1 systemd System and Service Manager
3 CHANGES WITH 236 in spe:
5 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
6 in v235 has been extended to also set the dummy.ko module option
7 numdummies=0, preventing the kernel from automatically creating
8 dummy0. All dummy interfaces must now be explicitly created.
10 * Unknown '%' specifiers in configuration files are now rejected. This
11 applies to units and tmpfiles.d configuration. Any percent characters
12 that are followed by a letter or digit that are not supposed to be
13 interpreted as the beginning of a specifier should be escaped by
14 doubling ("%%"). (So "size=5%" is still accepted, as well as
15 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
16 valid specifiers today.)
18 * systemd-resolved now maintains a new dynamic
19 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
20 recommended to make /etc/resolv.conf a symlink to it. This file
21 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
22 includes dynamically acquired search domains, achieving more correct
23 DNS resolution by software that bypasses local DNS APIs such as NSS.
25 * The "uaccess" udev tag has been dropped from /dev/kvm and
26 /dev/dri/renderD*. These devices now have the 0666 permissions by
27 default (but this may be changed at build-time). /dev/dri/renderD*
28 will now be owned by the "render" group along with /dev/kfd.
30 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
31 systemd-journal-gatewayd.service and
32 systemd-journal-upload.service. This means "nss-systemd" must be
33 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
34 services are resolved properly.
36 * In /etc/fstab two new mount options are now understood:
37 x-systemd.makefs and x-systemd.growfs. The former has the effect that
38 the configured file system is formatted before it is mounted, the
39 latter that the file system is resized to the full block device size
40 after it is mounted (i.e. if the file system is smaller than the
41 partition it resides on, it's grown). This is similar to the fsck
42 logic in /etc/fstab, and pulls in systemd-makefs@.service and
43 systemd-growfs@.service as necessary, similar to
44 systemd-fsck@.service. Resizing is currently only supported on ext4
47 * In systemd-networkd, the IPv6 RA logic now optionally may announce
48 DNS server and domain information.
50 * Support for the LUKS2 on-disk format for encrypted partitions has
51 been added. This requires libcryptsetup2 during compilation and
54 * The systemd --user instance will now signal "readiness" when its
55 basic.target unit has been reached, instead of when the run queue ran
56 empty for the first time.
58 * Tmpfiles.d with user configuration are now also supported.
59 systemd-tmpfiles gained a new --user switch, and snippets placed in
60 ~/.config/user-tmpfiles.d/ and corresponding directories will be
61 executed by systemd-tmpfiles --user running in the new
62 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
63 running in the user session.
65 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
66 %S resolves to the top-level state directory (/var/lib for the system
67 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
68 top-level cache directory (/var/cache for the system instance,
69 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
70 logs directory (/var/log for the system instance,
71 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
72 existing %t specifier, that resolves to the top-level runtime
73 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
76 * journalctl learnt a new parameter --output-fields= for limiting the
77 set of journal fields to output in verbose and JSON output modes.
79 * systemd-timesyncd's configuration file gained a new option
80 RootDistanceMaxSec= for setting the maximum root distance of servers
81 it'll use, as well as the new options PollIntervalMinSec= and
82 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
84 * bootctl gained a new command "list" for listing all available boot
85 menu items on systems that follow the boot loader specification.
87 * systemctl gained a new --dry-run switch that shows what would be done
88 instead of doing it, and is currently supported by the shutdown and
91 * ConditionSecurity= can now detect the TOMOYO security module.
93 * Unit file [Install] sections are now also respected in unit drop-in
94 files. This is intended to be used by drop-ins under /usr/lib/.
96 * systemd-firstboot may now also set the initial keyboard mapping.
98 * Udev "changed" events for devices which are exposed as systemd
99 .device units are now propagated to units specified in
100 ReloadPropagatedFrom= as reload requests.
102 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
103 unit template name (i.e. a name in the form of 'foobar@.service',
104 without the instance component between the '@' and - the '.'), then
105 the escaped sysfs path of the device is automatically used as the
108 * SystemCallFilter= in unit files has been extended so that an "errno"
109 can be specified individually for each system call. Example:
110 SystemCallFilter=~uname:EILSEQ.
112 * The cgroup delegation logic has been substantially updated. Delegate=
113 now optionally takes a list of controllers (instead of a boolean, as
114 before), which lists the controllers to delegate at least.
116 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
118 * A new LogLevelMax= setting configures the maximum log level any
119 process of the service may log at (i.e. anything with a lesser
120 priority than what is specified is automatically dropped). A new
121 LogExtraFields= setting allows configuration of additional journal
122 fields to attach to all log records generated by any of the unit's
125 * New StandardInputData= and StandardInputText= settings along with the
126 new option StandardInput=data may be used to configure textual or
127 binary data that shall be passed to the executed service process via
128 standard input, encoded in-line in the unit file.
130 * StandardInput=, StandardOutput= and StandardError= may now be used to
131 connect stdin/stdout/stderr of executed processes directly with a
132 file or AF_UNIX socket in the file system, using the new "file:" option.
134 * A new unit file option CollectMode= has been added, that allows
135 tweaking the garbage collection logic for units. It may be used to
136 tell systemd to garbage collect units that have failed automatically
137 (normally it only GCs units that exited successfully). systemd-run
138 and systemd-mount expose this new functionality with a new -G option.
140 * "machinectl bind" may now be used to bind mount non-directories
141 (i.e. regularfiles, devices, fifos, sockets).
143 * systemd-analyze gained a new verb "calendar" for validating and
144 testing calendar time specifications to use for OnCalendar= in timer
145 units. Besides validating the expression it will calculate the next
146 time the specified expression would elapse.
148 * In addition to the pre-existing FailureAction= unit file setting
149 there's now SuccessAction=, for configuring a shutdown action to
150 execute when a unit completes successfully. This is useful in
151 particular inside containers that shall terminate after some workload
152 has been completed. Also, both options are now supported for all unit
153 types, not just services.
155 * networkds's IP rule support gained two new options
156 IncomingInterface= and OutgoingInterface= for configuring the incoming
157 and outgoing interfaces of configured rules. systemd-networkd also
158 gained support for "vxcan" network devices.
160 * networkd gained a new setting RequiredForOnline=, taking a
161 boolean. If set, systemd-wait-online will take it into consideration
162 when determining that the system is up, otherwise it will ignore the
163 interface for this purpose.
165 * The sd_notify() protocol gained support for a new operation: with
166 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
167 store again, ahead of POLLHUP or POLLERR when they are removed
170 * A new document UIDS-GIDS.md has been added to the source tree, that
171 documents the UID/GID range and assignment assumptions and
172 requirements of systemd.
174 * The watchdog device PID 1 will ping may now be configured through the
175 WatchdogDevice= configuration file setting, or by setting the
176 systemd.watchdog_service= kernel commandline option.
178 * systemd-resolved's gained support for registering DNS-SD services on
179 the local network using MulticastDNS. Services may either be
180 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
181 the same dir below /run, /usr/lib), or through its D-Bus API.
183 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
184 extend the effective start, runtime, and stop time. The service must
185 continue to send EXTEND_TIMEOUT_USEC within the period specified to
186 prevent the service manager from making the service as timedout.
188 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
189 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
190 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
191 Eggert, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri John
192 Ledkov, Dmitry Rozhkov, Edward A. James, Evgeny Vereshchagin, Florian
193 Klink, Franck Bui, Gwendal Grignou, Hans de Goede, Harald Hoyer, Hristo
194 Venev, Ikey Doherty, Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin,
195 José Bollo, Josef Andersson, juga0, Krzysztof Nowicki, Lars Karlitski,
196 Lars Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir
197 Rintel, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
198 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
199 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala, Max Resch,
200 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Mike
201 Gilbert, Muhammet Kara, Neil Brown, Olaf Hering, Ondrej Kozina, Patrik
202 Flykt, Patryk Kocielnik, Peter Hutterer, Piotr Drąg, Razvan Cojocaru,
203 Robin McCorkell, Roland Hieber, Sergey Ptashnick, Shawn Landden, Shuang
204 Liu, Simon Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner,
205 Susant Sahani, Sylvain Plantefève, Thomas Blume, Tinu Weber, Tom
206 Stellard, Topi Miettinen, Torsten Hilbrich, Vito Caputo, Vladislav
207 Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Zeal
214 * INCOMPATIBILITY: systemd-logind.service and other long-running
215 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
216 communication with the outside. This generally improves security of
217 the system, and is in almost all cases a safe and good choice, as
218 these services do not and should not provide any network-facing
219 functionality. However, systemd-logind uses the glibc NSS API to
220 query the user database. This creates problems on systems where NSS
221 is set up to directly consult network services for user database
222 lookups. In particular, this creates incompatibilities with the
223 "nss-nis" module, which attempts to directly contact the NIS/YP
224 network servers it is configured for, and will now consistently
225 fail. In such cases, it is possible to turn off IP sandboxing for
226 systemd-logind.service (set IPAddressDeny= in its [Service] section
227 to the empty string, via a .d/ unit file drop-in). Downstream
228 distributions might want to update their nss-nis packaging to include
229 such a drop-in snippet, accordingly, to hide this incompatibility
230 from the user. Another option is to make use of glibc's nscd service
231 to proxy such network requests through a privilege-separated, minimal
232 local caching daemon, or to switch to more modern technologies such
233 sssd, whose NSS hook-ups generally do not involve direct network
234 access. In general, we think it's definitely time to question the
235 implementation choices of nss-nis, i.e. whether it's a good idea
236 today to embed a network-facing loadable module into all local
237 processes that need to query the user database, including the most
238 trivial and benign ones, such as "ls". For more details about
239 IPAddressDeny= see below.
241 * A new modprobe.d drop-in is now shipped by default that sets the
242 bonding module option max_bonds=0. This overrides the kernel default,
243 to avoid conflicts and ambiguity as to whether or not bond0 should be
244 managed by systemd-networkd or not. This resolves multiple issues
245 with bond0 properties not being applied, when bond0 is configured
246 with systemd-networkd. Distributors may choose to not package this,
247 however in that case users will be prevented from correctly managing
248 bond0 interface using systemd-networkd.
250 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
251 which print the logging level and target of the system manager. They
252 complement the existing "set-log-level" and "set-log-target" verbs
253 used to change those values.
255 * journald.conf gained a new boolean setting ReadKMsg= which defaults
256 to on. If turned off kernel log messages will not be read by
257 systemd-journald or included in the logs. It also gained a new
258 setting LineMax= for configuring the maximum line length in
259 STDOUT/STDERR log streams. The new default for this value is 48K, up
260 from the previous hardcoded 2048.
262 * A new unit setting RuntimeDirectoryPreserve= has been added, which
263 allows more detailed control of what to do with a runtime directory
264 configured with RuntimeDirectory= (i.e. a directory below /run or
265 $XDG_RUNTIME_DIR) after a unit is stopped.
267 * The RuntimeDirectory= setting for units gained support for creating
268 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
269 one top-level directory.
271 * Units gained new options StateDirectory=, CacheDirectory=,
272 LogsDirectory= and ConfigurationDirectory= which are closely related
273 to RuntimeDirectory= but manage per-service directories below
274 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
275 possible to write unit files which when activated automatically gain
276 properly owned service specific directories in these locations, thus
277 making unit files self-contained and increasing compatibility with
278 stateless systems and factory reset where /etc or /var are
279 unpopulated at boot. Matching these new settings there's also
280 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
281 ConfigurationDirectoryMode= for configuring the access mode of these
282 directories. These settings are particularly useful in combination
283 with DynamicUser=yes as they provide secure, properly-owned,
284 writable, and stateful locations for storage, excluded from the
285 sandbox that such services live in otherwise.
287 * Automake support has been removed from this release. systemd is now
290 * systemd-journald will now aggressively cache client metadata during
291 runtime, speeding up log write performance under pressure. This comes
292 at a small price though: as much of the metadata is read
293 asynchronously from /proc/ (and isn't implicitly attached to log
294 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
295 metadata stored alongside a log entry might be slightly
296 out-of-date. Previously it could only be slightly newer than the log
297 message. The time window is small however, and given that the kernel
298 is unlikely to be improved anytime soon in this regard, this appears
301 * nss-myhostname/systemd-resolved will now by default synthesize an
302 A/AAAA resource record for the "_gateway" hostname, pointing to the
303 current default IP gateway. Previously it did that for the "gateway"
304 name, hampering adoption, as some distributions wanted to leave that
305 host name open for local use. The old behaviour may still be
306 requested at build time.
308 * systemd-networkd's [Address] section in .network files gained a new
309 Scope= setting for configuring the IP address scope. The [Network]
310 section gained a new boolean setting ConfigureWithoutCarrier= that
311 tells systemd-networkd to ignore link sensing when configuring the
312 device. The [DHCP] section gained a new Anonymize= boolean option for
313 turning on a number of options suggested in RFC 7844. A new
314 [RoutingPolicyRule] section has been added for configuring the IP
315 routing policy. The [Route] section has gained support for a new
316 Type= setting which permits configuring
317 blackhole/unreachable/prohibit routes.
319 * The [VRF] section in .netdev files gained a new Table= setting for
320 configuring the routing table to use. The [Tunnel] section gained a
321 new Independent= boolean field for configuring tunnels independent of
322 an underlying network interface. The [Bridge] section gained a new
323 GroupForwardMask= option for configuration of propagation of link
324 local frames between bridge ports.
326 * The WakeOnLan= setting in .link files gained support for a number of
327 new modes. A new TCP6SegmentationOffload= setting has been added for
328 configuring TCP/IPv6 hardware segmentation offload.
330 * The IPv6 RA sender implementation may now optionally send out RDNSS
331 and RDNSSL records to supply DNS configuration to peers.
333 * systemd-nspawn gained support for a new --system-call-filter= command
334 line option for adding and removing entries in the default system
335 call filter it applies. Moreover systemd-nspawn has been changed to
336 implement a system call whitelist instead of a blacklist.
338 * systemd-run gained support for a new --pipe command line option. If
339 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
340 are directly passed on to the activated transient service
341 executable. This allows invoking arbitrary processes as systemd
342 services (for example to take benefit of dependency management,
343 accounting management, resource management or log management that is
344 done automatically for services) — while still allowing them to be
345 integrated in a classic UNIX shell pipeline.
347 * When a service sends RELOAD=1 via sd_notify() and reload propagation
348 using ReloadPropagationTo= is configured, a reload is now propagated
349 to configured units. (Previously this was only done on explicitly
350 requested reloads, using "systemctl reload" or an equivalent
353 * For each service unit a restart counter is now kept: it is increased
354 each time the service is restarted due to Restart=, and may be
355 queried using "systemctl show -p NRestarts …".
357 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
358 @signal and @timer have been added, for usage with SystemCallFilter=
359 in unit files and the new --system-call-filter= command line option
360 of systemd-nspawn (see above).
362 * ExecStart= lines in unit files gained two new modifiers: when a
363 command line is prefixed with "!" the command will be executed as
364 configured, except for the credentials applied by
365 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
366 "+", but does still apply namespacing options unlike "+". There's
367 also "!!" now, which is mostly identical, but becomes a NOP on
368 systems that support ambient capabilities. This is useful to write
369 unit files that work with ambient capabilities where possible but
370 automatically fall back to traditional privilege dropping mechanisms
371 on systems where this is not supported.
373 * ListenNetlink= settings in socket units now support RDMA netlink
376 * A new unit file setting LockPersonality= has been added which permits
377 locking down the chosen execution domain ("personality") of a service
380 * A new special target "getty-pre.target" has been added, which is
381 ordered before all text logins, and may be used to order services
382 before textual logins acquire access to the console.
384 * systemd will now attempt to load the virtio-rng.ko kernel module very
385 early on if a VM environment supporting this is detected. This should
386 improve entropy during early boot in virtualized environments.
388 * A _netdev option is now supported in /etc/crypttab that operates in a
389 similar way as the same option in /etc/fstab: it permits configuring
390 encrypted devices that need to be ordered after the network is up.
391 Following this logic, two new special targets
392 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
393 added that are to cryptsetup.target what remote-fs.target and
394 remote-fs-pre.target are to local-fs.target.
396 * Service units gained a new UnsetEnvironment= setting which permits
397 unsetting specific environment variables for services that are
398 normally passed to it (for example in order to mask out locale
399 settings for specific services that can't deal with it).
401 * Units acquired a new boolean option IPAccounting=. When turned on, IP
402 traffic accounting (packet count as well as byte count) is done for
403 the service, and shown as part of "systemctl status" or "systemd-run
406 * Service units acquired two new options IPAddressAllow= and
407 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
408 for configuring a simple IP access control list for all sockets of
409 the unit. These options are available also on .slice and .socket
410 units, permitting flexible access list configuration for individual
411 services as well as groups of services (as defined by a slice unit),
412 including system-wide. Note that IP ACLs configured this way are
413 enforced on every single IPv4 and IPv6 socket created by any process
414 of the service unit, and apply to ingress as well as egress traffic.
416 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
417 structured log message is generated each time the unit is stopped,
418 containing information about the consumed resources of this
421 * A new setting KeyringMode= has been added to unit files, which may be
422 used to control how the kernel keyring is set up for executed
425 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
426 "systemctl kexec" and "systemctl exit" are now always asynchronous in
427 behaviour (that is: these commands return immediately after the
428 operation was enqueued instead of waiting for the operation to
429 complete). Previously, "systemctl poweroff" and "systemctl reboot"
430 were asynchronous on systems using systemd-logind (i.e. almost
431 always, and like they were on sysvinit), and the other three commands
432 were unconditionally synchronous. With this release this is cleaned
433 up, and callers will see the same asynchronous behaviour on all
434 systems for all five operations.
436 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
439 * .timer units now accept calendar specifications in other timezones
440 than UTC or the local timezone.
442 * The tmpfiles snippet var.conf has been changed to create
443 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
444 the "utmp" group already, and it appears to be generally understood
445 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
446 databases. Previously this was implemented correctly for all these
447 databases excepts btmp, which has been opened up like this now
448 too. Note that while the other databases are world-readable
449 (i.e. 0644), btmp is not and remains more restrictive.
451 * The systemd-resolve tool gained a new --reset-server-features
452 switch. When invoked like this systemd-resolved will forget
453 everything it learnt about the features supported by the configured
454 upstream DNS servers, and restarts the feature probing logic on the
455 next resolver look-up for them at the highest feature level
458 * The status dump systemd-resolved sends to the logs upon receiving
459 SIGUSR1 now also includes information about all DNS servers it is
460 configured to use, and the features levels it probed for them.
462 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
463 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
464 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
465 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
466 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
467 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
468 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
469 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
470 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
471 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
472 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
473 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
474 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
475 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
476 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
477 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
478 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
479 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
480 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
481 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
487 * Meson is now supported as build system in addition to Automake. It is
488 our plan to remove Automake in one of our next releases, so that
489 Meson becomes our exclusive build system. Hence, please start using
490 the Meson build system in your downstream packaging. There's plenty
491 of documentation around how to use Meson, the extremely brief
494 ./autogen.sh && ./configure && make && sudo make install
498 meson build && ninja -C build && sudo ninja -C build install
500 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
501 which permits configuring a timeout on the time a job is
502 running. This is particularly useful for setting timeouts on jobs for
505 * Unit files gained two new options ConditionUser= and ConditionGroup=
506 for conditionalizing units based on the identity of the user/group
507 running a systemd user instance.
509 * systemd-networkd now understands a new FlowLabel= setting in the
510 [VXLAN] section of .network files, as well as a Priority= in
511 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
512 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
513 gained support for configuration of GENEVE links, and IPv6 address
514 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
516 * .link files now understand a new Port= setting.
518 * systemd-networkd's DHCP support gained support for DHCP option 119
519 (domain search list).
521 * systemd-networkd gained support for serving IPv6 address ranges using
522 the Router Advertisment protocol. The new .network configuration
523 section [IPv6Prefix] may be used to configure the ranges to
524 serve. This is implemented based on a new, minimal, native server
525 implementation of RA.
527 * journalctl's --output= switch gained support for a new parameter
528 "short-iso-precise" for a mode where timestamps are shown as precise
531 * systemd-udevd's "net_id" builtin may now generate stable network
532 interface names from IBM PowerVM VIO devices as well as ACPI platform
535 * MulticastDNS support in systemd-resolved may now be explicitly
536 enabled/disabled using the new MulticastDNS= configuration file
539 * systemd-resolved may now optionally use libidn2 instead of the libidn
540 for processing internationalized domain names. Support for libidn2
541 should be considered experimental and should not be enabled by
544 * "machinectl pull-tar" and related call may now do verification of
545 downloaded images using SUSE-style .sha256 checksum files in addition
546 to the already existing support for validating using Ubuntu-style
549 * sd-bus gained support for a new sd_bus_message_appendv() call which
550 is va_list equivalent of sd_bus_message_append().
552 * sd-boot gained support for validating images using SHIM/MOK.
554 * The SMACK code learnt support for "onlycap".
556 * systemd-mount --umount is now much smarter in figuring out how to
557 properly unmount a device given its mount or device path.
559 * The code to call libnss_dns as a fallback from libnss_resolve when
560 the communication with systemd-resolved fails was removed. This
561 fallback was redundant and interfered with the [!UNAVAIL=return]
562 suffix. See nss-resolve(8) for the recommended configuration.
564 * systemd-logind may now be restarted without losing state. It stores
565 the file descriptors for devices it manages in the system manager
566 using the FDSTORE= mechanism. Please note that further changes in
567 other components may be required to make use of this (for example
568 Xorg has code to listen for stops of systemd-logind and terminate
569 itself when logind is stopped or restarted, in order to avoid using
570 stale file descriptors for graphical devices, which is now
571 counterproductive and must be reverted in order for restarts of
572 systemd-logind to be safe. See
573 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
575 * All kernel install plugins are called with the environment variable
576 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
577 /etc/machine-id. If the file is missing or empty, the variable is
578 empty and BOOT_DIR_ABS is the path of a temporary directory which is
579 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
580 is empty, all plugins should not put anything in BOOT_DIR_ABS.
582 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
583 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
584 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
585 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
586 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
587 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
588 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
589 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
590 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
591 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
592 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
593 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
594 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
595 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
596 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
597 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
598 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
599 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
600 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
601 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
602 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
603 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
604 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
605 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
606 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
607 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
608 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
609 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
610 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
617 * The "hybrid" control group mode has been modified to improve
618 compatibility with "legacy" cgroups-v1 setups. Specifically, the
619 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
620 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
621 cgroups-v1 hierarchy), the only externally visible change being that
622 the cgroups-v2 hierarchy is also mounted, to
623 /sys/fs/cgroup/unified. This should provide a large degree of
624 compatibility with "legacy" cgroups-v1, while taking benefit of the
625 better management capabilities of cgroups-v2.
627 * The default control group setup mode may be selected both a boot-time
628 via a set of kernel command line parameters (specifically:
629 systemd.unified_cgroup_hierarchy= and
630 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
631 default selected on the configure command line
632 (--with-default-hierarchy=). The upstream default is "hybrid"
633 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
634 this will change in a future systemd version to be "unified" (pure
635 cgroups-v2 mode). The third option for the compile time option is
636 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
637 distributions to default to "hybrid" mode for release distributions,
638 starting with v233. We recommend "unified" for development
639 distributions (specifically: distributions such as Fedora's rawhide)
640 as that's where things are headed in the long run. Use "legacy" for
641 greatest stability and compatibility only.
643 * Note one current limitation of "unified" and "hybrid" control group
644 setup modes: the kernel currently does not permit the systemd --user
645 instance (i.e. unprivileged code) to migrate processes between two
646 disconnected cgroup subtrees, even if both are managed and owned by
647 the user. This effectively means "systemd-run --user --scope" doesn't
648 work when invoked from outside of any "systemd --user" service or
649 scope. Specifically, it is not supported from session scopes. We are
650 working on fixing this in a future systemd version. (See #3388 for
651 further details about this.)
653 * DBus policy files are now installed into /usr rather than /etc. Make
654 sure your system has dbus >= 1.9.18 running before upgrading to this
655 version, or override the install path with --with-dbuspolicydir= .
657 * All python scripts shipped with systemd (specifically: the various
658 tests written in Python) now require Python 3.
660 * systemd unit tests can now run standalone (without the source or
661 build directories), and can be installed into /usr/lib/systemd/tests/
662 with 'make install-tests'.
664 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
665 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
668 * Support for the %c, %r, %R specifiers in unit files has been
669 removed. Specifiers are not supposed to be dependent on configuration
670 in the unit file itself (so that they resolve the same regardless
671 where used in the unit files), but these specifiers were influenced
672 by the Slice= option.
674 * The shell invoked by debug-shell.service now defaults to /bin/sh in
675 all cases. If distributions want to use a different shell for this
676 purpose (for example Fedora's /sbin/sushell) they need to specify
677 this explicitly at configure time using --with-debug-shell=.
679 * The confirmation spawn prompt has been reworked to offer the
682 (c)ontinue, proceed without asking anymore
683 (D)ump, show the state of the unit
684 (f)ail, don't execute the command and pretend it failed
686 (i)nfo, show a short summary of the unit
687 (j)obs, show jobs that are in progress
688 (s)kip, don't execute the command and pretend it succeeded
689 (y)es, execute the command
691 The 'n' choice for the confirmation spawn prompt has been removed,
692 because its meaning was confusing.
694 The prompt may now also be redirected to an alternative console by
695 specifying the console as parameter to systemd.confirm_spawn=.
697 * Services of Type=notify require a READY=1 notification to be sent
698 during startup. If no such message is sent, the service now fails,
699 even if the main process exited with a successful exit code.
701 * Services that fail to start up correctly now always have their
702 ExecStopPost= commands executed. Previously, they'd enter "failed"
703 state directly, without executing these commands.
705 * The option MulticastDNS= of network configuration files has acquired
706 an actual implementation. With MulticastDNS=yes a host can resolve
707 names of remote hosts and reply to mDNS A and AAAA requests.
709 * When units are about to be started an additional check is now done to
710 ensure that all dependencies of type BindsTo= (when used in
711 combination with After=) have been started.
713 * systemd-analyze gained a new verb "syscall-filter" which shows which
714 system call groups are defined for the SystemCallFilter= unit file
715 setting, and which system calls they contain.
717 * A new system call filter group "@filesystem" has been added,
718 consisting of various file system related system calls. Group
719 "@reboot" has been added, covering reboot, kexec and shutdown related
720 calls. Finally, group "@swap" has been added covering swap
721 configuration related calls.
723 * A new unit file option RestrictNamespaces= has been added that may be
724 used to restrict access to the various process namespace types the
725 Linux kernel provides. Specifically, it may be used to take away the
726 right for a service unit to create additional file system, network,
727 user, and other namespaces. This sandboxing option is particularly
728 relevant due to the high amount of recently discovered namespacing
729 related vulnerabilities in the kernel.
731 * systemd-udev's .link files gained support for a new AutoNegotiation=
732 setting for configuring Ethernet auto-negotiation.
734 * systemd-networkd's .network files gained support for a new
735 ListenPort= setting in the [DHCP] section to explicitly configure the
736 UDP client port the DHCP client shall listen on.
738 * .network files gained a new Unmanaged= boolean setting for explicitly
739 excluding one or more interfaces from management by systemd-networkd.
741 * The systemd-networkd ProxyARP= option has been renamed to
742 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
743 renamed to ReduceARPProxy=. The old names continue to be available
746 * systemd-networkd gained support for configuring IPv6 Proxy NDP
747 addresses via the new IPv6ProxyNDPAddress= .network file setting.
749 * systemd-networkd's bonding device support gained support for two new
750 configuration options ActiveSlave= and PrimarySlave=.
752 * The various options in the [Match] section of .network files gained
753 support for negative matching.
755 * New systemd-specific mount options are now understood in /etc/fstab:
757 x-systemd.mount-timeout= may be used to configure the maximum
758 permitted runtime of the mount command.
760 x-systemd.device-bound may be set to bind a mount point to its
761 backing device unit, in order to automatically remove a mount point
762 if its backing device is unplugged. This option may also be
763 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
764 on the block device, which is now automatically set for all CDROM
765 drives, so that mounted CDs are automatically unmounted when they are
766 removed from the drive.
768 x-systemd.after= and x-systemd.before= may be used to explicitly
769 order a mount after or before another unit or mount point.
771 * Enqueued start jobs for device units are now automatically garbage
772 collected if there are no jobs waiting for them anymore.
774 * systemctl list-jobs gained two new switches: with --after, for every
775 queued job the jobs it's waiting for are shown; with --before the
776 jobs which it's blocking are shown.
778 * systemd-nspawn gained support for ephemeral boots from disk images
779 (or in other words: --ephemeral and --image= may now be
780 combined). Moreover, ephemeral boots are now supported for normal
781 directories, even if the backing file system is not btrfs. Of course,
782 if the file system does not support file system snapshots or
783 reflinks, the initial copy operation will be relatively expensive, but
784 this should still be suitable for many use cases.
786 * Calendar time specifications in .timer units now support
787 specifications relative to the end of a month by using "~" instead of
788 "-" as separator between month and day. For example, "*-02~03" means
789 "the third last day in February". In addition a new syntax for
790 repeated events has been added using the "/" character. For example,
791 "9..17/2:00" means "every two hours from 9am to 5pm".
793 * systemd-socket-proxyd gained a new parameter --connections-max= for
794 configuring the maximum number of concurrent connections.
796 * sd-id128 gained a new API for generating unique IDs for the host in a
797 way that does not leak the machine ID. Specifically,
798 sd_id128_get_machine_app_specific() derives an ID based on the
799 machine ID a in well-defined, non-reversible, stable way. This is
800 useful whenever an identifier for the host is needed but where the
801 identifier shall not be useful to identify the system beyond the
802 scope of the application itself. (Internally this uses HMAC-SHA256 as
803 keyed hash function using the machine ID as input.)
805 * NotifyAccess= gained a new supported value "exec". When set
806 notifications are accepted from all processes systemd itself invoked,
807 including all control processes.
809 * .nspawn files gained support for defining overlay mounts using the
810 Overlay= and OverlayReadOnly= options. Previously this functionality
811 was only available on the systemd-nspawn command line.
813 * systemd-nspawn's --bind= and --overlay= options gained support for
814 bind/overlay mounts whose source lies within the container tree by
815 prefixing the source path with "+".
817 * systemd-nspawn's --bind= and --overlay= options gained support for
818 automatically allocating a temporary source directory in /var/tmp
819 that is removed when the container dies. Specifically, if the source
820 directory is specified as empty string this mechanism is selected. An
821 example usage is --overlay=+/var::/var, which creates an overlay
822 mount based on the original /var contained in the image, overlayed
823 with a temporary directory in the host's /var/tmp. This way changes
824 to /var are automatically flushed when the container shuts down.
826 * systemd-nspawn --image= option does now permit raw file system block
827 devices (in addition to images containing partition tables, as
830 * The disk image dissection logic in systemd-nspawn gained support for
831 automatically setting up LUKS encrypted as well as Verity protected
832 partitions. When a container is booted from an encrypted image the
833 passphrase is queried at start-up time. When a container with Verity
834 data is started, the root hash is search in a ".roothash" file
835 accompanying the disk image (alternatively, pass the root hash via
836 the new --root-hash= command line option).
838 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
839 be used to dissect disk images the same way as systemd-nspawn does
840 it, following the Bootable Partition Specification. It may even be
841 used to mount disk images with complex partition setups (including
842 LUKS and Verity partitions) to a local host directory, in order to
843 inspect them. This tool is not considered public API (yet), and is
844 thus not installed into /usr/bin. Please do not rely on its
845 existence, since it might go away or be changed in later systemd
848 * A new generator "systemd-verity-generator" has been added, similar in
849 style to "systemd-cryptsetup-generator", permitting automatic setup of
850 Verity root partitions when systemd boots up. In order to make use of
851 this your partition setup should follow the Discoverable Partitions
852 Specification, and the GPT partition ID of the root file system
853 partition should be identical to the upper 128bit of the Verity root
854 hash. The GPT partition ID of the Verity partition protecting it
855 should be the lower 128bit of the Verity root hash. If the partition
856 image follows this model it is sufficient to specify a single
857 "roothash=" kernel command line argument to both configure which root
858 image and verity partition to use as well as the root hash for
859 it. Note that systemd-nspawn's Verity support follows the same
860 semantics, meaning that disk images with proper Verity data in place
861 may be booted in containers with systemd-nspawn as well as on
862 physical systems via the verity generator. Also note that the "mkosi"
863 tool available at https://github.com/systemd/mkosi has been updated
864 to generate Verity protected disk images following this scheme. In
865 fact, it has been updated to generate disk images that optionally
866 implement a complete UEFI SecureBoot trust chain, involving a signed
867 kernel and initrd image that incorporates such a root hash as well as
868 a Verity-enabled root partition.
870 * The hardware database (hwdb) udev supports has been updated to carry
871 accelerometer quirks.
873 * All system services are now run with a fresh kernel keyring set up
874 for them. The invocation ID is stored by default in it, thus
875 providing a safe, non-overridable way to determine the invocation
878 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
879 options for bind mounting arbitrary paths in a service-specific
880 way. When these options are used, arbitrary host or service files and
881 directories may be mounted to arbitrary locations in the service's
884 * Documentation has been added that lists all of systemd's low-level
885 environment variables:
887 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
889 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
890 whether a specific socket file descriptor matches a specified socket
893 * systemd-firstboot has been updated to check for the
894 systemd.firstboot= kernel command line option. It accepts a boolean
895 and when set to false the first boot questions are skipped.
897 * systemd-fstab-generator has been updated to check for the
898 systemd.volatile= kernel command line option, which either takes an
899 optional boolean parameter or the special value "state". If used the
900 system may be booted in a "volatile" boot mode. Specifically,
901 "systemd.volatile" is used, the root directory will be mounted as
902 tmpfs, and only /usr is mounted from the actual root file system. If
903 "systemd.volatile=state" is used, the root directory will be mounted
904 as usual, but /var is mounted as tmpfs. This concept provides similar
905 functionality as systemd-nspawn's --volatile= option, but provides it
906 on physical boots. Use this option for implementing stateless
907 systems, or testing systems with all state and/or configuration reset
908 to the defaults. (Note though that many distributions are not
909 prepared to boot up without a populated /etc or /var, though.)
911 * systemd-gpt-auto-generator gained support for LUKS encrypted root
912 partitions. Previously it only supported LUKS encrypted partitions
913 for all other uses, except for the root partition itself.
915 * Socket units gained support for listening on AF_VSOCK sockets for
916 communication in virtualized QEMU environments.
918 * The "configure" script gained a new option --with-fallback-hostname=
919 for specifying the fallback hostname to use if none is configured in
920 /etc/hostname. For example, by specifying
921 --with-fallback-hostname=fedora it is possible to default to a
922 hostname of "fedora" on pristine installations.
924 * systemd-cgls gained support for a new --unit= switch for listing only
925 the control groups of a specific unit. Similar --user-unit= has been
926 added for listing only the control groups of a specific user unit.
928 * systemd-mount gained a new --umount switch for unmounting a mount or
929 automount point (and all mount/automount points below it).
931 * systemd will now refuse full configuration reloads (via systemctl
932 daemon-reload and related calls) unless at least 16MiB of free space
933 are available in /run. This is a safety precaution in order to ensure
934 that generators can safely operate after the reload completed.
936 * A new unit file option RootImage= has been added, which has a similar
937 effect as RootDirectory= but mounts the service's root directory from
938 a disk image instead of plain directory. This logic reuses the same
939 image dissection and mount logic that systemd-nspawn already uses,
940 and hence supports any disk images systemd-nspawn supports, including
941 those following the Discoverable Partition Specification, as well as
942 Verity enabled images. This option enables systemd to run system
943 services directly off disk images acting as resource bundles,
944 possibly even including full integrity data.
946 * A new MountAPIVFS= unit file option has been added, taking a boolean
947 argument. If enabled /proc, /sys and /dev (collectively called the
948 "API VFS") will be mounted for the service. This is only relevant if
949 RootDirectory= or RootImage= is used for the service, as these mounts
950 are of course in place in the host mount namespace anyway.
952 * systemd-nspawn gained support for a new --pivot-root= switch. If
953 specified the root directory within the container image is pivoted to
954 the specified mount point, while the original root disk is moved to a
955 different place. This option enables booting of ostree images
956 directly with systemd-nspawn.
958 * The systemd build scripts will no longer complain if the NTP server
959 addresses are not changed from the defaults. Google now supports
960 these NTP servers officially. We still recommend downstreams to
961 properly register an NTP pool with the NTP pool project though.
963 * coredumpctl gained a new "--reverse" option for printing the list
964 of coredumps in reverse order.
966 * coredumpctl will now show additional information about truncated and
967 inaccessible coredumps, as well as coredumps that are still being
968 processed. It also gained a new --quiet switch for suppressing
969 additional informational message in its output.
971 * coredumpctl gained support for only showing coredumps newer and/or
972 older than specific timestamps, using the new --since= and --until=
973 options, reminiscent of journalctl's options by the same name.
975 * The systemd-coredump logic has been improved so that it may be reused
976 to collect backtraces in non-compiled languages, for example in
977 scripting languages such as Python.
979 * machinectl will now show the UID shift of local containers, if user
980 namespacing is enabled for them.
982 * systemd will now optionally run "environment generator" binaries at
983 configuration load time. They may be used to add environment
984 variables to the environment block passed to services invoked. One
985 user environment generator is shipped by default that sets up
986 environment variables based on files dropped into /etc/environment.d
987 and ~/.config/environment.d/.
989 * systemd-resolved now includes the new, recently published 2017 DNSSEC
992 * hostnamed has been updated to report a new chassis type of
993 "convertible" to cover "foldable" laptops that can both act as a
994 tablet and as a laptop, such as various Lenovo Yoga devices.
996 Contributions from: Adrián López, Alexander Galanin, Alexander
997 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
998 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
999 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1000 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1001 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1002 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1003 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1004 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1005 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1006 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1007 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1008 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1009 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1010 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1011 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1012 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1013 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1014 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1015 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1016 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1017 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1018 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1019 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1020 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1021 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1022 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1023 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1026 — Berlin, 2017-03-01
1030 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1031 RestrictAddressFamilies= enabled. These sandboxing options should
1032 generally be compatible with the various external udev call-out
1033 binaries we are aware of, however there may be exceptions, in
1034 particular when exotic languages for these call-outs are used. In
1035 this case, consider turning off these settings locally.
1037 * The new RemoveIPC= option can be used to remove IPC objects owned by
1038 the user or group of a service when that service exits.
1040 * The new ProtectKernelModules= option can be used to disable explicit
1041 load and unload operations of kernel modules by a service. In
1042 addition access to /usr/lib/modules is removed if this option is set.
1044 * ProtectSystem= option gained a new value "strict", which causes the
1045 whole file system tree with the exception of /dev, /proc, and /sys,
1046 to be remounted read-only for a service.
1048 * The new ProtectKernelTunables= option can be used to disable
1049 modification of configuration files in /sys and /proc by a service.
1050 Various directories and files are remounted read-only, so access is
1051 restricted even if the file permissions would allow it.
1053 * The new ProtectControlGroups= option can be used to disable write
1054 access by a service to /sys/fs/cgroup.
1056 * Various systemd services have been hardened with
1057 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1058 RestrictAddressFamilies=.
1060 * Support for dynamically creating users for the lifetime of a service
1061 has been added. If DynamicUser=yes is specified, user and group IDs
1062 will be allocated from the range 61184..65519 for the lifetime of the
1063 service. They can be resolved using the new nss-systemd.so NSS
1064 module. The module must be enabled in /etc/nsswitch.conf. Services
1065 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1066 any resources allocated by the service will be cleaned up when the
1067 service exits. They also have ProtectHome=read-only and
1068 ProtectSystem=strict enabled, so they are not able to make any
1069 permanent modifications to the system.
1071 * The nss-systemd module also always resolves root and nobody, making
1072 it possible to have no /etc/passwd or /etc/group files in minimal
1073 container or chroot environments.
1075 * Services may be started with their own user namespace using the new
1076 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1077 under which the service is running are mapped. All other users are
1080 * Support for the cgroup namespace has been added to systemd-nspawn. If
1081 supported by kernel, the container system started by systemd-nspawn
1082 will have its own view of the cgroup hierarchy. This new behaviour
1083 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1085 * The new MemorySwapMax= option can be used to limit the maximum swap
1086 usage under the unified cgroup hierarchy.
1088 * Support for the CPU controller in the unified cgroup hierarchy has
1089 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1090 options. This controller requires out-of-tree patches for the kernel
1091 and the support is provisional.
1093 * Mount and automount units may now be created transiently
1094 (i.e. dynamically at runtime via the bus API, instead of requiring
1095 unit files in the file system).
1097 * systemd-mount is a new tool which may mount file systems – much like
1098 mount(8), optionally pulling in additional dependencies through
1099 transient .mount and .automount units. For example, this tool
1100 automatically runs fsck on a backing block device before mounting,
1101 and allows the automount logic to be used dynamically from the
1102 command line for establishing mount points. This tool is particularly
1103 useful when dealing with removable media, as it will ensure fsck is
1104 run – if necessary – before the first access and that the file system
1105 is quickly unmounted after each access by utilizing the automount
1106 logic. This maximizes the chance that the file system on the
1107 removable media stays in a clean state, and if it isn't in a clean
1108 state is fixed automatically.
1110 * LazyUnmount=yes option for mount units has been added to expose the
1111 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1114 * /efi will be used as the mount point of the EFI boot partition, if
1115 the directory is present, and the mount point was not configured
1116 through other means (e.g. fstab). If /efi directory does not exist,
1117 /boot will be used as before. This makes it easier to automatically
1118 mount the EFI partition on systems where /boot is used for something
1121 * When operating on GPT disk images for containers, systemd-nspawn will
1122 now mount the ESP to /boot or /efi according to the same rules as PID
1123 1 running on a host. This allows tools like "bootctl" to operate
1124 correctly within such containers, in order to make container images
1125 bootable on physical systems.
1127 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1129 * Two new user session targets have been added to support running
1130 graphical sessions under the systemd --user instance:
1131 graphical-session.target and graphical-session-pre.target. See
1132 systemd.special(7) for a description of how those targets should be
1135 * The vconsole initialization code has been significantly reworked to
1136 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1137 support unicode keymaps. Font and keymap configuration will now be
1138 copied to all allocated virtual consoles.
1140 * FreeBSD's bhyve virtualization is now detected.
1142 * Information recorded in the journal for core dumps now includes the
1143 contents of /proc/mountinfo and the command line of the process at
1144 the top of the process hierarchy (which is usually the init process
1147 * systemd-journal-gatewayd learned the --directory= option to serve
1148 files from the specified location.
1150 * journalctl --root=… can be used to peruse the journal in the
1151 /var/log/ directories inside of a container tree. This is similar to
1152 the existing --machine= option, but does not require the container to
1155 * The hardware database has been extended to support
1156 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1159 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1160 specify the click rate for mice which include a horizontal wheel with
1161 a click rate that is different than the one for the vertical wheel.
1163 * systemd-run gained a new --wait option that makes service execution
1164 synchronous. (Specifically, the command will not return until the
1165 specified service binary exited.)
1167 * systemctl gained a new --wait option that causes the start command to
1168 wait until the units being started have terminated again.
1170 * A new journal output mode "short-full" has been added which displays
1171 timestamps with abbreviated English day names and adds a timezone
1172 suffix. Those timestamps include more information than the default
1173 "short" output mode, and can be passed directly to journalctl's
1174 --since= and --until= options.
1176 * /etc/resolv.conf will be bind-mounted into containers started by
1177 systemd-nspawn, if possible, so any changes to resolv.conf contents
1178 are automatically propagated to the container.
1180 * The number of instances for socket-activated services originating
1181 from a single IP address can be limited with
1182 MaxConnectionsPerSource=, extending the existing setting of
1185 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1188 * .netdev and .network configuration can now be extended through
1191 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1192 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1193 can be enabled and disabled using the new UDPSegmentationOffload=,
1194 TCPSegmentationOffload=, GenericSegmentationOffload=,
1195 GenericReceiveOffload=, LargeReceiveOffload= options in the
1196 [Link] section of .link files.
1198 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1199 Port VLAN ID can be configured for bridge devices using the new STP=,
1200 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1201 section of .netdev files.
1203 * The route table to which routes received over DHCP or RA should be
1204 added can be configured with the new RouteTable= option in the [DHCP]
1205 and [IPv6AcceptRA] sections of .network files.
1207 * The Address Resolution Protocol can be disabled on links managed by
1208 systemd-networkd using the ARP=no setting in the [Link] section of
1211 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1212 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1213 encode information about the result and exit codes of the current
1214 service runtime cycle.
1216 * systemd-sysctl will now configure kernel parameters in the order
1217 they occur in the configuration files. This matches what sysctl
1218 has been traditionally doing.
1220 * kernel-install "plugins" that are executed to perform various
1221 tasks after a new kernel is added and before an old one is removed
1222 can now return a special value to terminate the procedure and
1223 prevent any later plugins from running.
1225 * Journald's SplitMode=login setting has been deprecated. It has been
1226 removed from documentation, and its use is discouraged. In a future
1227 release it will be completely removed, and made equivalent to current
1228 default of SplitMode=uid.
1230 * Storage=both option setting in /etc/systemd/coredump.conf has been
1231 removed. With fast LZ4 compression storing the core dump twice is not
1234 * The --share-system systemd-nspawn option has been replaced with an
1235 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1236 this functionality is discouraged. In addition the variables
1237 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1238 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1239 individual namespaces.
1241 * "machinectl list" now shows the IP address of running containers in
1242 the output, as well as OS release information.
1244 * "loginctl list" now shows the TTY of each session in the output.
1246 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1247 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1248 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1249 tracking objects in a "recursive" mode, where a single client can be
1250 counted multiple times, if it takes multiple references.
1252 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1253 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1254 process using sd-bus automatically exit if the bus connection is
1257 * Bus clients of the service manager may now "pin" loaded units into
1258 memory, by taking an explicit reference on them. This is useful to
1259 ensure the client can retrieve runtime data about the service even
1260 after the service completed execution. Taking such a reference is
1261 available only for privileged clients and should be helpful to watch
1262 running services in a race-free manner, and in particular collect
1263 information about exit statuses and results.
1265 * The nss-resolve module has been changed to strictly return UNAVAIL
1266 when communication via D-Bus with resolved failed, and NOTFOUND when
1267 a lookup completed but was negative. This means it is now possible to
1268 neatly configure fallbacks using nsswitch.conf result checking
1269 expressions. Taking benefit of this, the new recommended
1270 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1272 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1274 * A new setting CtrlAltDelBurstAction= has been added to
1275 /etc/systemd/system.conf which may be used to configure the precise
1276 behaviour if the user on the console presses Ctrl-Alt-Del more often
1277 than 7 times in 2s. Previously this would unconditionally result in
1278 an expedited, immediate reboot. With this new setting the precise
1279 operation may be configured in more detail, and also turned off
1282 * In .netdev files two new settings RemoteChecksumTx= and
1283 RemoteChecksumRx= are now understood that permit configuring the
1284 remote checksumming logic for VXLAN networks.
1286 * The service manager learnt a new "invocation ID" concept for invoked
1287 services. Each runtime cycle of a service will get a new invocation
1288 ID (a 128bit random UUID) assigned that identifies the current
1289 run of the service uniquely and globally. A new invocation ID
1290 is generated each time a service starts up. The journal will store
1291 the invocation ID of a service along with any logged messages, thus
1292 making the invocation ID useful for matching the online runtime of a
1293 service with the offline log data it generated in a safe way without
1294 relying on synchronized timestamps. In many ways this new service
1295 invocation ID concept is similar to the kernel's boot ID concept that
1296 uniquely and globally identifies the runtime of each boot. The
1297 invocation ID of a service is passed to the service itself via an
1298 environment variable ($INVOCATION_ID). A new bus call
1299 GetUnitByInvocationID() has been added that is similar to GetUnit()
1300 but instead of retrieving the bus path for a unit by its name
1301 retrieves it by its invocation ID. The returned path is valid only as
1302 long as the passed invocation ID is current.
1304 * systemd-resolved gained a new "DNSStubListener" setting in
1305 resolved.conf. It either takes a boolean value or the special values
1306 "udp" and "tcp", and configures whether to enable the stub DNS
1307 listener on 127.0.0.53:53.
1309 * IP addresses configured via networkd may now carry additional
1310 configuration settings supported by the kernel. New options include:
1311 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1312 PrefixRoute=, AutoJoin=.
1314 * The PAM configuration fragment file for "user@.service" shipped with
1315 systemd (i.e. the --user instance of systemd) has been stripped to
1316 the minimum necessary to make the system boot. Previously, it
1317 contained Fedora-specific stanzas that did not apply to other
1318 distributions. It is expected that downstream distributions add
1319 additional configuration lines, matching their needs to this file,
1320 using it only as rough template of what systemd itself needs. Note
1321 that this reduced fragment does not even include an invocation of
1322 pam_limits which most distributions probably want to add, even though
1323 systemd itself does not need it. (There's also the new build time
1324 option --with-pamconfdir=no to disable installation of the PAM
1327 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1328 capability is now also dropped from its set (in addition to
1329 CAP_SYS_MKNOD as before).
1331 * In service unit files it is now possible to connect a specific named
1332 file descriptor with stdin/stdout/stdout of an executed service. The
1333 name may be specified in matching .socket units using the
1334 FileDescriptorName= setting.
1336 * A number of journal settings may now be configured on the kernel
1337 command line. Specifically, the following options are now understood:
1338 systemd.journald.max_level_console=,
1339 systemd.journald.max_level_store=,
1340 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1341 systemd.journald.max_level_wall=.
1343 * "systemctl is-enabled --full" will now show by which symlinks a unit
1344 file is enabled in the unit dependency tree.
1346 * Support for VeraCrypt encrypted partitions has been added to the
1347 "cryptsetup" logic and /etc/crypttab.
1349 * systemd-detect-virt gained support for a new --private-users switch
1350 that checks whether the invoking processes are running inside a user
1351 namespace. Similar, a new special value "private-users" for the
1352 existing ConditionVirtualization= setting has been added, permitting
1353 skipping of specific units in user namespace environments.
1355 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1356 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1357 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1358 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1359 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1360 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1361 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1362 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1363 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1364 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1365 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1366 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1367 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1368 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1369 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1370 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1371 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1372 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1373 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1374 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1375 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1376 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1377 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1378 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1379 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1380 Jędrzejewski-Szmek, Zeal Jagannatha
1382 — Santa Fe, 2016-11-03
1386 * In service units the various ExecXYZ= settings have been extended
1387 with an additional special character as first argument of the
1388 assigned value: if the character '+' is used the specified command
1389 line it will be run with full privileges, regardless of User=,
1390 Group=, CapabilityBoundingSet= and similar options. The effect is
1391 similar to the existing PermissionsStartOnly= option, but allows
1392 configuration of this concept for each executed command line
1395 * Services may now alter the service watchdog timeout at runtime by
1396 sending a WATCHDOG_USEC= message via sd_notify().
1398 * MemoryLimit= and related unit settings now optionally take percentage
1399 specifications. The percentage is taken relative to the amount of
1400 physical memory in the system (or in case of containers, the assigned
1401 amount of memory). This allows scaling service resources neatly with
1402 the amount of RAM available on the system. Similarly, systemd-logind's
1403 RuntimeDirectorySize= option now also optionally takes percentage
1406 * In similar fashion TasksMax= takes percentage values now, too. The
1407 value is taken relative to the configured maximum number of processes
1408 on the system. The per-service task maximum has been changed to 15%
1409 using this functionality. (Effectively this is an increase of 512 →
1410 4915 for service units, given the kernel's default pid_max setting.)
1412 * Calendar time specifications in .timer units now understand a ".."
1413 syntax for time ranges. Example: "4..7:10" may now be used for
1414 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1417 * The InaccessableDirectories=, ReadOnlyDirectories= and
1418 ReadWriteDirectories= unit file settings have been renamed to
1419 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1420 applied to all kinds of file nodes, and not just directories, with
1421 the exception of symlinks. Specifically these settings may now be
1422 used on block and character device nodes, UNIX sockets and FIFOS as
1423 well as regular files. The old names of these settings remain
1424 available for compatibility.
1426 * systemd will now log about all service processes it kills forcibly
1427 (using SIGKILL) because they remained after the clean shutdown phase
1428 of the service completed. This should help identifying services that
1429 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1430 systemd-logind's configuration a similar log message is generated for
1431 processes killed at the end of each session due to this setting.
1433 * systemd will now set the $JOURNAL_STREAM environment variable for all
1434 services whose stdout/stderr are connected to the Journal (which
1435 effectively means by default: all services). The variable contains
1436 the device and inode number of the file descriptor used for
1437 stdout/stderr. This may be used by invoked programs to detect whether
1438 their stdout/stderr is connected to the Journal, in which case they
1439 can switch over to direct Journal communication, thus being able to
1440 pass extended, structured metadata along with their log messages. As
1441 one example, this is now used by glib's logging primitives.
1443 * When using systemd's default tmp.mount unit for /tmp, the mount point
1444 will now be established with the "nosuid" and "nodev" options. This
1445 avoids privilege escalation attacks that put traps and exploits into
1446 /tmp. However, this might cause problems if you e. g. put container
1447 images or overlays into /tmp; if you need this, override tmp.mount's
1448 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1451 * systemd now supports the "memory" cgroup controller also on
1454 * The systemd-cgtop tool now optionally takes a control group path as
1455 command line argument. If specified, the control group list shown is
1456 limited to subgroups of that group.
1458 * The SystemCallFilter= unit file setting gained support for
1459 pre-defined, named system call filter sets. For example
1460 SystemCallFilter=@clock is now an effective way to make all clock
1461 changing-related system calls unavailable to a service. A number of
1462 similar pre-defined groups are defined. Writing system call filters
1463 for system services is simplified substantially with this new
1464 concept. Accordingly, all of systemd's own, long-running services now
1465 enable system call filtering based on this, by default.
1467 * A new service setting MemoryDenyWriteExecute= has been added, taking
1468 a boolean value. If turned on, a service may no longer create memory
1469 mappings that are writable and executable at the same time. This
1470 enhances security for services where this is enabled as it becomes
1471 harder to dynamically write and then execute memory in exploited
1472 service processes. This option has been enabled for all of systemd's
1473 own long-running services.
1475 * A new RestrictRealtime= service setting has been added, taking a
1476 boolean argument. If set the service's processes may no longer
1477 acquire realtime scheduling. This improves security as realtime
1478 scheduling may otherwise be used to easily freeze the system.
1480 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1481 value. This may be used for requesting that the system manager inside
1482 of the container reports start-up completion to nspawn which then
1483 propagates this notification further to the service manager
1484 supervising nspawn itself. A related option NotifyReady= in .nspawn
1485 files has been added too. This functionality allows ordering of the
1486 start-up of multiple containers using the usual systemd ordering
1489 * machinectl gained a new command "stop" that is an alias for
1492 * systemd-resolved gained support for contacting DNS servers on
1493 link-local IPv6 addresses.
1495 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1496 its caches. A method call for requesting the same operation has been
1497 added to the bus API too, and is made available via "systemd-resolve
1500 * systemd-resolve gained a new --status switch. If passed a brief
1501 summary of the used DNS configuration with per-interface information
1504 * resolved.conf gained a new Cache= boolean option, defaulting to
1505 on. If turned off local DNS caching is disabled. This comes with a
1506 performance penalty in particular when DNSSEC is enabled. Note that
1507 resolved disables its internal caching implicitly anyway, when the
1508 configured DNS server is on a host-local IP address such as ::1 or
1509 127.0.0.1, thus automatically avoiding double local caching.
1511 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1512 for DNS requests. This improves compatibility with local programs
1513 that do not use the libc NSS or systemd-resolved's bus APIs for name
1514 resolution. This minimal DNS service is only available to local
1515 programs and does not implement the full DNS protocol, but enough to
1516 cover local DNS clients. A new, static resolv.conf file, listing just
1517 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1518 now recommended to make /etc/resolv.conf a symlink to this file in
1519 order to route all DNS lookups to systemd-resolved, regardless if
1520 done via NSS, the bus API or raw DNS packets. Note that this local
1521 DNS service is not as fully featured as the libc NSS or
1522 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1523 used to deliver link-local address information (as this implies
1524 sending a local interface index along), LLMNR/mDNS support via this
1525 interface is severely restricted. It is thus strongly recommended for
1526 all applications to use the libc NSS API or native systemd-resolved
1529 * systemd-networkd's bridge support learned a new setting
1530 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1531 in .network files has been added for configuring VLAN bridging in
1532 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1534 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1535 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1536 now be acquired without relying on DHCPv6. Two new options
1537 UseDomains= and UseDNS= have been added to configure this behaviour.
1539 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1540 renamed IPv6AcceptRA=, without altering its behaviour. The old
1541 setting name remains available for compatibility reasons.
1543 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1544 Key=, InputKey= and OutputKey=.
1546 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1547 interface configuration.
1549 * "systemctl edit" may now be used to create new unit files by
1550 specifying the --force switch.
1552 * sd-event gained a new function sd_event_get_iteration() for
1553 requesting the current iteration counter of the event loop. It starts
1554 at zero and is increased by one with each event loop iteration.
1556 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1557 file. It can be used in lieu of %systemd_requires in packages which
1558 don't use any systemd functionality and are intended to be installed
1559 in minimal containers without systemd present. This macro provides
1560 ordering dependencies to ensure that if the package is installed in
1561 the same rpm transaction as systemd, systemd will be installed before
1562 the scriptlets for the package are executed, allowing unit presets
1565 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1566 been added to simplify packaging of generators.
1568 * The os-release file gained VERSION_CODENAME field for the
1569 distribution nickname (e.g. VERSION_CODENAME=woody).
1571 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1572 can be set to disable parsing of metadata and the creation
1573 of persistent symlinks for that device.
1575 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1576 to make them available to logged-in users has been reverted.
1578 * Much of the common code of the various systemd components is now
1579 built into an internal shared library libsystemd-shared-231.so
1580 (incorporating the systemd version number in the name, to be updated
1581 with future releases) that the components link to. This should
1582 decrease systemd footprint both in memory during runtime and on
1583 disk. Note that the shared library is not for public use, and is
1584 neither API not ABI stable, but is likely to change with every new
1585 released update. Packagers need to make sure that binaries
1586 linking to libsystemd-shared.so are updated in step with the
1589 * Configuration for "mkosi" is now part of the systemd
1590 repository. mkosi is a tool to easily build legacy-free OS images,
1591 and is available on github: https://github.com/systemd/mkosi. If
1592 "mkosi" is invoked in the build tree a new raw OS image is generated
1593 incorporating the systemd sources currently being worked on and a
1594 clean, fresh distribution installation. The generated OS image may be
1595 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1596 UEFI PC. This functionality is particularly useful to easily test
1597 local changes made to systemd in a pristine, defined environment. See
1598 HACKING for details.
1600 * configure learned the --with-support-url= option to specify the
1601 distribution's bugtracker.
1603 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1604 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1605 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1606 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1607 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1608 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1609 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1610 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1611 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1612 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1613 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1614 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1615 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1616 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1617 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1618 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1619 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1620 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1621 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1623 — Berlin, 2016-07-25
1627 * DNSSEC is now turned on by default in systemd-resolved (in
1628 "allow-downgrade" mode), but may be turned off during compile time by
1629 passing "--with-default-dnssec=no" to "configure" (and of course,
1630 during runtime with DNSSEC= in resolved.conf). We recommend
1631 downstreams to leave this on at least during development cycles and
1632 report any issues with the DNSSEC logic upstream. We are very
1633 interested in collecting feedback about the DNSSEC validator and its
1634 limitations in the wild. Note however, that DNSSEC support is
1635 probably nothing downstreams should turn on in stable distros just
1636 yet, as it might create incompatibilities with a few DNS servers and
1637 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1638 automatically whenever we detect such incompatible setups, but there
1639 might be systems we do not cover yet. Hence: please help us testing
1640 the DNSSEC code, leave this on where you can, report back, but then
1641 again don't consider turning this on in your stable, LTS or
1642 production release just yet. (Note that you have to enable
1643 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1644 and its DNSSEC mode for host name resolution from local
1647 * systemd-resolve conveniently resolves DANE records with the --tlsa
1648 option and OPENPGPKEY records with the --openpgp option. It also
1649 supports dumping raw DNS record data via the new --raw= switch.
1651 * systemd-logind will now by default terminate user processes that are
1652 part of the user session scope unit (session-XX.scope) when the user
1653 logs out. This behavior is controlled by the KillUserProcesses=
1654 setting in logind.conf, and the previous default of "no" is now
1655 changed to "yes". This means that user sessions will be properly
1656 cleaned up after, but additional steps are necessary to allow
1657 intentionally long-running processes to survive logout.
1659 While the user is logged in at least once, user@.service is running,
1660 and any service that should survive the end of any individual login
1661 session can be started at a user service or scope using systemd-run.
1662 systemd-run(1) man page has been extended with an example which shows
1663 how to run screen in a scope unit underneath user@.service. The same
1664 command works for tmux.
1666 After the user logs out of all sessions, user@.service will be
1667 terminated too, by default, unless the user has "lingering" enabled.
1668 To effectively allow users to run long-term tasks even if they are
1669 logged out, lingering must be enabled for them. See loginctl(1) for
1670 details. The default polkit policy was modified to allow users to
1671 set lingering for themselves without authentication.
1673 Previous defaults can be restored at compile time by the
1674 --without-kill-user-processes option to "configure".
1676 * systemd-logind gained new configuration settings SessionsMax= and
1677 InhibitorsMax=, both with a default of 8192. It will not register new
1678 user sessions or inhibitors above this limit.
1680 * systemd-logind will now reload configuration on SIGHUP.
1682 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1683 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1684 enable. Also, support for the "io" cgroup controller in the unified
1685 hierarchy has been added, so that the "memory", "pids" and "io" are
1686 now the controllers that are supported on the unified hierarchy.
1688 WARNING: it is not possible to use previous systemd versions with
1689 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1690 is necessary to also update systemd in the initramfs if using the
1691 unified hierarchy. An updated SELinux policy is also required.
1693 * LLDP support has been extended, and both passive (receive-only) and
1694 active (sender) modes are supported. Passive mode ("routers-only") is
1695 enabled by default in systemd-networkd. Active LLDP mode is enabled
1696 by default for containers on the internal network. The "networkctl
1697 lldp" command may be used to list information gathered. "networkctl
1698 status" will also show basic LLDP information on connected peers now.
1700 * The IAID and DUID unique identifier sent in DHCP requests may now be
1701 configured for the system and each .network file managed by
1702 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1704 * systemd-networkd gained support for configuring proxy ARP support for
1705 each interface, via the ProxyArp= setting in .network files. It also
1706 gained support for configuring the multicast querier feature of
1707 bridge devices, via the new MulticastQuerier= setting in .netdev
1708 files. Similarly, snooping on the IGMP traffic can be controlled
1709 via the new setting MulticastSnooping=.
1711 A new setting PreferredLifetime= has been added for addresses
1712 configured in .network file to configure the lifetime intended for an
1715 The systemd-networkd DHCP server gained the option EmitRouter=, which
1716 defaults to yes, to configure whether the DHCP Option 3 (Router)
1719 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1720 systemd-socket-activate and installed into /usr/bin. It is now fully
1723 * systemd-journald now uses separate threads to flush changes to disk
1724 when closing journal files, thus reducing impact of slow disk I/O on
1725 logging performance.
1727 * The sd-journal API gained two new calls
1728 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1729 can be used to open journal files using file descriptors instead of
1730 file or directory paths. sd_journal_open_container() has been
1731 deprecated, sd_journal_open_directory_fd() should be used instead
1732 with the flag SD_JOURNAL_OS_ROOT.
1734 * journalctl learned a new output mode "-o short-unix" that outputs log
1735 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1736 UTC). It also gained support for a new --no-hostname setting to
1737 suppress the hostname column in the family of "short" output modes.
1739 * systemd-ask-password now optionally skips printing of the password to
1740 stdout with --no-output which can be useful in scripts.
1742 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1743 (devices tagged with ID_MAKER_TOOL) are now tagged with
1744 "uaccess" and are available to logged in users.
1746 * The DeviceAllow= unit setting now supports specifiers (with "%").
1748 * "systemctl show" gained a new --value switch, which allows print a
1749 only the contents of a specific unit property, without also printing
1750 the property's name. Similar support was added to "show*" verbs
1751 of loginctl and machinectl that output "key=value" lists.
1753 * A new unit type "generated" was added for files dynamically generated
1754 by generator tools. Similarly, a new unit type "transient" is used
1755 for unit files created using the runtime API. "systemctl enable" will
1756 refuse to operate on such files.
1758 * A new command "systemctl revert" has been added that may be used to
1759 revert to the vendor version of a unit file, in case local changes
1760 have been made by adding drop-ins or overriding the unit file.
1762 * "machinectl clean" gained a new verb to automatically remove all or
1763 just hidden container images.
1765 * systemd-tmpfiles gained support for a new line type "e" for emptying
1766 directories, if they exist, without creating them if they don't.
1768 * systemd-nspawn gained support for automatically patching the UID/GIDs
1769 of the owners and the ACLs of all files and directories in a
1770 container tree to match the UID/GID user namespacing range selected
1771 for the container invocation. This mode is enabled via the new
1772 --private-users-chown switch. It also gained support for
1773 automatically choosing a free, previously unused UID/GID range when
1774 starting a container, via the new --private-users=pick setting (which
1775 implies --private-users-chown). Together, these options for the first
1776 time make user namespacing for nspawn containers fully automatic and
1777 thus deployable. The systemd-nspawn@.service template unit file has
1778 been changed to use this functionality by default.
1780 * systemd-nspawn gained a new --network-zone= switch, that allows
1781 creating ad-hoc virtual Ethernet links between multiple containers,
1782 that only exist as long as at least one container referencing them is
1783 running. This allows easy connecting of multiple containers with a
1784 common link that implements an Ethernet broadcast domain. Each of
1785 these network "zones" may be named relatively freely by the user, and
1786 may be referenced by any number of containers, but each container may
1787 only reference one of these "zones". On the lower level, this is
1788 implemented by an automatically managed bridge network interface for
1789 each zone, that is created when the first container referencing its
1790 zone is created and removed when the last one referencing its zone
1793 * The default start timeout may now be configured on the kernel command
1794 line via systemd.default_timeout_start_sec=. It was already
1795 configurable via the DefaultTimeoutStartSec= option in
1796 /etc/systemd/system.conf.
1798 * Socket units gained a new TriggerLimitIntervalSec= and
1799 TriggerLimitBurst= setting to configure a limit on the activation
1800 rate of the socket unit.
1802 * The LimitNICE= setting now optionally takes normal UNIX nice values
1803 in addition to the raw integer limit value. If the specified
1804 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1805 value is understood as UNIX nice value. If not prefixed like this it
1806 is understood as raw RLIMIT_NICE limit.
1808 * Note that the effect of the PrivateDevices= unit file setting changed
1809 slightly with this release: the per-device /dev file system will be
1810 mounted read-only from this version on, and will have "noexec"
1811 set. This (minor) change of behavior might cause some (exceptional)
1812 legacy software to break, when PrivateDevices=yes is set for its
1813 service. Please leave PrivateDevices= off if you run into problems
1816 * systemd-bootchart has been split out to a separate repository:
1817 https://github.com/systemd/systemd-bootchart
1819 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1820 merged into the kernel in its current form.
1822 * The compatibility libraries libsystemd-daemon.so,
1823 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1824 which have been deprecated since systemd-209 have been removed along
1825 with the corresponding pkg-config files. All symbols provided by
1826 those libraries are provided by libsystemd.so.
1828 * The Capabilities= unit file setting has been removed (it is ignored
1829 for backwards compatibility). AmbientCapabilities= and
1830 CapabilityBoundingSet= should be used instead.
1832 * A new special target has been added, initrd-root-device.target,
1833 which creates a synchronization point for dependencies of the root
1834 device in early userspace. Initramfs builders must ensure that this
1835 target is now included in early userspace.
1837 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1838 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1839 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1840 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1841 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1842 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1843 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1844 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1845 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1846 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1847 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1848 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1849 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1850 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1851 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1852 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1853 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1854 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1855 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1856 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1857 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1858 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1859 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1860 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1861 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1864 — Fairfax, 2016-05-21
1868 * The systemd-resolved DNS resolver service has gained a substantial
1869 set of new features, most prominently it may now act as a DNSSEC
1870 validating stub resolver. DNSSEC mode is currently turned off by
1871 default, but is expected to be turned on by default in one of the
1872 next releases. For now, we invite everybody to test the DNSSEC logic
1873 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1874 service also gained a full set of D-Bus interfaces, including calls
1875 to configure DNS and DNSSEC settings per link (for use by external
1876 network management software). systemd-resolved and systemd-networkd
1877 now distinguish between "search" and "routing" domains. The former
1878 are used to qualify single-label names, the latter are used purely
1879 for routing lookups within certain domains to specific links.
1880 resolved now also synthesizes RRs for all entries from /etc/hosts.
1882 * The systemd-resolve tool (which is a client utility for
1883 systemd-resolved) has been improved considerably and is now fully
1884 supported and documented. Hence it has moved from /usr/lib/systemd to
1887 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1890 * The coredump collection logic has been reworked: when a coredump is
1891 collected it is now written to disk, compressed and processed
1892 (including stacktrace extraction) from a new instantiated service
1893 systemd-coredump@.service, instead of directly from the
1894 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1895 processing large coredumps can take up a substantial amount of
1896 resources and time, and this previously happened entirely outside of
1897 systemd's service supervision. With the new logic the core_pattern
1898 hook only does minimal metadata collection before passing off control
1899 to the new instantiated service, which is configured with a time
1900 limit, a nice level and other settings to minimize negative impact on
1901 the rest of the system. Also note that the new logic will honour the
1902 RLIMIT_CORE setting of the crashed process, which now allows users
1903 and processes to turn off coredumping for their processes by setting
1906 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1907 and all forked processes by default. Previously, PID 1 would leave
1908 the setting at "0" for all processes, as set by the kernel. Note that
1909 the resource limit traditionally has no effect on the generated
1910 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1911 logic is used. Since the limit is now honoured (see above) its
1912 default has been changed so that the coredumping logic is enabled by
1913 default for all processes, while allowing specific opt-out.
1915 * When the stacktrace is extracted from processes of system users, this
1916 is now done as "systemd-coredump" user, in order to sandbox this
1917 potentially security sensitive parsing operation. (Note that when
1918 processing coredumps of normal users this is done under the user ID
1919 of process that crashed, as before.) Packagers should take notice
1920 that it is now necessary to create the "systemd-coredump" system user
1921 and group at package installation time.
1923 * The systemd-activate socket activation testing tool gained support
1924 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1925 and --seqpacket switches. It also has been extended to support both
1926 new-style and inetd-style file descriptor passing. Use the new
1927 --inetd switch to request inetd-style file descriptor passing.
1929 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1930 variable, which takes a boolean value. If set to false, ANSI color
1931 output is disabled in the tools even when run on a terminal that
1934 * The VXLAN support in networkd now supports two new settings
1935 DestinationPort= and PortRange=.
1937 * A new systemd.machine_id= kernel command line switch has been added,
1938 that may be used to set the machine ID in /etc/machine-id if it is
1939 not initialized yet. This command line option has no effect if the
1940 file is already initialized.
1942 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1943 specified command line as PID 2 rather than PID 1 in the
1944 container. In this mode PID 1 is a minimal stub init process that
1945 implements the special POSIX and Linux semantics of PID 1 regarding
1946 signal and child process management. Note that this stub init process
1947 is implemented in nspawn itself and requires no support from the
1948 container image. This new logic is useful to support running
1949 arbitrary commands in the container, as normal processes are
1950 generally not prepared to run as PID 1.
1952 * systemd-nspawn gained a new --chdir= switch for setting the current
1953 working directory for the process started in the container.
1955 * "journalctl /dev/sda" will now output all kernel log messages for
1956 specified device from the current boot, in addition to all devices
1957 that are parents of it. This should make log output about devices
1958 pretty useful, as long as kernel drivers attach enough metadata to
1959 the log messages. (The usual SATA drivers do.)
1961 * The sd-journal API gained two new calls
1962 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1963 that report whether log data from /run or /var has been found.
1965 * journalctl gained a new switch "--fields" that prints all journal
1966 record field names currently in use in the journal. This is backed
1967 by two new sd-journal API calls sd_journal_enumerate_fields() and
1968 sd_journal_restart_fields().
1970 * Most configurable timeouts in systemd now expect an argument of
1971 "infinity" to turn them off, instead of "0" as before. The semantics
1972 from now on is that a timeout of "0" means "now", and "infinity"
1973 means "never". To maintain backwards compatibility, "0" continues to
1974 turn off previously existing timeout settings.
1976 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1977 try-reload-or-restart" to clarify what it actually does: the "try"
1978 logic applies to both reloading and restarting, not just restarting.
1979 The old name continues to be accepted for compatibility.
1981 * On boot-up, when PID 1 detects that the system clock is behind the
1982 release date of the systemd version in use, the clock is now set
1983 to the latter. Previously, this was already done in timesyncd, in order
1984 to avoid running with clocks set to the various clock epochs such as
1985 1902, 1938 or 1970. With this change the logic is now done in PID 1
1986 in addition to timesyncd during early boot-up, so that it is enforced
1987 before the first process is spawned by systemd. Note that the logic
1988 in timesyncd remains, as it is more comprehensive and ensures
1989 clock monotonicity by maintaining a persistent timestamp file in
1990 /var. Since /var is generally not available in earliest boot or the
1991 initrd, this part of the logic remains in timesyncd, and is not done
1994 * Support for tweaking details in net_cls.class_id through the
1995 NetClass= configuration directive has been removed, as the kernel
1996 people have decided to deprecate that controller in cgroup v2.
1997 Userspace tools such as nftables are moving over to setting rules
1998 that are specific to the full cgroup path of a task, which obsoletes
1999 these controllers anyway. The NetClass= directive is kept around for
2000 legacy compatibility reasons. For a more in-depth description of the
2001 kernel change, please refer to the respective upstream commit:
2003 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2005 * A new service setting RuntimeMaxSec= has been added that may be used
2006 to specify a maximum runtime for a service. If the timeout is hit, the
2007 service is terminated and put into a failure state.
2009 * A new service setting AmbientCapabilities= has been added. It allows
2010 configuration of additional Linux process capabilities that are
2011 passed to the activated processes. This is only available on very
2014 * The process resource limit settings in service units may now be used
2015 to configure hard and soft limits individually.
2017 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2018 expose support for gcc's __attribute__((cleanup())) C extension.
2019 Specifically, for many object destructor functions alternative
2020 versions have been added that have names suffixed with "p" and take a
2021 pointer to a pointer to the object to destroy, instead of just a
2022 pointer to the object itself. This is useful because these destructor
2023 functions may be used directly as parameters to the cleanup
2024 construct. Internally, systemd has been a heavy user of this GCC
2025 extension for a long time, and with this change similar support is
2026 now available to consumers of the library outside of systemd. Note
2027 that by using this extension in your sources compatibility with old
2028 and strictly ANSI compatible C compilers is lost. However, all gcc or
2029 LLVM versions of recent years support this extension.
2031 * Timer units gained support for a new setting RandomizedDelaySec= that
2032 allows configuring some additional randomized delay to the configured
2033 time. This is useful to spread out timer events to avoid load peaks in
2034 clusters or larger setups.
2036 * Calendar time specifications now support sub-second accuracy.
2038 * Socket units now support listening on SCTP and UDP-lite protocol
2041 * The sd-event API now comes with a full set of man pages.
2043 * Older versions of systemd contained experimental support for
2044 compressing journal files and coredumps with the LZ4 compressor that
2045 was not compatible with the lz4 binary (due to API limitations of the
2046 lz4 library). This support has been removed; only support for files
2047 compatible with the lz4 binary remains. This LZ4 logic is now
2048 officially supported and no longer considered experimental.
2050 * The dkr image import logic has been removed again from importd. dkr's
2051 micro-services focus doesn't fit into the machine image focus of
2052 importd, and quickly got out of date with the upstream dkr API.
2054 * Creation of the /run/lock/lockdev/ directory was dropped from
2055 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2056 been available for many years. If you still need this, you need to
2057 create your own tmpfiles.d config file with:
2059 d /run/lock/lockdev 0775 root lock -
2061 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2062 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2063 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2064 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2065 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2066 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2067 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2068 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2069 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2070 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2071 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2072 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2073 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2074 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2075 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2076 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2077 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2078 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2079 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2081 — Berlin, 2016-02-11
2085 * A number of properties previously only settable in unit
2086 files are now also available as properties to set when
2087 creating transient units programmatically via the bus, as it
2088 is exposed with systemd-run's --property=
2089 setting. Specifically, these are: SyslogIdentifier=,
2090 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2091 EnvironmentFile=, ReadWriteDirectories=,
2092 ReadOnlyDirectories=, InaccessibleDirectories=,
2093 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2095 * When creating transient services via the bus API it is now
2096 possible to pass in a set of file descriptors to use as
2097 STDIN/STDOUT/STDERR for the invoked process.
2099 * Slice units may now be created transiently via the bus APIs,
2100 similar to the way service and scope units may already be
2101 created transiently.
2103 * Wherever systemd expects a calendar timestamp specification
2104 (like in journalctl's --since= and --until= switches) UTC
2105 timestamps are now supported. Timestamps suffixed with "UTC"
2106 are now considered to be in Universal Time Coordinated
2107 instead of the local timezone. Also, timestamps may now
2108 optionally be specified with sub-second accuracy. Both of
2109 these additions also apply to recurring calendar event
2110 specification, such as OnCalendar= in timer units.
2112 * journalctl gained a new "--sync" switch that asks the
2113 journal daemon to write all so far unwritten log messages to
2114 disk and sync the files, before returning.
2116 * systemd-tmpfiles learned two new line types "q" and "Q" that
2117 operate like "v", but also set up a basic btrfs quota
2118 hierarchy when used on a btrfs file system with quota
2121 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2122 instead of a subvolume (even on a btrfs file system) if the
2123 root directory is a plain directory, and not a
2124 subvolume. This should simplify things with certain chroot()
2125 environments which are not aware of the concept of btrfs
2128 * systemd-detect-virt gained a new --chroot switch to detect
2129 whether execution takes place in a chroot() environment.
2131 * CPUAffinity= now takes CPU index ranges in addition to
2134 * The various memory-related resource limit settings (such as
2135 LimitAS=) now understand the usual K, M, G, ... suffixes to
2136 the base of 1024 (IEC). Similar, the time-related resource
2137 limit settings understand the usual min, h, day, ...
2140 * There's a new system.conf setting DefaultTasksMax= to
2141 control the default TasksMax= setting for services and
2142 scopes running on the system. (TasksMax= is the primary
2143 setting that exposes the "pids" cgroup controller on systemd
2144 and was introduced in the previous systemd release.) The
2145 setting now defaults to 512, which means services that are
2146 not explicitly configured otherwise will only be able to
2147 create 512 processes or threads at maximum, from this
2148 version on. Note that this means that thread- or
2149 process-heavy services might need to be reconfigured to set
2150 TasksMax= to a higher value. It is sufficient to set
2151 TasksMax= in these specific unit files to a higher value, or
2152 even "infinity". Similar, there's now a logind.conf setting
2153 UserTasksMax= that defaults to 4096 and limits the total
2154 number of processes or tasks each user may own
2155 concurrently. nspawn containers also have the TasksMax=
2156 value set by default now, to 8192. Note that all of this
2157 only has an effect if the "pids" cgroup controller is
2158 enabled in the kernel. The general benefit of these changes
2159 should be a more robust and safer system, that provides a
2160 certain amount of per-service fork() bomb protection.
2162 * systemd-nspawn gained the new --network-veth-extra= switch
2163 to define additional and arbitrarily-named virtual Ethernet
2164 links between the host and the container.
2166 * A new service execution setting PassEnvironment= has been
2167 added that allows importing select environment variables
2168 from PID1's environment block into the environment block of
2171 * Timer units gained support for a new RemainAfterElapse=
2172 setting which takes a boolean argument. It defaults to on,
2173 exposing behaviour unchanged to previous releases. If set to
2174 off, timer units are unloaded after they elapsed if they
2175 cannot elapse again. This is particularly useful for
2176 transient timer units, which shall not stay around longer
2177 than until they first elapse.
2179 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2180 default now (the kernel default is 16). This is beneficial
2181 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2182 allows substantially larger numbers of queued
2183 datagrams. This should increase the capability of systemd to
2184 parallelize boot-up, as logging and sd_notify() are unlikely
2185 to stall execution anymore. If you need to change the value
2186 from the new defaults, use the usual sysctl.d/ snippets.
2188 * The compression framing format used by the journal or
2189 coredump processing has changed to be in line with what the
2190 official LZ4 tools generate. LZ4 compression support in
2191 systemd was considered unsupported previously, as the format
2192 was not compatible with the normal tools. With this release
2193 this has changed now, and it is hence safe for downstream
2194 distributions to turn it on. While not compressing as well
2195 as the XZ, LZ4 is substantially faster, which makes
2196 it a good default choice for the compression logic in the
2197 journal and in coredump handling.
2199 * Any reference to /etc/mtab has been dropped from
2200 systemd. The file has been obsolete since a while, but
2201 systemd refused to work on systems where it was incorrectly
2202 set up (it should be a symlink or non-existent). Please make
2203 sure to update to util-linux 2.27.1 or newer in conjunction
2204 with this systemd release, which also drops any reference to
2205 /etc/mtab. If you maintain a distribution make sure that no
2206 software you package still references it, as this is a
2207 likely source of bugs. There's also a glibc bug pending,
2208 asking for removal of any reference to this obsolete file:
2210 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2212 Note that only util-linux versions built with
2213 --enable-libmount-force-mountinfo are supported.
2215 * Support for the ".snapshot" unit type has been removed. This
2216 feature turned out to be little useful and little used, and
2217 has now been removed from the core and from systemctl.
2219 * The dependency types RequiresOverridable= and
2220 RequisiteOverridable= have been removed from systemd. They
2221 have been used only very sparingly to our knowledge and
2222 other options that provide a similar effect (such as
2223 systemctl --mode=ignore-dependencies) are much more useful
2224 and commonly used. Moreover, they were only half-way
2225 implemented as the option to control behaviour regarding
2226 these dependencies was never added to systemctl. By removing
2227 these dependency types the execution engine becomes a bit
2228 simpler. Unit files that use these dependencies should be
2229 changed to use the non-Overridable dependency types
2230 instead. In fact, when parsing unit files with these
2231 options, that's what systemd will automatically convert them
2232 too, but it will also warn, asking users to fix the unit
2233 files accordingly. Removal of these dependency types should
2234 only affect a negligible number of unit files in the wild.
2236 * Behaviour of networkd's IPForward= option changed
2237 (again). It will no longer maintain a per-interface setting,
2238 but propagate one way from interfaces where this is enabled
2239 to the global kernel setting. The global setting will be
2240 enabled when requested by a network that is set up, but
2241 never be disabled again. This change was made to make sure
2242 IPv4 and IPv6 behaviour regarding packet forwarding is
2243 similar (as the Linux IPv6 stack does not support
2244 per-interface control of this setting) and to minimize
2247 * In unit files the behaviour of %u, %U, %h, %s has
2248 changed. These specifiers will now unconditionally resolve
2249 to the various user database fields of the user that the
2250 systemd instance is running as, instead of the user
2251 configured in the specific unit via User=. Note that this
2252 effectively doesn't change much, as resolving of these
2253 specifiers was already turned off in the --system instance
2254 of systemd, as we cannot do NSS lookups from PID 1. In the
2255 --user instance of systemd these specifiers where correctly
2256 resolved, but hardly made any sense, since the user instance
2257 lacks privileges to do user switches anyway, and User= is
2258 hence useless. Moreover, even in the --user instance of
2259 systemd behaviour was awkward as it would only take settings
2260 from User= assignment placed before the specifier into
2261 account. In order to unify and simplify the logic around
2262 this the specifiers will now always resolve to the
2263 credentials of the user invoking the manager (which in case
2264 of PID 1 is the root user).
2266 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2267 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2268 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2269 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2270 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2271 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2272 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2273 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2274 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2275 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2276 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2277 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2278 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2279 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2282 — Berlin, 2015-11-18
2286 * systemd now depends on util-linux v2.27. More specifically,
2287 the newly added mount monitor feature in libmount now
2288 replaces systemd's former own implementation.
2290 * libmount mandates /etc/mtab not to be regular file, and
2291 systemd now enforces this condition at early boot.
2292 /etc/mtab has been deprecated and warned about for a very
2293 long time, so systems running systemd should already have
2294 stopped having this file around as anything else than a
2295 symlink to /proc/self/mounts.
2297 * Support for the "pids" cgroup controller has been added. It
2298 allows accounting the number of tasks in a cgroup and
2299 enforcing limits on it. This adds two new setting
2300 TasksAccounting= and TasksMax= to each unit, as well as a
2301 global option DefaultTasksAccounting=.
2303 * Support for the "net_cls" cgroup controller has been added.
2304 It allows assigning a net class ID to each task in the
2305 cgroup, which can then be used in firewall rules and traffic
2306 shaping configurations. Note that the kernel netfilter net
2307 class code does not currently work reliably for ingress
2308 packets on unestablished sockets.
2310 This adds a new config directive called NetClass= to CGroup
2311 enabled units. Allowed values are positive numbers for fixed
2312 assignments and "auto" for picking a free value
2315 * 'systemctl is-system-running' now returns 'offline' if the
2316 system is not booted with systemd. This command can now be
2317 used as a substitute for 'systemd-notify --booted'.
2319 * Watchdog timeouts have been increased to 3 minutes for all
2320 in-tree service files. Apparently, disk IO issues are more
2321 frequent than we hoped, and user reported >1 minute waiting
2324 * 'machine-id-commit' functionality has been merged into
2325 'machine-id-setup --commit'. The separate binary has been
2328 * The WorkingDirectory= directive in unit files may now be set
2329 to the special value '~'. In this case, the working
2330 directory is set to the home directory of the user
2331 configured in User=.
2333 * "machinectl shell" will now open the shell in the home
2334 directory of the selected user by default.
2336 * The CrashChVT= configuration file setting is renamed to
2337 CrashChangeVT=, following our usual logic of not
2338 abbreviating unnecessarily. The old directive is still
2339 supported for compat reasons. Also, this directive now takes
2340 an integer value between 1 and 63, or a boolean value. The
2341 formerly supported '-1' value for disabling stays around for
2344 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2345 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2346 RootDirectory= properties can now be set for transient
2349 * The systemd-analyze tool gained a new "set-log-target" verb
2350 to change the logging target the system manager logs to
2351 dynamically during runtime. This is similar to how
2352 "systemd-analyze set-log-level" already changes the log
2355 * In nspawn /sys is now mounted as tmpfs, with only a selected
2356 set of subdirectories mounted in from the real sysfs. This
2357 enhances security slightly, and is useful for ensuring user
2358 namespaces work correctly.
2360 * Support for USB FunctionFS activation has been added. This
2361 allows implementation of USB gadget services that are
2362 activated as soon as they are requested, so that they don't
2363 have to run continuously, similar to classic socket
2366 * The "systemctl exit" command now optionally takes an
2367 additional parameter that sets the exit code to return from
2368 the systemd manager when exiting. This is only relevant when
2369 running the systemd user instance, or when running the
2370 system instance in a container.
2372 * sd-bus gained the new API calls sd_bus_path_encode_many()
2373 and sd_bus_path_decode_many() that allow easy encoding and
2374 decoding of multiple identifier strings inside a D-Bus
2375 object path. Another new call sd_bus_default_flush_close()
2376 has been added to flush and close per-thread default
2379 * systemd-cgtop gained support for a -M/--machine= switch to
2380 show the control groups within a certain container only.
2382 * "systemctl kill" gained support for an optional --fail
2383 switch. If specified the requested operation will fail of no
2384 processes have been killed, because the unit had no
2385 processes attached, or similar.
2387 * A new systemd.crash_reboot=1 kernel command line option has
2388 been added that triggers a reboot after crashing. This can
2389 also be set through CrashReboot= in systemd.conf.
2391 * The RuntimeDirectory= setting now understands unit
2392 specifiers like %i or %f.
2394 * A new (still internal) library API sd-ipv4acd has been added,
2395 that implements address conflict detection for IPv4. It's
2396 based on code from sd-ipv4ll, and will be useful for
2397 detecting DHCP address conflicts.
2399 * File descriptors passed during socket activation may now be
2400 named. A new API sd_listen_fds_with_names() is added to
2401 access the names. The default names may be overridden,
2402 either in the .socket file using the FileDescriptorName=
2403 parameter, or by passing FDNAME= when storing the file
2404 descriptors using sd_notify().
2406 * systemd-networkd gained support for:
2408 - Setting the IPv6 Router Advertisement settings via
2409 IPv6AcceptRouterAdvertisements= in .network files.
2411 - Configuring the HelloTimeSec=, MaxAgeSec= and
2412 ForwardDelaySec= bridge parameters in .netdev files.
2414 - Configuring PreferredSource= for static routes in
2417 * The "ask-password" framework used to query for LUKS harddisk
2418 passwords or SSL passwords during boot gained support for
2419 caching passwords in the kernel keyring, if it is
2420 available. This makes sure that the user only has to type in
2421 a passphrase once if there are multiple objects to unlock
2422 with the same one. Previously, such password caching was
2423 available only when Plymouth was used; this moves the
2424 caching logic into the systemd codebase itself. The
2425 "systemd-ask-password" utility gained a new --keyname=
2426 switch to control which kernel keyring key to use for
2427 caching a password in. This functionality is also useful for
2428 enabling display managers such as gdm to automatically
2429 unlock the user's GNOME keyring if its passphrase, the
2430 user's password and the harddisk password are the same, if
2431 gdm-autologin is used.
2433 * When downloading tar or raw images using "machinectl
2434 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2435 file is now also downloaded, if it is available and stored
2436 next to the image file.
2438 * Units of type ".socket" gained a new boolean setting
2439 Writable= which is only useful in conjunction with
2440 ListenSpecial=. If true, enables opening the specified
2441 special file in O_RDWR mode rather than O_RDONLY mode.
2443 * systemd-rfkill has been reworked to become a singleton
2444 service that is activated through /dev/rfkill on each rfkill
2445 state change and saves the settings to disk. This way,
2446 systemd-rfkill is now compatible with devices that exist
2447 only intermittendly, and even restores state if the previous
2448 system shutdown was abrupt rather than clean.
2450 * The journal daemon gained support for vacuuming old journal
2451 files controlled by the number of files that shall remain,
2452 in addition to the already existing control by size and by
2453 date. This is useful as journal interleaving performance
2454 degrades with too many separate journal files, and allows
2455 putting an effective limit on them. The new setting defaults
2456 to 100, but this may be changed by setting SystemMaxFiles=
2457 and RuntimeMaxFiles= in journald.conf. Also, the
2458 "journalctl" tool gained the new --vacuum-files= switch to
2459 manually vacuum journal files to leave only the specified
2460 number of files in place.
2462 * udev will now create /dev/disk/by-path links for ATA devices
2463 on kernels where that is supported.
2465 * Galician, Serbian, Turkish and Korean translations were added.
2467 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2468 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2469 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2470 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2471 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2472 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2473 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2474 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2475 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2476 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2477 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2478 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2479 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2480 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2481 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2482 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2483 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2484 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2486 — Berlin, 2015-10-07
2490 * The DHCP implementation of systemd-networkd gained a set of
2493 - The DHCP server now supports emitting DNS and NTP
2494 information. It may be enabled and configured via
2495 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2496 and NTP information is enabled, but no servers are
2497 configured, the corresponding uplink information (if there
2498 is any) is propagated.
2500 - Server and client now support transmission and reception
2501 of timezone information. It can be configured via the
2502 newly introduced network options UseTimezone=,
2503 EmitTimezone=, and Timezone=. Transmission of timezone
2504 information is enabled between host and containers by
2505 default now: the container will change its local timezone
2506 to what the host has set.
2508 - Lease timeouts can now be configured via
2509 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2511 - The DHCP server improved on the stability of
2512 leases. Clients are more likely to get the same lease
2513 information back, even if the server loses state.
2515 - The DHCP server supports two new configuration options to
2516 control the lease address pool metrics, PoolOffset= and
2519 * The encapsulation limit of tunnels in systemd-networkd may
2520 now be configured via 'EncapsulationLimit='. It allows
2521 modifying the maximum additional levels of encapsulation
2522 that are permitted to be prepended to a packet.
2524 * systemd now supports the concept of user buses replacing
2525 session buses, if used with dbus-1.10 (and enabled via dbus
2526 --enable-user-session). It previously only supported this on
2527 kdbus-enabled systems, and this release expands this to
2528 'dbus-daemon' systems.
2530 * systemd-networkd now supports predictable interface names
2533 * systemd now optionally supports the new Linux kernel
2534 "unified" control group hierarchy. If enabled via the kernel
2535 command-line option 'systemd.unified_cgroup_hierarchy=1',
2536 systemd will try to mount the unified cgroup hierarchy
2537 directly on /sys/fs/cgroup. If not enabled, or not
2538 available, systemd will fall back to the legacy cgroup
2539 hierarchy setup, as before. Host system and containers can
2540 mix and match legacy and unified hierarchies as they
2541 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2542 environment variable to individually select the hierarchy to
2543 use for executed containers. By default, nspawn will use the
2544 unified hierarchy for the containers if the host uses the
2545 unified hierarchy, and the legacy hierarchy otherwise.
2546 Please note that at this point the unified hierarchy is an
2547 experimental kernel feature and is likely to change in one
2548 of the next kernel releases. Therefore, it should not be
2549 enabled by default in downstream distributions yet. The
2550 minimum required kernel version for the unified hierarchy to
2551 work is 4.2. Note that when the unified hierarchy is used
2552 for the first time delegated access to controllers is
2553 safe. Because of this systemd-nspawn containers will get
2554 access to controllers now, as will systemd user
2555 sessions. This means containers and user sessions may now
2556 manage their own resources, partitioning up what the system
2559 * A new special scope unit "init.scope" has been introduced
2560 that encapsulates PID 1 of the system. It may be used to
2561 determine resource usage and enforce resource limits on PID
2562 1 itself. PID 1 hence moved out of the root of the control
2565 * The cgtop tool gained support for filtering out kernel
2566 threads when counting tasks in a control group. Also, the
2567 count of processes is now recursively summed up by
2568 default. Two options -k and --recursive= have been added to
2569 revert to old behaviour. The tool has also been updated to
2570 work correctly in containers now.
2572 * systemd-nspawn's --bind= and --bind-ro= options have been
2573 extended to allow creation of non-recursive bind mounts.
2575 * libsystemd gained two new calls sd_pid_get_cgroup() and
2576 sd_peer_get_cgroup() which return the control group path of
2577 a process or peer of a connected AF_UNIX socket. This
2578 function call is particularly useful when implementing
2579 delegated subtrees support in the control group hierarchy.
2581 * The "sd-event" event loop API of libsystemd now supports
2582 correct dequeuing of real-time signals, without losing
2585 * When systemd requests a PolicyKit decision when managing
2586 units it will now add additional fields to the request,
2587 including unit name and desired operation. This enables more
2588 powerful PolicyKit policies, that make decisions depending
2589 on these parameters.
2591 * nspawn learnt support for .nspawn settings files, that may
2592 accompany the image files or directories of containers, and
2593 may contain additional settings for the container. This is
2594 an alternative to configuring container parameters via the
2595 nspawn command line.
2597 Contributions from: Cristian Rodríguez, Daniel Mack, David
2598 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2599 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2600 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2601 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2602 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2603 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2604 Andersen, Tom Gundersen, Torstein Husebø
2606 — Berlin, 2015-09-08
2610 * machinectl gained a new verb 'shell' which opens a fresh
2611 shell on the target container or the host. It is similar to
2612 the existing 'login' command of machinectl, but spawns the
2613 shell directly without prompting for username or
2614 password. The pseudo machine '.host' now refers to the local
2615 host and is used by default. Hence, 'machinectl shell' can
2616 be used as replacement for 'su -' which spawns a session as
2617 a fresh systemd unit in a way that is fully isolated from
2618 the originating session.
2620 * systemd-networkd learned to cope with private-zone DHCP
2621 options and allows other programs to query the values.
2623 * SELinux access control when enabling/disabling units is no
2624 longer enforced with this release. The previous
2625 implementation was incorrect, and a new corrected
2626 implementation is not yet available. As unit file operations
2627 are still protected via PolicyKit and D-Bus policy this is
2628 not a security problem. Yet, distributions which care about
2629 optimal SELinux support should probably not stabilize on
2632 * sd-bus gained support for matches of type "arg0has=", that
2633 test for membership of strings in string arrays sent in bus
2636 * systemd-resolved now dumps the contents of its DNS and LLMNR
2637 caches to the logs on reception of the SIGUSR1 signal. This
2638 is useful to debug DNS behaviour.
2640 * The coredumpctl tool gained a new --directory= option to
2641 operate on journal files in a specific directory.
2643 * "systemctl reboot" and related commands gained a new
2644 "--message=" option which may be used to set a free-text
2645 wall message when shutting down or rebooting the
2646 system. This message is also logged, which is useful for
2647 figuring out the reason for a reboot or shutdown a
2650 * The "systemd-resolve-host" tool's -i switch now takes
2651 network interface numbers as alternative to interface names.
2653 * A new unit file setting for services has been introduced:
2654 UtmpMode= allows configuration of how precisely systemd
2655 handles utmp and wtmp entries for the service if this is
2656 enabled. This allows writing services that appear similar to
2657 user sessions in the output of the "w", "who", "last" and
2660 * systemd-resolved will now locally synthesize DNS resource
2661 records for the "localhost" and "gateway" domains as well as
2662 the local hostname. This should ensure that clients querying
2663 RRs via resolved will get similar results as those going via
2664 NSS, if nss-myhostname is enabled.
2666 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2667 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2668 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2669 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2670 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2671 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2672 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2673 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2674 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2675 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2676 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2677 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2679 — Berlin, 2015-08-27
2683 * The systemd-efi-boot-generator functionality was merged into
2684 systemd-gpt-auto-generator.
2686 * systemd-networkd now supports Group Policy for vxlan
2687 devices. It can be enabled via the new boolean configuration
2688 option called 'GroupPolicyExtension='.
2690 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2691 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2692 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2694 — Berlin, 2015-07-31
2698 * The python-systemd code has been removed from the systemd repository.
2699 A new repository has been created which accommodates the code from
2700 now on, and we kindly ask distributions to create a separate package
2701 for this: https://github.com/systemd/python-systemd
2703 * The systemd daemon will now reload its main configuration
2704 (/etc/systemd/system.conf) on daemon-reload.
2706 * sd-dhcp now exposes vendor specific extensions via
2707 sd_dhcp_lease_get_vendor_specific().
2709 * systemd-networkd gained a number of new configuration options.
2711 - A new boolean configuration option for TAP devices called
2712 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2713 device, thus allowing to send and receive GSO packets.
2715 - A new tunnel configuration option called 'CopyDSCP='.
2716 If enabled, the DSCP field of ip6 tunnels is copied into the
2717 decapsulated packet.
2719 - A set of boolean bridge configuration options were added.
2720 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2721 and 'UnicastFlood=' are now parsed by networkd and applied to the
2722 respective bridge link device via the respective IFLA_BRPORT_*
2725 - A new string configuration option to override the hostname sent
2726 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2727 is true, networkd will use the configured hostname instead of the
2728 system hostname when sending DHCP requests.
2730 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2731 networkd will configure the IPv6 flow-label of the tunnel device
2732 according to RFC2460.
2734 - The 'macvtap' virtual network devices are now supported, similar to
2735 the already supported 'macvlan' devices.
2737 * systemd-resolved now implements RFC5452 to improve resilience against
2738 cache poisoning. Additionally, source port randomization is enabled
2739 by default to further protect against DNS spoofing attacks.
2741 * nss-mymachines now supports translating UIDs and GIDs of running
2742 containers with user-namespaces enabled. If a container 'foo'
2743 translates a host uid 'UID' to the container uid 'TUID', then
2744 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2745 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2746 mapped as 'vg-foo-TGID'.
2748 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2749 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2750 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2751 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2752 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2753 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2754 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2755 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2756 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2757 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2759 — Berlin, 2015-07-29
2763 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2764 There are no known issues with current sysfs, and udev does not need
2765 or should be used to work around such bugs.
2767 * udev does no longer enable USB HID power management. Several reports
2768 indicate, that some devices cannot handle that setting.
2770 * The udev accelerometer helper was removed. The functionality
2771 is now fully included in iio-sensor-proxy. But this means,
2772 older iio-sensor-proxy versions will no longer provide
2773 accelerometer/orientation data with this systemd version.
2774 Please upgrade iio-sensor-proxy to version 1.0.
2776 * networkd gained a new configuration option IPv6PrivacyExtensions=
2777 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2778 for Stateless Address") on selected networks.
2780 * For the sake of fewer build-time dependencies and less code in the
2781 main repository, the python bindings are about to be removed in the
2782 next release. A new repository has been created which accommodates
2783 the code from now on, and we kindly ask distributions to create a
2784 separate package for this. The removal will take place in v223.
2786 https://github.com/systemd/python-systemd
2788 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2789 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2790 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2791 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2792 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2793 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2794 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2795 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2796 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2797 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2799 — Berlin, 2015-07-07
2803 * The sd-bus.h and sd-event.h APIs have now been declared
2804 stable and have been added to the official interface of
2805 libsystemd.so. sd-bus implements an alternative D-Bus client
2806 library, that is relatively easy to use, very efficient and
2807 supports both classic D-Bus as well as kdbus as transport
2808 backend. sd-event is a generic event loop abstraction that
2809 is built around Linux epoll, but adds features such as event
2810 prioritization or efficient timer handling. Both APIs are good
2811 choices for C programs looking for a bus and/or event loop
2812 implementation that is minimal and does not have to be
2813 portable to other kernels.
2815 * kdbus support is no longer compile-time optional. It is now
2816 always built-in. However, it can still be disabled at
2817 runtime using the kdbus=0 kernel command line setting, and
2818 that setting may be changed to default to off, by specifying
2819 --disable-kdbus at build-time. Note though that the kernel
2820 command line setting has no effect if the kdbus.ko kernel
2821 module is not installed, in which case kdbus is (obviously)
2822 also disabled. We encourage all downstream distributions to
2823 begin testing kdbus by adding it to the kernel images in the
2824 development distributions, and leaving kdbus support in
2827 * The minimal required util-linux version has been bumped to
2830 * Support for chkconfig (--enable-chkconfig) was removed in
2831 favor of calling an abstraction tool
2832 /lib/systemd/systemd-sysv-install. This needs to be
2833 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2834 in README for details.
2836 * If there's a systemd unit and a SysV init script for the
2837 same service name, and the user executes "systemctl enable"
2838 for it (or a related call), then this will now enable both
2839 (or execute the related operation on both), not just the
2842 * The libudev API documentation has been converted from gtkdoc
2845 * gudev has been removed from the systemd tree, it is now an
2848 * The systemd-cgtop tool learnt a new --raw switch to generate
2849 "raw" (machine parsable) output.
2851 * networkd's IPForwarding= .network file setting learnt the
2852 new setting "kernel", which ensures that networkd does not
2853 change the IP forwarding sysctl from the default kernel
2856 * The systemd-logind bus API now exposes a new boolean
2857 property "Docked" that reports whether logind considers the
2858 system "docked", i.e. connected to a docking station or not.
2860 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2861 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2862 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2863 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2864 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2865 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2866 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2867 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2868 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2869 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2870 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2871 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2872 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2873 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2874 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2875 Fink, Zbigniew Jędrzejewski-Szmek
2877 — Berlin, 2015-06-19
2881 * The gudev library has been extracted into a separate repository
2882 available at: https://git.gnome.org/browse/libgudev/
2883 It is now managed as part of the Gnome project. Distributions
2884 are recommended to pass --disable-gudev to systemd and use
2885 gudev from the Gnome project instead. gudev is still included
2886 in systemd, for now. It will be removed soon, though. Please
2887 also see the announcement-thread on systemd-devel:
2888 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2890 * systemd now exposes a CPUUsageNSec= property for each
2891 service unit on the bus, that contains the overall consumed
2892 CPU time of a service (the sum of what each process of the
2893 service consumed). This value is only available if
2894 CPUAccounting= is turned on for a service, and is then shown
2895 in the "systemctl status" output.
2897 * Support for configuring alternative mappings of the old SysV
2898 runlevels to systemd targets has been removed. They are now
2899 hardcoded in a way that runlevels 2, 3, 4 all map to
2900 multi-user.target and 5 to graphical.target (which
2901 previously was already the default behaviour).
2903 * The auto-mounter logic gained support for mount point
2904 expiry, using a new TimeoutIdleSec= setting in .automount
2905 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2907 * The EFI System Partition (ESP) as mounted to /boot by
2908 systemd-efi-boot-generator will now be unmounted
2909 automatically after 2 minutes of not being used. This should
2910 minimize the risk of ESP corruptions.
2912 * New /etc/fstab options x-systemd.requires= and
2913 x-systemd.requires-mounts-for= are now supported to express
2914 additional dependencies for mounts. This is useful for
2915 journalling file systems that support external journal
2916 devices or overlay file systems that require underlying file
2917 systems to be mounted.
2919 * systemd does not support direct live-upgrades (via systemctl
2920 daemon-reexec) from versions older than v44 anymore. As no
2921 distribution we are aware of shipped such old versions in a
2922 stable release this should not be problematic.
2924 * When systemd forks off a new per-connection service instance
2925 it will now set the $REMOTE_ADDR environment variable to the
2926 remote IP address, and $REMOTE_PORT environment variable to
2927 the remote IP port. This behaviour is similar to the
2928 corresponding environment variables defined by CGI.
2930 * systemd-networkd gained support for uplink failure
2931 detection. The BindCarrier= option allows binding interface
2932 configuration dynamically to the link sense of other
2933 interfaces. This is useful to achieve behaviour like in
2936 * systemd-networkd gained support for configuring the DHCP
2937 client identifier to use when requesting leases.
2939 * systemd-networkd now has a per-network UseNTP= option to
2940 configure whether NTP server information acquired via DHCP
2941 is passed on to services like systemd-timesyncd.
2943 * systemd-networkd gained support for vti6 tunnels.
2945 * Note that systemd-networkd manages the sysctl variable
2946 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2947 it is configured for since v219. The variable controls IP
2948 forwarding, and is a per-interface alternative to the global
2949 /proc/sys/net/ipv[46]/ip_forward. This setting is
2950 configurable in the IPForward= option, which defaults to
2951 "no". This means if networkd is used for an interface it is
2952 no longer sufficient to set the global sysctl option to turn
2953 on IP forwarding! Instead, the .network file option
2954 IPForward= needs to be turned on! Note that the
2955 implementation of this behaviour was broken in v219 and has
2958 * Many bonding and vxlan options are now configurable in
2961 * systemd-nspawn gained a new --property= setting to set unit
2962 properties for the container scope. This is useful for
2963 setting resource parameters (e.g. "CPUShares=500") on
2964 containers started from the command line.
2966 * systemd-nspawn gained a new --private-users= switch to make
2967 use of user namespacing available on recent Linux kernels.
2969 * systemd-nspawn may now be called as part of a shell pipeline
2970 in which case the pipes used for stdin and stdout are passed
2971 directly to the process invoked in the container, without
2972 indirection via a pseudo tty.
2974 * systemd-nspawn gained a new switch to control the UNIX
2975 signal to use when killing the init process of the container
2978 * systemd-nspawn gained a new --overlay= switch for mounting
2979 overlay file systems into the container using the new kernel
2982 * When a container image is imported via systemd-importd and
2983 the host file system is not btrfs, a loopback block device
2984 file is created in /var/lib/machines.raw with a btrfs file
2985 system inside. It is then mounted to /var/lib/machines to
2986 enable btrfs features for container management. The loopback
2987 file and btrfs file system is grown as needed when container
2988 images are imported via systemd-importd.
2990 * systemd-machined/systemd-importd gained support for btrfs
2991 quota, to enforce container disk space limits on disk. This
2992 is exposed in "machinectl set-limit".
2994 * systemd-importd now can import containers from local .tar,
2995 .raw and .qcow2 images, and export them to .tar and .raw. It
2996 can also import dkr v2 images now from the network (on top
2999 * systemd-importd gained support for verifying downloaded
3000 images with gpg2 (previously only gpg1 was supported).
3002 * systemd-machined, systemd-logind, systemd: most bus calls
3003 are now accessible to unprivileged processes via
3004 PolicyKit. Also, systemd-logind will now allow users to kill
3005 their own sessions without further privileges or
3008 * systemd-shutdownd has been removed. This service was
3009 previously responsible for implementing scheduled shutdowns
3010 as exposed in /usr/bin/shutdown's time parameter. This
3011 functionality has now been moved into systemd-logind and is
3012 accessible via a bus interface.
3014 * "systemctl reboot" gained a new switch --firmware-setup that
3015 can be used to reboot into the EFI firmware setup, if that
3016 is available. systemd-logind now exposes an API on the bus
3017 to trigger such reboots, in case graphical desktop UIs want
3018 to cover this functionality.
3020 * "systemctl enable", "systemctl disable" and "systemctl mask"
3021 now support a new "--now" switch. If specified the units
3022 that are enabled will also be started, and the ones
3023 disabled/masked also stopped.
3025 * The Gummiboot EFI boot loader tool has been merged into
3026 systemd, and renamed to "systemd-boot". The bootctl tool has been
3027 updated to support systemd-boot.
3029 * An EFI kernel stub has been added that may be used to create
3030 kernel EFI binaries that contain not only the actual kernel,
3031 but also an initrd, boot splash, command line and OS release
3032 information. This combined binary can then be signed as a
3033 single image, so that the firmware can verify it all in one
3034 step. systemd-boot has special support for EFI binaries created
3035 like this and can extract OS release information from them
3036 and show them in the boot menu. This functionality is useful
3037 to implement cryptographically verified boot schemes.
3039 * Optional support has been added to systemd-fsck to pass
3040 fsck's progress report to an AF_UNIX socket in the file
3043 * udev will no longer create device symlinks for all block
3044 devices by default. A blacklist for excluding special block
3045 devices from this logic has been turned into a whitelist
3046 that requires picking block devices explicitly that require
3049 * A new (currently still internal) API sd-device.h has been
3050 added to libsystemd. This modernized API is supposed to
3051 replace libudev eventually. In fact, already much of libudev
3052 is now just a wrapper around sd-device.h.
3054 * A new hwdb database for storing metadata about pointing
3055 stick devices has been added.
3057 * systemd-tmpfiles gained support for setting file attributes
3058 similar to the "chattr" tool with new 'h' and 'H' lines.
3060 * systemd-journald will no longer unconditionally set the
3061 btrfs NOCOW flag on new journal files. This is instead done
3062 with tmpfiles snippet using the new 'h' line type. This
3063 allows easy disabling of this logic, by masking the
3064 journal-nocow.conf tmpfiles file.
3066 * systemd-journald will now translate audit message types to
3067 human readable identifiers when writing them to the
3068 journal. This should improve readability of audit messages.
3070 * The LUKS logic gained support for the offset= and skip=
3071 options in /etc/crypttab, as previously implemented by
3074 * /usr/lib/os-release gained a new optional field VARIANT= for
3075 distributions that support multiple variants (such as a
3076 desktop edition, a server edition, ...)
3078 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3079 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3080 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3081 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3082 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3083 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3084 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3085 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3086 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3087 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3088 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3089 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3090 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3091 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3092 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3093 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3094 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3095 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3096 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3097 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3098 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3099 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3100 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3101 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3102 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3103 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3104 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3106 — Berlin, 2015-05-22
3110 * Introduce a new API "sd-hwdb.h" for querying the hardware
3111 metadata database. With this minimal interface one can query
3112 and enumerate the udev hwdb, decoupled from the old libudev
3113 library. libudev's interface for this is now only a wrapper
3114 around sd-hwdb. A new tool systemd-hwdb has been added to
3115 interface with and update the database.
3117 * When any of systemd's tools copies files (for example due to
3118 tmpfiles' C lines) a btrfs reflink will attempted first,
3119 before bytewise copying is done.
3121 * systemd-nspawn gained a new --ephemeral switch. When
3122 specified a btrfs snapshot is taken of the container's root
3123 directory, and immediately removed when the container
3124 terminates again. Thus, a container can be started whose
3125 changes never alter the container's root directory, and are
3126 lost on container termination. This switch can also be used
3127 for starting a container off the root file system of the
3128 host without affecting the host OS. This switch is only
3129 available on btrfs file systems.
3131 * systemd-nspawn gained a new --template= switch. It takes the
3132 path to a container tree to use as template for the tree
3133 specified via --directory=, should that directory be
3134 missing. This allows instantiating containers dynamically,
3135 on first run. This switch is only available on btrfs file
3138 * When a .mount unit refers to a mount point on which multiple
3139 mounts are stacked, and the .mount unit is stopped all of
3140 the stacked mount points will now be unmounted until no
3141 mount point remains.
3143 * systemd now has an explicit notion of supported and
3144 unsupported unit types. Jobs enqueued for unsupported unit
3145 types will now fail with an "unsupported" error code. More
3146 specifically .swap, .automount and .device units are not
3147 supported in containers, .busname units are not supported on
3148 non-kdbus systems. .swap and .automount are also not
3149 supported if their respective kernel compile time options
3152 * machinectl gained support for two new "copy-from" and
3153 "copy-to" commands for copying files from a running
3154 container to the host or vice versa.
3156 * machinectl gained support for a new "bind" command to bind
3157 mount host directories into local containers. This is
3158 currently only supported for nspawn containers.
3160 * networkd gained support for configuring bridge forwarding
3161 database entries (fdb) from .network files.
3163 * A new tiny daemon "systemd-importd" has been added that can
3164 download container images in tar, raw, qcow2 or dkr formats,
3165 and make them available locally in /var/lib/machines, so
3166 that they can run as nspawn containers. The daemon can GPG
3167 verify the downloads (not supported for dkr, since it has no
3168 provisions for verifying downloads). It will transparently
3169 decompress bz2, xz, gzip compressed downloads if necessary,
3170 and restore sparse files on disk. The daemon uses privilege
3171 separation to ensure the actual download logic runs with
3172 fewer privileges than the daemon itself. machinectl has
3173 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3174 make the functionality of importd available to the
3175 user. With this in place the Fedora and Ubuntu "Cloud"
3176 images can be downloaded and booted as containers unmodified
3177 (the Fedora images lack the appropriate GPG signature files
3178 currently, so they cannot be verified, but this will change
3179 soon, hopefully). Note that downloading images is currently
3180 only fully supported on btrfs.
3182 * machinectl is now able to list container images found in
3183 /var/lib/machines, along with some metadata about sizes of
3184 disk and similar. If the directory is located on btrfs and
3185 quota is enabled, this includes quota display. A new command
3186 "image-status" has been added that shows additional
3187 information about images.
3189 * machinectl is now able to clone container images
3190 efficiently, if the underlying file system (btrfs) supports
3191 it, with the new "machinectl clone" command. It also
3192 gained commands for renaming and removing images, as well as
3193 marking them read-only or read-write (supported also on
3194 legacy file systems).
3196 * networkd gained support for collecting LLDP network
3197 announcements, from hardware that supports this. This is
3198 shown in networkctl output.
3200 * systemd-run gained support for a new -t (--pty) switch for
3201 invoking a binary on a pty whose input and output is
3202 connected to the invoking terminal. This allows executing
3203 processes as system services while interactively
3204 communicating with them via the terminal. Most interestingly
3205 this is supported across container boundaries. Invoking
3206 "systemd-run -t /bin/bash" is an alternative to running a
3207 full login session, the difference being that the former
3208 will not register a session, nor go through the PAM session
3211 * tmpfiles gained support for a new "v" line type for creating
3212 btrfs subvolumes. If the underlying file system is a legacy
3213 file system, this automatically degrades to creating a
3214 normal directory. Among others /var/lib/machines is now
3215 created like this at boot, should it be missing.
3217 * The directory /var/lib/containers/ has been deprecated and
3218 been replaced by /var/lib/machines. The term "machines" has
3219 been used in the systemd context as generic term for both
3220 VMs and containers, and hence appears more appropriate for
3221 this, as the directory can also contain raw images bootable
3224 * systemd-nspawn when invoked with -M but without --directory=
3225 or --image= is now capable of searching for the container
3226 root directory, subvolume or disk image automatically, in
3227 /var/lib/machines. systemd-nspawn@.service has been updated
3228 to make use of this, thus allowing it to be used for raw
3231 * A new machines.target unit has been introduced that is
3232 supposed to group all containers/VMs invoked as services on
3233 the system. systemd-nspawn@.service has been updated to
3234 integrate with that.
3236 * machinectl gained a new "start" command, for invoking a
3237 container as a service. "machinectl start foo" is mostly
3238 equivalent to "systemctl start systemd-nspawn@foo.service",
3239 but handles escaping in a nicer way.
3241 * systemd-nspawn will now mount most of the cgroupfs tree
3242 read-only into each container, with the exception of the
3243 container's own subtree in the name=systemd hierarchy.
3245 * journald now sets the special FS_NOCOW file flag for its
3246 journal files. This should improve performance on btrfs, by
3247 avoiding heavy fragmentation when journald's write-pattern
3248 is used on COW file systems. It degrades btrfs' data
3249 integrity guarantees for the files to the same levels as for
3250 ext3/ext4 however. This should be OK though as journald does
3251 its own data integrity checks and all its objects are
3252 checksummed on disk. Also, journald should handle btrfs disk
3253 full events a lot more gracefully now, by processing SIGBUS
3254 errors, and not relying on fallocate() anymore.
3256 * When journald detects that journal files it is writing to
3257 have been deleted it will immediately start new journal
3260 * systemd now provides a way to store file descriptors
3261 per-service in PID 1. This is useful for daemons to ensure
3262 that fds they require are not lost during a daemon
3263 restart. The fds are passed to the daemon on the next
3264 invocation in the same way socket activation fds are
3265 passed. This is now used by journald to ensure that the
3266 various sockets connected to all the system's stdout/stderr
3267 are not lost when journald is restarted. File descriptors
3268 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3269 an extension to sd_notify(). Note that a limit is enforced
3270 on the number of fds a service can store in PID 1, and it
3271 defaults to 0, so that no fds may be stored, unless this is
3272 explicitly turned on.
3274 * The default TERM variable to use for units connected to a
3275 terminal, when no other value is explicitly is set is now
3276 vt220 rather than vt102. This should be fairly safe still,
3277 but allows PgUp/PgDn work.
3279 * The /etc/crypttab option header= as known from Debian is now
3282 * "loginctl user-status" and "loginctl session-status" will
3283 now show the last 10 lines of log messages of the
3284 user/session following the status output. Similar,
3285 "machinectl status" will show the last 10 log lines
3286 associated with a virtual machine or container
3287 service. (Note that this is usually not the log messages
3288 done in the VM/container itself, but simply what the
3289 container manager logs. For nspawn this includes all console
3292 * "loginctl session-status" without further argument will now
3293 show the status of the session of the caller. Similar,
3294 "lock-session", "unlock-session", "activate",
3295 "enable-linger", "disable-linger" may now be called without
3296 session/user parameter in which case they apply to the
3297 caller's session/user.
3299 * An X11 session scriptlet is now shipped that uploads
3300 $DISPLAY and $XAUTHORITY into the environment of the systemd
3301 --user daemon if a session begins. This should improve
3302 compatibility with X11 enabled applications run as systemd
3305 * Generators are now subject to masking via /etc and /run, the
3306 same way as unit files.
3308 * networkd .network files gained support for configuring
3309 per-link IPv4/IPv6 packet forwarding as well as IPv4
3310 masquerading. This is by default turned on for veth links to
3311 containers, as registered by systemd-nspawn. This means that
3312 nspawn containers run with --network-veth will now get
3313 automatic routed access to the host's networks without any
3314 further configuration or setup, as long as networkd runs on
3317 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3318 or UDP posts of a container on the host. With this in place
3319 it is possible to run containers with private veth links
3320 (--network-veth), and have their functionality exposed on
3321 the host as if their services were running directly on the
3324 * systemd-nspawn's --network-veth switch now gained a short
3325 version "-n", since with the changes above it is now truly
3326 useful out-of-the-box. The systemd-nspawn@.service has been
3327 updated to make use of it too by default.
3329 * systemd-nspawn will now maintain a per-image R/W lock, to
3330 ensure that the same image is not started more than once
3331 writable. (It's OK to run an image multiple times
3332 simultaneously in read-only mode.)
3334 * systemd-nspawn's --image= option is now capable of
3335 dissecting and booting MBR and GPT disk images that contain
3336 only a single active Linux partition. Previously it
3337 supported only GPT disk images with proper GPT type
3338 IDs. This allows running cloud images from major
3339 distributions directly with systemd-nspawn, without
3342 * In addition to collecting mouse dpi data in the udev
3343 hardware database, there's now support for collecting angle
3344 information for mouse scroll wheels. The database is
3345 supposed to guarantee similar scrolling behavior on mice
3346 that it knows about. There's also support for collecting
3347 information about Touchpad types.
3349 * udev's input_id built-in will now also collect touch screen
3350 dimension data and attach it to probed devices.
3352 * /etc/os-release gained support for a Distribution Privacy
3355 * networkd gained support for creating "ipvlan", "gretap",
3356 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3358 * systemd-tmpfiles gained support for "a" lines for setting
3361 * systemd-nspawn will now mount /tmp in the container to
3362 tmpfs, automatically.
3364 * systemd now exposes the memory.usage_in_bytes cgroup
3365 attribute and shows it for each service in the "systemctl
3366 status" output, if available.
3368 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3369 immediate reboot is triggered. This useful if shutdown is
3370 hung and is unable to complete, to expedite the
3371 operation. Note that this kind of reboot will still unmount
3372 all file systems, and hence should not result in fsck being
3375 * A .device unit for an optical block device will now be
3376 considered active only when a medium is in the drive. Also,
3377 mount units are now bound to their backing devices thus
3378 triggering automatic unmounting when devices become
3379 unavailable. With this in place systemd will now
3380 automatically unmount left-over mounts when a CD-ROM is
3381 ejected or an USB stick is yanked from the system.
3383 * networkd-wait-online now has support for waiting for
3384 specific interfaces only (with globbing), and for giving up
3385 after a configurable timeout.
3387 * networkd now exits when idle. It will be automatically
3388 restarted as soon as interfaces show up, are removed or
3389 change state. networkd will stay around as long as there is
3390 at least one DHCP state machine or similar around, that keep
3393 * networkd may now configure IPv6 link-local addressing in
3394 addition to IPv4 link-local addressing.
3396 * The IPv6 "token" for use in SLAAC may now be configured for
3397 each .network interface in networkd.
3399 * Routes configured with networkd may now be assigned a scope
3402 * networkd's [Match] sections now support globbing and lists
3403 of multiple space-separated matches per item.
3405 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3406 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3407 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3408 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3409 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3410 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3411 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3412 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3413 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3414 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3415 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3416 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3417 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3418 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3419 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3420 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3421 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3422 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3423 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3424 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3425 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3426 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3427 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3428 Hoffmann, Zbigniew Jędrzejewski-Szmek
3430 — Berlin, 2015-02-16
3434 * When querying unit file enablement status (for example via
3435 "systemctl is-enabled"), a new state "indirect" is now known
3436 which indicates that a unit might not be enabled itself, but
3437 another unit listed in its Also= setting might be.
3439 * Similar to the various existing ConditionXYZ= settings for
3440 units, there are now matching AssertXYZ= settings. While
3441 failing conditions cause a unit to be skipped, but its job
3442 to succeed, failing assertions declared like this will cause
3443 a unit start operation and its job to fail.
3445 * hostnamed now knows a new chassis type "embedded".
3447 * systemctl gained a new "edit" command. When used on a unit
3448 file, this allows extending unit files with .d/ drop-in
3449 configuration snippets or editing the full file (after
3450 copying it from /usr/lib to /etc). This will invoke the
3451 user's editor (as configured with $EDITOR), and reload the
3452 modified configuration after editing.
3454 * "systemctl status" now shows the suggested enablement state
3455 for a unit, as declared in the (usually vendor-supplied)
3456 system preset files.
3458 * nss-myhostname will now resolve the single-label host name
3459 "gateway" to the locally configured default IP routing
3460 gateways, ordered by their metrics. This assigns a stable
3461 name to the used gateways, regardless which ones are
3462 currently configured. Note that the name will only be
3463 resolved after all other name sources (if nss-myhostname is
3464 configured properly) and should hence not negatively impact
3465 systems that use the single-label host name "gateway" in
3468 * systemd-inhibit now allows filtering by mode when listing
3471 * Scope and service units gained a new "Delegate" boolean
3472 property, which, when set, allows processes running inside the
3473 unit to further partition resources. This is primarily
3474 useful for systemd user instances as well as container
3477 * journald will now pick up audit messages directly from
3478 the kernel, and log them like any other log message. The
3479 audit fields are split up and fully indexed. This means that
3480 journalctl in many ways is now a (nicer!) alternative to
3481 ausearch, the traditional audit client. Note that this
3482 implements only a minimal audit client. If you want the
3483 special audit modes like reboot-on-log-overflow, please use
3484 the traditional auditd instead, which can be used in
3485 parallel to journald.
3487 * The ConditionSecurity= unit file option now understands the
3488 special string "audit" to check whether auditing is
3491 * journalctl gained two new commands --vacuum-size= and
3492 --vacuum-time= to delete old journal files until the
3493 remaining ones take up no more than the specified size on disk,
3494 or are not older than the specified time.
3496 * A new, native PPPoE library has been added to sd-network,
3497 systemd's library of light-weight networking protocols. This
3498 library will be used in a future version of networkd to
3499 enable PPPoE communication without an external pppd daemon.
3501 * The busctl tool now understands a new "capture" verb that
3502 works similar to "monitor", but writes a packet capture
3503 trace to STDOUT that can be redirected to a file which is
3504 compatible with libcap's capture file format. This can then
3505 be loaded in Wireshark and similar tools to inspect bus
3508 * The busctl tool now understands a new "tree" verb that shows
3509 the object trees of a specific service on the bus, or of all
3512 * The busctl tool now understands a new "introspect" verb that
3513 shows all interfaces and members of objects on the bus,
3514 including their signature and values. This is particularly
3515 useful to get more information about bus objects shown by
3516 the new "busctl tree" command.
3518 * The busctl tool now understands new verbs "call",
3519 "set-property" and "get-property" for invoking bus method
3520 calls, setting and getting bus object properties in a
3523 * busctl gained a new --augment-creds= argument that controls
3524 whether the tool shall augment credential information it
3525 gets from the bus with data from /proc, in a possibly
3528 * nspawn's --link-journal= switch gained two new values
3529 "try-guest" and "try-host" that work like "guest" and
3530 "host", but do not fail if the host has no persistent
3531 journalling enabled. -j is now equivalent to
3532 --link-journal=try-guest.
3534 * macvlan network devices created by nspawn will now have
3535 stable MAC addresses.
3537 * A new SmackProcessLabel= unit setting has been added, which
3538 controls the SMACK security label processes forked off by
3539 the respective unit shall use.
3541 * If compiled with --enable-xkbcommon, systemd-localed will
3542 verify x11 keymap settings by compiling the given keymap. It
3543 will spew out warnings if the compilation fails. This
3544 requires libxkbcommon to be installed.
3546 * When a coredump is collected, a larger number of metadata
3547 fields is now collected and included in the journal records
3548 created for it. More specifically, control group membership,
3549 environment variables, memory maps, working directory,
3550 chroot directory, /proc/$PID/status, and a list of open file
3551 descriptors is now stored in the log entry.
3553 * The udev hwdb now contains DPI information for mice. For
3556 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3558 * All systemd programs that read standalone configuration
3559 files in /etc now also support a corresponding series of
3560 .conf.d configuration directories in /etc/, /run/,
3561 /usr/local/lib/, /usr/lib/, and (if configured with
3562 --enable-split-usr) /lib/. In particular, the following
3563 configuration files now have corresponding configuration
3564 directories: system.conf user.conf, logind.conf,
3565 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3566 resolved.conf, timesyncd.conf, journal-remote.conf, and
3567 journal-upload.conf. Note that distributions should use the
3568 configuration directories in /usr/lib/; the directories in
3569 /etc/ are reserved for the system administrator.
3571 * systemd-rfkill will no longer take the rfkill device name
3572 into account when storing rfkill state on disk, as the name
3573 might be dynamically assigned and not stable. Instead, the
3574 ID_PATH udev variable combined with the rfkill type (wlan,
3575 bluetooth, ...) is used.
3577 * A new service systemd-machine-id-commit.service has been
3578 added. When used on systems where /etc is read-only during
3579 boot, and /etc/machine-id is not initialized (but an empty
3580 file), this service will copy the temporary machine ID
3581 created as replacement into /etc after the system is fully
3582 booted up. This is useful for systems that are freshly
3583 installed with a non-initialized machine ID, but should get
3584 a fixed machine ID for subsequent boots.
3586 * networkd's .netdev files now provide a large set of
3587 configuration parameters for VXLAN devices. Similarly, the
3588 bridge port cost parameter is now configurable in .network
3589 files. There's also new support for configuring IP source
3590 routing. networkd .link files gained support for a new
3591 OriginalName= match that is useful to match against the
3592 original interface name the kernel assigned. .network files
3593 may include MTU= and MACAddress= fields for altering the MTU
3594 and MAC address while being connected to a specific network
3597 * The LUKS logic gained supported for configuring
3598 UUID-specific key files. There's also new support for naming
3599 LUKS device from the kernel command line, using the new
3600 luks.name= argument.
3602 * Timer units may now be transiently created via the bus API
3603 (this was previously already available for scope and service
3604 units). In addition it is now possible to create multiple
3605 transient units at the same time with a single bus call. The
3606 "systemd-run" tool has been updated to make use of this for
3607 running commands on a specified time, in at(1)-style.
3609 * tmpfiles gained support for "t" lines, for assigning
3610 extended attributes to files. Among other uses this may be
3611 used to assign SMACK labels to files.
3613 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3614 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3615 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3616 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3617 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3618 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3619 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3620 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3621 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3622 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3623 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3624 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3625 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3626 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3627 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3628 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3629 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3630 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3632 — Berlin, 2014-12-10
3636 * journalctl gained the new options -t/--identifier= to match
3637 on the syslog identifier (aka "tag"), as well as --utc to
3638 show log timestamps in the UTC timezone. journalctl now also
3639 accepts -n/--lines=all to disable line capping in a pager.
3641 * journalctl gained a new switch, --flush, that synchronously
3642 flushes logs from /run/log/journal to /var/log/journal if
3643 persistent storage is enabled. systemd-journal-flush.service
3644 now waits until the operation is complete.
3646 * Services can notify the manager before they start a reload
3647 (by sending RELOADING=1) or shutdown (by sending
3648 STOPPING=1). This allows the manager to track and show the
3649 internal state of daemons and closes a race condition when
3650 the process is still running but has closed its D-Bus
3653 * Services with Type=oneshot do not have to have any ExecStart
3656 * User units are now loaded also from
3657 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3658 /run/systemd/user directory that was already previously
3659 supported, but is under the control of the user.
3661 * Job timeouts (i.e. time-outs on the time a job that is
3662 queued stays in the run queue) can now optionally result in
3663 immediate reboot or power-off actions (JobTimeoutAction= and
3664 JobTimeoutRebootArgument=). This is useful on ".target"
3665 units, to limit the maximum time a target remains
3666 undispatched in the run queue, and to trigger an emergency
3667 operation in such a case. This is now used by default to
3668 turn off the system if boot-up (as defined by everything in
3669 basic.target) hangs and does not complete for at least
3670 15min. Also, if power-off or reboot hang for at least 30min
3671 an immediate power-off/reboot operation is triggered. This
3672 functionality is particularly useful to increase reliability
3673 on embedded devices, but also on laptops which might
3674 accidentally get powered on when carried in a backpack and
3675 whose boot stays stuck in a hard disk encryption passphrase
3678 * systemd-logind can be configured to also handle lid switch
3679 events even when the machine is docked or multiple displays
3680 are attached (HandleLidSwitchDocked= option).
3682 * A helper binary and a service have been added which can be
3683 used to resume from hibernation in the initramfs. A
3684 generator will parse the resume= option on the kernel
3685 command line to trigger resume.
3687 * A user console daemon systemd-consoled has been
3688 added. Currently, it is a preview, and will so far open a
3689 single terminal on each session of the user marked as
3690 Desktop=systemd-console.
3692 * Route metrics can be specified for DHCP routes added by
3695 * The SELinux context of socket-activated services can be set
3696 from the information provided by the networking stack
3697 (SELinuxContextFromNet= option).
3699 * Userspace firmware loading support has been removed and
3700 the minimum supported kernel version is thus bumped to 3.7.
3702 * Timeout for udev workers has been increased from 1 to 3
3703 minutes, but a warning will be printed after 1 minute to
3704 help diagnose kernel modules that take a long time to load.
3706 * Udev rules can now remove tags on devices with TAG-="foobar".
3708 * systemd's readahead implementation has been removed. In many
3709 circumstances it didn't give expected benefits even for
3710 rotational disk drives and was becoming less relevant in the
3711 age of SSDs. As none of the developers has been using
3712 rotating media anymore, and nobody stepped up to actively
3713 maintain this component of systemd it has now been removed.
3715 * Swap units can use Options= to specify discard options.
3716 Discard options specified for swaps in /etc/fstab are now
3719 * Docker containers are now detected as a separate type of
3722 * The Password Agent protocol gained support for queries where
3723 the user input is shown, useful e.g. for user names.
3724 systemd-ask-password gained a new --echo option to turn that
3727 * The default sysctl.d/ snippets will now set:
3729 net.core.default_qdisc = fq_codel
3731 This selects Fair Queuing Controlled Delay as the default
3732 queuing discipline for network interfaces. fq_codel helps
3733 fight the network bufferbloat problem. It is believed to be
3734 a good default with no tuning required for most workloads.
3735 Downstream distributions may override this choice. On 10Gbit
3736 servers that do not do forwarding, "fq" may perform better.
3737 Systems without a good clocksource should use "pfifo_fast".
3739 * If kdbus is enabled during build a new option BusPolicy= is
3740 available for service units, that allows locking all service
3741 processes into a stricter bus policy, in order to limit
3742 access to various bus services, or even hide most of them
3743 from the service's view entirely.
3745 * networkctl will now show the .network and .link file
3746 networkd has applied to a specific interface.
3748 * sd-login gained a new API call sd_session_get_desktop() to
3749 query which desktop environment has been selected for a
3752 * UNIX utmp support is now compile-time optional to support
3753 legacy-free systems.
3755 * systemctl gained two new commands "add-wants" and
3756 "add-requires" for pulling in units from specific targets
3759 * If the word "rescue" is specified on the kernel command line
3760 the system will now boot into rescue mode (aka
3761 rescue.target), which was previously available only by
3762 specifying "1" or "systemd.unit=rescue.target" on the kernel
3763 command line. This new kernel command line option nicely
3764 mirrors the already existing "emergency" kernel command line
3767 * New kernel command line options mount.usr=, mount.usrflags=,
3768 mount.usrfstype= have been added that match root=, rootflags=,
3769 rootfstype= but allow mounting a specific file system to
3772 * The $NOTIFY_SOCKET is now also passed to control processes of
3773 services, not only the main process.
3775 * This version reenables support for fsck's -l switch. This
3776 means at least version v2.25 of util-linux is required for
3777 operation, otherwise dead-locks on device nodes may
3778 occur. Again: you need to update util-linux to at least
3779 v2.25 when updating systemd to v217.
3781 * The "multi-seat-x" tool has been removed from systemd, as
3782 its functionality has been integrated into X servers 1.16,
3783 and the tool is hence redundant. It is recommended to update
3784 display managers invoking this tool to simply invoke X
3785 directly from now on, again.
3787 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3788 message flag has been added for all of systemd's PolicyKit
3789 authenticated method calls has been added. In particular
3790 this now allows optional interactive authorization via
3791 PolicyKit for many of PID1's privileged operations such as
3792 unit file enabling and disabling.
3794 * "udevadm hwdb --update" learnt a new switch "--usr" for
3795 placing the rebuilt hardware database in /usr instead of
3796 /etc. When used only hardware database entries stored in
3797 /usr will be used, and any user database entries in /etc are
3798 ignored. This functionality is useful for vendors to ship a
3799 pre-built database on systems where local configuration is
3800 unnecessary or unlikely.
3802 * Calendar time specifications in .timer units now also
3803 understand the strings "semi-annually", "quarterly" and
3804 "minutely" as shortcuts (in addition to the preexisting
3805 "anually", "hourly", ...).
3807 * systemd-tmpfiles will now correctly create files in /dev
3808 at boot which are marked for creation only at boot. It is
3809 recommended to always create static device nodes with 'c!'
3810 and 'b!', so that they are created only at boot and not
3811 overwritten at runtime.
3813 * When the watchdog logic is used for a service (WatchdogSec=)
3814 and the watchdog timeout is hit the service will now be
3815 terminated with SIGABRT (instead of just SIGTERM), in order
3816 to make sure a proper coredump and backtrace is
3817 generated. This ensures that hanging services will result in
3818 similar coredump/backtrace behaviour as services that hit a
3821 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3822 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3823 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3824 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3825 Herrmann, David Sommerseth, David Strauss, Emil Renner
3826 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3827 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3828 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3829 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3830 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3831 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3832 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3833 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3834 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3835 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3836 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3837 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3838 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3839 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3840 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3841 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3844 — Berlin, 2014-10-28
3848 * timedated no longer reads NTP implementation unit names from
3849 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3850 implementations should add a
3852 Conflicts=systemd-timesyncd.service
3854 to their unit files to take over and replace systemd's NTP
3855 default functionality.
3857 * systemd-sysusers gained a new line type "r" for configuring
3858 which UID/GID ranges to allocate system users/groups
3859 from. Lines of type "u" may now add an additional column
3860 that specifies the home directory for the system user to be
3861 created. Also, systemd-sysusers may now optionally read user
3862 information from STDIN instead of a file. This is useful for
3863 invoking it from RPM preinst scriptlets that need to create
3864 users before the first RPM file is installed since these
3865 files might need to be owned by them. A new
3866 %sysusers_create_inline RPM macro has been introduced to do
3867 just that. systemd-sysusers now updates the shadow files as
3868 well as the user/group databases, which should enhance
3869 compatibility with certain tools like grpck.
3871 * A number of bus APIs of PID 1 now optionally consult
3872 PolicyKit to permit access for otherwise unprivileged
3873 clients under certain conditions. Note that this currently
3874 doesn't support interactive authentication yet, but this is
3875 expected to be added eventually, too.
3877 * /etc/machine-info now has new fields for configuring the
3878 deployment environment of the machine, as well as the
3879 location of the machine. hostnamectl has been updated with
3880 new command to update these fields.
3882 * systemd-timesyncd has been updated to automatically acquire
3883 NTP server information from systemd-networkd, which might
3884 have been discovered via DHCP.
3886 * systemd-resolved now includes a caching DNS stub resolver
3887 and a complete LLMNR name resolution implementation. A new
3888 NSS module "nss-resolve" has been added which can be used
3889 instead of glibc's own "nss-dns" to resolve hostnames via
3890 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3891 be resolved via systemd-resolved D-Bus APIs. In contrast to
3892 the glibc internal resolver systemd-resolved is aware of
3893 multi-homed system, and keeps DNS server and caches separate
3894 and per-interface. Queries are sent simultaneously on all
3895 interfaces that have DNS servers configured, in order to
3896 properly handle VPNs and local LANs which might resolve
3897 separate sets of domain names. systemd-resolved may acquire
3898 DNS server information from systemd-networkd automatically,
3899 which in turn might have discovered them via DHCP. A tool
3900 "systemd-resolve-host" has been added that may be used to
3901 query the DNS logic in resolved. systemd-resolved implements
3902 IDNA and automatically uses IDNA or UTF-8 encoding depending
3903 on whether classic DNS or LLMNR is used as transport. In the
3904 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3905 implementation to systemd-resolved.
3907 * A new NSS module nss-mymachines has been added, that
3908 automatically resolves the names of all local registered
3909 containers to their respective IP addresses.
3911 * A new client tool "networkctl" for systemd-networkd has been
3912 added. It currently is entirely passive and will query
3913 networking configuration from udev, rtnetlink and networkd,
3914 and present it to the user in a very friendly
3915 way. Eventually, we hope to extend it to become a full
3916 control utility for networkd.
3918 * .socket units gained a new DeferAcceptSec= setting that
3919 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3920 TCP. Similarly, support for controlling TCP keep-alive
3921 settings has been added (KeepAliveTimeSec=,
3922 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3923 turning off Nagle's algorithm on TCP has been added
3926 * logind learned a new session type "web", for use in projects
3927 like Cockpit which register web clients as PAM sessions.
3929 * timer units with at least one OnCalendar= setting will now
3930 be started only after time-sync.target has been
3931 reached. This way they will not elapse before the system
3932 clock has been corrected by a local NTP client or
3933 similar. This is particular useful on RTC-less embedded
3934 machines, that come up with an invalid system clock.
3936 * systemd-nspawn's --network-veth= switch should now result in
3937 stable MAC addresses for both the outer and the inner side
3940 * systemd-nspawn gained a new --volatile= switch for running
3941 container instances with /etc or /var unpopulated.
3943 * The kdbus client code has been updated to use the new Linux
3944 3.17 memfd subsystem instead of the old kdbus-specific one.
3946 * systemd-networkd's DHCP client and server now support
3947 FORCERENEW. There are also new configuration options to
3948 configure the vendor client identifier and broadcast mode
3951 * systemd will no longer inform the kernel about the current
3952 timezone, as this is necessarily incorrect and racy as the
3953 kernel has no understanding of DST and similar
3954 concepts. This hence means FAT timestamps will be always
3955 considered UTC, similar to what Android is already
3956 doing. Also, when the RTC is configured to the local time
3957 (rather than UTC) systemd will never synchronize back to it,
3958 as this might confuse Windows at a later boot.
3960 * systemd-analyze gained a new command "verify" for offline
3961 validation of unit files.
3963 * systemd-networkd gained support for a couple of additional
3964 settings for bonding networking setups. Also, the metric for
3965 statically configured routes may now be configured. For
3966 network interfaces where this is appropriate the peer IP
3967 address may now be configured.
3969 * systemd-networkd's DHCP client will no longer request
3970 broadcasting by default, as this tripped up some networks.
3971 For hardware where broadcast is required the feature should
3972 be switched back on using RequestBroadcast=yes.
3974 * systemd-networkd will now set up IPv4LL addresses (when
3975 enabled) even if DHCP is configured successfully.
3977 * udev will now default to respect network device names given
3978 by the kernel when the kernel indicates that these are
3979 predictable. This behavior can be tweaked by changing
3980 NamePolicy= in the relevant .link file.
3982 * A new library systemd-terminal has been added that
3983 implements full TTY stream parsing and rendering. This
3984 library is supposed to be used later on for implementing a
3985 full userspace VT subsystem, replacing the current kernel
3988 * A new tool systemd-journal-upload has been added to push
3989 journal data to a remote system running
3990 systemd-journal-remote.
3992 * journald will no longer forward all local data to another
3993 running syslog daemon. This change has been made because
3994 rsyslog (which appears to be the most commonly used syslog
3995 implementation these days) no longer makes use of this, and
3996 instead pulls the data out of the journal on its own. Since
3997 forwarding the messages to a non-existent syslog server is
3998 more expensive than we assumed we have now turned this
3999 off. If you run a syslog server that is not a recent rsyslog
4000 version, you have to turn this option on again
4001 (ForwardToSyslog= in journald.conf).
4003 * journald now optionally supports the LZ4 compressor for
4004 larger journal fields. This compressor should perform much
4005 better than XZ which was the previous default.
4007 * machinectl now shows the IP addresses of local containers,
4008 if it knows them, plus the interface name of the container.
4010 * A new tool "systemd-escape" has been added that makes it
4011 easy to escape strings to build unit names and similar.
4013 * sd_notify() messages may now include a new ERRNO= field
4014 which is parsed and collected by systemd and shown among the
4015 "systemctl status" output for a service.
4017 * A new component "systemd-firstboot" has been added that
4018 queries the most basic systemd information (timezone,
4019 hostname, root password) interactively on first
4020 boot. Alternatively it may also be used to provision these
4021 things offline on OS images installed into directories.
4023 * The default sysctl.d/ snippets will now set
4025 net.ipv4.conf.default.promote_secondaries=1
4027 This has the benefit of no flushing secondary IP addresses
4028 when primary addresses are removed.
4030 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4031 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4032 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4033 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4034 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4035 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4036 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4037 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4038 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4039 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4040 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4041 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4042 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4043 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4044 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4046 — Berlin, 2014-08-19
4050 * A new tool systemd-sysusers has been added. This tool
4051 creates system users and groups in /etc/passwd and
4052 /etc/group, based on static declarative system user/group
4053 definitions in /usr/lib/sysusers.d/. This is useful to
4054 enable factory resets and volatile systems that boot up with
4055 an empty /etc directory, and thus need system users and
4056 groups created during early boot. systemd now also ships
4057 with two default sysusers.d/ files for the most basic
4058 users and groups systemd and the core operating system
4061 * A new tmpfiles snippet has been added that rebuilds the
4062 essential files in /etc on boot, should they be missing.
4064 * A directive for ensuring automatic clean-up of
4065 /var/cache/man/ has been removed from the default
4066 configuration. This line should now be shipped by the man
4067 implementation. The necessary change has been made to the
4068 man-db implementation. Note that you need to update your man
4069 implementation to one that ships this line, otherwise no
4070 automatic clean-up of /var/cache/man will take place.
4072 * A new condition ConditionNeedsUpdate= has been added that
4073 may conditionalize services to only run when /etc or /var
4074 are "older" than the vendor operating system resources in
4075 /usr. This is useful for reconstructing or updating /etc
4076 after an offline update of /usr or a factory reset, on the
4077 next reboot. Services that want to run once after such an
4078 update or reset should use this condition and order
4079 themselves before the new systemd-update-done.service, which
4080 will mark the two directories as fully updated. A number of
4081 service files have been added making use of this, to rebuild
4082 the udev hardware database, the journald message catalog and
4083 dynamic loader cache (ldconfig). The systemd-sysusers tool
4084 described above also makes use of this now. With this in
4085 place it is now possible to start up a minimal operating
4086 system with /etc empty cleanly. For more information on the
4087 concepts involved see this recent blog story:
4089 http://0pointer.de/blog/projects/stateless.html
4091 * A new system group "input" has been introduced, and all
4092 input device nodes get this group assigned. This is useful
4093 for system-level software to get access to input devices. It
4094 complements what is already done for "audio" and "video".
4096 * systemd-networkd learnt minimal DHCPv4 server support in
4097 addition to the existing DHCPv4 client support. It also
4098 learnt DHCPv6 client and IPv6 Router Solicitation client
4099 support. The DHCPv4 client gained support for static routes
4100 passed in from the server. Note that the [DHCPv4] section
4101 known in older systemd-networkd versions has been renamed to
4102 [DHCP] and is now also used by the DHCPv6 client. Existing
4103 .network files using settings of this section should be
4104 updated, though compatibility is maintained. Optionally, the
4105 client hostname may now be sent to the DHCP server.
4107 * networkd gained support for vxlan virtual networks as well
4108 as tun/tap and dummy devices.
4110 * networkd gained support for automatic allocation of address
4111 ranges for interfaces from a system-wide pool of
4112 addresses. This is useful for dynamically managing a large
4113 number of interfaces with a single network configuration
4114 file. In particular this is useful to easily assign
4115 appropriate IP addresses to the veth links of a large number
4116 of nspawn instances.
4118 * RPM macros for processing sysusers, sysctl and binfmt
4119 drop-in snippets at package installation time have been
4122 * The /etc/os-release file should now be placed in
4123 /usr/lib/os-release. The old location is automatically
4124 created as symlink. /usr/lib is the more appropriate
4125 location of this file, since it shall actually describe the
4126 vendor operating system shipped in /usr, and not the
4127 configuration stored in /etc.
4129 * .mount units gained a new boolean SloppyOptions= setting
4130 that maps to mount(8)'s -s option which enables permissive
4131 parsing of unknown mount options.
4133 * tmpfiles learnt a new "L+" directive which creates a symlink
4134 but (unlike "L") deletes a pre-existing file first, should
4135 it already exist and not already be the correct
4136 symlink. Similarly, "b+", "c+" and "p+" directives have been
4137 added as well, which create block and character devices, as
4138 well as fifos in the filesystem, possibly removing any
4139 pre-existing files of different types.
4141 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4142 'argument' field (which so far specified the source to
4143 symlink/copy the files from) is now optional. If omitted the
4144 same file os copied from /usr/share/factory/ suffixed by the
4145 full destination path. This is useful for populating /etc
4146 with essential files, by copying them from vendor defaults
4147 shipped in /usr/share/factory/etc.
4149 * A new command "systemctl preset-all" has been added that
4150 applies the service preset settings to all installed unit
4151 files. A new switch --preset-mode= has been added that
4152 controls whether only enable or only disable operations
4155 * A new command "systemctl is-system-running" has been added
4156 that allows checking the overall state of the system, for
4157 example whether it is fully up and running.
4159 * When the system boots up with an empty /etc, the equivalent
4160 to "systemctl preset-all" is executed during early boot, to
4161 make sure all default services are enabled after a factory
4164 * systemd now contains a minimal preset file that enables the
4165 most basic services systemd ships by default.
4167 * Unit files' [Install] section gained a new DefaultInstance=
4168 field for defining the default instance to create if a
4169 template unit is enabled with no instance specified.
4171 * A new passive target cryptsetup-pre.target has been added
4172 that may be used by services that need to make they run and
4173 finish before the first LUKS cryptographic device is set up.
4175 * The /dev/loop-control and /dev/btrfs-control device nodes
4176 are now owned by the "disk" group by default, opening up
4177 access to this group.
4179 * systemd-coredump will now automatically generate a
4180 stack trace of all core dumps taking place on the system,
4181 based on elfutils' libdw library. This stack trace is logged
4184 * systemd-coredump may now optionally store coredumps directly
4185 on disk (in /var/lib/systemd/coredump, possibly compressed),
4186 instead of storing them unconditionally in the journal. This
4187 mode is the new default. A new configuration file
4188 /etc/systemd/coredump.conf has been added to configure this
4189 and other parameters of systemd-coredump.
4191 * coredumpctl gained a new "info" verb to show details about a
4192 specific coredump. A new switch "-1" has also been added
4193 that makes sure to only show information about the most
4194 recent entry instead of all entries. Also, as the tool is
4195 generally useful now the "systemd-" prefix of the binary
4196 name has been removed. Distributions that want to maintain
4197 compatibility with the old name should add a symlink from
4198 the old name to the new name.
4200 * journald's SplitMode= now defaults to "uid". This makes sure
4201 that unprivileged users can access their own coredumps with
4202 coredumpctl without restrictions.
4204 * New kernel command line options "systemd.wants=" (for
4205 pulling an additional unit during boot), "systemd.mask="
4206 (for masking a specific unit for the boot), and
4207 "systemd.debug-shell" (for enabling the debug shell on tty9)
4208 have been added. This is implemented in the new generator
4209 "systemd-debug-generator".
4211 * systemd-nspawn will now by default filter a couple of
4212 syscalls for containers, among them those required for
4213 kernel module loading, direct x86 IO port access, swap
4214 management, and kexec. Most importantly though
4215 open_by_handle_at() is now prohibited for containers,
4216 closing a hole similar to a recently discussed vulnerability
4217 in docker regarding access to files on file hierarchies the
4218 container should normally not have access to. Note that, for
4219 nspawn, we generally make no security claims anyway (and
4220 this is explicitly documented in the man page), so this is
4221 just a fix for one of the most obvious problems.
4223 * A new man page file-hierarchy(7) has been added that
4224 contains a minimized, modernized version of the file system
4225 layout systemd expects, similar in style to the FHS
4226 specification or hier(5). A new tool systemd-path(1) has
4227 been added to query many of these paths for the local
4230 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4231 longer done. Since the directory now has a per-user size
4232 limit, and is cleaned on logout this appears unnecessary,
4233 in particular since this now brings the lifecycle of this
4234 directory closer in line with how IPC objects are handled.
4236 * systemd.pc now exports a number of additional directories,
4237 including $libdir (which is useful to identify the library
4238 path for the primary architecture of the system), and a
4239 couple of drop-in directories.
4241 * udev's predictable network interface names now use the dev_port
4242 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4243 distinguish between ports of the same PCI function. dev_id should
4244 only be used for ports using the same HW address, hence the need
4247 * machined has been updated to export the OS version of a
4248 container (read from /etc/os-release and
4249 /usr/lib/os-release) on the bus. This is now shown in
4250 "machinectl status" for a machine.
4252 * A new service setting RestartForceExitStatus= has been
4253 added. If configured to a set of exit signals or process
4254 return values, the service will be restarted when the main
4255 daemon process exits with any of them, regardless of the
4258 * systemctl's -H switch for connecting to remote systemd
4259 machines has been extended so that it may be used to
4260 directly connect to a specific container on the
4261 host. "systemctl -H root@foobar:waldi" will now connect as
4262 user "root" to host "foobar", and then proceed directly to
4263 the container named "waldi". Note that currently you have to
4264 authenticate as user "root" for this to work, as entering
4265 containers is a privileged operation.
4267 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4268 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4269 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4270 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4271 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4272 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4273 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4274 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4275 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4276 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4277 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4278 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4280 — Berlin, 2014-07-03
4284 * As an experimental feature, udev now tries to lock the
4285 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4286 executes events for the disk or any of its partitions.
4287 Applications like partitioning programs can lock the
4288 disk device node (flock(LOCK_EX)) and claim temporary
4289 device ownership that way; udev will entirely skip all event
4290 handling for this disk and its partitions. If the disk
4291 was opened for writing, the close will trigger a partition
4292 table rescan in udev's "watch" facility, and if needed
4293 synthesize "change" events for the disk and all its partitions.
4294 This is now unconditionally enabled, and if it turns out to
4295 cause major problems, we might turn it on only for specific
4296 devices, or might need to disable it entirely. Device Mapper
4297 devices are excluded from this logic.
4299 * We temporarily dropped the "-l" switch for fsck invocations,
4300 since they collide with the flock() logic above. util-linux
4301 upstream has been changed already to avoid this conflict,
4302 and we will readd "-l" as soon as util-linux with this
4303 change has been released.
4305 * The dependency on libattr has been removed. Since a long
4306 time, the extended attribute calls have moved to glibc, and
4307 libattr is thus unnecessary.
4309 * Virtualization detection works without privileges now. This
4310 means the systemd-detect-virt binary no longer requires
4311 CAP_SYS_PTRACE file capabilities, and our daemons can run
4312 with fewer privileges.
4314 * systemd-networkd now runs under its own "systemd-network"
4315 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4316 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4317 loses the ability to write to files owned by root this way.
4319 * Similarly, systemd-resolved now runs under its own
4320 "systemd-resolve" user with no capabilities remaining.
4322 * Similarly, systemd-bus-proxyd now runs under its own
4323 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4325 * systemd-networkd gained support for setting up "veth"
4326 virtual Ethernet devices for container connectivity, as well
4327 as GRE and VTI tunnels.
4329 * systemd-networkd will no longer automatically attempt to
4330 manually load kernel modules necessary for certain tunnel
4331 transports. Instead, it is assumed the kernel loads them
4332 automatically when required. This only works correctly on
4333 very new kernels. On older kernels, please consider adding
4334 the kernel modules to /etc/modules-load.d/ as a work-around.
4336 * The resolv.conf file systemd-resolved generates has been
4337 moved to /run/systemd/resolve/. If you have a symlink from
4338 /etc/resolv.conf, it might be necessary to correct it.
4340 * Two new service settings, ProtectHome= and ProtectSystem=,
4341 have been added. When enabled, they will make the user data
4342 (such as /home) inaccessible or read-only and the system
4343 (such as /usr) read-only, for specific services. This allows
4344 very light-weight per-service sandboxing to avoid
4345 modifications of user data or system files from
4346 services. These two new switches have been enabled for all
4347 of systemd's long-running services, where appropriate.
4349 * Socket units gained new SocketUser= and SocketGroup=
4350 settings to set the owner user and group of AF_UNIX sockets
4351 and FIFOs in the file system.
4353 * Socket units gained a new RemoveOnStop= setting. If enabled,
4354 all FIFOS and sockets in the file system will be removed
4355 when the specific socket unit is stopped.
4357 * Socket units gained a new Symlinks= setting. It takes a list
4358 of symlinks to create to file system sockets or FIFOs
4359 created by the specific Unix sockets. This is useful to
4360 manage symlinks to socket nodes with the same life-cycle as
4363 * The /dev/log socket and /dev/initctl FIFO have been moved to
4364 /run, and have been replaced by symlinks. This allows
4365 connecting to these facilities even if PrivateDevices=yes is
4366 used for a service (which makes /dev/log itself unavailable,
4367 but /run is left). This also has the benefit of ensuring
4368 that /dev only contains device nodes, directories and
4369 symlinks, and nothing else.
4371 * sd-daemon gained two new calls sd_pid_notify() and
4372 sd_pid_notifyf(). They are similar to sd_notify() and
4373 sd_notifyf(), but allow overriding of the source PID of
4374 notification messages if permissions permit this. This is
4375 useful to send notify messages on behalf of a different
4376 process (for example, the parent process). The
4377 systemd-notify tool has been updated to make use of this
4378 when sending messages (so that notification messages now
4379 originate from the shell script invoking systemd-notify and
4380 not the systemd-notify process itself. This should minimize
4381 a race where systemd fails to associate notification
4382 messages to services when the originating process already
4385 * A new "on-abnormal" setting for Restart= has been added. If
4386 set, it will result in automatic restarts on all "abnormal"
4387 reasons for a process to exit, which includes unclean
4388 signals, core dumps, timeouts and watchdog timeouts, but
4389 does not include clean and unclean exit codes or clean
4390 signals. Restart=on-abnormal is an alternative for
4391 Restart=on-failure for services that shall be able to
4392 terminate and avoid restarts on certain errors, by
4393 indicating so with an unclean exit code. Restart=on-failure
4394 or Restart=on-abnormal is now the recommended setting for
4395 all long-running services.
4397 * If the InaccessibleDirectories= service setting points to a
4398 mount point (or if there are any submounts contained within
4399 it), it is now attempted to completely unmount it, to make
4400 the file systems truly unavailable for the respective
4403 * The ReadOnlyDirectories= service setting and
4404 systemd-nspawn's --read-only parameter are now recursively
4405 applied to all submounts, too.
4407 * Mount units may now be created transiently via the bus APIs.
4409 * The support for SysV and LSB init scripts has been removed
4410 from the systemd daemon itself. Instead, it is now
4411 implemented as a generator that creates native systemd units
4412 from these scripts when needed. This enables us to remove a
4413 substantial amount of legacy code from PID 1, following the
4414 fact that many distributions only ship a very small number
4415 of LSB/SysV init scripts nowadays.
4417 * Privileged Xen (dom0) domains are not considered
4418 virtualization anymore by the virtualization detection
4419 logic. After all, they generally have unrestricted access to
4420 the hardware and usually are used to manage the unprivileged
4423 * systemd-tmpfiles gained a new "C" line type, for copying
4424 files or entire directories.
4426 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4427 lines. So far, they have been non-globbing versions of the
4428 latter, and have thus been redundant. In future, it is
4429 recommended to only use "z". "m" has hence been removed
4430 from the documentation, even though it stays supported.
4432 * A tmpfiles snippet to recreate the most basic structure in
4433 /var has been added. This is enough to create the /var/run →
4434 /run symlink and create a couple of structural
4435 directories. This allows systems to boot up with an empty or
4436 volatile /var. Of course, while with this change, the core OS
4437 now is capable with dealing with a volatile /var, not all
4438 user services are ready for it. However, we hope that sooner
4439 or later, many service daemons will be changed upstream so
4440 that they are able to automatically create their necessary
4441 directories in /var at boot, should they be missing. This is
4442 the first step to allow state-less systems that only require
4443 the vendor image for /usr to boot.
4445 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4446 empty tmpfs instance to a specific directory. This is
4447 particularly useful for making use of the automatic
4448 reconstruction of /var (see above), by passing --tmpfs=/var.
4450 * Access modes specified in tmpfiles snippets may now be
4451 prefixed with "~", which indicates that they shall be masked
4452 by whether the existing file or directory is currently
4453 writable, readable or executable at all. Also, if specified,
4454 the sgid/suid/sticky bits will be masked for all
4457 * A new passive target unit "network-pre.target" has been
4458 added which is useful for services that shall run before any
4459 network is configured, for example firewall scripts.
4461 * The "floppy" group that previously owned the /dev/fd*
4462 devices is no longer used. The "disk" group is now used
4463 instead. Distributions should probably deprecate usage of
4466 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4467 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4468 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4469 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4470 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4471 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4474 — Berlin, 2014-06-11
4478 * A new "systemd-timesyncd" daemon has been added for
4479 synchronizing the system clock across the network. It
4480 implements an SNTP client. In contrast to NTP
4481 implementations such as chrony or the NTP reference server,
4482 this only implements a client side, and does not bother with
4483 the full NTP complexity, focusing only on querying time from
4484 one remote server and synchronizing the local clock to
4485 it. Unless you intend to serve NTP to networked clients or
4486 want to connect to local hardware clocks, this simple NTP
4487 client should be more than appropriate for most
4488 installations. The daemon runs with minimal privileges, and
4489 has been hooked up with networkd to only operate when
4490 network connectivity is available. The daemon saves the
4491 current clock to disk every time a new NTP sync has been
4492 acquired, and uses this to possibly correct the system clock
4493 early at bootup, in order to accommodate for systems that
4494 lack an RTC such as the Raspberry Pi and embedded devices,
4495 and to make sure that time monotonically progresses on these
4496 systems, even if it is not always correct. To make use of
4497 this daemon, a new system user and group "systemd-timesync"
4498 needs to be created on installation of systemd.
4500 * The queue "seqnum" interface of libudev has been disabled, as
4501 it was generally incompatible with device namespacing as
4502 sequence numbers of devices go "missing" if the devices are
4503 part of a different namespace.
4505 * "systemctl list-timers" and "systemctl list-sockets" gained
4506 a --recursive switch for showing units of these types also
4507 for all local containers, similar in style to the already
4508 supported --recursive switch for "systemctl list-units".
4510 * A new RebootArgument= setting has been added for service
4511 units, which may be used to specify a kernel reboot argument
4512 to use when triggering reboots with StartLimitAction=.
4514 * A new FailureAction= setting has been added for service
4515 units which may be used to specify an operation to trigger
4516 when a service fails. This works similarly to
4517 StartLimitAction=, but unlike it, controls what is done
4518 immediately rather than only after several attempts to
4519 restart the service in question.
4521 * hostnamed got updated to also expose the kernel name,
4522 release, and version on the bus. This is useful for
4523 executing commands like hostnamectl with the -H switch.
4524 systemd-analyze makes use of this to properly display
4525 details when running non-locally.
4527 * The bootchart tool can now show cgroup information in the
4528 graphs it generates.
4530 * The CFS CPU quota cgroup attribute is now exposed for
4531 services. The new CPUQuota= switch has been added for this
4532 which takes a percentage value. Setting this will have the
4533 result that a service may never get more CPU time than the
4534 specified percentage, even if the machine is otherwise idle.
4536 * systemd-networkd learned IPIP and SIT tunnel support.
4538 * LSB init scripts exposing a dependency on $network will now
4539 get a dependency on network-online.target rather than simply
4540 network.target. This should bring LSB handling closer to
4541 what it was on SysV systems.
4543 * A new fsck.repair= kernel option has been added to control
4544 how fsck shall deal with unclean file systems at boot.
4546 * The (.ini) configuration file parser will now silently
4547 ignore sections whose name begins with "X-". This may be
4548 used to maintain application-specific extension sections in unit
4551 * machined gained a new API to query the IP addresses of
4552 registered containers. "machinectl status" has been updated
4553 to show these addresses in its output.
4555 * A new call sd_uid_get_display() has been added to the
4556 sd-login APIs for querying the "primary" session of a
4557 user. The "primary" session of the user is elected from the
4558 user's sessions and generally a graphical session is
4559 preferred over a text one.
4561 * A minimal systemd-resolved daemon has been added. It
4562 currently simply acts as a companion to systemd-networkd and
4563 manages resolv.conf based on per-interface DNS
4564 configuration, possibly supplied via DHCP. In the long run
4565 we hope to extend this into a local DNSSEC enabled DNS and
4568 * The systemd-networkd-wait-online tool is now enabled by
4569 default. It will delay network-online.target until a network
4570 connection has been configured. The tool primarily integrates
4571 with networkd, but will also make a best effort to make sense
4572 of network configuration performed in some other way.
4574 * Two new service options StartupCPUShares= and
4575 StartupBlockIOWeight= have been added that work similarly to
4576 CPUShares= and BlockIOWeight= however only apply during
4577 system startup. This is useful to prioritize certain services
4578 differently during bootup than during normal runtime.
4580 * hostnamed has been changed to prefer the statically
4581 configured hostname in /etc/hostname (unless set to
4582 'localhost' or empty) over any dynamic one supplied by
4583 dhcp. With this change, the rules for picking the hostname
4584 match more closely the rules of other configuration settings
4585 where the local administrator's configuration in /etc always
4586 overrides any other settings.
4588 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4589 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4590 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4591 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4592 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4593 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4594 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4595 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4596 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4597 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4598 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4599 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4600 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4601 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4602 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4603 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4606 — Beijing, 2014-05-28
4610 * When restoring the screen brightness at boot, stay away from
4611 the darkest setting or from the lowest 5% of the available
4612 range, depending on which is the larger value of both. This
4613 should effectively protect the user from rebooting into a
4614 black screen, should the brightness have been set to minimum
4617 * sd-login gained a new sd_machine_get_class() call to
4618 determine the class ("vm" or "container") of a machine
4619 registered with machined.
4621 * sd-login gained new calls
4622 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4623 to query the identity of the peer of a local AF_UNIX
4624 connection. They operate similarly to their sd_pid_get_xyz()
4627 * PID 1 will now maintain a system-wide system state engine
4628 with the states "starting", "running", "degraded",
4629 "maintenance", "stopping". These states are bound to system
4630 startup, normal runtime, runtime with at least one failed
4631 service, rescue/emergency mode and system shutdown. This
4632 state is shown in the "systemctl status" output when no unit
4633 name is passed. It is useful to determine system state, in
4634 particularly when doing so for many systems or containers at
4637 * A new command "list-machines" has been added to "systemctl"
4638 that lists all local OS containers and shows their system
4639 state (see above), if systemd runs inside of them.
4641 * systemctl gained a new "-r" switch to recursively enumerate
4642 units on all local containers, when used with the
4643 "list-unit" command (which is the default one that is
4644 executed when no parameters are specified).
4646 * The GPT automatic partition discovery logic will now honour
4647 two GPT partition flags: one may be set on a partition to
4648 cause it to be mounted read-only, and the other may be set
4649 on a partition to ignore it during automatic discovery.
4651 * Two new GPT type UUIDs have been added for automatic root
4652 partition discovery, for 32-bit and 64-bit ARM. This is not
4653 particularly useful for discovering the root directory on
4654 these architectures during bare-metal boots (since UEFI is
4655 not common there), but still very useful to allow booting of
4656 ARM disk images in nspawn with the -i option.
4658 * MAC addresses of interfaces created with nspawn's
4659 --network-interface= switch will now be generated from the
4660 machine name, and thus be stable between multiple invocations
4663 * logind will now automatically remove all IPC objects owned
4664 by a user if she or he fully logs out. This makes sure that
4665 users who are logged out cannot continue to consume IPC
4666 resources. This covers SysV memory, semaphores and message
4667 queues as well as POSIX shared memory and message
4668 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4669 limits. With this functionality, that is corrected. This may
4670 be turned off by using the RemoveIPC= switch of logind.conf.
4672 * The systemd-machine-id-setup and tmpfiles tools gained a
4673 --root= switch to operate on a specific root directory,
4676 * journald can now forward logged messages to the TTYs of all
4677 logged in users ("wall"). This is the default for all
4678 emergency messages now.
4680 * A new tool systemd-journal-remote has been added to stream
4681 journal log messages across the network.
4683 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4684 controller trees are mounted into it. Note that the
4685 directories mounted beneath it are not read-only. This is a
4686 security measure and is particularly useful because glibc
4687 actually includes a search logic to pick any tmpfs it can
4688 find to implement shm_open() if /dev/shm is not available
4689 (which it might very well be in namespaced setups).
4691 * machinectl gained a new "poweroff" command to cleanly power
4692 down a local OS container.
4694 * The PrivateDevices= unit file setting will now also drop the
4695 CAP_MKNOD capability from the capability bound set, and
4696 imply DevicePolicy=closed.
4698 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4699 comprehensively on all long-running systemd services where
4700 this is appropriate.
4702 * systemd-udevd will now run in a disassociated mount
4703 namespace. To mount directories from udev rules, make sure to
4704 pull in mount units via SYSTEMD_WANTS properties.
4706 * The kdbus support gained support for uploading policy into
4707 the kernel. sd-bus gained support for creating "monitoring"
4708 connections that can eavesdrop into all bus communication
4709 for debugging purposes.
4711 * Timestamps may now be specified in seconds since the UNIX
4712 epoch Jan 1st, 1970 by specifying "@" followed by the value
4715 * Native tcpwrap support in systemd has been removed. tcpwrap
4716 is old code, not really maintained anymore and has serious
4717 shortcomings, and better options such as firewalls
4718 exist. For setups that require tcpwrap usage, please
4719 consider invoking your socket-activated service via tcpd,
4720 like on traditional inetd.
4722 * A new system.conf configuration option
4723 DefaultTimerAccuracySec= has been added that controls the
4724 default AccuracySec= setting of .timer units.
4726 * Timer units gained a new WakeSystem= switch. If enabled,
4727 timers configured this way will cause the system to resume
4728 from system suspend (if the system supports that, which most
4731 * Timer units gained a new Persistent= switch. If enabled,
4732 timers configured this way will save to disk when they have
4733 been last triggered. This information is then used on next
4734 reboot to possible execute overdue timer events, that
4735 could not take place because the system was powered off.
4736 This enables simple anacron-like behaviour for timer units.
4738 * systemctl's "list-timers" will now also list the time a
4739 timer unit was last triggered in addition to the next time
4740 it will be triggered.
4742 * systemd-networkd will now assign predictable IPv4LL
4743 addresses to its local interfaces.
4745 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4746 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4747 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4748 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4749 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4750 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4751 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4752 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4755 — Berlin, 2014-03-25
4759 * A new unit file setting RestrictAddressFamilies= has been
4760 added to restrict which socket address families unit
4761 processes gain access to. This takes address family names
4762 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4763 attack surface of services via exotic protocol stacks. This
4764 is built on seccomp system call filters.
4766 * Two new unit file settings RuntimeDirectory= and
4767 RuntimeDirectoryMode= have been added that may be used to
4768 manage a per-daemon runtime directories below /run. This is
4769 an alternative for setting up directory permissions with
4770 tmpfiles snippets, and has the advantage that the runtime
4771 directory's lifetime is bound to the daemon runtime and that
4772 the daemon starts up with an empty directory each time. This
4773 is particularly useful when writing services that drop
4774 privileges using the User= or Group= setting.
4776 * The DeviceAllow= unit setting now supports globbing for
4777 matching against device group names.
4779 * The systemd configuration file system.conf gained new
4780 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4781 DefaultMemoryAccounting= to globally turn on/off accounting
4782 for specific resources (cgroups) for all units. These
4783 settings may still be overridden individually in each unit
4786 * systemd-gpt-auto-generator is now able to discover /srv and
4787 root partitions in addition to /home and swap partitions. It
4788 also supports LUKS-encrypted partitions now. With this in
4789 place, automatic discovery of partitions to mount following
4790 the Discoverable Partitions Specification
4791 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4792 is now a lot more complete. This allows booting without
4793 /etc/fstab and without root= on the kernel command line on
4794 systems prepared appropriately.
4796 * systemd-nspawn gained a new --image= switch which allows
4797 booting up disk images and Linux installations on any block
4798 device that follow the Discoverable Partitions Specification
4799 (see above). This means that installations made with
4800 appropriately updated installers may now be started and
4801 deployed using container managers, completely
4802 unmodified. (We hope that libvirt-lxc will add support for
4803 this feature soon, too.)
4805 * systemd-nspawn gained a new --network-macvlan= setting to
4806 set up a private macvlan interface for the
4807 container. Similarly, systemd-networkd gained a new
4808 Kind=macvlan setting in .netdev files.
4810 * systemd-networkd now supports configuring local addresses
4813 * A new tool systemd-network-wait-online has been added to
4814 synchronously wait for network connectivity using
4817 * The sd-bus.h bus API gained a new sd_bus_track object for
4818 tracking the life-cycle of bus peers. Note that sd-bus.h is
4819 still not a public API though (unless you specify
4820 --enable-kdbus on the configure command line, which however
4821 voids your warranty and you get no API stability guarantee).
4823 * The $XDG_RUNTIME_DIR runtime directories for each user are
4824 now individual tmpfs instances, which has the benefit of
4825 introducing separate pools for each user, with individual
4826 size limits, and thus making sure that unprivileged clients
4827 can no longer negatively impact the system or other users by
4828 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4829 RuntimeDirectorySize= has been introduced that allows
4830 controlling the default size limit for all users. It
4831 defaults to 10% of the available physical memory. This is no
4832 replacement for quotas on tmpfs though (which the kernel
4833 still does not support), as /dev/shm and /tmp are still
4834 shared resources used by both the system and unprivileged
4837 * logind will now automatically turn off automatic suspending
4838 on laptop lid close when more than one display is
4839 connected. This was previously expected to be implemented
4840 individually in desktop environments (such as GNOME),
4841 however has been added to logind now, in order to fix a
4842 boot-time race where a desktop environment might not have
4843 been started yet and thus not been able to take an inhibitor
4844 lock at the time where logind already suspends the system
4845 due to a closed lid.
4847 * logind will now wait at least 30s after each system
4848 suspend/resume cycle, and 3min after system boot before
4849 suspending the system due to a closed laptop lid. This
4850 should give USB docking stations and similar enough time to
4851 be probed and configured after system resume and boot in
4852 order to then act as suspend blocker.
4854 * systemd-run gained a new --property= setting which allows
4855 initialization of resource control properties (and others)
4856 for the created scope or service unit. Example: "systemd-run
4857 --property=BlockIOWeight=10 updatedb" may be used to run
4858 updatedb at a low block IO scheduling weight.
4860 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4861 now also work in --scope mode.
4863 * When systemd is compiled with kdbus support, basic support
4864 for enforced policies is now in place. (Note that enabling
4865 kdbus still voids your warranty and no API compatibility
4868 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4869 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4870 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4871 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4872 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4873 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4874 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4875 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4876 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4877 Zbigniew Jędrzejewski-Szmek
4879 — Berlin, 2014-03-12
4883 * systemd will now relabel /dev after loading the SMACK policy
4884 according to SMACK rules.
4886 * A new unit file option AppArmorProfile= has been added to
4887 set the AppArmor profile for the processes of a unit.
4889 * A new condition check ConditionArchitecture= has been added
4890 to conditionalize units based on the system architecture, as
4891 reported by uname()'s "machine" field.
4893 * systemd-networkd now supports matching on the system
4894 virtualization, architecture, kernel command line, host name
4897 * logind is now a lot more aggressive when suspending the
4898 machine due to a closed laptop lid. Instead of acting only
4899 on the lid close action, it will continuously watch the lid
4900 status and act on it. This is useful for laptops where the
4901 power button is on the outside of the chassis so that it can
4902 be reached without opening the lid (such as the Lenovo
4903 Yoga). On those machines, logind will now immediately
4904 re-suspend the machine if the power button has been
4905 accidentally pressed while the laptop was suspended and in a
4906 backpack or similar.
4908 * logind will now watch SW_DOCK switches and inhibit reaction
4909 to the lid switch if it is pressed. This means that logind
4910 will not suspend the machine anymore if the lid is closed
4911 and the system is docked, if the laptop supports SW_DOCK
4912 notifications via the input layer. Note that ACPI docking
4913 stations do not generate this currently. Also note that this
4914 logic is usually not fully sufficient and Desktop
4915 Environments should take a lid switch inhibitor lock when an
4916 external display is connected, as systemd will not watch
4919 * nspawn will now make use of the devices cgroup controller by
4920 default, and only permit creation of and access to the usual
4921 API device nodes like /dev/null or /dev/random, as well as
4922 access to (but not creation of) the pty devices.
4924 * We will now ship a default .network file for
4925 systemd-networkd that automatically configures DHCP for
4926 network interfaces created by nspawn's --network-veth or
4927 --network-bridge= switches.
4929 * systemd will now understand the usual M, K, G, T suffixes
4930 according to SI conventions (i.e. to the base 1000) when
4931 referring to throughput and hardware metrics. It will stay
4932 with IEC conventions (i.e. to the base 1024) for software
4933 metrics, according to what is customary according to
4934 Wikipedia. We explicitly document which base applies for
4935 each configuration option.
4937 * The DeviceAllow= setting in unit files now supports a syntax
4938 to whitelist an entire group of devices node majors at once,
4939 based on the /proc/devices listing. For example, with the
4940 string "char-pts", it is now possible to whitelist all
4941 current and future pseudo-TTYs at once.
4943 * sd-event learned a new "post" event source. Event sources of
4944 this type are triggered by the dispatching of any event
4945 source of a type that is not "post". This is useful for
4946 implementing clean-up and check event sources that are
4947 triggered by other work being done in the program.
4949 * systemd-networkd is no longer statically enabled, but uses
4950 the usual [Install] sections so that it can be
4951 enabled/disabled using systemctl. It still is enabled by
4954 * When creating a veth interface pair with systemd-nspawn, the
4955 host side will now be prefixed with "vb-" if
4956 --network-bridge= is used, and with "ve-" if --network-veth
4957 is used. This way, it is easy to distinguish these cases on
4958 the host, for example to apply different configuration to
4959 them with systemd-networkd.
4961 * The compatibility libraries for libsystemd-journal.so,
4962 libsystem-id128.so, libsystemd-login.so and
4963 libsystemd-daemon.so do not make use of IFUNC
4964 anymore. Instead, we now build libsystemd.so multiple times
4965 under these alternative names. This means that the footprint
4966 is drastically increased, but given that these are
4967 transitional compatibility libraries, this should not matter
4968 much. This change has been made necessary to support the ARM
4969 platform for these compatibility libraries, as the ARM
4970 toolchain is not really at the same level as the toolchain
4971 for other architectures like x86 and does not support
4972 IFUNC. Please make sure to use --enable-compat-libs only
4973 during a transitional period!
4975 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4976 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4977 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4978 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4979 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4980 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4981 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4982 Zbigniew Jędrzejewski-Szmek
4984 — Berlin, 2014-02-24
4988 * A new component "systemd-networkd" has been added that can
4989 be used to configure local network interfaces statically or
4990 via DHCP. It is capable of bringing up bridges, VLANs, and
4991 bonding. Currently, no hook-ups for interactive network
4992 configuration are provided. Use this for your initrd,
4993 container, embedded, or server setup if you need a simple,
4994 yet powerful, network configuration solution. This
4995 configuration subsystem is quite nifty, as it allows wildcard
4996 hotplug matching in interfaces. For example, with a single
4997 configuration snippet, you can configure that all Ethernet
4998 interfaces showing up are automatically added to a bridge,
4999 or similar. It supports link-sensing and more.
5001 * A new tool "systemd-socket-proxyd" has been added which can
5002 act as a bidirectional proxy for TCP sockets. This is
5003 useful for adding socket activation support to services that
5004 do not actually support socket activation, including virtual
5005 machines and the like.
5007 * Add a new tool to save/restore rfkill state on
5010 * Save/restore state of keyboard backlights in addition to
5011 display backlights on shutdown/boot.
5013 * udev learned a new SECLABEL{} construct to label device
5014 nodes with a specific security label when they appear. For
5015 now, only SECLABEL{selinux} is supported, but the syntax is
5016 prepared for additional security frameworks.
5018 * udev gained a new scheme to configure link-level attributes
5019 from files in /etc/systemd/network/*.link. These files can
5020 match against MAC address, device path, driver name and type,
5021 and will apply attributes like the naming policy, link speed,
5022 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5023 address assignment policy (randomized, ...).
5025 * The configuration of network interface naming rules for
5026 "permanent interface names" has changed: a new NamePolicy=
5027 setting in the [Link] section of .link files determines the
5028 priority of possible naming schemes (onboard, slot, MAC,
5029 path). The default value of this setting is determined by
5030 /usr/lib/net/links/99-default.link. Old
5031 80-net-name-slot.rules udev configuration file has been
5032 removed, so local configuration overriding this file should
5033 be adapted to override 99-default.link instead.
5035 * When the User= switch is used in a unit file, also
5036 initialize $SHELL= based on the user database entry.
5038 * systemd no longer depends on libdbus. All communication is
5039 now done with sd-bus, systemd's low-level bus library
5042 * kdbus support has been added to PID 1 itself. When kdbus is
5043 enabled, this causes PID 1 to set up the system bus and
5044 enable support for a new ".busname" unit type that
5045 encapsulates bus name activation on kdbus. It works a little
5046 bit like ".socket" units, except for bus names. A new
5047 generator has been added that converts classic dbus1 service
5048 activation files automatically into native systemd .busname
5051 * sd-bus: add a light-weight vtable implementation that allows
5052 defining objects on the bus with a simple static const
5053 vtable array of its methods, signals and properties.
5055 * systemd will not generate or install static dbus
5056 introspection data anymore to /usr/share/dbus-1/interfaces,
5057 as the precise format of these files is unclear, and
5058 nothing makes use of it.
5060 * A proxy daemon is now provided to proxy clients connecting
5061 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5062 compatibility with classic D-Bus.
5064 * A bus driver implementation has been added that supports the
5065 classic D-Bus bus driver calls on kdbus, also for
5066 compatibility purposes.
5068 * A new API "sd-event.h" has been added that implements a
5069 minimal event loop API built around epoll. It provides a
5070 couple of features that direct epoll usage is lacking:
5071 prioritization of events, scales to large numbers of timer
5072 events, per-event timer slack (accuracy), system-wide
5073 coalescing of timer events, exit handlers, watchdog
5074 supervision support using systemd's sd_notify() API, child
5077 * A new API "sd-rntl.h" has been added that provides an API
5078 around the route netlink interface of the kernel, similar in
5079 style to "sd-bus.h".
5081 * A new API "sd-dhcp-client.h" has been added that provides a
5082 small DHCPv4 client-side implementation. This is used by
5085 * There is a new kernel command line option
5086 "systemd.restore_state=0|1". When set to "0", none of the
5087 systemd tools will restore saved runtime state to hardware
5088 devices. More specifically, the rfkill and backlight states
5091 * The FsckPassNo= compatibility option in mount/service units
5092 has been removed. The fstab generator will now add the
5093 necessary dependencies automatically, and does not require
5094 PID1's support for that anymore.
5096 * journalctl gained a new switch, --list-boots, that lists
5097 recent boots with their times and boot IDs.
5099 * The various tools like systemctl, loginctl, timedatectl,
5100 busctl, systemd-run, ... have gained a new switch "-M" to
5101 connect to a specific, local OS container (as direct
5102 connection, without requiring SSH). This works on any
5103 container that is registered with machined, such as those
5104 created by libvirt-lxc or nspawn.
5106 * systemd-run and systemd-analyze also gained support for "-H"
5107 to connect to remote hosts via SSH. This is particularly
5108 useful for systemd-run because it enables queuing of jobs
5109 onto remote systems.
5111 * machinectl gained a new command "login" to open a getty
5112 login in any local container. This works with any container
5113 that is registered with machined (such as those created by
5114 libvirt-lxc or nspawn), and which runs systemd inside.
5116 * machinectl gained a new "reboot" command that may be used to
5117 trigger a reboot on a specific container that is registered
5118 with machined. This works on any container that runs an init
5119 system of some kind.
5121 * systemctl gained a new "list-timers" command to print a nice
5122 listing of installed timer units with the times they elapse
5125 * Alternative reboot() parameters may now be specified on the
5126 "systemctl reboot" command line and are passed to the
5127 reboot() system call.
5129 * systemctl gained a new --job-mode= switch to configure the
5130 mode to queue a job with. This is a more generic version of
5131 --fail, --irreversible, and --ignore-dependencies, which are
5132 still available but not advertised anymore.
5134 * /etc/systemd/system.conf gained new settings to configure
5135 various default timeouts of units, as well as the default
5136 start limit interval and burst. These may still be overridden
5139 * PID1 will now export on the bus profile data of the security
5140 policy upload process (such as the SELinux policy upload to
5143 * journald: when forwarding logs to the console, include
5144 timestamps (following the setting in
5145 /sys/module/printk/parameters/time).
5147 * OnCalendar= in timer units now understands the special
5148 strings "yearly" and "annually". (Both are equivalent)
5150 * The accuracy of timer units is now configurable with the new
5151 AccuracySec= setting. It defaults to 1min.
5153 * A new dependency type JoinsNamespaceOf= has been added that
5154 allows running two services within the same /tmp and network
5155 namespace, if PrivateNetwork= or PrivateTmp= are used.
5157 * A new command "cat" has been added to systemctl. It outputs
5158 the original unit file of a unit, and concatenates the
5159 contents of additional "drop-in" unit file snippets, so that
5160 the full configuration is shown.
5162 * systemctl now supports globbing on the various "list-xyz"
5163 commands, like "list-units" or "list-sockets", as well as on
5164 those commands which take multiple unit names.
5166 * journalctl's --unit= switch gained support for globbing.
5168 * All systemd daemons now make use of the watchdog logic so
5169 that systemd automatically notices when they hang.
5171 * If the $container_ttys environment variable is set,
5172 getty-generator will automatically spawn a getty for each
5173 listed tty. This is useful for container managers to request
5174 login gettys to be spawned on as many ttys as needed.
5176 * %h, %s, %U specifier support is not available anymore when
5177 used in unit files for PID 1. This is because NSS calls are
5178 not safe from PID 1. They stay available for --user
5179 instances of systemd, and as special case for the root user.
5181 * loginctl gained a new "--no-legend" switch to turn off output
5184 * The "sd-login.h" API gained three new calls:
5185 sd_session_is_remote(), sd_session_get_remote_user(),
5186 sd_session_get_remote_host() to query information about
5189 * The udev hardware database now also carries vendor/product
5190 information of SDIO devices.
5192 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5193 determine whether watchdog notifications are requested by
5196 * Socket-activated per-connection services now include a
5197 short description of the connection parameters in the
5200 * tmpfiles gained a new "--boot" option. When this is not used,
5201 only lines where the command character is not suffixed with
5202 "!" are executed. When this option is specified, those
5203 options are executed too. This partitions tmpfiles
5204 directives into those that can be safely executed at any
5205 time, and those which should be run only at boot (for
5206 example, a line that creates /run/nologin).
5208 * A new API "sd-resolve.h" has been added which provides a simple
5209 asynchronous wrapper around glibc NSS host name resolution
5210 calls, such as getaddrinfo(). In contrast to glibc's
5211 getaddrinfo_a(), it does not use signals. In contrast to most
5212 other asynchronous name resolution libraries, this one does
5213 not reimplement DNS, but reuses NSS, so that alternate
5214 host name resolution systems continue to work, such as mDNS,
5215 LDAP, etc. This API is based on libasyncns, but it has been
5216 cleaned up for inclusion in systemd.
5218 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5219 "sd-daemon.h" are no longer found in individual libraries
5220 libsystemd-journal.so, libsystemd-login.so,
5221 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5222 merged them into a single library, libsystemd.so, which
5223 provides all symbols. The reason for this is cyclic
5224 dependencies, as these libraries tend to use each other's
5225 symbols. So far, we have managed to workaround that by linking
5226 a copy of a good part of our code into each of these
5227 libraries again and again, which, however, makes certain
5228 things hard to do, like sharing static variables. Also, it
5229 substantially increases footprint. With this change, there
5230 is only one library for the basic APIs systemd
5231 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5232 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5233 library as well, however are subject to the --enable-kdbus
5234 switch (see below). Note that "sd-dhcp-client.h" is not part
5235 of this library (this is because it only consumes, never
5236 provides, services of/to other APIs). To make the transition
5237 easy from the separate libraries to the unified one, we
5238 provide the --enable-compat-libs compile-time switch which
5239 will generate stub libraries that are compatible with the
5240 old ones but redirect all calls to the new one.
5242 * All of the kdbus logic and the new APIs "sd-bus.h",
5243 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5244 and "sd-utf8.h" are compile-time optional via the
5245 "--enable-kdbus" switch, and they are not compiled in by
5246 default. To make use of kdbus, you have to explicitly enable
5247 the switch. Note however, that neither the kernel nor the
5248 userspace API for all of this is considered stable yet. We
5249 want to maintain the freedom to still change the APIs for
5250 now. By specifying this build-time switch, you acknowledge
5251 that you are aware of the instability of the current
5254 * Also, note that while kdbus is pretty much complete,
5255 it lacks one thing: proper policy support. This means you
5256 can build a fully working system with all features; however,
5257 it will be highly insecure. Policy support will be added in
5258 one of the next releases, at the same time that we will
5259 declare the APIs stable.
5261 * When the kernel command line argument "kdbus" is specified,
5262 systemd will automatically load the kdbus.ko kernel module. At
5263 this stage of development, it is only useful for testing kdbus
5264 and should not be used in production. Note: if "--enable-kdbus"
5265 is specified, and the kdbus.ko kernel module is available, and
5266 "kdbus" is added to the kernel command line, the entire system
5267 runs with kdbus instead of dbus-daemon, with the above mentioned
5268 problem of missing the system policy enforcement. Also a future
5269 version of kdbus.ko or a newer systemd will not be compatible with
5270 each other, and will unlikely be able to boot the machine if only
5271 one of them is updated.
5273 * systemctl gained a new "import-environment" command which
5274 uploads the caller's environment (or parts thereof) into the
5275 service manager so that it is inherited by services started
5276 by the manager. This is useful to upload variables like
5277 $DISPLAY into the user service manager.
5279 * A new PrivateDevices= switch has been added to service units
5280 which allows running a service with a namespaced /dev
5281 directory that does not contain any device nodes for
5282 physical devices. More specifically, it only includes devices
5283 such as /dev/null, /dev/urandom, and /dev/zero which are API
5286 * logind has been extended to support behaviour like VT
5287 switching on seats that do not support a VT. This makes
5288 multi-session available on seats that are not the first seat
5289 (seat0), and on systems where kernel support for VTs has
5290 been disabled at compile-time.
5292 * If a process holds a delay lock for system sleep or shutdown
5293 and fails to release it in time, we will now log its
5294 identity. This makes it easier to identify processes that
5295 cause slow suspends or power-offs.
5297 * When parsing /etc/crypttab, support for a new key-slot=
5298 option as supported by Debian is added. It allows indicating
5299 which LUKS slot to use on disk, speeding up key loading.
5301 * The sd_journald_sendv() API call has been checked and
5302 officially declared to be async-signal-safe so that it may
5303 be invoked from signal handlers for logging purposes.
5305 * Boot-time status output is now enabled automatically after a
5306 short timeout if boot does not progress, in order to give
5307 the user an indication what she or he is waiting for.
5309 * The boot-time output has been improved to show how much time
5310 remains until jobs expire.
5312 * The KillMode= switch in service units gained a new possible
5313 value "mixed". If set, and the unit is shut down, then the
5314 initial SIGTERM signal is sent only to the main daemon
5315 process, while the following SIGKILL signal is sent to
5316 all remaining processes of the service.
5318 * When a scope unit is registered, a new property "Controller"
5319 may be set. If set to a valid bus name, systemd will send a
5320 RequestStop() signal to this name when it would like to shut
5321 down the scope. This may be used to hook manager logic into
5322 the shutdown logic of scope units. Also, scope units may now
5323 be put in a special "abandoned" state, in which case the
5324 manager process which created them takes no further
5325 responsibilities for it.
5327 * When reading unit files, systemd will now verify
5328 the access mode of these files, and warn about certain
5329 suspicious combinations. This has been added to make it
5330 easier to track down packaging bugs where unit files are
5331 marked executable or world-writable.
5333 * systemd-nspawn gained a new "--setenv=" switch to set
5334 container-wide environment variables. The similar option in
5335 systemd-activate was renamed from "--environment=" to
5336 "--setenv=" for consistency.
5338 * systemd-nspawn has been updated to create a new kdbus domain
5339 for each container that is invoked, thus allowing each
5340 container to have its own set of system and user buses,
5341 independent of the host.
5343 * systemd-nspawn gained a new --drop-capability= switch to run
5344 the container with less capabilities than the default. Both
5345 --drop-capability= and --capability= now take the special
5346 string "all" for dropping or keeping all capabilities.
5348 * systemd-nspawn gained new switches for executing containers
5349 with specific SELinux labels set.
5351 * systemd-nspawn gained a new --quiet switch to not generate
5352 any additional output but the container's own console
5355 * systemd-nspawn gained a new --share-system switch to run a
5356 container without PID namespacing enabled.
5358 * systemd-nspawn gained a new --register= switch to control
5359 whether the container is registered with systemd-machined or
5360 not. This is useful for containers that do not run full
5361 OS images, but only specific apps.
5363 * systemd-nspawn gained a new --keep-unit which may be used
5364 when invoked as the only program from a service unit, and
5365 results in registration of the unit service itself in
5366 systemd-machined, instead of a newly opened scope unit.
5368 * systemd-nspawn gained a new --network-interface= switch for
5369 moving arbitrary interfaces to the container. The new
5370 --network-veth switch creates a virtual Ethernet connection
5371 between host and container. The new --network-bridge=
5372 switch then allows assigning the host side of this virtual
5373 Ethernet connection to a bridge device.
5375 * systemd-nspawn gained a new --personality= switch for
5376 setting the kernel personality for the container. This is
5377 useful when running a 32-bit container on a 64-bit host. A
5378 similar option Personality= is now also available for service
5381 * logind will now also track a "Desktop" identifier for each
5382 session which encodes the desktop environment of it. This is
5383 useful for desktop environments that want to identify
5384 multiple running sessions of itself easily.
5386 * A new SELinuxContext= setting for service units has been
5387 added that allows setting a specific SELinux execution
5388 context for a service.
5390 * Most systemd client tools will now honour $SYSTEMD_LESS for
5391 settings of the "less" pager. By default, these tools will
5392 override $LESS to allow certain operations to work, such as
5393 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5394 influence this logic.
5396 * systemd's "seccomp" hook-up has been changed to make use of
5397 the libseccomp library instead of using its own
5398 implementation. This has benefits for portability among
5401 * For usage together with SystemCallFilter=, a new
5402 SystemCallErrorNumber= setting has been introduced that
5403 allows configuration of a system error number to be returned
5404 on filtered system calls, instead of immediately killing the
5405 process. Also, SystemCallArchitectures= has been added to
5406 limit access to system calls of a particular architecture
5407 (in order to turn off support for unused secondary
5408 architectures). There is also a global
5409 SystemCallArchitectures= setting in system.conf now to turn
5410 off support for non-native system calls system-wide.
5412 * systemd requires a kernel with a working name_to_handle_at(),
5413 please see the kernel config requirements in the README file.
5415 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5416 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5417 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5418 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5419 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5420 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5421 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5422 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5423 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5424 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5425 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5426 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5427 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5428 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5429 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5430 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5431 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5432 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5433 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5434 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5435 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5436 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5437 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5438 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5440 — Berlin, 2014-02-20
5444 * logind has gained support for facilitating privileged input
5445 and drm device access for unprivileged clients. This work is
5446 useful to allow Wayland display servers (and similar
5447 programs, such as kmscon) to run under the user's ID and
5448 access input and drm devices which are normally
5449 protected. When this is used (and the kernel is new enough)
5450 logind will "mute" IO on the file descriptors passed to
5451 Wayland as long as it is in the background and "unmute" it
5452 if it returns into the foreground. This allows secure
5453 session switching without allowing background sessions to
5454 eavesdrop on input and display data. This also introduces
5455 session switching support if VT support is turned off in the
5456 kernel, and on seats that are not seat0.
5458 * A new kernel command line option luks.options= is understood
5459 now which allows specifying LUKS options for usage for LUKS
5460 encrypted partitions specified with luks.uuid=.
5462 * tmpfiles.d(5) snippets may now use specifier expansion in
5463 path names. More specifically %m, %b, %H, %v, are now
5464 replaced by the local machine id, boot id, hostname, and
5465 kernel version number.
5467 * A new tmpfiles.d(5) command "m" has been introduced which
5468 may be used to change the owner/group/access mode of a file
5469 or directory if it exists, but do nothing if it does not.
5471 * This release removes high-level support for the
5472 MemorySoftLimit= cgroup setting. The underlying kernel
5473 cgroup attribute memory.soft_limit= is currently badly
5474 designed and likely to be removed from the kernel API in its
5475 current form, hence we should not expose it for now.
5477 * The memory.use_hierarchy cgroup attribute is now enabled for
5478 all cgroups systemd creates in the memory cgroup
5479 hierarchy. This option is likely to be come the built-in
5480 default in the kernel anyway, and the non-hierarchical mode
5481 never made much sense in the intrinsically hierarchical
5484 * A new field _SYSTEMD_SLICE= is logged along with all journal
5485 messages containing the slice a message was generated
5486 from. This is useful to allow easy per-customer filtering of
5487 logs among other things.
5489 * systemd-journald will no longer adjust the group of journal
5490 files it creates to the "systemd-journal" group. Instead we
5491 rely on the journal directory to be owned by the
5492 "systemd-journal" group, and its setgid bit set, so that the
5493 kernel file system layer will automatically enforce that
5494 journal files inherit this group assignment. The reason for
5495 this change is that we cannot allow NSS look-ups from
5496 journald which would be necessary to resolve
5497 "systemd-journal" to a numeric GID, because this might
5498 create deadlocks if NSS involves synchronous queries to
5499 other daemons (such as nscd, or sssd) which in turn are
5500 logging clients of journald and might block on it, which
5501 would then dead lock. A tmpfiles.d(5) snippet included in
5502 systemd will make sure the setgid bit and group are
5503 properly set on the journal directory if it exists on every
5504 boot. However, we recommend adjusting it manually after
5505 upgrades too (or from RPM scriptlets), so that the change is
5506 not delayed until next reboot.
5508 * Backlight and random seed files in /var/lib/ have moved into
5509 the /var/lib/systemd/ directory, in order to centralize all
5510 systemd generated files in one directory.
5512 * Boot time performance measurements (as displayed by
5513 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5514 performance information if that's available to determine how
5515 much time BIOS and boot loader initialization required. With
5516 a sufficiently new BIOS you hence no longer need to boot
5517 with Gummiboot to get access to such information.
5519 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5520 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5521 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5522 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5523 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5524 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5525 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5527 — Berlin, 2013-10-02
5531 * The Restart= option for services now understands a new
5532 on-watchdog setting, which will restart the service
5533 automatically if the service stops sending out watchdog keep
5534 alive messages (as configured with WatchdogSec=).
5536 * The getty generator (which is responsible for bringing up a
5537 getty on configured serial consoles) will no longer only
5538 start a getty on the primary kernel console but on all
5539 others, too. This makes the order in which console= is
5540 specified on the kernel command line less important.
5542 * libsystemd-logind gained a new sd_session_get_vt() call to
5543 retrieve the VT number of a session.
5545 * If the option "tries=0" is set for an entry of /etc/crypttab
5546 its passphrase is queried indefinitely instead of any
5547 maximum number of tries.
5549 * If a service with a configure PID file terminates its PID
5550 file will now be removed automatically if it still exists
5551 afterwards. This should put an end to stale PID files.
5553 * systemd-run will now also take relative binary path names
5554 for execution and no longer insists on absolute paths.
5556 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5557 paths that are optionally prefixed with "-" to indicate that
5558 it should not be considered a failure if they do not exist.
5560 * journalctl -o (and similar commands) now understands a new
5561 output mode "short-precise", it is similar to "short" but
5562 shows timestamps with usec accuracy.
5564 * The option "discard" (as known from Debian) is now
5565 synonymous to "allow-discards" in /etc/crypttab. In fact,
5566 "discard" is preferred now (since it is easier to remember
5569 * Some licensing clean-ups were made, so that more code is now
5570 LGPL-2.1 licensed than before.
5572 * A minimal tool to save/restore the display backlight
5573 brightness across reboots has been added. It will store the
5574 backlight setting as late as possible at shutdown, and
5575 restore it as early as possible during reboot.
5577 * A logic to automatically discover and enable home and swap
5578 partitions on GPT disks has been added. With this in place
5579 /etc/fstab becomes optional for many setups as systemd can
5580 discover certain partitions located on the root disk
5581 automatically. Home partitions are recognized under their
5582 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5583 partitions are recognized under their GPT type ID
5584 0657fd6da4ab43c484e50933c84b4f4f.
5586 * systemd will no longer pass any environment from the kernel
5587 or initrd to system services. If you want to set an
5588 environment for all services, do so via the kernel command
5589 line systemd.setenv= assignment.
5591 * The systemd-sysctl tool no longer natively reads the file
5592 /etc/sysctl.conf. If desired, the file should be symlinked
5593 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5594 legacy support by a symlink rather than built-in code, it
5595 also makes the otherwise hidden order of application of the
5596 different files visible. (Note that this partly reverts to a
5597 pre-198 application order of sysctl knobs!)
5599 * The "systemctl set-log-level" and "systemctl dump" commands
5600 have been moved to systemd-analyze.
5602 * systemd-run learned the new --remain-after-exit switch,
5603 which causes the scope unit not to be cleaned up
5604 automatically after the process terminated.
5606 * tmpfiles learned a new --exclude-prefix= switch to exclude
5607 certain paths from operation.
5609 * journald will now automatically flush all messages to disk
5610 as soon as a message at the log level CRIT, ALERT or EMERG
5613 Contributions from: Andrew Cook, Brandon Philips, Christian
5614 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5615 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5616 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5617 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5618 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5619 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5620 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5621 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5622 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5623 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5624 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5625 William Giokas, Zbigniew Jędrzejewski-Szmek
5627 — Berlin, 2013-09-13
5631 * The documentation has been updated to cover the various new
5632 concepts introduced with 205.
5634 * Unit files now understand the new %v specifier which
5635 resolves to the kernel version string as returned by "uname
5638 * systemctl now supports filtering the unit list output by
5639 load state, active state and sub state, using the new
5642 * "systemctl status" will now show the results of the
5643 condition checks (like ConditionPathExists= and similar) of
5644 the last start attempts of the unit. They are also logged to
5647 * "journalctl -b" may now be used to look for boot output of a
5648 specific boot. Try "journalctl -b -1" for the previous boot,
5649 but the syntax is substantially more powerful.
5651 * "journalctl --show-cursor" has been added which prints the
5652 cursor string the last shown log line. This may then be used
5653 with the new "journalctl --after-cursor=" switch to continue
5654 browsing logs from that point on.
5656 * "journalctl --force" may now be used to force regeneration
5659 * Creation of "dead" device nodes has been moved from udev
5660 into kmod and tmpfiles. Previously, udev would read the kmod
5661 databases to pre-generate dead device nodes based on meta
5662 information contained in kernel modules, so that these would
5663 be auto-loaded on access rather then at boot. As this
5664 does not really have much to do with the exposing actual
5665 kernel devices to userspace this has always been slightly
5666 alien in the udev codebase. Following the new scheme kmod
5667 will now generate a runtime snippet for tmpfiles from the
5668 module meta information and it now is tmpfiles' job to the
5669 create the nodes. This also allows overriding access and
5670 other parameters for the nodes using the usual tmpfiles
5671 facilities. As side effect this allows us to remove the
5672 CAP_SYS_MKNOD capability bit from udevd entirely.
5674 * logind's device ACLs may now be applied to these "dead"
5675 devices nodes too, thus finally allowing managed access to
5676 devices such as /dev/snd/sequencer without loading the
5677 backing module right-away.
5679 * A new RPM macro has been added that may be used to apply
5680 tmpfiles configuration during package installation.
5682 * systemd-detect-virt and ConditionVirtualization= now can
5683 detect User-Mode-Linux machines (UML).
5685 * journald will now implicitly log the effective capabilities
5686 set of processes in the message metadata.
5688 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5690 * The initrd interface has been simplified (more specifically,
5691 support for passing performance data via environment
5692 variables and fsck results via files in /run has been
5693 removed). These features were non-essential, and are
5694 nowadays available in a much nicer way by having systemd in
5695 the initrd serialize its state and have the hosts systemd
5696 deserialize it again.
5698 * The udev "keymap" data files and tools to apply keyboard
5699 specific mappings of scan to key codes, and force-release
5700 scan code lists have been entirely replaced by a udev
5701 "keyboard" builtin and a hwdb data file.
5703 * systemd will now honour the kernel's "quiet" command line
5704 argument also during late shutdown, resulting in a
5705 completely silent shutdown when used.
5707 * There's now an option to control the SO_REUSEPORT socket
5708 option in .socket units.
5710 * Instance units will now automatically get a per-template
5711 subslice of system.slice unless something else is explicitly
5712 configured. For example, instances of sshd@.service will now
5713 implicitly be placed in system-sshd.slice rather than
5714 system.slice as before.
5716 * Test coverage support may now be enabled at build time.
5718 Contributions from: Dave Reisner, Frederic Crozat, Harald
5719 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5720 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5721 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5722 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5723 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5724 Giokas, Zbigniew Jędrzejewski-Szmek
5726 — Berlin, 2013-07-23
5730 * Two new unit types have been introduced:
5732 Scope units are very similar to service units, however, are
5733 created out of pre-existing processes — instead of PID 1
5734 forking off the processes. By using scope units it is
5735 possible for system services and applications to group their
5736 own child processes (worker processes) in a powerful way
5737 which then maybe used to organize them, or kill them
5738 together, or apply resource limits on them.
5740 Slice units may be used to partition system resources in an
5741 hierarchical fashion and then assign other units to them. By
5742 default there are now three slices: system.slice (for all
5743 system services), user.slice (for all user sessions),
5744 machine.slice (for VMs and containers).
5746 Slices and scopes have been introduced primarily in
5747 context of the work to move cgroup handling to a
5748 single-writer scheme, where only PID 1
5749 creates/removes/manages cgroups.
5751 * There's a new concept of "transient" units. In contrast to
5752 normal units these units are created via an API at runtime,
5753 not from configuration from disk. More specifically this
5754 means it is now possible to run arbitrary programs as
5755 independent services, with all execution parameters passed
5756 in via bus APIs rather than read from disk. Transient units
5757 make systemd substantially more dynamic then it ever was,
5758 and useful as a general batch manager.
5760 * logind has been updated to make use of scope and slice units
5761 for managing user sessions. As a user logs in he will get
5762 his own private slice unit, to which all sessions are added
5763 as scope units. We also added support for automatically
5764 adding an instance of user@.service for the user into the
5765 slice. Effectively logind will no longer create cgroup
5766 hierarchies on its own now, it will defer entirely to PID 1
5767 for this by means of scope, service and slice units. Since
5768 user sessions this way become entities managed by PID 1
5769 the output of "systemctl" is now a lot more comprehensive.
5771 * A new mini-daemon "systemd-machined" has been added which
5772 may be used by virtualization managers to register local
5773 VMs/containers. nspawn has been updated accordingly, and
5774 libvirt will be updated shortly. machined will collect a bit
5775 of meta information about the VMs/containers, and assign
5776 them their own scope unit (see above). The collected
5777 meta-data is then made available via the "machinectl" tool,
5778 and exposed in "ps" and similar tools. machined/machinectl
5779 is compile-time optional.
5781 * As discussed earlier, the low-level cgroup configuration
5782 options ControlGroup=, ControlGroupModify=,
5783 ControlGroupPersistent=, ControlGroupAttribute= have been
5784 removed. Please use high-level attribute settings instead as
5785 well as slice units.
5787 * A new bus call SetUnitProperties() has been added to alter
5788 various runtime parameters of a unit. This is primarily
5789 useful to alter cgroup parameters dynamically in a nice way,
5790 but will be extended later on to make more properties
5791 modifiable at runtime. systemctl gained a new set-properties
5792 command that wraps this call.
5794 * A new tool "systemd-run" has been added which can be used to
5795 run arbitrary command lines as transient services or scopes,
5796 while configuring a number of settings via the command
5797 line. This tool is currently very basic, however already
5798 very useful. We plan to extend this tool to even allow
5799 queuing of execution jobs with time triggers from the
5800 command line, similar in fashion to "at".
5802 * nspawn will now inform the user explicitly that kernels with
5803 audit enabled break containers, and suggest the user to turn
5806 * Support for detecting the IMA and AppArmor security
5807 frameworks with ConditionSecurity= has been added.
5809 * journalctl gained a new "-k" switch for showing only kernel
5810 messages, mimicking dmesg output; in addition to "--user"
5811 and "--system" switches for showing only user's own logs
5814 * systemd-delta can now show information about drop-in
5815 snippets extending unit files.
5817 * libsystemd-bus has been substantially updated but is still
5818 not available as public API.
5820 * systemd will now look for the "debug" argument on the kernel
5821 command line and enable debug logging, similar to what
5822 "systemd.log_level=debug" already did before.
5824 * "systemctl set-default", "systemctl get-default" has been
5825 added to configure the default.target symlink, which
5826 controls what to boot into by default.
5828 * "systemctl set-log-level" has been added as a convenient
5829 way to raise and lower systemd logging threshold.
5831 * "systemd-analyze plot" will now show the time the various
5832 generators needed for execution, as well as information
5833 about the unit file loading.
5835 * libsystemd-journal gained a new sd_journal_open_files() call
5836 for opening specific journal files. journactl also gained a
5837 new switch to expose this new functionality. Previously we
5838 only supported opening all files from a directory, or all
5839 files from the system, as opening individual files only is
5840 racy due to journal file rotation.
5842 * systemd gained the new DefaultEnvironment= setting in
5843 /etc/systemd/system.conf to set environment variables for
5846 * If a privileged process logs a journal message with the
5847 OBJECT_PID= field set, then journald will automatically
5848 augment this with additional OBJECT_UID=, OBJECT_GID=,
5849 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5850 system services want to log events about specific client
5851 processes. journactl/systemctl has been updated to make use
5852 of this information if all log messages regarding a specific
5855 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5856 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5857 Reisner, David Coppa, David King, David Strauss, Eelco
5858 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5859 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5860 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5861 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5862 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5863 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5864 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5865 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5866 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5867 Łukasz Stelmach, 장동준
5871 * The Python bindings gained some minimal support for the APIs
5872 exposed by libsystemd-logind.
5874 * ConditionSecurity= gained support for detecting SMACK. Since
5875 this condition already supports SELinux and AppArmor we only
5876 miss IMA for this. Patches welcome!
5878 Contributions from: Karol Lewandowski, Lennart Poettering,
5879 Zbigniew Jędrzejewski-Szmek
5883 * systemd-nspawn will now create /etc/resolv.conf if
5884 necessary, before bind-mounting the host's file onto it.
5886 * systemd-nspawn will now store meta information about a
5887 container on the container's cgroup as extended attribute
5888 fields, including the root directory.
5890 * The cgroup hierarchy has been reworked in many ways. All
5891 objects any of the components systemd creates in the cgroup
5892 tree are now suffixed. More specifically, user sessions are
5893 now placed in cgroups suffixed with ".session", users in
5894 cgroups suffixed with ".user", and nspawn containers in
5895 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5896 names are now escaped in a simple scheme to avoid collision
5897 of userspace object names with kernel filenames. This work
5898 is preparation for making these objects relocatable in the
5899 cgroup tree, in order to allow easy resource partitioning of
5900 these objects without causing naming conflicts.
5902 * systemctl list-dependencies gained the new switches
5903 --plain, --reverse, --after and --before.
5905 * systemd-inhibit now shows the process name of processes that
5906 have taken an inhibitor lock.
5908 * nss-myhostname will now also resolve "localhost"
5909 implicitly. This makes /etc/hosts an optional file and
5910 nicely handles that on IPv6 ::1 maps to both "localhost" and
5913 * libsystemd-logind.so gained a new call
5914 sd_get_machine_names() to enumerate running containers and
5915 VMs (currently only supported by very new libvirt and
5916 nspawn). sd_login_monitor can now be used to watch
5917 VMs/containers coming and going.
5919 * .include is not allowed recursively anymore, and only in
5920 unit files. Usually it is better to use drop-in snippets in
5921 .d/*.conf anyway, as introduced with systemd 198.
5923 * systemd-analyze gained a new "critical-chain" command that
5924 determines the slowest chain of units run during system
5925 boot-up. It is very useful for tracking down where
5926 optimizing boot time is the most beneficial.
5928 * systemd will no longer allow manipulating service paths in
5929 the name=systemd:/system cgroup tree using ControlGroup= in
5930 units. (But is still fine with it in all other dirs.)
5932 * There's a new systemd-nspawn@.service service file that may
5933 be used to easily run nspawn containers as system
5934 services. With the container's root directory in
5935 /var/lib/container/foobar it is now sufficient to run
5936 "systemctl start systemd-nspawn@foobar.service" to boot it.
5938 * systemd-cgls gained a new parameter "--machine" to list only
5939 the processes within a certain container.
5941 * ConditionSecurity= now can check for "apparmor". We still
5942 are lacking checks for SMACK and IMA for this condition
5943 check though. Patches welcome!
5945 * A new configuration file /etc/systemd/sleep.conf has been
5946 added that may be used to configure which kernel operation
5947 systemd is supposed to execute when "suspend", "hibernate"
5948 or "hybrid-sleep" is requested. This makes the new kernel
5949 "freeze" state accessible to the user.
5951 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5952 the passed argument if applicable.
5954 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5955 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5956 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5957 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5958 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5959 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5960 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5965 * The output of 'systemctl list-jobs' got some polishing. The
5966 '--type=' argument may now be passed more than once. A new
5967 command 'systemctl list-sockets' has been added which shows
5968 a list of kernel sockets systemd is listening on with the
5969 socket units they belong to, plus the units these socket
5972 * The experimental libsystemd-bus library got substantial
5973 updates to work in conjunction with the (also experimental)
5974 kdbus kernel project. It works well enough to exchange
5975 messages with some sophistication. Note that kdbus is not
5976 ready yet, and the library is mostly an elaborate test case
5977 for now, and not installable.
5979 * systemd gained a new unit 'systemd-static-nodes.service'
5980 that generates static device nodes earlier during boot, and
5981 can run in conjunction with udev.
5983 * libsystemd-login gained a new call sd_pid_get_user_unit()
5984 to retrieve the user systemd unit a process is running
5985 in. This is useful for systems where systemd is used as
5988 * systemd-nspawn now places all containers in the new /machine
5989 top-level cgroup directory in the name=systemd
5990 hierarchy. libvirt will soon do the same, so that we get a
5991 uniform separation of /system, /user and /machine for system
5992 services, user processes and containers/virtual
5993 machines. This new cgroup hierarchy is also useful to stick
5994 stable names to specific container instances, which can be
5995 recognized later this way (this name may be controlled
5996 via systemd-nspawn's new -M switch). libsystemd-login also
5997 gained a new call sd_pid_get_machine_name() to retrieve the
5998 name of the container/VM a specific process belongs to.
6000 * bootchart can now store its data in the journal.
6002 * libsystemd-journal gained a new call
6003 sd_journal_add_conjunction() for AND expressions to the
6004 matching logic. This can be used to express more complex
6005 logical expressions.
6007 * journactl can now take multiple --unit= and --user-unit=
6010 * The cryptsetup logic now understands the "luks.key=" kernel
6011 command line switch for specifying a file to read the
6012 decryption key from. Also, if a configured key file is not
6013 found the tool will now automatically fall back to prompting
6016 * Python systemd.journal module was updated to wrap recently
6017 added functions from libsystemd-journal. The interface was
6018 changed to bring the low level interface in s.j._Reader
6019 closer to the C API, and the high level interface in
6020 s.j.Reader was updated to wrap and convert all data about
6023 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6024 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6025 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6026 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6027 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6028 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6032 * journalctl --update-catalog now understands a new --root=
6033 option to operate on catalogs found in a different root
6036 * During shutdown after systemd has terminated all running
6037 services a final killing loop kills all remaining left-over
6038 processes. We will now print the name of these processes
6039 when we send SIGKILL to them, since this usually indicates a
6042 * If /etc/crypttab refers to password files stored on
6043 configured mount points automatic dependencies will now be
6044 generated to ensure the specific mount is established first
6045 before the key file is attempted to be read.
6047 * 'systemctl status' will now show information about the
6048 network sockets a socket unit is listening on.
6050 * 'systemctl status' will also shown information about any
6051 drop-in configuration file for units. (Drop-In configuration
6052 files in this context are files such as
6053 /etc/systemd/systemd/foobar.service.d/*.conf)
6055 * systemd-cgtop now optionally shows summed up CPU times of
6056 cgroups. Press '%' while running cgtop to switch between
6057 percentage and absolute mode. This is useful to determine
6058 which cgroups use up the most CPU time over the entire
6059 runtime of the system. systemd-cgtop has also been updated
6060 to be 'pipeable' for processing with further shell tools.
6062 * 'hostnamectl set-hostname' will now allow setting of FQDN
6065 * The formatting and parsing of time span values has been
6066 changed. The parser now understands fractional expressions
6067 such as "5.5h". The formatter will now output fractional
6068 expressions for all time spans under 1min, i.e. "5.123456s"
6069 rather than "5s 123ms 456us". For time spans under 1s
6070 millisecond values are shown, for those under 1ms
6071 microsecond values are shown. This should greatly improve
6072 all time-related output of systemd.
6074 * libsystemd-login and libsystemd-journal gained new
6075 functions for querying the poll() events mask and poll()
6076 timeout value for integration into arbitrary event
6079 * localectl gained the ability to list available X11 keymaps
6080 (models, layouts, variants, options).
6082 * 'systemd-analyze dot' gained the ability to filter for
6083 specific units via shell-style globs, to create smaller,
6084 more useful graphs. I.e. it is now possible to create simple
6085 graphs of all the dependencies between only target units, or
6086 of all units that Avahi has dependencies with.
6088 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6089 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6090 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6091 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6092 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6093 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6094 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6098 * The boot-time readahead implementation for rotating media
6099 will now read the read-ahead data in multiple passes which
6100 consist of all read requests made in equidistant time
6101 intervals. This means instead of strictly reading read-ahead
6102 data in its physical order on disk we now try to find a
6103 middle ground between physical and access time order.
6105 * /etc/os-release files gained a new BUILD_ID= field for usage
6106 on operating systems that provide continuous builds of OS
6109 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6110 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6111 William Douglas, Zbigniew Jędrzejewski-Szmek
6115 * systemd-python gained an API exposing libsystemd-daemon.
6117 * The SMACK setup logic gained support for uploading CIPSO
6120 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6121 ReadOnlyDirectories= and InaccessibleDirectories= has
6122 changed. The private /tmp and /var/tmp directories are now
6123 shared by all processes of a service (which means
6124 ExecStartPre= may now leave data in /tmp that ExecStart= of
6125 the same service can still access). When a service is
6126 stopped its temporary directories are immediately deleted
6127 (normal clean-up with tmpfiles is still done in addition to
6130 * By default, systemd will now set a couple of sysctl
6131 variables in the kernel: the safe sysrq options are turned
6132 on, IP route verification is turned on, and source routing
6133 disabled. The recently added hardlink and softlink
6134 protection of the kernel is turned on. These settings should
6135 be reasonably safe, and good defaults for all new systems.
6137 * The predictable network naming logic may now be turned off
6138 with a new kernel command line switch: net.ifnames=0.
6140 * A new libsystemd-bus module has been added that implements a
6141 pretty complete D-Bus client library. For details see:
6143 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6145 * journald will now explicitly flush the journal files to disk
6146 at the latest 5min after each write. The file will then also
6147 be marked offline until the next write. This should increase
6148 reliability in case of a crash. The synchronization delay
6149 can be configured via SyncIntervalSec= in journald.conf.
6151 * There's a new remote-fs-setup.target unit that can be used
6152 to pull in specific services when at least one remote file
6153 system is to be mounted.
6155 * There are new targets timers.target and paths.target as
6156 canonical targets to pull user timer and path units in
6157 from. This complements sockets.target with a similar
6158 purpose for socket units.
6160 * libudev gained a new call udev_device_set_attribute_value()
6161 to set sysfs attributes of a device.
6163 * The udev daemon now sets the default number of worker
6164 processes executed in parallel based on the number of available
6165 CPUs instead of the amount of available RAM. This is supposed
6166 to provide a more reliable default and limit a too aggressive
6167 parallelism for setups with 1000s of devices connected.
6169 Contributions from: Auke Kok, Colin Walters, Cristian
6170 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6171 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6172 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6173 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6174 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6175 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6176 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6177 Zbigniew Jędrzejewski-Szmek
6181 * Configuration of unit files may now be extended via drop-in
6182 files without having to edit/override the unit files
6183 themselves. More specifically, if the administrator wants to
6184 change one value for a service file foobar.service he can
6185 now do so by dropping in a configuration snippet into
6186 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6187 will load all these snippets and apply them on top of the
6188 main unit configuration file, possibly extending or
6189 overriding its settings. Using these drop-in snippets is
6190 generally nicer than the two earlier options for changing
6191 unit files locally: copying the files from
6192 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6193 them there; or creating a new file in /etc/systemd/system/
6194 that incorporates the original one via ".include". Drop-in
6195 snippets into these .d/ directories can be placed in any
6196 directory systemd looks for units in, and the usual
6197 overriding semantics between /usr/lib, /etc and /run apply
6200 * Most unit file settings which take lists of items can now be
6201 reset by assigning the empty string to them. For example,
6202 normally, settings such as Environment=FOO=BAR append a new
6203 environment variable assignment to the environment block,
6204 each time they are used. By assigning Environment= the empty
6205 string the environment block can be reset to empty. This is
6206 particularly useful with the .d/*.conf drop-in snippets
6207 mentioned above, since this adds the ability to reset list
6208 settings from vendor unit files via these drop-ins.
6210 * systemctl gained a new "list-dependencies" command for
6211 listing the dependencies of a unit recursively.
6213 * Inhibitors are now honored and listed by "systemctl
6214 suspend", "systemctl poweroff" (and similar) too, not only
6215 GNOME. These commands will also list active sessions by
6218 * Resource limits (as exposed by the various control group
6219 controllers) can now be controlled dynamically at runtime
6220 for all units. More specifically, you can now use a command
6221 like "systemctl set-cgroup-attr foobar.service cpu.shares
6222 2000" to alter the CPU shares a specific service gets. These
6223 settings are stored persistently on disk, and thus allow the
6224 administrator to easily adjust the resource usage of
6225 services with a few simple commands. This dynamic resource
6226 management logic is also available to other programs via the
6227 bus. Almost any kernel cgroup attribute and controller is
6230 * systemd-vconsole-setup will now copy all font settings to
6231 all allocated VTs, where it previously applied them only to
6234 * libsystemd-login gained the new sd_session_get_tty() API
6237 * This release drops support for a few legacy or
6238 distribution-specific LSB facility names when parsing init
6239 scripts: $x-display-manager, $mail-transfer-agent,
6240 $mail-transport-agent, $mail-transfer-agent, $smtp,
6241 $null. Also, the mail-transfer-agent.target unit backing
6242 this has been removed. Distributions which want to retain
6243 compatibility with this should carry the burden for
6244 supporting this themselves and patch support for these back
6245 in, if they really need to. Also, the facilities $syslog and
6246 $local_fs are now ignored, since systemd does not support
6247 early-boot LSB init scripts anymore, and these facilities
6248 are implied anyway for normal services. syslog.target has
6251 * There are new bus calls on PID1's Manager object for
6252 cancelling jobs, and removing snapshot units. Previously,
6253 both calls were only available on the Job and Snapshot
6256 * systemd-journal-gatewayd gained SSL support.
6258 * The various "environment" files, such as /etc/locale.conf
6259 now support continuation lines with a backslash ("\") as
6260 last character in the line, similarly in style (but different)
6261 to how this is supported in shells.
6263 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6264 now implicitly appended to every log entry logged. systemctl
6265 has been updated to filter by this field when operating on a
6266 user systemd instance.
6268 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6269 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6270 the container. This makes it easier to boot unmodified
6271 Fedora systems in a container, which however still requires
6272 audit=0 to be passed on the kernel command line. Auditing in
6273 kernel and userspace is unfortunately still too broken in
6274 context of containers, hence we recommend compiling it out
6275 of the kernel or using audit=0. Hopefully this will be fixed
6276 one day for good in the kernel.
6278 * nspawn gained the new --bind= and --bind-ro= parameters to
6279 bind mount specific directories from the host into the
6282 * nspawn will now mount its own devpts file system instance
6283 into the container, in order not to leak pty devices from
6284 the host into the container.
6286 * systemd will now read the firmware boot time performance
6287 information from the EFI variables, if the used boot loader
6288 supports this, and takes it into account for boot performance
6289 analysis via "systemd-analyze". This is currently supported
6290 only in conjunction with Gummiboot, but could be supported
6291 by other boot loaders too. For details see:
6293 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6295 * A new generator has been added that automatically mounts the
6296 EFI System Partition (ESP) to /boot, if that directory
6297 exists, is empty, and no other file system has been
6298 configured to be mounted there.
6300 * logind will now send out PrepareForSleep(false) out
6301 unconditionally, after coming back from suspend. This may be
6302 used by applications as asynchronous notification for
6303 system resume events.
6305 * "systemctl unlock-sessions" has been added, that allows
6306 unlocking the screens of all user sessions at once, similar
6307 to how "systemctl lock-sessions" already locked all users
6308 sessions. This is backed by a new D-Bus call UnlockSessions().
6310 * "loginctl seat-status" will now show the master device of a
6311 seat. (i.e. the device of a seat that needs to be around for
6312 the seat to be considered available, usually the graphics
6315 * tmpfiles gained a new "X" line type, that allows
6316 configuration of files and directories (with wildcards) that
6317 shall be excluded from automatic cleanup ("aging").
6319 * udev default rules set the device node permissions now only
6320 at "add" events, and do not change them any longer with a
6321 later "change" event.
6323 * The log messages for lid events and power/sleep keypresses
6324 now carry a message ID.
6326 * We now have a substantially larger unit test suite, but this
6327 continues to be work in progress.
6329 * udevadm hwdb gained a new --root= parameter to change the
6330 root directory to operate relative to.
6332 * logind will now issue a background sync() request to the kernel
6333 early at shutdown, so that dirty buffers are flushed to disk early
6334 instead of at the last moment, in order to optimize shutdown
6337 * A new bootctl tool has been added that is an interface for
6338 certain boot loader operations. This is currently a preview
6339 and is likely to be extended into a small mechanism daemon
6340 like timedated, localed, hostnamed, and can be used by
6341 graphical UIs to enumerate available boot options, and
6342 request boot into firmware operations.
6344 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6345 the rest of the package. It also has been updated to work
6346 correctly in initrds.
6348 * Policykit previously has been runtime optional, and is now
6349 also compile time optional via a configure switch.
6351 * systemd-analyze has been reimplemented in C. Also "systemctl
6352 dot" has moved into systemd-analyze.
6354 * "systemctl status" with no further parameters will now print
6355 the status of all active or failed units.
6357 * Operations such as "systemctl start" can now be executed
6358 with a new mode "--irreversible" which may be used to queue
6359 operations that cannot accidentally be reversed by a later
6360 job queuing. This is by default used to make shutdown
6361 requests more robust.
6363 * The Python API of systemd now gained a new module for
6364 reading journal files.
6366 * A new tool kernel-install has been added that can install
6367 kernel images according to the Boot Loader Specification:
6369 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6371 * Boot time console output has been improved to provide
6372 animated boot time output for hanging jobs.
6374 * A new tool systemd-activate has been added which can be used
6375 to test socket activation with, directly from the command
6376 line. This should make it much easier to test and debug
6377 socket activation in daemons.
6379 * journalctl gained a new "--reverse" (or -r) option to show
6380 journal output in reverse order (i.e. newest line first).
6382 * journalctl gained a new "--pager-end" (or -e) option to jump
6383 to immediately jump to the end of the journal in the
6384 pager. This is only supported in conjunction with "less".
6386 * journalctl gained a new "--user-unit=" option, that works
6387 similarly to "--unit=" but filters for user units rather than
6390 * A number of unit files to ease adoption of systemd in
6391 initrds has been added. This moves some minimal logic from
6392 the various initrd implementations into systemd proper.
6394 * The journal files are now owned by a new group
6395 "systemd-journal", which exists specifically to allow access
6396 to the journal, and nothing else. Previously, we used the
6397 "adm" group for that, which however possibly covers more
6398 than just journal/log file access. This new group is now
6399 already used by systemd-journal-gatewayd to ensure this
6400 daemon gets access to the journal files and as little else
6401 as possible. Note that "make install" will also set FS ACLs
6402 up for /var/log/journal to give "adm" and "wheel" read
6403 access to it, in addition to "systemd-journal" which owns
6404 the journal files. We recommend that packaging scripts also
6405 add read access to "adm" + "wheel" to /var/log/journal, and
6406 all existing/future journal files. To normal users and
6407 administrators little changes, however packagers need to
6408 ensure to create the "systemd-journal" system group at
6409 package installation time.
6411 * The systemd-journal-gatewayd now runs as unprivileged user
6412 systemd-journal-gateway:systemd-journal-gateway. Packaging
6413 scripts need to create these system user/group at
6416 * timedated now exposes a new boolean property CanNTP that
6417 indicates whether a local NTP service is available or not.
6419 * systemd-detect-virt will now also detect xen PVs
6421 * The pstore file system is now mounted by default, if it is
6424 * In addition to the SELinux and IMA policies we will now also
6425 load SMACK policies at early boot.
6427 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6428 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6429 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6430 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6431 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6432 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6433 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6434 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6435 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6436 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6437 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6438 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6439 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6440 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6444 * Timer units now support calendar time events in addition to
6445 monotonic time events. That means you can now trigger a unit
6446 based on a calendar time specification such as "Thu,Fri
6447 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6448 or fifth day of any month of the year 2013, given that it is
6449 a thursday or friday. This brings timer event support
6450 considerably closer to cron's capabilities. For details on
6451 the supported calendar time specification language see
6454 * udev now supports a number of different naming policies for
6455 network interfaces for predictable names, and a combination
6456 of these policies is now the default. Please see this wiki
6457 document for details:
6459 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6461 * Auke Kok's bootchart implementation has been added to the
6462 systemd tree. It is an optional component that can graph the
6463 boot in quite some detail. It is one of the best bootchart
6464 implementations around and minimal in its code and
6467 * nss-myhostname has been integrated into the systemd source
6468 tree. nss-myhostname guarantees that the local hostname
6469 always stays resolvable via NSS. It has been a weak
6470 requirement of systemd-hostnamed since a long time, and
6471 since its code is actually trivial we decided to just
6472 include it in systemd's source tree. It can be turned off
6473 with a configure switch.
6475 * The read-ahead logic is now capable of properly detecting
6476 whether a btrfs file system is on SSD or rotating media, in
6477 order to optimize the read-ahead scheme. Previously, it was
6478 only capable of detecting this on traditional file systems
6481 * In udev, additional device properties are now read from the
6482 IAB in addition to the OUI database. Also, Bluetooth company
6483 identities are attached to the devices as well.
6485 * In service files %U may be used as specifier that is
6486 replaced by the configured user name of the service.
6488 * nspawn may now be invoked without a controlling TTY. This
6489 makes it suitable for invocation as its own service. This
6490 may be used to set up a simple containerized server system
6491 using only core OS tools.
6493 * systemd and nspawn can now accept socket file descriptors
6494 when they are started for socket activation. This enables
6495 implementation of socket activated nspawn
6496 containers. i.e. think about autospawning an entire OS image
6497 when the first SSH or HTTP connection is received. We expect
6498 that similar functionality will also be added to libvirt-lxc
6501 * journalctl will now suppress ANSI color codes when
6502 presenting log data.
6504 * systemctl will no longer show control group information for
6505 a unit if the control group is empty anyway.
6507 * logind can now automatically suspend/hibernate/shutdown the
6510 * /etc/machine-info and hostnamed now also expose the chassis
6511 type of the system. This can be used to determine whether
6512 the local system is a laptop, desktop, handset or
6513 tablet. This information may either be configured by the
6514 user/vendor or is automatically determined from ACPI and DMI
6515 information if possible.
6517 * A number of PolicyKit actions are now bound together with
6518 "imply" rules. This should simplify creating UIs because
6519 many actions will now authenticate similar ones as well.
6521 * Unit files learnt a new condition ConditionACPower= which
6522 may be used to conditionalize a unit depending on whether an
6523 AC power source is connected or not, of whether the system
6524 is running on battery power.
6526 * systemctl gained a new "is-failed" verb that may be used in
6527 shell scripts and suchlike to check whether a specific unit
6528 is in the "failed" state.
6530 * The EnvironmentFile= setting in unit files now supports file
6531 globbing, and can hence be used to easily read a number of
6532 environment files at once.
6534 * systemd will no longer detect and recognize specific
6535 distributions. All distribution-specific #ifdeffery has been
6536 removed, systemd is now fully generic and
6537 distribution-agnostic. Effectively, not too much is lost as
6538 a lot of the code is still accessible via explicit configure
6539 switches. However, support for some distribution specific
6540 legacy configuration file formats has been dropped. We
6541 recommend distributions to simply adopt the configuration
6542 files everybody else uses now and convert the old
6543 configuration from packaging scripts. Most distributions
6544 already did that. If that's not possible or desirable,
6545 distributions are welcome to forward port the specific
6546 pieces of code locally from the git history.
6548 * When logging a message about a unit systemd will now always
6549 log the unit name in the message meta data.
6551 * localectl will now also discover system locale data that is
6552 not stored in locale archives, but directly unpacked.
6554 * logind will no longer unconditionally use framebuffer
6555 devices as seat masters, i.e. as devices that are required
6556 to be existing before a seat is considered preset. Instead,
6557 it will now look for all devices that are tagged as
6558 "seat-master" in udev. By default, framebuffer devices will
6559 be marked as such, but depending on local systems, other
6560 devices might be marked as well. This may be used to
6561 integrate graphics cards using closed source drivers (such
6562 as NVidia ones) more nicely into logind. Note however, that
6563 we recommend using the open source NVidia drivers instead,
6564 and no udev rules for the closed-source drivers will be
6565 shipped from us upstream.
6567 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6568 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6569 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6570 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6571 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6572 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6573 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6574 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6575 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6576 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6577 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6578 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6583 * udev gained support for loading additional device properties
6584 from an indexed database that is keyed by vendor/product IDs
6585 and similar device identifiers. For the beginning this
6586 "hwdb" is populated with data from the well-known PCI and
6587 USB database, but also includes PNP, ACPI and OID data. In
6588 the longer run this indexed database shall grow into
6589 becoming the one central database for non-essential
6590 userspace device metadata. Previously, data from the PCI/USB
6591 database was only attached to select devices, since the
6592 lookup was a relatively expensive operation due to O(n) time
6593 complexity (with n being the number of entries in the
6594 database). Since this is now O(1), we decided to add in this
6595 data for all devices where this is available, by
6596 default. Note that the indexed database needs to be rebuilt
6597 when new data files are installed. To achieve this you need
6598 to update your packaging scripts to invoke "udevadm hwdb
6599 --update" after installation of hwdb data files. For
6600 RPM-based distributions we introduced the new
6601 %udev_hwdb_update macro for this purpose.
6603 * The Journal gained support for the "Message Catalog", an
6604 indexed database to link up additional information with
6605 journal entries. For further details please check:
6607 https://www.freedesktop.org/wiki/Software/systemd/catalog
6609 The indexed message catalog database also needs to be
6610 rebuilt after installation of message catalog files. Use
6611 "journalctl --update-catalog" for this. For RPM-based
6612 distributions we introduced the %journal_catalog_update
6613 macro for this purpose.
6615 * The Python Journal bindings gained support for the standard
6616 Python logging framework.
6618 * The Journal API gained new functions for checking whether
6619 the underlying file system of a journal file is capable of
6620 properly reporting file change notifications, or whether
6621 applications that want to reflect journal changes "live"
6622 need to recheck journal files continuously in appropriate
6625 * It is now possible to set the "age" field for tmpfiles
6626 entries to 0, indicating that files matching this entry
6627 shall always be removed when the directories are cleaned up.
6629 * coredumpctl gained a new "gdb" verb which invokes gdb
6630 right-away on the selected coredump.
6632 * There's now support for "hybrid sleep" on kernels that
6633 support this, in addition to "suspend" and "hibernate". Use
6634 "systemctl hybrid-sleep" to make use of this.
6636 * logind's HandleSuspendKey= setting (and related settings)
6637 now gained support for a new "lock" setting to simply
6638 request the screen lock on all local sessions, instead of
6639 actually executing a suspend or hibernation.
6641 * systemd will now mount the EFI variables file system by
6644 * Socket units now gained support for configuration of the
6645 SMACK security label.
6647 * timedatectl will now output the time of the last and next
6648 daylight saving change.
6650 * We dropped support for various legacy and distro-specific
6651 concepts, such as insserv, early-boot SysV services
6652 (i.e. those for non-standard runlevels such as 'b' or 'S')
6653 or ArchLinux /etc/rc.conf support. We recommend the
6654 distributions who still need support this to either continue
6655 to maintain the necessary patches downstream, or find a
6656 different solution. (Talk to us if you have questions!)
6658 * Various systemd components will now bypass PolicyKit checks
6659 for root and otherwise handle properly if PolicyKit is not
6660 found to be around. This should fix most issues for
6661 PolicyKit-less systems. Quite frankly this should have been
6662 this way since day one. It is absolutely our intention to
6663 make systemd work fine on PolicyKit-less systems, and we
6664 consider it a bug if something does not work as it should if
6665 PolicyKit is not around.
6667 * For embedded systems it is now possible to build udev and
6668 systemd without blkid and/or kmod support.
6670 * "systemctl switch-root" is now capable of switching root
6671 more than once. I.e. in addition to transitions from the
6672 initrd to the host OS it is now possible to transition to
6673 further OS images from the host. This is useful to implement
6674 offline updating tools.
6676 * Various other additions have been made to the RPM macros
6677 shipped with systemd. Use %udev_rules_update() after
6678 installing new udev rules files. %_udevhwdbdir,
6679 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6680 %_sysctldir are now available which resolve to the right
6681 directories for packages to place various data files in.
6683 * journalctl gained the new --full switch (in addition to
6684 --all, to disable ellipsation for long messages.
6686 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6687 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6688 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6689 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6690 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6691 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6692 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6693 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6694 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6698 * journalctl gained new --since= and --until= switches to
6699 filter by time. It also now supports nice filtering for
6700 units via --unit=/-u.
6702 * Type=oneshot services may use ExecReload= and do the
6705 * The journal daemon now supports time-based rotation and
6706 vacuuming, in addition to the usual disk-space based
6709 * The journal will now index the available field values for
6710 each field name. This enables clients to show pretty drop
6711 downs of available match values when filtering. The bash
6712 completion of journalctl has been updated
6713 accordingly. journalctl gained a new switch -F to list all
6714 values a certain field takes in the journal database.
6716 * More service events are now written as structured messages
6717 to the journal, and made recognizable via message IDs.
6719 * The timedated, localed and hostnamed mini-services which
6720 previously only provided support for changing time, locale
6721 and hostname settings from graphical DEs such as GNOME now
6722 also have a minimal (but very useful) text-based client
6723 utility each. This is probably the nicest way to changing
6724 these settings from the command line now, especially since
6725 it lists available options and is fully integrated with bash
6728 * There's now a new tool "systemd-coredumpctl" to list and
6729 extract coredumps from the journal.
6731 * We now install a README each in /var/log/ and
6732 /etc/rc.d/init.d explaining where the system logs and init
6733 scripts went. This hopefully should help folks who go to
6734 that dirs and look into the otherwise now empty void and
6735 scratch their heads.
6737 * When user-services are invoked (by systemd --user) the
6738 $MANAGERPID env var is set to the PID of systemd.
6740 * SIGRTMIN+24 when sent to a --user instance will now result
6741 in immediate termination of systemd.
6743 * gatewayd received numerous feature additions such as a
6744 "follow" mode, for live syncing and filtering.
6746 * browse.html now allows filtering and showing detailed
6747 information on specific entries. Keyboard navigation and
6748 mouse screen support has been added.
6750 * gatewayd/journalctl now supports HTML5/JSON
6751 Server-Sent-Events as output.
6753 * The SysV init script compatibility logic will now
6754 heuristically determine whether a script supports the
6755 "reload" verb, and only then make this available as
6758 * "systemctl status --follow" has been removed, use "journalctl
6761 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6762 have been removed since they are hardly useful to be
6765 * And I'd like to take the opportunity to specifically mention
6766 Zbigniew for his great contributions. Zbigniew, you rock!
6768 Contributions from: Andrew Eikum, Christian Hesse, Colin
6769 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6770 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6771 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6772 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6773 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6774 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6778 * If /etc/vconsole.conf is non-existent or empty we will no
6779 longer load any console font or key map at boot by
6780 default. Instead the kernel defaults will be left
6781 intact. This is definitely the right thing to do, as no
6782 configuration should mean no configuration, and hard-coding
6783 font names that are different on all archs is probably a bad
6784 idea. Also, the kernel default key map and font should be
6785 good enough for most cases anyway, and mostly identical to
6786 the userspace fonts/key maps we previously overloaded them
6787 with. If distributions want to continue to default to a
6788 non-kernel font or key map they should ship a default
6789 /etc/vconsole.conf with the appropriate contents.
6791 Contributions from: Colin Walters, Daniel J Walsh, Dave
6792 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6793 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6797 * journalctl gained a new --cursor= switch to show entries
6798 starting from the specified location in the journal.
6800 * We now enforce a size limit on journal entry fields exported
6801 with "-o json" in journalctl. Fields larger than 4K will be
6802 assigned null. This can be turned off with --all.
6804 * An (optional) journal gateway daemon is now available as
6805 "systemd-journal-gatewayd.service". This service provides
6806 access to the journal via HTTP and JSON. This functionality
6807 will be used to implement live log synchronization in both
6808 pull and push modes, but has various other users too, such
6809 as easy log access for debugging of embedded devices. Right
6810 now it is already useful to retrieve the journal via HTTP:
6812 # systemctl start systemd-journal-gatewayd.service
6813 # wget http://localhost:19531/entries
6815 This will download the journal contents in a
6816 /var/log/messages compatible format. The same as JSON:
6818 # curl -H"Accept: application/json" http://localhost:19531/entries
6820 This service is also accessible via a web browser where a
6821 single static HTML5 app is served that uses the JSON logic
6822 to enable the user to do some basic browsing of the
6823 journal. This will be extended later on. Here's an example
6824 screenshot of this app in its current state:
6826 http://0pointer.de/public/journal-gatewayd
6828 Contributions from: Kay Sievers, Lennart Poettering, Robert
6829 Milasan, Tom Gundersen
6833 * The bash completion logic is now available for journalctl
6836 * We do not mount the "cpuset" controller anymore together with
6837 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6838 started if no parameters are assigned to it. "cpuset" hence
6839 broke code that assumed it could create "cpu" groups and
6842 * journalctl -f will now subscribe to terminal size changes,
6843 and line break accordingly.
6845 Contributions from: Dave Reisner, Kay Sievers, Lennart
6846 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6850 * nspawn will now create a symlink /etc/localtime in the
6851 container environment, copying the host's timezone
6852 setting. Previously this has been done via a bind mount, but
6853 since symlinks cannot be bind mounted this has now been
6854 changed to create/update the appropriate symlink.
6856 * journalctl -n's line number argument is now optional, and
6857 will default to 10 if omitted.
6859 * journald will now log the maximum size the journal files may
6860 take up on disk. This is particularly useful if the default
6861 built-in logic of determining this parameter from the file
6862 system size is used. Use "systemctl status
6863 systemd-journald.service" to see this information.
6865 * The multi-seat X wrapper tool has been stripped down. As X
6866 is now capable of enumerating graphics devices via udev in a
6867 seat-aware way the wrapper is not strictly necessary
6868 anymore. A stripped down temporary stop-gap is still shipped
6869 until the upstream display managers have been updated to
6870 fully support the new X logic. Expect this wrapper to be
6871 removed entirely in one of the next releases.
6873 * HandleSleepKey= in logind.conf has been split up into
6874 HandleSuspendKey= and HandleHibernateKey=. The old setting
6875 is not available anymore. X11 and the kernel are
6876 distinguishing between these keys and we should too. This
6877 also means the inhibition lock for these keys has been split
6880 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6881 Poettering, Lukas Nykryn, Václav Pavlín
6885 * Whenever a unit changes state we will now log this to the
6886 journal and show along the unit's own log output in
6889 * ConditionPathIsMountPoint= can now properly detect bind
6890 mount points too. (Previously, a bind mount of one file
6891 system to another place in the same file system could not be
6892 detected as mount, since they shared struct stat's st_dev
6895 * We will now mount the cgroup controllers cpu, cpuacct,
6896 cpuset and the controllers net_cls, net_prio together by
6899 * nspawn containers will now have a virtualized boot
6900 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6901 over with a randomized ID at container initialization). This
6902 has the effect of making "journalctl -b" do the right thing
6905 * The JSON output journal serialization has been updated not
6906 to generate "endless" list objects anymore, but rather one
6907 JSON object per line. This is more in line how most JSON
6908 parsers expect JSON objects. The new output mode
6909 "json-pretty" has been added to provide similar output, but
6910 neatly aligned for readability by humans.
6912 * We dropped all explicit sync() invocations in the shutdown
6913 code. The kernel does this implicitly anyway in the kernel
6914 reboot() syscall. halt(8)'s -n option is now a compatibility
6917 * We now support virtualized reboot() in containers, as
6918 supported by newer kernels. We will fall back to exit() if
6919 CAP_SYS_REBOOT is not available to the container. Also,
6920 nspawn makes use of this now and will actually reboot the
6921 container if the containerized OS asks for that.
6923 * journalctl will only show local log output by default
6924 now. Use --merge (-m) to show remote log output, too.
6926 * libsystemd-journal gained the new sd_journal_get_usage()
6927 call to determine the current disk usage of all journal
6928 files. This is exposed in the new "journalctl --disk-usage"
6931 * journald gained a new configuration setting SplitMode= in
6932 journald.conf which may be used to control how user journals
6933 are split off. See journald.conf(5) for details.
6935 * A new condition type ConditionFileNotEmpty= has been added.
6937 * tmpfiles' "w" lines now support file globbing, to write
6938 multiple files at once.
6940 * We added Python bindings for the journal submission
6941 APIs. More Python APIs for a number of selected APIs will
6942 likely follow. Note that we intend to add native bindings
6943 only for the Python language, as we consider it common
6944 enough to deserve bindings shipped within systemd. There are
6945 various projects outside of systemd that provide bindings
6946 for languages such as PHP or Lua.
6948 * Many conditions will now resolve specifiers such as %i. In
6949 addition, PathChanged= and related directives of .path units
6950 now support specifiers as well.
6952 * There's now a new RPM macro definition for the system preset
6955 * journald will now warn if it ca not forward a message to the
6956 syslog daemon because its socket is full.
6958 * timedated will no longer write or process /etc/timezone,
6959 except on Debian. As we do not support late mounted /usr
6960 anymore /etc/localtime always being a symlink is now safe,
6961 and hence the information in /etc/timezone is not necessary
6964 * logind will now always reserve one VT for a text getty (VT6
6965 by default). Previously if more than 6 X sessions where
6966 started they took up all the VTs with auto-spawned gettys,
6967 so that no text gettys were available anymore.
6969 * udev will now automatically inform the btrfs kernel logic
6970 about btrfs RAID components showing up. This should make
6971 simple hotplug based btrfs RAID assembly work.
6973 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6974 (but not for its children which will stay at the kernel
6975 default). This should allow setups with a lot more listening
6978 * systemd will now always pass the configured timezone to the
6979 kernel at boot. timedated will do the same when the timezone
6982 * logind's inhibition logic has been updated. By default,
6983 logind will now handle the lid switch, the power and sleep
6984 keys all the time, even in graphical sessions. If DEs want
6985 to handle these events on their own they should take the new
6986 handle-power-key, handle-sleep-key and handle-lid-switch
6987 inhibitors during their runtime. A simple way to achieve
6988 that is to invoke the DE wrapped in an invocation of:
6990 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6992 * Access to unit operations is now checked via SELinux taking
6993 the unit file label and client process label into account.
6995 * systemd will now notify the administrator in the journal
6996 when he over-mounts a non-empty directory.
6998 * There are new specifiers that are resolved in unit files,
6999 for the host name (%H), the machine ID (%m) and the boot ID
7002 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7003 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7004 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7005 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7006 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7007 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7008 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7012 * Support for reading structured kernel messages from
7013 /dev/kmsg has now been added and is enabled by default.
7015 * Support for reading kernel messages from /proc/kmsg has now
7016 been removed. If you want kernel messages in the journal
7017 make sure to run a recent kernel (>= 3.5) that supports
7018 reading structured messages from /dev/kmsg (see
7019 above). /proc/kmsg is now exclusive property of classic
7020 syslog daemons again.
7022 * The libudev API gained the new
7023 udev_device_new_from_device_id() call.
7025 * The logic for file system namespace (ReadOnlyDirectory=,
7026 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7027 require pivot_root() anymore. This means fewer temporary
7028 directories are created below /tmp for this feature.
7030 * nspawn containers will now see and receive all submounts
7031 made on the host OS below the root file system of the
7034 * Forward Secure Sealing is now supported for Journal files,
7035 which provide cryptographical sealing of journal files so
7036 that attackers cannot alter log history anymore without this
7037 being detectable. Lennart will soon post a blog story about
7038 this explaining it in more detail.
7040 * There are two new service settings RestartPreventExitStatus=
7041 and SuccessExitStatus= which allow configuration of exit
7042 status (exit code or signal) which will be excepted from the
7043 restart logic, resp. consider successful.
7045 * journalctl gained the new --verify switch that can be used
7046 to check the integrity of the structure of journal files and
7047 (if Forward Secure Sealing is enabled) the contents of
7050 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7051 and similar symlinks pre-created. This makes running shells
7052 as container init process a lot more fun.
7054 * The fstab support can now handle PARTUUID= and PARTLABEL=
7057 * A new ConditionHost= condition has been added to match
7058 against the hostname (with globs) and machine ID. This is
7059 useful for clusters where a single OS image is used to
7060 provision a large number of hosts which shall run slightly
7061 different sets of services.
7063 * Services which hit the restart limit will now be placed in a
7066 Contributions from: Bertram Poettering, Dave Reisner, Huang
7067 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7068 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7072 * When running in --user mode systemd will now become a
7073 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7074 tree a lot more organized.
7076 * A new PartOf= unit dependency type has been introduced that
7077 may be used to group services in a natural way.
7079 * "systemctl enable" may now be used to enable instances of
7082 * journalctl now prints error log levels in red, and
7083 warning/notice log levels in bright white. It also supports
7084 filtering by log level now.
7086 * cgtop gained a new -n switch (similar to top), to configure
7087 the maximum number of iterations to run for. It also gained
7088 -b, to run in batch mode (accepting no input).
7090 * The suffix ".service" may now be omitted on most systemctl
7091 command lines involving service unit names.
7093 * There's a new bus call in logind to lock all sessions, as
7094 well as a loginctl verb for it "lock-sessions".
7096 * libsystemd-logind.so gained a new call sd_journal_perror()
7097 that works similar to libc perror() but logs to the journal
7098 and encodes structured information about the error number.
7100 * /etc/crypttab entries now understand the new keyfile-size=
7103 * shutdown(8) now can send a (configurable) wall message when
7104 a shutdown is cancelled.
7106 * The mount propagation mode for the root file system will now
7107 default to "shared", which is useful to make containers work
7108 nicely out-of-the-box so that they receive new mounts from
7109 the host. This can be undone locally by running "mount
7110 --make-rprivate /" if needed.
7112 * The prefdm.service file has been removed. Distributions
7113 should maintain this unit downstream if they intend to keep
7114 it around. However, we recommend writing normal unit files
7115 for display managers instead.
7117 * Since systemd is a crucial part of the OS we will now
7118 default to a number of compiler switches that improve
7119 security (hardening) such as read-only relocations, stack
7120 protection, and suchlike.
7122 * The TimeoutSec= setting for services is now split into
7123 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7124 of individual time outs for the start and the stop phase of
7127 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7128 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7129 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7130 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7131 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7132 Gundersen, Zbigniew Jędrzejewski-Szmek
7136 * The journal and id128 C APIs are now fully documented as man
7139 * Extra safety checks have been added when transitioning from
7140 the initial RAM disk to the main system to avoid accidental
7143 * /etc/crypttab entries now understand the new keyfile-offset=
7146 * systemctl -t can now be used to filter by unit load state.
7148 * The journal C API gained the new sd_journal_wait() call to
7149 make writing synchronous journal clients easier.
7151 * journalctl gained the new -D switch to show journals from a
7154 * journalctl now displays a special marker between log
7155 messages of two different boots.
7157 * The journal is now explicitly flushed to /var via a service
7158 systemd-journal-flush.service, rather than implicitly simply
7159 by seeing /var/log/journal to be writable.
7161 * journalctl (and the journal C APIs) can now match for much
7162 more complex expressions, with alternatives and
7165 * When transitioning from the initial RAM disk to the main
7166 system we will now kill all processes in a killing spree to
7167 ensure no processes stay around by accident.
7169 * Three new specifiers may be used in unit files: %u, %h, %s
7170 resolve to the user name, user home directory resp. user
7171 shell. This is useful for running systemd user instances.
7173 * We now automatically rotate journal files if their data
7174 object hash table gets a fill level > 75%. We also size the
7175 hash table based on the configured maximum file size. This
7176 together should lower hash collisions drastically and thus
7177 speed things up a bit.
7179 * journalctl gained the new "--header" switch to introspect
7180 header data of journal files.
7182 * A new setting SystemCallFilters= has been added to services
7183 which may be used to apply blacklists or whitelists to
7184 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7186 * nspawn gained a new --link-journal= switch (and quicker: -j)
7187 to link the container journal with the host. This makes it
7188 very easy to centralize log viewing on the host for all
7189 guests while still keeping the journal files separated.
7191 * Many bugfixes and optimizations
7193 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7194 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7195 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7200 * Several tools now understand kernel command line arguments,
7201 which are only read when run in an initial RAM disk. They
7202 usually follow closely their normal counterparts, but are
7205 * There's a new tool to analyze the readahead files that are
7206 automatically generated at boot. Use:
7208 /usr/lib/systemd/systemd-readahead analyze /.readahead
7210 * We now provide an early debug shell on tty9 if this enabled. Use:
7212 systemctl enable debug-shell.service
7214 * All plymouth related units have been moved into the Plymouth
7215 package. Please make sure to upgrade your Plymouth version
7218 * systemd-tmpfiles now supports getting passed the basename of
7219 a configuration file only, in which case it will look for it
7220 in all appropriate directories automatically.
7222 * udevadm info now takes a /dev or /sys path as argument, and
7223 does the right thing. Example:
7225 udevadm info /dev/sda
7226 udevadm info /sys/class/block/sda
7228 * systemctl now prints a warning if a unit is stopped but a
7229 unit that might trigger it continues to run. Example: a
7230 service is stopped but the socket that activates it is left
7233 * "systemctl status" will now mention if the log output was
7234 shortened due to rotation since a service has been started.
7236 * The journal API now exposes functions to determine the
7237 "cutoff" times due to rotation.
7239 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7240 immediately flushing of runtime logs to /var if possible,
7241 resp. for triggering immediate rotation of the journal
7244 * It is now considered an error if a service is attempted to
7245 be stopped that is not loaded.
7247 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7249 * systemd-analyze now supports Python 3
7251 * tmpfiles now supports cleaning up directories via aging
7252 where the first level dirs are always kept around but
7253 directories beneath it automatically aged. This is enabled
7254 by prefixing the age field with '~'.
7256 * Seat objects now expose CanGraphical, CanTTY properties
7257 which is required to deal with very fast bootups where the
7258 display manager might be running before the graphics drivers
7259 completed initialization.
7261 * Seat objects now expose a State property.
7263 * We now include RPM macros for service enabling/disabling
7264 based on the preset logic. We recommend RPM based
7265 distributions to make use of these macros if possible. This
7266 makes it simpler to reuse RPM spec files across
7269 * We now make sure that the collected systemd unit name is
7270 always valid when services log to the journal via
7273 * There's a new man page kernel-command-line(7) detailing all
7274 command line options we understand.
7276 * The fstab generator may now be disabled at boot by passing
7277 fstab=0 on the kernel command line.
7279 * A new kernel command line option modules-load= is now understood
7280 to load a specific kernel module statically, early at boot.
7282 * Unit names specified on the systemctl command line are now
7283 automatically escaped as needed. Also, if file system or
7284 device paths are specified they are automatically turned
7285 into the appropriate mount or device unit names. Example:
7287 systemctl status /home
7288 systemctl status /dev/sda
7290 * The SysVConsole= configuration option has been removed from
7291 system.conf parsing.
7293 * The SysV search path is no longer exported on the D-Bus
7296 * The Names= option has been removed from unit file parsing.
7298 * There's a new man page bootup(7) detailing the boot process.
7300 * Every unit and every generator we ship with systemd now
7301 comes with full documentation. The self-explanatory boot is
7304 * A couple of services gained "systemd-" prefixes in their
7305 name if they wrap systemd code, rather than only external
7306 code. Among them fsck@.service which is now
7307 systemd-fsck@.service.
7309 * The HaveWatchdog property has been removed from the D-Bus
7312 * systemd.confirm_spawn= on the kernel command line should now
7315 * There's a new man page crypttab(5) which details all options
7316 we actually understand.
7318 * systemd-nspawn gained a new --capability= switch to pass
7319 additional capabilities to the container.
7321 * timedated will now read known NTP implementation unit names
7322 from /usr/lib/systemd/ntp-units.d/*.list,
7323 systemd-timedated-ntp.target has been removed.
7325 * journalctl gained a new switch "-b" that lists log data of
7326 the current boot only.
7328 * The notify socket is in the abstract namespace again, in
7329 order to support daemons which chroot() at start-up.
7331 * There is a new Storage= configuration option for journald
7332 which allows configuration of where log data should go. This
7333 also provides a way to disable journal logging entirely, so
7334 that data collected is only forwarded to the console, the
7335 kernel log buffer or another syslog implementation.
7337 * Many bugfixes and optimizations
7339 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7340 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7341 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7342 Shawn Landden, Tom Gundersen
7346 * "systemctl help <unit>" now shows the man page if one is
7349 * Several new man pages have been added.
7351 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7352 MaxLevelConsole= can now be specified in
7353 journald.conf. These options allow reducing the amount of
7354 data stored on disk or forwarded by the log level.
7356 * TimerSlackNSec= can now be specified in system.conf for
7357 PID1. This allows system-wide power savings.
7359 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7360 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7365 * logind is now capable of (optionally) handling power and
7366 sleep keys as well as the lid switch.
7368 * journalctl now understands the syntax "journalctl
7369 /usr/bin/avahi-daemon" to get all log output of a specific
7372 * CapabilityBoundingSet= in system.conf now also influences
7373 the capability bound set of usermode helpers of the kernel.
7375 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7376 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7377 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7378 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7382 * Note that we skipped 139 releases here in order to set the
7383 new version to something that is greater than both udev's
7384 and systemd's most recent version number.
7386 * udev: all udev sources are merged into the systemd source tree now.
7387 All future udev development will happen in the systemd tree. It
7388 is still fully supported to use the udev daemon and tools without
7389 systemd running, like in initramfs or other init systems. Building
7390 udev though, will require the *build* of the systemd tree, but
7391 udev can be properly *run* without systemd.
7393 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7394 should be used to create dead device nodes as workarounds for broken
7397 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7398 no longer supported. udev_monitor_new_from_netlink() needs to be
7399 used to subscribe to events.
7401 * udev: when udevd is started by systemd, processes which are left
7402 behind by forking them off of udev rules, are unconditionally cleaned
7403 up and killed now after the event handling has finished. Services or
7404 daemons must be started as systemd services. Services can be
7405 pulled-in by udev to get started, but they can no longer be directly
7406 forked by udev rules.
7408 * udev: the daemon binary is called systemd-udevd now and installed
7409 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7410 to adapt to that, create symlink, or rename the binary after building
7413 * libudev no longer provides these symbols:
7414 udev_monitor_from_socket()
7415 udev_queue_get_failed_list_entry()
7416 udev_get_{dev,sys,run}_path()
7417 The versions number was bumped and symbol versioning introduced.
7419 * systemd-loginctl and systemd-journalctl have been renamed
7420 to loginctl and journalctl to match systemctl.
7422 * The config files: /etc/systemd/systemd-logind.conf and
7423 /etc/systemd/systemd-journald.conf have been renamed to
7424 logind.conf and journald.conf. Package updates should rename
7425 the files to the new names on upgrade.
7427 * For almost all files the license is now LGPL2.1+, changed
7428 from the previous GPL2.0+. Exceptions are some minor stuff
7429 of udev (which will be changed to LGPL2.1 eventually, too),
7430 and the MIT licensed sd-daemon.[ch] library that is suitable
7431 to be used as drop-in files.
7433 * systemd and logind now handle system sleep states, in
7434 particular suspending and hibernating.
7436 * logind now implements a sleep/shutdown/idle inhibiting logic
7437 suitable for a variety of uses. Soonishly Lennart will blog
7438 about this in more detail.
7440 * var-run.mount and var-lock.mount are no longer provided
7441 (which previously bind mounted these directories to their new
7442 places). Distributions which have not converted these
7443 directories to symlinks should consider stealing these files
7444 from git history and add them downstream.
7446 * We introduced the Documentation= field for units and added
7447 this to all our shipped units. This is useful to make it
7448 easier to explore the boot and the purpose of the various
7451 * All smaller setup units (such as
7452 systemd-vconsole-setup.service) now detect properly if they
7453 are run in a container and are skipped when
7454 appropriate. This guarantees an entirely noise-free boot in
7455 Linux container environments such as systemd-nspawn.
7457 * A framework for implementing offline system updates is now
7458 integrated, for details see:
7459 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7461 * A new service type Type=idle is available now which helps us
7462 avoiding ugly interleaving of getty output and boot status
7465 * There's now a system-wide CapabilityBoundingSet= option to
7466 globally reduce the set of capabilities for the
7467 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7468 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7469 even CAP_NET_ADMIN system-wide for secure systems.
7471 * There are now system-wide DefaultLimitXXX= options to
7472 globally change the defaults of the various resource limits
7473 for all units started by PID 1.
7475 * Harald Hoyer's systemd test suite has been integrated into
7476 systemd which allows easy testing of systemd builds in qemu
7477 and nspawn. (This is really awesome! Ask us for details!)
7479 * The fstab parser is now implemented as generator, not inside
7482 * systemctl will now warn you if .mount units generated from
7483 /etc/fstab are out of date due to changes in fstab that
7484 have not been read by systemd yet.
7486 * systemd is now suitable for usage in initrds. Dracut has
7487 already been updated to make use of this. With this in place
7488 initrds get a slight bit faster but primarily are much
7489 easier to introspect and debug since "systemctl status" in
7490 the host system can be used to introspect initrd services,
7491 and the journal from the initrd is kept around too.
7493 * systemd-delta has been added, a tool to explore differences
7494 between user/admin configuration and vendor defaults.
7496 * PrivateTmp= now affects both /tmp and /var/tmp.
7498 * Boot time status messages are now much prettier and feature
7499 proper english language. Booting up systemd has never been
7502 * Read-ahead pack files now include the inode number of all
7503 files to pre-cache. When the inode changes the pre-caching
7504 is not attempted. This should be nicer to deal with updated
7505 packages which might result in changes of read-ahead
7508 * We now temporaritly lower the kernel's read_ahead_kb variable
7509 when collecting read-ahead data to ensure the kernel's
7510 built-in read-ahead does not add noise to our measurements
7511 of necessary blocks to pre-cache.
7513 * There's now RequiresMountsFor= to add automatic dependencies
7514 for all mounts necessary for a specific file system path.
7516 * MountAuto= and SwapAuto= have been removed from
7517 system.conf. Mounting file systems at boot has to take place
7520 * nspawn now learned a new switch --uuid= to set the machine
7521 ID on the command line.
7523 * nspawn now learned the -b switch to automatically search
7526 * vt102 is now the default TERM for serial TTYs, upgraded from
7529 * systemd-logind now works on VT-less systems.
7531 * The build tree has been reorganized. The individual
7532 components now have directories of their own.
7534 * A new condition type ConditionPathIsReadWrite= is now available.
7536 * nspawn learned the new -C switch to create cgroups for the
7537 container in other hierarchies.
7539 * We now have support for hardware watchdogs, configurable in
7542 * The scheduled shutdown logic now has a public API.
7544 * We now mount /tmp as tmpfs by default, but this can be
7545 masked and /etc/fstab can override it.
7547 * Since udisks does not make use of /media anymore we are not
7548 mounting a tmpfs on it anymore.
7550 * journalctl gained a new --local switch to only interleave
7551 locally generated journal files.
7553 * We can now load the IMA policy at boot automatically.
7555 * The GTK tools have been split off into a systemd-ui.
7557 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7558 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7559 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7560 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7561 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7562 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7563 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7564 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7565 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7570 * This is mostly a bugfix release
7572 * Support optional initialization of the machine ID from the
7573 KVM or container configured UUID.
7575 * Support immediate reboots with "systemctl reboot -ff"
7577 * Show /etc/os-release data in systemd-analyze output
7579 * Many bugfixes for the journal, including endianness fixes and
7580 ensuring that disk space enforcement works
7582 * sd-login.h is C++ compatible again
7584 * Extend the /etc/os-release format on request of the Debian
7587 * We now refuse non-UTF8 strings used in various configuration
7588 and unit files. This is done to ensure we do not pass invalid
7589 data over D-Bus or expose it elsewhere.
7591 * Register Mimo USB Screens as suitable for automatic seat
7594 * Read SELinux client context from journal clients in a race
7597 * Reorder configuration file lookup order. /etc now always
7598 overrides /run in order to allow the administrator to always
7599 and unconditionally override vendor-supplied or
7600 automatically generated data.
7602 * The various user visible bits of the journal now have man
7603 pages. We still lack man pages for the journal API calls
7606 * We now ship all man pages in HTML format again in the
7609 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7610 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7611 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7616 * This is mostly a bugfix release
7618 * systems lacking /etc/os-release are no longer supported.
7620 * Various functionality updates to libsystemd-login.so
7622 * Track class of PAM logins to distinguish greeters from
7625 Contributions from: Kay Sievers, Lennart Poettering, Michael
7630 * This is an important bugfix release for v41.
7632 * Building man pages is now optional which should be useful
7633 for those building systemd from git but unwilling to install
7636 * Watchdog support for supervising services is now usable. In
7637 a future release support for hardware watchdogs
7638 (i.e. /dev/watchdog) will be added building on this.
7640 * Service start rate limiting is now configurable and can be
7641 turned off per service. When a start rate limit is hit a
7642 reboot can automatically be triggered.
7644 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7646 Contributions from: Benjamin Franzke, Bill Nottingham,
7647 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7648 Schmidt, Michał Górny, Piotr Drąg
7652 * The systemd binary is installed /usr/lib/systemd/systemd now;
7653 An existing /sbin/init symlink needs to be adapted with the
7656 * The code that loads kernel modules has been ported to invoke
7657 libkmod directly, instead of modprobe. This means we do not
7658 support systems with module-init-tools anymore.
7660 * Watchdog support is now already useful, but still not
7663 * A new kernel command line option systemd.setenv= is
7664 understood to set system wide environment variables
7665 dynamically at boot.
7667 * We now limit the set of capabilities of systemd-journald.
7669 * We now set SIGPIPE to ignore by default, since it only is
7670 useful in shell pipelines, and has little use in general
7671 code. This can be disabled with IgnoreSIPIPE=no in unit
7674 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7675 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7680 * This is mostly a bugfix release
7682 * We now expose the reason why a service failed in the
7683 "Result" D-Bus property.
7685 * Rudimentary service watchdog support (will be completed over
7686 the next few releases.)
7688 * When systemd forks off in order execute some service we will
7689 now immediately changes its argv[0] to reflect which process
7690 it will execute. This is useful to minimize the time window
7691 with a generic argv[0], which makes bootcharts more useful
7693 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7694 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7695 Mike Kazantsev, Ray Strode
7699 * This is mostly a test release, but incorporates many
7702 * New systemd-cgtop tool to show control groups by their
7705 * Linking against libacl for ACLs is optional again. If
7706 disabled, support tracking device access for active logins
7707 goes becomes unavailable, and so does access to the user
7708 journals by the respective users.
7710 * If a group "adm" exists, journal files are automatically
7711 owned by them, thus allow members of this group full access
7712 to the system journal as well as all user journals.
7714 * The journal now stores the SELinux context of the logging
7715 client for all entries.
7717 * Add C++ inclusion guards to all public headers
7719 * New output mode "cat" in the journal to print only text
7720 messages, without any meta data like date or time.
7722 * Include tiny X server wrapper as a temporary stop-gap to
7723 teach XOrg udev display enumeration. This is used by display
7724 managers such as gdm, and will go away as soon as XOrg
7725 learned native udev hotplugging for display devices.
7727 * Add new systemd-cat tool for executing arbitrary programs
7728 with STDERR/STDOUT connected to the journal. Can also act as
7729 BSD logger replacement, and does so by default.
7731 * Optionally store all locally generated coredumps in the
7732 journal along with meta data.
7734 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7735 writing short strings to files (for usage for /sys), and for
7736 creating symlinks, character and block device nodes.
7738 * New unit file option ControlGroupPersistent= to make cgroups
7739 persistent, following the mechanisms outlined in
7740 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7742 * Support multiple local RTCs in a sane way
7744 * No longer monopolize IO when replaying readahead data on
7745 rotating disks, since we might starve non-file-system IO to
7746 death, since fanotify() will not see accesses done by blkid,
7749 * Do not show kernel threads in systemd-cgls anymore, unless
7750 requested with new -k switch.
7752 Contributions from: Dan Horák, Kay Sievers, Lennart
7753 Poettering, Michal Schmidt
7757 * This is mostly a test release, but incorporates many
7760 * The git repository moved to:
7761 git://anongit.freedesktop.org/systemd/systemd
7762 ssh://git.freedesktop.org/git/systemd/systemd
7764 * First release with the journal
7765 http://0pointer.de/blog/projects/the-journal.html
7767 * The journal replaces both systemd-kmsg-syslogd and
7768 systemd-stdout-bridge.
7770 * New sd_pid_get_unit() API call in libsystemd-logind
7772 * Many systemadm clean-ups
7774 * Introduce remote-fs-pre.target which is ordered before all
7775 remote mounts and may be used to start services before all
7778 * Added Mageia support
7780 * Add bash completion for systemd-loginctl
7782 * Actively monitor PID file creation for daemons which exit in
7783 the parent process before having finished writing the PID
7784 file in the daemon process. Daemons which do this need to be
7785 fixed (i.e. PID file creation must have finished before the
7786 parent exits), but we now react a bit more gracefully to them.
7788 * Add colourful boot output, mimicking the well-known output
7789 of existing distributions.
7791 * New option PassCredentials= for socket units, for
7792 compatibility with a recent kernel ABI breakage.
7794 * /etc/rc.local is now hooked in via a generator binary, and
7795 thus will no longer act as synchronization point during
7798 * systemctl list-unit-files now supports --root=.
7800 * systemd-tmpfiles now understands two new commands: z, Z for
7801 relabelling files according to the SELinux database. This is
7802 useful to apply SELinux labels to specific files in /sys,
7805 * Output of SysV services is now forwarded to both the console
7806 and the journal by default, not only just the console.
7808 * New man pages for all APIs from libsystemd-login.
7810 * The build tree got reorganized and the build system is a
7811 lot more modular allowing embedded setups to specifically
7812 select the components of systemd they are interested in.
7814 * Support for Linux systems lacking the kernel VT subsystem is
7817 * configure's --with-rootdir= got renamed to
7818 --with-rootprefix= to follow the naming used by udev and
7821 * Unless specified otherwise we will now install to /usr instead
7822 of /usr/local by default.
7824 * Processes with '@' in argv[0][0] are now excluded from the
7825 final shut-down killing spree, following the logic explained
7827 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7829 * All processes remaining in a service cgroup when we enter
7830 the START or START_PRE states are now killed with
7831 SIGKILL. That means it is no longer possible to spawn
7832 background processes from ExecStart= lines (which was never
7833 supported anyway, and bad style).
7835 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7836 reloading of units together.
7838 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7839 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7840 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7841 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7842 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob