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 * elogind-resolved's DNSSEC support gained support for RFC 8080
189 (Ed25519 keys and signatures).
191 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
192 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
193 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
194 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
195 John Ledkov, Dmitry Rozhkov, Edward A. James, Evgeny Vereshchagin,
196 Florian Klink, Franck Bui, Gwendal Grignou, Hans de Goede, Harald
197 Hoyer, Hristo Venev, Ikey Doherty, Jakub Wilk, Jérémy Rosen, Jiahui
198 Xie, John Lin, José Bollo, Josef Andersson, juga0, Krzysztof Nowicki,
199 Lars Karlitski, Lars Kellogg-Stedman, Lauri Tirkkonen, Lennart
200 Poettering, Lubomir Rintel, Luca Bruno, Lucas Werkmeister, Lukáš
201 Nykrýn, Lukáš Říha, Lukasz Rubaszewski, Maciej S. Szmigiero, Mantas
202 Mikulėnas, Marcus Folkesson, Martin Steuer, Mathieu Trudel-Lapierre,
203 Matija Skala, Matthias-Christian Ott, Max Resch, Michael Biebl, Michael
204 Vogt, Michal Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil
205 Brown, Olaf Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik,
206 Peter Hutterer, Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland
207 Hieber, Saran Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang
208 Liu, Simon Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner,
209 Susant Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago
210 Salem Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten
211 Hilbrich, Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe,
212 Zbigniew Jędrzejewski-Szmek, Zeal Jagannatha
218 * INCOMPATIBILITY: systemd-logind.service and other long-running
219 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
220 communication with the outside. This generally improves security of
221 the system, and is in almost all cases a safe and good choice, as
222 these services do not and should not provide any network-facing
223 functionality. However, systemd-logind uses the glibc NSS API to
224 query the user database. This creates problems on systems where NSS
225 is set up to directly consult network services for user database
226 lookups. In particular, this creates incompatibilities with the
227 "nss-nis" module, which attempts to directly contact the NIS/YP
228 network servers it is configured for, and will now consistently
229 fail. In such cases, it is possible to turn off IP sandboxing for
230 systemd-logind.service (set IPAddressDeny= in its [Service] section
231 to the empty string, via a .d/ unit file drop-in). Downstream
232 distributions might want to update their nss-nis packaging to include
233 such a drop-in snippet, accordingly, to hide this incompatibility
234 from the user. Another option is to make use of glibc's nscd service
235 to proxy such network requests through a privilege-separated, minimal
236 local caching daemon, or to switch to more modern technologies such
237 sssd, whose NSS hook-ups generally do not involve direct network
238 access. In general, we think it's definitely time to question the
239 implementation choices of nss-nis, i.e. whether it's a good idea
240 today to embed a network-facing loadable module into all local
241 processes that need to query the user database, including the most
242 trivial and benign ones, such as "ls". For more details about
243 IPAddressDeny= see below.
245 * A new modprobe.d drop-in is now shipped by default that sets the
246 bonding module option max_bonds=0. This overrides the kernel default,
247 to avoid conflicts and ambiguity as to whether or not bond0 should be
248 managed by systemd-networkd or not. This resolves multiple issues
249 with bond0 properties not being applied, when bond0 is configured
250 with systemd-networkd. Distributors may choose to not package this,
251 however in that case users will be prevented from correctly managing
252 bond0 interface using systemd-networkd.
254 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
255 which print the logging level and target of the system manager. They
256 complement the existing "set-log-level" and "set-log-target" verbs
257 used to change those values.
259 * journald.conf gained a new boolean setting ReadKMsg= which defaults
260 to on. If turned off kernel log messages will not be read by
261 systemd-journald or included in the logs. It also gained a new
262 setting LineMax= for configuring the maximum line length in
263 STDOUT/STDERR log streams. The new default for this value is 48K, up
264 from the previous hardcoded 2048.
266 * A new unit setting RuntimeDirectoryPreserve= has been added, which
267 allows more detailed control of what to do with a runtime directory
268 configured with RuntimeDirectory= (i.e. a directory below /run or
269 $XDG_RUNTIME_DIR) after a unit is stopped.
271 * The RuntimeDirectory= setting for units gained support for creating
272 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
273 one top-level directory.
275 * Units gained new options StateDirectory=, CacheDirectory=,
276 LogsDirectory= and ConfigurationDirectory= which are closely related
277 to RuntimeDirectory= but manage per-service directories below
278 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
279 possible to write unit files which when activated automatically gain
280 properly owned service specific directories in these locations, thus
281 making unit files self-contained and increasing compatibility with
282 stateless systems and factory reset where /etc or /var are
283 unpopulated at boot. Matching these new settings there's also
284 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
285 ConfigurationDirectoryMode= for configuring the access mode of these
286 directories. These settings are particularly useful in combination
287 with DynamicUser=yes as they provide secure, properly-owned,
288 writable, and stateful locations for storage, excluded from the
289 sandbox that such services live in otherwise.
291 * Automake support has been removed from this release. systemd is now
294 * systemd-journald will now aggressively cache client metadata during
295 runtime, speeding up log write performance under pressure. This comes
296 at a small price though: as much of the metadata is read
297 asynchronously from /proc/ (and isn't implicitly attached to log
298 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
299 metadata stored alongside a log entry might be slightly
300 out-of-date. Previously it could only be slightly newer than the log
301 message. The time window is small however, and given that the kernel
302 is unlikely to be improved anytime soon in this regard, this appears
305 * nss-myhostname/systemd-resolved will now by default synthesize an
306 A/AAAA resource record for the "_gateway" hostname, pointing to the
307 current default IP gateway. Previously it did that for the "gateway"
308 name, hampering adoption, as some distributions wanted to leave that
309 host name open for local use. The old behaviour may still be
310 requested at build time.
312 * systemd-networkd's [Address] section in .network files gained a new
313 Scope= setting for configuring the IP address scope. The [Network]
314 section gained a new boolean setting ConfigureWithoutCarrier= that
315 tells systemd-networkd to ignore link sensing when configuring the
316 device. The [DHCP] section gained a new Anonymize= boolean option for
317 turning on a number of options suggested in RFC 7844. A new
318 [RoutingPolicyRule] section has been added for configuring the IP
319 routing policy. The [Route] section has gained support for a new
320 Type= setting which permits configuring
321 blackhole/unreachable/prohibit routes.
323 * The [VRF] section in .netdev files gained a new Table= setting for
324 configuring the routing table to use. The [Tunnel] section gained a
325 new Independent= boolean field for configuring tunnels independent of
326 an underlying network interface. The [Bridge] section gained a new
327 GroupForwardMask= option for configuration of propagation of link
328 local frames between bridge ports.
330 * The WakeOnLan= setting in .link files gained support for a number of
331 new modes. A new TCP6SegmentationOffload= setting has been added for
332 configuring TCP/IPv6 hardware segmentation offload.
334 * The IPv6 RA sender implementation may now optionally send out RDNSS
335 and RDNSSL records to supply DNS configuration to peers.
337 * systemd-nspawn gained support for a new --system-call-filter= command
338 line option for adding and removing entries in the default system
339 call filter it applies. Moreover systemd-nspawn has been changed to
340 implement a system call whitelist instead of a blacklist.
342 * systemd-run gained support for a new --pipe command line option. If
343 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
344 are directly passed on to the activated transient service
345 executable. This allows invoking arbitrary processes as systemd
346 services (for example to take benefit of dependency management,
347 accounting management, resource management or log management that is
348 done automatically for services) — while still allowing them to be
349 integrated in a classic UNIX shell pipeline.
351 * When a service sends RELOAD=1 via sd_notify() and reload propagation
352 using ReloadPropagationTo= is configured, a reload is now propagated
353 to configured units. (Previously this was only done on explicitly
354 requested reloads, using "systemctl reload" or an equivalent
357 * For each service unit a restart counter is now kept: it is increased
358 each time the service is restarted due to Restart=, and may be
359 queried using "systemctl show -p NRestarts …".
361 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
362 @signal and @timer have been added, for usage with SystemCallFilter=
363 in unit files and the new --system-call-filter= command line option
364 of systemd-nspawn (see above).
366 * ExecStart= lines in unit files gained two new modifiers: when a
367 command line is prefixed with "!" the command will be executed as
368 configured, except for the credentials applied by
369 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
370 "+", but does still apply namespacing options unlike "+". There's
371 also "!!" now, which is mostly identical, but becomes a NOP on
372 systems that support ambient capabilities. This is useful to write
373 unit files that work with ambient capabilities where possible but
374 automatically fall back to traditional privilege dropping mechanisms
375 on systems where this is not supported.
377 * ListenNetlink= settings in socket units now support RDMA netlink
380 * A new unit file setting LockPersonality= has been added which permits
381 locking down the chosen execution domain ("personality") of a service
384 * A new special target "getty-pre.target" has been added, which is
385 ordered before all text logins, and may be used to order services
386 before textual logins acquire access to the console.
388 * systemd will now attempt to load the virtio-rng.ko kernel module very
389 early on if a VM environment supporting this is detected. This should
390 improve entropy during early boot in virtualized environments.
392 * A _netdev option is now supported in /etc/crypttab that operates in a
393 similar way as the same option in /etc/fstab: it permits configuring
394 encrypted devices that need to be ordered after the network is up.
395 Following this logic, two new special targets
396 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
397 added that are to cryptsetup.target what remote-fs.target and
398 remote-fs-pre.target are to local-fs.target.
400 * Service units gained a new UnsetEnvironment= setting which permits
401 unsetting specific environment variables for services that are
402 normally passed to it (for example in order to mask out locale
403 settings for specific services that can't deal with it).
405 * Units acquired a new boolean option IPAccounting=. When turned on, IP
406 traffic accounting (packet count as well as byte count) is done for
407 the service, and shown as part of "systemctl status" or "systemd-run
410 * Service units acquired two new options IPAddressAllow= and
411 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
412 for configuring a simple IP access control list for all sockets of
413 the unit. These options are available also on .slice and .socket
414 units, permitting flexible access list configuration for individual
415 services as well as groups of services (as defined by a slice unit),
416 including system-wide. Note that IP ACLs configured this way are
417 enforced on every single IPv4 and IPv6 socket created by any process
418 of the service unit, and apply to ingress as well as egress traffic.
420 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
421 structured log message is generated each time the unit is stopped,
422 containing information about the consumed resources of this
425 * A new setting KeyringMode= has been added to unit files, which may be
426 used to control how the kernel keyring is set up for executed
429 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
430 "systemctl kexec" and "systemctl exit" are now always asynchronous in
431 behaviour (that is: these commands return immediately after the
432 operation was enqueued instead of waiting for the operation to
433 complete). Previously, "systemctl poweroff" and "systemctl reboot"
434 were asynchronous on systems using systemd-logind (i.e. almost
435 always, and like they were on sysvinit), and the other three commands
436 were unconditionally synchronous. With this release this is cleaned
437 up, and callers will see the same asynchronous behaviour on all
438 systems for all five operations.
440 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
443 * .timer units now accept calendar specifications in other timezones
444 than UTC or the local timezone.
446 * The tmpfiles snippet var.conf has been changed to create
447 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
448 the "utmp" group already, and it appears to be generally understood
449 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
450 databases. Previously this was implemented correctly for all these
451 databases excepts btmp, which has been opened up like this now
452 too. Note that while the other databases are world-readable
453 (i.e. 0644), btmp is not and remains more restrictive.
455 * The systemd-resolve tool gained a new --reset-server-features
456 switch. When invoked like this systemd-resolved will forget
457 everything it learnt about the features supported by the configured
458 upstream DNS servers, and restarts the feature probing logic on the
459 next resolver look-up for them at the highest feature level
462 * The status dump systemd-resolved sends to the logs upon receiving
463 SIGUSR1 now also includes information about all DNS servers it is
464 configured to use, and the features levels it probed for them.
466 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
467 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
468 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
469 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
470 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
471 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
472 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
473 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
474 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
475 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
476 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
477 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
478 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
479 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
480 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
481 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
482 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
483 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
484 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
485 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
491 * Meson is now supported as build system in addition to Automake. It is
492 our plan to remove Automake in one of our next releases, so that
493 Meson becomes our exclusive build system. Hence, please start using
494 the Meson build system in your downstream packaging. There's plenty
495 of documentation around how to use Meson, the extremely brief
498 ./autogen.sh && ./configure && make && sudo make install
502 meson build && ninja -C build && sudo ninja -C build install
504 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
505 which permits configuring a timeout on the time a job is
506 running. This is particularly useful for setting timeouts on jobs for
509 * Unit files gained two new options ConditionUser= and ConditionGroup=
510 for conditionalizing units based on the identity of the user/group
511 running a systemd user instance.
513 * systemd-networkd now understands a new FlowLabel= setting in the
514 [VXLAN] section of .network files, as well as a Priority= in
515 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
516 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
517 gained support for configuration of GENEVE links, and IPv6 address
518 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
520 * .link files now understand a new Port= setting.
522 * systemd-networkd's DHCP support gained support for DHCP option 119
523 (domain search list).
525 * systemd-networkd gained support for serving IPv6 address ranges using
526 the Router Advertisment protocol. The new .network configuration
527 section [IPv6Prefix] may be used to configure the ranges to
528 serve. This is implemented based on a new, minimal, native server
529 implementation of RA.
531 * journalctl's --output= switch gained support for a new parameter
532 "short-iso-precise" for a mode where timestamps are shown as precise
535 * systemd-udevd's "net_id" builtin may now generate stable network
536 interface names from IBM PowerVM VIO devices as well as ACPI platform
539 * MulticastDNS support in systemd-resolved may now be explicitly
540 enabled/disabled using the new MulticastDNS= configuration file
543 * systemd-resolved may now optionally use libidn2 instead of the libidn
544 for processing internationalized domain names. Support for libidn2
545 should be considered experimental and should not be enabled by
548 * "machinectl pull-tar" and related call may now do verification of
549 downloaded images using SUSE-style .sha256 checksum files in addition
550 to the already existing support for validating using Ubuntu-style
553 * sd-bus gained support for a new sd_bus_message_appendv() call which
554 is va_list equivalent of sd_bus_message_append().
556 * sd-boot gained support for validating images using SHIM/MOK.
558 * The SMACK code learnt support for "onlycap".
560 * systemd-mount --umount is now much smarter in figuring out how to
561 properly unmount a device given its mount or device path.
563 * The code to call libnss_dns as a fallback from libnss_resolve when
564 the communication with systemd-resolved fails was removed. This
565 fallback was redundant and interfered with the [!UNAVAIL=return]
566 suffix. See nss-resolve(8) for the recommended configuration.
568 * systemd-logind may now be restarted without losing state. It stores
569 the file descriptors for devices it manages in the system manager
570 using the FDSTORE= mechanism. Please note that further changes in
571 other components may be required to make use of this (for example
572 Xorg has code to listen for stops of systemd-logind and terminate
573 itself when logind is stopped or restarted, in order to avoid using
574 stale file descriptors for graphical devices, which is now
575 counterproductive and must be reverted in order for restarts of
576 systemd-logind to be safe. See
577 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
579 * All kernel install plugins are called with the environment variable
580 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
581 /etc/machine-id. If the file is missing or empty, the variable is
582 empty and BOOT_DIR_ABS is the path of a temporary directory which is
583 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
584 is empty, all plugins should not put anything in BOOT_DIR_ABS.
586 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
587 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
588 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
589 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
590 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
591 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
592 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
593 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
594 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
595 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
596 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
597 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
598 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
599 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
600 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
601 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
602 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
603 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
604 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
605 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
606 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
607 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
608 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
609 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
610 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
611 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
612 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
613 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
614 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
621 * The "hybrid" control group mode has been modified to improve
622 compatibility with "legacy" cgroups-v1 setups. Specifically, the
623 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
624 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
625 cgroups-v1 hierarchy), the only externally visible change being that
626 the cgroups-v2 hierarchy is also mounted, to
627 /sys/fs/cgroup/unified. This should provide a large degree of
628 compatibility with "legacy" cgroups-v1, while taking benefit of the
629 better management capabilities of cgroups-v2.
631 * The default control group setup mode may be selected both a boot-time
632 via a set of kernel command line parameters (specifically:
633 systemd.unified_cgroup_hierarchy= and
634 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
635 default selected on the configure command line
636 (--with-default-hierarchy=). The upstream default is "hybrid"
637 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
638 this will change in a future systemd version to be "unified" (pure
639 cgroups-v2 mode). The third option for the compile time option is
640 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
641 distributions to default to "hybrid" mode for release distributions,
642 starting with v233. We recommend "unified" for development
643 distributions (specifically: distributions such as Fedora's rawhide)
644 as that's where things are headed in the long run. Use "legacy" for
645 greatest stability and compatibility only.
647 * Note one current limitation of "unified" and "hybrid" control group
648 setup modes: the kernel currently does not permit the systemd --user
649 instance (i.e. unprivileged code) to migrate processes between two
650 disconnected cgroup subtrees, even if both are managed and owned by
651 the user. This effectively means "systemd-run --user --scope" doesn't
652 work when invoked from outside of any "systemd --user" service or
653 scope. Specifically, it is not supported from session scopes. We are
654 working on fixing this in a future systemd version. (See #3388 for
655 further details about this.)
657 * DBus policy files are now installed into /usr rather than /etc. Make
658 sure your system has dbus >= 1.9.18 running before upgrading to this
659 version, or override the install path with --with-dbuspolicydir= .
661 * All python scripts shipped with systemd (specifically: the various
662 tests written in Python) now require Python 3.
664 * systemd unit tests can now run standalone (without the source or
665 build directories), and can be installed into /usr/lib/systemd/tests/
666 with 'make install-tests'.
668 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
669 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
672 * Support for the %c, %r, %R specifiers in unit files has been
673 removed. Specifiers are not supposed to be dependent on configuration
674 in the unit file itself (so that they resolve the same regardless
675 where used in the unit files), but these specifiers were influenced
676 by the Slice= option.
678 * The shell invoked by debug-shell.service now defaults to /bin/sh in
679 all cases. If distributions want to use a different shell for this
680 purpose (for example Fedora's /sbin/sushell) they need to specify
681 this explicitly at configure time using --with-debug-shell=.
683 * The confirmation spawn prompt has been reworked to offer the
686 (c)ontinue, proceed without asking anymore
687 (D)ump, show the state of the unit
688 (f)ail, don't execute the command and pretend it failed
690 (i)nfo, show a short summary of the unit
691 (j)obs, show jobs that are in progress
692 (s)kip, don't execute the command and pretend it succeeded
693 (y)es, execute the command
695 The 'n' choice for the confirmation spawn prompt has been removed,
696 because its meaning was confusing.
698 The prompt may now also be redirected to an alternative console by
699 specifying the console as parameter to systemd.confirm_spawn=.
701 * Services of Type=notify require a READY=1 notification to be sent
702 during startup. If no such message is sent, the service now fails,
703 even if the main process exited with a successful exit code.
705 * Services that fail to start up correctly now always have their
706 ExecStopPost= commands executed. Previously, they'd enter "failed"
707 state directly, without executing these commands.
709 * The option MulticastDNS= of network configuration files has acquired
710 an actual implementation. With MulticastDNS=yes a host can resolve
711 names of remote hosts and reply to mDNS A and AAAA requests.
713 * When units are about to be started an additional check is now done to
714 ensure that all dependencies of type BindsTo= (when used in
715 combination with After=) have been started.
717 * systemd-analyze gained a new verb "syscall-filter" which shows which
718 system call groups are defined for the SystemCallFilter= unit file
719 setting, and which system calls they contain.
721 * A new system call filter group "@filesystem" has been added,
722 consisting of various file system related system calls. Group
723 "@reboot" has been added, covering reboot, kexec and shutdown related
724 calls. Finally, group "@swap" has been added covering swap
725 configuration related calls.
727 * A new unit file option RestrictNamespaces= has been added that may be
728 used to restrict access to the various process namespace types the
729 Linux kernel provides. Specifically, it may be used to take away the
730 right for a service unit to create additional file system, network,
731 user, and other namespaces. This sandboxing option is particularly
732 relevant due to the high amount of recently discovered namespacing
733 related vulnerabilities in the kernel.
735 * systemd-udev's .link files gained support for a new AutoNegotiation=
736 setting for configuring Ethernet auto-negotiation.
738 * systemd-networkd's .network files gained support for a new
739 ListenPort= setting in the [DHCP] section to explicitly configure the
740 UDP client port the DHCP client shall listen on.
742 * .network files gained a new Unmanaged= boolean setting for explicitly
743 excluding one or more interfaces from management by systemd-networkd.
745 * The systemd-networkd ProxyARP= option has been renamed to
746 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
747 renamed to ReduceARPProxy=. The old names continue to be available
750 * systemd-networkd gained support for configuring IPv6 Proxy NDP
751 addresses via the new IPv6ProxyNDPAddress= .network file setting.
753 * systemd-networkd's bonding device support gained support for two new
754 configuration options ActiveSlave= and PrimarySlave=.
756 * The various options in the [Match] section of .network files gained
757 support for negative matching.
759 * New systemd-specific mount options are now understood in /etc/fstab:
761 x-systemd.mount-timeout= may be used to configure the maximum
762 permitted runtime of the mount command.
764 x-systemd.device-bound may be set to bind a mount point to its
765 backing device unit, in order to automatically remove a mount point
766 if its backing device is unplugged. This option may also be
767 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
768 on the block device, which is now automatically set for all CDROM
769 drives, so that mounted CDs are automatically unmounted when they are
770 removed from the drive.
772 x-systemd.after= and x-systemd.before= may be used to explicitly
773 order a mount after or before another unit or mount point.
775 * Enqueued start jobs for device units are now automatically garbage
776 collected if there are no jobs waiting for them anymore.
778 * systemctl list-jobs gained two new switches: with --after, for every
779 queued job the jobs it's waiting for are shown; with --before the
780 jobs which it's blocking are shown.
782 * systemd-nspawn gained support for ephemeral boots from disk images
783 (or in other words: --ephemeral and --image= may now be
784 combined). Moreover, ephemeral boots are now supported for normal
785 directories, even if the backing file system is not btrfs. Of course,
786 if the file system does not support file system snapshots or
787 reflinks, the initial copy operation will be relatively expensive, but
788 this should still be suitable for many use cases.
790 * Calendar time specifications in .timer units now support
791 specifications relative to the end of a month by using "~" instead of
792 "-" as separator between month and day. For example, "*-02~03" means
793 "the third last day in February". In addition a new syntax for
794 repeated events has been added using the "/" character. For example,
795 "9..17/2:00" means "every two hours from 9am to 5pm".
797 * systemd-socket-proxyd gained a new parameter --connections-max= for
798 configuring the maximum number of concurrent connections.
800 * sd-id128 gained a new API for generating unique IDs for the host in a
801 way that does not leak the machine ID. Specifically,
802 sd_id128_get_machine_app_specific() derives an ID based on the
803 machine ID a in well-defined, non-reversible, stable way. This is
804 useful whenever an identifier for the host is needed but where the
805 identifier shall not be useful to identify the system beyond the
806 scope of the application itself. (Internally this uses HMAC-SHA256 as
807 keyed hash function using the machine ID as input.)
809 * NotifyAccess= gained a new supported value "exec". When set
810 notifications are accepted from all processes systemd itself invoked,
811 including all control processes.
813 * .nspawn files gained support for defining overlay mounts using the
814 Overlay= and OverlayReadOnly= options. Previously this functionality
815 was only available on the systemd-nspawn command line.
817 * systemd-nspawn's --bind= and --overlay= options gained support for
818 bind/overlay mounts whose source lies within the container tree by
819 prefixing the source path with "+".
821 * systemd-nspawn's --bind= and --overlay= options gained support for
822 automatically allocating a temporary source directory in /var/tmp
823 that is removed when the container dies. Specifically, if the source
824 directory is specified as empty string this mechanism is selected. An
825 example usage is --overlay=+/var::/var, which creates an overlay
826 mount based on the original /var contained in the image, overlayed
827 with a temporary directory in the host's /var/tmp. This way changes
828 to /var are automatically flushed when the container shuts down.
830 * systemd-nspawn --image= option does now permit raw file system block
831 devices (in addition to images containing partition tables, as
834 * The disk image dissection logic in systemd-nspawn gained support for
835 automatically setting up LUKS encrypted as well as Verity protected
836 partitions. When a container is booted from an encrypted image the
837 passphrase is queried at start-up time. When a container with Verity
838 data is started, the root hash is search in a ".roothash" file
839 accompanying the disk image (alternatively, pass the root hash via
840 the new --root-hash= command line option).
842 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
843 be used to dissect disk images the same way as systemd-nspawn does
844 it, following the Bootable Partition Specification. It may even be
845 used to mount disk images with complex partition setups (including
846 LUKS and Verity partitions) to a local host directory, in order to
847 inspect them. This tool is not considered public API (yet), and is
848 thus not installed into /usr/bin. Please do not rely on its
849 existence, since it might go away or be changed in later systemd
852 * A new generator "systemd-verity-generator" has been added, similar in
853 style to "systemd-cryptsetup-generator", permitting automatic setup of
854 Verity root partitions when systemd boots up. In order to make use of
855 this your partition setup should follow the Discoverable Partitions
856 Specification, and the GPT partition ID of the root file system
857 partition should be identical to the upper 128bit of the Verity root
858 hash. The GPT partition ID of the Verity partition protecting it
859 should be the lower 128bit of the Verity root hash. If the partition
860 image follows this model it is sufficient to specify a single
861 "roothash=" kernel command line argument to both configure which root
862 image and verity partition to use as well as the root hash for
863 it. Note that systemd-nspawn's Verity support follows the same
864 semantics, meaning that disk images with proper Verity data in place
865 may be booted in containers with systemd-nspawn as well as on
866 physical systems via the verity generator. Also note that the "mkosi"
867 tool available at https://github.com/systemd/mkosi has been updated
868 to generate Verity protected disk images following this scheme. In
869 fact, it has been updated to generate disk images that optionally
870 implement a complete UEFI SecureBoot trust chain, involving a signed
871 kernel and initrd image that incorporates such a root hash as well as
872 a Verity-enabled root partition.
874 * The hardware database (hwdb) udev supports has been updated to carry
875 accelerometer quirks.
877 * All system services are now run with a fresh kernel keyring set up
878 for them. The invocation ID is stored by default in it, thus
879 providing a safe, non-overridable way to determine the invocation
882 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
883 options for bind mounting arbitrary paths in a service-specific
884 way. When these options are used, arbitrary host or service files and
885 directories may be mounted to arbitrary locations in the service's
888 * Documentation has been added that lists all of systemd's low-level
889 environment variables:
891 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
893 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
894 whether a specific socket file descriptor matches a specified socket
897 * systemd-firstboot has been updated to check for the
898 systemd.firstboot= kernel command line option. It accepts a boolean
899 and when set to false the first boot questions are skipped.
901 * systemd-fstab-generator has been updated to check for the
902 systemd.volatile= kernel command line option, which either takes an
903 optional boolean parameter or the special value "state". If used the
904 system may be booted in a "volatile" boot mode. Specifically,
905 "systemd.volatile" is used, the root directory will be mounted as
906 tmpfs, and only /usr is mounted from the actual root file system. If
907 "systemd.volatile=state" is used, the root directory will be mounted
908 as usual, but /var is mounted as tmpfs. This concept provides similar
909 functionality as systemd-nspawn's --volatile= option, but provides it
910 on physical boots. Use this option for implementing stateless
911 systems, or testing systems with all state and/or configuration reset
912 to the defaults. (Note though that many distributions are not
913 prepared to boot up without a populated /etc or /var, though.)
915 * systemd-gpt-auto-generator gained support for LUKS encrypted root
916 partitions. Previously it only supported LUKS encrypted partitions
917 for all other uses, except for the root partition itself.
919 * Socket units gained support for listening on AF_VSOCK sockets for
920 communication in virtualized QEMU environments.
922 * The "configure" script gained a new option --with-fallback-hostname=
923 for specifying the fallback hostname to use if none is configured in
924 /etc/hostname. For example, by specifying
925 --with-fallback-hostname=fedora it is possible to default to a
926 hostname of "fedora" on pristine installations.
928 * systemd-cgls gained support for a new --unit= switch for listing only
929 the control groups of a specific unit. Similar --user-unit= has been
930 added for listing only the control groups of a specific user unit.
932 * systemd-mount gained a new --umount switch for unmounting a mount or
933 automount point (and all mount/automount points below it).
935 * systemd will now refuse full configuration reloads (via systemctl
936 daemon-reload and related calls) unless at least 16MiB of free space
937 are available in /run. This is a safety precaution in order to ensure
938 that generators can safely operate after the reload completed.
940 * A new unit file option RootImage= has been added, which has a similar
941 effect as RootDirectory= but mounts the service's root directory from
942 a disk image instead of plain directory. This logic reuses the same
943 image dissection and mount logic that systemd-nspawn already uses,
944 and hence supports any disk images systemd-nspawn supports, including
945 those following the Discoverable Partition Specification, as well as
946 Verity enabled images. This option enables systemd to run system
947 services directly off disk images acting as resource bundles,
948 possibly even including full integrity data.
950 * A new MountAPIVFS= unit file option has been added, taking a boolean
951 argument. If enabled /proc, /sys and /dev (collectively called the
952 "API VFS") will be mounted for the service. This is only relevant if
953 RootDirectory= or RootImage= is used for the service, as these mounts
954 are of course in place in the host mount namespace anyway.
956 * systemd-nspawn gained support for a new --pivot-root= switch. If
957 specified the root directory within the container image is pivoted to
958 the specified mount point, while the original root disk is moved to a
959 different place. This option enables booting of ostree images
960 directly with systemd-nspawn.
962 * The systemd build scripts will no longer complain if the NTP server
963 addresses are not changed from the defaults. Google now supports
964 these NTP servers officially. We still recommend downstreams to
965 properly register an NTP pool with the NTP pool project though.
967 * coredumpctl gained a new "--reverse" option for printing the list
968 of coredumps in reverse order.
970 * coredumpctl will now show additional information about truncated and
971 inaccessible coredumps, as well as coredumps that are still being
972 processed. It also gained a new --quiet switch for suppressing
973 additional informational message in its output.
975 * coredumpctl gained support for only showing coredumps newer and/or
976 older than specific timestamps, using the new --since= and --until=
977 options, reminiscent of journalctl's options by the same name.
979 * The systemd-coredump logic has been improved so that it may be reused
980 to collect backtraces in non-compiled languages, for example in
981 scripting languages such as Python.
983 * machinectl will now show the UID shift of local containers, if user
984 namespacing is enabled for them.
986 * systemd will now optionally run "environment generator" binaries at
987 configuration load time. They may be used to add environment
988 variables to the environment block passed to services invoked. One
989 user environment generator is shipped by default that sets up
990 environment variables based on files dropped into /etc/environment.d
991 and ~/.config/environment.d/.
993 * systemd-resolved now includes the new, recently published 2017 DNSSEC
996 * hostnamed has been updated to report a new chassis type of
997 "convertible" to cover "foldable" laptops that can both act as a
998 tablet and as a laptop, such as various Lenovo Yoga devices.
1000 Contributions from: Adrián López, Alexander Galanin, Alexander
1001 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1002 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1003 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1004 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1005 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1006 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1007 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1008 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1009 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1010 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1011 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1012 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1013 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1014 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1015 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1016 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1017 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1018 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1019 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1020 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1021 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1022 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1023 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1024 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1025 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1026 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1027 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1030 — Berlin, 2017-03-01
1034 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1035 RestrictAddressFamilies= enabled. These sandboxing options should
1036 generally be compatible with the various external udev call-out
1037 binaries we are aware of, however there may be exceptions, in
1038 particular when exotic languages for these call-outs are used. In
1039 this case, consider turning off these settings locally.
1041 * The new RemoveIPC= option can be used to remove IPC objects owned by
1042 the user or group of a service when that service exits.
1044 * The new ProtectKernelModules= option can be used to disable explicit
1045 load and unload operations of kernel modules by a service. In
1046 addition access to /usr/lib/modules is removed if this option is set.
1048 * ProtectSystem= option gained a new value "strict", which causes the
1049 whole file system tree with the exception of /dev, /proc, and /sys,
1050 to be remounted read-only for a service.
1052 * The new ProtectKernelTunables= option can be used to disable
1053 modification of configuration files in /sys and /proc by a service.
1054 Various directories and files are remounted read-only, so access is
1055 restricted even if the file permissions would allow it.
1057 * The new ProtectControlGroups= option can be used to disable write
1058 access by a service to /sys/fs/cgroup.
1060 * Various systemd services have been hardened with
1061 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1062 RestrictAddressFamilies=.
1064 * Support for dynamically creating users for the lifetime of a service
1065 has been added. If DynamicUser=yes is specified, user and group IDs
1066 will be allocated from the range 61184..65519 for the lifetime of the
1067 service. They can be resolved using the new nss-systemd.so NSS
1068 module. The module must be enabled in /etc/nsswitch.conf. Services
1069 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1070 any resources allocated by the service will be cleaned up when the
1071 service exits. They also have ProtectHome=read-only and
1072 ProtectSystem=strict enabled, so they are not able to make any
1073 permanent modifications to the system.
1075 * The nss-systemd module also always resolves root and nobody, making
1076 it possible to have no /etc/passwd or /etc/group files in minimal
1077 container or chroot environments.
1079 * Services may be started with their own user namespace using the new
1080 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1081 under which the service is running are mapped. All other users are
1084 * Support for the cgroup namespace has been added to systemd-nspawn. If
1085 supported by kernel, the container system started by systemd-nspawn
1086 will have its own view of the cgroup hierarchy. This new behaviour
1087 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1089 * The new MemorySwapMax= option can be used to limit the maximum swap
1090 usage under the unified cgroup hierarchy.
1092 * Support for the CPU controller in the unified cgroup hierarchy has
1093 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1094 options. This controller requires out-of-tree patches for the kernel
1095 and the support is provisional.
1097 * Mount and automount units may now be created transiently
1098 (i.e. dynamically at runtime via the bus API, instead of requiring
1099 unit files in the file system).
1101 * systemd-mount is a new tool which may mount file systems – much like
1102 mount(8), optionally pulling in additional dependencies through
1103 transient .mount and .automount units. For example, this tool
1104 automatically runs fsck on a backing block device before mounting,
1105 and allows the automount logic to be used dynamically from the
1106 command line for establishing mount points. This tool is particularly
1107 useful when dealing with removable media, as it will ensure fsck is
1108 run – if necessary – before the first access and that the file system
1109 is quickly unmounted after each access by utilizing the automount
1110 logic. This maximizes the chance that the file system on the
1111 removable media stays in a clean state, and if it isn't in a clean
1112 state is fixed automatically.
1114 * LazyUnmount=yes option for mount units has been added to expose the
1115 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1118 * /efi will be used as the mount point of the EFI boot partition, if
1119 the directory is present, and the mount point was not configured
1120 through other means (e.g. fstab). If /efi directory does not exist,
1121 /boot will be used as before. This makes it easier to automatically
1122 mount the EFI partition on systems where /boot is used for something
1125 * When operating on GPT disk images for containers, systemd-nspawn will
1126 now mount the ESP to /boot or /efi according to the same rules as PID
1127 1 running on a host. This allows tools like "bootctl" to operate
1128 correctly within such containers, in order to make container images
1129 bootable on physical systems.
1131 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1133 * Two new user session targets have been added to support running
1134 graphical sessions under the systemd --user instance:
1135 graphical-session.target and graphical-session-pre.target. See
1136 systemd.special(7) for a description of how those targets should be
1139 * The vconsole initialization code has been significantly reworked to
1140 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1141 support unicode keymaps. Font and keymap configuration will now be
1142 copied to all allocated virtual consoles.
1144 * FreeBSD's bhyve virtualization is now detected.
1146 * Information recorded in the journal for core dumps now includes the
1147 contents of /proc/mountinfo and the command line of the process at
1148 the top of the process hierarchy (which is usually the init process
1151 * systemd-journal-gatewayd learned the --directory= option to serve
1152 files from the specified location.
1154 * journalctl --root=… can be used to peruse the journal in the
1155 /var/log/ directories inside of a container tree. This is similar to
1156 the existing --machine= option, but does not require the container to
1159 * The hardware database has been extended to support
1160 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1163 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1164 specify the click rate for mice which include a horizontal wheel with
1165 a click rate that is different than the one for the vertical wheel.
1167 * systemd-run gained a new --wait option that makes service execution
1168 synchronous. (Specifically, the command will not return until the
1169 specified service binary exited.)
1171 * systemctl gained a new --wait option that causes the start command to
1172 wait until the units being started have terminated again.
1174 * A new journal output mode "short-full" has been added which displays
1175 timestamps with abbreviated English day names and adds a timezone
1176 suffix. Those timestamps include more information than the default
1177 "short" output mode, and can be passed directly to journalctl's
1178 --since= and --until= options.
1180 * /etc/resolv.conf will be bind-mounted into containers started by
1181 systemd-nspawn, if possible, so any changes to resolv.conf contents
1182 are automatically propagated to the container.
1184 * The number of instances for socket-activated services originating
1185 from a single IP address can be limited with
1186 MaxConnectionsPerSource=, extending the existing setting of
1189 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1192 * .netdev and .network configuration can now be extended through
1195 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1196 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1197 can be enabled and disabled using the new UDPSegmentationOffload=,
1198 TCPSegmentationOffload=, GenericSegmentationOffload=,
1199 GenericReceiveOffload=, LargeReceiveOffload= options in the
1200 [Link] section of .link files.
1202 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1203 Port VLAN ID can be configured for bridge devices using the new STP=,
1204 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1205 section of .netdev files.
1207 * The route table to which routes received over DHCP or RA should be
1208 added can be configured with the new RouteTable= option in the [DHCP]
1209 and [IPv6AcceptRA] sections of .network files.
1211 * The Address Resolution Protocol can be disabled on links managed by
1212 systemd-networkd using the ARP=no setting in the [Link] section of
1215 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1216 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1217 encode information about the result and exit codes of the current
1218 service runtime cycle.
1220 * systemd-sysctl will now configure kernel parameters in the order
1221 they occur in the configuration files. This matches what sysctl
1222 has been traditionally doing.
1224 * kernel-install "plugins" that are executed to perform various
1225 tasks after a new kernel is added and before an old one is removed
1226 can now return a special value to terminate the procedure and
1227 prevent any later plugins from running.
1229 * Journald's SplitMode=login setting has been deprecated. It has been
1230 removed from documentation, and its use is discouraged. In a future
1231 release it will be completely removed, and made equivalent to current
1232 default of SplitMode=uid.
1234 * Storage=both option setting in /etc/systemd/coredump.conf has been
1235 removed. With fast LZ4 compression storing the core dump twice is not
1238 * The --share-system systemd-nspawn option has been replaced with an
1239 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1240 this functionality is discouraged. In addition the variables
1241 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1242 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1243 individual namespaces.
1245 * "machinectl list" now shows the IP address of running containers in
1246 the output, as well as OS release information.
1248 * "loginctl list" now shows the TTY of each session in the output.
1250 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1251 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1252 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1253 tracking objects in a "recursive" mode, where a single client can be
1254 counted multiple times, if it takes multiple references.
1256 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1257 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1258 process using sd-bus automatically exit if the bus connection is
1261 * Bus clients of the service manager may now "pin" loaded units into
1262 memory, by taking an explicit reference on them. This is useful to
1263 ensure the client can retrieve runtime data about the service even
1264 after the service completed execution. Taking such a reference is
1265 available only for privileged clients and should be helpful to watch
1266 running services in a race-free manner, and in particular collect
1267 information about exit statuses and results.
1269 * The nss-resolve module has been changed to strictly return UNAVAIL
1270 when communication via D-Bus with resolved failed, and NOTFOUND when
1271 a lookup completed but was negative. This means it is now possible to
1272 neatly configure fallbacks using nsswitch.conf result checking
1273 expressions. Taking benefit of this, the new recommended
1274 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1276 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1278 * A new setting CtrlAltDelBurstAction= has been added to
1279 /etc/systemd/system.conf which may be used to configure the precise
1280 behaviour if the user on the console presses Ctrl-Alt-Del more often
1281 than 7 times in 2s. Previously this would unconditionally result in
1282 an expedited, immediate reboot. With this new setting the precise
1283 operation may be configured in more detail, and also turned off
1286 * In .netdev files two new settings RemoteChecksumTx= and
1287 RemoteChecksumRx= are now understood that permit configuring the
1288 remote checksumming logic for VXLAN networks.
1290 * The service manager learnt a new "invocation ID" concept for invoked
1291 services. Each runtime cycle of a service will get a new invocation
1292 ID (a 128bit random UUID) assigned that identifies the current
1293 run of the service uniquely and globally. A new invocation ID
1294 is generated each time a service starts up. The journal will store
1295 the invocation ID of a service along with any logged messages, thus
1296 making the invocation ID useful for matching the online runtime of a
1297 service with the offline log data it generated in a safe way without
1298 relying on synchronized timestamps. In many ways this new service
1299 invocation ID concept is similar to the kernel's boot ID concept that
1300 uniquely and globally identifies the runtime of each boot. The
1301 invocation ID of a service is passed to the service itself via an
1302 environment variable ($INVOCATION_ID). A new bus call
1303 GetUnitByInvocationID() has been added that is similar to GetUnit()
1304 but instead of retrieving the bus path for a unit by its name
1305 retrieves it by its invocation ID. The returned path is valid only as
1306 long as the passed invocation ID is current.
1308 * systemd-resolved gained a new "DNSStubListener" setting in
1309 resolved.conf. It either takes a boolean value or the special values
1310 "udp" and "tcp", and configures whether to enable the stub DNS
1311 listener on 127.0.0.53:53.
1313 * IP addresses configured via networkd may now carry additional
1314 configuration settings supported by the kernel. New options include:
1315 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1316 PrefixRoute=, AutoJoin=.
1318 * The PAM configuration fragment file for "user@.service" shipped with
1319 systemd (i.e. the --user instance of systemd) has been stripped to
1320 the minimum necessary to make the system boot. Previously, it
1321 contained Fedora-specific stanzas that did not apply to other
1322 distributions. It is expected that downstream distributions add
1323 additional configuration lines, matching their needs to this file,
1324 using it only as rough template of what systemd itself needs. Note
1325 that this reduced fragment does not even include an invocation of
1326 pam_limits which most distributions probably want to add, even though
1327 systemd itself does not need it. (There's also the new build time
1328 option --with-pamconfdir=no to disable installation of the PAM
1331 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1332 capability is now also dropped from its set (in addition to
1333 CAP_SYS_MKNOD as before).
1335 * In service unit files it is now possible to connect a specific named
1336 file descriptor with stdin/stdout/stdout of an executed service. The
1337 name may be specified in matching .socket units using the
1338 FileDescriptorName= setting.
1340 * A number of journal settings may now be configured on the kernel
1341 command line. Specifically, the following options are now understood:
1342 systemd.journald.max_level_console=,
1343 systemd.journald.max_level_store=,
1344 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1345 systemd.journald.max_level_wall=.
1347 * "systemctl is-enabled --full" will now show by which symlinks a unit
1348 file is enabled in the unit dependency tree.
1350 * Support for VeraCrypt encrypted partitions has been added to the
1351 "cryptsetup" logic and /etc/crypttab.
1353 * systemd-detect-virt gained support for a new --private-users switch
1354 that checks whether the invoking processes are running inside a user
1355 namespace. Similar, a new special value "private-users" for the
1356 existing ConditionVirtualization= setting has been added, permitting
1357 skipping of specific units in user namespace environments.
1359 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1360 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1361 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1362 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1363 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1364 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1365 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1366 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1367 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1368 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1369 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1370 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1371 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1372 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1373 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1374 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1375 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1376 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1377 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1378 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1379 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1380 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1381 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1382 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1383 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1384 Jędrzejewski-Szmek, Zeal Jagannatha
1386 — Santa Fe, 2016-11-03
1390 * In service units the various ExecXYZ= settings have been extended
1391 with an additional special character as first argument of the
1392 assigned value: if the character '+' is used the specified command
1393 line it will be run with full privileges, regardless of User=,
1394 Group=, CapabilityBoundingSet= and similar options. The effect is
1395 similar to the existing PermissionsStartOnly= option, but allows
1396 configuration of this concept for each executed command line
1399 * Services may now alter the service watchdog timeout at runtime by
1400 sending a WATCHDOG_USEC= message via sd_notify().
1402 * MemoryLimit= and related unit settings now optionally take percentage
1403 specifications. The percentage is taken relative to the amount of
1404 physical memory in the system (or in case of containers, the assigned
1405 amount of memory). This allows scaling service resources neatly with
1406 the amount of RAM available on the system. Similarly, systemd-logind's
1407 RuntimeDirectorySize= option now also optionally takes percentage
1410 * In similar fashion TasksMax= takes percentage values now, too. The
1411 value is taken relative to the configured maximum number of processes
1412 on the system. The per-service task maximum has been changed to 15%
1413 using this functionality. (Effectively this is an increase of 512 →
1414 4915 for service units, given the kernel's default pid_max setting.)
1416 * Calendar time specifications in .timer units now understand a ".."
1417 syntax for time ranges. Example: "4..7:10" may now be used for
1418 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1421 * The InaccessableDirectories=, ReadOnlyDirectories= and
1422 ReadWriteDirectories= unit file settings have been renamed to
1423 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1424 applied to all kinds of file nodes, and not just directories, with
1425 the exception of symlinks. Specifically these settings may now be
1426 used on block and character device nodes, UNIX sockets and FIFOS as
1427 well as regular files. The old names of these settings remain
1428 available for compatibility.
1430 * systemd will now log about all service processes it kills forcibly
1431 (using SIGKILL) because they remained after the clean shutdown phase
1432 of the service completed. This should help identifying services that
1433 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1434 systemd-logind's configuration a similar log message is generated for
1435 processes killed at the end of each session due to this setting.
1437 * systemd will now set the $JOURNAL_STREAM environment variable for all
1438 services whose stdout/stderr are connected to the Journal (which
1439 effectively means by default: all services). The variable contains
1440 the device and inode number of the file descriptor used for
1441 stdout/stderr. This may be used by invoked programs to detect whether
1442 their stdout/stderr is connected to the Journal, in which case they
1443 can switch over to direct Journal communication, thus being able to
1444 pass extended, structured metadata along with their log messages. As
1445 one example, this is now used by glib's logging primitives.
1447 * When using systemd's default tmp.mount unit for /tmp, the mount point
1448 will now be established with the "nosuid" and "nodev" options. This
1449 avoids privilege escalation attacks that put traps and exploits into
1450 /tmp. However, this might cause problems if you e. g. put container
1451 images or overlays into /tmp; if you need this, override tmp.mount's
1452 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1455 * systemd now supports the "memory" cgroup controller also on
1458 * The systemd-cgtop tool now optionally takes a control group path as
1459 command line argument. If specified, the control group list shown is
1460 limited to subgroups of that group.
1462 * The SystemCallFilter= unit file setting gained support for
1463 pre-defined, named system call filter sets. For example
1464 SystemCallFilter=@clock is now an effective way to make all clock
1465 changing-related system calls unavailable to a service. A number of
1466 similar pre-defined groups are defined. Writing system call filters
1467 for system services is simplified substantially with this new
1468 concept. Accordingly, all of systemd's own, long-running services now
1469 enable system call filtering based on this, by default.
1471 * A new service setting MemoryDenyWriteExecute= has been added, taking
1472 a boolean value. If turned on, a service may no longer create memory
1473 mappings that are writable and executable at the same time. This
1474 enhances security for services where this is enabled as it becomes
1475 harder to dynamically write and then execute memory in exploited
1476 service processes. This option has been enabled for all of systemd's
1477 own long-running services.
1479 * A new RestrictRealtime= service setting has been added, taking a
1480 boolean argument. If set the service's processes may no longer
1481 acquire realtime scheduling. This improves security as realtime
1482 scheduling may otherwise be used to easily freeze the system.
1484 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1485 value. This may be used for requesting that the system manager inside
1486 of the container reports start-up completion to nspawn which then
1487 propagates this notification further to the service manager
1488 supervising nspawn itself. A related option NotifyReady= in .nspawn
1489 files has been added too. This functionality allows ordering of the
1490 start-up of multiple containers using the usual systemd ordering
1493 * machinectl gained a new command "stop" that is an alias for
1496 * systemd-resolved gained support for contacting DNS servers on
1497 link-local IPv6 addresses.
1499 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1500 its caches. A method call for requesting the same operation has been
1501 added to the bus API too, and is made available via "systemd-resolve
1504 * systemd-resolve gained a new --status switch. If passed a brief
1505 summary of the used DNS configuration with per-interface information
1508 * resolved.conf gained a new Cache= boolean option, defaulting to
1509 on. If turned off local DNS caching is disabled. This comes with a
1510 performance penalty in particular when DNSSEC is enabled. Note that
1511 resolved disables its internal caching implicitly anyway, when the
1512 configured DNS server is on a host-local IP address such as ::1 or
1513 127.0.0.1, thus automatically avoiding double local caching.
1515 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1516 for DNS requests. This improves compatibility with local programs
1517 that do not use the libc NSS or systemd-resolved's bus APIs for name
1518 resolution. This minimal DNS service is only available to local
1519 programs and does not implement the full DNS protocol, but enough to
1520 cover local DNS clients. A new, static resolv.conf file, listing just
1521 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1522 now recommended to make /etc/resolv.conf a symlink to this file in
1523 order to route all DNS lookups to systemd-resolved, regardless if
1524 done via NSS, the bus API or raw DNS packets. Note that this local
1525 DNS service is not as fully featured as the libc NSS or
1526 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1527 used to deliver link-local address information (as this implies
1528 sending a local interface index along), LLMNR/mDNS support via this
1529 interface is severely restricted. It is thus strongly recommended for
1530 all applications to use the libc NSS API or native systemd-resolved
1533 * systemd-networkd's bridge support learned a new setting
1534 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1535 in .network files has been added for configuring VLAN bridging in
1536 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1538 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1539 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1540 now be acquired without relying on DHCPv6. Two new options
1541 UseDomains= and UseDNS= have been added to configure this behaviour.
1543 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1544 renamed IPv6AcceptRA=, without altering its behaviour. The old
1545 setting name remains available for compatibility reasons.
1547 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1548 Key=, InputKey= and OutputKey=.
1550 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1551 interface configuration.
1553 * "systemctl edit" may now be used to create new unit files by
1554 specifying the --force switch.
1556 * sd-event gained a new function sd_event_get_iteration() for
1557 requesting the current iteration counter of the event loop. It starts
1558 at zero and is increased by one with each event loop iteration.
1560 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1561 file. It can be used in lieu of %systemd_requires in packages which
1562 don't use any systemd functionality and are intended to be installed
1563 in minimal containers without systemd present. This macro provides
1564 ordering dependencies to ensure that if the package is installed in
1565 the same rpm transaction as systemd, systemd will be installed before
1566 the scriptlets for the package are executed, allowing unit presets
1569 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1570 been added to simplify packaging of generators.
1572 * The os-release file gained VERSION_CODENAME field for the
1573 distribution nickname (e.g. VERSION_CODENAME=woody).
1575 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1576 can be set to disable parsing of metadata and the creation
1577 of persistent symlinks for that device.
1579 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1580 to make them available to logged-in users has been reverted.
1582 * Much of the common code of the various systemd components is now
1583 built into an internal shared library libsystemd-shared-231.so
1584 (incorporating the systemd version number in the name, to be updated
1585 with future releases) that the components link to. This should
1586 decrease systemd footprint both in memory during runtime and on
1587 disk. Note that the shared library is not for public use, and is
1588 neither API not ABI stable, but is likely to change with every new
1589 released update. Packagers need to make sure that binaries
1590 linking to libsystemd-shared.so are updated in step with the
1593 * Configuration for "mkosi" is now part of the systemd
1594 repository. mkosi is a tool to easily build legacy-free OS images,
1595 and is available on github: https://github.com/systemd/mkosi. If
1596 "mkosi" is invoked in the build tree a new raw OS image is generated
1597 incorporating the systemd sources currently being worked on and a
1598 clean, fresh distribution installation. The generated OS image may be
1599 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1600 UEFI PC. This functionality is particularly useful to easily test
1601 local changes made to systemd in a pristine, defined environment. See
1602 HACKING for details.
1604 * configure learned the --with-support-url= option to specify the
1605 distribution's bugtracker.
1607 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1608 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1609 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1610 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1611 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1612 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1613 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1614 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1615 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1616 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1617 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1618 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1619 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1620 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1621 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1622 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1623 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1624 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1625 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1627 — Berlin, 2016-07-25
1631 * DNSSEC is now turned on by default in systemd-resolved (in
1632 "allow-downgrade" mode), but may be turned off during compile time by
1633 passing "--with-default-dnssec=no" to "configure" (and of course,
1634 during runtime with DNSSEC= in resolved.conf). We recommend
1635 downstreams to leave this on at least during development cycles and
1636 report any issues with the DNSSEC logic upstream. We are very
1637 interested in collecting feedback about the DNSSEC validator and its
1638 limitations in the wild. Note however, that DNSSEC support is
1639 probably nothing downstreams should turn on in stable distros just
1640 yet, as it might create incompatibilities with a few DNS servers and
1641 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1642 automatically whenever we detect such incompatible setups, but there
1643 might be systems we do not cover yet. Hence: please help us testing
1644 the DNSSEC code, leave this on where you can, report back, but then
1645 again don't consider turning this on in your stable, LTS or
1646 production release just yet. (Note that you have to enable
1647 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1648 and its DNSSEC mode for host name resolution from local
1651 * systemd-resolve conveniently resolves DANE records with the --tlsa
1652 option and OPENPGPKEY records with the --openpgp option. It also
1653 supports dumping raw DNS record data via the new --raw= switch.
1655 * systemd-logind will now by default terminate user processes that are
1656 part of the user session scope unit (session-XX.scope) when the user
1657 logs out. This behavior is controlled by the KillUserProcesses=
1658 setting in logind.conf, and the previous default of "no" is now
1659 changed to "yes". This means that user sessions will be properly
1660 cleaned up after, but additional steps are necessary to allow
1661 intentionally long-running processes to survive logout.
1663 While the user is logged in at least once, user@.service is running,
1664 and any service that should survive the end of any individual login
1665 session can be started at a user service or scope using systemd-run.
1666 systemd-run(1) man page has been extended with an example which shows
1667 how to run screen in a scope unit underneath user@.service. The same
1668 command works for tmux.
1670 After the user logs out of all sessions, user@.service will be
1671 terminated too, by default, unless the user has "lingering" enabled.
1672 To effectively allow users to run long-term tasks even if they are
1673 logged out, lingering must be enabled for them. See loginctl(1) for
1674 details. The default polkit policy was modified to allow users to
1675 set lingering for themselves without authentication.
1677 Previous defaults can be restored at compile time by the
1678 --without-kill-user-processes option to "configure".
1680 * systemd-logind gained new configuration settings SessionsMax= and
1681 InhibitorsMax=, both with a default of 8192. It will not register new
1682 user sessions or inhibitors above this limit.
1684 * systemd-logind will now reload configuration on SIGHUP.
1686 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1687 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1688 enable. Also, support for the "io" cgroup controller in the unified
1689 hierarchy has been added, so that the "memory", "pids" and "io" are
1690 now the controllers that are supported on the unified hierarchy.
1692 WARNING: it is not possible to use previous systemd versions with
1693 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1694 is necessary to also update systemd in the initramfs if using the
1695 unified hierarchy. An updated SELinux policy is also required.
1697 * LLDP support has been extended, and both passive (receive-only) and
1698 active (sender) modes are supported. Passive mode ("routers-only") is
1699 enabled by default in systemd-networkd. Active LLDP mode is enabled
1700 by default for containers on the internal network. The "networkctl
1701 lldp" command may be used to list information gathered. "networkctl
1702 status" will also show basic LLDP information on connected peers now.
1704 * The IAID and DUID unique identifier sent in DHCP requests may now be
1705 configured for the system and each .network file managed by
1706 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1708 * systemd-networkd gained support for configuring proxy ARP support for
1709 each interface, via the ProxyArp= setting in .network files. It also
1710 gained support for configuring the multicast querier feature of
1711 bridge devices, via the new MulticastQuerier= setting in .netdev
1712 files. Similarly, snooping on the IGMP traffic can be controlled
1713 via the new setting MulticastSnooping=.
1715 A new setting PreferredLifetime= has been added for addresses
1716 configured in .network file to configure the lifetime intended for an
1719 The systemd-networkd DHCP server gained the option EmitRouter=, which
1720 defaults to yes, to configure whether the DHCP Option 3 (Router)
1723 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1724 systemd-socket-activate and installed into /usr/bin. It is now fully
1727 * systemd-journald now uses separate threads to flush changes to disk
1728 when closing journal files, thus reducing impact of slow disk I/O on
1729 logging performance.
1731 * The sd-journal API gained two new calls
1732 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1733 can be used to open journal files using file descriptors instead of
1734 file or directory paths. sd_journal_open_container() has been
1735 deprecated, sd_journal_open_directory_fd() should be used instead
1736 with the flag SD_JOURNAL_OS_ROOT.
1738 * journalctl learned a new output mode "-o short-unix" that outputs log
1739 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1740 UTC). It also gained support for a new --no-hostname setting to
1741 suppress the hostname column in the family of "short" output modes.
1743 * systemd-ask-password now optionally skips printing of the password to
1744 stdout with --no-output which can be useful in scripts.
1746 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1747 (devices tagged with ID_MAKER_TOOL) are now tagged with
1748 "uaccess" and are available to logged in users.
1750 * The DeviceAllow= unit setting now supports specifiers (with "%").
1752 * "systemctl show" gained a new --value switch, which allows print a
1753 only the contents of a specific unit property, without also printing
1754 the property's name. Similar support was added to "show*" verbs
1755 of loginctl and machinectl that output "key=value" lists.
1757 * A new unit type "generated" was added for files dynamically generated
1758 by generator tools. Similarly, a new unit type "transient" is used
1759 for unit files created using the runtime API. "systemctl enable" will
1760 refuse to operate on such files.
1762 * A new command "systemctl revert" has been added that may be used to
1763 revert to the vendor version of a unit file, in case local changes
1764 have been made by adding drop-ins or overriding the unit file.
1766 * "machinectl clean" gained a new verb to automatically remove all or
1767 just hidden container images.
1769 * systemd-tmpfiles gained support for a new line type "e" for emptying
1770 directories, if they exist, without creating them if they don't.
1772 * systemd-nspawn gained support for automatically patching the UID/GIDs
1773 of the owners and the ACLs of all files and directories in a
1774 container tree to match the UID/GID user namespacing range selected
1775 for the container invocation. This mode is enabled via the new
1776 --private-users-chown switch. It also gained support for
1777 automatically choosing a free, previously unused UID/GID range when
1778 starting a container, via the new --private-users=pick setting (which
1779 implies --private-users-chown). Together, these options for the first
1780 time make user namespacing for nspawn containers fully automatic and
1781 thus deployable. The systemd-nspawn@.service template unit file has
1782 been changed to use this functionality by default.
1784 * systemd-nspawn gained a new --network-zone= switch, that allows
1785 creating ad-hoc virtual Ethernet links between multiple containers,
1786 that only exist as long as at least one container referencing them is
1787 running. This allows easy connecting of multiple containers with a
1788 common link that implements an Ethernet broadcast domain. Each of
1789 these network "zones" may be named relatively freely by the user, and
1790 may be referenced by any number of containers, but each container may
1791 only reference one of these "zones". On the lower level, this is
1792 implemented by an automatically managed bridge network interface for
1793 each zone, that is created when the first container referencing its
1794 zone is created and removed when the last one referencing its zone
1797 * The default start timeout may now be configured on the kernel command
1798 line via systemd.default_timeout_start_sec=. It was already
1799 configurable via the DefaultTimeoutStartSec= option in
1800 /etc/systemd/system.conf.
1802 * Socket units gained a new TriggerLimitIntervalSec= and
1803 TriggerLimitBurst= setting to configure a limit on the activation
1804 rate of the socket unit.
1806 * The LimitNICE= setting now optionally takes normal UNIX nice values
1807 in addition to the raw integer limit value. If the specified
1808 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1809 value is understood as UNIX nice value. If not prefixed like this it
1810 is understood as raw RLIMIT_NICE limit.
1812 * Note that the effect of the PrivateDevices= unit file setting changed
1813 slightly with this release: the per-device /dev file system will be
1814 mounted read-only from this version on, and will have "noexec"
1815 set. This (minor) change of behavior might cause some (exceptional)
1816 legacy software to break, when PrivateDevices=yes is set for its
1817 service. Please leave PrivateDevices= off if you run into problems
1820 * systemd-bootchart has been split out to a separate repository:
1821 https://github.com/systemd/systemd-bootchart
1823 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1824 merged into the kernel in its current form.
1826 * The compatibility libraries libsystemd-daemon.so,
1827 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1828 which have been deprecated since systemd-209 have been removed along
1829 with the corresponding pkg-config files. All symbols provided by
1830 those libraries are provided by libsystemd.so.
1832 * The Capabilities= unit file setting has been removed (it is ignored
1833 for backwards compatibility). AmbientCapabilities= and
1834 CapabilityBoundingSet= should be used instead.
1836 * A new special target has been added, initrd-root-device.target,
1837 which creates a synchronization point for dependencies of the root
1838 device in early userspace. Initramfs builders must ensure that this
1839 target is now included in early userspace.
1841 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1842 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1843 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1844 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1845 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1846 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1847 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1848 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1849 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1850 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1851 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1852 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1853 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1854 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1855 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1856 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1857 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1858 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1859 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1860 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1861 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1862 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1863 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1864 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1865 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1868 — Fairfax, 2016-05-21
1872 * The systemd-resolved DNS resolver service has gained a substantial
1873 set of new features, most prominently it may now act as a DNSSEC
1874 validating stub resolver. DNSSEC mode is currently turned off by
1875 default, but is expected to be turned on by default in one of the
1876 next releases. For now, we invite everybody to test the DNSSEC logic
1877 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1878 service also gained a full set of D-Bus interfaces, including calls
1879 to configure DNS and DNSSEC settings per link (for use by external
1880 network management software). systemd-resolved and systemd-networkd
1881 now distinguish between "search" and "routing" domains. The former
1882 are used to qualify single-label names, the latter are used purely
1883 for routing lookups within certain domains to specific links.
1884 resolved now also synthesizes RRs for all entries from /etc/hosts.
1886 * The systemd-resolve tool (which is a client utility for
1887 systemd-resolved) has been improved considerably and is now fully
1888 supported and documented. Hence it has moved from /usr/lib/systemd to
1891 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1894 * The coredump collection logic has been reworked: when a coredump is
1895 collected it is now written to disk, compressed and processed
1896 (including stacktrace extraction) from a new instantiated service
1897 systemd-coredump@.service, instead of directly from the
1898 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1899 processing large coredumps can take up a substantial amount of
1900 resources and time, and this previously happened entirely outside of
1901 systemd's service supervision. With the new logic the core_pattern
1902 hook only does minimal metadata collection before passing off control
1903 to the new instantiated service, which is configured with a time
1904 limit, a nice level and other settings to minimize negative impact on
1905 the rest of the system. Also note that the new logic will honour the
1906 RLIMIT_CORE setting of the crashed process, which now allows users
1907 and processes to turn off coredumping for their processes by setting
1910 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1911 and all forked processes by default. Previously, PID 1 would leave
1912 the setting at "0" for all processes, as set by the kernel. Note that
1913 the resource limit traditionally has no effect on the generated
1914 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1915 logic is used. Since the limit is now honoured (see above) its
1916 default has been changed so that the coredumping logic is enabled by
1917 default for all processes, while allowing specific opt-out.
1919 * When the stacktrace is extracted from processes of system users, this
1920 is now done as "systemd-coredump" user, in order to sandbox this
1921 potentially security sensitive parsing operation. (Note that when
1922 processing coredumps of normal users this is done under the user ID
1923 of process that crashed, as before.) Packagers should take notice
1924 that it is now necessary to create the "systemd-coredump" system user
1925 and group at package installation time.
1927 * The systemd-activate socket activation testing tool gained support
1928 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1929 and --seqpacket switches. It also has been extended to support both
1930 new-style and inetd-style file descriptor passing. Use the new
1931 --inetd switch to request inetd-style file descriptor passing.
1933 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1934 variable, which takes a boolean value. If set to false, ANSI color
1935 output is disabled in the tools even when run on a terminal that
1938 * The VXLAN support in networkd now supports two new settings
1939 DestinationPort= and PortRange=.
1941 * A new systemd.machine_id= kernel command line switch has been added,
1942 that may be used to set the machine ID in /etc/machine-id if it is
1943 not initialized yet. This command line option has no effect if the
1944 file is already initialized.
1946 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1947 specified command line as PID 2 rather than PID 1 in the
1948 container. In this mode PID 1 is a minimal stub init process that
1949 implements the special POSIX and Linux semantics of PID 1 regarding
1950 signal and child process management. Note that this stub init process
1951 is implemented in nspawn itself and requires no support from the
1952 container image. This new logic is useful to support running
1953 arbitrary commands in the container, as normal processes are
1954 generally not prepared to run as PID 1.
1956 * systemd-nspawn gained a new --chdir= switch for setting the current
1957 working directory for the process started in the container.
1959 * "journalctl /dev/sda" will now output all kernel log messages for
1960 specified device from the current boot, in addition to all devices
1961 that are parents of it. This should make log output about devices
1962 pretty useful, as long as kernel drivers attach enough metadata to
1963 the log messages. (The usual SATA drivers do.)
1965 * The sd-journal API gained two new calls
1966 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1967 that report whether log data from /run or /var has been found.
1969 * journalctl gained a new switch "--fields" that prints all journal
1970 record field names currently in use in the journal. This is backed
1971 by two new sd-journal API calls sd_journal_enumerate_fields() and
1972 sd_journal_restart_fields().
1974 * Most configurable timeouts in systemd now expect an argument of
1975 "infinity" to turn them off, instead of "0" as before. The semantics
1976 from now on is that a timeout of "0" means "now", and "infinity"
1977 means "never". To maintain backwards compatibility, "0" continues to
1978 turn off previously existing timeout settings.
1980 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1981 try-reload-or-restart" to clarify what it actually does: the "try"
1982 logic applies to both reloading and restarting, not just restarting.
1983 The old name continues to be accepted for compatibility.
1985 * On boot-up, when PID 1 detects that the system clock is behind the
1986 release date of the systemd version in use, the clock is now set
1987 to the latter. Previously, this was already done in timesyncd, in order
1988 to avoid running with clocks set to the various clock epochs such as
1989 1902, 1938 or 1970. With this change the logic is now done in PID 1
1990 in addition to timesyncd during early boot-up, so that it is enforced
1991 before the first process is spawned by systemd. Note that the logic
1992 in timesyncd remains, as it is more comprehensive and ensures
1993 clock monotonicity by maintaining a persistent timestamp file in
1994 /var. Since /var is generally not available in earliest boot or the
1995 initrd, this part of the logic remains in timesyncd, and is not done
1998 * Support for tweaking details in net_cls.class_id through the
1999 NetClass= configuration directive has been removed, as the kernel
2000 people have decided to deprecate that controller in cgroup v2.
2001 Userspace tools such as nftables are moving over to setting rules
2002 that are specific to the full cgroup path of a task, which obsoletes
2003 these controllers anyway. The NetClass= directive is kept around for
2004 legacy compatibility reasons. For a more in-depth description of the
2005 kernel change, please refer to the respective upstream commit:
2007 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2009 * A new service setting RuntimeMaxSec= has been added that may be used
2010 to specify a maximum runtime for a service. If the timeout is hit, the
2011 service is terminated and put into a failure state.
2013 * A new service setting AmbientCapabilities= has been added. It allows
2014 configuration of additional Linux process capabilities that are
2015 passed to the activated processes. This is only available on very
2018 * The process resource limit settings in service units may now be used
2019 to configure hard and soft limits individually.
2021 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2022 expose support for gcc's __attribute__((cleanup())) C extension.
2023 Specifically, for many object destructor functions alternative
2024 versions have been added that have names suffixed with "p" and take a
2025 pointer to a pointer to the object to destroy, instead of just a
2026 pointer to the object itself. This is useful because these destructor
2027 functions may be used directly as parameters to the cleanup
2028 construct. Internally, systemd has been a heavy user of this GCC
2029 extension for a long time, and with this change similar support is
2030 now available to consumers of the library outside of systemd. Note
2031 that by using this extension in your sources compatibility with old
2032 and strictly ANSI compatible C compilers is lost. However, all gcc or
2033 LLVM versions of recent years support this extension.
2035 * Timer units gained support for a new setting RandomizedDelaySec= that
2036 allows configuring some additional randomized delay to the configured
2037 time. This is useful to spread out timer events to avoid load peaks in
2038 clusters or larger setups.
2040 * Calendar time specifications now support sub-second accuracy.
2042 * Socket units now support listening on SCTP and UDP-lite protocol
2045 * The sd-event API now comes with a full set of man pages.
2047 * Older versions of systemd contained experimental support for
2048 compressing journal files and coredumps with the LZ4 compressor that
2049 was not compatible with the lz4 binary (due to API limitations of the
2050 lz4 library). This support has been removed; only support for files
2051 compatible with the lz4 binary remains. This LZ4 logic is now
2052 officially supported and no longer considered experimental.
2054 * The dkr image import logic has been removed again from importd. dkr's
2055 micro-services focus doesn't fit into the machine image focus of
2056 importd, and quickly got out of date with the upstream dkr API.
2058 * Creation of the /run/lock/lockdev/ directory was dropped from
2059 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2060 been available for many years. If you still need this, you need to
2061 create your own tmpfiles.d config file with:
2063 d /run/lock/lockdev 0775 root lock -
2065 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2066 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2067 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2068 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2069 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2070 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2071 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2072 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2073 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2074 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2075 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2076 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2077 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2078 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2079 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2080 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2081 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2082 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2083 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2085 — Berlin, 2016-02-11
2089 * A number of properties previously only settable in unit
2090 files are now also available as properties to set when
2091 creating transient units programmatically via the bus, as it
2092 is exposed with systemd-run's --property=
2093 setting. Specifically, these are: SyslogIdentifier=,
2094 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2095 EnvironmentFile=, ReadWriteDirectories=,
2096 ReadOnlyDirectories=, InaccessibleDirectories=,
2097 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2099 * When creating transient services via the bus API it is now
2100 possible to pass in a set of file descriptors to use as
2101 STDIN/STDOUT/STDERR for the invoked process.
2103 * Slice units may now be created transiently via the bus APIs,
2104 similar to the way service and scope units may already be
2105 created transiently.
2107 * Wherever systemd expects a calendar timestamp specification
2108 (like in journalctl's --since= and --until= switches) UTC
2109 timestamps are now supported. Timestamps suffixed with "UTC"
2110 are now considered to be in Universal Time Coordinated
2111 instead of the local timezone. Also, timestamps may now
2112 optionally be specified with sub-second accuracy. Both of
2113 these additions also apply to recurring calendar event
2114 specification, such as OnCalendar= in timer units.
2116 * journalctl gained a new "--sync" switch that asks the
2117 journal daemon to write all so far unwritten log messages to
2118 disk and sync the files, before returning.
2120 * systemd-tmpfiles learned two new line types "q" and "Q" that
2121 operate like "v", but also set up a basic btrfs quota
2122 hierarchy when used on a btrfs file system with quota
2125 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2126 instead of a subvolume (even on a btrfs file system) if the
2127 root directory is a plain directory, and not a
2128 subvolume. This should simplify things with certain chroot()
2129 environments which are not aware of the concept of btrfs
2132 * systemd-detect-virt gained a new --chroot switch to detect
2133 whether execution takes place in a chroot() environment.
2135 * CPUAffinity= now takes CPU index ranges in addition to
2138 * The various memory-related resource limit settings (such as
2139 LimitAS=) now understand the usual K, M, G, ... suffixes to
2140 the base of 1024 (IEC). Similar, the time-related resource
2141 limit settings understand the usual min, h, day, ...
2144 * There's a new system.conf setting DefaultTasksMax= to
2145 control the default TasksMax= setting for services and
2146 scopes running on the system. (TasksMax= is the primary
2147 setting that exposes the "pids" cgroup controller on systemd
2148 and was introduced in the previous systemd release.) The
2149 setting now defaults to 512, which means services that are
2150 not explicitly configured otherwise will only be able to
2151 create 512 processes or threads at maximum, from this
2152 version on. Note that this means that thread- or
2153 process-heavy services might need to be reconfigured to set
2154 TasksMax= to a higher value. It is sufficient to set
2155 TasksMax= in these specific unit files to a higher value, or
2156 even "infinity". Similar, there's now a logind.conf setting
2157 UserTasksMax= that defaults to 4096 and limits the total
2158 number of processes or tasks each user may own
2159 concurrently. nspawn containers also have the TasksMax=
2160 value set by default now, to 8192. Note that all of this
2161 only has an effect if the "pids" cgroup controller is
2162 enabled in the kernel. The general benefit of these changes
2163 should be a more robust and safer system, that provides a
2164 certain amount of per-service fork() bomb protection.
2166 * systemd-nspawn gained the new --network-veth-extra= switch
2167 to define additional and arbitrarily-named virtual Ethernet
2168 links between the host and the container.
2170 * A new service execution setting PassEnvironment= has been
2171 added that allows importing select environment variables
2172 from PID1's environment block into the environment block of
2175 * Timer units gained support for a new RemainAfterElapse=
2176 setting which takes a boolean argument. It defaults to on,
2177 exposing behaviour unchanged to previous releases. If set to
2178 off, timer units are unloaded after they elapsed if they
2179 cannot elapse again. This is particularly useful for
2180 transient timer units, which shall not stay around longer
2181 than until they first elapse.
2183 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2184 default now (the kernel default is 16). This is beneficial
2185 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2186 allows substantially larger numbers of queued
2187 datagrams. This should increase the capability of systemd to
2188 parallelize boot-up, as logging and sd_notify() are unlikely
2189 to stall execution anymore. If you need to change the value
2190 from the new defaults, use the usual sysctl.d/ snippets.
2192 * The compression framing format used by the journal or
2193 coredump processing has changed to be in line with what the
2194 official LZ4 tools generate. LZ4 compression support in
2195 systemd was considered unsupported previously, as the format
2196 was not compatible with the normal tools. With this release
2197 this has changed now, and it is hence safe for downstream
2198 distributions to turn it on. While not compressing as well
2199 as the XZ, LZ4 is substantially faster, which makes
2200 it a good default choice for the compression logic in the
2201 journal and in coredump handling.
2203 * Any reference to /etc/mtab has been dropped from
2204 systemd. The file has been obsolete since a while, but
2205 systemd refused to work on systems where it was incorrectly
2206 set up (it should be a symlink or non-existent). Please make
2207 sure to update to util-linux 2.27.1 or newer in conjunction
2208 with this systemd release, which also drops any reference to
2209 /etc/mtab. If you maintain a distribution make sure that no
2210 software you package still references it, as this is a
2211 likely source of bugs. There's also a glibc bug pending,
2212 asking for removal of any reference to this obsolete file:
2214 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2216 Note that only util-linux versions built with
2217 --enable-libmount-force-mountinfo are supported.
2219 * Support for the ".snapshot" unit type has been removed. This
2220 feature turned out to be little useful and little used, and
2221 has now been removed from the core and from systemctl.
2223 * The dependency types RequiresOverridable= and
2224 RequisiteOverridable= have been removed from systemd. They
2225 have been used only very sparingly to our knowledge and
2226 other options that provide a similar effect (such as
2227 systemctl --mode=ignore-dependencies) are much more useful
2228 and commonly used. Moreover, they were only half-way
2229 implemented as the option to control behaviour regarding
2230 these dependencies was never added to systemctl. By removing
2231 these dependency types the execution engine becomes a bit
2232 simpler. Unit files that use these dependencies should be
2233 changed to use the non-Overridable dependency types
2234 instead. In fact, when parsing unit files with these
2235 options, that's what systemd will automatically convert them
2236 too, but it will also warn, asking users to fix the unit
2237 files accordingly. Removal of these dependency types should
2238 only affect a negligible number of unit files in the wild.
2240 * Behaviour of networkd's IPForward= option changed
2241 (again). It will no longer maintain a per-interface setting,
2242 but propagate one way from interfaces where this is enabled
2243 to the global kernel setting. The global setting will be
2244 enabled when requested by a network that is set up, but
2245 never be disabled again. This change was made to make sure
2246 IPv4 and IPv6 behaviour regarding packet forwarding is
2247 similar (as the Linux IPv6 stack does not support
2248 per-interface control of this setting) and to minimize
2251 * In unit files the behaviour of %u, %U, %h, %s has
2252 changed. These specifiers will now unconditionally resolve
2253 to the various user database fields of the user that the
2254 systemd instance is running as, instead of the user
2255 configured in the specific unit via User=. Note that this
2256 effectively doesn't change much, as resolving of these
2257 specifiers was already turned off in the --system instance
2258 of systemd, as we cannot do NSS lookups from PID 1. In the
2259 --user instance of systemd these specifiers where correctly
2260 resolved, but hardly made any sense, since the user instance
2261 lacks privileges to do user switches anyway, and User= is
2262 hence useless. Moreover, even in the --user instance of
2263 systemd behaviour was awkward as it would only take settings
2264 from User= assignment placed before the specifier into
2265 account. In order to unify and simplify the logic around
2266 this the specifiers will now always resolve to the
2267 credentials of the user invoking the manager (which in case
2268 of PID 1 is the root user).
2270 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2271 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2272 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2273 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2274 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2275 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2276 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2277 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2278 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2279 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2280 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2281 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2282 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2283 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2286 — Berlin, 2015-11-18
2290 * systemd now depends on util-linux v2.27. More specifically,
2291 the newly added mount monitor feature in libmount now
2292 replaces systemd's former own implementation.
2294 * libmount mandates /etc/mtab not to be regular file, and
2295 systemd now enforces this condition at early boot.
2296 /etc/mtab has been deprecated and warned about for a very
2297 long time, so systems running systemd should already have
2298 stopped having this file around as anything else than a
2299 symlink to /proc/self/mounts.
2301 * Support for the "pids" cgroup controller has been added. It
2302 allows accounting the number of tasks in a cgroup and
2303 enforcing limits on it. This adds two new setting
2304 TasksAccounting= and TasksMax= to each unit, as well as a
2305 global option DefaultTasksAccounting=.
2307 * Support for the "net_cls" cgroup controller has been added.
2308 It allows assigning a net class ID to each task in the
2309 cgroup, which can then be used in firewall rules and traffic
2310 shaping configurations. Note that the kernel netfilter net
2311 class code does not currently work reliably for ingress
2312 packets on unestablished sockets.
2314 This adds a new config directive called NetClass= to CGroup
2315 enabled units. Allowed values are positive numbers for fixed
2316 assignments and "auto" for picking a free value
2319 * 'systemctl is-system-running' now returns 'offline' if the
2320 system is not booted with systemd. This command can now be
2321 used as a substitute for 'systemd-notify --booted'.
2323 * Watchdog timeouts have been increased to 3 minutes for all
2324 in-tree service files. Apparently, disk IO issues are more
2325 frequent than we hoped, and user reported >1 minute waiting
2328 * 'machine-id-commit' functionality has been merged into
2329 'machine-id-setup --commit'. The separate binary has been
2332 * The WorkingDirectory= directive in unit files may now be set
2333 to the special value '~'. In this case, the working
2334 directory is set to the home directory of the user
2335 configured in User=.
2337 * "machinectl shell" will now open the shell in the home
2338 directory of the selected user by default.
2340 * The CrashChVT= configuration file setting is renamed to
2341 CrashChangeVT=, following our usual logic of not
2342 abbreviating unnecessarily. The old directive is still
2343 supported for compat reasons. Also, this directive now takes
2344 an integer value between 1 and 63, or a boolean value. The
2345 formerly supported '-1' value for disabling stays around for
2348 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2349 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2350 RootDirectory= properties can now be set for transient
2353 * The systemd-analyze tool gained a new "set-log-target" verb
2354 to change the logging target the system manager logs to
2355 dynamically during runtime. This is similar to how
2356 "systemd-analyze set-log-level" already changes the log
2359 * In nspawn /sys is now mounted as tmpfs, with only a selected
2360 set of subdirectories mounted in from the real sysfs. This
2361 enhances security slightly, and is useful for ensuring user
2362 namespaces work correctly.
2364 * Support for USB FunctionFS activation has been added. This
2365 allows implementation of USB gadget services that are
2366 activated as soon as they are requested, so that they don't
2367 have to run continuously, similar to classic socket
2370 * The "systemctl exit" command now optionally takes an
2371 additional parameter that sets the exit code to return from
2372 the systemd manager when exiting. This is only relevant when
2373 running the systemd user instance, or when running the
2374 system instance in a container.
2376 * sd-bus gained the new API calls sd_bus_path_encode_many()
2377 and sd_bus_path_decode_many() that allow easy encoding and
2378 decoding of multiple identifier strings inside a D-Bus
2379 object path. Another new call sd_bus_default_flush_close()
2380 has been added to flush and close per-thread default
2383 * systemd-cgtop gained support for a -M/--machine= switch to
2384 show the control groups within a certain container only.
2386 * "systemctl kill" gained support for an optional --fail
2387 switch. If specified the requested operation will fail of no
2388 processes have been killed, because the unit had no
2389 processes attached, or similar.
2391 * A new systemd.crash_reboot=1 kernel command line option has
2392 been added that triggers a reboot after crashing. This can
2393 also be set through CrashReboot= in systemd.conf.
2395 * The RuntimeDirectory= setting now understands unit
2396 specifiers like %i or %f.
2398 * A new (still internal) library API sd-ipv4acd has been added,
2399 that implements address conflict detection for IPv4. It's
2400 based on code from sd-ipv4ll, and will be useful for
2401 detecting DHCP address conflicts.
2403 * File descriptors passed during socket activation may now be
2404 named. A new API sd_listen_fds_with_names() is added to
2405 access the names. The default names may be overridden,
2406 either in the .socket file using the FileDescriptorName=
2407 parameter, or by passing FDNAME= when storing the file
2408 descriptors using sd_notify().
2410 * systemd-networkd gained support for:
2412 - Setting the IPv6 Router Advertisement settings via
2413 IPv6AcceptRouterAdvertisements= in .network files.
2415 - Configuring the HelloTimeSec=, MaxAgeSec= and
2416 ForwardDelaySec= bridge parameters in .netdev files.
2418 - Configuring PreferredSource= for static routes in
2421 * The "ask-password" framework used to query for LUKS harddisk
2422 passwords or SSL passwords during boot gained support for
2423 caching passwords in the kernel keyring, if it is
2424 available. This makes sure that the user only has to type in
2425 a passphrase once if there are multiple objects to unlock
2426 with the same one. Previously, such password caching was
2427 available only when Plymouth was used; this moves the
2428 caching logic into the systemd codebase itself. The
2429 "systemd-ask-password" utility gained a new --keyname=
2430 switch to control which kernel keyring key to use for
2431 caching a password in. This functionality is also useful for
2432 enabling display managers such as gdm to automatically
2433 unlock the user's GNOME keyring if its passphrase, the
2434 user's password and the harddisk password are the same, if
2435 gdm-autologin is used.
2437 * When downloading tar or raw images using "machinectl
2438 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2439 file is now also downloaded, if it is available and stored
2440 next to the image file.
2442 * Units of type ".socket" gained a new boolean setting
2443 Writable= which is only useful in conjunction with
2444 ListenSpecial=. If true, enables opening the specified
2445 special file in O_RDWR mode rather than O_RDONLY mode.
2447 * systemd-rfkill has been reworked to become a singleton
2448 service that is activated through /dev/rfkill on each rfkill
2449 state change and saves the settings to disk. This way,
2450 systemd-rfkill is now compatible with devices that exist
2451 only intermittendly, and even restores state if the previous
2452 system shutdown was abrupt rather than clean.
2454 * The journal daemon gained support for vacuuming old journal
2455 files controlled by the number of files that shall remain,
2456 in addition to the already existing control by size and by
2457 date. This is useful as journal interleaving performance
2458 degrades with too many separate journal files, and allows
2459 putting an effective limit on them. The new setting defaults
2460 to 100, but this may be changed by setting SystemMaxFiles=
2461 and RuntimeMaxFiles= in journald.conf. Also, the
2462 "journalctl" tool gained the new --vacuum-files= switch to
2463 manually vacuum journal files to leave only the specified
2464 number of files in place.
2466 * udev will now create /dev/disk/by-path links for ATA devices
2467 on kernels where that is supported.
2469 * Galician, Serbian, Turkish and Korean translations were added.
2471 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2472 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2473 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2474 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2475 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2476 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2477 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2478 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2479 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2480 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2481 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2482 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2483 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2484 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2485 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2486 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2487 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2488 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2490 — Berlin, 2015-10-07
2494 * The DHCP implementation of systemd-networkd gained a set of
2497 - The DHCP server now supports emitting DNS and NTP
2498 information. It may be enabled and configured via
2499 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2500 and NTP information is enabled, but no servers are
2501 configured, the corresponding uplink information (if there
2502 is any) is propagated.
2504 - Server and client now support transmission and reception
2505 of timezone information. It can be configured via the
2506 newly introduced network options UseTimezone=,
2507 EmitTimezone=, and Timezone=. Transmission of timezone
2508 information is enabled between host and containers by
2509 default now: the container will change its local timezone
2510 to what the host has set.
2512 - Lease timeouts can now be configured via
2513 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2515 - The DHCP server improved on the stability of
2516 leases. Clients are more likely to get the same lease
2517 information back, even if the server loses state.
2519 - The DHCP server supports two new configuration options to
2520 control the lease address pool metrics, PoolOffset= and
2523 * The encapsulation limit of tunnels in systemd-networkd may
2524 now be configured via 'EncapsulationLimit='. It allows
2525 modifying the maximum additional levels of encapsulation
2526 that are permitted to be prepended to a packet.
2528 * systemd now supports the concept of user buses replacing
2529 session buses, if used with dbus-1.10 (and enabled via dbus
2530 --enable-user-session). It previously only supported this on
2531 kdbus-enabled systems, and this release expands this to
2532 'dbus-daemon' systems.
2534 * systemd-networkd now supports predictable interface names
2537 * systemd now optionally supports the new Linux kernel
2538 "unified" control group hierarchy. If enabled via the kernel
2539 command-line option 'systemd.unified_cgroup_hierarchy=1',
2540 systemd will try to mount the unified cgroup hierarchy
2541 directly on /sys/fs/cgroup. If not enabled, or not
2542 available, systemd will fall back to the legacy cgroup
2543 hierarchy setup, as before. Host system and containers can
2544 mix and match legacy and unified hierarchies as they
2545 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2546 environment variable to individually select the hierarchy to
2547 use for executed containers. By default, nspawn will use the
2548 unified hierarchy for the containers if the host uses the
2549 unified hierarchy, and the legacy hierarchy otherwise.
2550 Please note that at this point the unified hierarchy is an
2551 experimental kernel feature and is likely to change in one
2552 of the next kernel releases. Therefore, it should not be
2553 enabled by default in downstream distributions yet. The
2554 minimum required kernel version for the unified hierarchy to
2555 work is 4.2. Note that when the unified hierarchy is used
2556 for the first time delegated access to controllers is
2557 safe. Because of this systemd-nspawn containers will get
2558 access to controllers now, as will systemd user
2559 sessions. This means containers and user sessions may now
2560 manage their own resources, partitioning up what the system
2563 * A new special scope unit "init.scope" has been introduced
2564 that encapsulates PID 1 of the system. It may be used to
2565 determine resource usage and enforce resource limits on PID
2566 1 itself. PID 1 hence moved out of the root of the control
2569 * The cgtop tool gained support for filtering out kernel
2570 threads when counting tasks in a control group. Also, the
2571 count of processes is now recursively summed up by
2572 default. Two options -k and --recursive= have been added to
2573 revert to old behaviour. The tool has also been updated to
2574 work correctly in containers now.
2576 * systemd-nspawn's --bind= and --bind-ro= options have been
2577 extended to allow creation of non-recursive bind mounts.
2579 * libsystemd gained two new calls sd_pid_get_cgroup() and
2580 sd_peer_get_cgroup() which return the control group path of
2581 a process or peer of a connected AF_UNIX socket. This
2582 function call is particularly useful when implementing
2583 delegated subtrees support in the control group hierarchy.
2585 * The "sd-event" event loop API of libsystemd now supports
2586 correct dequeuing of real-time signals, without losing
2589 * When systemd requests a PolicyKit decision when managing
2590 units it will now add additional fields to the request,
2591 including unit name and desired operation. This enables more
2592 powerful PolicyKit policies, that make decisions depending
2593 on these parameters.
2595 * nspawn learnt support for .nspawn settings files, that may
2596 accompany the image files or directories of containers, and
2597 may contain additional settings for the container. This is
2598 an alternative to configuring container parameters via the
2599 nspawn command line.
2601 Contributions from: Cristian Rodríguez, Daniel Mack, David
2602 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2603 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2604 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2605 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2606 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2607 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2608 Andersen, Tom Gundersen, Torstein Husebø
2610 — Berlin, 2015-09-08
2614 * machinectl gained a new verb 'shell' which opens a fresh
2615 shell on the target container or the host. It is similar to
2616 the existing 'login' command of machinectl, but spawns the
2617 shell directly without prompting for username or
2618 password. The pseudo machine '.host' now refers to the local
2619 host and is used by default. Hence, 'machinectl shell' can
2620 be used as replacement for 'su -' which spawns a session as
2621 a fresh systemd unit in a way that is fully isolated from
2622 the originating session.
2624 * systemd-networkd learned to cope with private-zone DHCP
2625 options and allows other programs to query the values.
2627 * SELinux access control when enabling/disabling units is no
2628 longer enforced with this release. The previous
2629 implementation was incorrect, and a new corrected
2630 implementation is not yet available. As unit file operations
2631 are still protected via PolicyKit and D-Bus policy this is
2632 not a security problem. Yet, distributions which care about
2633 optimal SELinux support should probably not stabilize on
2636 * sd-bus gained support for matches of type "arg0has=", that
2637 test for membership of strings in string arrays sent in bus
2640 * systemd-resolved now dumps the contents of its DNS and LLMNR
2641 caches to the logs on reception of the SIGUSR1 signal. This
2642 is useful to debug DNS behaviour.
2644 * The coredumpctl tool gained a new --directory= option to
2645 operate on journal files in a specific directory.
2647 * "systemctl reboot" and related commands gained a new
2648 "--message=" option which may be used to set a free-text
2649 wall message when shutting down or rebooting the
2650 system. This message is also logged, which is useful for
2651 figuring out the reason for a reboot or shutdown a
2654 * The "systemd-resolve-host" tool's -i switch now takes
2655 network interface numbers as alternative to interface names.
2657 * A new unit file setting for services has been introduced:
2658 UtmpMode= allows configuration of how precisely systemd
2659 handles utmp and wtmp entries for the service if this is
2660 enabled. This allows writing services that appear similar to
2661 user sessions in the output of the "w", "who", "last" and
2664 * systemd-resolved will now locally synthesize DNS resource
2665 records for the "localhost" and "gateway" domains as well as
2666 the local hostname. This should ensure that clients querying
2667 RRs via resolved will get similar results as those going via
2668 NSS, if nss-myhostname is enabled.
2670 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2671 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2672 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2673 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2674 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2675 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2676 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2677 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2678 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2679 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2680 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2681 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2683 — Berlin, 2015-08-27
2687 * The systemd-efi-boot-generator functionality was merged into
2688 systemd-gpt-auto-generator.
2690 * systemd-networkd now supports Group Policy for vxlan
2691 devices. It can be enabled via the new boolean configuration
2692 option called 'GroupPolicyExtension='.
2694 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2695 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2696 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2698 — Berlin, 2015-07-31
2702 * The python-systemd code has been removed from the systemd repository.
2703 A new repository has been created which accommodates the code from
2704 now on, and we kindly ask distributions to create a separate package
2705 for this: https://github.com/systemd/python-systemd
2707 * The systemd daemon will now reload its main configuration
2708 (/etc/systemd/system.conf) on daemon-reload.
2710 * sd-dhcp now exposes vendor specific extensions via
2711 sd_dhcp_lease_get_vendor_specific().
2713 * systemd-networkd gained a number of new configuration options.
2715 - A new boolean configuration option for TAP devices called
2716 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2717 device, thus allowing to send and receive GSO packets.
2719 - A new tunnel configuration option called 'CopyDSCP='.
2720 If enabled, the DSCP field of ip6 tunnels is copied into the
2721 decapsulated packet.
2723 - A set of boolean bridge configuration options were added.
2724 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2725 and 'UnicastFlood=' are now parsed by networkd and applied to the
2726 respective bridge link device via the respective IFLA_BRPORT_*
2729 - A new string configuration option to override the hostname sent
2730 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2731 is true, networkd will use the configured hostname instead of the
2732 system hostname when sending DHCP requests.
2734 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2735 networkd will configure the IPv6 flow-label of the tunnel device
2736 according to RFC2460.
2738 - The 'macvtap' virtual network devices are now supported, similar to
2739 the already supported 'macvlan' devices.
2741 * systemd-resolved now implements RFC5452 to improve resilience against
2742 cache poisoning. Additionally, source port randomization is enabled
2743 by default to further protect against DNS spoofing attacks.
2745 * nss-mymachines now supports translating UIDs and GIDs of running
2746 containers with user-namespaces enabled. If a container 'foo'
2747 translates a host uid 'UID' to the container uid 'TUID', then
2748 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2749 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2750 mapped as 'vg-foo-TGID'.
2752 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2753 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2754 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2755 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2756 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2757 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2758 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2759 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2760 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2761 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2763 — Berlin, 2015-07-29
2767 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2768 There are no known issues with current sysfs, and udev does not need
2769 or should be used to work around such bugs.
2771 * udev does no longer enable USB HID power management. Several reports
2772 indicate, that some devices cannot handle that setting.
2774 * The udev accelerometer helper was removed. The functionality
2775 is now fully included in iio-sensor-proxy. But this means,
2776 older iio-sensor-proxy versions will no longer provide
2777 accelerometer/orientation data with this systemd version.
2778 Please upgrade iio-sensor-proxy to version 1.0.
2780 * networkd gained a new configuration option IPv6PrivacyExtensions=
2781 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2782 for Stateless Address") on selected networks.
2784 * For the sake of fewer build-time dependencies and less code in the
2785 main repository, the python bindings are about to be removed in the
2786 next release. A new repository has been created which accommodates
2787 the code from now on, and we kindly ask distributions to create a
2788 separate package for this. The removal will take place in v223.
2790 https://github.com/systemd/python-systemd
2792 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2793 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2794 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2795 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2796 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2797 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2798 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2799 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2800 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2801 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2803 — Berlin, 2015-07-07
2807 * The sd-bus.h and sd-event.h APIs have now been declared
2808 stable and have been added to the official interface of
2809 libsystemd.so. sd-bus implements an alternative D-Bus client
2810 library, that is relatively easy to use, very efficient and
2811 supports both classic D-Bus as well as kdbus as transport
2812 backend. sd-event is a generic event loop abstraction that
2813 is built around Linux epoll, but adds features such as event
2814 prioritization or efficient timer handling. Both APIs are good
2815 choices for C programs looking for a bus and/or event loop
2816 implementation that is minimal and does not have to be
2817 portable to other kernels.
2819 * kdbus support is no longer compile-time optional. It is now
2820 always built-in. However, it can still be disabled at
2821 runtime using the kdbus=0 kernel command line setting, and
2822 that setting may be changed to default to off, by specifying
2823 --disable-kdbus at build-time. Note though that the kernel
2824 command line setting has no effect if the kdbus.ko kernel
2825 module is not installed, in which case kdbus is (obviously)
2826 also disabled. We encourage all downstream distributions to
2827 begin testing kdbus by adding it to the kernel images in the
2828 development distributions, and leaving kdbus support in
2831 * The minimal required util-linux version has been bumped to
2834 * Support for chkconfig (--enable-chkconfig) was removed in
2835 favor of calling an abstraction tool
2836 /lib/systemd/systemd-sysv-install. This needs to be
2837 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2838 in README for details.
2840 * If there's a systemd unit and a SysV init script for the
2841 same service name, and the user executes "systemctl enable"
2842 for it (or a related call), then this will now enable both
2843 (or execute the related operation on both), not just the
2846 * The libudev API documentation has been converted from gtkdoc
2849 * gudev has been removed from the systemd tree, it is now an
2852 * The systemd-cgtop tool learnt a new --raw switch to generate
2853 "raw" (machine parsable) output.
2855 * networkd's IPForwarding= .network file setting learnt the
2856 new setting "kernel", which ensures that networkd does not
2857 change the IP forwarding sysctl from the default kernel
2860 * The systemd-logind bus API now exposes a new boolean
2861 property "Docked" that reports whether logind considers the
2862 system "docked", i.e. connected to a docking station or not.
2864 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2865 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2866 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2867 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2868 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2869 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2870 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2871 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2872 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2873 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2874 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2875 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2876 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2877 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2878 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2879 Fink, Zbigniew Jędrzejewski-Szmek
2881 — Berlin, 2015-06-19
2885 * The gudev library has been extracted into a separate repository
2886 available at: https://git.gnome.org/browse/libgudev/
2887 It is now managed as part of the Gnome project. Distributions
2888 are recommended to pass --disable-gudev to systemd and use
2889 gudev from the Gnome project instead. gudev is still included
2890 in systemd, for now. It will be removed soon, though. Please
2891 also see the announcement-thread on systemd-devel:
2892 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2894 * systemd now exposes a CPUUsageNSec= property for each
2895 service unit on the bus, that contains the overall consumed
2896 CPU time of a service (the sum of what each process of the
2897 service consumed). This value is only available if
2898 CPUAccounting= is turned on for a service, and is then shown
2899 in the "systemctl status" output.
2901 * Support for configuring alternative mappings of the old SysV
2902 runlevels to systemd targets has been removed. They are now
2903 hardcoded in a way that runlevels 2, 3, 4 all map to
2904 multi-user.target and 5 to graphical.target (which
2905 previously was already the default behaviour).
2907 * The auto-mounter logic gained support for mount point
2908 expiry, using a new TimeoutIdleSec= setting in .automount
2909 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2911 * The EFI System Partition (ESP) as mounted to /boot by
2912 systemd-efi-boot-generator will now be unmounted
2913 automatically after 2 minutes of not being used. This should
2914 minimize the risk of ESP corruptions.
2916 * New /etc/fstab options x-systemd.requires= and
2917 x-systemd.requires-mounts-for= are now supported to express
2918 additional dependencies for mounts. This is useful for
2919 journalling file systems that support external journal
2920 devices or overlay file systems that require underlying file
2921 systems to be mounted.
2923 * systemd does not support direct live-upgrades (via systemctl
2924 daemon-reexec) from versions older than v44 anymore. As no
2925 distribution we are aware of shipped such old versions in a
2926 stable release this should not be problematic.
2928 * When systemd forks off a new per-connection service instance
2929 it will now set the $REMOTE_ADDR environment variable to the
2930 remote IP address, and $REMOTE_PORT environment variable to
2931 the remote IP port. This behaviour is similar to the
2932 corresponding environment variables defined by CGI.
2934 * systemd-networkd gained support for uplink failure
2935 detection. The BindCarrier= option allows binding interface
2936 configuration dynamically to the link sense of other
2937 interfaces. This is useful to achieve behaviour like in
2940 * systemd-networkd gained support for configuring the DHCP
2941 client identifier to use when requesting leases.
2943 * systemd-networkd now has a per-network UseNTP= option to
2944 configure whether NTP server information acquired via DHCP
2945 is passed on to services like systemd-timesyncd.
2947 * systemd-networkd gained support for vti6 tunnels.
2949 * Note that systemd-networkd manages the sysctl variable
2950 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2951 it is configured for since v219. The variable controls IP
2952 forwarding, and is a per-interface alternative to the global
2953 /proc/sys/net/ipv[46]/ip_forward. This setting is
2954 configurable in the IPForward= option, which defaults to
2955 "no". This means if networkd is used for an interface it is
2956 no longer sufficient to set the global sysctl option to turn
2957 on IP forwarding! Instead, the .network file option
2958 IPForward= needs to be turned on! Note that the
2959 implementation of this behaviour was broken in v219 and has
2962 * Many bonding and vxlan options are now configurable in
2965 * systemd-nspawn gained a new --property= setting to set unit
2966 properties for the container scope. This is useful for
2967 setting resource parameters (e.g. "CPUShares=500") on
2968 containers started from the command line.
2970 * systemd-nspawn gained a new --private-users= switch to make
2971 use of user namespacing available on recent Linux kernels.
2973 * systemd-nspawn may now be called as part of a shell pipeline
2974 in which case the pipes used for stdin and stdout are passed
2975 directly to the process invoked in the container, without
2976 indirection via a pseudo tty.
2978 * systemd-nspawn gained a new switch to control the UNIX
2979 signal to use when killing the init process of the container
2982 * systemd-nspawn gained a new --overlay= switch for mounting
2983 overlay file systems into the container using the new kernel
2986 * When a container image is imported via systemd-importd and
2987 the host file system is not btrfs, a loopback block device
2988 file is created in /var/lib/machines.raw with a btrfs file
2989 system inside. It is then mounted to /var/lib/machines to
2990 enable btrfs features for container management. The loopback
2991 file and btrfs file system is grown as needed when container
2992 images are imported via systemd-importd.
2994 * systemd-machined/systemd-importd gained support for btrfs
2995 quota, to enforce container disk space limits on disk. This
2996 is exposed in "machinectl set-limit".
2998 * systemd-importd now can import containers from local .tar,
2999 .raw and .qcow2 images, and export them to .tar and .raw. It
3000 can also import dkr v2 images now from the network (on top
3003 * systemd-importd gained support for verifying downloaded
3004 images with gpg2 (previously only gpg1 was supported).
3006 * systemd-machined, systemd-logind, systemd: most bus calls
3007 are now accessible to unprivileged processes via
3008 PolicyKit. Also, systemd-logind will now allow users to kill
3009 their own sessions without further privileges or
3012 * systemd-shutdownd has been removed. This service was
3013 previously responsible for implementing scheduled shutdowns
3014 as exposed in /usr/bin/shutdown's time parameter. This
3015 functionality has now been moved into systemd-logind and is
3016 accessible via a bus interface.
3018 * "systemctl reboot" gained a new switch --firmware-setup that
3019 can be used to reboot into the EFI firmware setup, if that
3020 is available. systemd-logind now exposes an API on the bus
3021 to trigger such reboots, in case graphical desktop UIs want
3022 to cover this functionality.
3024 * "systemctl enable", "systemctl disable" and "systemctl mask"
3025 now support a new "--now" switch. If specified the units
3026 that are enabled will also be started, and the ones
3027 disabled/masked also stopped.
3029 * The Gummiboot EFI boot loader tool has been merged into
3030 systemd, and renamed to "systemd-boot". The bootctl tool has been
3031 updated to support systemd-boot.
3033 * An EFI kernel stub has been added that may be used to create
3034 kernel EFI binaries that contain not only the actual kernel,
3035 but also an initrd, boot splash, command line and OS release
3036 information. This combined binary can then be signed as a
3037 single image, so that the firmware can verify it all in one
3038 step. systemd-boot has special support for EFI binaries created
3039 like this and can extract OS release information from them
3040 and show them in the boot menu. This functionality is useful
3041 to implement cryptographically verified boot schemes.
3043 * Optional support has been added to systemd-fsck to pass
3044 fsck's progress report to an AF_UNIX socket in the file
3047 * udev will no longer create device symlinks for all block
3048 devices by default. A blacklist for excluding special block
3049 devices from this logic has been turned into a whitelist
3050 that requires picking block devices explicitly that require
3053 * A new (currently still internal) API sd-device.h has been
3054 added to libsystemd. This modernized API is supposed to
3055 replace libudev eventually. In fact, already much of libudev
3056 is now just a wrapper around sd-device.h.
3058 * A new hwdb database for storing metadata about pointing
3059 stick devices has been added.
3061 * systemd-tmpfiles gained support for setting file attributes
3062 similar to the "chattr" tool with new 'h' and 'H' lines.
3064 * systemd-journald will no longer unconditionally set the
3065 btrfs NOCOW flag on new journal files. This is instead done
3066 with tmpfiles snippet using the new 'h' line type. This
3067 allows easy disabling of this logic, by masking the
3068 journal-nocow.conf tmpfiles file.
3070 * systemd-journald will now translate audit message types to
3071 human readable identifiers when writing them to the
3072 journal. This should improve readability of audit messages.
3074 * The LUKS logic gained support for the offset= and skip=
3075 options in /etc/crypttab, as previously implemented by
3078 * /usr/lib/os-release gained a new optional field VARIANT= for
3079 distributions that support multiple variants (such as a
3080 desktop edition, a server edition, ...)
3082 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3083 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3084 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3085 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3086 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3087 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3088 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3089 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3090 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3091 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3092 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3093 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3094 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3095 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3096 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3097 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3098 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3099 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3100 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3101 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3102 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3103 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3104 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3105 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3106 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3107 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3108 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3110 — Berlin, 2015-05-22
3114 * Introduce a new API "sd-hwdb.h" for querying the hardware
3115 metadata database. With this minimal interface one can query
3116 and enumerate the udev hwdb, decoupled from the old libudev
3117 library. libudev's interface for this is now only a wrapper
3118 around sd-hwdb. A new tool systemd-hwdb has been added to
3119 interface with and update the database.
3121 * When any of systemd's tools copies files (for example due to
3122 tmpfiles' C lines) a btrfs reflink will attempted first,
3123 before bytewise copying is done.
3125 * systemd-nspawn gained a new --ephemeral switch. When
3126 specified a btrfs snapshot is taken of the container's root
3127 directory, and immediately removed when the container
3128 terminates again. Thus, a container can be started whose
3129 changes never alter the container's root directory, and are
3130 lost on container termination. This switch can also be used
3131 for starting a container off the root file system of the
3132 host without affecting the host OS. This switch is only
3133 available on btrfs file systems.
3135 * systemd-nspawn gained a new --template= switch. It takes the
3136 path to a container tree to use as template for the tree
3137 specified via --directory=, should that directory be
3138 missing. This allows instantiating containers dynamically,
3139 on first run. This switch is only available on btrfs file
3142 * When a .mount unit refers to a mount point on which multiple
3143 mounts are stacked, and the .mount unit is stopped all of
3144 the stacked mount points will now be unmounted until no
3145 mount point remains.
3147 * systemd now has an explicit notion of supported and
3148 unsupported unit types. Jobs enqueued for unsupported unit
3149 types will now fail with an "unsupported" error code. More
3150 specifically .swap, .automount and .device units are not
3151 supported in containers, .busname units are not supported on
3152 non-kdbus systems. .swap and .automount are also not
3153 supported if their respective kernel compile time options
3156 * machinectl gained support for two new "copy-from" and
3157 "copy-to" commands for copying files from a running
3158 container to the host or vice versa.
3160 * machinectl gained support for a new "bind" command to bind
3161 mount host directories into local containers. This is
3162 currently only supported for nspawn containers.
3164 * networkd gained support for configuring bridge forwarding
3165 database entries (fdb) from .network files.
3167 * A new tiny daemon "systemd-importd" has been added that can
3168 download container images in tar, raw, qcow2 or dkr formats,
3169 and make them available locally in /var/lib/machines, so
3170 that they can run as nspawn containers. The daemon can GPG
3171 verify the downloads (not supported for dkr, since it has no
3172 provisions for verifying downloads). It will transparently
3173 decompress bz2, xz, gzip compressed downloads if necessary,
3174 and restore sparse files on disk. The daemon uses privilege
3175 separation to ensure the actual download logic runs with
3176 fewer privileges than the daemon itself. machinectl has
3177 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3178 make the functionality of importd available to the
3179 user. With this in place the Fedora and Ubuntu "Cloud"
3180 images can be downloaded and booted as containers unmodified
3181 (the Fedora images lack the appropriate GPG signature files
3182 currently, so they cannot be verified, but this will change
3183 soon, hopefully). Note that downloading images is currently
3184 only fully supported on btrfs.
3186 * machinectl is now able to list container images found in
3187 /var/lib/machines, along with some metadata about sizes of
3188 disk and similar. If the directory is located on btrfs and
3189 quota is enabled, this includes quota display. A new command
3190 "image-status" has been added that shows additional
3191 information about images.
3193 * machinectl is now able to clone container images
3194 efficiently, if the underlying file system (btrfs) supports
3195 it, with the new "machinectl clone" command. It also
3196 gained commands for renaming and removing images, as well as
3197 marking them read-only or read-write (supported also on
3198 legacy file systems).
3200 * networkd gained support for collecting LLDP network
3201 announcements, from hardware that supports this. This is
3202 shown in networkctl output.
3204 * systemd-run gained support for a new -t (--pty) switch for
3205 invoking a binary on a pty whose input and output is
3206 connected to the invoking terminal. This allows executing
3207 processes as system services while interactively
3208 communicating with them via the terminal. Most interestingly
3209 this is supported across container boundaries. Invoking
3210 "systemd-run -t /bin/bash" is an alternative to running a
3211 full login session, the difference being that the former
3212 will not register a session, nor go through the PAM session
3215 * tmpfiles gained support for a new "v" line type for creating
3216 btrfs subvolumes. If the underlying file system is a legacy
3217 file system, this automatically degrades to creating a
3218 normal directory. Among others /var/lib/machines is now
3219 created like this at boot, should it be missing.
3221 * The directory /var/lib/containers/ has been deprecated and
3222 been replaced by /var/lib/machines. The term "machines" has
3223 been used in the systemd context as generic term for both
3224 VMs and containers, and hence appears more appropriate for
3225 this, as the directory can also contain raw images bootable
3228 * systemd-nspawn when invoked with -M but without --directory=
3229 or --image= is now capable of searching for the container
3230 root directory, subvolume or disk image automatically, in
3231 /var/lib/machines. systemd-nspawn@.service has been updated
3232 to make use of this, thus allowing it to be used for raw
3235 * A new machines.target unit has been introduced that is
3236 supposed to group all containers/VMs invoked as services on
3237 the system. systemd-nspawn@.service has been updated to
3238 integrate with that.
3240 * machinectl gained a new "start" command, for invoking a
3241 container as a service. "machinectl start foo" is mostly
3242 equivalent to "systemctl start systemd-nspawn@foo.service",
3243 but handles escaping in a nicer way.
3245 * systemd-nspawn will now mount most of the cgroupfs tree
3246 read-only into each container, with the exception of the
3247 container's own subtree in the name=systemd hierarchy.
3249 * journald now sets the special FS_NOCOW file flag for its
3250 journal files. This should improve performance on btrfs, by
3251 avoiding heavy fragmentation when journald's write-pattern
3252 is used on COW file systems. It degrades btrfs' data
3253 integrity guarantees for the files to the same levels as for
3254 ext3/ext4 however. This should be OK though as journald does
3255 its own data integrity checks and all its objects are
3256 checksummed on disk. Also, journald should handle btrfs disk
3257 full events a lot more gracefully now, by processing SIGBUS
3258 errors, and not relying on fallocate() anymore.
3260 * When journald detects that journal files it is writing to
3261 have been deleted it will immediately start new journal
3264 * systemd now provides a way to store file descriptors
3265 per-service in PID 1. This is useful for daemons to ensure
3266 that fds they require are not lost during a daemon
3267 restart. The fds are passed to the daemon on the next
3268 invocation in the same way socket activation fds are
3269 passed. This is now used by journald to ensure that the
3270 various sockets connected to all the system's stdout/stderr
3271 are not lost when journald is restarted. File descriptors
3272 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3273 an extension to sd_notify(). Note that a limit is enforced
3274 on the number of fds a service can store in PID 1, and it
3275 defaults to 0, so that no fds may be stored, unless this is
3276 explicitly turned on.
3278 * The default TERM variable to use for units connected to a
3279 terminal, when no other value is explicitly is set is now
3280 vt220 rather than vt102. This should be fairly safe still,
3281 but allows PgUp/PgDn work.
3283 * The /etc/crypttab option header= as known from Debian is now
3286 * "loginctl user-status" and "loginctl session-status" will
3287 now show the last 10 lines of log messages of the
3288 user/session following the status output. Similar,
3289 "machinectl status" will show the last 10 log lines
3290 associated with a virtual machine or container
3291 service. (Note that this is usually not the log messages
3292 done in the VM/container itself, but simply what the
3293 container manager logs. For nspawn this includes all console
3296 * "loginctl session-status" without further argument will now
3297 show the status of the session of the caller. Similar,
3298 "lock-session", "unlock-session", "activate",
3299 "enable-linger", "disable-linger" may now be called without
3300 session/user parameter in which case they apply to the
3301 caller's session/user.
3303 * An X11 session scriptlet is now shipped that uploads
3304 $DISPLAY and $XAUTHORITY into the environment of the systemd
3305 --user daemon if a session begins. This should improve
3306 compatibility with X11 enabled applications run as systemd
3309 * Generators are now subject to masking via /etc and /run, the
3310 same way as unit files.
3312 * networkd .network files gained support for configuring
3313 per-link IPv4/IPv6 packet forwarding as well as IPv4
3314 masquerading. This is by default turned on for veth links to
3315 containers, as registered by systemd-nspawn. This means that
3316 nspawn containers run with --network-veth will now get
3317 automatic routed access to the host's networks without any
3318 further configuration or setup, as long as networkd runs on
3321 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3322 or UDP posts of a container on the host. With this in place
3323 it is possible to run containers with private veth links
3324 (--network-veth), and have their functionality exposed on
3325 the host as if their services were running directly on the
3328 * systemd-nspawn's --network-veth switch now gained a short
3329 version "-n", since with the changes above it is now truly
3330 useful out-of-the-box. The systemd-nspawn@.service has been
3331 updated to make use of it too by default.
3333 * systemd-nspawn will now maintain a per-image R/W lock, to
3334 ensure that the same image is not started more than once
3335 writable. (It's OK to run an image multiple times
3336 simultaneously in read-only mode.)
3338 * systemd-nspawn's --image= option is now capable of
3339 dissecting and booting MBR and GPT disk images that contain
3340 only a single active Linux partition. Previously it
3341 supported only GPT disk images with proper GPT type
3342 IDs. This allows running cloud images from major
3343 distributions directly with systemd-nspawn, without
3346 * In addition to collecting mouse dpi data in the udev
3347 hardware database, there's now support for collecting angle
3348 information for mouse scroll wheels. The database is
3349 supposed to guarantee similar scrolling behavior on mice
3350 that it knows about. There's also support for collecting
3351 information about Touchpad types.
3353 * udev's input_id built-in will now also collect touch screen
3354 dimension data and attach it to probed devices.
3356 * /etc/os-release gained support for a Distribution Privacy
3359 * networkd gained support for creating "ipvlan", "gretap",
3360 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3362 * systemd-tmpfiles gained support for "a" lines for setting
3365 * systemd-nspawn will now mount /tmp in the container to
3366 tmpfs, automatically.
3368 * systemd now exposes the memory.usage_in_bytes cgroup
3369 attribute and shows it for each service in the "systemctl
3370 status" output, if available.
3372 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3373 immediate reboot is triggered. This useful if shutdown is
3374 hung and is unable to complete, to expedite the
3375 operation. Note that this kind of reboot will still unmount
3376 all file systems, and hence should not result in fsck being
3379 * A .device unit for an optical block device will now be
3380 considered active only when a medium is in the drive. Also,
3381 mount units are now bound to their backing devices thus
3382 triggering automatic unmounting when devices become
3383 unavailable. With this in place systemd will now
3384 automatically unmount left-over mounts when a CD-ROM is
3385 ejected or an USB stick is yanked from the system.
3387 * networkd-wait-online now has support for waiting for
3388 specific interfaces only (with globbing), and for giving up
3389 after a configurable timeout.
3391 * networkd now exits when idle. It will be automatically
3392 restarted as soon as interfaces show up, are removed or
3393 change state. networkd will stay around as long as there is
3394 at least one DHCP state machine or similar around, that keep
3397 * networkd may now configure IPv6 link-local addressing in
3398 addition to IPv4 link-local addressing.
3400 * The IPv6 "token" for use in SLAAC may now be configured for
3401 each .network interface in networkd.
3403 * Routes configured with networkd may now be assigned a scope
3406 * networkd's [Match] sections now support globbing and lists
3407 of multiple space-separated matches per item.
3409 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3410 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3411 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3412 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3413 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3414 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3415 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3416 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3417 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3418 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3419 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3420 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3421 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3422 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3423 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3424 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3425 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3426 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3427 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3428 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3429 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3430 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3431 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3432 Hoffmann, Zbigniew Jędrzejewski-Szmek
3434 — Berlin, 2015-02-16
3438 * When querying unit file enablement status (for example via
3439 "systemctl is-enabled"), a new state "indirect" is now known
3440 which indicates that a unit might not be enabled itself, but
3441 another unit listed in its Also= setting might be.
3443 * Similar to the various existing ConditionXYZ= settings for
3444 units, there are now matching AssertXYZ= settings. While
3445 failing conditions cause a unit to be skipped, but its job
3446 to succeed, failing assertions declared like this will cause
3447 a unit start operation and its job to fail.
3449 * hostnamed now knows a new chassis type "embedded".
3451 * systemctl gained a new "edit" command. When used on a unit
3452 file, this allows extending unit files with .d/ drop-in
3453 configuration snippets or editing the full file (after
3454 copying it from /usr/lib to /etc). This will invoke the
3455 user's editor (as configured with $EDITOR), and reload the
3456 modified configuration after editing.
3458 * "systemctl status" now shows the suggested enablement state
3459 for a unit, as declared in the (usually vendor-supplied)
3460 system preset files.
3462 * nss-myhostname will now resolve the single-label host name
3463 "gateway" to the locally configured default IP routing
3464 gateways, ordered by their metrics. This assigns a stable
3465 name to the used gateways, regardless which ones are
3466 currently configured. Note that the name will only be
3467 resolved after all other name sources (if nss-myhostname is
3468 configured properly) and should hence not negatively impact
3469 systems that use the single-label host name "gateway" in
3472 * systemd-inhibit now allows filtering by mode when listing
3475 * Scope and service units gained a new "Delegate" boolean
3476 property, which, when set, allows processes running inside the
3477 unit to further partition resources. This is primarily
3478 useful for systemd user instances as well as container
3481 * journald will now pick up audit messages directly from
3482 the kernel, and log them like any other log message. The
3483 audit fields are split up and fully indexed. This means that
3484 journalctl in many ways is now a (nicer!) alternative to
3485 ausearch, the traditional audit client. Note that this
3486 implements only a minimal audit client. If you want the
3487 special audit modes like reboot-on-log-overflow, please use
3488 the traditional auditd instead, which can be used in
3489 parallel to journald.
3491 * The ConditionSecurity= unit file option now understands the
3492 special string "audit" to check whether auditing is
3495 * journalctl gained two new commands --vacuum-size= and
3496 --vacuum-time= to delete old journal files until the
3497 remaining ones take up no more than the specified size on disk,
3498 or are not older than the specified time.
3500 * A new, native PPPoE library has been added to sd-network,
3501 systemd's library of light-weight networking protocols. This
3502 library will be used in a future version of networkd to
3503 enable PPPoE communication without an external pppd daemon.
3505 * The busctl tool now understands a new "capture" verb that
3506 works similar to "monitor", but writes a packet capture
3507 trace to STDOUT that can be redirected to a file which is
3508 compatible with libcap's capture file format. This can then
3509 be loaded in Wireshark and similar tools to inspect bus
3512 * The busctl tool now understands a new "tree" verb that shows
3513 the object trees of a specific service on the bus, or of all
3516 * The busctl tool now understands a new "introspect" verb that
3517 shows all interfaces and members of objects on the bus,
3518 including their signature and values. This is particularly
3519 useful to get more information about bus objects shown by
3520 the new "busctl tree" command.
3522 * The busctl tool now understands new verbs "call",
3523 "set-property" and "get-property" for invoking bus method
3524 calls, setting and getting bus object properties in a
3527 * busctl gained a new --augment-creds= argument that controls
3528 whether the tool shall augment credential information it
3529 gets from the bus with data from /proc, in a possibly
3532 * nspawn's --link-journal= switch gained two new values
3533 "try-guest" and "try-host" that work like "guest" and
3534 "host", but do not fail if the host has no persistent
3535 journalling enabled. -j is now equivalent to
3536 --link-journal=try-guest.
3538 * macvlan network devices created by nspawn will now have
3539 stable MAC addresses.
3541 * A new SmackProcessLabel= unit setting has been added, which
3542 controls the SMACK security label processes forked off by
3543 the respective unit shall use.
3545 * If compiled with --enable-xkbcommon, systemd-localed will
3546 verify x11 keymap settings by compiling the given keymap. It
3547 will spew out warnings if the compilation fails. This
3548 requires libxkbcommon to be installed.
3550 * When a coredump is collected, a larger number of metadata
3551 fields is now collected and included in the journal records
3552 created for it. More specifically, control group membership,
3553 environment variables, memory maps, working directory,
3554 chroot directory, /proc/$PID/status, and a list of open file
3555 descriptors is now stored in the log entry.
3557 * The udev hwdb now contains DPI information for mice. For
3560 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3562 * All systemd programs that read standalone configuration
3563 files in /etc now also support a corresponding series of
3564 .conf.d configuration directories in /etc/, /run/,
3565 /usr/local/lib/, /usr/lib/, and (if configured with
3566 --enable-split-usr) /lib/. In particular, the following
3567 configuration files now have corresponding configuration
3568 directories: system.conf user.conf, logind.conf,
3569 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3570 resolved.conf, timesyncd.conf, journal-remote.conf, and
3571 journal-upload.conf. Note that distributions should use the
3572 configuration directories in /usr/lib/; the directories in
3573 /etc/ are reserved for the system administrator.
3575 * systemd-rfkill will no longer take the rfkill device name
3576 into account when storing rfkill state on disk, as the name
3577 might be dynamically assigned and not stable. Instead, the
3578 ID_PATH udev variable combined with the rfkill type (wlan,
3579 bluetooth, ...) is used.
3581 * A new service systemd-machine-id-commit.service has been
3582 added. When used on systems where /etc is read-only during
3583 boot, and /etc/machine-id is not initialized (but an empty
3584 file), this service will copy the temporary machine ID
3585 created as replacement into /etc after the system is fully
3586 booted up. This is useful for systems that are freshly
3587 installed with a non-initialized machine ID, but should get
3588 a fixed machine ID for subsequent boots.
3590 * networkd's .netdev files now provide a large set of
3591 configuration parameters for VXLAN devices. Similarly, the
3592 bridge port cost parameter is now configurable in .network
3593 files. There's also new support for configuring IP source
3594 routing. networkd .link files gained support for a new
3595 OriginalName= match that is useful to match against the
3596 original interface name the kernel assigned. .network files
3597 may include MTU= and MACAddress= fields for altering the MTU
3598 and MAC address while being connected to a specific network
3601 * The LUKS logic gained supported for configuring
3602 UUID-specific key files. There's also new support for naming
3603 LUKS device from the kernel command line, using the new
3604 luks.name= argument.
3606 * Timer units may now be transiently created via the bus API
3607 (this was previously already available for scope and service
3608 units). In addition it is now possible to create multiple
3609 transient units at the same time with a single bus call. The
3610 "systemd-run" tool has been updated to make use of this for
3611 running commands on a specified time, in at(1)-style.
3613 * tmpfiles gained support for "t" lines, for assigning
3614 extended attributes to files. Among other uses this may be
3615 used to assign SMACK labels to files.
3617 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3618 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3619 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3620 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3621 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3622 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3623 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3624 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3625 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3626 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3627 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3628 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3629 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3630 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3631 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3632 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3633 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3634 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3636 — Berlin, 2014-12-10
3640 * journalctl gained the new options -t/--identifier= to match
3641 on the syslog identifier (aka "tag"), as well as --utc to
3642 show log timestamps in the UTC timezone. journalctl now also
3643 accepts -n/--lines=all to disable line capping in a pager.
3645 * journalctl gained a new switch, --flush, that synchronously
3646 flushes logs from /run/log/journal to /var/log/journal if
3647 persistent storage is enabled. systemd-journal-flush.service
3648 now waits until the operation is complete.
3650 * Services can notify the manager before they start a reload
3651 (by sending RELOADING=1) or shutdown (by sending
3652 STOPPING=1). This allows the manager to track and show the
3653 internal state of daemons and closes a race condition when
3654 the process is still running but has closed its D-Bus
3657 * Services with Type=oneshot do not have to have any ExecStart
3660 * User units are now loaded also from
3661 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3662 /run/systemd/user directory that was already previously
3663 supported, but is under the control of the user.
3665 * Job timeouts (i.e. time-outs on the time a job that is
3666 queued stays in the run queue) can now optionally result in
3667 immediate reboot or power-off actions (JobTimeoutAction= and
3668 JobTimeoutRebootArgument=). This is useful on ".target"
3669 units, to limit the maximum time a target remains
3670 undispatched in the run queue, and to trigger an emergency
3671 operation in such a case. This is now used by default to
3672 turn off the system if boot-up (as defined by everything in
3673 basic.target) hangs and does not complete for at least
3674 15min. Also, if power-off or reboot hang for at least 30min
3675 an immediate power-off/reboot operation is triggered. This
3676 functionality is particularly useful to increase reliability
3677 on embedded devices, but also on laptops which might
3678 accidentally get powered on when carried in a backpack and
3679 whose boot stays stuck in a hard disk encryption passphrase
3682 * systemd-logind can be configured to also handle lid switch
3683 events even when the machine is docked or multiple displays
3684 are attached (HandleLidSwitchDocked= option).
3686 * A helper binary and a service have been added which can be
3687 used to resume from hibernation in the initramfs. A
3688 generator will parse the resume= option on the kernel
3689 command line to trigger resume.
3691 * A user console daemon systemd-consoled has been
3692 added. Currently, it is a preview, and will so far open a
3693 single terminal on each session of the user marked as
3694 Desktop=systemd-console.
3696 * Route metrics can be specified for DHCP routes added by
3699 * The SELinux context of socket-activated services can be set
3700 from the information provided by the networking stack
3701 (SELinuxContextFromNet= option).
3703 * Userspace firmware loading support has been removed and
3704 the minimum supported kernel version is thus bumped to 3.7.
3706 * Timeout for udev workers has been increased from 1 to 3
3707 minutes, but a warning will be printed after 1 minute to
3708 help diagnose kernel modules that take a long time to load.
3710 * Udev rules can now remove tags on devices with TAG-="foobar".
3712 * systemd's readahead implementation has been removed. In many
3713 circumstances it didn't give expected benefits even for
3714 rotational disk drives and was becoming less relevant in the
3715 age of SSDs. As none of the developers has been using
3716 rotating media anymore, and nobody stepped up to actively
3717 maintain this component of systemd it has now been removed.
3719 * Swap units can use Options= to specify discard options.
3720 Discard options specified for swaps in /etc/fstab are now
3723 * Docker containers are now detected as a separate type of
3726 * The Password Agent protocol gained support for queries where
3727 the user input is shown, useful e.g. for user names.
3728 systemd-ask-password gained a new --echo option to turn that
3731 * The default sysctl.d/ snippets will now set:
3733 net.core.default_qdisc = fq_codel
3735 This selects Fair Queuing Controlled Delay as the default
3736 queuing discipline for network interfaces. fq_codel helps
3737 fight the network bufferbloat problem. It is believed to be
3738 a good default with no tuning required for most workloads.
3739 Downstream distributions may override this choice. On 10Gbit
3740 servers that do not do forwarding, "fq" may perform better.
3741 Systems without a good clocksource should use "pfifo_fast".
3743 * If kdbus is enabled during build a new option BusPolicy= is
3744 available for service units, that allows locking all service
3745 processes into a stricter bus policy, in order to limit
3746 access to various bus services, or even hide most of them
3747 from the service's view entirely.
3749 * networkctl will now show the .network and .link file
3750 networkd has applied to a specific interface.
3752 * sd-login gained a new API call sd_session_get_desktop() to
3753 query which desktop environment has been selected for a
3756 * UNIX utmp support is now compile-time optional to support
3757 legacy-free systems.
3759 * systemctl gained two new commands "add-wants" and
3760 "add-requires" for pulling in units from specific targets
3763 * If the word "rescue" is specified on the kernel command line
3764 the system will now boot into rescue mode (aka
3765 rescue.target), which was previously available only by
3766 specifying "1" or "systemd.unit=rescue.target" on the kernel
3767 command line. This new kernel command line option nicely
3768 mirrors the already existing "emergency" kernel command line
3771 * New kernel command line options mount.usr=, mount.usrflags=,
3772 mount.usrfstype= have been added that match root=, rootflags=,
3773 rootfstype= but allow mounting a specific file system to
3776 * The $NOTIFY_SOCKET is now also passed to control processes of
3777 services, not only the main process.
3779 * This version reenables support for fsck's -l switch. This
3780 means at least version v2.25 of util-linux is required for
3781 operation, otherwise dead-locks on device nodes may
3782 occur. Again: you need to update util-linux to at least
3783 v2.25 when updating systemd to v217.
3785 * The "multi-seat-x" tool has been removed from systemd, as
3786 its functionality has been integrated into X servers 1.16,
3787 and the tool is hence redundant. It is recommended to update
3788 display managers invoking this tool to simply invoke X
3789 directly from now on, again.
3791 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3792 message flag has been added for all of systemd's PolicyKit
3793 authenticated method calls has been added. In particular
3794 this now allows optional interactive authorization via
3795 PolicyKit for many of PID1's privileged operations such as
3796 unit file enabling and disabling.
3798 * "udevadm hwdb --update" learnt a new switch "--usr" for
3799 placing the rebuilt hardware database in /usr instead of
3800 /etc. When used only hardware database entries stored in
3801 /usr will be used, and any user database entries in /etc are
3802 ignored. This functionality is useful for vendors to ship a
3803 pre-built database on systems where local configuration is
3804 unnecessary or unlikely.
3806 * Calendar time specifications in .timer units now also
3807 understand the strings "semi-annually", "quarterly" and
3808 "minutely" as shortcuts (in addition to the preexisting
3809 "anually", "hourly", ...).
3811 * systemd-tmpfiles will now correctly create files in /dev
3812 at boot which are marked for creation only at boot. It is
3813 recommended to always create static device nodes with 'c!'
3814 and 'b!', so that they are created only at boot and not
3815 overwritten at runtime.
3817 * When the watchdog logic is used for a service (WatchdogSec=)
3818 and the watchdog timeout is hit the service will now be
3819 terminated with SIGABRT (instead of just SIGTERM), in order
3820 to make sure a proper coredump and backtrace is
3821 generated. This ensures that hanging services will result in
3822 similar coredump/backtrace behaviour as services that hit a
3825 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3826 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3827 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3828 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3829 Herrmann, David Sommerseth, David Strauss, Emil Renner
3830 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3831 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3832 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3833 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3834 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3835 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3836 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3837 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3838 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3839 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3840 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3841 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3842 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3843 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3844 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3845 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3848 — Berlin, 2014-10-28
3852 * timedated no longer reads NTP implementation unit names from
3853 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3854 implementations should add a
3856 Conflicts=systemd-timesyncd.service
3858 to their unit files to take over and replace systemd's NTP
3859 default functionality.
3861 * systemd-sysusers gained a new line type "r" for configuring
3862 which UID/GID ranges to allocate system users/groups
3863 from. Lines of type "u" may now add an additional column
3864 that specifies the home directory for the system user to be
3865 created. Also, systemd-sysusers may now optionally read user
3866 information from STDIN instead of a file. This is useful for
3867 invoking it from RPM preinst scriptlets that need to create
3868 users before the first RPM file is installed since these
3869 files might need to be owned by them. A new
3870 %sysusers_create_inline RPM macro has been introduced to do
3871 just that. systemd-sysusers now updates the shadow files as
3872 well as the user/group databases, which should enhance
3873 compatibility with certain tools like grpck.
3875 * A number of bus APIs of PID 1 now optionally consult
3876 PolicyKit to permit access for otherwise unprivileged
3877 clients under certain conditions. Note that this currently
3878 doesn't support interactive authentication yet, but this is
3879 expected to be added eventually, too.
3881 * /etc/machine-info now has new fields for configuring the
3882 deployment environment of the machine, as well as the
3883 location of the machine. hostnamectl has been updated with
3884 new command to update these fields.
3886 * systemd-timesyncd has been updated to automatically acquire
3887 NTP server information from systemd-networkd, which might
3888 have been discovered via DHCP.
3890 * systemd-resolved now includes a caching DNS stub resolver
3891 and a complete LLMNR name resolution implementation. A new
3892 NSS module "nss-resolve" has been added which can be used
3893 instead of glibc's own "nss-dns" to resolve hostnames via
3894 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3895 be resolved via systemd-resolved D-Bus APIs. In contrast to
3896 the glibc internal resolver systemd-resolved is aware of
3897 multi-homed system, and keeps DNS server and caches separate
3898 and per-interface. Queries are sent simultaneously on all
3899 interfaces that have DNS servers configured, in order to
3900 properly handle VPNs and local LANs which might resolve
3901 separate sets of domain names. systemd-resolved may acquire
3902 DNS server information from systemd-networkd automatically,
3903 which in turn might have discovered them via DHCP. A tool
3904 "systemd-resolve-host" has been added that may be used to
3905 query the DNS logic in resolved. systemd-resolved implements
3906 IDNA and automatically uses IDNA or UTF-8 encoding depending
3907 on whether classic DNS or LLMNR is used as transport. In the
3908 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3909 implementation to systemd-resolved.
3911 * A new NSS module nss-mymachines has been added, that
3912 automatically resolves the names of all local registered
3913 containers to their respective IP addresses.
3915 * A new client tool "networkctl" for systemd-networkd has been
3916 added. It currently is entirely passive and will query
3917 networking configuration from udev, rtnetlink and networkd,
3918 and present it to the user in a very friendly
3919 way. Eventually, we hope to extend it to become a full
3920 control utility for networkd.
3922 * .socket units gained a new DeferAcceptSec= setting that
3923 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3924 TCP. Similarly, support for controlling TCP keep-alive
3925 settings has been added (KeepAliveTimeSec=,
3926 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3927 turning off Nagle's algorithm on TCP has been added
3930 * logind learned a new session type "web", for use in projects
3931 like Cockpit which register web clients as PAM sessions.
3933 * timer units with at least one OnCalendar= setting will now
3934 be started only after time-sync.target has been
3935 reached. This way they will not elapse before the system
3936 clock has been corrected by a local NTP client or
3937 similar. This is particular useful on RTC-less embedded
3938 machines, that come up with an invalid system clock.
3940 * systemd-nspawn's --network-veth= switch should now result in
3941 stable MAC addresses for both the outer and the inner side
3944 * systemd-nspawn gained a new --volatile= switch for running
3945 container instances with /etc or /var unpopulated.
3947 * The kdbus client code has been updated to use the new Linux
3948 3.17 memfd subsystem instead of the old kdbus-specific one.
3950 * systemd-networkd's DHCP client and server now support
3951 FORCERENEW. There are also new configuration options to
3952 configure the vendor client identifier and broadcast mode
3955 * systemd will no longer inform the kernel about the current
3956 timezone, as this is necessarily incorrect and racy as the
3957 kernel has no understanding of DST and similar
3958 concepts. This hence means FAT timestamps will be always
3959 considered UTC, similar to what Android is already
3960 doing. Also, when the RTC is configured to the local time
3961 (rather than UTC) systemd will never synchronize back to it,
3962 as this might confuse Windows at a later boot.
3964 * systemd-analyze gained a new command "verify" for offline
3965 validation of unit files.
3967 * systemd-networkd gained support for a couple of additional
3968 settings for bonding networking setups. Also, the metric for
3969 statically configured routes may now be configured. For
3970 network interfaces where this is appropriate the peer IP
3971 address may now be configured.
3973 * systemd-networkd's DHCP client will no longer request
3974 broadcasting by default, as this tripped up some networks.
3975 For hardware where broadcast is required the feature should
3976 be switched back on using RequestBroadcast=yes.
3978 * systemd-networkd will now set up IPv4LL addresses (when
3979 enabled) even if DHCP is configured successfully.
3981 * udev will now default to respect network device names given
3982 by the kernel when the kernel indicates that these are
3983 predictable. This behavior can be tweaked by changing
3984 NamePolicy= in the relevant .link file.
3986 * A new library systemd-terminal has been added that
3987 implements full TTY stream parsing and rendering. This
3988 library is supposed to be used later on for implementing a
3989 full userspace VT subsystem, replacing the current kernel
3992 * A new tool systemd-journal-upload has been added to push
3993 journal data to a remote system running
3994 systemd-journal-remote.
3996 * journald will no longer forward all local data to another
3997 running syslog daemon. This change has been made because
3998 rsyslog (which appears to be the most commonly used syslog
3999 implementation these days) no longer makes use of this, and
4000 instead pulls the data out of the journal on its own. Since
4001 forwarding the messages to a non-existent syslog server is
4002 more expensive than we assumed we have now turned this
4003 off. If you run a syslog server that is not a recent rsyslog
4004 version, you have to turn this option on again
4005 (ForwardToSyslog= in journald.conf).
4007 * journald now optionally supports the LZ4 compressor for
4008 larger journal fields. This compressor should perform much
4009 better than XZ which was the previous default.
4011 * machinectl now shows the IP addresses of local containers,
4012 if it knows them, plus the interface name of the container.
4014 * A new tool "systemd-escape" has been added that makes it
4015 easy to escape strings to build unit names and similar.
4017 * sd_notify() messages may now include a new ERRNO= field
4018 which is parsed and collected by systemd and shown among the
4019 "systemctl status" output for a service.
4021 * A new component "systemd-firstboot" has been added that
4022 queries the most basic systemd information (timezone,
4023 hostname, root password) interactively on first
4024 boot. Alternatively it may also be used to provision these
4025 things offline on OS images installed into directories.
4027 * The default sysctl.d/ snippets will now set
4029 net.ipv4.conf.default.promote_secondaries=1
4031 This has the benefit of no flushing secondary IP addresses
4032 when primary addresses are removed.
4034 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4035 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4036 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4037 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4038 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4039 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4040 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4041 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4042 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4043 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4044 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4045 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4046 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4047 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4048 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4050 — Berlin, 2014-08-19
4054 * A new tool systemd-sysusers has been added. This tool
4055 creates system users and groups in /etc/passwd and
4056 /etc/group, based on static declarative system user/group
4057 definitions in /usr/lib/sysusers.d/. This is useful to
4058 enable factory resets and volatile systems that boot up with
4059 an empty /etc directory, and thus need system users and
4060 groups created during early boot. systemd now also ships
4061 with two default sysusers.d/ files for the most basic
4062 users and groups systemd and the core operating system
4065 * A new tmpfiles snippet has been added that rebuilds the
4066 essential files in /etc on boot, should they be missing.
4068 * A directive for ensuring automatic clean-up of
4069 /var/cache/man/ has been removed from the default
4070 configuration. This line should now be shipped by the man
4071 implementation. The necessary change has been made to the
4072 man-db implementation. Note that you need to update your man
4073 implementation to one that ships this line, otherwise no
4074 automatic clean-up of /var/cache/man will take place.
4076 * A new condition ConditionNeedsUpdate= has been added that
4077 may conditionalize services to only run when /etc or /var
4078 are "older" than the vendor operating system resources in
4079 /usr. This is useful for reconstructing or updating /etc
4080 after an offline update of /usr or a factory reset, on the
4081 next reboot. Services that want to run once after such an
4082 update or reset should use this condition and order
4083 themselves before the new systemd-update-done.service, which
4084 will mark the two directories as fully updated. A number of
4085 service files have been added making use of this, to rebuild
4086 the udev hardware database, the journald message catalog and
4087 dynamic loader cache (ldconfig). The systemd-sysusers tool
4088 described above also makes use of this now. With this in
4089 place it is now possible to start up a minimal operating
4090 system with /etc empty cleanly. For more information on the
4091 concepts involved see this recent blog story:
4093 http://0pointer.de/blog/projects/stateless.html
4095 * A new system group "input" has been introduced, and all
4096 input device nodes get this group assigned. This is useful
4097 for system-level software to get access to input devices. It
4098 complements what is already done for "audio" and "video".
4100 * systemd-networkd learnt minimal DHCPv4 server support in
4101 addition to the existing DHCPv4 client support. It also
4102 learnt DHCPv6 client and IPv6 Router Solicitation client
4103 support. The DHCPv4 client gained support for static routes
4104 passed in from the server. Note that the [DHCPv4] section
4105 known in older systemd-networkd versions has been renamed to
4106 [DHCP] and is now also used by the DHCPv6 client. Existing
4107 .network files using settings of this section should be
4108 updated, though compatibility is maintained. Optionally, the
4109 client hostname may now be sent to the DHCP server.
4111 * networkd gained support for vxlan virtual networks as well
4112 as tun/tap and dummy devices.
4114 * networkd gained support for automatic allocation of address
4115 ranges for interfaces from a system-wide pool of
4116 addresses. This is useful for dynamically managing a large
4117 number of interfaces with a single network configuration
4118 file. In particular this is useful to easily assign
4119 appropriate IP addresses to the veth links of a large number
4120 of nspawn instances.
4122 * RPM macros for processing sysusers, sysctl and binfmt
4123 drop-in snippets at package installation time have been
4126 * The /etc/os-release file should now be placed in
4127 /usr/lib/os-release. The old location is automatically
4128 created as symlink. /usr/lib is the more appropriate
4129 location of this file, since it shall actually describe the
4130 vendor operating system shipped in /usr, and not the
4131 configuration stored in /etc.
4133 * .mount units gained a new boolean SloppyOptions= setting
4134 that maps to mount(8)'s -s option which enables permissive
4135 parsing of unknown mount options.
4137 * tmpfiles learnt a new "L+" directive which creates a symlink
4138 but (unlike "L") deletes a pre-existing file first, should
4139 it already exist and not already be the correct
4140 symlink. Similarly, "b+", "c+" and "p+" directives have been
4141 added as well, which create block and character devices, as
4142 well as fifos in the filesystem, possibly removing any
4143 pre-existing files of different types.
4145 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4146 'argument' field (which so far specified the source to
4147 symlink/copy the files from) is now optional. If omitted the
4148 same file os copied from /usr/share/factory/ suffixed by the
4149 full destination path. This is useful for populating /etc
4150 with essential files, by copying them from vendor defaults
4151 shipped in /usr/share/factory/etc.
4153 * A new command "systemctl preset-all" has been added that
4154 applies the service preset settings to all installed unit
4155 files. A new switch --preset-mode= has been added that
4156 controls whether only enable or only disable operations
4159 * A new command "systemctl is-system-running" has been added
4160 that allows checking the overall state of the system, for
4161 example whether it is fully up and running.
4163 * When the system boots up with an empty /etc, the equivalent
4164 to "systemctl preset-all" is executed during early boot, to
4165 make sure all default services are enabled after a factory
4168 * systemd now contains a minimal preset file that enables the
4169 most basic services systemd ships by default.
4171 * Unit files' [Install] section gained a new DefaultInstance=
4172 field for defining the default instance to create if a
4173 template unit is enabled with no instance specified.
4175 * A new passive target cryptsetup-pre.target has been added
4176 that may be used by services that need to make they run and
4177 finish before the first LUKS cryptographic device is set up.
4179 * The /dev/loop-control and /dev/btrfs-control device nodes
4180 are now owned by the "disk" group by default, opening up
4181 access to this group.
4183 * systemd-coredump will now automatically generate a
4184 stack trace of all core dumps taking place on the system,
4185 based on elfutils' libdw library. This stack trace is logged
4188 * systemd-coredump may now optionally store coredumps directly
4189 on disk (in /var/lib/systemd/coredump, possibly compressed),
4190 instead of storing them unconditionally in the journal. This
4191 mode is the new default. A new configuration file
4192 /etc/systemd/coredump.conf has been added to configure this
4193 and other parameters of systemd-coredump.
4195 * coredumpctl gained a new "info" verb to show details about a
4196 specific coredump. A new switch "-1" has also been added
4197 that makes sure to only show information about the most
4198 recent entry instead of all entries. Also, as the tool is
4199 generally useful now the "systemd-" prefix of the binary
4200 name has been removed. Distributions that want to maintain
4201 compatibility with the old name should add a symlink from
4202 the old name to the new name.
4204 * journald's SplitMode= now defaults to "uid". This makes sure
4205 that unprivileged users can access their own coredumps with
4206 coredumpctl without restrictions.
4208 * New kernel command line options "systemd.wants=" (for
4209 pulling an additional unit during boot), "systemd.mask="
4210 (for masking a specific unit for the boot), and
4211 "systemd.debug-shell" (for enabling the debug shell on tty9)
4212 have been added. This is implemented in the new generator
4213 "systemd-debug-generator".
4215 * systemd-nspawn will now by default filter a couple of
4216 syscalls for containers, among them those required for
4217 kernel module loading, direct x86 IO port access, swap
4218 management, and kexec. Most importantly though
4219 open_by_handle_at() is now prohibited for containers,
4220 closing a hole similar to a recently discussed vulnerability
4221 in docker regarding access to files on file hierarchies the
4222 container should normally not have access to. Note that, for
4223 nspawn, we generally make no security claims anyway (and
4224 this is explicitly documented in the man page), so this is
4225 just a fix for one of the most obvious problems.
4227 * A new man page file-hierarchy(7) has been added that
4228 contains a minimized, modernized version of the file system
4229 layout systemd expects, similar in style to the FHS
4230 specification or hier(5). A new tool systemd-path(1) has
4231 been added to query many of these paths for the local
4234 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4235 longer done. Since the directory now has a per-user size
4236 limit, and is cleaned on logout this appears unnecessary,
4237 in particular since this now brings the lifecycle of this
4238 directory closer in line with how IPC objects are handled.
4240 * systemd.pc now exports a number of additional directories,
4241 including $libdir (which is useful to identify the library
4242 path for the primary architecture of the system), and a
4243 couple of drop-in directories.
4245 * udev's predictable network interface names now use the dev_port
4246 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4247 distinguish between ports of the same PCI function. dev_id should
4248 only be used for ports using the same HW address, hence the need
4251 * machined has been updated to export the OS version of a
4252 container (read from /etc/os-release and
4253 /usr/lib/os-release) on the bus. This is now shown in
4254 "machinectl status" for a machine.
4256 * A new service setting RestartForceExitStatus= has been
4257 added. If configured to a set of exit signals or process
4258 return values, the service will be restarted when the main
4259 daemon process exits with any of them, regardless of the
4262 * systemctl's -H switch for connecting to remote systemd
4263 machines has been extended so that it may be used to
4264 directly connect to a specific container on the
4265 host. "systemctl -H root@foobar:waldi" will now connect as
4266 user "root" to host "foobar", and then proceed directly to
4267 the container named "waldi". Note that currently you have to
4268 authenticate as user "root" for this to work, as entering
4269 containers is a privileged operation.
4271 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4272 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4273 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4274 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4275 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4276 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4277 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4278 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4279 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4280 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4281 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4282 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4284 — Berlin, 2014-07-03
4288 * As an experimental feature, udev now tries to lock the
4289 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4290 executes events for the disk or any of its partitions.
4291 Applications like partitioning programs can lock the
4292 disk device node (flock(LOCK_EX)) and claim temporary
4293 device ownership that way; udev will entirely skip all event
4294 handling for this disk and its partitions. If the disk
4295 was opened for writing, the close will trigger a partition
4296 table rescan in udev's "watch" facility, and if needed
4297 synthesize "change" events for the disk and all its partitions.
4298 This is now unconditionally enabled, and if it turns out to
4299 cause major problems, we might turn it on only for specific
4300 devices, or might need to disable it entirely. Device Mapper
4301 devices are excluded from this logic.
4303 * We temporarily dropped the "-l" switch for fsck invocations,
4304 since they collide with the flock() logic above. util-linux
4305 upstream has been changed already to avoid this conflict,
4306 and we will readd "-l" as soon as util-linux with this
4307 change has been released.
4309 * The dependency on libattr has been removed. Since a long
4310 time, the extended attribute calls have moved to glibc, and
4311 libattr is thus unnecessary.
4313 * Virtualization detection works without privileges now. This
4314 means the systemd-detect-virt binary no longer requires
4315 CAP_SYS_PTRACE file capabilities, and our daemons can run
4316 with fewer privileges.
4318 * systemd-networkd now runs under its own "systemd-network"
4319 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4320 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4321 loses the ability to write to files owned by root this way.
4323 * Similarly, systemd-resolved now runs under its own
4324 "systemd-resolve" user with no capabilities remaining.
4326 * Similarly, systemd-bus-proxyd now runs under its own
4327 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4329 * systemd-networkd gained support for setting up "veth"
4330 virtual Ethernet devices for container connectivity, as well
4331 as GRE and VTI tunnels.
4333 * systemd-networkd will no longer automatically attempt to
4334 manually load kernel modules necessary for certain tunnel
4335 transports. Instead, it is assumed the kernel loads them
4336 automatically when required. This only works correctly on
4337 very new kernels. On older kernels, please consider adding
4338 the kernel modules to /etc/modules-load.d/ as a work-around.
4340 * The resolv.conf file systemd-resolved generates has been
4341 moved to /run/systemd/resolve/. If you have a symlink from
4342 /etc/resolv.conf, it might be necessary to correct it.
4344 * Two new service settings, ProtectHome= and ProtectSystem=,
4345 have been added. When enabled, they will make the user data
4346 (such as /home) inaccessible or read-only and the system
4347 (such as /usr) read-only, for specific services. This allows
4348 very light-weight per-service sandboxing to avoid
4349 modifications of user data or system files from
4350 services. These two new switches have been enabled for all
4351 of systemd's long-running services, where appropriate.
4353 * Socket units gained new SocketUser= and SocketGroup=
4354 settings to set the owner user and group of AF_UNIX sockets
4355 and FIFOs in the file system.
4357 * Socket units gained a new RemoveOnStop= setting. If enabled,
4358 all FIFOS and sockets in the file system will be removed
4359 when the specific socket unit is stopped.
4361 * Socket units gained a new Symlinks= setting. It takes a list
4362 of symlinks to create to file system sockets or FIFOs
4363 created by the specific Unix sockets. This is useful to
4364 manage symlinks to socket nodes with the same life-cycle as
4367 * The /dev/log socket and /dev/initctl FIFO have been moved to
4368 /run, and have been replaced by symlinks. This allows
4369 connecting to these facilities even if PrivateDevices=yes is
4370 used for a service (which makes /dev/log itself unavailable,
4371 but /run is left). This also has the benefit of ensuring
4372 that /dev only contains device nodes, directories and
4373 symlinks, and nothing else.
4375 * sd-daemon gained two new calls sd_pid_notify() and
4376 sd_pid_notifyf(). They are similar to sd_notify() and
4377 sd_notifyf(), but allow overriding of the source PID of
4378 notification messages if permissions permit this. This is
4379 useful to send notify messages on behalf of a different
4380 process (for example, the parent process). The
4381 systemd-notify tool has been updated to make use of this
4382 when sending messages (so that notification messages now
4383 originate from the shell script invoking systemd-notify and
4384 not the systemd-notify process itself. This should minimize
4385 a race where systemd fails to associate notification
4386 messages to services when the originating process already
4389 * A new "on-abnormal" setting for Restart= has been added. If
4390 set, it will result in automatic restarts on all "abnormal"
4391 reasons for a process to exit, which includes unclean
4392 signals, core dumps, timeouts and watchdog timeouts, but
4393 does not include clean and unclean exit codes or clean
4394 signals. Restart=on-abnormal is an alternative for
4395 Restart=on-failure for services that shall be able to
4396 terminate and avoid restarts on certain errors, by
4397 indicating so with an unclean exit code. Restart=on-failure
4398 or Restart=on-abnormal is now the recommended setting for
4399 all long-running services.
4401 * If the InaccessibleDirectories= service setting points to a
4402 mount point (or if there are any submounts contained within
4403 it), it is now attempted to completely unmount it, to make
4404 the file systems truly unavailable for the respective
4407 * The ReadOnlyDirectories= service setting and
4408 systemd-nspawn's --read-only parameter are now recursively
4409 applied to all submounts, too.
4411 * Mount units may now be created transiently via the bus APIs.
4413 * The support for SysV and LSB init scripts has been removed
4414 from the systemd daemon itself. Instead, it is now
4415 implemented as a generator that creates native systemd units
4416 from these scripts when needed. This enables us to remove a
4417 substantial amount of legacy code from PID 1, following the
4418 fact that many distributions only ship a very small number
4419 of LSB/SysV init scripts nowadays.
4421 * Privileged Xen (dom0) domains are not considered
4422 virtualization anymore by the virtualization detection
4423 logic. After all, they generally have unrestricted access to
4424 the hardware and usually are used to manage the unprivileged
4427 * systemd-tmpfiles gained a new "C" line type, for copying
4428 files or entire directories.
4430 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4431 lines. So far, they have been non-globbing versions of the
4432 latter, and have thus been redundant. In future, it is
4433 recommended to only use "z". "m" has hence been removed
4434 from the documentation, even though it stays supported.
4436 * A tmpfiles snippet to recreate the most basic structure in
4437 /var has been added. This is enough to create the /var/run →
4438 /run symlink and create a couple of structural
4439 directories. This allows systems to boot up with an empty or
4440 volatile /var. Of course, while with this change, the core OS
4441 now is capable with dealing with a volatile /var, not all
4442 user services are ready for it. However, we hope that sooner
4443 or later, many service daemons will be changed upstream so
4444 that they are able to automatically create their necessary
4445 directories in /var at boot, should they be missing. This is
4446 the first step to allow state-less systems that only require
4447 the vendor image for /usr to boot.
4449 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4450 empty tmpfs instance to a specific directory. This is
4451 particularly useful for making use of the automatic
4452 reconstruction of /var (see above), by passing --tmpfs=/var.
4454 * Access modes specified in tmpfiles snippets may now be
4455 prefixed with "~", which indicates that they shall be masked
4456 by whether the existing file or directory is currently
4457 writable, readable or executable at all. Also, if specified,
4458 the sgid/suid/sticky bits will be masked for all
4461 * A new passive target unit "network-pre.target" has been
4462 added which is useful for services that shall run before any
4463 network is configured, for example firewall scripts.
4465 * The "floppy" group that previously owned the /dev/fd*
4466 devices is no longer used. The "disk" group is now used
4467 instead. Distributions should probably deprecate usage of
4470 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4471 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4472 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4473 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4474 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4475 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4478 — Berlin, 2014-06-11
4482 * A new "systemd-timesyncd" daemon has been added for
4483 synchronizing the system clock across the network. It
4484 implements an SNTP client. In contrast to NTP
4485 implementations such as chrony or the NTP reference server,
4486 this only implements a client side, and does not bother with
4487 the full NTP complexity, focusing only on querying time from
4488 one remote server and synchronizing the local clock to
4489 it. Unless you intend to serve NTP to networked clients or
4490 want to connect to local hardware clocks, this simple NTP
4491 client should be more than appropriate for most
4492 installations. The daemon runs with minimal privileges, and
4493 has been hooked up with networkd to only operate when
4494 network connectivity is available. The daemon saves the
4495 current clock to disk every time a new NTP sync has been
4496 acquired, and uses this to possibly correct the system clock
4497 early at bootup, in order to accommodate for systems that
4498 lack an RTC such as the Raspberry Pi and embedded devices,
4499 and to make sure that time monotonically progresses on these
4500 systems, even if it is not always correct. To make use of
4501 this daemon, a new system user and group "systemd-timesync"
4502 needs to be created on installation of systemd.
4504 * The queue "seqnum" interface of libudev has been disabled, as
4505 it was generally incompatible with device namespacing as
4506 sequence numbers of devices go "missing" if the devices are
4507 part of a different namespace.
4509 * "systemctl list-timers" and "systemctl list-sockets" gained
4510 a --recursive switch for showing units of these types also
4511 for all local containers, similar in style to the already
4512 supported --recursive switch for "systemctl list-units".
4514 * A new RebootArgument= setting has been added for service
4515 units, which may be used to specify a kernel reboot argument
4516 to use when triggering reboots with StartLimitAction=.
4518 * A new FailureAction= setting has been added for service
4519 units which may be used to specify an operation to trigger
4520 when a service fails. This works similarly to
4521 StartLimitAction=, but unlike it, controls what is done
4522 immediately rather than only after several attempts to
4523 restart the service in question.
4525 * hostnamed got updated to also expose the kernel name,
4526 release, and version on the bus. This is useful for
4527 executing commands like hostnamectl with the -H switch.
4528 systemd-analyze makes use of this to properly display
4529 details when running non-locally.
4531 * The bootchart tool can now show cgroup information in the
4532 graphs it generates.
4534 * The CFS CPU quota cgroup attribute is now exposed for
4535 services. The new CPUQuota= switch has been added for this
4536 which takes a percentage value. Setting this will have the
4537 result that a service may never get more CPU time than the
4538 specified percentage, even if the machine is otherwise idle.
4540 * systemd-networkd learned IPIP and SIT tunnel support.
4542 * LSB init scripts exposing a dependency on $network will now
4543 get a dependency on network-online.target rather than simply
4544 network.target. This should bring LSB handling closer to
4545 what it was on SysV systems.
4547 * A new fsck.repair= kernel option has been added to control
4548 how fsck shall deal with unclean file systems at boot.
4550 * The (.ini) configuration file parser will now silently
4551 ignore sections whose name begins with "X-". This may be
4552 used to maintain application-specific extension sections in unit
4555 * machined gained a new API to query the IP addresses of
4556 registered containers. "machinectl status" has been updated
4557 to show these addresses in its output.
4559 * A new call sd_uid_get_display() has been added to the
4560 sd-login APIs for querying the "primary" session of a
4561 user. The "primary" session of the user is elected from the
4562 user's sessions and generally a graphical session is
4563 preferred over a text one.
4565 * A minimal systemd-resolved daemon has been added. It
4566 currently simply acts as a companion to systemd-networkd and
4567 manages resolv.conf based on per-interface DNS
4568 configuration, possibly supplied via DHCP. In the long run
4569 we hope to extend this into a local DNSSEC enabled DNS and
4572 * The systemd-networkd-wait-online tool is now enabled by
4573 default. It will delay network-online.target until a network
4574 connection has been configured. The tool primarily integrates
4575 with networkd, but will also make a best effort to make sense
4576 of network configuration performed in some other way.
4578 * Two new service options StartupCPUShares= and
4579 StartupBlockIOWeight= have been added that work similarly to
4580 CPUShares= and BlockIOWeight= however only apply during
4581 system startup. This is useful to prioritize certain services
4582 differently during bootup than during normal runtime.
4584 * hostnamed has been changed to prefer the statically
4585 configured hostname in /etc/hostname (unless set to
4586 'localhost' or empty) over any dynamic one supplied by
4587 dhcp. With this change, the rules for picking the hostname
4588 match more closely the rules of other configuration settings
4589 where the local administrator's configuration in /etc always
4590 overrides any other settings.
4592 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4593 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4594 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4595 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4596 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4597 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4598 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4599 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4600 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4601 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4602 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4603 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4604 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4605 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4606 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4607 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4610 — Beijing, 2014-05-28
4614 * When restoring the screen brightness at boot, stay away from
4615 the darkest setting or from the lowest 5% of the available
4616 range, depending on which is the larger value of both. This
4617 should effectively protect the user from rebooting into a
4618 black screen, should the brightness have been set to minimum
4621 * sd-login gained a new sd_machine_get_class() call to
4622 determine the class ("vm" or "container") of a machine
4623 registered with machined.
4625 * sd-login gained new calls
4626 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4627 to query the identity of the peer of a local AF_UNIX
4628 connection. They operate similarly to their sd_pid_get_xyz()
4631 * PID 1 will now maintain a system-wide system state engine
4632 with the states "starting", "running", "degraded",
4633 "maintenance", "stopping". These states are bound to system
4634 startup, normal runtime, runtime with at least one failed
4635 service, rescue/emergency mode and system shutdown. This
4636 state is shown in the "systemctl status" output when no unit
4637 name is passed. It is useful to determine system state, in
4638 particularly when doing so for many systems or containers at
4641 * A new command "list-machines" has been added to "systemctl"
4642 that lists all local OS containers and shows their system
4643 state (see above), if systemd runs inside of them.
4645 * systemctl gained a new "-r" switch to recursively enumerate
4646 units on all local containers, when used with the
4647 "list-unit" command (which is the default one that is
4648 executed when no parameters are specified).
4650 * The GPT automatic partition discovery logic will now honour
4651 two GPT partition flags: one may be set on a partition to
4652 cause it to be mounted read-only, and the other may be set
4653 on a partition to ignore it during automatic discovery.
4655 * Two new GPT type UUIDs have been added for automatic root
4656 partition discovery, for 32-bit and 64-bit ARM. This is not
4657 particularly useful for discovering the root directory on
4658 these architectures during bare-metal boots (since UEFI is
4659 not common there), but still very useful to allow booting of
4660 ARM disk images in nspawn with the -i option.
4662 * MAC addresses of interfaces created with nspawn's
4663 --network-interface= switch will now be generated from the
4664 machine name, and thus be stable between multiple invocations
4667 * logind will now automatically remove all IPC objects owned
4668 by a user if she or he fully logs out. This makes sure that
4669 users who are logged out cannot continue to consume IPC
4670 resources. This covers SysV memory, semaphores and message
4671 queues as well as POSIX shared memory and message
4672 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4673 limits. With this functionality, that is corrected. This may
4674 be turned off by using the RemoveIPC= switch of logind.conf.
4676 * The systemd-machine-id-setup and tmpfiles tools gained a
4677 --root= switch to operate on a specific root directory,
4680 * journald can now forward logged messages to the TTYs of all
4681 logged in users ("wall"). This is the default for all
4682 emergency messages now.
4684 * A new tool systemd-journal-remote has been added to stream
4685 journal log messages across the network.
4687 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4688 controller trees are mounted into it. Note that the
4689 directories mounted beneath it are not read-only. This is a
4690 security measure and is particularly useful because glibc
4691 actually includes a search logic to pick any tmpfs it can
4692 find to implement shm_open() if /dev/shm is not available
4693 (which it might very well be in namespaced setups).
4695 * machinectl gained a new "poweroff" command to cleanly power
4696 down a local OS container.
4698 * The PrivateDevices= unit file setting will now also drop the
4699 CAP_MKNOD capability from the capability bound set, and
4700 imply DevicePolicy=closed.
4702 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4703 comprehensively on all long-running systemd services where
4704 this is appropriate.
4706 * systemd-udevd will now run in a disassociated mount
4707 namespace. To mount directories from udev rules, make sure to
4708 pull in mount units via SYSTEMD_WANTS properties.
4710 * The kdbus support gained support for uploading policy into
4711 the kernel. sd-bus gained support for creating "monitoring"
4712 connections that can eavesdrop into all bus communication
4713 for debugging purposes.
4715 * Timestamps may now be specified in seconds since the UNIX
4716 epoch Jan 1st, 1970 by specifying "@" followed by the value
4719 * Native tcpwrap support in systemd has been removed. tcpwrap
4720 is old code, not really maintained anymore and has serious
4721 shortcomings, and better options such as firewalls
4722 exist. For setups that require tcpwrap usage, please
4723 consider invoking your socket-activated service via tcpd,
4724 like on traditional inetd.
4726 * A new system.conf configuration option
4727 DefaultTimerAccuracySec= has been added that controls the
4728 default AccuracySec= setting of .timer units.
4730 * Timer units gained a new WakeSystem= switch. If enabled,
4731 timers configured this way will cause the system to resume
4732 from system suspend (if the system supports that, which most
4735 * Timer units gained a new Persistent= switch. If enabled,
4736 timers configured this way will save to disk when they have
4737 been last triggered. This information is then used on next
4738 reboot to possible execute overdue timer events, that
4739 could not take place because the system was powered off.
4740 This enables simple anacron-like behaviour for timer units.
4742 * systemctl's "list-timers" will now also list the time a
4743 timer unit was last triggered in addition to the next time
4744 it will be triggered.
4746 * systemd-networkd will now assign predictable IPv4LL
4747 addresses to its local interfaces.
4749 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4750 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4751 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4752 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4753 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4754 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4755 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4756 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4759 — Berlin, 2014-03-25
4763 * A new unit file setting RestrictAddressFamilies= has been
4764 added to restrict which socket address families unit
4765 processes gain access to. This takes address family names
4766 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4767 attack surface of services via exotic protocol stacks. This
4768 is built on seccomp system call filters.
4770 * Two new unit file settings RuntimeDirectory= and
4771 RuntimeDirectoryMode= have been added that may be used to
4772 manage a per-daemon runtime directories below /run. This is
4773 an alternative for setting up directory permissions with
4774 tmpfiles snippets, and has the advantage that the runtime
4775 directory's lifetime is bound to the daemon runtime and that
4776 the daemon starts up with an empty directory each time. This
4777 is particularly useful when writing services that drop
4778 privileges using the User= or Group= setting.
4780 * The DeviceAllow= unit setting now supports globbing for
4781 matching against device group names.
4783 * The systemd configuration file system.conf gained new
4784 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4785 DefaultMemoryAccounting= to globally turn on/off accounting
4786 for specific resources (cgroups) for all units. These
4787 settings may still be overridden individually in each unit
4790 * systemd-gpt-auto-generator is now able to discover /srv and
4791 root partitions in addition to /home and swap partitions. It
4792 also supports LUKS-encrypted partitions now. With this in
4793 place, automatic discovery of partitions to mount following
4794 the Discoverable Partitions Specification
4795 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4796 is now a lot more complete. This allows booting without
4797 /etc/fstab and without root= on the kernel command line on
4798 systems prepared appropriately.
4800 * systemd-nspawn gained a new --image= switch which allows
4801 booting up disk images and Linux installations on any block
4802 device that follow the Discoverable Partitions Specification
4803 (see above). This means that installations made with
4804 appropriately updated installers may now be started and
4805 deployed using container managers, completely
4806 unmodified. (We hope that libvirt-lxc will add support for
4807 this feature soon, too.)
4809 * systemd-nspawn gained a new --network-macvlan= setting to
4810 set up a private macvlan interface for the
4811 container. Similarly, systemd-networkd gained a new
4812 Kind=macvlan setting in .netdev files.
4814 * systemd-networkd now supports configuring local addresses
4817 * A new tool systemd-network-wait-online has been added to
4818 synchronously wait for network connectivity using
4821 * The sd-bus.h bus API gained a new sd_bus_track object for
4822 tracking the life-cycle of bus peers. Note that sd-bus.h is
4823 still not a public API though (unless you specify
4824 --enable-kdbus on the configure command line, which however
4825 voids your warranty and you get no API stability guarantee).
4827 * The $XDG_RUNTIME_DIR runtime directories for each user are
4828 now individual tmpfs instances, which has the benefit of
4829 introducing separate pools for each user, with individual
4830 size limits, and thus making sure that unprivileged clients
4831 can no longer negatively impact the system or other users by
4832 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4833 RuntimeDirectorySize= has been introduced that allows
4834 controlling the default size limit for all users. It
4835 defaults to 10% of the available physical memory. This is no
4836 replacement for quotas on tmpfs though (which the kernel
4837 still does not support), as /dev/shm and /tmp are still
4838 shared resources used by both the system and unprivileged
4841 * logind will now automatically turn off automatic suspending
4842 on laptop lid close when more than one display is
4843 connected. This was previously expected to be implemented
4844 individually in desktop environments (such as GNOME),
4845 however has been added to logind now, in order to fix a
4846 boot-time race where a desktop environment might not have
4847 been started yet and thus not been able to take an inhibitor
4848 lock at the time where logind already suspends the system
4849 due to a closed lid.
4851 * logind will now wait at least 30s after each system
4852 suspend/resume cycle, and 3min after system boot before
4853 suspending the system due to a closed laptop lid. This
4854 should give USB docking stations and similar enough time to
4855 be probed and configured after system resume and boot in
4856 order to then act as suspend blocker.
4858 * systemd-run gained a new --property= setting which allows
4859 initialization of resource control properties (and others)
4860 for the created scope or service unit. Example: "systemd-run
4861 --property=BlockIOWeight=10 updatedb" may be used to run
4862 updatedb at a low block IO scheduling weight.
4864 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4865 now also work in --scope mode.
4867 * When systemd is compiled with kdbus support, basic support
4868 for enforced policies is now in place. (Note that enabling
4869 kdbus still voids your warranty and no API compatibility
4872 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4873 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4874 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4875 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4876 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4877 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4878 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4879 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4880 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4881 Zbigniew Jędrzejewski-Szmek
4883 — Berlin, 2014-03-12
4887 * systemd will now relabel /dev after loading the SMACK policy
4888 according to SMACK rules.
4890 * A new unit file option AppArmorProfile= has been added to
4891 set the AppArmor profile for the processes of a unit.
4893 * A new condition check ConditionArchitecture= has been added
4894 to conditionalize units based on the system architecture, as
4895 reported by uname()'s "machine" field.
4897 * systemd-networkd now supports matching on the system
4898 virtualization, architecture, kernel command line, host name
4901 * logind is now a lot more aggressive when suspending the
4902 machine due to a closed laptop lid. Instead of acting only
4903 on the lid close action, it will continuously watch the lid
4904 status and act on it. This is useful for laptops where the
4905 power button is on the outside of the chassis so that it can
4906 be reached without opening the lid (such as the Lenovo
4907 Yoga). On those machines, logind will now immediately
4908 re-suspend the machine if the power button has been
4909 accidentally pressed while the laptop was suspended and in a
4910 backpack or similar.
4912 * logind will now watch SW_DOCK switches and inhibit reaction
4913 to the lid switch if it is pressed. This means that logind
4914 will not suspend the machine anymore if the lid is closed
4915 and the system is docked, if the laptop supports SW_DOCK
4916 notifications via the input layer. Note that ACPI docking
4917 stations do not generate this currently. Also note that this
4918 logic is usually not fully sufficient and Desktop
4919 Environments should take a lid switch inhibitor lock when an
4920 external display is connected, as systemd will not watch
4923 * nspawn will now make use of the devices cgroup controller by
4924 default, and only permit creation of and access to the usual
4925 API device nodes like /dev/null or /dev/random, as well as
4926 access to (but not creation of) the pty devices.
4928 * We will now ship a default .network file for
4929 systemd-networkd that automatically configures DHCP for
4930 network interfaces created by nspawn's --network-veth or
4931 --network-bridge= switches.
4933 * systemd will now understand the usual M, K, G, T suffixes
4934 according to SI conventions (i.e. to the base 1000) when
4935 referring to throughput and hardware metrics. It will stay
4936 with IEC conventions (i.e. to the base 1024) for software
4937 metrics, according to what is customary according to
4938 Wikipedia. We explicitly document which base applies for
4939 each configuration option.
4941 * The DeviceAllow= setting in unit files now supports a syntax
4942 to whitelist an entire group of devices node majors at once,
4943 based on the /proc/devices listing. For example, with the
4944 string "char-pts", it is now possible to whitelist all
4945 current and future pseudo-TTYs at once.
4947 * sd-event learned a new "post" event source. Event sources of
4948 this type are triggered by the dispatching of any event
4949 source of a type that is not "post". This is useful for
4950 implementing clean-up and check event sources that are
4951 triggered by other work being done in the program.
4953 * systemd-networkd is no longer statically enabled, but uses
4954 the usual [Install] sections so that it can be
4955 enabled/disabled using systemctl. It still is enabled by
4958 * When creating a veth interface pair with systemd-nspawn, the
4959 host side will now be prefixed with "vb-" if
4960 --network-bridge= is used, and with "ve-" if --network-veth
4961 is used. This way, it is easy to distinguish these cases on
4962 the host, for example to apply different configuration to
4963 them with systemd-networkd.
4965 * The compatibility libraries for libsystemd-journal.so,
4966 libsystem-id128.so, libsystemd-login.so and
4967 libsystemd-daemon.so do not make use of IFUNC
4968 anymore. Instead, we now build libsystemd.so multiple times
4969 under these alternative names. This means that the footprint
4970 is drastically increased, but given that these are
4971 transitional compatibility libraries, this should not matter
4972 much. This change has been made necessary to support the ARM
4973 platform for these compatibility libraries, as the ARM
4974 toolchain is not really at the same level as the toolchain
4975 for other architectures like x86 and does not support
4976 IFUNC. Please make sure to use --enable-compat-libs only
4977 during a transitional period!
4979 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4980 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4981 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4982 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4983 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4984 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4985 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4986 Zbigniew Jędrzejewski-Szmek
4988 — Berlin, 2014-02-24
4992 * A new component "systemd-networkd" has been added that can
4993 be used to configure local network interfaces statically or
4994 via DHCP. It is capable of bringing up bridges, VLANs, and
4995 bonding. Currently, no hook-ups for interactive network
4996 configuration are provided. Use this for your initrd,
4997 container, embedded, or server setup if you need a simple,
4998 yet powerful, network configuration solution. This
4999 configuration subsystem is quite nifty, as it allows wildcard
5000 hotplug matching in interfaces. For example, with a single
5001 configuration snippet, you can configure that all Ethernet
5002 interfaces showing up are automatically added to a bridge,
5003 or similar. It supports link-sensing and more.
5005 * A new tool "systemd-socket-proxyd" has been added which can
5006 act as a bidirectional proxy for TCP sockets. This is
5007 useful for adding socket activation support to services that
5008 do not actually support socket activation, including virtual
5009 machines and the like.
5011 * Add a new tool to save/restore rfkill state on
5014 * Save/restore state of keyboard backlights in addition to
5015 display backlights on shutdown/boot.
5017 * udev learned a new SECLABEL{} construct to label device
5018 nodes with a specific security label when they appear. For
5019 now, only SECLABEL{selinux} is supported, but the syntax is
5020 prepared for additional security frameworks.
5022 * udev gained a new scheme to configure link-level attributes
5023 from files in /etc/systemd/network/*.link. These files can
5024 match against MAC address, device path, driver name and type,
5025 and will apply attributes like the naming policy, link speed,
5026 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5027 address assignment policy (randomized, ...).
5029 * The configuration of network interface naming rules for
5030 "permanent interface names" has changed: a new NamePolicy=
5031 setting in the [Link] section of .link files determines the
5032 priority of possible naming schemes (onboard, slot, MAC,
5033 path). The default value of this setting is determined by
5034 /usr/lib/net/links/99-default.link. Old
5035 80-net-name-slot.rules udev configuration file has been
5036 removed, so local configuration overriding this file should
5037 be adapted to override 99-default.link instead.
5039 * When the User= switch is used in a unit file, also
5040 initialize $SHELL= based on the user database entry.
5042 * systemd no longer depends on libdbus. All communication is
5043 now done with sd-bus, systemd's low-level bus library
5046 * kdbus support has been added to PID 1 itself. When kdbus is
5047 enabled, this causes PID 1 to set up the system bus and
5048 enable support for a new ".busname" unit type that
5049 encapsulates bus name activation on kdbus. It works a little
5050 bit like ".socket" units, except for bus names. A new
5051 generator has been added that converts classic dbus1 service
5052 activation files automatically into native systemd .busname
5055 * sd-bus: add a light-weight vtable implementation that allows
5056 defining objects on the bus with a simple static const
5057 vtable array of its methods, signals and properties.
5059 * systemd will not generate or install static dbus
5060 introspection data anymore to /usr/share/dbus-1/interfaces,
5061 as the precise format of these files is unclear, and
5062 nothing makes use of it.
5064 * A proxy daemon is now provided to proxy clients connecting
5065 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5066 compatibility with classic D-Bus.
5068 * A bus driver implementation has been added that supports the
5069 classic D-Bus bus driver calls on kdbus, also for
5070 compatibility purposes.
5072 * A new API "sd-event.h" has been added that implements a
5073 minimal event loop API built around epoll. It provides a
5074 couple of features that direct epoll usage is lacking:
5075 prioritization of events, scales to large numbers of timer
5076 events, per-event timer slack (accuracy), system-wide
5077 coalescing of timer events, exit handlers, watchdog
5078 supervision support using systemd's sd_notify() API, child
5081 * A new API "sd-rntl.h" has been added that provides an API
5082 around the route netlink interface of the kernel, similar in
5083 style to "sd-bus.h".
5085 * A new API "sd-dhcp-client.h" has been added that provides a
5086 small DHCPv4 client-side implementation. This is used by
5089 * There is a new kernel command line option
5090 "systemd.restore_state=0|1". When set to "0", none of the
5091 systemd tools will restore saved runtime state to hardware
5092 devices. More specifically, the rfkill and backlight states
5095 * The FsckPassNo= compatibility option in mount/service units
5096 has been removed. The fstab generator will now add the
5097 necessary dependencies automatically, and does not require
5098 PID1's support for that anymore.
5100 * journalctl gained a new switch, --list-boots, that lists
5101 recent boots with their times and boot IDs.
5103 * The various tools like systemctl, loginctl, timedatectl,
5104 busctl, systemd-run, ... have gained a new switch "-M" to
5105 connect to a specific, local OS container (as direct
5106 connection, without requiring SSH). This works on any
5107 container that is registered with machined, such as those
5108 created by libvirt-lxc or nspawn.
5110 * systemd-run and systemd-analyze also gained support for "-H"
5111 to connect to remote hosts via SSH. This is particularly
5112 useful for systemd-run because it enables queuing of jobs
5113 onto remote systems.
5115 * machinectl gained a new command "login" to open a getty
5116 login in any local container. This works with any container
5117 that is registered with machined (such as those created by
5118 libvirt-lxc or nspawn), and which runs systemd inside.
5120 * machinectl gained a new "reboot" command that may be used to
5121 trigger a reboot on a specific container that is registered
5122 with machined. This works on any container that runs an init
5123 system of some kind.
5125 * systemctl gained a new "list-timers" command to print a nice
5126 listing of installed timer units with the times they elapse
5129 * Alternative reboot() parameters may now be specified on the
5130 "systemctl reboot" command line and are passed to the
5131 reboot() system call.
5133 * systemctl gained a new --job-mode= switch to configure the
5134 mode to queue a job with. This is a more generic version of
5135 --fail, --irreversible, and --ignore-dependencies, which are
5136 still available but not advertised anymore.
5138 * /etc/systemd/system.conf gained new settings to configure
5139 various default timeouts of units, as well as the default
5140 start limit interval and burst. These may still be overridden
5143 * PID1 will now export on the bus profile data of the security
5144 policy upload process (such as the SELinux policy upload to
5147 * journald: when forwarding logs to the console, include
5148 timestamps (following the setting in
5149 /sys/module/printk/parameters/time).
5151 * OnCalendar= in timer units now understands the special
5152 strings "yearly" and "annually". (Both are equivalent)
5154 * The accuracy of timer units is now configurable with the new
5155 AccuracySec= setting. It defaults to 1min.
5157 * A new dependency type JoinsNamespaceOf= has been added that
5158 allows running two services within the same /tmp and network
5159 namespace, if PrivateNetwork= or PrivateTmp= are used.
5161 * A new command "cat" has been added to systemctl. It outputs
5162 the original unit file of a unit, and concatenates the
5163 contents of additional "drop-in" unit file snippets, so that
5164 the full configuration is shown.
5166 * systemctl now supports globbing on the various "list-xyz"
5167 commands, like "list-units" or "list-sockets", as well as on
5168 those commands which take multiple unit names.
5170 * journalctl's --unit= switch gained support for globbing.
5172 * All systemd daemons now make use of the watchdog logic so
5173 that systemd automatically notices when they hang.
5175 * If the $container_ttys environment variable is set,
5176 getty-generator will automatically spawn a getty for each
5177 listed tty. This is useful for container managers to request
5178 login gettys to be spawned on as many ttys as needed.
5180 * %h, %s, %U specifier support is not available anymore when
5181 used in unit files for PID 1. This is because NSS calls are
5182 not safe from PID 1. They stay available for --user
5183 instances of systemd, and as special case for the root user.
5185 * loginctl gained a new "--no-legend" switch to turn off output
5188 * The "sd-login.h" API gained three new calls:
5189 sd_session_is_remote(), sd_session_get_remote_user(),
5190 sd_session_get_remote_host() to query information about
5193 * The udev hardware database now also carries vendor/product
5194 information of SDIO devices.
5196 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5197 determine whether watchdog notifications are requested by
5200 * Socket-activated per-connection services now include a
5201 short description of the connection parameters in the
5204 * tmpfiles gained a new "--boot" option. When this is not used,
5205 only lines where the command character is not suffixed with
5206 "!" are executed. When this option is specified, those
5207 options are executed too. This partitions tmpfiles
5208 directives into those that can be safely executed at any
5209 time, and those which should be run only at boot (for
5210 example, a line that creates /run/nologin).
5212 * A new API "sd-resolve.h" has been added which provides a simple
5213 asynchronous wrapper around glibc NSS host name resolution
5214 calls, such as getaddrinfo(). In contrast to glibc's
5215 getaddrinfo_a(), it does not use signals. In contrast to most
5216 other asynchronous name resolution libraries, this one does
5217 not reimplement DNS, but reuses NSS, so that alternate
5218 host name resolution systems continue to work, such as mDNS,
5219 LDAP, etc. This API is based on libasyncns, but it has been
5220 cleaned up for inclusion in systemd.
5222 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5223 "sd-daemon.h" are no longer found in individual libraries
5224 libsystemd-journal.so, libsystemd-login.so,
5225 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5226 merged them into a single library, libsystemd.so, which
5227 provides all symbols. The reason for this is cyclic
5228 dependencies, as these libraries tend to use each other's
5229 symbols. So far, we have managed to workaround that by linking
5230 a copy of a good part of our code into each of these
5231 libraries again and again, which, however, makes certain
5232 things hard to do, like sharing static variables. Also, it
5233 substantially increases footprint. With this change, there
5234 is only one library for the basic APIs systemd
5235 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5236 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5237 library as well, however are subject to the --enable-kdbus
5238 switch (see below). Note that "sd-dhcp-client.h" is not part
5239 of this library (this is because it only consumes, never
5240 provides, services of/to other APIs). To make the transition
5241 easy from the separate libraries to the unified one, we
5242 provide the --enable-compat-libs compile-time switch which
5243 will generate stub libraries that are compatible with the
5244 old ones but redirect all calls to the new one.
5246 * All of the kdbus logic and the new APIs "sd-bus.h",
5247 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5248 and "sd-utf8.h" are compile-time optional via the
5249 "--enable-kdbus" switch, and they are not compiled in by
5250 default. To make use of kdbus, you have to explicitly enable
5251 the switch. Note however, that neither the kernel nor the
5252 userspace API for all of this is considered stable yet. We
5253 want to maintain the freedom to still change the APIs for
5254 now. By specifying this build-time switch, you acknowledge
5255 that you are aware of the instability of the current
5258 * Also, note that while kdbus is pretty much complete,
5259 it lacks one thing: proper policy support. This means you
5260 can build a fully working system with all features; however,
5261 it will be highly insecure. Policy support will be added in
5262 one of the next releases, at the same time that we will
5263 declare the APIs stable.
5265 * When the kernel command line argument "kdbus" is specified,
5266 systemd will automatically load the kdbus.ko kernel module. At
5267 this stage of development, it is only useful for testing kdbus
5268 and should not be used in production. Note: if "--enable-kdbus"
5269 is specified, and the kdbus.ko kernel module is available, and
5270 "kdbus" is added to the kernel command line, the entire system
5271 runs with kdbus instead of dbus-daemon, with the above mentioned
5272 problem of missing the system policy enforcement. Also a future
5273 version of kdbus.ko or a newer systemd will not be compatible with
5274 each other, and will unlikely be able to boot the machine if only
5275 one of them is updated.
5277 * systemctl gained a new "import-environment" command which
5278 uploads the caller's environment (or parts thereof) into the
5279 service manager so that it is inherited by services started
5280 by the manager. This is useful to upload variables like
5281 $DISPLAY into the user service manager.
5283 * A new PrivateDevices= switch has been added to service units
5284 which allows running a service with a namespaced /dev
5285 directory that does not contain any device nodes for
5286 physical devices. More specifically, it only includes devices
5287 such as /dev/null, /dev/urandom, and /dev/zero which are API
5290 * logind has been extended to support behaviour like VT
5291 switching on seats that do not support a VT. This makes
5292 multi-session available on seats that are not the first seat
5293 (seat0), and on systems where kernel support for VTs has
5294 been disabled at compile-time.
5296 * If a process holds a delay lock for system sleep or shutdown
5297 and fails to release it in time, we will now log its
5298 identity. This makes it easier to identify processes that
5299 cause slow suspends or power-offs.
5301 * When parsing /etc/crypttab, support for a new key-slot=
5302 option as supported by Debian is added. It allows indicating
5303 which LUKS slot to use on disk, speeding up key loading.
5305 * The sd_journald_sendv() API call has been checked and
5306 officially declared to be async-signal-safe so that it may
5307 be invoked from signal handlers for logging purposes.
5309 * Boot-time status output is now enabled automatically after a
5310 short timeout if boot does not progress, in order to give
5311 the user an indication what she or he is waiting for.
5313 * The boot-time output has been improved to show how much time
5314 remains until jobs expire.
5316 * The KillMode= switch in service units gained a new possible
5317 value "mixed". If set, and the unit is shut down, then the
5318 initial SIGTERM signal is sent only to the main daemon
5319 process, while the following SIGKILL signal is sent to
5320 all remaining processes of the service.
5322 * When a scope unit is registered, a new property "Controller"
5323 may be set. If set to a valid bus name, systemd will send a
5324 RequestStop() signal to this name when it would like to shut
5325 down the scope. This may be used to hook manager logic into
5326 the shutdown logic of scope units. Also, scope units may now
5327 be put in a special "abandoned" state, in which case the
5328 manager process which created them takes no further
5329 responsibilities for it.
5331 * When reading unit files, systemd will now verify
5332 the access mode of these files, and warn about certain
5333 suspicious combinations. This has been added to make it
5334 easier to track down packaging bugs where unit files are
5335 marked executable or world-writable.
5337 * systemd-nspawn gained a new "--setenv=" switch to set
5338 container-wide environment variables. The similar option in
5339 systemd-activate was renamed from "--environment=" to
5340 "--setenv=" for consistency.
5342 * systemd-nspawn has been updated to create a new kdbus domain
5343 for each container that is invoked, thus allowing each
5344 container to have its own set of system and user buses,
5345 independent of the host.
5347 * systemd-nspawn gained a new --drop-capability= switch to run
5348 the container with less capabilities than the default. Both
5349 --drop-capability= and --capability= now take the special
5350 string "all" for dropping or keeping all capabilities.
5352 * systemd-nspawn gained new switches for executing containers
5353 with specific SELinux labels set.
5355 * systemd-nspawn gained a new --quiet switch to not generate
5356 any additional output but the container's own console
5359 * systemd-nspawn gained a new --share-system switch to run a
5360 container without PID namespacing enabled.
5362 * systemd-nspawn gained a new --register= switch to control
5363 whether the container is registered with systemd-machined or
5364 not. This is useful for containers that do not run full
5365 OS images, but only specific apps.
5367 * systemd-nspawn gained a new --keep-unit which may be used
5368 when invoked as the only program from a service unit, and
5369 results in registration of the unit service itself in
5370 systemd-machined, instead of a newly opened scope unit.
5372 * systemd-nspawn gained a new --network-interface= switch for
5373 moving arbitrary interfaces to the container. The new
5374 --network-veth switch creates a virtual Ethernet connection
5375 between host and container. The new --network-bridge=
5376 switch then allows assigning the host side of this virtual
5377 Ethernet connection to a bridge device.
5379 * systemd-nspawn gained a new --personality= switch for
5380 setting the kernel personality for the container. This is
5381 useful when running a 32-bit container on a 64-bit host. A
5382 similar option Personality= is now also available for service
5385 * logind will now also track a "Desktop" identifier for each
5386 session which encodes the desktop environment of it. This is
5387 useful for desktop environments that want to identify
5388 multiple running sessions of itself easily.
5390 * A new SELinuxContext= setting for service units has been
5391 added that allows setting a specific SELinux execution
5392 context for a service.
5394 * Most systemd client tools will now honour $SYSTEMD_LESS for
5395 settings of the "less" pager. By default, these tools will
5396 override $LESS to allow certain operations to work, such as
5397 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5398 influence this logic.
5400 * systemd's "seccomp" hook-up has been changed to make use of
5401 the libseccomp library instead of using its own
5402 implementation. This has benefits for portability among
5405 * For usage together with SystemCallFilter=, a new
5406 SystemCallErrorNumber= setting has been introduced that
5407 allows configuration of a system error number to be returned
5408 on filtered system calls, instead of immediately killing the
5409 process. Also, SystemCallArchitectures= has been added to
5410 limit access to system calls of a particular architecture
5411 (in order to turn off support for unused secondary
5412 architectures). There is also a global
5413 SystemCallArchitectures= setting in system.conf now to turn
5414 off support for non-native system calls system-wide.
5416 * systemd requires a kernel with a working name_to_handle_at(),
5417 please see the kernel config requirements in the README file.
5419 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5420 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5421 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5422 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5423 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5424 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5425 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5426 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5427 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5428 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5429 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5430 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5431 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5432 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5433 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5434 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5435 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5436 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5437 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5438 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5439 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5440 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5441 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5442 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5444 — Berlin, 2014-02-20
5448 * logind has gained support for facilitating privileged input
5449 and drm device access for unprivileged clients. This work is
5450 useful to allow Wayland display servers (and similar
5451 programs, such as kmscon) to run under the user's ID and
5452 access input and drm devices which are normally
5453 protected. When this is used (and the kernel is new enough)
5454 logind will "mute" IO on the file descriptors passed to
5455 Wayland as long as it is in the background and "unmute" it
5456 if it returns into the foreground. This allows secure
5457 session switching without allowing background sessions to
5458 eavesdrop on input and display data. This also introduces
5459 session switching support if VT support is turned off in the
5460 kernel, and on seats that are not seat0.
5462 * A new kernel command line option luks.options= is understood
5463 now which allows specifying LUKS options for usage for LUKS
5464 encrypted partitions specified with luks.uuid=.
5466 * tmpfiles.d(5) snippets may now use specifier expansion in
5467 path names. More specifically %m, %b, %H, %v, are now
5468 replaced by the local machine id, boot id, hostname, and
5469 kernel version number.
5471 * A new tmpfiles.d(5) command "m" has been introduced which
5472 may be used to change the owner/group/access mode of a file
5473 or directory if it exists, but do nothing if it does not.
5475 * This release removes high-level support for the
5476 MemorySoftLimit= cgroup setting. The underlying kernel
5477 cgroup attribute memory.soft_limit= is currently badly
5478 designed and likely to be removed from the kernel API in its
5479 current form, hence we should not expose it for now.
5481 * The memory.use_hierarchy cgroup attribute is now enabled for
5482 all cgroups systemd creates in the memory cgroup
5483 hierarchy. This option is likely to be come the built-in
5484 default in the kernel anyway, and the non-hierarchical mode
5485 never made much sense in the intrinsically hierarchical
5488 * A new field _SYSTEMD_SLICE= is logged along with all journal
5489 messages containing the slice a message was generated
5490 from. This is useful to allow easy per-customer filtering of
5491 logs among other things.
5493 * systemd-journald will no longer adjust the group of journal
5494 files it creates to the "systemd-journal" group. Instead we
5495 rely on the journal directory to be owned by the
5496 "systemd-journal" group, and its setgid bit set, so that the
5497 kernel file system layer will automatically enforce that
5498 journal files inherit this group assignment. The reason for
5499 this change is that we cannot allow NSS look-ups from
5500 journald which would be necessary to resolve
5501 "systemd-journal" to a numeric GID, because this might
5502 create deadlocks if NSS involves synchronous queries to
5503 other daemons (such as nscd, or sssd) which in turn are
5504 logging clients of journald and might block on it, which
5505 would then dead lock. A tmpfiles.d(5) snippet included in
5506 systemd will make sure the setgid bit and group are
5507 properly set on the journal directory if it exists on every
5508 boot. However, we recommend adjusting it manually after
5509 upgrades too (or from RPM scriptlets), so that the change is
5510 not delayed until next reboot.
5512 * Backlight and random seed files in /var/lib/ have moved into
5513 the /var/lib/systemd/ directory, in order to centralize all
5514 systemd generated files in one directory.
5516 * Boot time performance measurements (as displayed by
5517 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5518 performance information if that's available to determine how
5519 much time BIOS and boot loader initialization required. With
5520 a sufficiently new BIOS you hence no longer need to boot
5521 with Gummiboot to get access to such information.
5523 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5524 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5525 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5526 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5527 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5528 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5529 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5531 — Berlin, 2013-10-02
5535 * The Restart= option for services now understands a new
5536 on-watchdog setting, which will restart the service
5537 automatically if the service stops sending out watchdog keep
5538 alive messages (as configured with WatchdogSec=).
5540 * The getty generator (which is responsible for bringing up a
5541 getty on configured serial consoles) will no longer only
5542 start a getty on the primary kernel console but on all
5543 others, too. This makes the order in which console= is
5544 specified on the kernel command line less important.
5546 * libsystemd-logind gained a new sd_session_get_vt() call to
5547 retrieve the VT number of a session.
5549 * If the option "tries=0" is set for an entry of /etc/crypttab
5550 its passphrase is queried indefinitely instead of any
5551 maximum number of tries.
5553 * If a service with a configure PID file terminates its PID
5554 file will now be removed automatically if it still exists
5555 afterwards. This should put an end to stale PID files.
5557 * systemd-run will now also take relative binary path names
5558 for execution and no longer insists on absolute paths.
5560 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5561 paths that are optionally prefixed with "-" to indicate that
5562 it should not be considered a failure if they do not exist.
5564 * journalctl -o (and similar commands) now understands a new
5565 output mode "short-precise", it is similar to "short" but
5566 shows timestamps with usec accuracy.
5568 * The option "discard" (as known from Debian) is now
5569 synonymous to "allow-discards" in /etc/crypttab. In fact,
5570 "discard" is preferred now (since it is easier to remember
5573 * Some licensing clean-ups were made, so that more code is now
5574 LGPL-2.1 licensed than before.
5576 * A minimal tool to save/restore the display backlight
5577 brightness across reboots has been added. It will store the
5578 backlight setting as late as possible at shutdown, and
5579 restore it as early as possible during reboot.
5581 * A logic to automatically discover and enable home and swap
5582 partitions on GPT disks has been added. With this in place
5583 /etc/fstab becomes optional for many setups as systemd can
5584 discover certain partitions located on the root disk
5585 automatically. Home partitions are recognized under their
5586 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5587 partitions are recognized under their GPT type ID
5588 0657fd6da4ab43c484e50933c84b4f4f.
5590 * systemd will no longer pass any environment from the kernel
5591 or initrd to system services. If you want to set an
5592 environment for all services, do so via the kernel command
5593 line systemd.setenv= assignment.
5595 * The systemd-sysctl tool no longer natively reads the file
5596 /etc/sysctl.conf. If desired, the file should be symlinked
5597 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5598 legacy support by a symlink rather than built-in code, it
5599 also makes the otherwise hidden order of application of the
5600 different files visible. (Note that this partly reverts to a
5601 pre-198 application order of sysctl knobs!)
5603 * The "systemctl set-log-level" and "systemctl dump" commands
5604 have been moved to systemd-analyze.
5606 * systemd-run learned the new --remain-after-exit switch,
5607 which causes the scope unit not to be cleaned up
5608 automatically after the process terminated.
5610 * tmpfiles learned a new --exclude-prefix= switch to exclude
5611 certain paths from operation.
5613 * journald will now automatically flush all messages to disk
5614 as soon as a message at the log level CRIT, ALERT or EMERG
5617 Contributions from: Andrew Cook, Brandon Philips, Christian
5618 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5619 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5620 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5621 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5622 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5623 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5624 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5625 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5626 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5627 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5628 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5629 William Giokas, Zbigniew Jędrzejewski-Szmek
5631 — Berlin, 2013-09-13
5635 * The documentation has been updated to cover the various new
5636 concepts introduced with 205.
5638 * Unit files now understand the new %v specifier which
5639 resolves to the kernel version string as returned by "uname
5642 * systemctl now supports filtering the unit list output by
5643 load state, active state and sub state, using the new
5646 * "systemctl status" will now show the results of the
5647 condition checks (like ConditionPathExists= and similar) of
5648 the last start attempts of the unit. They are also logged to
5651 * "journalctl -b" may now be used to look for boot output of a
5652 specific boot. Try "journalctl -b -1" for the previous boot,
5653 but the syntax is substantially more powerful.
5655 * "journalctl --show-cursor" has been added which prints the
5656 cursor string the last shown log line. This may then be used
5657 with the new "journalctl --after-cursor=" switch to continue
5658 browsing logs from that point on.
5660 * "journalctl --force" may now be used to force regeneration
5663 * Creation of "dead" device nodes has been moved from udev
5664 into kmod and tmpfiles. Previously, udev would read the kmod
5665 databases to pre-generate dead device nodes based on meta
5666 information contained in kernel modules, so that these would
5667 be auto-loaded on access rather then at boot. As this
5668 does not really have much to do with the exposing actual
5669 kernel devices to userspace this has always been slightly
5670 alien in the udev codebase. Following the new scheme kmod
5671 will now generate a runtime snippet for tmpfiles from the
5672 module meta information and it now is tmpfiles' job to the
5673 create the nodes. This also allows overriding access and
5674 other parameters for the nodes using the usual tmpfiles
5675 facilities. As side effect this allows us to remove the
5676 CAP_SYS_MKNOD capability bit from udevd entirely.
5678 * logind's device ACLs may now be applied to these "dead"
5679 devices nodes too, thus finally allowing managed access to
5680 devices such as /dev/snd/sequencer without loading the
5681 backing module right-away.
5683 * A new RPM macro has been added that may be used to apply
5684 tmpfiles configuration during package installation.
5686 * systemd-detect-virt and ConditionVirtualization= now can
5687 detect User-Mode-Linux machines (UML).
5689 * journald will now implicitly log the effective capabilities
5690 set of processes in the message metadata.
5692 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5694 * The initrd interface has been simplified (more specifically,
5695 support for passing performance data via environment
5696 variables and fsck results via files in /run has been
5697 removed). These features were non-essential, and are
5698 nowadays available in a much nicer way by having systemd in
5699 the initrd serialize its state and have the hosts systemd
5700 deserialize it again.
5702 * The udev "keymap" data files and tools to apply keyboard
5703 specific mappings of scan to key codes, and force-release
5704 scan code lists have been entirely replaced by a udev
5705 "keyboard" builtin and a hwdb data file.
5707 * systemd will now honour the kernel's "quiet" command line
5708 argument also during late shutdown, resulting in a
5709 completely silent shutdown when used.
5711 * There's now an option to control the SO_REUSEPORT socket
5712 option in .socket units.
5714 * Instance units will now automatically get a per-template
5715 subslice of system.slice unless something else is explicitly
5716 configured. For example, instances of sshd@.service will now
5717 implicitly be placed in system-sshd.slice rather than
5718 system.slice as before.
5720 * Test coverage support may now be enabled at build time.
5722 Contributions from: Dave Reisner, Frederic Crozat, Harald
5723 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5724 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5725 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5726 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5727 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5728 Giokas, Zbigniew Jędrzejewski-Szmek
5730 — Berlin, 2013-07-23
5734 * Two new unit types have been introduced:
5736 Scope units are very similar to service units, however, are
5737 created out of pre-existing processes — instead of PID 1
5738 forking off the processes. By using scope units it is
5739 possible for system services and applications to group their
5740 own child processes (worker processes) in a powerful way
5741 which then maybe used to organize them, or kill them
5742 together, or apply resource limits on them.
5744 Slice units may be used to partition system resources in an
5745 hierarchical fashion and then assign other units to them. By
5746 default there are now three slices: system.slice (for all
5747 system services), user.slice (for all user sessions),
5748 machine.slice (for VMs and containers).
5750 Slices and scopes have been introduced primarily in
5751 context of the work to move cgroup handling to a
5752 single-writer scheme, where only PID 1
5753 creates/removes/manages cgroups.
5755 * There's a new concept of "transient" units. In contrast to
5756 normal units these units are created via an API at runtime,
5757 not from configuration from disk. More specifically this
5758 means it is now possible to run arbitrary programs as
5759 independent services, with all execution parameters passed
5760 in via bus APIs rather than read from disk. Transient units
5761 make systemd substantially more dynamic then it ever was,
5762 and useful as a general batch manager.
5764 * logind has been updated to make use of scope and slice units
5765 for managing user sessions. As a user logs in he will get
5766 his own private slice unit, to which all sessions are added
5767 as scope units. We also added support for automatically
5768 adding an instance of user@.service for the user into the
5769 slice. Effectively logind will no longer create cgroup
5770 hierarchies on its own now, it will defer entirely to PID 1
5771 for this by means of scope, service and slice units. Since
5772 user sessions this way become entities managed by PID 1
5773 the output of "systemctl" is now a lot more comprehensive.
5775 * A new mini-daemon "systemd-machined" has been added which
5776 may be used by virtualization managers to register local
5777 VMs/containers. nspawn has been updated accordingly, and
5778 libvirt will be updated shortly. machined will collect a bit
5779 of meta information about the VMs/containers, and assign
5780 them their own scope unit (see above). The collected
5781 meta-data is then made available via the "machinectl" tool,
5782 and exposed in "ps" and similar tools. machined/machinectl
5783 is compile-time optional.
5785 * As discussed earlier, the low-level cgroup configuration
5786 options ControlGroup=, ControlGroupModify=,
5787 ControlGroupPersistent=, ControlGroupAttribute= have been
5788 removed. Please use high-level attribute settings instead as
5789 well as slice units.
5791 * A new bus call SetUnitProperties() has been added to alter
5792 various runtime parameters of a unit. This is primarily
5793 useful to alter cgroup parameters dynamically in a nice way,
5794 but will be extended later on to make more properties
5795 modifiable at runtime. systemctl gained a new set-properties
5796 command that wraps this call.
5798 * A new tool "systemd-run" has been added which can be used to
5799 run arbitrary command lines as transient services or scopes,
5800 while configuring a number of settings via the command
5801 line. This tool is currently very basic, however already
5802 very useful. We plan to extend this tool to even allow
5803 queuing of execution jobs with time triggers from the
5804 command line, similar in fashion to "at".
5806 * nspawn will now inform the user explicitly that kernels with
5807 audit enabled break containers, and suggest the user to turn
5810 * Support for detecting the IMA and AppArmor security
5811 frameworks with ConditionSecurity= has been added.
5813 * journalctl gained a new "-k" switch for showing only kernel
5814 messages, mimicking dmesg output; in addition to "--user"
5815 and "--system" switches for showing only user's own logs
5818 * systemd-delta can now show information about drop-in
5819 snippets extending unit files.
5821 * libsystemd-bus has been substantially updated but is still
5822 not available as public API.
5824 * systemd will now look for the "debug" argument on the kernel
5825 command line and enable debug logging, similar to what
5826 "systemd.log_level=debug" already did before.
5828 * "systemctl set-default", "systemctl get-default" has been
5829 added to configure the default.target symlink, which
5830 controls what to boot into by default.
5832 * "systemctl set-log-level" has been added as a convenient
5833 way to raise and lower systemd logging threshold.
5835 * "systemd-analyze plot" will now show the time the various
5836 generators needed for execution, as well as information
5837 about the unit file loading.
5839 * libsystemd-journal gained a new sd_journal_open_files() call
5840 for opening specific journal files. journactl also gained a
5841 new switch to expose this new functionality. Previously we
5842 only supported opening all files from a directory, or all
5843 files from the system, as opening individual files only is
5844 racy due to journal file rotation.
5846 * systemd gained the new DefaultEnvironment= setting in
5847 /etc/systemd/system.conf to set environment variables for
5850 * If a privileged process logs a journal message with the
5851 OBJECT_PID= field set, then journald will automatically
5852 augment this with additional OBJECT_UID=, OBJECT_GID=,
5853 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5854 system services want to log events about specific client
5855 processes. journactl/systemctl has been updated to make use
5856 of this information if all log messages regarding a specific
5859 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5860 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5861 Reisner, David Coppa, David King, David Strauss, Eelco
5862 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5863 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5864 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5865 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5866 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5867 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5868 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5869 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5870 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5871 Łukasz Stelmach, 장동준
5875 * The Python bindings gained some minimal support for the APIs
5876 exposed by libsystemd-logind.
5878 * ConditionSecurity= gained support for detecting SMACK. Since
5879 this condition already supports SELinux and AppArmor we only
5880 miss IMA for this. Patches welcome!
5882 Contributions from: Karol Lewandowski, Lennart Poettering,
5883 Zbigniew Jędrzejewski-Szmek
5887 * systemd-nspawn will now create /etc/resolv.conf if
5888 necessary, before bind-mounting the host's file onto it.
5890 * systemd-nspawn will now store meta information about a
5891 container on the container's cgroup as extended attribute
5892 fields, including the root directory.
5894 * The cgroup hierarchy has been reworked in many ways. All
5895 objects any of the components systemd creates in the cgroup
5896 tree are now suffixed. More specifically, user sessions are
5897 now placed in cgroups suffixed with ".session", users in
5898 cgroups suffixed with ".user", and nspawn containers in
5899 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5900 names are now escaped in a simple scheme to avoid collision
5901 of userspace object names with kernel filenames. This work
5902 is preparation for making these objects relocatable in the
5903 cgroup tree, in order to allow easy resource partitioning of
5904 these objects without causing naming conflicts.
5906 * systemctl list-dependencies gained the new switches
5907 --plain, --reverse, --after and --before.
5909 * systemd-inhibit now shows the process name of processes that
5910 have taken an inhibitor lock.
5912 * nss-myhostname will now also resolve "localhost"
5913 implicitly. This makes /etc/hosts an optional file and
5914 nicely handles that on IPv6 ::1 maps to both "localhost" and
5917 * libsystemd-logind.so gained a new call
5918 sd_get_machine_names() to enumerate running containers and
5919 VMs (currently only supported by very new libvirt and
5920 nspawn). sd_login_monitor can now be used to watch
5921 VMs/containers coming and going.
5923 * .include is not allowed recursively anymore, and only in
5924 unit files. Usually it is better to use drop-in snippets in
5925 .d/*.conf anyway, as introduced with systemd 198.
5927 * systemd-analyze gained a new "critical-chain" command that
5928 determines the slowest chain of units run during system
5929 boot-up. It is very useful for tracking down where
5930 optimizing boot time is the most beneficial.
5932 * systemd will no longer allow manipulating service paths in
5933 the name=systemd:/system cgroup tree using ControlGroup= in
5934 units. (But is still fine with it in all other dirs.)
5936 * There's a new systemd-nspawn@.service service file that may
5937 be used to easily run nspawn containers as system
5938 services. With the container's root directory in
5939 /var/lib/container/foobar it is now sufficient to run
5940 "systemctl start systemd-nspawn@foobar.service" to boot it.
5942 * systemd-cgls gained a new parameter "--machine" to list only
5943 the processes within a certain container.
5945 * ConditionSecurity= now can check for "apparmor". We still
5946 are lacking checks for SMACK and IMA for this condition
5947 check though. Patches welcome!
5949 * A new configuration file /etc/systemd/sleep.conf has been
5950 added that may be used to configure which kernel operation
5951 systemd is supposed to execute when "suspend", "hibernate"
5952 or "hybrid-sleep" is requested. This makes the new kernel
5953 "freeze" state accessible to the user.
5955 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5956 the passed argument if applicable.
5958 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5959 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5960 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5961 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5962 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5963 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5964 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5969 * The output of 'systemctl list-jobs' got some polishing. The
5970 '--type=' argument may now be passed more than once. A new
5971 command 'systemctl list-sockets' has been added which shows
5972 a list of kernel sockets systemd is listening on with the
5973 socket units they belong to, plus the units these socket
5976 * The experimental libsystemd-bus library got substantial
5977 updates to work in conjunction with the (also experimental)
5978 kdbus kernel project. It works well enough to exchange
5979 messages with some sophistication. Note that kdbus is not
5980 ready yet, and the library is mostly an elaborate test case
5981 for now, and not installable.
5983 * systemd gained a new unit 'systemd-static-nodes.service'
5984 that generates static device nodes earlier during boot, and
5985 can run in conjunction with udev.
5987 * libsystemd-login gained a new call sd_pid_get_user_unit()
5988 to retrieve the user systemd unit a process is running
5989 in. This is useful for systems where systemd is used as
5992 * systemd-nspawn now places all containers in the new /machine
5993 top-level cgroup directory in the name=systemd
5994 hierarchy. libvirt will soon do the same, so that we get a
5995 uniform separation of /system, /user and /machine for system
5996 services, user processes and containers/virtual
5997 machines. This new cgroup hierarchy is also useful to stick
5998 stable names to specific container instances, which can be
5999 recognized later this way (this name may be controlled
6000 via systemd-nspawn's new -M switch). libsystemd-login also
6001 gained a new call sd_pid_get_machine_name() to retrieve the
6002 name of the container/VM a specific process belongs to.
6004 * bootchart can now store its data in the journal.
6006 * libsystemd-journal gained a new call
6007 sd_journal_add_conjunction() for AND expressions to the
6008 matching logic. This can be used to express more complex
6009 logical expressions.
6011 * journactl can now take multiple --unit= and --user-unit=
6014 * The cryptsetup logic now understands the "luks.key=" kernel
6015 command line switch for specifying a file to read the
6016 decryption key from. Also, if a configured key file is not
6017 found the tool will now automatically fall back to prompting
6020 * Python systemd.journal module was updated to wrap recently
6021 added functions from libsystemd-journal. The interface was
6022 changed to bring the low level interface in s.j._Reader
6023 closer to the C API, and the high level interface in
6024 s.j.Reader was updated to wrap and convert all data about
6027 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6028 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6029 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6030 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6031 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6032 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6036 * journalctl --update-catalog now understands a new --root=
6037 option to operate on catalogs found in a different root
6040 * During shutdown after systemd has terminated all running
6041 services a final killing loop kills all remaining left-over
6042 processes. We will now print the name of these processes
6043 when we send SIGKILL to them, since this usually indicates a
6046 * If /etc/crypttab refers to password files stored on
6047 configured mount points automatic dependencies will now be
6048 generated to ensure the specific mount is established first
6049 before the key file is attempted to be read.
6051 * 'systemctl status' will now show information about the
6052 network sockets a socket unit is listening on.
6054 * 'systemctl status' will also shown information about any
6055 drop-in configuration file for units. (Drop-In configuration
6056 files in this context are files such as
6057 /etc/systemd/systemd/foobar.service.d/*.conf)
6059 * systemd-cgtop now optionally shows summed up CPU times of
6060 cgroups. Press '%' while running cgtop to switch between
6061 percentage and absolute mode. This is useful to determine
6062 which cgroups use up the most CPU time over the entire
6063 runtime of the system. systemd-cgtop has also been updated
6064 to be 'pipeable' for processing with further shell tools.
6066 * 'hostnamectl set-hostname' will now allow setting of FQDN
6069 * The formatting and parsing of time span values has been
6070 changed. The parser now understands fractional expressions
6071 such as "5.5h". The formatter will now output fractional
6072 expressions for all time spans under 1min, i.e. "5.123456s"
6073 rather than "5s 123ms 456us". For time spans under 1s
6074 millisecond values are shown, for those under 1ms
6075 microsecond values are shown. This should greatly improve
6076 all time-related output of systemd.
6078 * libsystemd-login and libsystemd-journal gained new
6079 functions for querying the poll() events mask and poll()
6080 timeout value for integration into arbitrary event
6083 * localectl gained the ability to list available X11 keymaps
6084 (models, layouts, variants, options).
6086 * 'systemd-analyze dot' gained the ability to filter for
6087 specific units via shell-style globs, to create smaller,
6088 more useful graphs. I.e. it is now possible to create simple
6089 graphs of all the dependencies between only target units, or
6090 of all units that Avahi has dependencies with.
6092 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6093 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6094 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6095 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6096 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6097 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6098 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6102 * The boot-time readahead implementation for rotating media
6103 will now read the read-ahead data in multiple passes which
6104 consist of all read requests made in equidistant time
6105 intervals. This means instead of strictly reading read-ahead
6106 data in its physical order on disk we now try to find a
6107 middle ground between physical and access time order.
6109 * /etc/os-release files gained a new BUILD_ID= field for usage
6110 on operating systems that provide continuous builds of OS
6113 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6114 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6115 William Douglas, Zbigniew Jędrzejewski-Szmek
6119 * systemd-python gained an API exposing libsystemd-daemon.
6121 * The SMACK setup logic gained support for uploading CIPSO
6124 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6125 ReadOnlyDirectories= and InaccessibleDirectories= has
6126 changed. The private /tmp and /var/tmp directories are now
6127 shared by all processes of a service (which means
6128 ExecStartPre= may now leave data in /tmp that ExecStart= of
6129 the same service can still access). When a service is
6130 stopped its temporary directories are immediately deleted
6131 (normal clean-up with tmpfiles is still done in addition to
6134 * By default, systemd will now set a couple of sysctl
6135 variables in the kernel: the safe sysrq options are turned
6136 on, IP route verification is turned on, and source routing
6137 disabled. The recently added hardlink and softlink
6138 protection of the kernel is turned on. These settings should
6139 be reasonably safe, and good defaults for all new systems.
6141 * The predictable network naming logic may now be turned off
6142 with a new kernel command line switch: net.ifnames=0.
6144 * A new libsystemd-bus module has been added that implements a
6145 pretty complete D-Bus client library. For details see:
6147 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6149 * journald will now explicitly flush the journal files to disk
6150 at the latest 5min after each write. The file will then also
6151 be marked offline until the next write. This should increase
6152 reliability in case of a crash. The synchronization delay
6153 can be configured via SyncIntervalSec= in journald.conf.
6155 * There's a new remote-fs-setup.target unit that can be used
6156 to pull in specific services when at least one remote file
6157 system is to be mounted.
6159 * There are new targets timers.target and paths.target as
6160 canonical targets to pull user timer and path units in
6161 from. This complements sockets.target with a similar
6162 purpose for socket units.
6164 * libudev gained a new call udev_device_set_attribute_value()
6165 to set sysfs attributes of a device.
6167 * The udev daemon now sets the default number of worker
6168 processes executed in parallel based on the number of available
6169 CPUs instead of the amount of available RAM. This is supposed
6170 to provide a more reliable default and limit a too aggressive
6171 parallelism for setups with 1000s of devices connected.
6173 Contributions from: Auke Kok, Colin Walters, Cristian
6174 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6175 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6176 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6177 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6178 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6179 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6180 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6181 Zbigniew Jędrzejewski-Szmek
6185 * Configuration of unit files may now be extended via drop-in
6186 files without having to edit/override the unit files
6187 themselves. More specifically, if the administrator wants to
6188 change one value for a service file foobar.service he can
6189 now do so by dropping in a configuration snippet into
6190 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6191 will load all these snippets and apply them on top of the
6192 main unit configuration file, possibly extending or
6193 overriding its settings. Using these drop-in snippets is
6194 generally nicer than the two earlier options for changing
6195 unit files locally: copying the files from
6196 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6197 them there; or creating a new file in /etc/systemd/system/
6198 that incorporates the original one via ".include". Drop-in
6199 snippets into these .d/ directories can be placed in any
6200 directory systemd looks for units in, and the usual
6201 overriding semantics between /usr/lib, /etc and /run apply
6204 * Most unit file settings which take lists of items can now be
6205 reset by assigning the empty string to them. For example,
6206 normally, settings such as Environment=FOO=BAR append a new
6207 environment variable assignment to the environment block,
6208 each time they are used. By assigning Environment= the empty
6209 string the environment block can be reset to empty. This is
6210 particularly useful with the .d/*.conf drop-in snippets
6211 mentioned above, since this adds the ability to reset list
6212 settings from vendor unit files via these drop-ins.
6214 * systemctl gained a new "list-dependencies" command for
6215 listing the dependencies of a unit recursively.
6217 * Inhibitors are now honored and listed by "systemctl
6218 suspend", "systemctl poweroff" (and similar) too, not only
6219 GNOME. These commands will also list active sessions by
6222 * Resource limits (as exposed by the various control group
6223 controllers) can now be controlled dynamically at runtime
6224 for all units. More specifically, you can now use a command
6225 like "systemctl set-cgroup-attr foobar.service cpu.shares
6226 2000" to alter the CPU shares a specific service gets. These
6227 settings are stored persistently on disk, and thus allow the
6228 administrator to easily adjust the resource usage of
6229 services with a few simple commands. This dynamic resource
6230 management logic is also available to other programs via the
6231 bus. Almost any kernel cgroup attribute and controller is
6234 * systemd-vconsole-setup will now copy all font settings to
6235 all allocated VTs, where it previously applied them only to
6238 * libsystemd-login gained the new sd_session_get_tty() API
6241 * This release drops support for a few legacy or
6242 distribution-specific LSB facility names when parsing init
6243 scripts: $x-display-manager, $mail-transfer-agent,
6244 $mail-transport-agent, $mail-transfer-agent, $smtp,
6245 $null. Also, the mail-transfer-agent.target unit backing
6246 this has been removed. Distributions which want to retain
6247 compatibility with this should carry the burden for
6248 supporting this themselves and patch support for these back
6249 in, if they really need to. Also, the facilities $syslog and
6250 $local_fs are now ignored, since systemd does not support
6251 early-boot LSB init scripts anymore, and these facilities
6252 are implied anyway for normal services. syslog.target has
6255 * There are new bus calls on PID1's Manager object for
6256 cancelling jobs, and removing snapshot units. Previously,
6257 both calls were only available on the Job and Snapshot
6260 * systemd-journal-gatewayd gained SSL support.
6262 * The various "environment" files, such as /etc/locale.conf
6263 now support continuation lines with a backslash ("\") as
6264 last character in the line, similarly in style (but different)
6265 to how this is supported in shells.
6267 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6268 now implicitly appended to every log entry logged. systemctl
6269 has been updated to filter by this field when operating on a
6270 user systemd instance.
6272 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6273 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6274 the container. This makes it easier to boot unmodified
6275 Fedora systems in a container, which however still requires
6276 audit=0 to be passed on the kernel command line. Auditing in
6277 kernel and userspace is unfortunately still too broken in
6278 context of containers, hence we recommend compiling it out
6279 of the kernel or using audit=0. Hopefully this will be fixed
6280 one day for good in the kernel.
6282 * nspawn gained the new --bind= and --bind-ro= parameters to
6283 bind mount specific directories from the host into the
6286 * nspawn will now mount its own devpts file system instance
6287 into the container, in order not to leak pty devices from
6288 the host into the container.
6290 * systemd will now read the firmware boot time performance
6291 information from the EFI variables, if the used boot loader
6292 supports this, and takes it into account for boot performance
6293 analysis via "systemd-analyze". This is currently supported
6294 only in conjunction with Gummiboot, but could be supported
6295 by other boot loaders too. For details see:
6297 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6299 * A new generator has been added that automatically mounts the
6300 EFI System Partition (ESP) to /boot, if that directory
6301 exists, is empty, and no other file system has been
6302 configured to be mounted there.
6304 * logind will now send out PrepareForSleep(false) out
6305 unconditionally, after coming back from suspend. This may be
6306 used by applications as asynchronous notification for
6307 system resume events.
6309 * "systemctl unlock-sessions" has been added, that allows
6310 unlocking the screens of all user sessions at once, similar
6311 to how "systemctl lock-sessions" already locked all users
6312 sessions. This is backed by a new D-Bus call UnlockSessions().
6314 * "loginctl seat-status" will now show the master device of a
6315 seat. (i.e. the device of a seat that needs to be around for
6316 the seat to be considered available, usually the graphics
6319 * tmpfiles gained a new "X" line type, that allows
6320 configuration of files and directories (with wildcards) that
6321 shall be excluded from automatic cleanup ("aging").
6323 * udev default rules set the device node permissions now only
6324 at "add" events, and do not change them any longer with a
6325 later "change" event.
6327 * The log messages for lid events and power/sleep keypresses
6328 now carry a message ID.
6330 * We now have a substantially larger unit test suite, but this
6331 continues to be work in progress.
6333 * udevadm hwdb gained a new --root= parameter to change the
6334 root directory to operate relative to.
6336 * logind will now issue a background sync() request to the kernel
6337 early at shutdown, so that dirty buffers are flushed to disk early
6338 instead of at the last moment, in order to optimize shutdown
6341 * A new bootctl tool has been added that is an interface for
6342 certain boot loader operations. This is currently a preview
6343 and is likely to be extended into a small mechanism daemon
6344 like timedated, localed, hostnamed, and can be used by
6345 graphical UIs to enumerate available boot options, and
6346 request boot into firmware operations.
6348 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6349 the rest of the package. It also has been updated to work
6350 correctly in initrds.
6352 * Policykit previously has been runtime optional, and is now
6353 also compile time optional via a configure switch.
6355 * systemd-analyze has been reimplemented in C. Also "systemctl
6356 dot" has moved into systemd-analyze.
6358 * "systemctl status" with no further parameters will now print
6359 the status of all active or failed units.
6361 * Operations such as "systemctl start" can now be executed
6362 with a new mode "--irreversible" which may be used to queue
6363 operations that cannot accidentally be reversed by a later
6364 job queuing. This is by default used to make shutdown
6365 requests more robust.
6367 * The Python API of systemd now gained a new module for
6368 reading journal files.
6370 * A new tool kernel-install has been added that can install
6371 kernel images according to the Boot Loader Specification:
6373 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6375 * Boot time console output has been improved to provide
6376 animated boot time output for hanging jobs.
6378 * A new tool systemd-activate has been added which can be used
6379 to test socket activation with, directly from the command
6380 line. This should make it much easier to test and debug
6381 socket activation in daemons.
6383 * journalctl gained a new "--reverse" (or -r) option to show
6384 journal output in reverse order (i.e. newest line first).
6386 * journalctl gained a new "--pager-end" (or -e) option to jump
6387 to immediately jump to the end of the journal in the
6388 pager. This is only supported in conjunction with "less".
6390 * journalctl gained a new "--user-unit=" option, that works
6391 similarly to "--unit=" but filters for user units rather than
6394 * A number of unit files to ease adoption of systemd in
6395 initrds has been added. This moves some minimal logic from
6396 the various initrd implementations into systemd proper.
6398 * The journal files are now owned by a new group
6399 "systemd-journal", which exists specifically to allow access
6400 to the journal, and nothing else. Previously, we used the
6401 "adm" group for that, which however possibly covers more
6402 than just journal/log file access. This new group is now
6403 already used by systemd-journal-gatewayd to ensure this
6404 daemon gets access to the journal files and as little else
6405 as possible. Note that "make install" will also set FS ACLs
6406 up for /var/log/journal to give "adm" and "wheel" read
6407 access to it, in addition to "systemd-journal" which owns
6408 the journal files. We recommend that packaging scripts also
6409 add read access to "adm" + "wheel" to /var/log/journal, and
6410 all existing/future journal files. To normal users and
6411 administrators little changes, however packagers need to
6412 ensure to create the "systemd-journal" system group at
6413 package installation time.
6415 * The systemd-journal-gatewayd now runs as unprivileged user
6416 systemd-journal-gateway:systemd-journal-gateway. Packaging
6417 scripts need to create these system user/group at
6420 * timedated now exposes a new boolean property CanNTP that
6421 indicates whether a local NTP service is available or not.
6423 * systemd-detect-virt will now also detect xen PVs
6425 * The pstore file system is now mounted by default, if it is
6428 * In addition to the SELinux and IMA policies we will now also
6429 load SMACK policies at early boot.
6431 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6432 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6433 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6434 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6435 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6436 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6437 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6438 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6439 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6440 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6441 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6442 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6443 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6444 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6448 * Timer units now support calendar time events in addition to
6449 monotonic time events. That means you can now trigger a unit
6450 based on a calendar time specification such as "Thu,Fri
6451 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6452 or fifth day of any month of the year 2013, given that it is
6453 a thursday or friday. This brings timer event support
6454 considerably closer to cron's capabilities. For details on
6455 the supported calendar time specification language see
6458 * udev now supports a number of different naming policies for
6459 network interfaces for predictable names, and a combination
6460 of these policies is now the default. Please see this wiki
6461 document for details:
6463 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6465 * Auke Kok's bootchart implementation has been added to the
6466 systemd tree. It is an optional component that can graph the
6467 boot in quite some detail. It is one of the best bootchart
6468 implementations around and minimal in its code and
6471 * nss-myhostname has been integrated into the systemd source
6472 tree. nss-myhostname guarantees that the local hostname
6473 always stays resolvable via NSS. It has been a weak
6474 requirement of systemd-hostnamed since a long time, and
6475 since its code is actually trivial we decided to just
6476 include it in systemd's source tree. It can be turned off
6477 with a configure switch.
6479 * The read-ahead logic is now capable of properly detecting
6480 whether a btrfs file system is on SSD or rotating media, in
6481 order to optimize the read-ahead scheme. Previously, it was
6482 only capable of detecting this on traditional file systems
6485 * In udev, additional device properties are now read from the
6486 IAB in addition to the OUI database. Also, Bluetooth company
6487 identities are attached to the devices as well.
6489 * In service files %U may be used as specifier that is
6490 replaced by the configured user name of the service.
6492 * nspawn may now be invoked without a controlling TTY. This
6493 makes it suitable for invocation as its own service. This
6494 may be used to set up a simple containerized server system
6495 using only core OS tools.
6497 * systemd and nspawn can now accept socket file descriptors
6498 when they are started for socket activation. This enables
6499 implementation of socket activated nspawn
6500 containers. i.e. think about autospawning an entire OS image
6501 when the first SSH or HTTP connection is received. We expect
6502 that similar functionality will also be added to libvirt-lxc
6505 * journalctl will now suppress ANSI color codes when
6506 presenting log data.
6508 * systemctl will no longer show control group information for
6509 a unit if the control group is empty anyway.
6511 * logind can now automatically suspend/hibernate/shutdown the
6514 * /etc/machine-info and hostnamed now also expose the chassis
6515 type of the system. This can be used to determine whether
6516 the local system is a laptop, desktop, handset or
6517 tablet. This information may either be configured by the
6518 user/vendor or is automatically determined from ACPI and DMI
6519 information if possible.
6521 * A number of PolicyKit actions are now bound together with
6522 "imply" rules. This should simplify creating UIs because
6523 many actions will now authenticate similar ones as well.
6525 * Unit files learnt a new condition ConditionACPower= which
6526 may be used to conditionalize a unit depending on whether an
6527 AC power source is connected or not, of whether the system
6528 is running on battery power.
6530 * systemctl gained a new "is-failed" verb that may be used in
6531 shell scripts and suchlike to check whether a specific unit
6532 is in the "failed" state.
6534 * The EnvironmentFile= setting in unit files now supports file
6535 globbing, and can hence be used to easily read a number of
6536 environment files at once.
6538 * systemd will no longer detect and recognize specific
6539 distributions. All distribution-specific #ifdeffery has been
6540 removed, systemd is now fully generic and
6541 distribution-agnostic. Effectively, not too much is lost as
6542 a lot of the code is still accessible via explicit configure
6543 switches. However, support for some distribution specific
6544 legacy configuration file formats has been dropped. We
6545 recommend distributions to simply adopt the configuration
6546 files everybody else uses now and convert the old
6547 configuration from packaging scripts. Most distributions
6548 already did that. If that's not possible or desirable,
6549 distributions are welcome to forward port the specific
6550 pieces of code locally from the git history.
6552 * When logging a message about a unit systemd will now always
6553 log the unit name in the message meta data.
6555 * localectl will now also discover system locale data that is
6556 not stored in locale archives, but directly unpacked.
6558 * logind will no longer unconditionally use framebuffer
6559 devices as seat masters, i.e. as devices that are required
6560 to be existing before a seat is considered preset. Instead,
6561 it will now look for all devices that are tagged as
6562 "seat-master" in udev. By default, framebuffer devices will
6563 be marked as such, but depending on local systems, other
6564 devices might be marked as well. This may be used to
6565 integrate graphics cards using closed source drivers (such
6566 as NVidia ones) more nicely into logind. Note however, that
6567 we recommend using the open source NVidia drivers instead,
6568 and no udev rules for the closed-source drivers will be
6569 shipped from us upstream.
6571 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6572 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6573 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6574 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6575 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6576 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6577 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6578 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6579 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6580 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6581 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6582 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6587 * udev gained support for loading additional device properties
6588 from an indexed database that is keyed by vendor/product IDs
6589 and similar device identifiers. For the beginning this
6590 "hwdb" is populated with data from the well-known PCI and
6591 USB database, but also includes PNP, ACPI and OID data. In
6592 the longer run this indexed database shall grow into
6593 becoming the one central database for non-essential
6594 userspace device metadata. Previously, data from the PCI/USB
6595 database was only attached to select devices, since the
6596 lookup was a relatively expensive operation due to O(n) time
6597 complexity (with n being the number of entries in the
6598 database). Since this is now O(1), we decided to add in this
6599 data for all devices where this is available, by
6600 default. Note that the indexed database needs to be rebuilt
6601 when new data files are installed. To achieve this you need
6602 to update your packaging scripts to invoke "udevadm hwdb
6603 --update" after installation of hwdb data files. For
6604 RPM-based distributions we introduced the new
6605 %udev_hwdb_update macro for this purpose.
6607 * The Journal gained support for the "Message Catalog", an
6608 indexed database to link up additional information with
6609 journal entries. For further details please check:
6611 https://www.freedesktop.org/wiki/Software/systemd/catalog
6613 The indexed message catalog database also needs to be
6614 rebuilt after installation of message catalog files. Use
6615 "journalctl --update-catalog" for this. For RPM-based
6616 distributions we introduced the %journal_catalog_update
6617 macro for this purpose.
6619 * The Python Journal bindings gained support for the standard
6620 Python logging framework.
6622 * The Journal API gained new functions for checking whether
6623 the underlying file system of a journal file is capable of
6624 properly reporting file change notifications, or whether
6625 applications that want to reflect journal changes "live"
6626 need to recheck journal files continuously in appropriate
6629 * It is now possible to set the "age" field for tmpfiles
6630 entries to 0, indicating that files matching this entry
6631 shall always be removed when the directories are cleaned up.
6633 * coredumpctl gained a new "gdb" verb which invokes gdb
6634 right-away on the selected coredump.
6636 * There's now support for "hybrid sleep" on kernels that
6637 support this, in addition to "suspend" and "hibernate". Use
6638 "systemctl hybrid-sleep" to make use of this.
6640 * logind's HandleSuspendKey= setting (and related settings)
6641 now gained support for a new "lock" setting to simply
6642 request the screen lock on all local sessions, instead of
6643 actually executing a suspend or hibernation.
6645 * systemd will now mount the EFI variables file system by
6648 * Socket units now gained support for configuration of the
6649 SMACK security label.
6651 * timedatectl will now output the time of the last and next
6652 daylight saving change.
6654 * We dropped support for various legacy and distro-specific
6655 concepts, such as insserv, early-boot SysV services
6656 (i.e. those for non-standard runlevels such as 'b' or 'S')
6657 or ArchLinux /etc/rc.conf support. We recommend the
6658 distributions who still need support this to either continue
6659 to maintain the necessary patches downstream, or find a
6660 different solution. (Talk to us if you have questions!)
6662 * Various systemd components will now bypass PolicyKit checks
6663 for root and otherwise handle properly if PolicyKit is not
6664 found to be around. This should fix most issues for
6665 PolicyKit-less systems. Quite frankly this should have been
6666 this way since day one. It is absolutely our intention to
6667 make systemd work fine on PolicyKit-less systems, and we
6668 consider it a bug if something does not work as it should if
6669 PolicyKit is not around.
6671 * For embedded systems it is now possible to build udev and
6672 systemd without blkid and/or kmod support.
6674 * "systemctl switch-root" is now capable of switching root
6675 more than once. I.e. in addition to transitions from the
6676 initrd to the host OS it is now possible to transition to
6677 further OS images from the host. This is useful to implement
6678 offline updating tools.
6680 * Various other additions have been made to the RPM macros
6681 shipped with systemd. Use %udev_rules_update() after
6682 installing new udev rules files. %_udevhwdbdir,
6683 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6684 %_sysctldir are now available which resolve to the right
6685 directories for packages to place various data files in.
6687 * journalctl gained the new --full switch (in addition to
6688 --all, to disable ellipsation for long messages.
6690 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6691 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6692 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6693 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6694 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6695 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6696 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6697 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6698 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6702 * journalctl gained new --since= and --until= switches to
6703 filter by time. It also now supports nice filtering for
6704 units via --unit=/-u.
6706 * Type=oneshot services may use ExecReload= and do the
6709 * The journal daemon now supports time-based rotation and
6710 vacuuming, in addition to the usual disk-space based
6713 * The journal will now index the available field values for
6714 each field name. This enables clients to show pretty drop
6715 downs of available match values when filtering. The bash
6716 completion of journalctl has been updated
6717 accordingly. journalctl gained a new switch -F to list all
6718 values a certain field takes in the journal database.
6720 * More service events are now written as structured messages
6721 to the journal, and made recognizable via message IDs.
6723 * The timedated, localed and hostnamed mini-services which
6724 previously only provided support for changing time, locale
6725 and hostname settings from graphical DEs such as GNOME now
6726 also have a minimal (but very useful) text-based client
6727 utility each. This is probably the nicest way to changing
6728 these settings from the command line now, especially since
6729 it lists available options and is fully integrated with bash
6732 * There's now a new tool "systemd-coredumpctl" to list and
6733 extract coredumps from the journal.
6735 * We now install a README each in /var/log/ and
6736 /etc/rc.d/init.d explaining where the system logs and init
6737 scripts went. This hopefully should help folks who go to
6738 that dirs and look into the otherwise now empty void and
6739 scratch their heads.
6741 * When user-services are invoked (by systemd --user) the
6742 $MANAGERPID env var is set to the PID of systemd.
6744 * SIGRTMIN+24 when sent to a --user instance will now result
6745 in immediate termination of systemd.
6747 * gatewayd received numerous feature additions such as a
6748 "follow" mode, for live syncing and filtering.
6750 * browse.html now allows filtering and showing detailed
6751 information on specific entries. Keyboard navigation and
6752 mouse screen support has been added.
6754 * gatewayd/journalctl now supports HTML5/JSON
6755 Server-Sent-Events as output.
6757 * The SysV init script compatibility logic will now
6758 heuristically determine whether a script supports the
6759 "reload" verb, and only then make this available as
6762 * "systemctl status --follow" has been removed, use "journalctl
6765 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6766 have been removed since they are hardly useful to be
6769 * And I'd like to take the opportunity to specifically mention
6770 Zbigniew for his great contributions. Zbigniew, you rock!
6772 Contributions from: Andrew Eikum, Christian Hesse, Colin
6773 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6774 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6775 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6776 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6777 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6778 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6782 * If /etc/vconsole.conf is non-existent or empty we will no
6783 longer load any console font or key map at boot by
6784 default. Instead the kernel defaults will be left
6785 intact. This is definitely the right thing to do, as no
6786 configuration should mean no configuration, and hard-coding
6787 font names that are different on all archs is probably a bad
6788 idea. Also, the kernel default key map and font should be
6789 good enough for most cases anyway, and mostly identical to
6790 the userspace fonts/key maps we previously overloaded them
6791 with. If distributions want to continue to default to a
6792 non-kernel font or key map they should ship a default
6793 /etc/vconsole.conf with the appropriate contents.
6795 Contributions from: Colin Walters, Daniel J Walsh, Dave
6796 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6797 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6801 * journalctl gained a new --cursor= switch to show entries
6802 starting from the specified location in the journal.
6804 * We now enforce a size limit on journal entry fields exported
6805 with "-o json" in journalctl. Fields larger than 4K will be
6806 assigned null. This can be turned off with --all.
6808 * An (optional) journal gateway daemon is now available as
6809 "systemd-journal-gatewayd.service". This service provides
6810 access to the journal via HTTP and JSON. This functionality
6811 will be used to implement live log synchronization in both
6812 pull and push modes, but has various other users too, such
6813 as easy log access for debugging of embedded devices. Right
6814 now it is already useful to retrieve the journal via HTTP:
6816 # systemctl start systemd-journal-gatewayd.service
6817 # wget http://localhost:19531/entries
6819 This will download the journal contents in a
6820 /var/log/messages compatible format. The same as JSON:
6822 # curl -H"Accept: application/json" http://localhost:19531/entries
6824 This service is also accessible via a web browser where a
6825 single static HTML5 app is served that uses the JSON logic
6826 to enable the user to do some basic browsing of the
6827 journal. This will be extended later on. Here's an example
6828 screenshot of this app in its current state:
6830 http://0pointer.de/public/journal-gatewayd
6832 Contributions from: Kay Sievers, Lennart Poettering, Robert
6833 Milasan, Tom Gundersen
6837 * The bash completion logic is now available for journalctl
6840 * We do not mount the "cpuset" controller anymore together with
6841 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6842 started if no parameters are assigned to it. "cpuset" hence
6843 broke code that assumed it could create "cpu" groups and
6846 * journalctl -f will now subscribe to terminal size changes,
6847 and line break accordingly.
6849 Contributions from: Dave Reisner, Kay Sievers, Lennart
6850 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6854 * nspawn will now create a symlink /etc/localtime in the
6855 container environment, copying the host's timezone
6856 setting. Previously this has been done via a bind mount, but
6857 since symlinks cannot be bind mounted this has now been
6858 changed to create/update the appropriate symlink.
6860 * journalctl -n's line number argument is now optional, and
6861 will default to 10 if omitted.
6863 * journald will now log the maximum size the journal files may
6864 take up on disk. This is particularly useful if the default
6865 built-in logic of determining this parameter from the file
6866 system size is used. Use "systemctl status
6867 systemd-journald.service" to see this information.
6869 * The multi-seat X wrapper tool has been stripped down. As X
6870 is now capable of enumerating graphics devices via udev in a
6871 seat-aware way the wrapper is not strictly necessary
6872 anymore. A stripped down temporary stop-gap is still shipped
6873 until the upstream display managers have been updated to
6874 fully support the new X logic. Expect this wrapper to be
6875 removed entirely in one of the next releases.
6877 * HandleSleepKey= in logind.conf has been split up into
6878 HandleSuspendKey= and HandleHibernateKey=. The old setting
6879 is not available anymore. X11 and the kernel are
6880 distinguishing between these keys and we should too. This
6881 also means the inhibition lock for these keys has been split
6884 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6885 Poettering, Lukas Nykryn, Václav Pavlín
6889 * Whenever a unit changes state we will now log this to the
6890 journal and show along the unit's own log output in
6893 * ConditionPathIsMountPoint= can now properly detect bind
6894 mount points too. (Previously, a bind mount of one file
6895 system to another place in the same file system could not be
6896 detected as mount, since they shared struct stat's st_dev
6899 * We will now mount the cgroup controllers cpu, cpuacct,
6900 cpuset and the controllers net_cls, net_prio together by
6903 * nspawn containers will now have a virtualized boot
6904 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6905 over with a randomized ID at container initialization). This
6906 has the effect of making "journalctl -b" do the right thing
6909 * The JSON output journal serialization has been updated not
6910 to generate "endless" list objects anymore, but rather one
6911 JSON object per line. This is more in line how most JSON
6912 parsers expect JSON objects. The new output mode
6913 "json-pretty" has been added to provide similar output, but
6914 neatly aligned for readability by humans.
6916 * We dropped all explicit sync() invocations in the shutdown
6917 code. The kernel does this implicitly anyway in the kernel
6918 reboot() syscall. halt(8)'s -n option is now a compatibility
6921 * We now support virtualized reboot() in containers, as
6922 supported by newer kernels. We will fall back to exit() if
6923 CAP_SYS_REBOOT is not available to the container. Also,
6924 nspawn makes use of this now and will actually reboot the
6925 container if the containerized OS asks for that.
6927 * journalctl will only show local log output by default
6928 now. Use --merge (-m) to show remote log output, too.
6930 * libsystemd-journal gained the new sd_journal_get_usage()
6931 call to determine the current disk usage of all journal
6932 files. This is exposed in the new "journalctl --disk-usage"
6935 * journald gained a new configuration setting SplitMode= in
6936 journald.conf which may be used to control how user journals
6937 are split off. See journald.conf(5) for details.
6939 * A new condition type ConditionFileNotEmpty= has been added.
6941 * tmpfiles' "w" lines now support file globbing, to write
6942 multiple files at once.
6944 * We added Python bindings for the journal submission
6945 APIs. More Python APIs for a number of selected APIs will
6946 likely follow. Note that we intend to add native bindings
6947 only for the Python language, as we consider it common
6948 enough to deserve bindings shipped within systemd. There are
6949 various projects outside of systemd that provide bindings
6950 for languages such as PHP or Lua.
6952 * Many conditions will now resolve specifiers such as %i. In
6953 addition, PathChanged= and related directives of .path units
6954 now support specifiers as well.
6956 * There's now a new RPM macro definition for the system preset
6959 * journald will now warn if it ca not forward a message to the
6960 syslog daemon because its socket is full.
6962 * timedated will no longer write or process /etc/timezone,
6963 except on Debian. As we do not support late mounted /usr
6964 anymore /etc/localtime always being a symlink is now safe,
6965 and hence the information in /etc/timezone is not necessary
6968 * logind will now always reserve one VT for a text getty (VT6
6969 by default). Previously if more than 6 X sessions where
6970 started they took up all the VTs with auto-spawned gettys,
6971 so that no text gettys were available anymore.
6973 * udev will now automatically inform the btrfs kernel logic
6974 about btrfs RAID components showing up. This should make
6975 simple hotplug based btrfs RAID assembly work.
6977 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6978 (but not for its children which will stay at the kernel
6979 default). This should allow setups with a lot more listening
6982 * systemd will now always pass the configured timezone to the
6983 kernel at boot. timedated will do the same when the timezone
6986 * logind's inhibition logic has been updated. By default,
6987 logind will now handle the lid switch, the power and sleep
6988 keys all the time, even in graphical sessions. If DEs want
6989 to handle these events on their own they should take the new
6990 handle-power-key, handle-sleep-key and handle-lid-switch
6991 inhibitors during their runtime. A simple way to achieve
6992 that is to invoke the DE wrapped in an invocation of:
6994 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6996 * Access to unit operations is now checked via SELinux taking
6997 the unit file label and client process label into account.
6999 * systemd will now notify the administrator in the journal
7000 when he over-mounts a non-empty directory.
7002 * There are new specifiers that are resolved in unit files,
7003 for the host name (%H), the machine ID (%m) and the boot ID
7006 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7007 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7008 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7009 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7010 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7011 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7012 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7016 * Support for reading structured kernel messages from
7017 /dev/kmsg has now been added and is enabled by default.
7019 * Support for reading kernel messages from /proc/kmsg has now
7020 been removed. If you want kernel messages in the journal
7021 make sure to run a recent kernel (>= 3.5) that supports
7022 reading structured messages from /dev/kmsg (see
7023 above). /proc/kmsg is now exclusive property of classic
7024 syslog daemons again.
7026 * The libudev API gained the new
7027 udev_device_new_from_device_id() call.
7029 * The logic for file system namespace (ReadOnlyDirectory=,
7030 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7031 require pivot_root() anymore. This means fewer temporary
7032 directories are created below /tmp for this feature.
7034 * nspawn containers will now see and receive all submounts
7035 made on the host OS below the root file system of the
7038 * Forward Secure Sealing is now supported for Journal files,
7039 which provide cryptographical sealing of journal files so
7040 that attackers cannot alter log history anymore without this
7041 being detectable. Lennart will soon post a blog story about
7042 this explaining it in more detail.
7044 * There are two new service settings RestartPreventExitStatus=
7045 and SuccessExitStatus= which allow configuration of exit
7046 status (exit code or signal) which will be excepted from the
7047 restart logic, resp. consider successful.
7049 * journalctl gained the new --verify switch that can be used
7050 to check the integrity of the structure of journal files and
7051 (if Forward Secure Sealing is enabled) the contents of
7054 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7055 and similar symlinks pre-created. This makes running shells
7056 as container init process a lot more fun.
7058 * The fstab support can now handle PARTUUID= and PARTLABEL=
7061 * A new ConditionHost= condition has been added to match
7062 against the hostname (with globs) and machine ID. This is
7063 useful for clusters where a single OS image is used to
7064 provision a large number of hosts which shall run slightly
7065 different sets of services.
7067 * Services which hit the restart limit will now be placed in a
7070 Contributions from: Bertram Poettering, Dave Reisner, Huang
7071 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7072 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7076 * When running in --user mode systemd will now become a
7077 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7078 tree a lot more organized.
7080 * A new PartOf= unit dependency type has been introduced that
7081 may be used to group services in a natural way.
7083 * "systemctl enable" may now be used to enable instances of
7086 * journalctl now prints error log levels in red, and
7087 warning/notice log levels in bright white. It also supports
7088 filtering by log level now.
7090 * cgtop gained a new -n switch (similar to top), to configure
7091 the maximum number of iterations to run for. It also gained
7092 -b, to run in batch mode (accepting no input).
7094 * The suffix ".service" may now be omitted on most systemctl
7095 command lines involving service unit names.
7097 * There's a new bus call in logind to lock all sessions, as
7098 well as a loginctl verb for it "lock-sessions".
7100 * libsystemd-logind.so gained a new call sd_journal_perror()
7101 that works similar to libc perror() but logs to the journal
7102 and encodes structured information about the error number.
7104 * /etc/crypttab entries now understand the new keyfile-size=
7107 * shutdown(8) now can send a (configurable) wall message when
7108 a shutdown is cancelled.
7110 * The mount propagation mode for the root file system will now
7111 default to "shared", which is useful to make containers work
7112 nicely out-of-the-box so that they receive new mounts from
7113 the host. This can be undone locally by running "mount
7114 --make-rprivate /" if needed.
7116 * The prefdm.service file has been removed. Distributions
7117 should maintain this unit downstream if they intend to keep
7118 it around. However, we recommend writing normal unit files
7119 for display managers instead.
7121 * Since systemd is a crucial part of the OS we will now
7122 default to a number of compiler switches that improve
7123 security (hardening) such as read-only relocations, stack
7124 protection, and suchlike.
7126 * The TimeoutSec= setting for services is now split into
7127 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7128 of individual time outs for the start and the stop phase of
7131 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7132 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7133 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7134 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7135 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7136 Gundersen, Zbigniew Jędrzejewski-Szmek
7140 * The journal and id128 C APIs are now fully documented as man
7143 * Extra safety checks have been added when transitioning from
7144 the initial RAM disk to the main system to avoid accidental
7147 * /etc/crypttab entries now understand the new keyfile-offset=
7150 * systemctl -t can now be used to filter by unit load state.
7152 * The journal C API gained the new sd_journal_wait() call to
7153 make writing synchronous journal clients easier.
7155 * journalctl gained the new -D switch to show journals from a
7158 * journalctl now displays a special marker between log
7159 messages of two different boots.
7161 * The journal is now explicitly flushed to /var via a service
7162 systemd-journal-flush.service, rather than implicitly simply
7163 by seeing /var/log/journal to be writable.
7165 * journalctl (and the journal C APIs) can now match for much
7166 more complex expressions, with alternatives and
7169 * When transitioning from the initial RAM disk to the main
7170 system we will now kill all processes in a killing spree to
7171 ensure no processes stay around by accident.
7173 * Three new specifiers may be used in unit files: %u, %h, %s
7174 resolve to the user name, user home directory resp. user
7175 shell. This is useful for running systemd user instances.
7177 * We now automatically rotate journal files if their data
7178 object hash table gets a fill level > 75%. We also size the
7179 hash table based on the configured maximum file size. This
7180 together should lower hash collisions drastically and thus
7181 speed things up a bit.
7183 * journalctl gained the new "--header" switch to introspect
7184 header data of journal files.
7186 * A new setting SystemCallFilters= has been added to services
7187 which may be used to apply blacklists or whitelists to
7188 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7190 * nspawn gained a new --link-journal= switch (and quicker: -j)
7191 to link the container journal with the host. This makes it
7192 very easy to centralize log viewing on the host for all
7193 guests while still keeping the journal files separated.
7195 * Many bugfixes and optimizations
7197 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7198 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7199 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7204 * Several tools now understand kernel command line arguments,
7205 which are only read when run in an initial RAM disk. They
7206 usually follow closely their normal counterparts, but are
7209 * There's a new tool to analyze the readahead files that are
7210 automatically generated at boot. Use:
7212 /usr/lib/systemd/systemd-readahead analyze /.readahead
7214 * We now provide an early debug shell on tty9 if this enabled. Use:
7216 systemctl enable debug-shell.service
7218 * All plymouth related units have been moved into the Plymouth
7219 package. Please make sure to upgrade your Plymouth version
7222 * systemd-tmpfiles now supports getting passed the basename of
7223 a configuration file only, in which case it will look for it
7224 in all appropriate directories automatically.
7226 * udevadm info now takes a /dev or /sys path as argument, and
7227 does the right thing. Example:
7229 udevadm info /dev/sda
7230 udevadm info /sys/class/block/sda
7232 * systemctl now prints a warning if a unit is stopped but a
7233 unit that might trigger it continues to run. Example: a
7234 service is stopped but the socket that activates it is left
7237 * "systemctl status" will now mention if the log output was
7238 shortened due to rotation since a service has been started.
7240 * The journal API now exposes functions to determine the
7241 "cutoff" times due to rotation.
7243 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7244 immediately flushing of runtime logs to /var if possible,
7245 resp. for triggering immediate rotation of the journal
7248 * It is now considered an error if a service is attempted to
7249 be stopped that is not loaded.
7251 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7253 * systemd-analyze now supports Python 3
7255 * tmpfiles now supports cleaning up directories via aging
7256 where the first level dirs are always kept around but
7257 directories beneath it automatically aged. This is enabled
7258 by prefixing the age field with '~'.
7260 * Seat objects now expose CanGraphical, CanTTY properties
7261 which is required to deal with very fast bootups where the
7262 display manager might be running before the graphics drivers
7263 completed initialization.
7265 * Seat objects now expose a State property.
7267 * We now include RPM macros for service enabling/disabling
7268 based on the preset logic. We recommend RPM based
7269 distributions to make use of these macros if possible. This
7270 makes it simpler to reuse RPM spec files across
7273 * We now make sure that the collected systemd unit name is
7274 always valid when services log to the journal via
7277 * There's a new man page kernel-command-line(7) detailing all
7278 command line options we understand.
7280 * The fstab generator may now be disabled at boot by passing
7281 fstab=0 on the kernel command line.
7283 * A new kernel command line option modules-load= is now understood
7284 to load a specific kernel module statically, early at boot.
7286 * Unit names specified on the systemctl command line are now
7287 automatically escaped as needed. Also, if file system or
7288 device paths are specified they are automatically turned
7289 into the appropriate mount or device unit names. Example:
7291 systemctl status /home
7292 systemctl status /dev/sda
7294 * The SysVConsole= configuration option has been removed from
7295 system.conf parsing.
7297 * The SysV search path is no longer exported on the D-Bus
7300 * The Names= option has been removed from unit file parsing.
7302 * There's a new man page bootup(7) detailing the boot process.
7304 * Every unit and every generator we ship with systemd now
7305 comes with full documentation. The self-explanatory boot is
7308 * A couple of services gained "systemd-" prefixes in their
7309 name if they wrap systemd code, rather than only external
7310 code. Among them fsck@.service which is now
7311 systemd-fsck@.service.
7313 * The HaveWatchdog property has been removed from the D-Bus
7316 * systemd.confirm_spawn= on the kernel command line should now
7319 * There's a new man page crypttab(5) which details all options
7320 we actually understand.
7322 * systemd-nspawn gained a new --capability= switch to pass
7323 additional capabilities to the container.
7325 * timedated will now read known NTP implementation unit names
7326 from /usr/lib/systemd/ntp-units.d/*.list,
7327 systemd-timedated-ntp.target has been removed.
7329 * journalctl gained a new switch "-b" that lists log data of
7330 the current boot only.
7332 * The notify socket is in the abstract namespace again, in
7333 order to support daemons which chroot() at start-up.
7335 * There is a new Storage= configuration option for journald
7336 which allows configuration of where log data should go. This
7337 also provides a way to disable journal logging entirely, so
7338 that data collected is only forwarded to the console, the
7339 kernel log buffer or another syslog implementation.
7341 * Many bugfixes and optimizations
7343 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7344 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7345 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7346 Shawn Landden, Tom Gundersen
7350 * "systemctl help <unit>" now shows the man page if one is
7353 * Several new man pages have been added.
7355 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7356 MaxLevelConsole= can now be specified in
7357 journald.conf. These options allow reducing the amount of
7358 data stored on disk or forwarded by the log level.
7360 * TimerSlackNSec= can now be specified in system.conf for
7361 PID1. This allows system-wide power savings.
7363 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7364 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7369 * logind is now capable of (optionally) handling power and
7370 sleep keys as well as the lid switch.
7372 * journalctl now understands the syntax "journalctl
7373 /usr/bin/avahi-daemon" to get all log output of a specific
7376 * CapabilityBoundingSet= in system.conf now also influences
7377 the capability bound set of usermode helpers of the kernel.
7379 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7380 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7381 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7382 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7386 * Note that we skipped 139 releases here in order to set the
7387 new version to something that is greater than both udev's
7388 and systemd's most recent version number.
7390 * udev: all udev sources are merged into the systemd source tree now.
7391 All future udev development will happen in the systemd tree. It
7392 is still fully supported to use the udev daemon and tools without
7393 systemd running, like in initramfs or other init systems. Building
7394 udev though, will require the *build* of the systemd tree, but
7395 udev can be properly *run* without systemd.
7397 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7398 should be used to create dead device nodes as workarounds for broken
7401 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7402 no longer supported. udev_monitor_new_from_netlink() needs to be
7403 used to subscribe to events.
7405 * udev: when udevd is started by systemd, processes which are left
7406 behind by forking them off of udev rules, are unconditionally cleaned
7407 up and killed now after the event handling has finished. Services or
7408 daemons must be started as systemd services. Services can be
7409 pulled-in by udev to get started, but they can no longer be directly
7410 forked by udev rules.
7412 * udev: the daemon binary is called systemd-udevd now and installed
7413 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7414 to adapt to that, create symlink, or rename the binary after building
7417 * libudev no longer provides these symbols:
7418 udev_monitor_from_socket()
7419 udev_queue_get_failed_list_entry()
7420 udev_get_{dev,sys,run}_path()
7421 The versions number was bumped and symbol versioning introduced.
7423 * systemd-loginctl and systemd-journalctl have been renamed
7424 to loginctl and journalctl to match systemctl.
7426 * The config files: /etc/systemd/systemd-logind.conf and
7427 /etc/systemd/systemd-journald.conf have been renamed to
7428 logind.conf and journald.conf. Package updates should rename
7429 the files to the new names on upgrade.
7431 * For almost all files the license is now LGPL2.1+, changed
7432 from the previous GPL2.0+. Exceptions are some minor stuff
7433 of udev (which will be changed to LGPL2.1 eventually, too),
7434 and the MIT licensed sd-daemon.[ch] library that is suitable
7435 to be used as drop-in files.
7437 * systemd and logind now handle system sleep states, in
7438 particular suspending and hibernating.
7440 * logind now implements a sleep/shutdown/idle inhibiting logic
7441 suitable for a variety of uses. Soonishly Lennart will blog
7442 about this in more detail.
7444 * var-run.mount and var-lock.mount are no longer provided
7445 (which previously bind mounted these directories to their new
7446 places). Distributions which have not converted these
7447 directories to symlinks should consider stealing these files
7448 from git history and add them downstream.
7450 * We introduced the Documentation= field for units and added
7451 this to all our shipped units. This is useful to make it
7452 easier to explore the boot and the purpose of the various
7455 * All smaller setup units (such as
7456 systemd-vconsole-setup.service) now detect properly if they
7457 are run in a container and are skipped when
7458 appropriate. This guarantees an entirely noise-free boot in
7459 Linux container environments such as systemd-nspawn.
7461 * A framework for implementing offline system updates is now
7462 integrated, for details see:
7463 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7465 * A new service type Type=idle is available now which helps us
7466 avoiding ugly interleaving of getty output and boot status
7469 * There's now a system-wide CapabilityBoundingSet= option to
7470 globally reduce the set of capabilities for the
7471 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7472 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7473 even CAP_NET_ADMIN system-wide for secure systems.
7475 * There are now system-wide DefaultLimitXXX= options to
7476 globally change the defaults of the various resource limits
7477 for all units started by PID 1.
7479 * Harald Hoyer's systemd test suite has been integrated into
7480 systemd which allows easy testing of systemd builds in qemu
7481 and nspawn. (This is really awesome! Ask us for details!)
7483 * The fstab parser is now implemented as generator, not inside
7486 * systemctl will now warn you if .mount units generated from
7487 /etc/fstab are out of date due to changes in fstab that
7488 have not been read by systemd yet.
7490 * systemd is now suitable for usage in initrds. Dracut has
7491 already been updated to make use of this. With this in place
7492 initrds get a slight bit faster but primarily are much
7493 easier to introspect and debug since "systemctl status" in
7494 the host system can be used to introspect initrd services,
7495 and the journal from the initrd is kept around too.
7497 * systemd-delta has been added, a tool to explore differences
7498 between user/admin configuration and vendor defaults.
7500 * PrivateTmp= now affects both /tmp and /var/tmp.
7502 * Boot time status messages are now much prettier and feature
7503 proper english language. Booting up systemd has never been
7506 * Read-ahead pack files now include the inode number of all
7507 files to pre-cache. When the inode changes the pre-caching
7508 is not attempted. This should be nicer to deal with updated
7509 packages which might result in changes of read-ahead
7512 * We now temporaritly lower the kernel's read_ahead_kb variable
7513 when collecting read-ahead data to ensure the kernel's
7514 built-in read-ahead does not add noise to our measurements
7515 of necessary blocks to pre-cache.
7517 * There's now RequiresMountsFor= to add automatic dependencies
7518 for all mounts necessary for a specific file system path.
7520 * MountAuto= and SwapAuto= have been removed from
7521 system.conf. Mounting file systems at boot has to take place
7524 * nspawn now learned a new switch --uuid= to set the machine
7525 ID on the command line.
7527 * nspawn now learned the -b switch to automatically search
7530 * vt102 is now the default TERM for serial TTYs, upgraded from
7533 * systemd-logind now works on VT-less systems.
7535 * The build tree has been reorganized. The individual
7536 components now have directories of their own.
7538 * A new condition type ConditionPathIsReadWrite= is now available.
7540 * nspawn learned the new -C switch to create cgroups for the
7541 container in other hierarchies.
7543 * We now have support for hardware watchdogs, configurable in
7546 * The scheduled shutdown logic now has a public API.
7548 * We now mount /tmp as tmpfs by default, but this can be
7549 masked and /etc/fstab can override it.
7551 * Since udisks does not make use of /media anymore we are not
7552 mounting a tmpfs on it anymore.
7554 * journalctl gained a new --local switch to only interleave
7555 locally generated journal files.
7557 * We can now load the IMA policy at boot automatically.
7559 * The GTK tools have been split off into a systemd-ui.
7561 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7562 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7563 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7564 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7565 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7566 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7567 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7568 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7569 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7574 * This is mostly a bugfix release
7576 * Support optional initialization of the machine ID from the
7577 KVM or container configured UUID.
7579 * Support immediate reboots with "systemctl reboot -ff"
7581 * Show /etc/os-release data in systemd-analyze output
7583 * Many bugfixes for the journal, including endianness fixes and
7584 ensuring that disk space enforcement works
7586 * sd-login.h is C++ compatible again
7588 * Extend the /etc/os-release format on request of the Debian
7591 * We now refuse non-UTF8 strings used in various configuration
7592 and unit files. This is done to ensure we do not pass invalid
7593 data over D-Bus or expose it elsewhere.
7595 * Register Mimo USB Screens as suitable for automatic seat
7598 * Read SELinux client context from journal clients in a race
7601 * Reorder configuration file lookup order. /etc now always
7602 overrides /run in order to allow the administrator to always
7603 and unconditionally override vendor-supplied or
7604 automatically generated data.
7606 * The various user visible bits of the journal now have man
7607 pages. We still lack man pages for the journal API calls
7610 * We now ship all man pages in HTML format again in the
7613 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7614 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7615 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7620 * This is mostly a bugfix release
7622 * systems lacking /etc/os-release are no longer supported.
7624 * Various functionality updates to libsystemd-login.so
7626 * Track class of PAM logins to distinguish greeters from
7629 Contributions from: Kay Sievers, Lennart Poettering, Michael
7634 * This is an important bugfix release for v41.
7636 * Building man pages is now optional which should be useful
7637 for those building systemd from git but unwilling to install
7640 * Watchdog support for supervising services is now usable. In
7641 a future release support for hardware watchdogs
7642 (i.e. /dev/watchdog) will be added building on this.
7644 * Service start rate limiting is now configurable and can be
7645 turned off per service. When a start rate limit is hit a
7646 reboot can automatically be triggered.
7648 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7650 Contributions from: Benjamin Franzke, Bill Nottingham,
7651 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7652 Schmidt, Michał Górny, Piotr Drąg
7656 * The systemd binary is installed /usr/lib/systemd/systemd now;
7657 An existing /sbin/init symlink needs to be adapted with the
7660 * The code that loads kernel modules has been ported to invoke
7661 libkmod directly, instead of modprobe. This means we do not
7662 support systems with module-init-tools anymore.
7664 * Watchdog support is now already useful, but still not
7667 * A new kernel command line option systemd.setenv= is
7668 understood to set system wide environment variables
7669 dynamically at boot.
7671 * We now limit the set of capabilities of systemd-journald.
7673 * We now set SIGPIPE to ignore by default, since it only is
7674 useful in shell pipelines, and has little use in general
7675 code. This can be disabled with IgnoreSIPIPE=no in unit
7678 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7679 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7684 * This is mostly a bugfix release
7686 * We now expose the reason why a service failed in the
7687 "Result" D-Bus property.
7689 * Rudimentary service watchdog support (will be completed over
7690 the next few releases.)
7692 * When systemd forks off in order execute some service we will
7693 now immediately changes its argv[0] to reflect which process
7694 it will execute. This is useful to minimize the time window
7695 with a generic argv[0], which makes bootcharts more useful
7697 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7698 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7699 Mike Kazantsev, Ray Strode
7703 * This is mostly a test release, but incorporates many
7706 * New systemd-cgtop tool to show control groups by their
7709 * Linking against libacl for ACLs is optional again. If
7710 disabled, support tracking device access for active logins
7711 goes becomes unavailable, and so does access to the user
7712 journals by the respective users.
7714 * If a group "adm" exists, journal files are automatically
7715 owned by them, thus allow members of this group full access
7716 to the system journal as well as all user journals.
7718 * The journal now stores the SELinux context of the logging
7719 client for all entries.
7721 * Add C++ inclusion guards to all public headers
7723 * New output mode "cat" in the journal to print only text
7724 messages, without any meta data like date or time.
7726 * Include tiny X server wrapper as a temporary stop-gap to
7727 teach XOrg udev display enumeration. This is used by display
7728 managers such as gdm, and will go away as soon as XOrg
7729 learned native udev hotplugging for display devices.
7731 * Add new systemd-cat tool for executing arbitrary programs
7732 with STDERR/STDOUT connected to the journal. Can also act as
7733 BSD logger replacement, and does so by default.
7735 * Optionally store all locally generated coredumps in the
7736 journal along with meta data.
7738 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7739 writing short strings to files (for usage for /sys), and for
7740 creating symlinks, character and block device nodes.
7742 * New unit file option ControlGroupPersistent= to make cgroups
7743 persistent, following the mechanisms outlined in
7744 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7746 * Support multiple local RTCs in a sane way
7748 * No longer monopolize IO when replaying readahead data on
7749 rotating disks, since we might starve non-file-system IO to
7750 death, since fanotify() will not see accesses done by blkid,
7753 * Do not show kernel threads in systemd-cgls anymore, unless
7754 requested with new -k switch.
7756 Contributions from: Dan Horák, Kay Sievers, Lennart
7757 Poettering, Michal Schmidt
7761 * This is mostly a test release, but incorporates many
7764 * The git repository moved to:
7765 git://anongit.freedesktop.org/systemd/systemd
7766 ssh://git.freedesktop.org/git/systemd/systemd
7768 * First release with the journal
7769 http://0pointer.de/blog/projects/the-journal.html
7771 * The journal replaces both systemd-kmsg-syslogd and
7772 systemd-stdout-bridge.
7774 * New sd_pid_get_unit() API call in libsystemd-logind
7776 * Many systemadm clean-ups
7778 * Introduce remote-fs-pre.target which is ordered before all
7779 remote mounts and may be used to start services before all
7782 * Added Mageia support
7784 * Add bash completion for systemd-loginctl
7786 * Actively monitor PID file creation for daemons which exit in
7787 the parent process before having finished writing the PID
7788 file in the daemon process. Daemons which do this need to be
7789 fixed (i.e. PID file creation must have finished before the
7790 parent exits), but we now react a bit more gracefully to them.
7792 * Add colourful boot output, mimicking the well-known output
7793 of existing distributions.
7795 * New option PassCredentials= for socket units, for
7796 compatibility with a recent kernel ABI breakage.
7798 * /etc/rc.local is now hooked in via a generator binary, and
7799 thus will no longer act as synchronization point during
7802 * systemctl list-unit-files now supports --root=.
7804 * systemd-tmpfiles now understands two new commands: z, Z for
7805 relabelling files according to the SELinux database. This is
7806 useful to apply SELinux labels to specific files in /sys,
7809 * Output of SysV services is now forwarded to both the console
7810 and the journal by default, not only just the console.
7812 * New man pages for all APIs from libsystemd-login.
7814 * The build tree got reorganized and the build system is a
7815 lot more modular allowing embedded setups to specifically
7816 select the components of systemd they are interested in.
7818 * Support for Linux systems lacking the kernel VT subsystem is
7821 * configure's --with-rootdir= got renamed to
7822 --with-rootprefix= to follow the naming used by udev and
7825 * Unless specified otherwise we will now install to /usr instead
7826 of /usr/local by default.
7828 * Processes with '@' in argv[0][0] are now excluded from the
7829 final shut-down killing spree, following the logic explained
7831 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7833 * All processes remaining in a service cgroup when we enter
7834 the START or START_PRE states are now killed with
7835 SIGKILL. That means it is no longer possible to spawn
7836 background processes from ExecStart= lines (which was never
7837 supported anyway, and bad style).
7839 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7840 reloading of units together.
7842 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7843 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7844 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7845 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7846 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob