5273 has been removed. The fstab generator will now add the
5274 necessary dependencies automatically, and does not require
5275 PID1's support for that anymore.
5277 * journalctl gained a new switch, --list-boots, that lists
5278 recent boots with their times and boot IDs.
5280 * The various tools like systemctl, loginctl, timedatectl,
5281 busctl, systemd-run, ... have gained a new switch "-M" to
5282 connect to a specific, local OS container (as direct
5283 connection, without requiring SSH). This works on any
5284 container that is registered with machined, such as those
5285 created by libvirt-lxc or nspawn.
5287 * systemd-run and systemd-analyze also gained support for "-H"
5288 to connect to remote hosts via SSH. This is particularly
5289 useful for systemd-run because it enables queuing of jobs
5290 onto remote systems.
5292 * machinectl gained a new command "login" to open a getty
5293 login in any local container. This works with any container
5294 that is registered with machined (such as those created by
5295 libvirt-lxc or nspawn), and which runs systemd inside.
5297 * machinectl gained a new "reboot" command that may be used to
5298 trigger a reboot on a specific container that is registered
5299 with machined. This works on any container that runs an init
5300 system of some kind.
5302 * systemctl gained a new "list-timers" command to print a nice
5303 listing of installed timer units with the times they elapse
5306 * Alternative reboot() parameters may now be specified on the
5307 "systemctl reboot" command line and are passed to the
5308 reboot() system call.
5310 * systemctl gained a new --job-mode= switch to configure the
5311 mode to queue a job with. This is a more generic version of
5312 --fail, --irreversible, and --ignore-dependencies, which are
5313 still available but not advertised anymore.
5315 * /etc/systemd/system.conf gained new settings to configure
5316 various default timeouts of units, as well as the default
5317 start limit interval and burst. These may still be overridden
5320 * PID1 will now export on the bus profile data of the security
5321 policy upload process (such as the SELinux policy upload to
5324 * journald: when forwarding logs to the console, include
5325 timestamps (following the setting in
5326 /sys/module/printk/parameters/time).
5328 * OnCalendar= in timer units now understands the special
5329 strings "yearly" and "annually". (Both are equivalent)
5331 * The accuracy of timer units is now configurable with the new
5332 AccuracySec= setting. It defaults to 1min.
5334 * A new dependency type JoinsNamespaceOf= has been added that
5335 allows running two services within the same /tmp and network
5336 namespace, if PrivateNetwork= or PrivateTmp= are used.
5338 * A new command "cat" has been added to systemctl. It outputs
5339 the original unit file of a unit, and concatenates the
5340 contents of additional "drop-in" unit file snippets, so that
5341 the full configuration is shown.
5343 * systemctl now supports globbing on the various "list-xyz"
5344 commands, like "list-units" or "list-sockets", as well as on
5345 those commands which take multiple unit names.
5347 * journalctl's --unit= switch gained support for globbing.
5349 * All systemd daemons now make use of the watchdog logic so
5350 that systemd automatically notices when they hang.
5352 * If the $container_ttys environment variable is set,
5353 getty-generator will automatically spawn a getty for each
5354 listed tty. This is useful for container managers to request
5355 login gettys to be spawned on as many ttys as needed.
5357 * %h, %s, %U specifier support is not available anymore when
5358 used in unit files for PID 1. This is because NSS calls are
5359 not safe from PID 1. They stay available for --user
5360 instances of systemd, and as special case for the root user.
5362 * loginctl gained a new "--no-legend" switch to turn off output
5365 * The "sd-login.h" API gained three new calls:
5366 sd_session_is_remote(), sd_session_get_remote_user(),
5367 sd_session_get_remote_host() to query information about
5370 * The udev hardware database now also carries vendor/product
5371 information of SDIO devices.
5373 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5374 determine whether watchdog notifications are requested by
5377 * Socket-activated per-connection services now include a
5378 short description of the connection parameters in the
5381 * tmpfiles gained a new "--boot" option. When this is not used,
5382 only lines where the command character is not suffixed with
5383 "!" are executed. When this option is specified, those
5384 options are executed too. This partitions tmpfiles
5385 directives into those that can be safely executed at any
5386 time, and those which should be run only at boot (for
5387 example, a line that creates /run/nologin).
5389 * A new API "sd-resolve.h" has been added which provides a simple
5390 asynchronous wrapper around glibc NSS host name resolution
5391 calls, such as getaddrinfo(). In contrast to glibc's
5392 getaddrinfo_a(), it does not use signals. In contrast to most
5393 other asynchronous name resolution libraries, this one does
5394 not reimplement DNS, but reuses NSS, so that alternate
5395 host name resolution systems continue to work, such as mDNS,
5396 LDAP, etc. This API is based on libasyncns, but it has been
5397 cleaned up for inclusion in systemd.
5399 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5400 "sd-daemon.h" are no longer found in individual libraries
5401 libsystemd-journal.so, libsystemd-login.so,
5402 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5403 merged them into a single library, libsystemd.so, which
5404 provides all symbols. The reason for this is cyclic
5405 dependencies, as these libraries tend to use each other's
5406 symbols. So far, we have managed to workaround that by linking
5407 a copy of a good part of our code into each of these
5408 libraries again and again, which, however, makes certain
5409 things hard to do, like sharing static variables. Also, it
5410 substantially increases footprint. With this change, there
5411 is only one library for the basic APIs systemd
5412 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5413 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5414 library as well, however are subject to the --enable-kdbus
5415 switch (see below). Note that "sd-dhcp-client.h" is not part
5416 of this library (this is because it only consumes, never
5417 provides, services of/to other APIs). To make the transition
5418 easy from the separate libraries to the unified one, we
5419 provide the --enable-compat-libs compile-time switch which
5420 will generate stub libraries that are compatible with the
5421 old ones but redirect all calls to the new one.
5423 * All of the kdbus logic and the new APIs "sd-bus.h",
5424 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5425 and "sd-utf8.h" are compile-time optional via the
5426 "--enable-kdbus" switch, and they are not compiled in by
5427 default. To make use of kdbus, you have to explicitly enable
5428 the switch. Note however, that neither the kernel nor the
5429 userspace API for all of this is considered stable yet. We
5430 want to maintain the freedom to still change the APIs for
5431 now. By specifying this build-time switch, you acknowledge
5432 that you are aware of the instability of the current
5435 * Also, note that while kdbus is pretty much complete,
5436 it lacks one thing: proper policy support. This means you
5437 can build a fully working system with all features; however,
5438 it will be highly insecure. Policy support will be added in
5439 one of the next releases, at the same time that we will
5440 declare the APIs stable.
5442 * When the kernel command line argument "kdbus" is specified,
5443 systemd will automatically load the kdbus.ko kernel module. At
5444 this stage of development, it is only useful for testing kdbus
5445 and should not be used in production. Note: if "--enable-kdbus"
5446 is specified, and the kdbus.ko kernel module is available, and
5447 "kdbus" is added to the kernel command line, the entire system
5448 runs with kdbus instead of dbus-daemon, with the above mentioned
5449 problem of missing the system policy enforcement. Also a future
5450 version of kdbus.ko or a newer systemd will not be compatible with
5451 each other, and will unlikely be able to boot the machine if only
5452 one of them is updated.
5454 * systemctl gained a new "import-environment" command which
5455 uploads the caller's environment (or parts thereof) into the
5456 service manager so that it is inherited by services started
5457 by the manager. This is useful to upload variables like
5458 $DISPLAY into the user service manager.
5460 * A new PrivateDevices= switch has been added to service units
5461 which allows running a service with a namespaced /dev
5462 directory that does not contain any device nodes for
5463 physical devices. More specifically, it only includes devices
5464 such as /dev/null, /dev/urandom, and /dev/zero which are API
5467 * logind has been extended to support behaviour like VT
5468 switching on seats that do not support a VT. This makes
5469 multi-session available on seats that are not the first seat
5470 (seat0), and on systems where kernel support for VTs has
5471 been disabled at compile-time.
5473 * If a process holds a delay lock for system sleep or shutdown
5474 and fails to release it in time, we will now log its
5475 identity. This makes it easier to identify processes that
5476 cause slow suspends or power-offs.
5478 * When parsing /etc/crypttab, support for a new key-slot=
5479 option as supported by Debian is added. It allows indicating
5480 which LUKS slot to use on disk, speeding up key loading.
5482 * The sd_journal_sendv() API call has been checked and
5483 officially declared to be async-signal-safe so that it may
5484 be invoked from signal handlers for logging purposes.
5486 * Boot-time status output is now enabled automatically after a
5487 short timeout if boot does not progress, in order to give
5488 the user an indication what she or he is waiting for.
5490 * The boot-time output has been improved to show how much time
5491 remains until jobs expire.
5493 * The KillMode= switch in service units gained a new possible
5494 value "mixed". If set, and the unit is shut down, then the
5495 initial SIGTERM signal is sent only to the main daemon
5496 process, while the following SIGKILL signal is sent to
5497 all remaining processes of the service.
5499 * When a scope unit is registered, a new property "Controller"
5500 may be set. If set to a valid bus name, systemd will send a
5501 RequestStop() signal to this name when it would like to shut
5502 down the scope. This may be used to hook manager logic into
5503 the shutdown logic of scope units. Also, scope units may now
5504 be put in a special "abandoned" state, in which case the
5505 manager process which created them takes no further
5506 responsibilities for it.
5508 * When reading unit files, systemd will now verify
5509 the access mode of these files, and warn about certain
5510 suspicious combinations. This has been added to make it
5511 easier to track down packaging bugs where unit files are
5512 marked executable or world-writable.
5514 * systemd-nspawn gained a new "--setenv=" switch to set
5515 container-wide environment variables. The similar option in
5516 systemd-activate was renamed from "--environment=" to
5517 "--setenv=" for consistency.
5519 * systemd-nspawn has been updated to create a new kdbus domain
5520 for each container that is invoked, thus allowing each
5521 container to have its own set of system and user buses,
5522 independent of the host.
5524 * systemd-nspawn gained a new --drop-capability= switch to run
5525 the container with less capabilities than the default. Both
5526 --drop-capability= and --capability= now take the special
5527 string "all" for dropping or keeping all capabilities.
5529 * systemd-nspawn gained new switches for executing containers
5530 with specific SELinux labels set.
5532 * systemd-nspawn gained a new --quiet switch to not generate
5533 any additional output but the container's own console
5536 * systemd-nspawn gained a new --share-system switch to run a
5537 container without PID namespacing enabled.
5539 * systemd-nspawn gained a new --register= switch to control
5540 whether the container is registered with systemd-machined or
5541 not. This is useful for containers that do not run full
5542 OS images, but only specific apps.
5544 * systemd-nspawn gained a new --keep-unit which may be used
5545 when invoked as the only program from a service unit, and
5546 results in registration of the unit service itself in
5547 systemd-machined, instead of a newly opened scope unit.
5549 * systemd-nspawn gained a new --network-interface= switch for
5550 moving arbitrary interfaces to the container. The new
5551 --network-veth switch creates a virtual Ethernet connection
5552 between host and container. The new --network-bridge=
5553 switch then allows assigning the host side of this virtual
5554 Ethernet connection to a bridge device.
5556 * systemd-nspawn gained a new --personality= switch for
5557 setting the kernel personality for the container. This is
5558 useful when running a 32-bit container on a 64-bit host. A
5559 similar option Personality= is now also available for service
5562 * logind will now also track a "Desktop" identifier for each
5563 session which encodes the desktop environment of it. This is
5564 useful for desktop environments that want to identify
5565 multiple running sessions of itself easily.
5567 * A new SELinuxContext= setting for service units has been
5568 added that allows setting a specific SELinux execution
5569 context for a service.
5571 * Most systemd client tools will now honour $SYSTEMD_LESS for
5572 settings of the "less" pager. By default, these tools will
5573 override $LESS to allow certain operations to work, such as
5574 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5575 influence this logic.
5577 * systemd's "seccomp" hook-up has been changed to make use of
5578 the libseccomp library instead of using its own
5579 implementation. This has benefits for portability among
5582 * For usage together with SystemCallFilter=, a new
5583 SystemCallErrorNumber= setting has been introduced that
5584 allows configuration of a system error number to be returned
5585 on filtered system calls, instead of immediately killing the
5586 process. Also, SystemCallArchitectures= has been added to
5587 limit access to system calls of a particular architecture
5588 (in order to turn off support for unused secondary
5589 architectures). There is also a global
5590 SystemCallArchitectures= setting in system.conf now to turn
5591 off support for non-native system calls system-wide.
5593 * systemd requires a kernel with a working name_to_handle_at(),
5594 please see the kernel config requirements in the README file.
5596 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5597 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5598 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5599 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5600 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5601 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5602 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5603 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5604 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5605 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5606 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5607 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5608 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5609 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5610 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5611 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5612 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5613 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5614 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5615 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5616 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5617 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5618 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5619 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5621 — Berlin, 2014-02-20
5625 * logind has gained support for facilitating privileged input
5626 and drm device access for unprivileged clients. This work is
5627 useful to allow Wayland display servers (and similar
5628 programs, such as kmscon) to run under the user's ID and
5629 access input and drm devices which are normally
5630 protected. When this is used (and the kernel is new enough)
5631 logind will "mute" IO on the file descriptors passed to
5632 Wayland as long as it is in the background and "unmute" it
5633 if it returns into the foreground. This allows secure
5634 session switching without allowing background sessions to
5635 eavesdrop on input and display data. This also introduces
5636 session switching support if VT support is turned off in the
5637 kernel, and on seats that are not seat0.
5639 * A new kernel command line option luks.options= is understood
5640 now which allows specifying LUKS options for usage for LUKS
5641 encrypted partitions specified with luks.uuid=.
5643 * tmpfiles.d(5) snippets may now use specifier expansion in
5644 path names. More specifically %m, %b, %H, %v, are now
5645 replaced by the local machine id, boot id, hostname, and
5646 kernel version number.
5648 * A new tmpfiles.d(5) command "m" has been introduced which
5649 may be used to change the owner/group/access mode of a file
5650 or directory if it exists, but do nothing if it does not.
5652 * This release removes high-level support for the
5653 MemorySoftLimit= cgroup setting. The underlying kernel
5654 cgroup attribute memory.soft_limit= is currently badly
5655 designed and likely to be removed from the kernel API in its
5656 current form, hence we should not expose it for now.
5658 * The memory.use_hierarchy cgroup attribute is now enabled for
5659 all cgroups systemd creates in the memory cgroup
5660 hierarchy. This option is likely to be come the built-in
5661 default in the kernel anyway, and the non-hierarchical mode
5662 never made much sense in the intrinsically hierarchical
5665 * A new field _SYSTEMD_SLICE= is logged along with all journal
5666 messages containing the slice a message was generated
5667 from. This is useful to allow easy per-customer filtering of
5668 logs among other things.
5670 * systemd-journald will no longer adjust the group of journal
5671 files it creates to the "systemd-journal" group. Instead we
5672 rely on the journal directory to be owned by the
5673 "systemd-journal" group, and its setgid bit set, so that the
5674 kernel file system layer will automatically enforce that
5675 journal files inherit this group assignment. The reason for
5676 this change is that we cannot allow NSS look-ups from
5677 journald which would be necessary to resolve
5678 "systemd-journal" to a numeric GID, because this might
5679 create deadlocks if NSS involves synchronous queries to
5680 other daemons (such as nscd, or sssd) which in turn are
5681 logging clients of journald and might block on it, which
5682 would then dead lock. A tmpfiles.d(5) snippet included in
5683 systemd will make sure the setgid bit and group are
5684 properly set on the journal directory if it exists on every
5685 boot. However, we recommend adjusting it manually after
5686 upgrades too (or from RPM scriptlets), so that the change is
5687 not delayed until next reboot.
5689 * Backlight and random seed files in /var/lib/ have moved into
5690 the /var/lib/systemd/ directory, in order to centralize all
5691 systemd generated files in one directory.
5693 * Boot time performance measurements (as displayed by
5694 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5695 performance information if that's available to determine how
5696 much time BIOS and boot loader initialization required. With
5697 a sufficiently new BIOS you hence no longer need to boot
5698 with Gummiboot to get access to such information.
5700 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5701 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5702 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5703 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5704 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5705 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5706 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5708 — Berlin, 2013-10-02
5712 * The Restart= option for services now understands a new
5713 on-watchdog setting, which will restart the service
5714 automatically if the service stops sending out watchdog keep
5715 alive messages (as configured with WatchdogSec=).
5717 * The getty generator (which is responsible for bringing up a
5718 getty on configured serial consoles) will no longer only
5719 start a getty on the primary kernel console but on all
5720 others, too. This makes the order in which console= is
5721 specified on the kernel command line less important.
5723 * libsystemd-logind gained a new sd_session_get_vt() call to
5724 retrieve the VT number of a session.
5726 * If the option "tries=0" is set for an entry of /etc/crypttab
5727 its passphrase is queried indefinitely instead of any
5728 maximum number of tries.
5730 * If a service with a configure PID file terminates its PID
5731 file will now be removed automatically if it still exists
5732 afterwards. This should put an end to stale PID files.
5734 * systemd-run will now also take relative binary path names
5735 for execution and no longer insists on absolute paths.
5737 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5738 paths that are optionally prefixed with "-" to indicate that
5739 it should not be considered a failure if they do not exist.
5741 * journalctl -o (and similar commands) now understands a new
5742 output mode "short-precise", it is similar to "short" but
5743 shows timestamps with usec accuracy.
5745 * The option "discard" (as known from Debian) is now
5746 synonymous to "allow-discards" in /etc/crypttab. In fact,
5747 "discard" is preferred now (since it is easier to remember
5750 * Some licensing clean-ups were made, so that more code is now
5751 LGPL-2.1 licensed than before.
5753 * A minimal tool to save/restore the display backlight
5754 brightness across reboots has been added. It will store the
5755 backlight setting as late as possible at shutdown, and
5756 restore it as early as possible during reboot.
5758 * A logic to automatically discover and enable home and swap
5759 partitions on GPT disks has been added. With this in place
5760 /etc/fstab becomes optional for many setups as systemd can
5761 discover certain partitions located on the root disk
5762 automatically. Home partitions are recognized under their
5763 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5764 partitions are recognized under their GPT type ID
5765 0657fd6da4ab43c484e50933c84b4f4f.
5767 * systemd will no longer pass any environment from the kernel
5768 or initrd to system services. If you want to set an
5769 environment for all services, do so via the kernel command
5770 line systemd.setenv= assignment.
5772 * The systemd-sysctl tool no longer natively reads the file
5773 /etc/sysctl.conf. If desired, the file should be symlinked
5774 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5775 legacy support by a symlink rather than built-in code, it
5776 also makes the otherwise hidden order of application of the
5777 different files visible. (Note that this partly reverts to a
5778 pre-198 application order of sysctl knobs!)
5780 * The "systemctl set-log-level" and "systemctl dump" commands
5781 have been moved to systemd-analyze.
5783 * systemd-run learned the new --remain-after-exit switch,
5784 which causes the scope unit not to be cleaned up
5785 automatically after the process terminated.
5787 * tmpfiles learned a new --exclude-prefix= switch to exclude
5788 certain paths from operation.
5790 * journald will now automatically flush all messages to disk
5791 as soon as a message at the log level CRIT, ALERT or EMERG
5794 Contributions from: Andrew Cook, Brandon Philips, Christian
5795 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5796 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5797 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5798 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5799 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5800 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5801 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5802 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5803 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5804 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5805 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5806 William Giokas, Zbigniew Jędrzejewski-Szmek
5808 — Berlin, 2013-09-13
5812 * The documentation has been updated to cover the various new
5813 concepts introduced with 205.
5815 * Unit files now understand the new %v specifier which
5816 resolves to the kernel version string as returned by "uname
5819 * systemctl now supports filtering the unit list output by
5820 load state, active state and sub state, using the new
5823 * "systemctl status" will now show the results of the
5824 condition checks (like ConditionPathExists= and similar) of
5825 the last start attempts of the unit. They are also logged to
5828 * "journalctl -b" may now be used to look for boot output of a
5829 specific boot. Try "journalctl -b -1" for the previous boot,
5830 but the syntax is substantially more powerful.
5832 * "journalctl --show-cursor" has been added which prints the
5833 cursor string the last shown log line. This may then be used
5834 with the new "journalctl --after-cursor=" switch to continue
5835 browsing logs from that point on.
5837 * "journalctl --force" may now be used to force regeneration
5840 * Creation of "dead" device nodes has been moved from udev
5841 into kmod and tmpfiles. Previously, udev would read the kmod
5842 databases to pre-generate dead device nodes based on meta
5843 information contained in kernel modules, so that these would
5844 be auto-loaded on access rather then at boot. As this
5845 does not really have much to do with the exposing actual
5846 kernel devices to userspace this has always been slightly
5847 alien in the udev codebase. Following the new scheme kmod
5848 will now generate a runtime snippet for tmpfiles from the
5849 module meta information and it now is tmpfiles' job to the
5850 create the nodes. This also allows overriding access and
5851 other parameters for the nodes using the usual tmpfiles
5852 facilities. As side effect this allows us to remove the
5853 CAP_SYS_MKNOD capability bit from udevd entirely.
5855 * logind's device ACLs may now be applied to these "dead"
5856 devices nodes too, thus finally allowing managed access to
5857 devices such as /dev/snd/sequencer without loading the
5858 backing module right-away.
5860 * A new RPM macro has been added that may be used to apply
5861 tmpfiles configuration during package installation.
5863 * systemd-detect-virt and ConditionVirtualization= now can
5864 detect User-Mode-Linux machines (UML).
5866 * journald will now implicitly log the effective capabilities
5867 set of processes in the message metadata.
5869 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5871 * The initrd interface has been simplified (more specifically,
5872 support for passing performance data via environment
5873 variables and fsck results via files in /run has been
5874 removed). These features were non-essential, and are
5875 nowadays available in a much nicer way by having systemd in
5876 the initrd serialize its state and have the hosts systemd
5877 deserialize it again.
5879 * The udev "keymap" data files and tools to apply keyboard
5880 specific mappings of scan to key codes, and force-release
5881 scan code lists have been entirely replaced by a udev
5882 "keyboard" builtin and a hwdb data file.
5884 * systemd will now honour the kernel's "quiet" command line
5885 argument also during late shutdown, resulting in a
5886 completely silent shutdown when used.
5888 * There's now an option to control the SO_REUSEPORT socket
5889 option in .socket units.
5891 * Instance units will now automatically get a per-template
5892 subslice of system.slice unless something else is explicitly
5893 configured. For example, instances of sshd@.service will now
5894 implicitly be placed in system-sshd.slice rather than
5895 system.slice as before.
5897 * Test coverage support may now be enabled at build time.
5899 Contributions from: Dave Reisner, Frederic Crozat, Harald
5900 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5901 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5902 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5903 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5904 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5905 Giokas, Zbigniew Jędrzejewski-Szmek
5907 — Berlin, 2013-07-23
5911 * Two new unit types have been introduced:
5913 Scope units are very similar to service units, however, are
5914 created out of pre-existing processes — instead of PID 1
5915 forking off the processes. By using scope units it is
5916 possible for system services and applications to group their
5917 own child processes (worker processes) in a powerful way
5918 which then maybe used to organize them, or kill them
5919 together, or apply resource limits on them.
5921 Slice units may be used to partition system resources in an
5922 hierarchical fashion and then assign other units to them. By
5923 default there are now three slices: system.slice (for all
5924 system services), user.slice (for all user sessions),
5925 machine.slice (for VMs and containers).
5927 Slices and scopes have been introduced primarily in
5928 context of the work to move cgroup handling to a
5929 single-writer scheme, where only PID 1
5930 creates/removes/manages cgroups.
5932 * There's a new concept of "transient" units. In contrast to
5933 normal units these units are created via an API at runtime,
5934 not from configuration from disk. More specifically this
5935 means it is now possible to run arbitrary programs as
5936 independent services, with all execution parameters passed
5937 in via bus APIs rather than read from disk. Transient units
5938 make systemd substantially more dynamic then it ever was,
5939 and useful as a general batch manager.
5941 * logind has been updated to make use of scope and slice units
5942 for managing user sessions. As a user logs in he will get
5943 his own private slice unit, to which all sessions are added
5944 as scope units. We also added support for automatically
5945 adding an instance of user@.service for the user into the
5946 slice. Effectively logind will no longer create cgroup
5947 hierarchies on its own now, it will defer entirely to PID 1
5948 for this by means of scope, service and slice units. Since
5949 user sessions this way become entities managed by PID 1
5950 the output of "systemctl" is now a lot more comprehensive.
5952 * A new mini-daemon "systemd-machined" has been added which
5953 may be used by virtualization managers to register local
5954 VMs/containers. nspawn has been updated accordingly, and
5955 libvirt will be updated shortly. machined will collect a bit
5956 of meta information about the VMs/containers, and assign
5957 them their own scope unit (see above). The collected
5958 meta-data is then made available via the "machinectl" tool,
5959 and exposed in "ps" and similar tools. machined/machinectl
5960 is compile-time optional.
5962 * As discussed earlier, the low-level cgroup configuration
5963 options ControlGroup=, ControlGroupModify=,
5964 ControlGroupPersistent=, ControlGroupAttribute= have been
5965 removed. Please use high-level attribute settings instead as
5966 well as slice units.
5968 * A new bus call SetUnitProperties() has been added to alter
5969 various runtime parameters of a unit. This is primarily
5970 useful to alter cgroup parameters dynamically in a nice way,
5971 but will be extended later on to make more properties
5972 modifiable at runtime. systemctl gained a new set-properties
5973 command that wraps this call.
5975 * A new tool "systemd-run" has been added which can be used to
5976 run arbitrary command lines as transient services or scopes,
5977 while configuring a number of settings via the command
5978 line. This tool is currently very basic, however already
5979 very useful. We plan to extend this tool to even allow
5980 queuing of execution jobs with time triggers from the
5981 command line, similar in fashion to "at".
5983 * nspawn will now inform the user explicitly that kernels with
5984 audit enabled break containers, and suggest the user to turn
5987 * Support for detecting the IMA and AppArmor security
5988 frameworks with ConditionSecurity= has been added.
5990 * journalctl gained a new "-k" switch for showing only kernel
5991 messages, mimicking dmesg output; in addition to "--user"
5992 and "--system" switches for showing only user's own logs
5995 * systemd-delta can now show information about drop-in
5996 snippets extending unit files.
5998 * libsystemd-bus has been substantially updated but is still
5999 not available as public API.
6001 * systemd will now look for the "debug" argument on the kernel
6002 command line and enable debug logging, similar to what
6003 "systemd.log_level=debug" already did before.
6005 * "systemctl set-default", "systemctl get-default" has been
6006 added to configure the default.target symlink, which
6007 controls what to boot into by default.
6009 * "systemctl set-log-level" has been added as a convenient
6010 way to raise and lower systemd logging threshold.
6012 * "systemd-analyze plot" will now show the time the various
6013 generators needed for execution, as well as information
6014 about the unit file loading.
6016 * libsystemd-journal gained a new sd_journal_open_files() call
6017 for opening specific journal files. journactl also gained a
6018 new switch to expose this new functionality. Previously we
6019 only supported opening all files from a directory, or all
6020 files from the system, as opening individual files only is
6021 racy due to journal file rotation.
6023 * systemd gained the new DefaultEnvironment= setting in
6024 /etc/systemd/system.conf to set environment variables for
6027 * If a privileged process logs a journal message with the
6028 OBJECT_PID= field set, then journald will automatically
6029 augment this with additional OBJECT_UID=, OBJECT_GID=,
6030 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6031 system services want to log events about specific client
6032 processes. journactl/systemctl has been updated to make use
6033 of this information if all log messages regarding a specific
6036 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6037 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6038 Reisner, David Coppa, David King, David Strauss, Eelco
6039 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6040 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6041 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6042 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6043 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6044 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6045 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6046 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6047 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6048 Łukasz Stelmach, 장동준
6052 * The Python bindings gained some minimal support for the APIs
6053 exposed by libsystemd-logind.
6055 * ConditionSecurity= gained support for detecting SMACK. Since
6056 this condition already supports SELinux and AppArmor we only
6057 miss IMA for this. Patches welcome!
6059 Contributions from: Karol Lewandowski, Lennart Poettering,
6060 Zbigniew Jędrzejewski-Szmek
6064 * systemd-nspawn will now create /etc/resolv.conf if
6065 necessary, before bind-mounting the host's file onto it.
6067 * systemd-nspawn will now store meta information about a
6068 container on the container's cgroup as extended attribute
6069 fields, including the root directory.
6071 * The cgroup hierarchy has been reworked in many ways. All
6072 objects any of the components systemd creates in the cgroup
6073 tree are now suffixed. More specifically, user sessions are
6074 now placed in cgroups suffixed with ".session", users in
6075 cgroups suffixed with ".user", and nspawn containers in
6076 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6077 names are now escaped in a simple scheme to avoid collision
6078 of userspace object names with kernel filenames. This work
6079 is preparation for making these objects relocatable in the
6080 cgroup tree, in order to allow easy resource partitioning of
6081 these objects without causing naming conflicts.
6083 * systemctl list-dependencies gained the new switches
6084 --plain, --reverse, --after and --before.
6086 * systemd-inhibit now shows the process name of processes that
6087 have taken an inhibitor lock.
6089 * nss-myhostname will now also resolve "localhost"
6090 implicitly. This makes /etc/hosts an optional file and
6091 nicely handles that on IPv6 ::1 maps to both "localhost" and
6094 * libsystemd-logind.so gained a new call
6095 sd_get_machine_names() to enumerate running containers and
6096 VMs (currently only supported by very new libvirt and
6097 nspawn). sd_login_monitor can now be used to watch
6098 VMs/containers coming and going.
6100 * .include is not allowed recursively anymore, and only in
6101 unit files. Usually it is better to use drop-in snippets in
6102 .d/*.conf anyway, as introduced with systemd 198.
6104 * systemd-analyze gained a new "critical-chain" command that
6105 determines the slowest chain of units run during system
6106 boot-up. It is very useful for tracking down where
6107 optimizing boot time is the most beneficial.
6109 * systemd will no longer allow manipulating service paths in
6110 the name=systemd:/system cgroup tree using ControlGroup= in
6111 units. (But is still fine with it in all other dirs.)
6113 * There's a new systemd-nspawn@.service service file that may
6114 be used to easily run nspawn containers as system
6115 services. With the container's root directory in
6116 /var/lib/container/foobar it is now sufficient to run
6117 "systemctl start systemd-nspawn@foobar.service" to boot it.
6119 * systemd-cgls gained a new parameter "--machine" to list only
6120 the processes within a certain container.
6122 * ConditionSecurity= now can check for "apparmor". We still
6123 are lacking checks for SMACK and IMA for this condition
6124 check though. Patches welcome!
6126 * A new configuration file /etc/systemd/sleep.conf has been
6127 added that may be used to configure which kernel operation
6128 systemd is supposed to execute when "suspend", "hibernate"
6129 or "hybrid-sleep" is requested. This makes the new kernel
6130 "freeze" state accessible to the user.
6132 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6133 the passed argument if applicable.
6135 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6136 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6137 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6138 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6139 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6140 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6141 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6146 * The output of 'systemctl list-jobs' got some polishing. The
6147 '--type=' argument may now be passed more than once. A new
6148 command 'systemctl list-sockets' has been added which shows
6149 a list of kernel sockets systemd is listening on with the
6150 socket units they belong to, plus the units these socket
6153 * The experimental libsystemd-bus library got substantial
6154 updates to work in conjunction with the (also experimental)
6155 kdbus kernel project. It works well enough to exchange
6156 messages with some sophistication. Note that kdbus is not
6157 ready yet, and the library is mostly an elaborate test case
6158 for now, and not installable.
6160 * systemd gained a new unit 'systemd-static-nodes.service'
6161 that generates static device nodes earlier during boot, and
6162 can run in conjunction with udev.
6164 * libsystemd-login gained a new call sd_pid_get_user_unit()
6165 to retrieve the user systemd unit a process is running
6166 in. This is useful for systems where systemd is used as
6169 * systemd-nspawn now places all containers in the new /machine
6170 top-level cgroup directory in the name=systemd
6171 hierarchy. libvirt will soon do the same, so that we get a
6172 uniform separation of /system, /user and /machine for system
6173 services, user processes and containers/virtual
6174 machines. This new cgroup hierarchy is also useful to stick
6175 stable names to specific container instances, which can be
6176 recognized later this way (this name may be controlled
6177 via systemd-nspawn's new -M switch). libsystemd-login also
6178 gained a new call sd_pid_get_machine_name() to retrieve the
6179 name of the container/VM a specific process belongs to.
6181 * bootchart can now store its data in the journal.
6183 * libsystemd-journal gained a new call
6184 sd_journal_add_conjunction() for AND expressions to the
6185 matching logic. This can be used to express more complex
6186 logical expressions.
6188 * journactl can now take multiple --unit= and --user-unit=
6191 * The cryptsetup logic now understands the "luks.key=" kernel
6192 command line switch for specifying a file to read the
6193 decryption key from. Also, if a configured key file is not
6194 found the tool will now automatically fall back to prompting
6197 * Python systemd.journal module was updated to wrap recently
6198 added functions from libsystemd-journal. The interface was
6199 changed to bring the low level interface in s.j._Reader
6200 closer to the C API, and the high level interface in
6201 s.j.Reader was updated to wrap and convert all data about
6204 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6205 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6206 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6207 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6208 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6209 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6213 * journalctl --update-catalog now understands a new --root=
6214 option to operate on catalogs found in a different root
6217 * During shutdown after systemd has terminated all running
6218 services a final killing loop kills all remaining left-over
6219 processes. We will now print the name of these processes
6220 when we send SIGKILL to them, since this usually indicates a
6223 * If /etc/crypttab refers to password files stored on
6224 configured mount points automatic dependencies will now be
6225 generated to ensure the specific mount is established first
6226 before the key file is attempted to be read.
6228 * 'systemctl status' will now show information about the
6229 network sockets a socket unit is listening on.
6231 * 'systemctl status' will also shown information about any
6232 drop-in configuration file for units. (Drop-In configuration
6233 files in this context are files such as
6234 /etc/systemd/systemd/foobar.service.d/*.conf)
6236 * systemd-cgtop now optionally shows summed up CPU times of
6237 cgroups. Press '%' while running cgtop to switch between
6238 percentage and absolute mode. This is useful to determine
6239 which cgroups use up the most CPU time over the entire
6240 runtime of the system. systemd-cgtop has also been updated
6241 to be 'pipeable' for processing with further shell tools.
6243 * 'hostnamectl set-hostname' will now allow setting of FQDN
6246 * The formatting and parsing of time span values has been
6247 changed. The parser now understands fractional expressions
6248 such as "5.5h". The formatter will now output fractional
6249 expressions for all time spans under 1min, i.e. "5.123456s"
6250 rather than "5s 123ms 456us". For time spans under 1s
6251 millisecond values are shown, for those under 1ms
6252 microsecond values are shown. This should greatly improve
6253 all time-related output of systemd.
6255 * libsystemd-login and libsystemd-journal gained new
6256 functions for querying the poll() events mask and poll()
6257 timeout value for integration into arbitrary event
6260 * localectl gained the ability to list available X11 keymaps
6261 (models, layouts, variants, options).
6263 * 'systemd-analyze dot' gained the ability to filter for
6264 specific units via shell-style globs, to create smaller,
6265 more useful graphs. I.e. it is now possible to create simple
6266 graphs of all the dependencies between only target units, or
6267 of all units that Avahi has dependencies with.
6269 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6270 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6271 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6272 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6273 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6274 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6275 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6279 * The boot-time readahead implementation for rotating media
6280 will now read the read-ahead data in multiple passes which
6281 consist of all read requests made in equidistant time
6282 intervals. This means instead of strictly reading read-ahead
6283 data in its physical order on disk we now try to find a
6284 middle ground between physical and access time order.
6286 * /etc/os-release files gained a new BUILD_ID= field for usage
6287 on operating systems that provide continuous builds of OS
6290 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6291 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6292 William Douglas, Zbigniew Jędrzejewski-Szmek
6296 * systemd-python gained an API exposing libsystemd-daemon.
6298 * The SMACK setup logic gained support for uploading CIPSO
6301 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6302 ReadOnlyDirectories= and InaccessibleDirectories= has
6303 changed. The private /tmp and /var/tmp directories are now
6304 shared by all processes of a service (which means
6305 ExecStartPre= may now leave data in /tmp that ExecStart= of
6306 the same service can still access). When a service is
6307 stopped its temporary directories are immediately deleted
6308 (normal clean-up with tmpfiles is still done in addition to
6311 * By default, systemd will now set a couple of sysctl
6312 variables in the kernel: the safe sysrq options are turned
6313 on, IP route verification is turned on, and source routing
6314 disabled. The recently added hardlink and softlink
6315 protection of the kernel is turned on. These settings should
6316 be reasonably safe, and good defaults for all new systems.
6318 * The predictable network naming logic may now be turned off
6319 with a new kernel command line switch: net.ifnames=0.
6321 * A new libsystemd-bus module has been added that implements a
6322 pretty complete D-Bus client library. For details see:
6324 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6326 * journald will now explicitly flush the journal files to disk
6327 at the latest 5min after each write. The file will then also
6328 be marked offline until the next write. This should increase
6329 reliability in case of a crash. The synchronization delay
6330 can be configured via SyncIntervalSec= in journald.conf.
6332 * There's a new remote-fs-setup.target unit that can be used
6333 to pull in specific services when at least one remote file
6334 system is to be mounted.
6336 * There are new targets timers.target and paths.target as
6337 canonical targets to pull user timer and path units in
6338 from. This complements sockets.target with a similar
6339 purpose for socket units.
6341 * libudev gained a new call udev_device_set_attribute_value()
6342 to set sysfs attributes of a device.
6344 * The udev daemon now sets the default number of worker
6345 processes executed in parallel based on the number of available
6346 CPUs instead of the amount of available RAM. This is supposed
6347 to provide a more reliable default and limit a too aggressive
6348 parallelism for setups with 1000s of devices connected.
6350 Contributions from: Auke Kok, Colin Walters, Cristian
6351 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6352 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6353 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6354 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6355 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6356 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6357 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6358 Zbigniew Jędrzejewski-Szmek
6362 * Configuration of unit files may now be extended via drop-in
6363 files without having to edit/override the unit files
6364 themselves. More specifically, if the administrator wants to
6365 change one value for a service file foobar.service he can
6366 now do so by dropping in a configuration snippet into
6367 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6368 will load all these snippets and apply them on top of the
6369 main unit configuration file, possibly extending or
6370 overriding its settings. Using these drop-in snippets is
6371 generally nicer than the two earlier options for changing
6372 unit files locally: copying the files from
6373 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6374 them there; or creating a new file in /etc/systemd/system/
6375 that incorporates the original one via ".include". Drop-in
6376 snippets into these .d/ directories can be placed in any
6377 directory systemd looks for units in, and the usual
6378 overriding semantics between /usr/lib, /etc and /run apply
6381 * Most unit file settings which take lists of items can now be
6382 reset by assigning the empty string to them. For example,
6383 normally, settings such as Environment=FOO=BAR append a new
6384 environment variable assignment to the environment block,
6385 each time they are used. By assigning Environment= the empty
6386 string the environment block can be reset to empty. This is
6387 particularly useful with the .d/*.conf drop-in snippets
6388 mentioned above, since this adds the ability to reset list
6389 settings from vendor unit files via these drop-ins.
6391 * systemctl gained a new "list-dependencies" command for
6392 listing the dependencies of a unit recursively.
6394 * Inhibitors are now honored and listed by "systemctl
6395 suspend", "systemctl poweroff" (and similar) too, not only
6396 GNOME. These commands will also list active sessions by
6399 * Resource limits (as exposed by the various control group
6400 controllers) can now be controlled dynamically at runtime
6401 for all units. More specifically, you can now use a command
6402 like "systemctl set-cgroup-attr foobar.service cpu.shares
6403 2000" to alter the CPU shares a specific service gets. These
6404 settings are stored persistently on disk, and thus allow the
6405 administrator to easily adjust the resource usage of
6406 services with a few simple commands. This dynamic resource
6407 management logic is also available to other programs via the
6408 bus. Almost any kernel cgroup attribute and controller is
6411 * systemd-vconsole-setup will now copy all font settings to
6412 all allocated VTs, where it previously applied them only to
6415 * libsystemd-login gained the new sd_session_get_tty() API
6418 * This release drops support for a few legacy or
6419 distribution-specific LSB facility names when parsing init
6420 scripts: $x-display-manager, $mail-transfer-agent,
6421 $mail-transport-agent, $mail-transfer-agent, $smtp,
6422 $null. Also, the mail-transfer-agent.target unit backing
6423 this has been removed. Distributions which want to retain
6424 compatibility with this should carry the burden for
6425 supporting this themselves and patch support for these back
6426 in, if they really need to. Also, the facilities $syslog and
6427 $local_fs are now ignored, since systemd does not support
6428 early-boot LSB init scripts anymore, and these facilities
6429 are implied anyway for normal services. syslog.target has
6432 * There are new bus calls on PID1's Manager object for
6433 cancelling jobs, and removing snapshot units. Previously,
6434 both calls were only available on the Job and Snapshot
6437 * systemd-journal-gatewayd gained SSL support.
6439 * The various "environment" files, such as /etc/locale.conf
6440 now support continuation lines with a backslash ("\") as
6441 last character in the line, similarly in style (but different)
6442 to how this is supported in shells.
6444 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6445 now implicitly appended to every log entry logged. systemctl
6446 has been updated to filter by this field when operating on a
6447 user systemd instance.
6449 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6450 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6451 the container. This makes it easier to boot unmodified
6452 Fedora systems in a container, which however still requires
6453 audit=0 to be passed on the kernel command line. Auditing in
6454 kernel and userspace is unfortunately still too broken in
6455 context of containers, hence we recommend compiling it out
6456 of the kernel or using audit=0. Hopefully this will be fixed
6457 one day for good in the kernel.
6459 * nspawn gained the new --bind= and --bind-ro= parameters to
6460 bind mount specific directories from the host into the
6463 * nspawn will now mount its own devpts file system instance
6464 into the container, in order not to leak pty devices from
6465 the host into the container.
6467 * systemd will now read the firmware boot time performance
6468 information from the EFI variables, if the used boot loader
6469 supports this, and takes it into account for boot performance
6470 analysis via "systemd-analyze". This is currently supported
6471 only in conjunction with Gummiboot, but could be supported
6472 by other boot loaders too. For details see:
6474 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6476 * A new generator has been added that automatically mounts the
6477 EFI System Partition (ESP) to /boot, if that directory
6478 exists, is empty, and no other file system has been
6479 configured to be mounted there.
6481 * logind will now send out PrepareForSleep(false) out
6482 unconditionally, after coming back from suspend. This may be
6483 used by applications as asynchronous notification for
6484 system resume events.
6486 * "systemctl unlock-sessions" has been added, that allows
6487 unlocking the screens of all user sessions at once, similar
6488 to how "systemctl lock-sessions" already locked all users
6489 sessions. This is backed by a new D-Bus call UnlockSessions().
6491 * "loginctl seat-status" will now show the master device of a
6492 seat. (i.e. the device of a seat that needs to be around for
6493 the seat to be considered available, usually the graphics
6496 * tmpfiles gained a new "X" line type, that allows
6497 configuration of files and directories (with wildcards) that
6498 shall be excluded from automatic cleanup ("aging").
6500 * udev default rules set the device node permissions now only
6501 at "add" events, and do not change them any longer with a
6502 later "change" event.
6504 * The log messages for lid events and power/sleep keypresses
6505 now carry a message ID.
6507 * We now have a substantially larger unit test suite, but this
6508 continues to be work in progress.
6510 * udevadm hwdb gained a new --root= parameter to change the
6511 root directory to operate relative to.
6513 * logind will now issue a background sync() request to the kernel
6514 early at shutdown, so that dirty buffers are flushed to disk early
6515 instead of at the last moment, in order to optimize shutdown
6518 * A new bootctl tool has been added that is an interface for
6519 certain boot loader operations. This is currently a preview
6520 and is likely to be extended into a small mechanism daemon
6521 like timedated, localed, hostnamed, and can be used by
6522 graphical UIs to enumerate available boot options, and
6523 request boot into firmware operations.
6525 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6526 the rest of the package. It also has been updated to work
6527 correctly in initrds.
6529 * Policykit previously has been runtime optional, and is now
6530 also compile time optional via a configure switch.
6532 * systemd-analyze has been reimplemented in C. Also "systemctl
6533 dot" has moved into systemd-analyze.
6535 * "systemctl status" with no further parameters will now print
6536 the status of all active or failed units.
6538 * Operations such as "systemctl start" can now be executed
6539 with a new mode "--irreversible" which may be used to queue
6540 operations that cannot accidentally be reversed by a later
6541 job queuing. This is by default used to make shutdown
6542 requests more robust.
6544 * The Python API of systemd now gained a new module for
6545 reading journal files.
6547 * A new tool kernel-install has been added that can install
6548 kernel images according to the Boot Loader Specification:
6550 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6552 * Boot time console output has been improved to provide
6553 animated boot time output for hanging jobs.
6555 * A new tool systemd-activate has been added which can be used
6556 to test socket activation with, directly from the command
6557 line. This should make it much easier to test and debug
6558 socket activation in daemons.
6560 * journalctl gained a new "--reverse" (or -r) option to show
6561 journal output in reverse order (i.e. newest line first).
6563 * journalctl gained a new "--pager-end" (or -e) option to jump
6564 to immediately jump to the end of the journal in the
6565 pager. This is only supported in conjunction with "less".
6567 * journalctl gained a new "--user-unit=" option, that works
6568 similarly to "--unit=" but filters for user units rather than
6571 * A number of unit files to ease adoption of systemd in
6572 initrds has been added. This moves some minimal logic from
6573 the various initrd implementations into systemd proper.
6575 * The journal files are now owned by a new group
6576 "systemd-journal", which exists specifically to allow access
6577 to the journal, and nothing else. Previously, we used the
6578 "adm" group for that, which however possibly covers more
6579 than just journal/log file access. This new group is now
6580 already used by systemd-journal-gatewayd to ensure this
6581 daemon gets access to the journal files and as little else
6582 as possible. Note that "make install" will also set FS ACLs
6583 up for /var/log/journal to give "adm" and "wheel" read
6584 access to it, in addition to "systemd-journal" which owns
6585 the journal files. We recommend that packaging scripts also
6586 add read access to "adm" + "wheel" to /var/log/journal, and
6587 all existing/future journal files. To normal users and
6588 administrators little changes, however packagers need to
6589 ensure to create the "systemd-journal" system group at
6590 package installation time.
6592 * The systemd-journal-gatewayd now runs as unprivileged user
6593 systemd-journal-gateway:systemd-journal-gateway. Packaging
6594 scripts need to create these system user/group at
6597 * timedated now exposes a new boolean property CanNTP that
6598 indicates whether a local NTP service is available or not.
6600 * systemd-detect-virt will now also detect xen PVs
6602 * The pstore file system is now mounted by default, if it is
6605 * In addition to the SELinux and IMA policies we will now also
6606 load SMACK policies at early boot.
6608 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6609 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6610 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6611 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6612 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6613 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6614 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6615 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6616 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6617 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6618 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6619 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6620 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6621 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6625 * Timer units now support calendar time events in addition to
6626 monotonic time events. That means you can now trigger a unit
6627 based on a calendar time specification such as "Thu,Fri
6628 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6629 or fifth day of any month of the year 2013, given that it is
6630 a thursday or friday. This brings timer event support
6631 considerably closer to cron's capabilities. For details on
6632 the supported calendar time specification language see
6635 * udev now supports a number of different naming policies for
6636 network interfaces for predictable names, and a combination
6637 of these policies is now the default. Please see this wiki
6638 document for details:
6640 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6642 * Auke Kok's bootchart implementation has been added to the
6643 systemd tree. It is an optional component that can graph the
6644 boot in quite some detail. It is one of the best bootchart
6645 implementations around and minimal in its code and
6648 * nss-myhostname has been integrated into the systemd source
6649 tree. nss-myhostname guarantees that the local hostname
6650 always stays resolvable via NSS. It has been a weak
6651 requirement of systemd-hostnamed since a long time, and
6652 since its code is actually trivial we decided to just
6653 include it in systemd's source tree. It can be turned off
6654 with a configure switch.
6656 * The read-ahead logic is now capable of properly detecting
6657 whether a btrfs file system is on SSD or rotating media, in
6658 order to optimize the read-ahead scheme. Previously, it was
6659 only capable of detecting this on traditional file systems
6662 * In udev, additional device properties are now read from the
6663 IAB in addition to the OUI database. Also, Bluetooth company
6664 identities are attached to the devices as well.
6666 * In service files %U may be used as specifier that is
6667 replaced by the configured user name of the service.
6669 * nspawn may now be invoked without a controlling TTY. This
6670 makes it suitable for invocation as its own service. This
6671 may be used to set up a simple containerized server system
6672 using only core OS tools.
6674 * systemd and nspawn can now accept socket file descriptors
6675 when they are started for socket activation. This enables
6676 implementation of socket activated nspawn
6677 containers. i.e. think about autospawning an entire OS image
6678 when the first SSH or HTTP connection is received. We expect
6679 that similar functionality will also be added to libvirt-lxc
6682 * journalctl will now suppress ANSI color codes when
6683 presenting log data.
6685 * systemctl will no longer show control group information for
6686 a unit if the control group is empty anyway.
6688 * logind can now automatically suspend/hibernate/shutdown the
6691 * /etc/machine-info and hostnamed now also expose the chassis
6692 type of the system. This can be used to determine whether
6693 the local system is a laptop, desktop, handset or
6694 tablet. This information may either be configured by the
6695 user/vendor or is automatically determined from ACPI and DMI
6696 information if possible.
6698 * A number of PolicyKit actions are now bound together with
6699 "imply" rules. This should simplify creating UIs because
6700 many actions will now authenticate similar ones as well.
6702 * Unit files learnt a new condition ConditionACPower= which
6703 may be used to conditionalize a unit depending on whether an
6704 AC power source is connected or not, of whether the system
6705 is running on battery power.
6707 * systemctl gained a new "is-failed" verb that may be used in
6708 shell scripts and suchlike to check whether a specific unit
6709 is in the "failed" state.
6711 * The EnvironmentFile= setting in unit files now supports file
6712 globbing, and can hence be used to easily read a number of
6713 environment files at once.
6715 * systemd will no longer detect and recognize specific
6716 distributions. All distribution-specific #ifdeffery has been
6717 removed, systemd is now fully generic and
6718 distribution-agnostic. Effectively, not too much is lost as
6719 a lot of the code is still accessible via explicit configure
6720 switches. However, support for some distribution specific
6721 legacy configuration file formats has been dropped. We
6722 recommend distributions to simply adopt the configuration
6723 files everybody else uses now and convert the old
6724 configuration from packaging scripts. Most distributions
6725 already did that. If that's not possible or desirable,
6726 distributions are welcome to forward port the specific
6727 pieces of code locally from the git history.
6729 * When logging a message about a unit systemd will now always
6730 log the unit name in the message meta data.
6732 * localectl will now also discover system locale data that is
6733 not stored in locale archives, but directly unpacked.
6735 * logind will no longer unconditionally use framebuffer
6736 devices as seat masters, i.e. as devices that are required
6737 to be existing before a seat is considered preset. Instead,
6738 it will now look for all devices that are tagged as
6739 "seat-master" in udev. By default, framebuffer devices will
6740 be marked as such, but depending on local systems, other
6741 devices might be marked as well. This may be used to
6742 integrate graphics cards using closed source drivers (such
6743 as NVidia ones) more nicely into logind. Note however, that
6744 we recommend using the open source NVidia drivers instead,
6745 and no udev rules for the closed-source drivers will be
6746 shipped from us upstream.
6748 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6749 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6750 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6751 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6752 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6753 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6754 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6755 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6756 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6757 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6758 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6759 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6764 * udev gained support for loading additional device properties
6765 from an indexed database that is keyed by vendor/product IDs
6766 and similar device identifiers. For the beginning this
6767 "hwdb" is populated with data from the well-known PCI and
6768 USB database, but also includes PNP, ACPI and OID data. In
6769 the longer run this indexed database shall grow into
6770 becoming the one central database for non-essential
6771 userspace device metadata. Previously, data from the PCI/USB
6772 database was only attached to select devices, since the
6773 lookup was a relatively expensive operation due to O(n) time
6774 complexity (with n being the number of entries in the
6775 database). Since this is now O(1), we decided to add in this
6776 data for all devices where this is available, by
6777 default. Note that the indexed database needs to be rebuilt
6778 when new data files are installed. To achieve this you need
6779 to update your packaging scripts to invoke "udevadm hwdb
6780 --update" after installation of hwdb data files. For
6781 RPM-based distributions we introduced the new
6782 %udev_hwdb_update macro for this purpose.
6784 * The Journal gained support for the "Message Catalog", an
6785 indexed database to link up additional information with
6786 journal entries. For further details please check:
6788 https://www.freedesktop.org/wiki/Software/systemd/catalog
6790 The indexed message catalog database also needs to be
6791 rebuilt after installation of message catalog files. Use
6792 "journalctl --update-catalog" for this. For RPM-based
6793 distributions we introduced the %journal_catalog_update
6794 macro for this purpose.
6796 * The Python Journal bindings gained support for the standard
6797 Python logging framework.
6799 * The Journal API gained new functions for checking whether
6800 the underlying file system of a journal file is capable of
6801 properly reporting file change notifications, or whether
6802 applications that want to reflect journal changes "live"
6803 need to recheck journal files continuously in appropriate
6806 * It is now possible to set the "age" field for tmpfiles
6807 entries to 0, indicating that files matching this entry
6808 shall always be removed when the directories are cleaned up.
6810 * coredumpctl gained a new "gdb" verb which invokes gdb
6811 right-away on the selected coredump.
6813 * There's now support for "hybrid sleep" on kernels that
6814 support this, in addition to "suspend" and "hibernate". Use
6815 "systemctl hybrid-sleep" to make use of this.
6817 * logind's HandleSuspendKey= setting (and related settings)
6818 now gained support for a new "lock" setting to simply
6819 request the screen lock on all local sessions, instead of
6820 actually executing a suspend or hibernation.
6822 * systemd will now mount the EFI variables file system by
6825 * Socket units now gained support for configuration of the
6826 SMACK security label.
6828 * timedatectl will now output the time of the last and next
6829 daylight saving change.
6831 * We dropped support for various legacy and distro-specific
6832 concepts, such as insserv, early-boot SysV services
6833 (i.e. those for non-standard runlevels such as 'b' or 'S')
6834 or ArchLinux /etc/rc.conf support. We recommend the
6835 distributions who still need support this to either continue
6836 to maintain the necessary patches downstream, or find a
6837 different solution. (Talk to us if you have questions!)
6839 * Various systemd components will now bypass PolicyKit checks
6840 for root and otherwise handle properly if PolicyKit is not
6841 found to be around. This should fix most issues for
6842 PolicyKit-less systems. Quite frankly this should have been
6843 this way since day one. It is absolutely our intention to
6844 make systemd work fine on PolicyKit-less systems, and we
6845 consider it a bug if something does not work as it should if
6846 PolicyKit is not around.
6848 * For embedded systems it is now possible to build udev and
6849 systemd without blkid and/or kmod support.
6851 * "systemctl switch-root" is now capable of switching root
6852 more than once. I.e. in addition to transitions from the
6853 initrd to the host OS it is now possible to transition to
6854 further OS images from the host. This is useful to implement
6855 offline updating tools.
6857 * Various other additions have been made to the RPM macros
6858 shipped with systemd. Use %udev_rules_update() after
6859 installing new udev rules files. %_udevhwdbdir,
6860 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6861 %_sysctldir are now available which resolve to the right
6862 directories for packages to place various data files in.
6864 * journalctl gained the new --full switch (in addition to
6865 --all, to disable ellipsation for long messages.
6867 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6868 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6869 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6870 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6871 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6872 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6873 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6874 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6875 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6879 * journalctl gained new --since= and --until= switches to
6880 filter by time. It also now supports nice filtering for
6881 units via --unit=/-u.
6883 * Type=oneshot services may use ExecReload= and do the
6886 * The journal daemon now supports time-based rotation and
6887 vacuuming, in addition to the usual disk-space based
6890 * The journal will now index the available field values for
6891 each field name. This enables clients to show pretty drop
6892 downs of available match values when filtering. The bash
6893 completion of journalctl has been updated
6894 accordingly. journalctl gained a new switch -F to list all
6895 values a certain field takes in the journal database.
6897 * More service events are now written as structured messages
6898 to the journal, and made recognizable via message IDs.
6900 * The timedated, localed and hostnamed mini-services which
6901 previously only provided support for changing time, locale
6902 and hostname settings from graphical DEs such as GNOME now
6903 also have a minimal (but very useful) text-based client
6904 utility each. This is probably the nicest way to changing
6905 these settings from the command line now, especially since
6906 it lists available options and is fully integrated with bash
6909 * There's now a new tool "systemd-coredumpctl" to list and
6910 extract coredumps from the journal.
6912 * We now install a README each in /var/log/ and
6913 /etc/rc.d/init.d explaining where the system logs and init
6914 scripts went. This hopefully should help folks who go to
6915 that dirs and look into the otherwise now empty void and
6916 scratch their heads.
6918 * When user-services are invoked (by systemd --user) the
6919 $MANAGERPID env var is set to the PID of systemd.
6921 * SIGRTMIN+24 when sent to a --user instance will now result
6922 in immediate termination of systemd.
6924 * gatewayd received numerous feature additions such as a
6925 "follow" mode, for live syncing and filtering.
6927 * browse.html now allows filtering and showing detailed
6928 information on specific entries. Keyboard navigation and
6929 mouse screen support has been added.
6931 * gatewayd/journalctl now supports HTML5/JSON
6932 Server-Sent-Events as output.
6934 * The SysV init script compatibility logic will now
6935 heuristically determine whether a script supports the
6936 "reload" verb, and only then make this available as
6939 * "systemctl status --follow" has been removed, use "journalctl
6942 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6943 have been removed since they are hardly useful to be
6946 * And I'd like to take the opportunity to specifically mention
6947 Zbigniew for his great contributions. Zbigniew, you rock!
6949 Contributions from: Andrew Eikum, Christian Hesse, Colin
6950 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6951 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6952 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6953 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6954 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6955 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6959 * If /etc/vconsole.conf is non-existent or empty we will no
6960 longer load any console font or key map at boot by
6961 default. Instead the kernel defaults will be left
6962 intact. This is definitely the right thing to do, as no
6963 configuration should mean no configuration, and hard-coding
6964 font names that are different on all archs is probably a bad
6965 idea. Also, the kernel default key map and font should be
6966 good enough for most cases anyway, and mostly identical to
6967 the userspace fonts/key maps we previously overloaded them
6968 with. If distributions want to continue to default to a
6969 non-kernel font or key map they should ship a default
6970 /etc/vconsole.conf with the appropriate contents.
6972 Contributions from: Colin Walters, Daniel J Walsh, Dave
6973 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6974 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6978 * journalctl gained a new --cursor= switch to show entries
6979 starting from the specified location in the journal.
6981 * We now enforce a size limit on journal entry fields exported
6982 with "-o json" in journalctl. Fields larger than 4K will be
6983 assigned null. This can be turned off with --all.
6985 * An (optional) journal gateway daemon is now available as
6986 "systemd-journal-gatewayd.service". This service provides
6987 access to the journal via HTTP and JSON. This functionality
6988 will be used to implement live log synchronization in both
6989 pull and push modes, but has various other users too, such
6990 as easy log access for debugging of embedded devices. Right
6991 now it is already useful to retrieve the journal via HTTP:
6993 # systemctl start systemd-journal-gatewayd.service
6994 # wget http://localhost:19531/entries
6996 This will download the journal contents in a
6997 /var/log/messages compatible format. The same as JSON:
6999 # curl -H"Accept: application/json" http://localhost:19531/entries
7001 This service is also accessible via a web browser where a
7002 single static HTML5 app is served that uses the JSON logic
7003 to enable the user to do some basic browsing of the
7004 journal. This will be extended later on. Here's an example
7005 screenshot of this app in its current state:
7007 http://0pointer.de/public/journal-gatewayd
7009 Contributions from: Kay Sievers, Lennart Poettering, Robert
7010 Milasan, Tom Gundersen
7014 * The bash completion logic is now available for journalctl
7017 * We do not mount the "cpuset" controller anymore together with
7018 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7019 started if no parameters are assigned to it. "cpuset" hence
7020 broke code that assumed it could create "cpu" groups and
7023 * journalctl -f will now subscribe to terminal size changes,
7024 and line break accordingly.
7026 Contributions from: Dave Reisner, Kay Sievers, Lennart
7027 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7031 * nspawn will now create a symlink /etc/localtime in the
7032 container environment, copying the host's timezone
7033 setting. Previously this has been done via a bind mount, but
7034 since symlinks cannot be bind mounted this has now been
7035 changed to create/update the appropriate symlink.
7037 * journalctl -n's line number argument is now optional, and
7038 will default to 10 if omitted.
7040 * journald will now log the maximum size the journal files may
7041 take up on disk. This is particularly useful if the default
7042 built-in logic of determining this parameter from the file
7043 system size is used. Use "systemctl status
7044 systemd-journald.service" to see this information.
7046 * The multi-seat X wrapper tool has been stripped down. As X
7047 is now capable of enumerating graphics devices via udev in a
7048 seat-aware way the wrapper is not strictly necessary
7049 anymore. A stripped down temporary stop-gap is still shipped
7050 until the upstream display managers have been updated to
7051 fully support the new X logic. Expect this wrapper to be
7052 removed entirely in one of the next releases.
7054 * HandleSleepKey= in logind.conf has been split up into
7055 HandleSuspendKey= and HandleHibernateKey=. The old setting
7056 is not available anymore. X11 and the kernel are
7057 distinguishing between these keys and we should too. This
7058 also means the inhibition lock for these keys has been split
7061 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7062 Poettering, Lukas Nykryn, Václav Pavlín
7066 * Whenever a unit changes state we will now log this to the
7067 journal and show along the unit's own log output in
7070 * ConditionPathIsMountPoint= can now properly detect bind
7071 mount points too. (Previously, a bind mount of one file
7072 system to another place in the same file system could not be
7073 detected as mount, since they shared struct stat's st_dev
7076 * We will now mount the cgroup controllers cpu, cpuacct,
7077 cpuset and the controllers net_cls, net_prio together by
7080 * nspawn containers will now have a virtualized boot
7081 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7082 over with a randomized ID at container initialization). This
7083 has the effect of making "journalctl -b" do the right thing
7086 * The JSON output journal serialization has been updated not
7087 to generate "endless" list objects anymore, but rather one
7088 JSON object per line. This is more in line how most JSON
7089 parsers expect JSON objects. The new output mode
7090 "json-pretty" has been added to provide similar output, but
7091 neatly aligned for readability by humans.
7093 * We dropped all explicit sync() invocations in the shutdown
7094 code. The kernel does this implicitly anyway in the kernel
7095 reboot() syscall. halt(8)'s -n option is now a compatibility
7098 * We now support virtualized reboot() in containers, as
7099 supported by newer kernels. We will fall back to exit() if
7100 CAP_SYS_REBOOT is not available to the container. Also,
7101 nspawn makes use of this now and will actually reboot the
7102 container if the containerized OS asks for that.
7104 * journalctl will only show local log output by default
7105 now. Use --merge (-m) to show remote log output, too.
7107 * libsystemd-journal gained the new sd_journal_get_usage()
7108 call to determine the current disk usage of all journal
7109 files. This is exposed in the new "journalctl --disk-usage"
7112 * journald gained a new configuration setting SplitMode= in
7113 journald.conf which may be used to control how user journals
7114 are split off. See journald.conf(5) for details.
7116 * A new condition type ConditionFileNotEmpty= has been added.
7118 * tmpfiles' "w" lines now support file globbing, to write
7119 multiple files at once.
7121 * We added Python bindings for the journal submission
7122 APIs. More Python APIs for a number of selected APIs will
7123 likely follow. Note that we intend to add native bindings
7124 only for the Python language, as we consider it common
7125 enough to deserve bindings shipped within systemd. There are
7126 various projects outside of systemd that provide bindings
7127 for languages such as PHP or Lua.
7129 * Many conditions will now resolve specifiers such as %i. In
7130 addition, PathChanged= and related directives of .path units
7131 now support specifiers as well.
7133 * There's now a new RPM macro definition for the system preset
7136 * journald will now warn if it ca not forward a message to the
7137 syslog daemon because its socket is full.
7139 * timedated will no longer write or process /etc/timezone,
7140 except on Debian. As we do not support late mounted /usr
7141 anymore /etc/localtime always being a symlink is now safe,
7142 and hence the information in /etc/timezone is not necessary
7145 * logind will now always reserve one VT for a text getty (VT6
7146 by default). Previously if more than 6 X sessions where
7147 started they took up all the VTs with auto-spawned gettys,
7148 so that no text gettys were available anymore.
7150 * udev will now automatically inform the btrfs kernel logic
7151 about btrfs RAID components showing up. This should make
7152 simple hotplug based btrfs RAID assembly work.
7154 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7155 (but not for its children which will stay at the kernel
7156 default). This should allow setups with a lot more listening
7159 * systemd will now always pass the configured timezone to the
7160 kernel at boot. timedated will do the same when the timezone
7163 * logind's inhibition logic has been updated. By default,
7164 logind will now handle the lid switch, the power and sleep
7165 keys all the time, even in graphical sessions. If DEs want
7166 to handle these events on their own they should take the new
7167 handle-power-key, handle-sleep-key and handle-lid-switch
7168 inhibitors during their runtime. A simple way to achieve
7169 that is to invoke the DE wrapped in an invocation of:
7171 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7173 * Access to unit operations is now checked via SELinux taking
7174 the unit file label and client process label into account.
7176 * systemd will now notify the administrator in the journal
7177 when he over-mounts a non-empty directory.
7179 * There are new specifiers that are resolved in unit files,
7180 for the host name (%H), the machine ID (%m) and the boot ID
7183 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7184 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7185 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7186 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7187 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7188 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7189 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7193 * Support for reading structured kernel messages from
7194 /dev/kmsg has now been added and is enabled by default.
7196 * Support for reading kernel messages from /proc/kmsg has now
7197 been removed. If you want kernel messages in the journal
7198 make sure to run a recent kernel (>= 3.5) that supports
7199 reading structured messages from /dev/kmsg (see
7200 above). /proc/kmsg is now exclusive property of classic
7201 syslog daemons again.
7203 * The libudev API gained the new
7204 udev_device_new_from_device_id() call.
7206 * The logic for file system namespace (ReadOnlyDirectory=,
7207 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7208 require pivot_root() anymore. This means fewer temporary
7209 directories are created below /tmp for this feature.
7211 * nspawn containers will now see and receive all submounts
7212 made on the host OS below the root file system of the
7215 * Forward Secure Sealing is now supported for Journal files,
7216 which provide cryptographical sealing of journal files so
7217 that attackers cannot alter log history anymore without this
7218 being detectable. Lennart will soon post a blog story about
7219 this explaining it in more detail.
7221 * There are two new service settings RestartPreventExitStatus=
7222 and SuccessExitStatus= which allow configuration of exit
7223 status (exit code or signal) which will be excepted from the
7224 restart logic, resp. consider successful.
7226 * journalctl gained the new --verify switch that can be used
7227 to check the integrity of the structure of journal files and
7228 (if Forward Secure Sealing is enabled) the contents of
7231 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7232 and similar symlinks pre-created. This makes running shells
7233 as container init process a lot more fun.
7235 * The fstab support can now handle PARTUUID= and PARTLABEL=
7238 * A new ConditionHost= condition has been added to match
7239 against the hostname (with globs) and machine ID. This is
7240 useful for clusters where a single OS image is used to
7241 provision a large number of hosts which shall run slightly
7242 different sets of services.
7244 * Services which hit the restart limit will now be placed in a
7247 Contributions from: Bertram Poettering, Dave Reisner, Huang
7248 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7249 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7253 * When running in --user mode systemd will now become a
7254 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7255 tree a lot more organized.
7257 * A new PartOf= unit dependency type has been introduced that
7258 may be used to group services in a natural way.
7260 * "systemctl enable" may now be used to enable instances of
7263 * journalctl now prints error log levels in red, and
7264 warning/notice log levels in bright white. It also supports
7265 filtering by log level now.
7267 * cgtop gained a new -n switch (similar to top), to configure
7268 the maximum number of iterations to run for. It also gained
7269 -b, to run in batch mode (accepting no input).
7271 * The suffix ".service" may now be omitted on most systemctl
7272 command lines involving service unit names.
7274 * There's a new bus call in logind to lock all sessions, as
7275 well as a loginctl verb for it "lock-sessions".
7277 * libsystemd-logind.so gained a new call sd_journal_perror()
7278 that works similar to libc perror() but logs to the journal
7279 and encodes structured information about the error number.
7281 * /etc/crypttab entries now understand the new keyfile-size=
7284 * shutdown(8) now can send a (configurable) wall message when
7285 a shutdown is cancelled.
7287 * The mount propagation mode for the root file system will now
7288 default to "shared", which is useful to make containers work
7289 nicely out-of-the-box so that they receive new mounts from
7290 the host. This can be undone locally by running "mount
7291 --make-rprivate /" if needed.
7293 * The prefdm.service file has been removed. Distributions
7294 should maintain this unit downstream if they intend to keep
7295 it around. However, we recommend writing normal unit files
7296 for display managers instead.
7298 * Since systemd is a crucial part of the OS we will now
7299 default to a number of compiler switches that improve
7300 security (hardening) such as read-only relocations, stack
7301 protection, and suchlike.
7303 * The TimeoutSec= setting for services is now split into
7304 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7305 of individual time outs for the start and the stop phase of
7308 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7309 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7310 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7311 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7312 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7313 Gundersen, Zbigniew Jędrzejewski-Szmek
7317 * The journal and id128 C APIs are now fully documented as man
7320 * Extra safety checks have been added when transitioning from
7321 the initial RAM disk to the main system to avoid accidental
7324 * /etc/crypttab entries now understand the new keyfile-offset=
7327 * systemctl -t can now be used to filter by unit load state.
7329 * The journal C API gained the new sd_journal_wait() call to
7330 make writing synchronous journal clients easier.
7332 * journalctl gained the new -D switch to show journals from a
7335 * journalctl now displays a special marker between log
7336 messages of two different boots.
7338 * The journal is now explicitly flushed to /var via a service
7339 systemd-journal-flush.service, rather than implicitly simply
7340 by seeing /var/log/journal to be writable.
7342 * journalctl (and the journal C APIs) can now match for much
7343 more complex expressions, with alternatives and
7346 * When transitioning from the initial RAM disk to the main
7347 system we will now kill all processes in a killing spree to
7348 ensure no processes stay around by accident.
7350 * Three new specifiers may be used in unit files: %u, %h, %s
7351 resolve to the user name, user home directory resp. user
7352 shell. This is useful for running systemd user instances.
7354 * We now automatically rotate journal files if their data
7355 object hash table gets a fill level > 75%. We also size the
7356 hash table based on the configured maximum file size. This
7357 together should lower hash collisions drastically and thus
7358 speed things up a bit.
7360 * journalctl gained the new "--header" switch to introspect
7361 header data of journal files.
7363 * A new setting SystemCallFilters= has been added to services
7364 which may be used to apply blacklists or whitelists to
7365 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7367 * nspawn gained a new --link-journal= switch (and quicker: -j)
7368 to link the container journal with the host. This makes it
7369 very easy to centralize log viewing on the host for all
7370 guests while still keeping the journal files separated.
7372 * Many bugfixes and optimizations
7374 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7375 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7376 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7381 * Several tools now understand kernel command line arguments,
7382 which are only read when run in an initial RAM disk. They
7383 usually follow closely their normal counterparts, but are
7386 * There's a new tool to analyze the readahead files that are
7387 automatically generated at boot. Use:
7389 /usr/lib/systemd/systemd-readahead analyze /.readahead
7391 * We now provide an early debug shell on tty9 if this enabled. Use:
7393 systemctl enable debug-shell.service
7395 * All plymouth related units have been moved into the Plymouth
7396 package. Please make sure to upgrade your Plymouth version
7399 * systemd-tmpfiles now supports getting passed the basename of
7400 a configuration file only, in which case it will look for it
7401 in all appropriate directories automatically.
7403 * udevadm info now takes a /dev or /sys path as argument, and
7404 does the right thing. Example:
7406 udevadm info /dev/sda
7407 udevadm info /sys/class/block/sda
7409 * systemctl now prints a warning if a unit is stopped but a
7410 unit that might trigger it continues to run. Example: a
7411 service is stopped but the socket that activates it is left
7414 * "systemctl status" will now mention if the log output was
7415 shortened due to rotation since a service has been started.
7417 * The journal API now exposes functions to determine the
7418 "cutoff" times due to rotation.
7420 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7421 immediately flushing of runtime logs to /var if possible,
7422 resp. for triggering immediate rotation of the journal
7425 * It is now considered an error if a service is attempted to
7426 be stopped that is not loaded.
7428 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7430 * systemd-analyze now supports Python 3
7432 * tmpfiles now supports cleaning up directories via aging
7433 where the first level dirs are always kept around but
7434 directories beneath it automatically aged. This is enabled
7435 by prefixing the age field with '~'.
7437 * Seat objects now expose CanGraphical, CanTTY properties
7438 which is required to deal with very fast bootups where the
7439 display manager might be running before the graphics drivers
7440 completed initialization.
7442 * Seat objects now expose a State property.
7444 * We now include RPM macros for service enabling/disabling
7445 based on the preset logic. We recommend RPM based
7446 distributions to make use of these macros if possible. This
7447 makes it simpler to reuse RPM spec files across
7450 * We now make sure that the collected systemd unit name is
7451 always valid when services log to the journal via
7454 * There's a new man page kernel-command-line(7) detailing all
7455 command line options we understand.
7457 * The fstab generator may now be disabled at boot by passing
7458 fstab=0 on the kernel command line.
7460 * A new kernel command line option modules-load= is now understood
7461 to load a specific kernel module statically, early at boot.
7463 * Unit names specified on the systemctl command line are now
7464 automatically escaped as needed. Also, if file system or
7465 device paths are specified they are automatically turned
7466 into the appropriate mount or device unit names. Example:
7468 systemctl status /home
7469 systemctl status /dev/sda
7471 * The SysVConsole= configuration option has been removed from
7472 system.conf parsing.
7474 * The SysV search path is no longer exported on the D-Bus
7477 * The Names= option has been removed from unit file parsing.
7479 * There's a new man page bootup(7) detailing the boot process.
7481 * Every unit and every generator we ship with systemd now
7482 comes with full documentation. The self-explanatory boot is
7485 * A couple of services gained "systemd-" prefixes in their
7486 name if they wrap systemd code, rather than only external
7487 code. Among them fsck@.service which is now
7488 systemd-fsck@.service.
7490 * The HaveWatchdog property has been removed from the D-Bus
7493 * systemd.confirm_spawn= on the kernel command line should now
7496 * There's a new man page crypttab(5) which details all options
7497 we actually understand.
7499 * systemd-nspawn gained a new --capability= switch to pass
7500 additional capabilities to the container.
7502 * timedated will now read known NTP implementation unit names
7503 from /usr/lib/systemd/ntp-units.d/*.list,
7504 systemd-timedated-ntp.target has been removed.
7506 * journalctl gained a new switch "-b" that lists log data of
7507 the current boot only.
7509 * The notify socket is in the abstract namespace again, in
7510 order to support daemons which chroot() at start-up.
7512 * There is a new Storage= configuration option for journald
7513 which allows configuration of where log data should go. This
7514 also provides a way to disable journal logging entirely, so
7515 that data collected is only forwarded to the console, the
7516 kernel log buffer or another syslog implementation.
7518 * Many bugfixes and optimizations
7520 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7521 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7522 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7523 Shawn Landden, Tom Gundersen
7527 * "systemctl help <unit>" now shows the man page if one is
7530 * Several new man pages have been added.
7532 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7533 MaxLevelConsole= can now be specified in
7534 journald.conf. These options allow reducing the amount of
7535 data stored on disk or forwarded by the log level.
7537 * TimerSlackNSec= can now be specified in system.conf for
7538 PID1. This allows system-wide power savings.
7540 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7541 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7546 * logind is now capable of (optionally) handling power and
7547 sleep keys as well as the lid switch.
7549 * journalctl now understands the syntax "journalctl
7550 /usr/bin/avahi-daemon" to get all log output of a specific
7553 * CapabilityBoundingSet= in system.conf now also influences
7554 the capability bound set of usermode helpers of the kernel.
7556 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7557 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7558 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7559 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7563 * Note that we skipped 139 releases here in order to set the
7564 new version to something that is greater than both udev's
7565 and systemd's most recent version number.
7567 * udev: all udev sources are merged into the systemd source tree now.
7568 All future udev development will happen in the systemd tree. It
7569 is still fully supported to use the udev daemon and tools without
7570 systemd running, like in initramfs or other init systems. Building
7571 udev though, will require the *build* of the systemd tree, but
7572 udev can be properly *run* without systemd.
7574 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7575 should be used to create dead device nodes as workarounds for broken
7578 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7579 no longer supported. udev_monitor_new_from_netlink() needs to be
7580 used to subscribe to events.
7582 * udev: when udevd is started by systemd, processes which are left
7583 behind by forking them off of udev rules, are unconditionally cleaned
7584 up and killed now after the event handling has finished. Services or
7585 daemons must be started as systemd services. Services can be
7586 pulled-in by udev to get started, but they can no longer be directly
7587 forked by udev rules.
7589 * udev: the daemon binary is called systemd-udevd now and installed
7590 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7591 to adapt to that, create symlink, or rename the binary after building
7594 * libudev no longer provides these symbols:
7595 udev_monitor_from_socket()
7596 udev_queue_get_failed_list_entry()
7597 udev_get_{dev,sys,run}_path()
7598 The versions number was bumped and symbol versioning introduced.
7600 * systemd-loginctl and systemd-journalctl have been renamed
7601 to loginctl and journalctl to match systemctl.
7603 * The config files: /etc/systemd/systemd-logind.conf and
7604 /etc/systemd/systemd-journald.conf have been renamed to
7605 logind.conf and journald.conf. Package updates should rename
7606 the files to the new names on upgrade.
7608 * For almost all files the license is now LGPL2.1+, changed
7609 from the previous GPL2.0+. Exceptions are some minor stuff
7610 of udev (which will be changed to LGPL2.1 eventually, too),
7611 and the MIT licensed sd-daemon.[ch] library that is suitable
7612 to be used as drop-in files.
7614 * systemd and logind now handle system sleep states, in
7615 particular suspending and hibernating.
7617 * logind now implements a sleep/shutdown/idle inhibiting logic
7618 suitable for a variety of uses. Soonishly Lennart will blog
7619 about this in more detail.
7621 * var-run.mount and var-lock.mount are no longer provided
7622 (which previously bind mounted these directories to their new
7623 places). Distributions which have not converted these
7624 directories to symlinks should consider stealing these files
7625 from git history and add them downstream.
7627 * We introduced the Documentation= field for units and added
7628 this to all our shipped units. This is useful to make it
7629 easier to explore the boot and the purpose of the various
7632 * All smaller setup units (such as
7633 systemd-vconsole-setup.service) now detect properly if they
7634 are run in a container and are skipped when
7635 appropriate. This guarantees an entirely noise-free boot in
7636 Linux container environments such as systemd-nspawn.
7638 * A framework for implementing offline system updates is now
7639 integrated, for details see:
7640 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7642 * A new service type Type=idle is available now which helps us
7643 avoiding ugly interleaving of getty output and boot status
7646 * There's now a system-wide CapabilityBoundingSet= option to
7647 globally reduce the set of capabilities for the
7648 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7649 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7650 even CAP_NET_ADMIN system-wide for secure systems.
7652 * There are now system-wide DefaultLimitXXX= options to
7653 globally change the defaults of the various resource limits
7654 for all units started by PID 1.
7656 * Harald Hoyer's systemd test suite has been integrated into
7657 systemd which allows easy testing of systemd builds in qemu
7658 and nspawn. (This is really awesome! Ask us for details!)
7660 * The fstab parser is now implemented as generator, not inside
7663 * systemctl will now warn you if .mount units generated from
7664 /etc/fstab are out of date due to changes in fstab that
7665 have not been read by systemd yet.
7667 * systemd is now suitable for usage in initrds. Dracut has
7668 already been updated to make use of this. With this in place
7669 initrds get a slight bit faster but primarily are much
7670 easier to introspect and debug since "systemctl status" in
7671 the host system can be used to introspect initrd services,
7672 and the journal from the initrd is kept around too.
7674 * systemd-delta has been added, a tool to explore differences
7675 between user/admin configuration and vendor defaults.
7677 * PrivateTmp= now affects both /tmp and /var/tmp.
7679 * Boot time status messages are now much prettier and feature
7680 proper english language. Booting up systemd has never been
7683 * Read-ahead pack files now include the inode number of all
7684 files to pre-cache. When the inode changes the pre-caching
7685 is not attempted. This should be nicer to deal with updated
7686 packages which might result in changes of read-ahead
7689 * We now temporaritly lower the kernel's read_ahead_kb variable
7690 when collecting read-ahead data to ensure the kernel's
7691 built-in read-ahead does not add noise to our measurements
7692 of necessary blocks to pre-cache.
7694 * There's now RequiresMountsFor= to add automatic dependencies
7695 for all mounts necessary for a specific file system path.
7697 * MountAuto= and SwapAuto= have been removed from
7698 system.conf. Mounting file systems at boot has to take place
7701 * nspawn now learned a new switch --uuid= to set the machine
7702 ID on the command line.
7704 * nspawn now learned the -b switch to automatically search
7707 * vt102 is now the default TERM for serial TTYs, upgraded from
7710 * systemd-logind now works on VT-less systems.
7712 * The build tree has been reorganized. The individual
7713 components now have directories of their own.
7715 * A new condition type ConditionPathIsReadWrite= is now available.
7717 * nspawn learned the new -C switch to create cgroups for the
7718 container in other hierarchies.
7720 * We now have support for hardware watchdogs, configurable in
7723 * The scheduled shutdown logic now has a public API.
7725 * We now mount /tmp as tmpfs by default, but this can be
7726 masked and /etc/fstab can override it.
7728 * Since udisks does not make use of /media anymore we are not
7729 mounting a tmpfs on it anymore.
7731 * journalctl gained a new --local switch to only interleave
7732 locally generated journal files.
7734 * We can now load the IMA policy at boot automatically.
7736 * The GTK tools have been split off into a systemd-ui.
7738 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7739 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7740 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7741 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7742 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7743 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7744 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7745 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7746 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7751 * This is mostly a bugfix release
7753 * Support optional initialization of the machine ID from the
7754 KVM or container configured UUID.
7756 * Support immediate reboots with "systemctl reboot -ff"
7758 * Show /etc/os-release data in systemd-analyze output
7760 * Many bugfixes for the journal, including endianness fixes and
7761 ensuring that disk space enforcement works
7763 * sd-login.h is C++ compatible again
7765 * Extend the /etc/os-release format on request of the Debian
7768 * We now refuse non-UTF8 strings used in various configuration
7769 and unit files. This is done to ensure we do not pass invalid
7770 data over D-Bus or expose it elsewhere.
7772 * Register Mimo USB Screens as suitable for automatic seat
7775 * Read SELinux client context from journal clients in a race
7778 * Reorder configuration file lookup order. /etc now always
7779 overrides /run in order to allow the administrator to always
7780 and unconditionally override vendor-supplied or
7781 automatically generated data.
7783 * The various user visible bits of the journal now have man
7784 pages. We still lack man pages for the journal API calls
7787 * We now ship all man pages in HTML format again in the
7790 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7791 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7792 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7797 * This is mostly a bugfix release
7799 * systems lacking /etc/os-release are no longer supported.
7801 * Various functionality updates to libsystemd-login.so
7803 * Track class of PAM logins to distinguish greeters from
7806 Contributions from: Kay Sievers, Lennart Poettering, Michael
7811 * This is an important bugfix release for v41.
7813 * Building man pages is now optional which should be useful
7814 for those building systemd from git but unwilling to install
7817 * Watchdog support for supervising services is now usable. In
7818 a future release support for hardware watchdogs
7819 (i.e. /dev/watchdog) will be added building on this.
7821 * Service start rate limiting is now configurable and can be
7822 turned off per service. When a start rate limit is hit a
7823 reboot can automatically be triggered.
7825 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7827 Contributions from: Benjamin Franzke, Bill Nottingham,
7828 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7829 Schmidt, Michał Górny, Piotr Drąg
7833 * The systemd binary is installed /usr/lib/systemd/systemd now;
7834 An existing /sbin/init symlink needs to be adapted with the
7837 * The code that loads kernel modules has been ported to invoke
7838 libkmod directly, instead of modprobe. This means we do not
7839 support systems with module-init-tools anymore.
7841 * Watchdog support is now already useful, but still not
7844 * A new kernel command line option systemd.setenv= is
7845 understood to set system wide environment variables
7846 dynamically at boot.
7848 * We now limit the set of capabilities of systemd-journald.
7850 * We now set SIGPIPE to ignore by default, since it only is
7851 useful in shell pipelines, and has little use in general
7852 code. This can be disabled with IgnoreSIPIPE=no in unit
7855 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7856 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7861 * This is mostly a bugfix release
7863 * We now expose the reason why a service failed in the
7864 "Result" D-Bus property.
7866 * Rudimentary service watchdog support (will be completed over
7867 the next few releases.)
7869 * When systemd forks off in order execute some service we will
7870 now immediately changes its argv[0] to reflect which process
7871 it will execute. This is useful to minimize the time window
7872 with a generic argv[0], which makes bootcharts more useful
7874 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7875 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7876 Mike Kazantsev, Ray Strode
7880 * This is mostly a test release, but incorporates many
7883 * New systemd-cgtop tool to show control groups by their
7886 * Linking against libacl for ACLs is optional again. If
7887 disabled, support tracking device access for active logins
7888 goes becomes unavailable, and so does access to the user
7889 journals by the respective users.
7891 * If a group "adm" exists, journal files are automatically
7892 owned by them, thus allow members of this group full access
7893 to the system journal as well as all user journals.
7895 * The journal now stores the SELinux context of the logging
7896 client for all entries.
7898 * Add C++ inclusion guards to all public headers
7900 * New output mode "cat" in the journal to print only text
7901 messages, without any meta data like date or time.
7903 * Include tiny X server wrapper as a temporary stop-gap to
7904 teach XOrg udev display enumeration. This is used by display
7905 managers such as gdm, and will go away as soon as XOrg
7906 learned native udev hotplugging for display devices.
7908 * Add new systemd-cat tool for executing arbitrary programs
7909 with STDERR/STDOUT connected to the journal. Can also act as
7910 BSD logger replacement, and does so by default.
7912 * Optionally store all locally generated coredumps in the
7913 journal along with meta data.
7915 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7916 writing short strings to files (for usage for /sys), and for
7917 creating symlinks, character and block device nodes.
7919 * New unit file option ControlGroupPersistent= to make cgroups
7920 persistent, following the mechanisms outlined in
7921 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7923 * Support multiple local RTCs in a sane way
7925 * No longer monopolize IO when replaying readahead data on
7926 rotating disks, since we might starve non-file-system IO to
7927 death, since fanotify() will not see accesses done by blkid,
7930 * Do not show kernel threads in systemd-cgls anymore, unless
7931 requested with new -k switch.
7933 Contributions from: Dan Horák, Kay Sievers, Lennart
7934 Poettering, Michal Schmidt
7938 * This is mostly a test release, but incorporates many
7941 * The git repository moved to:
7942 git://anongit.freedesktop.org/systemd/systemd
7943 ssh://git.freedesktop.org/git/systemd/systemd
7945 * First release with the journal
7946 http://0pointer.de/blog/projects/the-journal.html
7948 * The journal replaces both systemd-kmsg-syslogd and
7949 systemd-stdout-bridge.
7951 * New sd_pid_get_unit() API call in libsystemd-logind
7953 * Many systemadm clean-ups
7955 * Introduce remote-fs-pre.target which is ordered before all
7956 remote mounts and may be used to start services before all
7959 * Added Mageia support
7961 * Add bash completion for systemd-loginctl
7963 * Actively monitor PID file creation for daemons which exit in
7964 the parent process before having finished writing the PID
7965 file in the daemon process. Daemons which do this need to be
7966 fixed (i.e. PID file creation must have finished before the
7967 parent exits), but we now react a bit more gracefully to them.
7969 * Add colourful boot output, mimicking the well-known output
7970 of existing distributions.
7972 * New option PassCredentials= for socket units, for
7973 compatibility with a recent kernel ABI breakage.
7975 * /etc/rc.local is now hooked in via a generator binary, and
7976 thus will no longer act as synchronization point during
7979 * systemctl list-unit-files now supports --root=.
7981 * systemd-tmpfiles now understands two new commands: z, Z for
7982 relabelling files according to the SELinux database. This is
7983 useful to apply SELinux labels to specific files in /sys,
7986 * Output of SysV services is now forwarded to both the console
7987 and the journal by default, not only just the console.
7989 * New man pages for all APIs from libsystemd-login.
7991 * The build tree got reorganized and the build system is a
7992 lot more modular allowing embedded setups to specifically
7993 select the components of systemd they are interested in.
7995 * Support for Linux systems lacking the kernel VT subsystem is
7998 * configure's --with-rootdir= got renamed to
7999 --with-rootprefix= to follow the naming used by udev and
8002 * Unless specified otherwise we will now install to /usr instead
8003 of /usr/local by default.
8005 * Processes with '@' in argv[0][0] are now excluded from the
8006 final shut-down killing spree, following the logic explained
8008 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8010 * All processes remaining in a service cgroup when we enter
8011 the START or START_PRE states are now killed with
8012 SIGKILL. That means it is no longer possible to spawn
8013 background processes from ExecStart= lines (which was never
8014 supported anyway, and bad style).
8016 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8017 reloading of units together.
8019 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8020 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8021 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8022 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8023 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob