1 systemd System and Service Manager
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 * systemd-resolved's DNSSEC support gained support for RFC 8080
189 (Ed25519 keys and signatures).
191 * The systemd-resolve command line tool gained a new set of options
192 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
193 --set-nta= and --revert to configure per-interface DNS configuration
194 dynamically during runtime. It's useful for pushing DNS information
195 into systemd-resolved from DNS hook scripts that various interface
196 managing software supports (such as pppd).
198 * systemd-nspawn gained a new --network-namespace-path= command line
199 option, which may be used to make a container join an existing
200 network namespace, by specifying a path to a "netns" file.
202 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
203 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
204 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
205 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
206 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
207 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
208 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
209 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
210 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
211 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
212 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
213 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
214 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
215 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
216 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
217 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
218 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
219 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
220 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
221 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
222 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
223 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
224 Jędrzejewski-Szmek, Zeal Jagannatha
230 * INCOMPATIBILITY: systemd-logind.service and other long-running
231 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
232 communication with the outside. This generally improves security of
233 the system, and is in almost all cases a safe and good choice, as
234 these services do not and should not provide any network-facing
235 functionality. However, systemd-logind uses the glibc NSS API to
236 query the user database. This creates problems on systems where NSS
237 is set up to directly consult network services for user database
238 lookups. In particular, this creates incompatibilities with the
239 "nss-nis" module, which attempts to directly contact the NIS/YP
240 network servers it is configured for, and will now consistently
241 fail. In such cases, it is possible to turn off IP sandboxing for
242 systemd-logind.service (set IPAddressDeny= in its [Service] section
243 to the empty string, via a .d/ unit file drop-in). Downstream
244 distributions might want to update their nss-nis packaging to include
245 such a drop-in snippet, accordingly, to hide this incompatibility
246 from the user. Another option is to make use of glibc's nscd service
247 to proxy such network requests through a privilege-separated, minimal
248 local caching daemon, or to switch to more modern technologies such
249 sssd, whose NSS hook-ups generally do not involve direct network
250 access. In general, we think it's definitely time to question the
251 implementation choices of nss-nis, i.e. whether it's a good idea
252 today to embed a network-facing loadable module into all local
253 processes that need to query the user database, including the most
254 trivial and benign ones, such as "ls". For more details about
255 IPAddressDeny= see below.
257 * A new modprobe.d drop-in is now shipped by default that sets the
258 bonding module option max_bonds=0. This overrides the kernel default,
259 to avoid conflicts and ambiguity as to whether or not bond0 should be
260 managed by systemd-networkd or not. This resolves multiple issues
261 with bond0 properties not being applied, when bond0 is configured
262 with systemd-networkd. Distributors may choose to not package this,
263 however in that case users will be prevented from correctly managing
264 bond0 interface using systemd-networkd.
266 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
267 which print the logging level and target of the system manager. They
268 complement the existing "set-log-level" and "set-log-target" verbs
269 used to change those values.
271 * journald.conf gained a new boolean setting ReadKMsg= which defaults
272 to on. If turned off kernel log messages will not be read by
273 systemd-journald or included in the logs. It also gained a new
274 setting LineMax= for configuring the maximum line length in
275 STDOUT/STDERR log streams. The new default for this value is 48K, up
276 from the previous hardcoded 2048.
278 * A new unit setting RuntimeDirectoryPreserve= has been added, which
279 allows more detailed control of what to do with a runtime directory
280 configured with RuntimeDirectory= (i.e. a directory below /run or
281 $XDG_RUNTIME_DIR) after a unit is stopped.
283 * The RuntimeDirectory= setting for units gained support for creating
284 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
285 one top-level directory.
287 * Units gained new options StateDirectory=, CacheDirectory=,
288 LogsDirectory= and ConfigurationDirectory= which are closely related
289 to RuntimeDirectory= but manage per-service directories below
290 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
291 possible to write unit files which when activated automatically gain
292 properly owned service specific directories in these locations, thus
293 making unit files self-contained and increasing compatibility with
294 stateless systems and factory reset where /etc or /var are
295 unpopulated at boot. Matching these new settings there's also
296 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
297 ConfigurationDirectoryMode= for configuring the access mode of these
298 directories. These settings are particularly useful in combination
299 with DynamicUser=yes as they provide secure, properly-owned,
300 writable, and stateful locations for storage, excluded from the
301 sandbox that such services live in otherwise.
303 * Automake support has been removed from this release. systemd is now
306 * systemd-journald will now aggressively cache client metadata during
307 runtime, speeding up log write performance under pressure. This comes
308 at a small price though: as much of the metadata is read
309 asynchronously from /proc/ (and isn't implicitly attached to log
310 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
311 metadata stored alongside a log entry might be slightly
312 out-of-date. Previously it could only be slightly newer than the log
313 message. The time window is small however, and given that the kernel
314 is unlikely to be improved anytime soon in this regard, this appears
317 * nss-myhostname/systemd-resolved will now by default synthesize an
318 A/AAAA resource record for the "_gateway" hostname, pointing to the
319 current default IP gateway. Previously it did that for the "gateway"
320 name, hampering adoption, as some distributions wanted to leave that
321 host name open for local use. The old behaviour may still be
322 requested at build time.
324 * systemd-networkd's [Address] section in .network files gained a new
325 Scope= setting for configuring the IP address scope. The [Network]
326 section gained a new boolean setting ConfigureWithoutCarrier= that
327 tells systemd-networkd to ignore link sensing when configuring the
328 device. The [DHCP] section gained a new Anonymize= boolean option for
329 turning on a number of options suggested in RFC 7844. A new
330 [RoutingPolicyRule] section has been added for configuring the IP
331 routing policy. The [Route] section has gained support for a new
332 Type= setting which permits configuring
333 blackhole/unreachable/prohibit routes.
335 * The [VRF] section in .netdev files gained a new Table= setting for
336 configuring the routing table to use. The [Tunnel] section gained a
337 new Independent= boolean field for configuring tunnels independent of
338 an underlying network interface. The [Bridge] section gained a new
339 GroupForwardMask= option for configuration of propagation of link
340 local frames between bridge ports.
342 * The WakeOnLan= setting in .link files gained support for a number of
343 new modes. A new TCP6SegmentationOffload= setting has been added for
344 configuring TCP/IPv6 hardware segmentation offload.
346 * The IPv6 RA sender implementation may now optionally send out RDNSS
347 and RDNSSL records to supply DNS configuration to peers.
349 * systemd-nspawn gained support for a new --system-call-filter= command
350 line option for adding and removing entries in the default system
351 call filter it applies. Moreover systemd-nspawn has been changed to
352 implement a system call whitelist instead of a blacklist.
354 * systemd-run gained support for a new --pipe command line option. If
355 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
356 are directly passed on to the activated transient service
357 executable. This allows invoking arbitrary processes as systemd
358 services (for example to take benefit of dependency management,
359 accounting management, resource management or log management that is
360 done automatically for services) — while still allowing them to be
361 integrated in a classic UNIX shell pipeline.
363 * When a service sends RELOAD=1 via sd_notify() and reload propagation
364 using ReloadPropagationTo= is configured, a reload is now propagated
365 to configured units. (Previously this was only done on explicitly
366 requested reloads, using "systemctl reload" or an equivalent
369 * For each service unit a restart counter is now kept: it is increased
370 each time the service is restarted due to Restart=, and may be
371 queried using "systemctl show -p NRestarts …".
373 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
374 @signal and @timer have been added, for usage with SystemCallFilter=
375 in unit files and the new --system-call-filter= command line option
376 of systemd-nspawn (see above).
378 * ExecStart= lines in unit files gained two new modifiers: when a
379 command line is prefixed with "!" the command will be executed as
380 configured, except for the credentials applied by
381 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
382 "+", but does still apply namespacing options unlike "+". There's
383 also "!!" now, which is mostly identical, but becomes a NOP on
384 systems that support ambient capabilities. This is useful to write
385 unit files that work with ambient capabilities where possible but
386 automatically fall back to traditional privilege dropping mechanisms
387 on systems where this is not supported.
389 * ListenNetlink= settings in socket units now support RDMA netlink
392 * A new unit file setting LockPersonality= has been added which permits
393 locking down the chosen execution domain ("personality") of a service
396 * A new special target "getty-pre.target" has been added, which is
397 ordered before all text logins, and may be used to order services
398 before textual logins acquire access to the console.
400 * systemd will now attempt to load the virtio-rng.ko kernel module very
401 early on if a VM environment supporting this is detected. This should
402 improve entropy during early boot in virtualized environments.
404 * A _netdev option is now supported in /etc/crypttab that operates in a
405 similar way as the same option in /etc/fstab: it permits configuring
406 encrypted devices that need to be ordered after the network is up.
407 Following this logic, two new special targets
408 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
409 added that are to cryptsetup.target what remote-fs.target and
410 remote-fs-pre.target are to local-fs.target.
412 * Service units gained a new UnsetEnvironment= setting which permits
413 unsetting specific environment variables for services that are
414 normally passed to it (for example in order to mask out locale
415 settings for specific services that can't deal with it).
417 * Units acquired a new boolean option IPAccounting=. When turned on, IP
418 traffic accounting (packet count as well as byte count) is done for
419 the service, and shown as part of "systemctl status" or "systemd-run
422 * Service units acquired two new options IPAddressAllow= and
423 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
424 for configuring a simple IP access control list for all sockets of
425 the unit. These options are available also on .slice and .socket
426 units, permitting flexible access list configuration for individual
427 services as well as groups of services (as defined by a slice unit),
428 including system-wide. Note that IP ACLs configured this way are
429 enforced on every single IPv4 and IPv6 socket created by any process
430 of the service unit, and apply to ingress as well as egress traffic.
432 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
433 structured log message is generated each time the unit is stopped,
434 containing information about the consumed resources of this
437 * A new setting KeyringMode= has been added to unit files, which may be
438 used to control how the kernel keyring is set up for executed
441 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
442 "systemctl kexec" and "systemctl exit" are now always asynchronous in
443 behaviour (that is: these commands return immediately after the
444 operation was enqueued instead of waiting for the operation to
445 complete). Previously, "systemctl poweroff" and "systemctl reboot"
446 were asynchronous on systems using systemd-logind (i.e. almost
447 always, and like they were on sysvinit), and the other three commands
448 were unconditionally synchronous. With this release this is cleaned
449 up, and callers will see the same asynchronous behaviour on all
450 systems for all five operations.
452 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
455 * .timer units now accept calendar specifications in other timezones
456 than UTC or the local timezone.
458 * The tmpfiles snippet var.conf has been changed to create
459 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
460 the "utmp" group already, and it appears to be generally understood
461 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
462 databases. Previously this was implemented correctly for all these
463 databases excepts btmp, which has been opened up like this now
464 too. Note that while the other databases are world-readable
465 (i.e. 0644), btmp is not and remains more restrictive.
467 * The systemd-resolve tool gained a new --reset-server-features
468 switch. When invoked like this systemd-resolved will forget
469 everything it learnt about the features supported by the configured
470 upstream DNS servers, and restarts the feature probing logic on the
471 next resolver look-up for them at the highest feature level
474 * The status dump systemd-resolved sends to the logs upon receiving
475 SIGUSR1 now also includes information about all DNS servers it is
476 configured to use, and the features levels it probed for them.
478 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
479 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
480 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
481 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
482 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
483 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
484 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
485 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
486 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
487 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
488 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
489 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
490 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
491 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
492 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
493 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
494 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
495 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
496 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
497 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
503 * Meson is now supported as build system in addition to Automake. It is
504 our plan to remove Automake in one of our next releases, so that
505 Meson becomes our exclusive build system. Hence, please start using
506 the Meson build system in your downstream packaging. There's plenty
507 of documentation around how to use Meson, the extremely brief
510 ./autogen.sh && ./configure && make && sudo make install
514 meson build && ninja -C build && sudo ninja -C build install
516 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
517 which permits configuring a timeout on the time a job is
518 running. This is particularly useful for setting timeouts on jobs for
521 * Unit files gained two new options ConditionUser= and ConditionGroup=
522 for conditionalizing units based on the identity of the user/group
523 running a systemd user instance.
525 * systemd-networkd now understands a new FlowLabel= setting in the
526 [VXLAN] section of .network files, as well as a Priority= in
527 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
528 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
529 gained support for configuration of GENEVE links, and IPv6 address
530 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
532 * .link files now understand a new Port= setting.
534 * systemd-networkd's DHCP support gained support for DHCP option 119
535 (domain search list).
537 * systemd-networkd gained support for serving IPv6 address ranges using
538 the Router Advertisment protocol. The new .network configuration
539 section [IPv6Prefix] may be used to configure the ranges to
540 serve. This is implemented based on a new, minimal, native server
541 implementation of RA.
543 * journalctl's --output= switch gained support for a new parameter
544 "short-iso-precise" for a mode where timestamps are shown as precise
547 * systemd-udevd's "net_id" builtin may now generate stable network
548 interface names from IBM PowerVM VIO devices as well as ACPI platform
551 * MulticastDNS support in systemd-resolved may now be explicitly
552 enabled/disabled using the new MulticastDNS= configuration file
555 * systemd-resolved may now optionally use libidn2 instead of the libidn
556 for processing internationalized domain names. Support for libidn2
557 should be considered experimental and should not be enabled by
560 * "machinectl pull-tar" and related call may now do verification of
561 downloaded images using SUSE-style .sha256 checksum files in addition
562 to the already existing support for validating using Ubuntu-style
565 * sd-bus gained support for a new sd_bus_message_appendv() call which
566 is va_list equivalent of sd_bus_message_append().
568 * sd-boot gained support for validating images using SHIM/MOK.
570 * The SMACK code learnt support for "onlycap".
572 * systemd-mount --umount is now much smarter in figuring out how to
573 properly unmount a device given its mount or device path.
575 * The code to call libnss_dns as a fallback from libnss_resolve when
576 the communication with systemd-resolved fails was removed. This
577 fallback was redundant and interfered with the [!UNAVAIL=return]
578 suffix. See nss-resolve(8) for the recommended configuration.
580 * systemd-logind may now be restarted without losing state. It stores
581 the file descriptors for devices it manages in the system manager
582 using the FDSTORE= mechanism. Please note that further changes in
583 other components may be required to make use of this (for example
584 Xorg has code to listen for stops of systemd-logind and terminate
585 itself when logind is stopped or restarted, in order to avoid using
586 stale file descriptors for graphical devices, which is now
587 counterproductive and must be reverted in order for restarts of
588 systemd-logind to be safe. See
589 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
591 * All kernel install plugins are called with the environment variable
592 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
593 /etc/machine-id. If the file is missing or empty, the variable is
594 empty and BOOT_DIR_ABS is the path of a temporary directory which is
595 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
596 is empty, all plugins should not put anything in BOOT_DIR_ABS.
598 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
599 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
600 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
601 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
602 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
603 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
604 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
605 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
606 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
607 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
608 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
609 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
610 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
611 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
612 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
613 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
614 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
615 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
616 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
617 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
618 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
619 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
620 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
621 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
622 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
623 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
624 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
625 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
626 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
633 * The "hybrid" control group mode has been modified to improve
634 compatibility with "legacy" cgroups-v1 setups. Specifically, the
635 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
636 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
637 cgroups-v1 hierarchy), the only externally visible change being that
638 the cgroups-v2 hierarchy is also mounted, to
639 /sys/fs/cgroup/unified. This should provide a large degree of
640 compatibility with "legacy" cgroups-v1, while taking benefit of the
641 better management capabilities of cgroups-v2.
643 * The default control group setup mode may be selected both a boot-time
644 via a set of kernel command line parameters (specifically:
645 systemd.unified_cgroup_hierarchy= and
646 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
647 default selected on the configure command line
648 (--with-default-hierarchy=). The upstream default is "hybrid"
649 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
650 this will change in a future systemd version to be "unified" (pure
651 cgroups-v2 mode). The third option for the compile time option is
652 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
653 distributions to default to "hybrid" mode for release distributions,
654 starting with v233. We recommend "unified" for development
655 distributions (specifically: distributions such as Fedora's rawhide)
656 as that's where things are headed in the long run. Use "legacy" for
657 greatest stability and compatibility only.
659 * Note one current limitation of "unified" and "hybrid" control group
660 setup modes: the kernel currently does not permit the systemd --user
661 instance (i.e. unprivileged code) to migrate processes between two
662 disconnected cgroup subtrees, even if both are managed and owned by
663 the user. This effectively means "systemd-run --user --scope" doesn't
664 work when invoked from outside of any "systemd --user" service or
665 scope. Specifically, it is not supported from session scopes. We are
666 working on fixing this in a future systemd version. (See #3388 for
667 further details about this.)
669 * DBus policy files are now installed into /usr rather than /etc. Make
670 sure your system has dbus >= 1.9.18 running before upgrading to this
671 version, or override the install path with --with-dbuspolicydir= .
673 * All python scripts shipped with systemd (specifically: the various
674 tests written in Python) now require Python 3.
676 * systemd unit tests can now run standalone (without the source or
677 build directories), and can be installed into /usr/lib/systemd/tests/
678 with 'make install-tests'.
680 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
681 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
684 * Support for the %c, %r, %R specifiers in unit files has been
685 removed. Specifiers are not supposed to be dependent on configuration
686 in the unit file itself (so that they resolve the same regardless
687 where used in the unit files), but these specifiers were influenced
688 by the Slice= option.
690 * The shell invoked by debug-shell.service now defaults to /bin/sh in
691 all cases. If distributions want to use a different shell for this
692 purpose (for example Fedora's /sbin/sushell) they need to specify
693 this explicitly at configure time using --with-debug-shell=.
695 * The confirmation spawn prompt has been reworked to offer the
698 (c)ontinue, proceed without asking anymore
699 (D)ump, show the state of the unit
700 (f)ail, don't execute the command and pretend it failed
702 (i)nfo, show a short summary of the unit
703 (j)obs, show jobs that are in progress
704 (s)kip, don't execute the command and pretend it succeeded
705 (y)es, execute the command
707 The 'n' choice for the confirmation spawn prompt has been removed,
708 because its meaning was confusing.
710 The prompt may now also be redirected to an alternative console by
711 specifying the console as parameter to systemd.confirm_spawn=.
713 * Services of Type=notify require a READY=1 notification to be sent
714 during startup. If no such message is sent, the service now fails,
715 even if the main process exited with a successful exit code.
717 * Services that fail to start up correctly now always have their
718 ExecStopPost= commands executed. Previously, they'd enter "failed"
719 state directly, without executing these commands.
721 * The option MulticastDNS= of network configuration files has acquired
722 an actual implementation. With MulticastDNS=yes a host can resolve
723 names of remote hosts and reply to mDNS A and AAAA requests.
725 * When units are about to be started an additional check is now done to
726 ensure that all dependencies of type BindsTo= (when used in
727 combination with After=) have been started.
729 * systemd-analyze gained a new verb "syscall-filter" which shows which
730 system call groups are defined for the SystemCallFilter= unit file
731 setting, and which system calls they contain.
733 * A new system call filter group "@filesystem" has been added,
734 consisting of various file system related system calls. Group
735 "@reboot" has been added, covering reboot, kexec and shutdown related
736 calls. Finally, group "@swap" has been added covering swap
737 configuration related calls.
739 * A new unit file option RestrictNamespaces= has been added that may be
740 used to restrict access to the various process namespace types the
741 Linux kernel provides. Specifically, it may be used to take away the
742 right for a service unit to create additional file system, network,
743 user, and other namespaces. This sandboxing option is particularly
744 relevant due to the high amount of recently discovered namespacing
745 related vulnerabilities in the kernel.
747 * systemd-udev's .link files gained support for a new AutoNegotiation=
748 setting for configuring Ethernet auto-negotiation.
750 * systemd-networkd's .network files gained support for a new
751 ListenPort= setting in the [DHCP] section to explicitly configure the
752 UDP client port the DHCP client shall listen on.
754 * .network files gained a new Unmanaged= boolean setting for explicitly
755 excluding one or more interfaces from management by systemd-networkd.
757 * The systemd-networkd ProxyARP= option has been renamed to
758 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
759 renamed to ReduceARPProxy=. The old names continue to be available
762 * systemd-networkd gained support for configuring IPv6 Proxy NDP
763 addresses via the new IPv6ProxyNDPAddress= .network file setting.
765 * systemd-networkd's bonding device support gained support for two new
766 configuration options ActiveSlave= and PrimarySlave=.
768 * The various options in the [Match] section of .network files gained
769 support for negative matching.
771 * New systemd-specific mount options are now understood in /etc/fstab:
773 x-systemd.mount-timeout= may be used to configure the maximum
774 permitted runtime of the mount command.
776 x-systemd.device-bound may be set to bind a mount point to its
777 backing device unit, in order to automatically remove a mount point
778 if its backing device is unplugged. This option may also be
779 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
780 on the block device, which is now automatically set for all CDROM
781 drives, so that mounted CDs are automatically unmounted when they are
782 removed from the drive.
784 x-systemd.after= and x-systemd.before= may be used to explicitly
785 order a mount after or before another unit or mount point.
787 * Enqueued start jobs for device units are now automatically garbage
788 collected if there are no jobs waiting for them anymore.
790 * systemctl list-jobs gained two new switches: with --after, for every
791 queued job the jobs it's waiting for are shown; with --before the
792 jobs which it's blocking are shown.
794 * systemd-nspawn gained support for ephemeral boots from disk images
795 (or in other words: --ephemeral and --image= may now be
796 combined). Moreover, ephemeral boots are now supported for normal
797 directories, even if the backing file system is not btrfs. Of course,
798 if the file system does not support file system snapshots or
799 reflinks, the initial copy operation will be relatively expensive, but
800 this should still be suitable for many use cases.
802 * Calendar time specifications in .timer units now support
803 specifications relative to the end of a month by using "~" instead of
804 "-" as separator between month and day. For example, "*-02~03" means
805 "the third last day in February". In addition a new syntax for
806 repeated events has been added using the "/" character. For example,
807 "9..17/2:00" means "every two hours from 9am to 5pm".
809 * systemd-socket-proxyd gained a new parameter --connections-max= for
810 configuring the maximum number of concurrent connections.
812 * sd-id128 gained a new API for generating unique IDs for the host in a
813 way that does not leak the machine ID. Specifically,
814 sd_id128_get_machine_app_specific() derives an ID based on the
815 machine ID a in well-defined, non-reversible, stable way. This is
816 useful whenever an identifier for the host is needed but where the
817 identifier shall not be useful to identify the system beyond the
818 scope of the application itself. (Internally this uses HMAC-SHA256 as
819 keyed hash function using the machine ID as input.)
821 * NotifyAccess= gained a new supported value "exec". When set
822 notifications are accepted from all processes systemd itself invoked,
823 including all control processes.
825 * .nspawn files gained support for defining overlay mounts using the
826 Overlay= and OverlayReadOnly= options. Previously this functionality
827 was only available on the systemd-nspawn command line.
829 * systemd-nspawn's --bind= and --overlay= options gained support for
830 bind/overlay mounts whose source lies within the container tree by
831 prefixing the source path with "+".
833 * systemd-nspawn's --bind= and --overlay= options gained support for
834 automatically allocating a temporary source directory in /var/tmp
835 that is removed when the container dies. Specifically, if the source
836 directory is specified as empty string this mechanism is selected. An
837 example usage is --overlay=+/var::/var, which creates an overlay
838 mount based on the original /var contained in the image, overlayed
839 with a temporary directory in the host's /var/tmp. This way changes
840 to /var are automatically flushed when the container shuts down.
842 * systemd-nspawn --image= option does now permit raw file system block
843 devices (in addition to images containing partition tables, as
846 * The disk image dissection logic in systemd-nspawn gained support for
847 automatically setting up LUKS encrypted as well as Verity protected
848 partitions. When a container is booted from an encrypted image the
849 passphrase is queried at start-up time. When a container with Verity
850 data is started, the root hash is search in a ".roothash" file
851 accompanying the disk image (alternatively, pass the root hash via
852 the new --root-hash= command line option).
854 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
855 be used to dissect disk images the same way as systemd-nspawn does
856 it, following the Bootable Partition Specification. It may even be
857 used to mount disk images with complex partition setups (including
858 LUKS and Verity partitions) to a local host directory, in order to
859 inspect them. This tool is not considered public API (yet), and is
860 thus not installed into /usr/bin. Please do not rely on its
861 existence, since it might go away or be changed in later systemd
864 * A new generator "systemd-verity-generator" has been added, similar in
865 style to "systemd-cryptsetup-generator", permitting automatic setup of
866 Verity root partitions when systemd boots up. In order to make use of
867 this your partition setup should follow the Discoverable Partitions
868 Specification, and the GPT partition ID of the root file system
869 partition should be identical to the upper 128bit of the Verity root
870 hash. The GPT partition ID of the Verity partition protecting it
871 should be the lower 128bit of the Verity root hash. If the partition
872 image follows this model it is sufficient to specify a single
873 "roothash=" kernel command line argument to both configure which root
874 image and verity partition to use as well as the root hash for
875 it. Note that systemd-nspawn's Verity support follows the same
876 semantics, meaning that disk images with proper Verity data in place
877 may be booted in containers with systemd-nspawn as well as on
878 physical systems via the verity generator. Also note that the "mkosi"
879 tool available at https://github.com/systemd/mkosi has been updated
880 to generate Verity protected disk images following this scheme. In
881 fact, it has been updated to generate disk images that optionally
882 implement a complete UEFI SecureBoot trust chain, involving a signed
883 kernel and initrd image that incorporates such a root hash as well as
884 a Verity-enabled root partition.
886 * The hardware database (hwdb) udev supports has been updated to carry
887 accelerometer quirks.
889 * All system services are now run with a fresh kernel keyring set up
890 for them. The invocation ID is stored by default in it, thus
891 providing a safe, non-overridable way to determine the invocation
894 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
895 options for bind mounting arbitrary paths in a service-specific
896 way. When these options are used, arbitrary host or service files and
897 directories may be mounted to arbitrary locations in the service's
900 * Documentation has been added that lists all of systemd's low-level
901 environment variables:
903 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
905 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
906 whether a specific socket file descriptor matches a specified socket
909 * systemd-firstboot has been updated to check for the
910 systemd.firstboot= kernel command line option. It accepts a boolean
911 and when set to false the first boot questions are skipped.
913 * systemd-fstab-generator has been updated to check for the
914 systemd.volatile= kernel command line option, which either takes an
915 optional boolean parameter or the special value "state". If used the
916 system may be booted in a "volatile" boot mode. Specifically,
917 "systemd.volatile" is used, the root directory will be mounted as
918 tmpfs, and only /usr is mounted from the actual root file system. If
919 "systemd.volatile=state" is used, the root directory will be mounted
920 as usual, but /var is mounted as tmpfs. This concept provides similar
921 functionality as systemd-nspawn's --volatile= option, but provides it
922 on physical boots. Use this option for implementing stateless
923 systems, or testing systems with all state and/or configuration reset
924 to the defaults. (Note though that many distributions are not
925 prepared to boot up without a populated /etc or /var, though.)
927 * systemd-gpt-auto-generator gained support for LUKS encrypted root
928 partitions. Previously it only supported LUKS encrypted partitions
929 for all other uses, except for the root partition itself.
931 * Socket units gained support for listening on AF_VSOCK sockets for
932 communication in virtualized QEMU environments.
934 * The "configure" script gained a new option --with-fallback-hostname=
935 for specifying the fallback hostname to use if none is configured in
936 /etc/hostname. For example, by specifying
937 --with-fallback-hostname=fedora it is possible to default to a
938 hostname of "fedora" on pristine installations.
940 * systemd-cgls gained support for a new --unit= switch for listing only
941 the control groups of a specific unit. Similar --user-unit= has been
942 added for listing only the control groups of a specific user unit.
944 * systemd-mount gained a new --umount switch for unmounting a mount or
945 automount point (and all mount/automount points below it).
947 * systemd will now refuse full configuration reloads (via systemctl
948 daemon-reload and related calls) unless at least 16MiB of free space
949 are available in /run. This is a safety precaution in order to ensure
950 that generators can safely operate after the reload completed.
952 * A new unit file option RootImage= has been added, which has a similar
953 effect as RootDirectory= but mounts the service's root directory from
954 a disk image instead of plain directory. This logic reuses the same
955 image dissection and mount logic that systemd-nspawn already uses,
956 and hence supports any disk images systemd-nspawn supports, including
957 those following the Discoverable Partition Specification, as well as
958 Verity enabled images. This option enables systemd to run system
959 services directly off disk images acting as resource bundles,
960 possibly even including full integrity data.
962 * A new MountAPIVFS= unit file option has been added, taking a boolean
963 argument. If enabled /proc, /sys and /dev (collectively called the
964 "API VFS") will be mounted for the service. This is only relevant if
965 RootDirectory= or RootImage= is used for the service, as these mounts
966 are of course in place in the host mount namespace anyway.
968 * systemd-nspawn gained support for a new --pivot-root= switch. If
969 specified the root directory within the container image is pivoted to
970 the specified mount point, while the original root disk is moved to a
971 different place. This option enables booting of ostree images
972 directly with systemd-nspawn.
974 * The systemd build scripts will no longer complain if the NTP server
975 addresses are not changed from the defaults. Google now supports
976 these NTP servers officially. We still recommend downstreams to
977 properly register an NTP pool with the NTP pool project though.
979 * coredumpctl gained a new "--reverse" option for printing the list
980 of coredumps in reverse order.
982 * coredumpctl will now show additional information about truncated and
983 inaccessible coredumps, as well as coredumps that are still being
984 processed. It also gained a new --quiet switch for suppressing
985 additional informational message in its output.
987 * coredumpctl gained support for only showing coredumps newer and/or
988 older than specific timestamps, using the new --since= and --until=
989 options, reminiscent of journalctl's options by the same name.
991 * The systemd-coredump logic has been improved so that it may be reused
992 to collect backtraces in non-compiled languages, for example in
993 scripting languages such as Python.
995 * machinectl will now show the UID shift of local containers, if user
996 namespacing is enabled for them.
998 * systemd will now optionally run "environment generator" binaries at
999 configuration load time. They may be used to add environment
1000 variables to the environment block passed to services invoked. One
1001 user environment generator is shipped by default that sets up
1002 environment variables based on files dropped into /etc/environment.d
1003 and ~/.config/environment.d/.
1005 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1008 * hostnamed has been updated to report a new chassis type of
1009 "convertible" to cover "foldable" laptops that can both act as a
1010 tablet and as a laptop, such as various Lenovo Yoga devices.
1012 Contributions from: Adrián López, Alexander Galanin, Alexander
1013 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1014 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1015 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1016 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1017 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1018 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1019 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1020 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1021 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1022 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1023 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1024 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1025 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1026 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1027 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1028 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1029 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1030 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1031 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1032 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1033 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1034 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1035 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1036 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1037 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1038 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1039 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1042 — Berlin, 2017-03-01
1046 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1047 RestrictAddressFamilies= enabled. These sandboxing options should
1048 generally be compatible with the various external udev call-out
1049 binaries we are aware of, however there may be exceptions, in
1050 particular when exotic languages for these call-outs are used. In
1051 this case, consider turning off these settings locally.
1053 * The new RemoveIPC= option can be used to remove IPC objects owned by
1054 the user or group of a service when that service exits.
1056 * The new ProtectKernelModules= option can be used to disable explicit
1057 load and unload operations of kernel modules by a service. In
1058 addition access to /usr/lib/modules is removed if this option is set.
1060 * ProtectSystem= option gained a new value "strict", which causes the
1061 whole file system tree with the exception of /dev, /proc, and /sys,
1062 to be remounted read-only for a service.
1064 * The new ProtectKernelTunables= option can be used to disable
1065 modification of configuration files in /sys and /proc by a service.
1066 Various directories and files are remounted read-only, so access is
1067 restricted even if the file permissions would allow it.
1069 * The new ProtectControlGroups= option can be used to disable write
1070 access by a service to /sys/fs/cgroup.
1072 * Various systemd services have been hardened with
1073 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1074 RestrictAddressFamilies=.
1076 * Support for dynamically creating users for the lifetime of a service
1077 has been added. If DynamicUser=yes is specified, user and group IDs
1078 will be allocated from the range 61184..65519 for the lifetime of the
1079 service. They can be resolved using the new nss-systemd.so NSS
1080 module. The module must be enabled in /etc/nsswitch.conf. Services
1081 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1082 any resources allocated by the service will be cleaned up when the
1083 service exits. They also have ProtectHome=read-only and
1084 ProtectSystem=strict enabled, so they are not able to make any
1085 permanent modifications to the system.
1087 * The nss-systemd module also always resolves root and nobody, making
1088 it possible to have no /etc/passwd or /etc/group files in minimal
1089 container or chroot environments.
1091 * Services may be started with their own user namespace using the new
1092 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1093 under which the service is running are mapped. All other users are
1096 * Support for the cgroup namespace has been added to systemd-nspawn. If
1097 supported by kernel, the container system started by systemd-nspawn
1098 will have its own view of the cgroup hierarchy. This new behaviour
1099 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1101 * The new MemorySwapMax= option can be used to limit the maximum swap
1102 usage under the unified cgroup hierarchy.
1104 * Support for the CPU controller in the unified cgroup hierarchy has
1105 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1106 options. This controller requires out-of-tree patches for the kernel
1107 and the support is provisional.
1109 * Mount and automount units may now be created transiently
1110 (i.e. dynamically at runtime via the bus API, instead of requiring
1111 unit files in the file system).
1113 * systemd-mount is a new tool which may mount file systems – much like
1114 mount(8), optionally pulling in additional dependencies through
1115 transient .mount and .automount units. For example, this tool
1116 automatically runs fsck on a backing block device before mounting,
1117 and allows the automount logic to be used dynamically from the
1118 command line for establishing mount points. This tool is particularly
1119 useful when dealing with removable media, as it will ensure fsck is
1120 run – if necessary – before the first access and that the file system
1121 is quickly unmounted after each access by utilizing the automount
1122 logic. This maximizes the chance that the file system on the
1123 removable media stays in a clean state, and if it isn't in a clean
1124 state is fixed automatically.
1126 * LazyUnmount=yes option for mount units has been added to expose the
1127 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1130 * /efi will be used as the mount point of the EFI boot partition, if
1131 the directory is present, and the mount point was not configured
1132 through other means (e.g. fstab). If /efi directory does not exist,
1133 /boot will be used as before. This makes it easier to automatically
1134 mount the EFI partition on systems where /boot is used for something
1137 * When operating on GPT disk images for containers, systemd-nspawn will
1138 now mount the ESP to /boot or /efi according to the same rules as PID
1139 1 running on a host. This allows tools like "bootctl" to operate
1140 correctly within such containers, in order to make container images
1141 bootable on physical systems.
1143 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1145 * Two new user session targets have been added to support running
1146 graphical sessions under the systemd --user instance:
1147 graphical-session.target and graphical-session-pre.target. See
1148 systemd.special(7) for a description of how those targets should be
1151 * The vconsole initialization code has been significantly reworked to
1152 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1153 support unicode keymaps. Font and keymap configuration will now be
1154 copied to all allocated virtual consoles.
1156 * FreeBSD's bhyve virtualization is now detected.
1158 * Information recorded in the journal for core dumps now includes the
1159 contents of /proc/mountinfo and the command line of the process at
1160 the top of the process hierarchy (which is usually the init process
1163 * systemd-journal-gatewayd learned the --directory= option to serve
1164 files from the specified location.
1166 * journalctl --root=… can be used to peruse the journal in the
1167 /var/log/ directories inside of a container tree. This is similar to
1168 the existing --machine= option, but does not require the container to
1171 * The hardware database has been extended to support
1172 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1175 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1176 specify the click rate for mice which include a horizontal wheel with
1177 a click rate that is different than the one for the vertical wheel.
1179 * systemd-run gained a new --wait option that makes service execution
1180 synchronous. (Specifically, the command will not return until the
1181 specified service binary exited.)
1183 * systemctl gained a new --wait option that causes the start command to
1184 wait until the units being started have terminated again.
1186 * A new journal output mode "short-full" has been added which displays
1187 timestamps with abbreviated English day names and adds a timezone
1188 suffix. Those timestamps include more information than the default
1189 "short" output mode, and can be passed directly to journalctl's
1190 --since= and --until= options.
1192 * /etc/resolv.conf will be bind-mounted into containers started by
1193 systemd-nspawn, if possible, so any changes to resolv.conf contents
1194 are automatically propagated to the container.
1196 * The number of instances for socket-activated services originating
1197 from a single IP address can be limited with
1198 MaxConnectionsPerSource=, extending the existing setting of
1201 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1204 * .netdev and .network configuration can now be extended through
1207 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1208 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1209 can be enabled and disabled using the new UDPSegmentationOffload=,
1210 TCPSegmentationOffload=, GenericSegmentationOffload=,
1211 GenericReceiveOffload=, LargeReceiveOffload= options in the
1212 [Link] section of .link files.
1214 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1215 Port VLAN ID can be configured for bridge devices using the new STP=,
1216 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1217 section of .netdev files.
1219 * The route table to which routes received over DHCP or RA should be
1220 added can be configured with the new RouteTable= option in the [DHCP]
1221 and [IPv6AcceptRA] sections of .network files.
1223 * The Address Resolution Protocol can be disabled on links managed by
1224 systemd-networkd using the ARP=no setting in the [Link] section of
1227 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1228 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1229 encode information about the result and exit codes of the current
1230 service runtime cycle.
1232 * systemd-sysctl will now configure kernel parameters in the order
1233 they occur in the configuration files. This matches what sysctl
1234 has been traditionally doing.
1236 * kernel-install "plugins" that are executed to perform various
1237 tasks after a new kernel is added and before an old one is removed
1238 can now return a special value to terminate the procedure and
1239 prevent any later plugins from running.
1241 * Journald's SplitMode=login setting has been deprecated. It has been
1242 removed from documentation, and its use is discouraged. In a future
1243 release it will be completely removed, and made equivalent to current
1244 default of SplitMode=uid.
1246 * Storage=both option setting in /etc/systemd/coredump.conf has been
1247 removed. With fast LZ4 compression storing the core dump twice is not
1250 * The --share-system systemd-nspawn option has been replaced with an
1251 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1252 this functionality is discouraged. In addition the variables
1253 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1254 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1255 individual namespaces.
1257 * "machinectl list" now shows the IP address of running containers in
1258 the output, as well as OS release information.
1260 * "loginctl list" now shows the TTY of each session in the output.
1262 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1263 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1264 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1265 tracking objects in a "recursive" mode, where a single client can be
1266 counted multiple times, if it takes multiple references.
1268 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1269 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1270 process using sd-bus automatically exit if the bus connection is
1273 * Bus clients of the service manager may now "pin" loaded units into
1274 memory, by taking an explicit reference on them. This is useful to
1275 ensure the client can retrieve runtime data about the service even
1276 after the service completed execution. Taking such a reference is
1277 available only for privileged clients and should be helpful to watch
1278 running services in a race-free manner, and in particular collect
1279 information about exit statuses and results.
1281 * The nss-resolve module has been changed to strictly return UNAVAIL
1282 when communication via D-Bus with resolved failed, and NOTFOUND when
1283 a lookup completed but was negative. This means it is now possible to
1284 neatly configure fallbacks using nsswitch.conf result checking
1285 expressions. Taking benefit of this, the new recommended
1286 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1288 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1290 * A new setting CtrlAltDelBurstAction= has been added to
1291 /etc/systemd/system.conf which may be used to configure the precise
1292 behaviour if the user on the console presses Ctrl-Alt-Del more often
1293 than 7 times in 2s. Previously this would unconditionally result in
1294 an expedited, immediate reboot. With this new setting the precise
1295 operation may be configured in more detail, and also turned off
1298 * In .netdev files two new settings RemoteChecksumTx= and
1299 RemoteChecksumRx= are now understood that permit configuring the
1300 remote checksumming logic for VXLAN networks.
1302 * The service manager learnt a new "invocation ID" concept for invoked
1303 services. Each runtime cycle of a service will get a new invocation
1304 ID (a 128bit random UUID) assigned that identifies the current
1305 run of the service uniquely and globally. A new invocation ID
1306 is generated each time a service starts up. The journal will store
1307 the invocation ID of a service along with any logged messages, thus
1308 making the invocation ID useful for matching the online runtime of a
1309 service with the offline log data it generated in a safe way without
1310 relying on synchronized timestamps. In many ways this new service
1311 invocation ID concept is similar to the kernel's boot ID concept that
1312 uniquely and globally identifies the runtime of each boot. The
1313 invocation ID of a service is passed to the service itself via an
1314 environment variable ($INVOCATION_ID). A new bus call
1315 GetUnitByInvocationID() has been added that is similar to GetUnit()
1316 but instead of retrieving the bus path for a unit by its name
1317 retrieves it by its invocation ID. The returned path is valid only as
1318 long as the passed invocation ID is current.
1320 * systemd-resolved gained a new "DNSStubListener" setting in
1321 resolved.conf. It either takes a boolean value or the special values
1322 "udp" and "tcp", and configures whether to enable the stub DNS
1323 listener on 127.0.0.53:53.
1325 * IP addresses configured via networkd may now carry additional
1326 configuration settings supported by the kernel. New options include:
1327 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1328 PrefixRoute=, AutoJoin=.
1330 * The PAM configuration fragment file for "user@.service" shipped with
1331 systemd (i.e. the --user instance of systemd) has been stripped to
1332 the minimum necessary to make the system boot. Previously, it
1333 contained Fedora-specific stanzas that did not apply to other
1334 distributions. It is expected that downstream distributions add
1335 additional configuration lines, matching their needs to this file,
1336 using it only as rough template of what systemd itself needs. Note
1337 that this reduced fragment does not even include an invocation of
1338 pam_limits which most distributions probably want to add, even though
1339 systemd itself does not need it. (There's also the new build time
1340 option --with-pamconfdir=no to disable installation of the PAM
1343 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1344 capability is now also dropped from its set (in addition to
1345 CAP_SYS_MKNOD as before).
1347 * In service unit files it is now possible to connect a specific named
1348 file descriptor with stdin/stdout/stdout of an executed service. The
1349 name may be specified in matching .socket units using the
1350 FileDescriptorName= setting.
1352 * A number of journal settings may now be configured on the kernel
1353 command line. Specifically, the following options are now understood:
1354 systemd.journald.max_level_console=,
1355 systemd.journald.max_level_store=,
1356 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1357 systemd.journald.max_level_wall=.
1359 * "systemctl is-enabled --full" will now show by which symlinks a unit
1360 file is enabled in the unit dependency tree.
1362 * Support for VeraCrypt encrypted partitions has been added to the
1363 "cryptsetup" logic and /etc/crypttab.
1365 * systemd-detect-virt gained support for a new --private-users switch
1366 that checks whether the invoking processes are running inside a user
1367 namespace. Similar, a new special value "private-users" for the
1368 existing ConditionVirtualization= setting has been added, permitting
1369 skipping of specific units in user namespace environments.
1371 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1372 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1373 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1374 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1375 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1376 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1377 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1378 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1379 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1380 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1381 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1382 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1383 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1384 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1385 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1386 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1387 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1388 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1389 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1390 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1391 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1392 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1393 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1394 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1395 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1396 Jędrzejewski-Szmek, Zeal Jagannatha
1398 — Santa Fe, 2016-11-03
1402 * In service units the various ExecXYZ= settings have been extended
1403 with an additional special character as first argument of the
1404 assigned value: if the character '+' is used the specified command
1405 line it will be run with full privileges, regardless of User=,
1406 Group=, CapabilityBoundingSet= and similar options. The effect is
1407 similar to the existing PermissionsStartOnly= option, but allows
1408 configuration of this concept for each executed command line
1411 * Services may now alter the service watchdog timeout at runtime by
1412 sending a WATCHDOG_USEC= message via sd_notify().
1414 * MemoryLimit= and related unit settings now optionally take percentage
1415 specifications. The percentage is taken relative to the amount of
1416 physical memory in the system (or in case of containers, the assigned
1417 amount of memory). This allows scaling service resources neatly with
1418 the amount of RAM available on the system. Similarly, systemd-logind's
1419 RuntimeDirectorySize= option now also optionally takes percentage
1422 * In similar fashion TasksMax= takes percentage values now, too. The
1423 value is taken relative to the configured maximum number of processes
1424 on the system. The per-service task maximum has been changed to 15%
1425 using this functionality. (Effectively this is an increase of 512 →
1426 4915 for service units, given the kernel's default pid_max setting.)
1428 * Calendar time specifications in .timer units now understand a ".."
1429 syntax for time ranges. Example: "4..7:10" may now be used for
1430 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1433 * The InaccessableDirectories=, ReadOnlyDirectories= and
1434 ReadWriteDirectories= unit file settings have been renamed to
1435 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1436 applied to all kinds of file nodes, and not just directories, with
1437 the exception of symlinks. Specifically these settings may now be
1438 used on block and character device nodes, UNIX sockets and FIFOS as
1439 well as regular files. The old names of these settings remain
1440 available for compatibility.
1442 * systemd will now log about all service processes it kills forcibly
1443 (using SIGKILL) because they remained after the clean shutdown phase
1444 of the service completed. This should help identifying services that
1445 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1446 systemd-logind's configuration a similar log message is generated for
1447 processes killed at the end of each session due to this setting.
1449 * systemd will now set the $JOURNAL_STREAM environment variable for all
1450 services whose stdout/stderr are connected to the Journal (which
1451 effectively means by default: all services). The variable contains
1452 the device and inode number of the file descriptor used for
1453 stdout/stderr. This may be used by invoked programs to detect whether
1454 their stdout/stderr is connected to the Journal, in which case they
1455 can switch over to direct Journal communication, thus being able to
1456 pass extended, structured metadata along with their log messages. As
1457 one example, this is now used by glib's logging primitives.
1459 * When using systemd's default tmp.mount unit for /tmp, the mount point
1460 will now be established with the "nosuid" and "nodev" options. This
1461 avoids privilege escalation attacks that put traps and exploits into
1462 /tmp. However, this might cause problems if you e. g. put container
1463 images or overlays into /tmp; if you need this, override tmp.mount's
1464 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1467 * systemd now supports the "memory" cgroup controller also on
1470 * The systemd-cgtop tool now optionally takes a control group path as
1471 command line argument. If specified, the control group list shown is
1472 limited to subgroups of that group.
1474 * The SystemCallFilter= unit file setting gained support for
1475 pre-defined, named system call filter sets. For example
1476 SystemCallFilter=@clock is now an effective way to make all clock
1477 changing-related system calls unavailable to a service. A number of
1478 similar pre-defined groups are defined. Writing system call filters
1479 for system services is simplified substantially with this new
1480 concept. Accordingly, all of systemd's own, long-running services now
1481 enable system call filtering based on this, by default.
1483 * A new service setting MemoryDenyWriteExecute= has been added, taking
1484 a boolean value. If turned on, a service may no longer create memory
1485 mappings that are writable and executable at the same time. This
1486 enhances security for services where this is enabled as it becomes
1487 harder to dynamically write and then execute memory in exploited
1488 service processes. This option has been enabled for all of systemd's
1489 own long-running services.
1491 * A new RestrictRealtime= service setting has been added, taking a
1492 boolean argument. If set the service's processes may no longer
1493 acquire realtime scheduling. This improves security as realtime
1494 scheduling may otherwise be used to easily freeze the system.
1496 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1497 value. This may be used for requesting that the system manager inside
1498 of the container reports start-up completion to nspawn which then
1499 propagates this notification further to the service manager
1500 supervising nspawn itself. A related option NotifyReady= in .nspawn
1501 files has been added too. This functionality allows ordering of the
1502 start-up of multiple containers using the usual systemd ordering
1505 * machinectl gained a new command "stop" that is an alias for
1508 * systemd-resolved gained support for contacting DNS servers on
1509 link-local IPv6 addresses.
1511 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1512 its caches. A method call for requesting the same operation has been
1513 added to the bus API too, and is made available via "systemd-resolve
1516 * systemd-resolve gained a new --status switch. If passed a brief
1517 summary of the used DNS configuration with per-interface information
1520 * resolved.conf gained a new Cache= boolean option, defaulting to
1521 on. If turned off local DNS caching is disabled. This comes with a
1522 performance penalty in particular when DNSSEC is enabled. Note that
1523 resolved disables its internal caching implicitly anyway, when the
1524 configured DNS server is on a host-local IP address such as ::1 or
1525 127.0.0.1, thus automatically avoiding double local caching.
1527 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1528 for DNS requests. This improves compatibility with local programs
1529 that do not use the libc NSS or systemd-resolved's bus APIs for name
1530 resolution. This minimal DNS service is only available to local
1531 programs and does not implement the full DNS protocol, but enough to
1532 cover local DNS clients. A new, static resolv.conf file, listing just
1533 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1534 now recommended to make /etc/resolv.conf a symlink to this file in
1535 order to route all DNS lookups to systemd-resolved, regardless if
1536 done via NSS, the bus API or raw DNS packets. Note that this local
1537 DNS service is not as fully featured as the libc NSS or
1538 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1539 used to deliver link-local address information (as this implies
1540 sending a local interface index along), LLMNR/mDNS support via this
1541 interface is severely restricted. It is thus strongly recommended for
1542 all applications to use the libc NSS API or native systemd-resolved
1545 * systemd-networkd's bridge support learned a new setting
1546 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1547 in .network files has been added for configuring VLAN bridging in
1548 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1550 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1551 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1552 now be acquired without relying on DHCPv6. Two new options
1553 UseDomains= and UseDNS= have been added to configure this behaviour.
1555 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1556 renamed IPv6AcceptRA=, without altering its behaviour. The old
1557 setting name remains available for compatibility reasons.
1559 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1560 Key=, InputKey= and OutputKey=.
1562 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1563 interface configuration.
1565 * "systemctl edit" may now be used to create new unit files by
1566 specifying the --force switch.
1568 * sd-event gained a new function sd_event_get_iteration() for
1569 requesting the current iteration counter of the event loop. It starts
1570 at zero and is increased by one with each event loop iteration.
1572 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1573 file. It can be used in lieu of %systemd_requires in packages which
1574 don't use any systemd functionality and are intended to be installed
1575 in minimal containers without systemd present. This macro provides
1576 ordering dependencies to ensure that if the package is installed in
1577 the same rpm transaction as systemd, systemd will be installed before
1578 the scriptlets for the package are executed, allowing unit presets
1581 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1582 been added to simplify packaging of generators.
1584 * The os-release file gained VERSION_CODENAME field for the
1585 distribution nickname (e.g. VERSION_CODENAME=woody).
1587 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1588 can be set to disable parsing of metadata and the creation
1589 of persistent symlinks for that device.
1591 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1592 to make them available to logged-in users has been reverted.
1594 * Much of the common code of the various systemd components is now
1595 built into an internal shared library libsystemd-shared-231.so
1596 (incorporating the systemd version number in the name, to be updated
1597 with future releases) that the components link to. This should
1598 decrease systemd footprint both in memory during runtime and on
1599 disk. Note that the shared library is not for public use, and is
1600 neither API not ABI stable, but is likely to change with every new
1601 released update. Packagers need to make sure that binaries
1602 linking to libsystemd-shared.so are updated in step with the
1605 * Configuration for "mkosi" is now part of the systemd
1606 repository. mkosi is a tool to easily build legacy-free OS images,
1607 and is available on github: https://github.com/systemd/mkosi. If
1608 "mkosi" is invoked in the build tree a new raw OS image is generated
1609 incorporating the systemd sources currently being worked on and a
1610 clean, fresh distribution installation. The generated OS image may be
1611 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1612 UEFI PC. This functionality is particularly useful to easily test
1613 local changes made to systemd in a pristine, defined environment. See
1614 HACKING for details.
1616 * configure learned the --with-support-url= option to specify the
1617 distribution's bugtracker.
1619 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1620 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1621 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1622 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1623 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1624 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1625 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1626 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1627 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1628 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1629 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1630 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1631 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1632 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1633 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1634 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1635 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1636 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1637 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1639 — Berlin, 2016-07-25
1643 * DNSSEC is now turned on by default in systemd-resolved (in
1644 "allow-downgrade" mode), but may be turned off during compile time by
1645 passing "--with-default-dnssec=no" to "configure" (and of course,
1646 during runtime with DNSSEC= in resolved.conf). We recommend
1647 downstreams to leave this on at least during development cycles and
1648 report any issues with the DNSSEC logic upstream. We are very
1649 interested in collecting feedback about the DNSSEC validator and its
1650 limitations in the wild. Note however, that DNSSEC support is
1651 probably nothing downstreams should turn on in stable distros just
1652 yet, as it might create incompatibilities with a few DNS servers and
1653 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1654 automatically whenever we detect such incompatible setups, but there
1655 might be systems we do not cover yet. Hence: please help us testing
1656 the DNSSEC code, leave this on where you can, report back, but then
1657 again don't consider turning this on in your stable, LTS or
1658 production release just yet. (Note that you have to enable
1659 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1660 and its DNSSEC mode for host name resolution from local
1663 * systemd-resolve conveniently resolves DANE records with the --tlsa
1664 option and OPENPGPKEY records with the --openpgp option. It also
1665 supports dumping raw DNS record data via the new --raw= switch.
1667 * systemd-logind will now by default terminate user processes that are
1668 part of the user session scope unit (session-XX.scope) when the user
1669 logs out. This behavior is controlled by the KillUserProcesses=
1670 setting in logind.conf, and the previous default of "no" is now
1671 changed to "yes". This means that user sessions will be properly
1672 cleaned up after, but additional steps are necessary to allow
1673 intentionally long-running processes to survive logout.
1675 While the user is logged in at least once, user@.service is running,
1676 and any service that should survive the end of any individual login
1677 session can be started at a user service or scope using systemd-run.
1678 systemd-run(1) man page has been extended with an example which shows
1679 how to run screen in a scope unit underneath user@.service. The same
1680 command works for tmux.
1682 After the user logs out of all sessions, user@.service will be
1683 terminated too, by default, unless the user has "lingering" enabled.
1684 To effectively allow users to run long-term tasks even if they are
1685 logged out, lingering must be enabled for them. See loginctl(1) for
1686 details. The default polkit policy was modified to allow users to
1687 set lingering for themselves without authentication.
1689 Previous defaults can be restored at compile time by the
1690 --without-kill-user-processes option to "configure".
1692 * systemd-logind gained new configuration settings SessionsMax= and
1693 InhibitorsMax=, both with a default of 8192. It will not register new
1694 user sessions or inhibitors above this limit.
1696 * systemd-logind will now reload configuration on SIGHUP.
1698 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1699 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1700 enable. Also, support for the "io" cgroup controller in the unified
1701 hierarchy has been added, so that the "memory", "pids" and "io" are
1702 now the controllers that are supported on the unified hierarchy.
1704 WARNING: it is not possible to use previous systemd versions with
1705 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1706 is necessary to also update systemd in the initramfs if using the
1707 unified hierarchy. An updated SELinux policy is also required.
1709 * LLDP support has been extended, and both passive (receive-only) and
1710 active (sender) modes are supported. Passive mode ("routers-only") is
1711 enabled by default in systemd-networkd. Active LLDP mode is enabled
1712 by default for containers on the internal network. The "networkctl
1713 lldp" command may be used to list information gathered. "networkctl
1714 status" will also show basic LLDP information on connected peers now.
1716 * The IAID and DUID unique identifier sent in DHCP requests may now be
1717 configured for the system and each .network file managed by
1718 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1720 * systemd-networkd gained support for configuring proxy ARP support for
1721 each interface, via the ProxyArp= setting in .network files. It also
1722 gained support for configuring the multicast querier feature of
1723 bridge devices, via the new MulticastQuerier= setting in .netdev
1724 files. Similarly, snooping on the IGMP traffic can be controlled
1725 via the new setting MulticastSnooping=.
1727 A new setting PreferredLifetime= has been added for addresses
1728 configured in .network file to configure the lifetime intended for an
1731 The systemd-networkd DHCP server gained the option EmitRouter=, which
1732 defaults to yes, to configure whether the DHCP Option 3 (Router)
1735 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1736 systemd-socket-activate and installed into /usr/bin. It is now fully
1739 * systemd-journald now uses separate threads to flush changes to disk
1740 when closing journal files, thus reducing impact of slow disk I/O on
1741 logging performance.
1743 * The sd-journal API gained two new calls
1744 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1745 can be used to open journal files using file descriptors instead of
1746 file or directory paths. sd_journal_open_container() has been
1747 deprecated, sd_journal_open_directory_fd() should be used instead
1748 with the flag SD_JOURNAL_OS_ROOT.
1750 * journalctl learned a new output mode "-o short-unix" that outputs log
1751 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1752 UTC). It also gained support for a new --no-hostname setting to
1753 suppress the hostname column in the family of "short" output modes.
1755 * systemd-ask-password now optionally skips printing of the password to
1756 stdout with --no-output which can be useful in scripts.
1758 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1759 (devices tagged with ID_MAKER_TOOL) are now tagged with
1760 "uaccess" and are available to logged in users.
1762 * The DeviceAllow= unit setting now supports specifiers (with "%").
1764 * "systemctl show" gained a new --value switch, which allows print a
1765 only the contents of a specific unit property, without also printing
1766 the property's name. Similar support was added to "show*" verbs
1767 of loginctl and machinectl that output "key=value" lists.
1769 * A new unit type "generated" was added for files dynamically generated
1770 by generator tools. Similarly, a new unit type "transient" is used
1771 for unit files created using the runtime API. "systemctl enable" will
1772 refuse to operate on such files.
1774 * A new command "systemctl revert" has been added that may be used to
1775 revert to the vendor version of a unit file, in case local changes
1776 have been made by adding drop-ins or overriding the unit file.
1778 * "machinectl clean" gained a new verb to automatically remove all or
1779 just hidden container images.
1781 * systemd-tmpfiles gained support for a new line type "e" for emptying
1782 directories, if they exist, without creating them if they don't.
1784 * systemd-nspawn gained support for automatically patching the UID/GIDs
1785 of the owners and the ACLs of all files and directories in a
1786 container tree to match the UID/GID user namespacing range selected
1787 for the container invocation. This mode is enabled via the new
1788 --private-users-chown switch. It also gained support for
1789 automatically choosing a free, previously unused UID/GID range when
1790 starting a container, via the new --private-users=pick setting (which
1791 implies --private-users-chown). Together, these options for the first
1792 time make user namespacing for nspawn containers fully automatic and
1793 thus deployable. The systemd-nspawn@.service template unit file has
1794 been changed to use this functionality by default.
1796 * systemd-nspawn gained a new --network-zone= switch, that allows
1797 creating ad-hoc virtual Ethernet links between multiple containers,
1798 that only exist as long as at least one container referencing them is
1799 running. This allows easy connecting of multiple containers with a
1800 common link that implements an Ethernet broadcast domain. Each of
1801 these network "zones" may be named relatively freely by the user, and
1802 may be referenced by any number of containers, but each container may
1803 only reference one of these "zones". On the lower level, this is
1804 implemented by an automatically managed bridge network interface for
1805 each zone, that is created when the first container referencing its
1806 zone is created and removed when the last one referencing its zone
1809 * The default start timeout may now be configured on the kernel command
1810 line via systemd.default_timeout_start_sec=. It was already
1811 configurable via the DefaultTimeoutStartSec= option in
1812 /etc/systemd/system.conf.
1814 * Socket units gained a new TriggerLimitIntervalSec= and
1815 TriggerLimitBurst= setting to configure a limit on the activation
1816 rate of the socket unit.
1818 * The LimitNICE= setting now optionally takes normal UNIX nice values
1819 in addition to the raw integer limit value. If the specified
1820 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1821 value is understood as UNIX nice value. If not prefixed like this it
1822 is understood as raw RLIMIT_NICE limit.
1824 * Note that the effect of the PrivateDevices= unit file setting changed
1825 slightly with this release: the per-device /dev file system will be
1826 mounted read-only from this version on, and will have "noexec"
1827 set. This (minor) change of behavior might cause some (exceptional)
1828 legacy software to break, when PrivateDevices=yes is set for its
1829 service. Please leave PrivateDevices= off if you run into problems
1832 * systemd-bootchart has been split out to a separate repository:
1833 https://github.com/systemd/systemd-bootchart
1835 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1836 merged into the kernel in its current form.
1838 * The compatibility libraries libsystemd-daemon.so,
1839 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1840 which have been deprecated since systemd-209 have been removed along
1841 with the corresponding pkg-config files. All symbols provided by
1842 those libraries are provided by libsystemd.so.
1844 * The Capabilities= unit file setting has been removed (it is ignored
1845 for backwards compatibility). AmbientCapabilities= and
1846 CapabilityBoundingSet= should be used instead.
1848 * A new special target has been added, initrd-root-device.target,
1849 which creates a synchronization point for dependencies of the root
1850 device in early userspace. Initramfs builders must ensure that this
1851 target is now included in early userspace.
1853 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1854 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1855 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1856 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1857 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1858 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1859 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1860 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1861 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1862 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1863 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1864 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1865 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1866 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1867 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1868 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1869 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1870 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1871 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1872 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1873 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1874 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1875 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1876 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1877 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1880 — Fairfax, 2016-05-21
1884 * The systemd-resolved DNS resolver service has gained a substantial
1885 set of new features, most prominently it may now act as a DNSSEC
1886 validating stub resolver. DNSSEC mode is currently turned off by
1887 default, but is expected to be turned on by default in one of the
1888 next releases. For now, we invite everybody to test the DNSSEC logic
1889 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1890 service also gained a full set of D-Bus interfaces, including calls
1891 to configure DNS and DNSSEC settings per link (for use by external
1892 network management software). systemd-resolved and systemd-networkd
1893 now distinguish between "search" and "routing" domains. The former
1894 are used to qualify single-label names, the latter are used purely
1895 for routing lookups within certain domains to specific links.
1896 resolved now also synthesizes RRs for all entries from /etc/hosts.
1898 * The systemd-resolve tool (which is a client utility for
1899 systemd-resolved) has been improved considerably and is now fully
1900 supported and documented. Hence it has moved from /usr/lib/systemd to
1903 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1906 * The coredump collection logic has been reworked: when a coredump is
1907 collected it is now written to disk, compressed and processed
1908 (including stacktrace extraction) from a new instantiated service
1909 systemd-coredump@.service, instead of directly from the
1910 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1911 processing large coredumps can take up a substantial amount of
1912 resources and time, and this previously happened entirely outside of
1913 systemd's service supervision. With the new logic the core_pattern
1914 hook only does minimal metadata collection before passing off control
1915 to the new instantiated service, which is configured with a time
1916 limit, a nice level and other settings to minimize negative impact on
1917 the rest of the system. Also note that the new logic will honour the
1918 RLIMIT_CORE setting of the crashed process, which now allows users
1919 and processes to turn off coredumping for their processes by setting
1922 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1923 and all forked processes by default. Previously, PID 1 would leave
1924 the setting at "0" for all processes, as set by the kernel. Note that
1925 the resource limit traditionally has no effect on the generated
1926 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1927 logic is used. Since the limit is now honoured (see above) its
1928 default has been changed so that the coredumping logic is enabled by
1929 default for all processes, while allowing specific opt-out.
1931 * When the stacktrace is extracted from processes of system users, this
1932 is now done as "systemd-coredump" user, in order to sandbox this
1933 potentially security sensitive parsing operation. (Note that when
1934 processing coredumps of normal users this is done under the user ID
1935 of process that crashed, as before.) Packagers should take notice
1936 that it is now necessary to create the "systemd-coredump" system user
1937 and group at package installation time.
1939 * The systemd-activate socket activation testing tool gained support
1940 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1941 and --seqpacket switches. It also has been extended to support both
1942 new-style and inetd-style file descriptor passing. Use the new
1943 --inetd switch to request inetd-style file descriptor passing.
1945 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1946 variable, which takes a boolean value. If set to false, ANSI color
1947 output is disabled in the tools even when run on a terminal that
1950 * The VXLAN support in networkd now supports two new settings
1951 DestinationPort= and PortRange=.
1953 * A new systemd.machine_id= kernel command line switch has been added,
1954 that may be used to set the machine ID in /etc/machine-id if it is
1955 not initialized yet. This command line option has no effect if the
1956 file is already initialized.
1958 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1959 specified command line as PID 2 rather than PID 1 in the
1960 container. In this mode PID 1 is a minimal stub init process that
1961 implements the special POSIX and Linux semantics of PID 1 regarding
1962 signal and child process management. Note that this stub init process
1963 is implemented in nspawn itself and requires no support from the
1964 container image. This new logic is useful to support running
1965 arbitrary commands in the container, as normal processes are
1966 generally not prepared to run as PID 1.
1968 * systemd-nspawn gained a new --chdir= switch for setting the current
1969 working directory for the process started in the container.
1971 * "journalctl /dev/sda" will now output all kernel log messages for
1972 specified device from the current boot, in addition to all devices
1973 that are parents of it. This should make log output about devices
1974 pretty useful, as long as kernel drivers attach enough metadata to
1975 the log messages. (The usual SATA drivers do.)
1977 * The sd-journal API gained two new calls
1978 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1979 that report whether log data from /run or /var has been found.
1981 * journalctl gained a new switch "--fields" that prints all journal
1982 record field names currently in use in the journal. This is backed
1983 by two new sd-journal API calls sd_journal_enumerate_fields() and
1984 sd_journal_restart_fields().
1986 * Most configurable timeouts in systemd now expect an argument of
1987 "infinity" to turn them off, instead of "0" as before. The semantics
1988 from now on is that a timeout of "0" means "now", and "infinity"
1989 means "never". To maintain backwards compatibility, "0" continues to
1990 turn off previously existing timeout settings.
1992 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1993 try-reload-or-restart" to clarify what it actually does: the "try"
1994 logic applies to both reloading and restarting, not just restarting.
1995 The old name continues to be accepted for compatibility.
1997 * On boot-up, when PID 1 detects that the system clock is behind the
1998 release date of the systemd version in use, the clock is now set
1999 to the latter. Previously, this was already done in timesyncd, in order
2000 to avoid running with clocks set to the various clock epochs such as
2001 1902, 1938 or 1970. With this change the logic is now done in PID 1
2002 in addition to timesyncd during early boot-up, so that it is enforced
2003 before the first process is spawned by systemd. Note that the logic
2004 in timesyncd remains, as it is more comprehensive and ensures
2005 clock monotonicity by maintaining a persistent timestamp file in
2006 /var. Since /var is generally not available in earliest boot or the
2007 initrd, this part of the logic remains in timesyncd, and is not done
2010 * Support for tweaking details in net_cls.class_id through the
2011 NetClass= configuration directive has been removed, as the kernel
2012 people have decided to deprecate that controller in cgroup v2.
2013 Userspace tools such as nftables are moving over to setting rules
2014 that are specific to the full cgroup path of a task, which obsoletes
2015 these controllers anyway. The NetClass= directive is kept around for
2016 legacy compatibility reasons. For a more in-depth description of the
2017 kernel change, please refer to the respective upstream commit:
2019 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2021 * A new service setting RuntimeMaxSec= has been added that may be used
2022 to specify a maximum runtime for a service. If the timeout is hit, the
2023 service is terminated and put into a failure state.
2025 * A new service setting AmbientCapabilities= has been added. It allows
2026 configuration of additional Linux process capabilities that are
2027 passed to the activated processes. This is only available on very
2030 * The process resource limit settings in service units may now be used
2031 to configure hard and soft limits individually.
2033 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2034 expose support for gcc's __attribute__((cleanup())) C extension.
2035 Specifically, for many object destructor functions alternative
2036 versions have been added that have names suffixed with "p" and take a
2037 pointer to a pointer to the object to destroy, instead of just a
2038 pointer to the object itself. This is useful because these destructor
2039 functions may be used directly as parameters to the cleanup
2040 construct. Internally, systemd has been a heavy user of this GCC
2041 extension for a long time, and with this change similar support is
2042 now available to consumers of the library outside of systemd. Note
2043 that by using this extension in your sources compatibility with old
2044 and strictly ANSI compatible C compilers is lost. However, all gcc or
2045 LLVM versions of recent years support this extension.
2047 * Timer units gained support for a new setting RandomizedDelaySec= that
2048 allows configuring some additional randomized delay to the configured
2049 time. This is useful to spread out timer events to avoid load peaks in
2050 clusters or larger setups.
2052 * Calendar time specifications now support sub-second accuracy.
2054 * Socket units now support listening on SCTP and UDP-lite protocol
2057 * The sd-event API now comes with a full set of man pages.
2059 * Older versions of systemd contained experimental support for
2060 compressing journal files and coredumps with the LZ4 compressor that
2061 was not compatible with the lz4 binary (due to API limitations of the
2062 lz4 library). This support has been removed; only support for files
2063 compatible with the lz4 binary remains. This LZ4 logic is now
2064 officially supported and no longer considered experimental.
2066 * The dkr image import logic has been removed again from importd. dkr's
2067 micro-services focus doesn't fit into the machine image focus of
2068 importd, and quickly got out of date with the upstream dkr API.
2070 * Creation of the /run/lock/lockdev/ directory was dropped from
2071 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2072 been available for many years. If you still need this, you need to
2073 create your own tmpfiles.d config file with:
2075 d /run/lock/lockdev 0775 root lock -
2077 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2078 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2079 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2080 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2081 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2082 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2083 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2084 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2085 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2086 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2087 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2088 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2089 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2090 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2091 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2092 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2093 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2094 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2095 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2097 — Berlin, 2016-02-11
2101 * A number of properties previously only settable in unit
2102 files are now also available as properties to set when
2103 creating transient units programmatically via the bus, as it
2104 is exposed with systemd-run's --property=
2105 setting. Specifically, these are: SyslogIdentifier=,
2106 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2107 EnvironmentFile=, ReadWriteDirectories=,
2108 ReadOnlyDirectories=, InaccessibleDirectories=,
2109 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2111 * When creating transient services via the bus API it is now
2112 possible to pass in a set of file descriptors to use as
2113 STDIN/STDOUT/STDERR for the invoked process.
2115 * Slice units may now be created transiently via the bus APIs,
2116 similar to the way service and scope units may already be
2117 created transiently.
2119 * Wherever systemd expects a calendar timestamp specification
2120 (like in journalctl's --since= and --until= switches) UTC
2121 timestamps are now supported. Timestamps suffixed with "UTC"
2122 are now considered to be in Universal Time Coordinated
2123 instead of the local timezone. Also, timestamps may now
2124 optionally be specified with sub-second accuracy. Both of
2125 these additions also apply to recurring calendar event
2126 specification, such as OnCalendar= in timer units.
2128 * journalctl gained a new "--sync" switch that asks the
2129 journal daemon to write all so far unwritten log messages to
2130 disk and sync the files, before returning.
2132 * systemd-tmpfiles learned two new line types "q" and "Q" that
2133 operate like "v", but also set up a basic btrfs quota
2134 hierarchy when used on a btrfs file system with quota
2137 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2138 instead of a subvolume (even on a btrfs file system) if the
2139 root directory is a plain directory, and not a
2140 subvolume. This should simplify things with certain chroot()
2141 environments which are not aware of the concept of btrfs
2144 * systemd-detect-virt gained a new --chroot switch to detect
2145 whether execution takes place in a chroot() environment.
2147 * CPUAffinity= now takes CPU index ranges in addition to
2150 * The various memory-related resource limit settings (such as
2151 LimitAS=) now understand the usual K, M, G, ... suffixes to
2152 the base of 1024 (IEC). Similar, the time-related resource
2153 limit settings understand the usual min, h, day, ...
2156 * There's a new system.conf setting DefaultTasksMax= to
2157 control the default TasksMax= setting for services and
2158 scopes running on the system. (TasksMax= is the primary
2159 setting that exposes the "pids" cgroup controller on systemd
2160 and was introduced in the previous systemd release.) The
2161 setting now defaults to 512, which means services that are
2162 not explicitly configured otherwise will only be able to
2163 create 512 processes or threads at maximum, from this
2164 version on. Note that this means that thread- or
2165 process-heavy services might need to be reconfigured to set
2166 TasksMax= to a higher value. It is sufficient to set
2167 TasksMax= in these specific unit files to a higher value, or
2168 even "infinity". Similar, there's now a logind.conf setting
2169 UserTasksMax= that defaults to 4096 and limits the total
2170 number of processes or tasks each user may own
2171 concurrently. nspawn containers also have the TasksMax=
2172 value set by default now, to 8192. Note that all of this
2173 only has an effect if the "pids" cgroup controller is
2174 enabled in the kernel. The general benefit of these changes
2175 should be a more robust and safer system, that provides a
2176 certain amount of per-service fork() bomb protection.
2178 * systemd-nspawn gained the new --network-veth-extra= switch
2179 to define additional and arbitrarily-named virtual Ethernet
2180 links between the host and the container.
2182 * A new service execution setting PassEnvironment= has been
2183 added that allows importing select environment variables
2184 from PID1's environment block into the environment block of
2187 * Timer units gained support for a new RemainAfterElapse=
2188 setting which takes a boolean argument. It defaults to on,
2189 exposing behaviour unchanged to previous releases. If set to
2190 off, timer units are unloaded after they elapsed if they
2191 cannot elapse again. This is particularly useful for
2192 transient timer units, which shall not stay around longer
2193 than until they first elapse.
2195 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2196 default now (the kernel default is 16). This is beneficial
2197 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2198 allows substantially larger numbers of queued
2199 datagrams. This should increase the capability of systemd to
2200 parallelize boot-up, as logging and sd_notify() are unlikely
2201 to stall execution anymore. If you need to change the value
2202 from the new defaults, use the usual sysctl.d/ snippets.
2204 * The compression framing format used by the journal or
2205 coredump processing has changed to be in line with what the
2206 official LZ4 tools generate. LZ4 compression support in
2207 systemd was considered unsupported previously, as the format
2208 was not compatible with the normal tools. With this release
2209 this has changed now, and it is hence safe for downstream
2210 distributions to turn it on. While not compressing as well
2211 as the XZ, LZ4 is substantially faster, which makes
2212 it a good default choice for the compression logic in the
2213 journal and in coredump handling.
2215 * Any reference to /etc/mtab has been dropped from
2216 systemd. The file has been obsolete since a while, but
2217 systemd refused to work on systems where it was incorrectly
2218 set up (it should be a symlink or non-existent). Please make
2219 sure to update to util-linux 2.27.1 or newer in conjunction
2220 with this systemd release, which also drops any reference to
2221 /etc/mtab. If you maintain a distribution make sure that no
2222 software you package still references it, as this is a
2223 likely source of bugs. There's also a glibc bug pending,
2224 asking for removal of any reference to this obsolete file:
2226 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2228 Note that only util-linux versions built with
2229 --enable-libmount-force-mountinfo are supported.
2231 * Support for the ".snapshot" unit type has been removed. This
2232 feature turned out to be little useful and little used, and
2233 has now been removed from the core and from systemctl.
2235 * The dependency types RequiresOverridable= and
2236 RequisiteOverridable= have been removed from systemd. They
2237 have been used only very sparingly to our knowledge and
2238 other options that provide a similar effect (such as
2239 systemctl --mode=ignore-dependencies) are much more useful
2240 and commonly used. Moreover, they were only half-way
2241 implemented as the option to control behaviour regarding
2242 these dependencies was never added to systemctl. By removing
2243 these dependency types the execution engine becomes a bit
2244 simpler. Unit files that use these dependencies should be
2245 changed to use the non-Overridable dependency types
2246 instead. In fact, when parsing unit files with these
2247 options, that's what systemd will automatically convert them
2248 too, but it will also warn, asking users to fix the unit
2249 files accordingly. Removal of these dependency types should
2250 only affect a negligible number of unit files in the wild.
2252 * Behaviour of networkd's IPForward= option changed
2253 (again). It will no longer maintain a per-interface setting,
2254 but propagate one way from interfaces where this is enabled
2255 to the global kernel setting. The global setting will be
2256 enabled when requested by a network that is set up, but
2257 never be disabled again. This change was made to make sure
2258 IPv4 and IPv6 behaviour regarding packet forwarding is
2259 similar (as the Linux IPv6 stack does not support
2260 per-interface control of this setting) and to minimize
2263 * In unit files the behaviour of %u, %U, %h, %s has
2264 changed. These specifiers will now unconditionally resolve
2265 to the various user database fields of the user that the
2266 systemd instance is running as, instead of the user
2267 configured in the specific unit via User=. Note that this
2268 effectively doesn't change much, as resolving of these
2269 specifiers was already turned off in the --system instance
2270 of systemd, as we cannot do NSS lookups from PID 1. In the
2271 --user instance of systemd these specifiers where correctly
2272 resolved, but hardly made any sense, since the user instance
2273 lacks privileges to do user switches anyway, and User= is
2274 hence useless. Moreover, even in the --user instance of
2275 systemd behaviour was awkward as it would only take settings
2276 from User= assignment placed before the specifier into
2277 account. In order to unify and simplify the logic around
2278 this the specifiers will now always resolve to the
2279 credentials of the user invoking the manager (which in case
2280 of PID 1 is the root user).
2282 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2283 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2284 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2285 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2286 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2287 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2288 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2289 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2290 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2291 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2292 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2293 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2294 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2295 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2298 — Berlin, 2015-11-18
2302 * systemd now depends on util-linux v2.27. More specifically,
2303 the newly added mount monitor feature in libmount now
2304 replaces systemd's former own implementation.
2306 * libmount mandates /etc/mtab not to be regular file, and
2307 systemd now enforces this condition at early boot.
2308 /etc/mtab has been deprecated and warned about for a very
2309 long time, so systems running systemd should already have
2310 stopped having this file around as anything else than a
2311 symlink to /proc/self/mounts.
2313 * Support for the "pids" cgroup controller has been added. It
2314 allows accounting the number of tasks in a cgroup and
2315 enforcing limits on it. This adds two new setting
2316 TasksAccounting= and TasksMax= to each unit, as well as a
2317 global option DefaultTasksAccounting=.
2319 * Support for the "net_cls" cgroup controller has been added.
2320 It allows assigning a net class ID to each task in the
2321 cgroup, which can then be used in firewall rules and traffic
2322 shaping configurations. Note that the kernel netfilter net
2323 class code does not currently work reliably for ingress
2324 packets on unestablished sockets.
2326 This adds a new config directive called NetClass= to CGroup
2327 enabled units. Allowed values are positive numbers for fixed
2328 assignments and "auto" for picking a free value
2331 * 'systemctl is-system-running' now returns 'offline' if the
2332 system is not booted with systemd. This command can now be
2333 used as a substitute for 'systemd-notify --booted'.
2335 * Watchdog timeouts have been increased to 3 minutes for all
2336 in-tree service files. Apparently, disk IO issues are more
2337 frequent than we hoped, and user reported >1 minute waiting
2340 * 'machine-id-commit' functionality has been merged into
2341 'machine-id-setup --commit'. The separate binary has been
2344 * The WorkingDirectory= directive in unit files may now be set
2345 to the special value '~'. In this case, the working
2346 directory is set to the home directory of the user
2347 configured in User=.
2349 * "machinectl shell" will now open the shell in the home
2350 directory of the selected user by default.
2352 * The CrashChVT= configuration file setting is renamed to
2353 CrashChangeVT=, following our usual logic of not
2354 abbreviating unnecessarily. The old directive is still
2355 supported for compat reasons. Also, this directive now takes
2356 an integer value between 1 and 63, or a boolean value. The
2357 formerly supported '-1' value for disabling stays around for
2360 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2361 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2362 RootDirectory= properties can now be set for transient
2365 * The systemd-analyze tool gained a new "set-log-target" verb
2366 to change the logging target the system manager logs to
2367 dynamically during runtime. This is similar to how
2368 "systemd-analyze set-log-level" already changes the log
2371 * In nspawn /sys is now mounted as tmpfs, with only a selected
2372 set of subdirectories mounted in from the real sysfs. This
2373 enhances security slightly, and is useful for ensuring user
2374 namespaces work correctly.
2376 * Support for USB FunctionFS activation has been added. This
2377 allows implementation of USB gadget services that are
2378 activated as soon as they are requested, so that they don't
2379 have to run continuously, similar to classic socket
2382 * The "systemctl exit" command now optionally takes an
2383 additional parameter that sets the exit code to return from
2384 the systemd manager when exiting. This is only relevant when
2385 running the systemd user instance, or when running the
2386 system instance in a container.
2388 * sd-bus gained the new API calls sd_bus_path_encode_many()
2389 and sd_bus_path_decode_many() that allow easy encoding and
2390 decoding of multiple identifier strings inside a D-Bus
2391 object path. Another new call sd_bus_default_flush_close()
2392 has been added to flush and close per-thread default
2395 * systemd-cgtop gained support for a -M/--machine= switch to
2396 show the control groups within a certain container only.
2398 * "systemctl kill" gained support for an optional --fail
2399 switch. If specified the requested operation will fail of no
2400 processes have been killed, because the unit had no
2401 processes attached, or similar.
2403 * A new systemd.crash_reboot=1 kernel command line option has
2404 been added that triggers a reboot after crashing. This can
2405 also be set through CrashReboot= in systemd.conf.
2407 * The RuntimeDirectory= setting now understands unit
2408 specifiers like %i or %f.
2410 * A new (still internal) library API sd-ipv4acd has been added,
2411 that implements address conflict detection for IPv4. It's
2412 based on code from sd-ipv4ll, and will be useful for
2413 detecting DHCP address conflicts.
2415 * File descriptors passed during socket activation may now be
2416 named. A new API sd_listen_fds_with_names() is added to
2417 access the names. The default names may be overridden,
2418 either in the .socket file using the FileDescriptorName=
2419 parameter, or by passing FDNAME= when storing the file
2420 descriptors using sd_notify().
2422 * systemd-networkd gained support for:
2424 - Setting the IPv6 Router Advertisement settings via
2425 IPv6AcceptRouterAdvertisements= in .network files.
2427 - Configuring the HelloTimeSec=, MaxAgeSec= and
2428 ForwardDelaySec= bridge parameters in .netdev files.
2430 - Configuring PreferredSource= for static routes in
2433 * The "ask-password" framework used to query for LUKS harddisk
2434 passwords or SSL passwords during boot gained support for
2435 caching passwords in the kernel keyring, if it is
2436 available. This makes sure that the user only has to type in
2437 a passphrase once if there are multiple objects to unlock
2438 with the same one. Previously, such password caching was
2439 available only when Plymouth was used; this moves the
2440 caching logic into the systemd codebase itself. The
2441 "systemd-ask-password" utility gained a new --keyname=
2442 switch to control which kernel keyring key to use for
2443 caching a password in. This functionality is also useful for
2444 enabling display managers such as gdm to automatically
2445 unlock the user's GNOME keyring if its passphrase, the
2446 user's password and the harddisk password are the same, if
2447 gdm-autologin is used.
2449 * When downloading tar or raw images using "machinectl
2450 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2451 file is now also downloaded, if it is available and stored
2452 next to the image file.
2454 * Units of type ".socket" gained a new boolean setting
2455 Writable= which is only useful in conjunction with
2456 ListenSpecial=. If true, enables opening the specified
2457 special file in O_RDWR mode rather than O_RDONLY mode.
2459 * systemd-rfkill has been reworked to become a singleton
2460 service that is activated through /dev/rfkill on each rfkill
2461 state change and saves the settings to disk. This way,
2462 systemd-rfkill is now compatible with devices that exist
2463 only intermittendly, and even restores state if the previous
2464 system shutdown was abrupt rather than clean.
2466 * The journal daemon gained support for vacuuming old journal
2467 files controlled by the number of files that shall remain,
2468 in addition to the already existing control by size and by
2469 date. This is useful as journal interleaving performance
2470 degrades with too many separate journal files, and allows
2471 putting an effective limit on them. The new setting defaults
2472 to 100, but this may be changed by setting SystemMaxFiles=
2473 and RuntimeMaxFiles= in journald.conf. Also, the
2474 "journalctl" tool gained the new --vacuum-files= switch to
2475 manually vacuum journal files to leave only the specified
2476 number of files in place.
2478 * udev will now create /dev/disk/by-path links for ATA devices
2479 on kernels where that is supported.
2481 * Galician, Serbian, Turkish and Korean translations were added.
2483 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2484 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2485 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2486 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2487 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2488 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2489 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2490 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2491 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2492 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2493 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2494 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2495 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2496 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2497 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2498 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2499 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2500 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2502 — Berlin, 2015-10-07
2506 * The DHCP implementation of systemd-networkd gained a set of
2509 - The DHCP server now supports emitting DNS and NTP
2510 information. It may be enabled and configured via
2511 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2512 and NTP information is enabled, but no servers are
2513 configured, the corresponding uplink information (if there
2514 is any) is propagated.
2516 - Server and client now support transmission and reception
2517 of timezone information. It can be configured via the
2518 newly introduced network options UseTimezone=,
2519 EmitTimezone=, and Timezone=. Transmission of timezone
2520 information is enabled between host and containers by
2521 default now: the container will change its local timezone
2522 to what the host has set.
2524 - Lease timeouts can now be configured via
2525 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2527 - The DHCP server improved on the stability of
2528 leases. Clients are more likely to get the same lease
2529 information back, even if the server loses state.
2531 - The DHCP server supports two new configuration options to
2532 control the lease address pool metrics, PoolOffset= and
2535 * The encapsulation limit of tunnels in systemd-networkd may
2536 now be configured via 'EncapsulationLimit='. It allows
2537 modifying the maximum additional levels of encapsulation
2538 that are permitted to be prepended to a packet.
2540 * systemd now supports the concept of user buses replacing
2541 session buses, if used with dbus-1.10 (and enabled via dbus
2542 --enable-user-session). It previously only supported this on
2543 kdbus-enabled systems, and this release expands this to
2544 'dbus-daemon' systems.
2546 * systemd-networkd now supports predictable interface names
2549 * systemd now optionally supports the new Linux kernel
2550 "unified" control group hierarchy. If enabled via the kernel
2551 command-line option 'systemd.unified_cgroup_hierarchy=1',
2552 systemd will try to mount the unified cgroup hierarchy
2553 directly on /sys/fs/cgroup. If not enabled, or not
2554 available, systemd will fall back to the legacy cgroup
2555 hierarchy setup, as before. Host system and containers can
2556 mix and match legacy and unified hierarchies as they
2557 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2558 environment variable to individually select the hierarchy to
2559 use for executed containers. By default, nspawn will use the
2560 unified hierarchy for the containers if the host uses the
2561 unified hierarchy, and the legacy hierarchy otherwise.
2562 Please note that at this point the unified hierarchy is an
2563 experimental kernel feature and is likely to change in one
2564 of the next kernel releases. Therefore, it should not be
2565 enabled by default in downstream distributions yet. The
2566 minimum required kernel version for the unified hierarchy to
2567 work is 4.2. Note that when the unified hierarchy is used
2568 for the first time delegated access to controllers is
2569 safe. Because of this systemd-nspawn containers will get
2570 access to controllers now, as will systemd user
2571 sessions. This means containers and user sessions may now
2572 manage their own resources, partitioning up what the system
2575 * A new special scope unit "init.scope" has been introduced
2576 that encapsulates PID 1 of the system. It may be used to
2577 determine resource usage and enforce resource limits on PID
2578 1 itself. PID 1 hence moved out of the root of the control
2581 * The cgtop tool gained support for filtering out kernel
2582 threads when counting tasks in a control group. Also, the
2583 count of processes is now recursively summed up by
2584 default. Two options -k and --recursive= have been added to
2585 revert to old behaviour. The tool has also been updated to
2586 work correctly in containers now.
2588 * systemd-nspawn's --bind= and --bind-ro= options have been
2589 extended to allow creation of non-recursive bind mounts.
2591 * libsystemd gained two new calls sd_pid_get_cgroup() and
2592 sd_peer_get_cgroup() which return the control group path of
2593 a process or peer of a connected AF_UNIX socket. This
2594 function call is particularly useful when implementing
2595 delegated subtrees support in the control group hierarchy.
2597 * The "sd-event" event loop API of libsystemd now supports
2598 correct dequeuing of real-time signals, without losing
2601 * When systemd requests a PolicyKit decision when managing
2602 units it will now add additional fields to the request,
2603 including unit name and desired operation. This enables more
2604 powerful PolicyKit policies, that make decisions depending
2605 on these parameters.
2607 * nspawn learnt support for .nspawn settings files, that may
2608 accompany the image files or directories of containers, and
2609 may contain additional settings for the container. This is
2610 an alternative to configuring container parameters via the
2611 nspawn command line.
2613 Contributions from: Cristian Rodríguez, Daniel Mack, David
2614 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2615 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2616 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2617 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2618 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2619 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2620 Andersen, Tom Gundersen, Torstein Husebø
2622 — Berlin, 2015-09-08
2626 * machinectl gained a new verb 'shell' which opens a fresh
2627 shell on the target container or the host. It is similar to
2628 the existing 'login' command of machinectl, but spawns the
2629 shell directly without prompting for username or
2630 password. The pseudo machine '.host' now refers to the local
2631 host and is used by default. Hence, 'machinectl shell' can
2632 be used as replacement for 'su -' which spawns a session as
2633 a fresh systemd unit in a way that is fully isolated from
2634 the originating session.
2636 * systemd-networkd learned to cope with private-zone DHCP
2637 options and allows other programs to query the values.
2639 * SELinux access control when enabling/disabling units is no
2640 longer enforced with this release. The previous
2641 implementation was incorrect, and a new corrected
2642 implementation is not yet available. As unit file operations
2643 are still protected via PolicyKit and D-Bus policy this is
2644 not a security problem. Yet, distributions which care about
2645 optimal SELinux support should probably not stabilize on
2648 * sd-bus gained support for matches of type "arg0has=", that
2649 test for membership of strings in string arrays sent in bus
2652 * systemd-resolved now dumps the contents of its DNS and LLMNR
2653 caches to the logs on reception of the SIGUSR1 signal. This
2654 is useful to debug DNS behaviour.
2656 * The coredumpctl tool gained a new --directory= option to
2657 operate on journal files in a specific directory.
2659 * "systemctl reboot" and related commands gained a new
2660 "--message=" option which may be used to set a free-text
2661 wall message when shutting down or rebooting the
2662 system. This message is also logged, which is useful for
2663 figuring out the reason for a reboot or shutdown a
2666 * The "systemd-resolve-host" tool's -i switch now takes
2667 network interface numbers as alternative to interface names.
2669 * A new unit file setting for services has been introduced:
2670 UtmpMode= allows configuration of how precisely systemd
2671 handles utmp and wtmp entries for the service if this is
2672 enabled. This allows writing services that appear similar to
2673 user sessions in the output of the "w", "who", "last" and
2676 * systemd-resolved will now locally synthesize DNS resource
2677 records for the "localhost" and "gateway" domains as well as
2678 the local hostname. This should ensure that clients querying
2679 RRs via resolved will get similar results as those going via
2680 NSS, if nss-myhostname is enabled.
2682 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2683 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2684 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2685 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2686 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2687 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2688 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2689 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2690 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2691 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2692 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2693 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2695 — Berlin, 2015-08-27
2699 * The systemd-efi-boot-generator functionality was merged into
2700 systemd-gpt-auto-generator.
2702 * systemd-networkd now supports Group Policy for vxlan
2703 devices. It can be enabled via the new boolean configuration
2704 option called 'GroupPolicyExtension='.
2706 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2707 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2708 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2710 — Berlin, 2015-07-31
2714 * The python-systemd code has been removed from the systemd repository.
2715 A new repository has been created which accommodates the code from
2716 now on, and we kindly ask distributions to create a separate package
2717 for this: https://github.com/systemd/python-systemd
2719 * The systemd daemon will now reload its main configuration
2720 (/etc/systemd/system.conf) on daemon-reload.
2722 * sd-dhcp now exposes vendor specific extensions via
2723 sd_dhcp_lease_get_vendor_specific().
2725 * systemd-networkd gained a number of new configuration options.
2727 - A new boolean configuration option for TAP devices called
2728 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2729 device, thus allowing to send and receive GSO packets.
2731 - A new tunnel configuration option called 'CopyDSCP='.
2732 If enabled, the DSCP field of ip6 tunnels is copied into the
2733 decapsulated packet.
2735 - A set of boolean bridge configuration options were added.
2736 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2737 and 'UnicastFlood=' are now parsed by networkd and applied to the
2738 respective bridge link device via the respective IFLA_BRPORT_*
2741 - A new string configuration option to override the hostname sent
2742 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2743 is true, networkd will use the configured hostname instead of the
2744 system hostname when sending DHCP requests.
2746 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2747 networkd will configure the IPv6 flow-label of the tunnel device
2748 according to RFC2460.
2750 - The 'macvtap' virtual network devices are now supported, similar to
2751 the already supported 'macvlan' devices.
2753 * systemd-resolved now implements RFC5452 to improve resilience against
2754 cache poisoning. Additionally, source port randomization is enabled
2755 by default to further protect against DNS spoofing attacks.
2757 * nss-mymachines now supports translating UIDs and GIDs of running
2758 containers with user-namespaces enabled. If a container 'foo'
2759 translates a host uid 'UID' to the container uid 'TUID', then
2760 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2761 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2762 mapped as 'vg-foo-TGID'.
2764 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2765 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2766 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2767 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2768 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2769 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2770 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2771 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2772 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2773 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2775 — Berlin, 2015-07-29
2779 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2780 There are no known issues with current sysfs, and udev does not need
2781 or should be used to work around such bugs.
2783 * udev does no longer enable USB HID power management. Several reports
2784 indicate, that some devices cannot handle that setting.
2786 * The udev accelerometer helper was removed. The functionality
2787 is now fully included in iio-sensor-proxy. But this means,
2788 older iio-sensor-proxy versions will no longer provide
2789 accelerometer/orientation data with this systemd version.
2790 Please upgrade iio-sensor-proxy to version 1.0.
2792 * networkd gained a new configuration option IPv6PrivacyExtensions=
2793 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2794 for Stateless Address") on selected networks.
2796 * For the sake of fewer build-time dependencies and less code in the
2797 main repository, the python bindings are about to be removed in the
2798 next release. A new repository has been created which accommodates
2799 the code from now on, and we kindly ask distributions to create a
2800 separate package for this. The removal will take place in v223.
2802 https://github.com/systemd/python-systemd
2804 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2805 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2806 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2807 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2808 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2809 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2810 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2811 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2812 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2813 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2815 — Berlin, 2015-07-07
2819 * The sd-bus.h and sd-event.h APIs have now been declared
2820 stable and have been added to the official interface of
2821 libsystemd.so. sd-bus implements an alternative D-Bus client
2822 library, that is relatively easy to use, very efficient and
2823 supports both classic D-Bus as well as kdbus as transport
2824 backend. sd-event is a generic event loop abstraction that
2825 is built around Linux epoll, but adds features such as event
2826 prioritization or efficient timer handling. Both APIs are good
2827 choices for C programs looking for a bus and/or event loop
2828 implementation that is minimal and does not have to be
2829 portable to other kernels.
2831 * kdbus support is no longer compile-time optional. It is now
2832 always built-in. However, it can still be disabled at
2833 runtime using the kdbus=0 kernel command line setting, and
2834 that setting may be changed to default to off, by specifying
2835 --disable-kdbus at build-time. Note though that the kernel
2836 command line setting has no effect if the kdbus.ko kernel
2837 module is not installed, in which case kdbus is (obviously)
2838 also disabled. We encourage all downstream distributions to
2839 begin testing kdbus by adding it to the kernel images in the
2840 development distributions, and leaving kdbus support in
2843 * The minimal required util-linux version has been bumped to
2846 * Support for chkconfig (--enable-chkconfig) was removed in
2847 favor of calling an abstraction tool
2848 /lib/systemd/systemd-sysv-install. This needs to be
2849 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2850 in README for details.
2852 * If there's a systemd unit and a SysV init script for the
2853 same service name, and the user executes "systemctl enable"
2854 for it (or a related call), then this will now enable both
2855 (or execute the related operation on both), not just the
2858 * The libudev API documentation has been converted from gtkdoc
2861 * gudev has been removed from the systemd tree, it is now an
2864 * The systemd-cgtop tool learnt a new --raw switch to generate
2865 "raw" (machine parsable) output.
2867 * networkd's IPForwarding= .network file setting learnt the
2868 new setting "kernel", which ensures that networkd does not
2869 change the IP forwarding sysctl from the default kernel
2872 * The systemd-logind bus API now exposes a new boolean
2873 property "Docked" that reports whether logind considers the
2874 system "docked", i.e. connected to a docking station or not.
2876 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2877 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2878 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2879 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2880 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2881 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2882 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2883 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2884 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2885 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2886 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2887 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2888 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2889 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2890 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2891 Fink, Zbigniew Jędrzejewski-Szmek
2893 — Berlin, 2015-06-19
2897 * The gudev library has been extracted into a separate repository
2898 available at: https://git.gnome.org/browse/libgudev/
2899 It is now managed as part of the Gnome project. Distributions
2900 are recommended to pass --disable-gudev to systemd and use
2901 gudev from the Gnome project instead. gudev is still included
2902 in systemd, for now. It will be removed soon, though. Please
2903 also see the announcement-thread on systemd-devel:
2904 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2906 * systemd now exposes a CPUUsageNSec= property for each
2907 service unit on the bus, that contains the overall consumed
2908 CPU time of a service (the sum of what each process of the
2909 service consumed). This value is only available if
2910 CPUAccounting= is turned on for a service, and is then shown
2911 in the "systemctl status" output.
2913 * Support for configuring alternative mappings of the old SysV
2914 runlevels to systemd targets has been removed. They are now
2915 hardcoded in a way that runlevels 2, 3, 4 all map to
2916 multi-user.target and 5 to graphical.target (which
2917 previously was already the default behaviour).
2919 * The auto-mounter logic gained support for mount point
2920 expiry, using a new TimeoutIdleSec= setting in .automount
2921 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2923 * The EFI System Partition (ESP) as mounted to /boot by
2924 systemd-efi-boot-generator will now be unmounted
2925 automatically after 2 minutes of not being used. This should
2926 minimize the risk of ESP corruptions.
2928 * New /etc/fstab options x-systemd.requires= and
2929 x-systemd.requires-mounts-for= are now supported to express
2930 additional dependencies for mounts. This is useful for
2931 journalling file systems that support external journal
2932 devices or overlay file systems that require underlying file
2933 systems to be mounted.
2935 * systemd does not support direct live-upgrades (via systemctl
2936 daemon-reexec) from versions older than v44 anymore. As no
2937 distribution we are aware of shipped such old versions in a
2938 stable release this should not be problematic.
2940 * When systemd forks off a new per-connection service instance
2941 it will now set the $REMOTE_ADDR environment variable to the
2942 remote IP address, and $REMOTE_PORT environment variable to
2943 the remote IP port. This behaviour is similar to the
2944 corresponding environment variables defined by CGI.
2946 * systemd-networkd gained support for uplink failure
2947 detection. The BindCarrier= option allows binding interface
2948 configuration dynamically to the link sense of other
2949 interfaces. This is useful to achieve behaviour like in
2952 * systemd-networkd gained support for configuring the DHCP
2953 client identifier to use when requesting leases.
2955 * systemd-networkd now has a per-network UseNTP= option to
2956 configure whether NTP server information acquired via DHCP
2957 is passed on to services like systemd-timesyncd.
2959 * systemd-networkd gained support for vti6 tunnels.
2961 * Note that systemd-networkd manages the sysctl variable
2962 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2963 it is configured for since v219. The variable controls IP
2964 forwarding, and is a per-interface alternative to the global
2965 /proc/sys/net/ipv[46]/ip_forward. This setting is
2966 configurable in the IPForward= option, which defaults to
2967 "no". This means if networkd is used for an interface it is
2968 no longer sufficient to set the global sysctl option to turn
2969 on IP forwarding! Instead, the .network file option
2970 IPForward= needs to be turned on! Note that the
2971 implementation of this behaviour was broken in v219 and has
2974 * Many bonding and vxlan options are now configurable in
2977 * systemd-nspawn gained a new --property= setting to set unit
2978 properties for the container scope. This is useful for
2979 setting resource parameters (e.g. "CPUShares=500") on
2980 containers started from the command line.
2982 * systemd-nspawn gained a new --private-users= switch to make
2983 use of user namespacing available on recent Linux kernels.
2985 * systemd-nspawn may now be called as part of a shell pipeline
2986 in which case the pipes used for stdin and stdout are passed
2987 directly to the process invoked in the container, without
2988 indirection via a pseudo tty.
2990 * systemd-nspawn gained a new switch to control the UNIX
2991 signal to use when killing the init process of the container
2994 * systemd-nspawn gained a new --overlay= switch for mounting
2995 overlay file systems into the container using the new kernel
2998 * When a container image is imported via systemd-importd and
2999 the host file system is not btrfs, a loopback block device
3000 file is created in /var/lib/machines.raw with a btrfs file
3001 system inside. It is then mounted to /var/lib/machines to
3002 enable btrfs features for container management. The loopback
3003 file and btrfs file system is grown as needed when container
3004 images are imported via systemd-importd.
3006 * systemd-machined/systemd-importd gained support for btrfs
3007 quota, to enforce container disk space limits on disk. This
3008 is exposed in "machinectl set-limit".
3010 * systemd-importd now can import containers from local .tar,
3011 .raw and .qcow2 images, and export them to .tar and .raw. It
3012 can also import dkr v2 images now from the network (on top
3015 * systemd-importd gained support for verifying downloaded
3016 images with gpg2 (previously only gpg1 was supported).
3018 * systemd-machined, systemd-logind, systemd: most bus calls
3019 are now accessible to unprivileged processes via
3020 PolicyKit. Also, systemd-logind will now allow users to kill
3021 their own sessions without further privileges or
3024 * systemd-shutdownd has been removed. This service was
3025 previously responsible for implementing scheduled shutdowns
3026 as exposed in /usr/bin/shutdown's time parameter. This
3027 functionality has now been moved into systemd-logind and is
3028 accessible via a bus interface.
3030 * "systemctl reboot" gained a new switch --firmware-setup that
3031 can be used to reboot into the EFI firmware setup, if that
3032 is available. systemd-logind now exposes an API on the bus
3033 to trigger such reboots, in case graphical desktop UIs want
3034 to cover this functionality.
3036 * "systemctl enable", "systemctl disable" and "systemctl mask"
3037 now support a new "--now" switch. If specified the units
3038 that are enabled will also be started, and the ones
3039 disabled/masked also stopped.
3041 * The Gummiboot EFI boot loader tool has been merged into
3042 systemd, and renamed to "systemd-boot". The bootctl tool has been
3043 updated to support systemd-boot.
3045 * An EFI kernel stub has been added that may be used to create
3046 kernel EFI binaries that contain not only the actual kernel,
3047 but also an initrd, boot splash, command line and OS release
3048 information. This combined binary can then be signed as a
3049 single image, so that the firmware can verify it all in one
3050 step. systemd-boot has special support for EFI binaries created
3051 like this and can extract OS release information from them
3052 and show them in the boot menu. This functionality is useful
3053 to implement cryptographically verified boot schemes.
3055 * Optional support has been added to systemd-fsck to pass
3056 fsck's progress report to an AF_UNIX socket in the file
3059 * udev will no longer create device symlinks for all block
3060 devices by default. A blacklist for excluding special block
3061 devices from this logic has been turned into a whitelist
3062 that requires picking block devices explicitly that require
3065 * A new (currently still internal) API sd-device.h has been
3066 added to libsystemd. This modernized API is supposed to
3067 replace libudev eventually. In fact, already much of libudev
3068 is now just a wrapper around sd-device.h.
3070 * A new hwdb database for storing metadata about pointing
3071 stick devices has been added.
3073 * systemd-tmpfiles gained support for setting file attributes
3074 similar to the "chattr" tool with new 'h' and 'H' lines.
3076 * systemd-journald will no longer unconditionally set the
3077 btrfs NOCOW flag on new journal files. This is instead done
3078 with tmpfiles snippet using the new 'h' line type. This
3079 allows easy disabling of this logic, by masking the
3080 journal-nocow.conf tmpfiles file.
3082 * systemd-journald will now translate audit message types to
3083 human readable identifiers when writing them to the
3084 journal. This should improve readability of audit messages.
3086 * The LUKS logic gained support for the offset= and skip=
3087 options in /etc/crypttab, as previously implemented by
3090 * /usr/lib/os-release gained a new optional field VARIANT= for
3091 distributions that support multiple variants (such as a
3092 desktop edition, a server edition, ...)
3094 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3095 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3096 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3097 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3098 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3099 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3100 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3101 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3102 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3103 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3104 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3105 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3106 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3107 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3108 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3109 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3110 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3111 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3112 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3113 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3114 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3115 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3116 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3117 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3118 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3119 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3120 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3122 — Berlin, 2015-05-22
3126 * Introduce a new API "sd-hwdb.h" for querying the hardware
3127 metadata database. With this minimal interface one can query
3128 and enumerate the udev hwdb, decoupled from the old libudev
3129 library. libudev's interface for this is now only a wrapper
3130 around sd-hwdb. A new tool systemd-hwdb has been added to
3131 interface with and update the database.
3133 * When any of systemd's tools copies files (for example due to
3134 tmpfiles' C lines) a btrfs reflink will attempted first,
3135 before bytewise copying is done.
3137 * systemd-nspawn gained a new --ephemeral switch. When
3138 specified a btrfs snapshot is taken of the container's root
3139 directory, and immediately removed when the container
3140 terminates again. Thus, a container can be started whose
3141 changes never alter the container's root directory, and are
3142 lost on container termination. This switch can also be used
3143 for starting a container off the root file system of the
3144 host without affecting the host OS. This switch is only
3145 available on btrfs file systems.
3147 * systemd-nspawn gained a new --template= switch. It takes the
3148 path to a container tree to use as template for the tree
3149 specified via --directory=, should that directory be
3150 missing. This allows instantiating containers dynamically,
3151 on first run. This switch is only available on btrfs file
3154 * When a .mount unit refers to a mount point on which multiple
3155 mounts are stacked, and the .mount unit is stopped all of
3156 the stacked mount points will now be unmounted until no
3157 mount point remains.
3159 * systemd now has an explicit notion of supported and
3160 unsupported unit types. Jobs enqueued for unsupported unit
3161 types will now fail with an "unsupported" error code. More
3162 specifically .swap, .automount and .device units are not
3163 supported in containers, .busname units are not supported on
3164 non-kdbus systems. .swap and .automount are also not
3165 supported if their respective kernel compile time options
3168 * machinectl gained support for two new "copy-from" and
3169 "copy-to" commands for copying files from a running
3170 container to the host or vice versa.
3172 * machinectl gained support for a new "bind" command to bind
3173 mount host directories into local containers. This is
3174 currently only supported for nspawn containers.
3176 * networkd gained support for configuring bridge forwarding
3177 database entries (fdb) from .network files.
3179 * A new tiny daemon "systemd-importd" has been added that can
3180 download container images in tar, raw, qcow2 or dkr formats,
3181 and make them available locally in /var/lib/machines, so
3182 that they can run as nspawn containers. The daemon can GPG
3183 verify the downloads (not supported for dkr, since it has no
3184 provisions for verifying downloads). It will transparently
3185 decompress bz2, xz, gzip compressed downloads if necessary,
3186 and restore sparse files on disk. The daemon uses privilege
3187 separation to ensure the actual download logic runs with
3188 fewer privileges than the daemon itself. machinectl has
3189 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3190 make the functionality of importd available to the
3191 user. With this in place the Fedora and Ubuntu "Cloud"
3192 images can be downloaded and booted as containers unmodified
3193 (the Fedora images lack the appropriate GPG signature files
3194 currently, so they cannot be verified, but this will change
3195 soon, hopefully). Note that downloading images is currently
3196 only fully supported on btrfs.
3198 * machinectl is now able to list container images found in
3199 /var/lib/machines, along with some metadata about sizes of
3200 disk and similar. If the directory is located on btrfs and
3201 quota is enabled, this includes quota display. A new command
3202 "image-status" has been added that shows additional
3203 information about images.
3205 * machinectl is now able to clone container images
3206 efficiently, if the underlying file system (btrfs) supports
3207 it, with the new "machinectl clone" command. It also
3208 gained commands for renaming and removing images, as well as
3209 marking them read-only or read-write (supported also on
3210 legacy file systems).
3212 * networkd gained support for collecting LLDP network
3213 announcements, from hardware that supports this. This is
3214 shown in networkctl output.
3216 * systemd-run gained support for a new -t (--pty) switch for
3217 invoking a binary on a pty whose input and output is
3218 connected to the invoking terminal. This allows executing
3219 processes as system services while interactively
3220 communicating with them via the terminal. Most interestingly
3221 this is supported across container boundaries. Invoking
3222 "systemd-run -t /bin/bash" is an alternative to running a
3223 full login session, the difference being that the former
3224 will not register a session, nor go through the PAM session
3227 * tmpfiles gained support for a new "v" line type for creating
3228 btrfs subvolumes. If the underlying file system is a legacy
3229 file system, this automatically degrades to creating a
3230 normal directory. Among others /var/lib/machines is now
3231 created like this at boot, should it be missing.
3233 * The directory /var/lib/containers/ has been deprecated and
3234 been replaced by /var/lib/machines. The term "machines" has
3235 been used in the systemd context as generic term for both
3236 VMs and containers, and hence appears more appropriate for
3237 this, as the directory can also contain raw images bootable
3240 * systemd-nspawn when invoked with -M but without --directory=
3241 or --image= is now capable of searching for the container
3242 root directory, subvolume or disk image automatically, in
3243 /var/lib/machines. systemd-nspawn@.service has been updated
3244 to make use of this, thus allowing it to be used for raw
3247 * A new machines.target unit has been introduced that is
3248 supposed to group all containers/VMs invoked as services on
3249 the system. systemd-nspawn@.service has been updated to
3250 integrate with that.
3252 * machinectl gained a new "start" command, for invoking a
3253 container as a service. "machinectl start foo" is mostly
3254 equivalent to "systemctl start systemd-nspawn@foo.service",
3255 but handles escaping in a nicer way.
3257 * systemd-nspawn will now mount most of the cgroupfs tree
3258 read-only into each container, with the exception of the
3259 container's own subtree in the name=systemd hierarchy.
3261 * journald now sets the special FS_NOCOW file flag for its
3262 journal files. This should improve performance on btrfs, by
3263 avoiding heavy fragmentation when journald's write-pattern
3264 is used on COW file systems. It degrades btrfs' data
3265 integrity guarantees for the files to the same levels as for
3266 ext3/ext4 however. This should be OK though as journald does
3267 its own data integrity checks and all its objects are
3268 checksummed on disk. Also, journald should handle btrfs disk
3269 full events a lot more gracefully now, by processing SIGBUS
3270 errors, and not relying on fallocate() anymore.
3272 * When journald detects that journal files it is writing to
3273 have been deleted it will immediately start new journal
3276 * systemd now provides a way to store file descriptors
3277 per-service in PID 1. This is useful for daemons to ensure
3278 that fds they require are not lost during a daemon
3279 restart. The fds are passed to the daemon on the next
3280 invocation in the same way socket activation fds are
3281 passed. This is now used by journald to ensure that the
3282 various sockets connected to all the system's stdout/stderr
3283 are not lost when journald is restarted. File descriptors
3284 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3285 an extension to sd_notify(). Note that a limit is enforced
3286 on the number of fds a service can store in PID 1, and it
3287 defaults to 0, so that no fds may be stored, unless this is
3288 explicitly turned on.
3290 * The default TERM variable to use for units connected to a
3291 terminal, when no other value is explicitly is set is now
3292 vt220 rather than vt102. This should be fairly safe still,
3293 but allows PgUp/PgDn work.
3295 * The /etc/crypttab option header= as known from Debian is now
3298 * "loginctl user-status" and "loginctl session-status" will
3299 now show the last 10 lines of log messages of the
3300 user/session following the status output. Similar,
3301 "machinectl status" will show the last 10 log lines
3302 associated with a virtual machine or container
3303 service. (Note that this is usually not the log messages
3304 done in the VM/container itself, but simply what the
3305 container manager logs. For nspawn this includes all console
3308 * "loginctl session-status" without further argument will now
3309 show the status of the session of the caller. Similar,
3310 "lock-session", "unlock-session", "activate",
3311 "enable-linger", "disable-linger" may now be called without
3312 session/user parameter in which case they apply to the
3313 caller's session/user.
3315 * An X11 session scriptlet is now shipped that uploads
3316 $DISPLAY and $XAUTHORITY into the environment of the systemd
3317 --user daemon if a session begins. This should improve
3318 compatibility with X11 enabled applications run as systemd
3321 * Generators are now subject to masking via /etc and /run, the
3322 same way as unit files.
3324 * networkd .network files gained support for configuring
3325 per-link IPv4/IPv6 packet forwarding as well as IPv4
3326 masquerading. This is by default turned on for veth links to
3327 containers, as registered by systemd-nspawn. This means that
3328 nspawn containers run with --network-veth will now get
3329 automatic routed access to the host's networks without any
3330 further configuration or setup, as long as networkd runs on
3333 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3334 or UDP posts of a container on the host. With this in place
3335 it is possible to run containers with private veth links
3336 (--network-veth), and have their functionality exposed on
3337 the host as if their services were running directly on the
3340 * systemd-nspawn's --network-veth switch now gained a short
3341 version "-n", since with the changes above it is now truly
3342 useful out-of-the-box. The systemd-nspawn@.service has been
3343 updated to make use of it too by default.
3345 * systemd-nspawn will now maintain a per-image R/W lock, to
3346 ensure that the same image is not started more than once
3347 writable. (It's OK to run an image multiple times
3348 simultaneously in read-only mode.)
3350 * systemd-nspawn's --image= option is now capable of
3351 dissecting and booting MBR and GPT disk images that contain
3352 only a single active Linux partition. Previously it
3353 supported only GPT disk images with proper GPT type
3354 IDs. This allows running cloud images from major
3355 distributions directly with systemd-nspawn, without
3358 * In addition to collecting mouse dpi data in the udev
3359 hardware database, there's now support for collecting angle
3360 information for mouse scroll wheels. The database is
3361 supposed to guarantee similar scrolling behavior on mice
3362 that it knows about. There's also support for collecting
3363 information about Touchpad types.
3365 * udev's input_id built-in will now also collect touch screen
3366 dimension data and attach it to probed devices.
3368 * /etc/os-release gained support for a Distribution Privacy
3371 * networkd gained support for creating "ipvlan", "gretap",
3372 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3374 * systemd-tmpfiles gained support for "a" lines for setting
3377 * systemd-nspawn will now mount /tmp in the container to
3378 tmpfs, automatically.
3380 * systemd now exposes the memory.usage_in_bytes cgroup
3381 attribute and shows it for each service in the "systemctl
3382 status" output, if available.
3384 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3385 immediate reboot is triggered. This useful if shutdown is
3386 hung and is unable to complete, to expedite the
3387 operation. Note that this kind of reboot will still unmount
3388 all file systems, and hence should not result in fsck being
3391 * A .device unit for an optical block device will now be
3392 considered active only when a medium is in the drive. Also,
3393 mount units are now bound to their backing devices thus
3394 triggering automatic unmounting when devices become
3395 unavailable. With this in place systemd will now
3396 automatically unmount left-over mounts when a CD-ROM is
3397 ejected or an USB stick is yanked from the system.
3399 * networkd-wait-online now has support for waiting for
3400 specific interfaces only (with globbing), and for giving up
3401 after a configurable timeout.
3403 * networkd now exits when idle. It will be automatically
3404 restarted as soon as interfaces show up, are removed or
3405 change state. networkd will stay around as long as there is
3406 at least one DHCP state machine or similar around, that keep
3409 * networkd may now configure IPv6 link-local addressing in
3410 addition to IPv4 link-local addressing.
3412 * The IPv6 "token" for use in SLAAC may now be configured for
3413 each .network interface in networkd.
3415 * Routes configured with networkd may now be assigned a scope
3418 * networkd's [Match] sections now support globbing and lists
3419 of multiple space-separated matches per item.
3421 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3422 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3423 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3424 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3425 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3426 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3427 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3428 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3429 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3430 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3431 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3432 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3433 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3434 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3435 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3436 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3437 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3438 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3439 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3440 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3441 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3442 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3443 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3444 Hoffmann, Zbigniew Jędrzejewski-Szmek
3446 — Berlin, 2015-02-16
3450 * When querying unit file enablement status (for example via
3451 "systemctl is-enabled"), a new state "indirect" is now known
3452 which indicates that a unit might not be enabled itself, but
3453 another unit listed in its Also= setting might be.
3455 * Similar to the various existing ConditionXYZ= settings for
3456 units, there are now matching AssertXYZ= settings. While
3457 failing conditions cause a unit to be skipped, but its job
3458 to succeed, failing assertions declared like this will cause
3459 a unit start operation and its job to fail.
3461 * hostnamed now knows a new chassis type "embedded".
3463 * systemctl gained a new "edit" command. When used on a unit
3464 file, this allows extending unit files with .d/ drop-in
3465 configuration snippets or editing the full file (after
3466 copying it from /usr/lib to /etc). This will invoke the
3467 user's editor (as configured with $EDITOR), and reload the
3468 modified configuration after editing.
3470 * "systemctl status" now shows the suggested enablement state
3471 for a unit, as declared in the (usually vendor-supplied)
3472 system preset files.
3474 * nss-myhostname will now resolve the single-label host name
3475 "gateway" to the locally configured default IP routing
3476 gateways, ordered by their metrics. This assigns a stable
3477 name to the used gateways, regardless which ones are
3478 currently configured. Note that the name will only be
3479 resolved after all other name sources (if nss-myhostname is
3480 configured properly) and should hence not negatively impact
3481 systems that use the single-label host name "gateway" in
3484 * systemd-inhibit now allows filtering by mode when listing
3487 * Scope and service units gained a new "Delegate" boolean
3488 property, which, when set, allows processes running inside the
3489 unit to further partition resources. This is primarily
3490 useful for systemd user instances as well as container
3493 * journald will now pick up audit messages directly from
3494 the kernel, and log them like any other log message. The
3495 audit fields are split up and fully indexed. This means that
3496 journalctl in many ways is now a (nicer!) alternative to
3497 ausearch, the traditional audit client. Note that this
3498 implements only a minimal audit client. If you want the
3499 special audit modes like reboot-on-log-overflow, please use
3500 the traditional auditd instead, which can be used in
3501 parallel to journald.
3503 * The ConditionSecurity= unit file option now understands the
3504 special string "audit" to check whether auditing is
3507 * journalctl gained two new commands --vacuum-size= and
3508 --vacuum-time= to delete old journal files until the
3509 remaining ones take up no more than the specified size on disk,
3510 or are not older than the specified time.
3512 * A new, native PPPoE library has been added to sd-network,
3513 systemd's library of light-weight networking protocols. This
3514 library will be used in a future version of networkd to
3515 enable PPPoE communication without an external pppd daemon.
3517 * The busctl tool now understands a new "capture" verb that
3518 works similar to "monitor", but writes a packet capture
3519 trace to STDOUT that can be redirected to a file which is
3520 compatible with libcap's capture file format. This can then
3521 be loaded in Wireshark and similar tools to inspect bus
3524 * The busctl tool now understands a new "tree" verb that shows
3525 the object trees of a specific service on the bus, or of all
3528 * The busctl tool now understands a new "introspect" verb that
3529 shows all interfaces and members of objects on the bus,
3530 including their signature and values. This is particularly
3531 useful to get more information about bus objects shown by
3532 the new "busctl tree" command.
3534 * The busctl tool now understands new verbs "call",
3535 "set-property" and "get-property" for invoking bus method
3536 calls, setting and getting bus object properties in a
3539 * busctl gained a new --augment-creds= argument that controls
3540 whether the tool shall augment credential information it
3541 gets from the bus with data from /proc, in a possibly
3544 * nspawn's --link-journal= switch gained two new values
3545 "try-guest" and "try-host" that work like "guest" and
3546 "host", but do not fail if the host has no persistent
3547 journalling enabled. -j is now equivalent to
3548 --link-journal=try-guest.
3550 * macvlan network devices created by nspawn will now have
3551 stable MAC addresses.
3553 * A new SmackProcessLabel= unit setting has been added, which
3554 controls the SMACK security label processes forked off by
3555 the respective unit shall use.
3557 * If compiled with --enable-xkbcommon, systemd-localed will
3558 verify x11 keymap settings by compiling the given keymap. It
3559 will spew out warnings if the compilation fails. This
3560 requires libxkbcommon to be installed.
3562 * When a coredump is collected, a larger number of metadata
3563 fields is now collected and included in the journal records
3564 created for it. More specifically, control group membership,
3565 environment variables, memory maps, working directory,
3566 chroot directory, /proc/$PID/status, and a list of open file
3567 descriptors is now stored in the log entry.
3569 * The udev hwdb now contains DPI information for mice. For
3572 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3574 * All systemd programs that read standalone configuration
3575 files in /etc now also support a corresponding series of
3576 .conf.d configuration directories in /etc/, /run/,
3577 /usr/local/lib/, /usr/lib/, and (if configured with
3578 --enable-split-usr) /lib/. In particular, the following
3579 configuration files now have corresponding configuration
3580 directories: system.conf user.conf, logind.conf,
3581 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3582 resolved.conf, timesyncd.conf, journal-remote.conf, and
3583 journal-upload.conf. Note that distributions should use the
3584 configuration directories in /usr/lib/; the directories in
3585 /etc/ are reserved for the system administrator.
3587 * systemd-rfkill will no longer take the rfkill device name
3588 into account when storing rfkill state on disk, as the name
3589 might be dynamically assigned and not stable. Instead, the
3590 ID_PATH udev variable combined with the rfkill type (wlan,
3591 bluetooth, ...) is used.
3593 * A new service systemd-machine-id-commit.service has been
3594 added. When used on systems where /etc is read-only during
3595 boot, and /etc/machine-id is not initialized (but an empty
3596 file), this service will copy the temporary machine ID
3597 created as replacement into /etc after the system is fully
3598 booted up. This is useful for systems that are freshly
3599 installed with a non-initialized machine ID, but should get
3600 a fixed machine ID for subsequent boots.
3602 * networkd's .netdev files now provide a large set of
3603 configuration parameters for VXLAN devices. Similarly, the
3604 bridge port cost parameter is now configurable in .network
3605 files. There's also new support for configuring IP source
3606 routing. networkd .link files gained support for a new
3607 OriginalName= match that is useful to match against the
3608 original interface name the kernel assigned. .network files
3609 may include MTU= and MACAddress= fields for altering the MTU
3610 and MAC address while being connected to a specific network
3613 * The LUKS logic gained supported for configuring
3614 UUID-specific key files. There's also new support for naming
3615 LUKS device from the kernel command line, using the new
3616 luks.name= argument.
3618 * Timer units may now be transiently created via the bus API
3619 (this was previously already available for scope and service
3620 units). In addition it is now possible to create multiple
3621 transient units at the same time with a single bus call. The
3622 "systemd-run" tool has been updated to make use of this for
3623 running commands on a specified time, in at(1)-style.
3625 * tmpfiles gained support for "t" lines, for assigning
3626 extended attributes to files. Among other uses this may be
3627 used to assign SMACK labels to files.
3629 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3630 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3631 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3632 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3633 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3634 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3635 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3636 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3637 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3638 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3639 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3640 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3641 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3642 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3643 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3644 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3645 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3646 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3648 — Berlin, 2014-12-10
3652 * journalctl gained the new options -t/--identifier= to match
3653 on the syslog identifier (aka "tag"), as well as --utc to
3654 show log timestamps in the UTC timezone. journalctl now also
3655 accepts -n/--lines=all to disable line capping in a pager.
3657 * journalctl gained a new switch, --flush, that synchronously
3658 flushes logs from /run/log/journal to /var/log/journal if
3659 persistent storage is enabled. systemd-journal-flush.service
3660 now waits until the operation is complete.
3662 * Services can notify the manager before they start a reload
3663 (by sending RELOADING=1) or shutdown (by sending
3664 STOPPING=1). This allows the manager to track and show the
3665 internal state of daemons and closes a race condition when
3666 the process is still running but has closed its D-Bus
3669 * Services with Type=oneshot do not have to have any ExecStart
3672 * User units are now loaded also from
3673 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3674 /run/systemd/user directory that was already previously
3675 supported, but is under the control of the user.
3677 * Job timeouts (i.e. time-outs on the time a job that is
3678 queued stays in the run queue) can now optionally result in
3679 immediate reboot or power-off actions (JobTimeoutAction= and
3680 JobTimeoutRebootArgument=). This is useful on ".target"
3681 units, to limit the maximum time a target remains
3682 undispatched in the run queue, and to trigger an emergency
3683 operation in such a case. This is now used by default to
3684 turn off the system if boot-up (as defined by everything in
3685 basic.target) hangs and does not complete for at least
3686 15min. Also, if power-off or reboot hang for at least 30min
3687 an immediate power-off/reboot operation is triggered. This
3688 functionality is particularly useful to increase reliability
3689 on embedded devices, but also on laptops which might
3690 accidentally get powered on when carried in a backpack and
3691 whose boot stays stuck in a hard disk encryption passphrase
3694 * systemd-logind can be configured to also handle lid switch
3695 events even when the machine is docked or multiple displays
3696 are attached (HandleLidSwitchDocked= option).
3698 * A helper binary and a service have been added which can be
3699 used to resume from hibernation in the initramfs. A
3700 generator will parse the resume= option on the kernel
3701 command line to trigger resume.
3703 * A user console daemon systemd-consoled has been
3704 added. Currently, it is a preview, and will so far open a
3705 single terminal on each session of the user marked as
3706 Desktop=systemd-console.
3708 * Route metrics can be specified for DHCP routes added by
3711 * The SELinux context of socket-activated services can be set
3712 from the information provided by the networking stack
3713 (SELinuxContextFromNet= option).
3715 * Userspace firmware loading support has been removed and
3716 the minimum supported kernel version is thus bumped to 3.7.
3718 * Timeout for udev workers has been increased from 1 to 3
3719 minutes, but a warning will be printed after 1 minute to
3720 help diagnose kernel modules that take a long time to load.
3722 * Udev rules can now remove tags on devices with TAG-="foobar".
3724 * systemd's readahead implementation has been removed. In many
3725 circumstances it didn't give expected benefits even for
3726 rotational disk drives and was becoming less relevant in the
3727 age of SSDs. As none of the developers has been using
3728 rotating media anymore, and nobody stepped up to actively
3729 maintain this component of systemd it has now been removed.
3731 * Swap units can use Options= to specify discard options.
3732 Discard options specified for swaps in /etc/fstab are now
3735 * Docker containers are now detected as a separate type of
3738 * The Password Agent protocol gained support for queries where
3739 the user input is shown, useful e.g. for user names.
3740 systemd-ask-password gained a new --echo option to turn that
3743 * The default sysctl.d/ snippets will now set:
3745 net.core.default_qdisc = fq_codel
3747 This selects Fair Queuing Controlled Delay as the default
3748 queuing discipline for network interfaces. fq_codel helps
3749 fight the network bufferbloat problem. It is believed to be
3750 a good default with no tuning required for most workloads.
3751 Downstream distributions may override this choice. On 10Gbit
3752 servers that do not do forwarding, "fq" may perform better.
3753 Systems without a good clocksource should use "pfifo_fast".
3755 * If kdbus is enabled during build a new option BusPolicy= is
3756 available for service units, that allows locking all service
3757 processes into a stricter bus policy, in order to limit
3758 access to various bus services, or even hide most of them
3759 from the service's view entirely.
3761 * networkctl will now show the .network and .link file
3762 networkd has applied to a specific interface.
3764 * sd-login gained a new API call sd_session_get_desktop() to
3765 query which desktop environment has been selected for a
3768 * UNIX utmp support is now compile-time optional to support
3769 legacy-free systems.
3771 * systemctl gained two new commands "add-wants" and
3772 "add-requires" for pulling in units from specific targets
3775 * If the word "rescue" is specified on the kernel command line
3776 the system will now boot into rescue mode (aka
3777 rescue.target), which was previously available only by
3778 specifying "1" or "systemd.unit=rescue.target" on the kernel
3779 command line. This new kernel command line option nicely
3780 mirrors the already existing "emergency" kernel command line
3783 * New kernel command line options mount.usr=, mount.usrflags=,
3784 mount.usrfstype= have been added that match root=, rootflags=,
3785 rootfstype= but allow mounting a specific file system to
3788 * The $NOTIFY_SOCKET is now also passed to control processes of
3789 services, not only the main process.
3791 * This version reenables support for fsck's -l switch. This
3792 means at least version v2.25 of util-linux is required for
3793 operation, otherwise dead-locks on device nodes may
3794 occur. Again: you need to update util-linux to at least
3795 v2.25 when updating systemd to v217.
3797 * The "multi-seat-x" tool has been removed from systemd, as
3798 its functionality has been integrated into X servers 1.16,
3799 and the tool is hence redundant. It is recommended to update
3800 display managers invoking this tool to simply invoke X
3801 directly from now on, again.
3803 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3804 message flag has been added for all of systemd's PolicyKit
3805 authenticated method calls has been added. In particular
3806 this now allows optional interactive authorization via
3807 PolicyKit for many of PID1's privileged operations such as
3808 unit file enabling and disabling.
3810 * "udevadm hwdb --update" learnt a new switch "--usr" for
3811 placing the rebuilt hardware database in /usr instead of
3812 /etc. When used only hardware database entries stored in
3813 /usr will be used, and any user database entries in /etc are
3814 ignored. This functionality is useful for vendors to ship a
3815 pre-built database on systems where local configuration is
3816 unnecessary or unlikely.
3818 * Calendar time specifications in .timer units now also
3819 understand the strings "semi-annually", "quarterly" and
3820 "minutely" as shortcuts (in addition to the preexisting
3821 "anually", "hourly", ...).
3823 * systemd-tmpfiles will now correctly create files in /dev
3824 at boot which are marked for creation only at boot. It is
3825 recommended to always create static device nodes with 'c!'
3826 and 'b!', so that they are created only at boot and not
3827 overwritten at runtime.
3829 * When the watchdog logic is used for a service (WatchdogSec=)
3830 and the watchdog timeout is hit the service will now be
3831 terminated with SIGABRT (instead of just SIGTERM), in order
3832 to make sure a proper coredump and backtrace is
3833 generated. This ensures that hanging services will result in
3834 similar coredump/backtrace behaviour as services that hit a
3837 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3838 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3839 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3840 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3841 Herrmann, David Sommerseth, David Strauss, Emil Renner
3842 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3843 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3844 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3845 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3846 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3847 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3848 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3849 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3850 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3851 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3852 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3853 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3854 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3855 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3856 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3857 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3860 — Berlin, 2014-10-28
3864 * timedated no longer reads NTP implementation unit names from
3865 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3866 implementations should add a
3868 Conflicts=systemd-timesyncd.service
3870 to their unit files to take over and replace systemd's NTP
3871 default functionality.
3873 * systemd-sysusers gained a new line type "r" for configuring
3874 which UID/GID ranges to allocate system users/groups
3875 from. Lines of type "u" may now add an additional column
3876 that specifies the home directory for the system user to be
3877 created. Also, systemd-sysusers may now optionally read user
3878 information from STDIN instead of a file. This is useful for
3879 invoking it from RPM preinst scriptlets that need to create
3880 users before the first RPM file is installed since these
3881 files might need to be owned by them. A new
3882 %sysusers_create_inline RPM macro has been introduced to do
3883 just that. systemd-sysusers now updates the shadow files as
3884 well as the user/group databases, which should enhance
3885 compatibility with certain tools like grpck.
3887 * A number of bus APIs of PID 1 now optionally consult
3888 PolicyKit to permit access for otherwise unprivileged
3889 clients under certain conditions. Note that this currently
3890 doesn't support interactive authentication yet, but this is
3891 expected to be added eventually, too.
3893 * /etc/machine-info now has new fields for configuring the
3894 deployment environment of the machine, as well as the
3895 location of the machine. hostnamectl has been updated with
3896 new command to update these fields.
3898 * systemd-timesyncd has been updated to automatically acquire
3899 NTP server information from systemd-networkd, which might
3900 have been discovered via DHCP.
3902 * systemd-resolved now includes a caching DNS stub resolver
3903 and a complete LLMNR name resolution implementation. A new
3904 NSS module "nss-resolve" has been added which can be used
3905 instead of glibc's own "nss-dns" to resolve hostnames via
3906 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3907 be resolved via systemd-resolved D-Bus APIs. In contrast to
3908 the glibc internal resolver systemd-resolved is aware of
3909 multi-homed system, and keeps DNS server and caches separate
3910 and per-interface. Queries are sent simultaneously on all
3911 interfaces that have DNS servers configured, in order to
3912 properly handle VPNs and local LANs which might resolve
3913 separate sets of domain names. systemd-resolved may acquire
3914 DNS server information from systemd-networkd automatically,
3915 which in turn might have discovered them via DHCP. A tool
3916 "systemd-resolve-host" has been added that may be used to
3917 query the DNS logic in resolved. systemd-resolved implements
3918 IDNA and automatically uses IDNA or UTF-8 encoding depending
3919 on whether classic DNS or LLMNR is used as transport. In the
3920 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3921 implementation to systemd-resolved.
3923 * A new NSS module nss-mymachines has been added, that
3924 automatically resolves the names of all local registered
3925 containers to their respective IP addresses.
3927 * A new client tool "networkctl" for systemd-networkd has been
3928 added. It currently is entirely passive and will query
3929 networking configuration from udev, rtnetlink and networkd,
3930 and present it to the user in a very friendly
3931 way. Eventually, we hope to extend it to become a full
3932 control utility for networkd.
3934 * .socket units gained a new DeferAcceptSec= setting that
3935 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3936 TCP. Similarly, support for controlling TCP keep-alive
3937 settings has been added (KeepAliveTimeSec=,
3938 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3939 turning off Nagle's algorithm on TCP has been added
3942 * logind learned a new session type "web", for use in projects
3943 like Cockpit which register web clients as PAM sessions.
3945 * timer units with at least one OnCalendar= setting will now
3946 be started only after time-sync.target has been
3947 reached. This way they will not elapse before the system
3948 clock has been corrected by a local NTP client or
3949 similar. This is particular useful on RTC-less embedded
3950 machines, that come up with an invalid system clock.
3952 * systemd-nspawn's --network-veth= switch should now result in
3953 stable MAC addresses for both the outer and the inner side
3956 * systemd-nspawn gained a new --volatile= switch for running
3957 container instances with /etc or /var unpopulated.
3959 * The kdbus client code has been updated to use the new Linux
3960 3.17 memfd subsystem instead of the old kdbus-specific one.
3962 * systemd-networkd's DHCP client and server now support
3963 FORCERENEW. There are also new configuration options to
3964 configure the vendor client identifier and broadcast mode
3967 * systemd will no longer inform the kernel about the current
3968 timezone, as this is necessarily incorrect and racy as the
3969 kernel has no understanding of DST and similar
3970 concepts. This hence means FAT timestamps will be always
3971 considered UTC, similar to what Android is already
3972 doing. Also, when the RTC is configured to the local time
3973 (rather than UTC) systemd will never synchronize back to it,
3974 as this might confuse Windows at a later boot.
3976 * systemd-analyze gained a new command "verify" for offline
3977 validation of unit files.
3979 * systemd-networkd gained support for a couple of additional
3980 settings for bonding networking setups. Also, the metric for
3981 statically configured routes may now be configured. For
3982 network interfaces where this is appropriate the peer IP
3983 address may now be configured.
3985 * systemd-networkd's DHCP client will no longer request
3986 broadcasting by default, as this tripped up some networks.
3987 For hardware where broadcast is required the feature should
3988 be switched back on using RequestBroadcast=yes.
3990 * systemd-networkd will now set up IPv4LL addresses (when
3991 enabled) even if DHCP is configured successfully.
3993 * udev will now default to respect network device names given
3994 by the kernel when the kernel indicates that these are
3995 predictable. This behavior can be tweaked by changing
3996 NamePolicy= in the relevant .link file.
3998 * A new library systemd-terminal has been added that
3999 implements full TTY stream parsing and rendering. This
4000 library is supposed to be used later on for implementing a
4001 full userspace VT subsystem, replacing the current kernel
4004 * A new tool systemd-journal-upload has been added to push
4005 journal data to a remote system running
4006 systemd-journal-remote.
4008 * journald will no longer forward all local data to another
4009 running syslog daemon. This change has been made because
4010 rsyslog (which appears to be the most commonly used syslog
4011 implementation these days) no longer makes use of this, and
4012 instead pulls the data out of the journal on its own. Since
4013 forwarding the messages to a non-existent syslog server is
4014 more expensive than we assumed we have now turned this
4015 off. If you run a syslog server that is not a recent rsyslog
4016 version, you have to turn this option on again
4017 (ForwardToSyslog= in journald.conf).
4019 * journald now optionally supports the LZ4 compressor for
4020 larger journal fields. This compressor should perform much
4021 better than XZ which was the previous default.
4023 * machinectl now shows the IP addresses of local containers,
4024 if it knows them, plus the interface name of the container.
4026 * A new tool "systemd-escape" has been added that makes it
4027 easy to escape strings to build unit names and similar.
4029 * sd_notify() messages may now include a new ERRNO= field
4030 which is parsed and collected by systemd and shown among the
4031 "systemctl status" output for a service.
4033 * A new component "systemd-firstboot" has been added that
4034 queries the most basic systemd information (timezone,
4035 hostname, root password) interactively on first
4036 boot. Alternatively it may also be used to provision these
4037 things offline on OS images installed into directories.
4039 * The default sysctl.d/ snippets will now set
4041 net.ipv4.conf.default.promote_secondaries=1
4043 This has the benefit of no flushing secondary IP addresses
4044 when primary addresses are removed.
4046 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4047 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4048 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4049 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4050 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4051 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4052 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4053 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4054 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4055 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4056 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4057 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4058 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4059 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4060 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4062 — Berlin, 2014-08-19
4066 * A new tool systemd-sysusers has been added. This tool
4067 creates system users and groups in /etc/passwd and
4068 /etc/group, based on static declarative system user/group
4069 definitions in /usr/lib/sysusers.d/. This is useful to
4070 enable factory resets and volatile systems that boot up with
4071 an empty /etc directory, and thus need system users and
4072 groups created during early boot. systemd now also ships
4073 with two default sysusers.d/ files for the most basic
4074 users and groups systemd and the core operating system
4077 * A new tmpfiles snippet has been added that rebuilds the
4078 essential files in /etc on boot, should they be missing.
4080 * A directive for ensuring automatic clean-up of
4081 /var/cache/man/ has been removed from the default
4082 configuration. This line should now be shipped by the man
4083 implementation. The necessary change has been made to the
4084 man-db implementation. Note that you need to update your man
4085 implementation to one that ships this line, otherwise no
4086 automatic clean-up of /var/cache/man will take place.
4088 * A new condition ConditionNeedsUpdate= has been added that
4089 may conditionalize services to only run when /etc or /var
4090 are "older" than the vendor operating system resources in
4091 /usr. This is useful for reconstructing or updating /etc
4092 after an offline update of /usr or a factory reset, on the
4093 next reboot. Services that want to run once after such an
4094 update or reset should use this condition and order
4095 themselves before the new systemd-update-done.service, which
4096 will mark the two directories as fully updated. A number of
4097 service files have been added making use of this, to rebuild
4098 the udev hardware database, the journald message catalog and
4099 dynamic loader cache (ldconfig). The systemd-sysusers tool
4100 described above also makes use of this now. With this in
4101 place it is now possible to start up a minimal operating
4102 system with /etc empty cleanly. For more information on the
4103 concepts involved see this recent blog story:
4105 http://0pointer.de/blog/projects/stateless.html
4107 * A new system group "input" has been introduced, and all
4108 input device nodes get this group assigned. This is useful
4109 for system-level software to get access to input devices. It
4110 complements what is already done for "audio" and "video".
4112 * systemd-networkd learnt minimal DHCPv4 server support in
4113 addition to the existing DHCPv4 client support. It also
4114 learnt DHCPv6 client and IPv6 Router Solicitation client
4115 support. The DHCPv4 client gained support for static routes
4116 passed in from the server. Note that the [DHCPv4] section
4117 known in older systemd-networkd versions has been renamed to
4118 [DHCP] and is now also used by the DHCPv6 client. Existing
4119 .network files using settings of this section should be
4120 updated, though compatibility is maintained. Optionally, the
4121 client hostname may now be sent to the DHCP server.
4123 * networkd gained support for vxlan virtual networks as well
4124 as tun/tap and dummy devices.
4126 * networkd gained support for automatic allocation of address
4127 ranges for interfaces from a system-wide pool of
4128 addresses. This is useful for dynamically managing a large
4129 number of interfaces with a single network configuration
4130 file. In particular this is useful to easily assign
4131 appropriate IP addresses to the veth links of a large number
4132 of nspawn instances.
4134 * RPM macros for processing sysusers, sysctl and binfmt
4135 drop-in snippets at package installation time have been
4138 * The /etc/os-release file should now be placed in
4139 /usr/lib/os-release. The old location is automatically
4140 created as symlink. /usr/lib is the more appropriate
4141 location of this file, since it shall actually describe the
4142 vendor operating system shipped in /usr, and not the
4143 configuration stored in /etc.
4145 * .mount units gained a new boolean SloppyOptions= setting
4146 that maps to mount(8)'s -s option which enables permissive
4147 parsing of unknown mount options.
4149 * tmpfiles learnt a new "L+" directive which creates a symlink
4150 but (unlike "L") deletes a pre-existing file first, should
4151 it already exist and not already be the correct
4152 symlink. Similarly, "b+", "c+" and "p+" directives have been
4153 added as well, which create block and character devices, as
4154 well as fifos in the filesystem, possibly removing any
4155 pre-existing files of different types.
4157 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4158 'argument' field (which so far specified the source to
4159 symlink/copy the files from) is now optional. If omitted the
4160 same file os copied from /usr/share/factory/ suffixed by the
4161 full destination path. This is useful for populating /etc
4162 with essential files, by copying them from vendor defaults
4163 shipped in /usr/share/factory/etc.
4165 * A new command "systemctl preset-all" has been added that
4166 applies the service preset settings to all installed unit
4167 files. A new switch --preset-mode= has been added that
4168 controls whether only enable or only disable operations
4171 * A new command "systemctl is-system-running" has been added
4172 that allows checking the overall state of the system, for
4173 example whether it is fully up and running.
4175 * When the system boots up with an empty /etc, the equivalent
4176 to "systemctl preset-all" is executed during early boot, to
4177 make sure all default services are enabled after a factory
4180 * systemd now contains a minimal preset file that enables the
4181 most basic services systemd ships by default.
4183 * Unit files' [Install] section gained a new DefaultInstance=
4184 field for defining the default instance to create if a
4185 template unit is enabled with no instance specified.
4187 * A new passive target cryptsetup-pre.target has been added
4188 that may be used by services that need to make they run and
4189 finish before the first LUKS cryptographic device is set up.
4191 * The /dev/loop-control and /dev/btrfs-control device nodes
4192 are now owned by the "disk" group by default, opening up
4193 access to this group.
4195 * systemd-coredump will now automatically generate a
4196 stack trace of all core dumps taking place on the system,
4197 based on elfutils' libdw library. This stack trace is logged
4200 * systemd-coredump may now optionally store coredumps directly
4201 on disk (in /var/lib/systemd/coredump, possibly compressed),
4202 instead of storing them unconditionally in the journal. This
4203 mode is the new default. A new configuration file
4204 /etc/systemd/coredump.conf has been added to configure this
4205 and other parameters of systemd-coredump.
4207 * coredumpctl gained a new "info" verb to show details about a
4208 specific coredump. A new switch "-1" has also been added
4209 that makes sure to only show information about the most
4210 recent entry instead of all entries. Also, as the tool is
4211 generally useful now the "systemd-" prefix of the binary
4212 name has been removed. Distributions that want to maintain
4213 compatibility with the old name should add a symlink from
4214 the old name to the new name.
4216 * journald's SplitMode= now defaults to "uid". This makes sure
4217 that unprivileged users can access their own coredumps with
4218 coredumpctl without restrictions.
4220 * New kernel command line options "systemd.wants=" (for
4221 pulling an additional unit during boot), "systemd.mask="
4222 (for masking a specific unit for the boot), and
4223 "systemd.debug-shell" (for enabling the debug shell on tty9)
4224 have been added. This is implemented in the new generator
4225 "systemd-debug-generator".
4227 * systemd-nspawn will now by default filter a couple of
4228 syscalls for containers, among them those required for
4229 kernel module loading, direct x86 IO port access, swap
4230 management, and kexec. Most importantly though
4231 open_by_handle_at() is now prohibited for containers,
4232 closing a hole similar to a recently discussed vulnerability
4233 in docker regarding access to files on file hierarchies the
4234 container should normally not have access to. Note that, for
4235 nspawn, we generally make no security claims anyway (and
4236 this is explicitly documented in the man page), so this is
4237 just a fix for one of the most obvious problems.
4239 * A new man page file-hierarchy(7) has been added that
4240 contains a minimized, modernized version of the file system
4241 layout systemd expects, similar in style to the FHS
4242 specification or hier(5). A new tool systemd-path(1) has
4243 been added to query many of these paths for the local
4246 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4247 longer done. Since the directory now has a per-user size
4248 limit, and is cleaned on logout this appears unnecessary,
4249 in particular since this now brings the lifecycle of this
4250 directory closer in line with how IPC objects are handled.
4252 * systemd.pc now exports a number of additional directories,
4253 including $libdir (which is useful to identify the library
4254 path for the primary architecture of the system), and a
4255 couple of drop-in directories.
4257 * udev's predictable network interface names now use the dev_port
4258 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4259 distinguish between ports of the same PCI function. dev_id should
4260 only be used for ports using the same HW address, hence the need
4263 * machined has been updated to export the OS version of a
4264 container (read from /etc/os-release and
4265 /usr/lib/os-release) on the bus. This is now shown in
4266 "machinectl status" for a machine.
4268 * A new service setting RestartForceExitStatus= has been
4269 added. If configured to a set of exit signals or process
4270 return values, the service will be restarted when the main
4271 daemon process exits with any of them, regardless of the
4274 * systemctl's -H switch for connecting to remote systemd
4275 machines has been extended so that it may be used to
4276 directly connect to a specific container on the
4277 host. "systemctl -H root@foobar:waldi" will now connect as
4278 user "root" to host "foobar", and then proceed directly to
4279 the container named "waldi". Note that currently you have to
4280 authenticate as user "root" for this to work, as entering
4281 containers is a privileged operation.
4283 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4284 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4285 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4286 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4287 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4288 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4289 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4290 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4291 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4292 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4293 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4294 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4296 — Berlin, 2014-07-03
4300 * As an experimental feature, udev now tries to lock the
4301 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4302 executes events for the disk or any of its partitions.
4303 Applications like partitioning programs can lock the
4304 disk device node (flock(LOCK_EX)) and claim temporary
4305 device ownership that way; udev will entirely skip all event
4306 handling for this disk and its partitions. If the disk
4307 was opened for writing, the close will trigger a partition
4308 table rescan in udev's "watch" facility, and if needed
4309 synthesize "change" events for the disk and all its partitions.
4310 This is now unconditionally enabled, and if it turns out to
4311 cause major problems, we might turn it on only for specific
4312 devices, or might need to disable it entirely. Device Mapper
4313 devices are excluded from this logic.
4315 * We temporarily dropped the "-l" switch for fsck invocations,
4316 since they collide with the flock() logic above. util-linux
4317 upstream has been changed already to avoid this conflict,
4318 and we will readd "-l" as soon as util-linux with this
4319 change has been released.
4321 * The dependency on libattr has been removed. Since a long
4322 time, the extended attribute calls have moved to glibc, and
4323 libattr is thus unnecessary.
4325 * Virtualization detection works without privileges now. This
4326 means the systemd-detect-virt binary no longer requires
4327 CAP_SYS_PTRACE file capabilities, and our daemons can run
4328 with fewer privileges.
4330 * systemd-networkd now runs under its own "systemd-network"
4331 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4332 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4333 loses the ability to write to files owned by root this way.
4335 * Similarly, systemd-resolved now runs under its own
4336 "systemd-resolve" user with no capabilities remaining.
4338 * Similarly, systemd-bus-proxyd now runs under its own
4339 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4341 * systemd-networkd gained support for setting up "veth"
4342 virtual Ethernet devices for container connectivity, as well
4343 as GRE and VTI tunnels.
4345 * systemd-networkd will no longer automatically attempt to
4346 manually load kernel modules necessary for certain tunnel
4347 transports. Instead, it is assumed the kernel loads them
4348 automatically when required. This only works correctly on
4349 very new kernels. On older kernels, please consider adding
4350 the kernel modules to /etc/modules-load.d/ as a work-around.
4352 * The resolv.conf file systemd-resolved generates has been
4353 moved to /run/systemd/resolve/. If you have a symlink from
4354 /etc/resolv.conf, it might be necessary to correct it.
4356 * Two new service settings, ProtectHome= and ProtectSystem=,
4357 have been added. When enabled, they will make the user data
4358 (such as /home) inaccessible or read-only and the system
4359 (such as /usr) read-only, for specific services. This allows
4360 very light-weight per-service sandboxing to avoid
4361 modifications of user data or system files from
4362 services. These two new switches have been enabled for all
4363 of systemd's long-running services, where appropriate.
4365 * Socket units gained new SocketUser= and SocketGroup=
4366 settings to set the owner user and group of AF_UNIX sockets
4367 and FIFOs in the file system.
4369 * Socket units gained a new RemoveOnStop= setting. If enabled,
4370 all FIFOS and sockets in the file system will be removed
4371 when the specific socket unit is stopped.
4373 * Socket units gained a new Symlinks= setting. It takes a list
4374 of symlinks to create to file system sockets or FIFOs
4375 created by the specific Unix sockets. This is useful to
4376 manage symlinks to socket nodes with the same life-cycle as
4379 * The /dev/log socket and /dev/initctl FIFO have been moved to
4380 /run, and have been replaced by symlinks. This allows
4381 connecting to these facilities even if PrivateDevices=yes is
4382 used for a service (which makes /dev/log itself unavailable,
4383 but /run is left). This also has the benefit of ensuring
4384 that /dev only contains device nodes, directories and
4385 symlinks, and nothing else.
4387 * sd-daemon gained two new calls sd_pid_notify() and
4388 sd_pid_notifyf(). They are similar to sd_notify() and
4389 sd_notifyf(), but allow overriding of the source PID of
4390 notification messages if permissions permit this. This is
4391 useful to send notify messages on behalf of a different
4392 process (for example, the parent process). The
4393 systemd-notify tool has been updated to make use of this
4394 when sending messages (so that notification messages now
4395 originate from the shell script invoking systemd-notify and
4396 not the systemd-notify process itself. This should minimize
4397 a race where systemd fails to associate notification
4398 messages to services when the originating process already
4401 * A new "on-abnormal" setting for Restart= has been added. If
4402 set, it will result in automatic restarts on all "abnormal"
4403 reasons for a process to exit, which includes unclean
4404 signals, core dumps, timeouts and watchdog timeouts, but
4405 does not include clean and unclean exit codes or clean
4406 signals. Restart=on-abnormal is an alternative for
4407 Restart=on-failure for services that shall be able to
4408 terminate and avoid restarts on certain errors, by
4409 indicating so with an unclean exit code. Restart=on-failure
4410 or Restart=on-abnormal is now the recommended setting for
4411 all long-running services.
4413 * If the InaccessibleDirectories= service setting points to a
4414 mount point (or if there are any submounts contained within
4415 it), it is now attempted to completely unmount it, to make
4416 the file systems truly unavailable for the respective
4419 * The ReadOnlyDirectories= service setting and
4420 systemd-nspawn's --read-only parameter are now recursively
4421 applied to all submounts, too.
4423 * Mount units may now be created transiently via the bus APIs.
4425 * The support for SysV and LSB init scripts has been removed
4426 from the systemd daemon itself. Instead, it is now
4427 implemented as a generator that creates native systemd units
4428 from these scripts when needed. This enables us to remove a
4429 substantial amount of legacy code from PID 1, following the
4430 fact that many distributions only ship a very small number
4431 of LSB/SysV init scripts nowadays.
4433 * Privileged Xen (dom0) domains are not considered
4434 virtualization anymore by the virtualization detection
4435 logic. After all, they generally have unrestricted access to
4436 the hardware and usually are used to manage the unprivileged
4439 * systemd-tmpfiles gained a new "C" line type, for copying
4440 files or entire directories.
4442 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4443 lines. So far, they have been non-globbing versions of the
4444 latter, and have thus been redundant. In future, it is
4445 recommended to only use "z". "m" has hence been removed
4446 from the documentation, even though it stays supported.
4448 * A tmpfiles snippet to recreate the most basic structure in
4449 /var has been added. This is enough to create the /var/run →
4450 /run symlink and create a couple of structural
4451 directories. This allows systems to boot up with an empty or
4452 volatile /var. Of course, while with this change, the core OS
4453 now is capable with dealing with a volatile /var, not all
4454 user services are ready for it. However, we hope that sooner
4455 or later, many service daemons will be changed upstream so
4456 that they are able to automatically create their necessary
4457 directories in /var at boot, should they be missing. This is
4458 the first step to allow state-less systems that only require
4459 the vendor image for /usr to boot.
4461 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4462 empty tmpfs instance to a specific directory. This is
4463 particularly useful for making use of the automatic
4464 reconstruction of /var (see above), by passing --tmpfs=/var.
4466 * Access modes specified in tmpfiles snippets may now be
4467 prefixed with "~", which indicates that they shall be masked
4468 by whether the existing file or directory is currently
4469 writable, readable or executable at all. Also, if specified,
4470 the sgid/suid/sticky bits will be masked for all
4473 * A new passive target unit "network-pre.target" has been
4474 added which is useful for services that shall run before any
4475 network is configured, for example firewall scripts.
4477 * The "floppy" group that previously owned the /dev/fd*
4478 devices is no longer used. The "disk" group is now used
4479 instead. Distributions should probably deprecate usage of
4482 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4483 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4484 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4485 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4486 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4487 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4490 — Berlin, 2014-06-11
4494 * A new "systemd-timesyncd" daemon has been added for
4495 synchronizing the system clock across the network. It
4496 implements an SNTP client. In contrast to NTP
4497 implementations such as chrony or the NTP reference server,
4498 this only implements a client side, and does not bother with
4499 the full NTP complexity, focusing only on querying time from
4500 one remote server and synchronizing the local clock to
4501 it. Unless you intend to serve NTP to networked clients or
4502 want to connect to local hardware clocks, this simple NTP
4503 client should be more than appropriate for most
4504 installations. The daemon runs with minimal privileges, and
4505 has been hooked up with networkd to only operate when
4506 network connectivity is available. The daemon saves the
4507 current clock to disk every time a new NTP sync has been
4508 acquired, and uses this to possibly correct the system clock
4509 early at bootup, in order to accommodate for systems that
4510 lack an RTC such as the Raspberry Pi and embedded devices,
4511 and to make sure that time monotonically progresses on these
4512 systems, even if it is not always correct. To make use of
4513 this daemon, a new system user and group "systemd-timesync"
4514 needs to be created on installation of systemd.
4516 * The queue "seqnum" interface of libudev has been disabled, as
4517 it was generally incompatible with device namespacing as
4518 sequence numbers of devices go "missing" if the devices are
4519 part of a different namespace.
4521 * "systemctl list-timers" and "systemctl list-sockets" gained
4522 a --recursive switch for showing units of these types also
4523 for all local containers, similar in style to the already
4524 supported --recursive switch for "systemctl list-units".
4526 * A new RebootArgument= setting has been added for service
4527 units, which may be used to specify a kernel reboot argument
4528 to use when triggering reboots with StartLimitAction=.
4530 * A new FailureAction= setting has been added for service
4531 units which may be used to specify an operation to trigger
4532 when a service fails. This works similarly to
4533 StartLimitAction=, but unlike it, controls what is done
4534 immediately rather than only after several attempts to
4535 restart the service in question.
4537 * hostnamed got updated to also expose the kernel name,
4538 release, and version on the bus. This is useful for
4539 executing commands like hostnamectl with the -H switch.
4540 systemd-analyze makes use of this to properly display
4541 details when running non-locally.
4543 * The bootchart tool can now show cgroup information in the
4544 graphs it generates.
4546 * The CFS CPU quota cgroup attribute is now exposed for
4547 services. The new CPUQuota= switch has been added for this
4548 which takes a percentage value. Setting this will have the
4549 result that a service may never get more CPU time than the
4550 specified percentage, even if the machine is otherwise idle.
4552 * systemd-networkd learned IPIP and SIT tunnel support.
4554 * LSB init scripts exposing a dependency on $network will now
4555 get a dependency on network-online.target rather than simply
4556 network.target. This should bring LSB handling closer to
4557 what it was on SysV systems.
4559 * A new fsck.repair= kernel option has been added to control
4560 how fsck shall deal with unclean file systems at boot.
4562 * The (.ini) configuration file parser will now silently
4563 ignore sections whose name begins with "X-". This may be
4564 used to maintain application-specific extension sections in unit
4567 * machined gained a new API to query the IP addresses of
4568 registered containers. "machinectl status" has been updated
4569 to show these addresses in its output.
4571 * A new call sd_uid_get_display() has been added to the
4572 sd-login APIs for querying the "primary" session of a
4573 user. The "primary" session of the user is elected from the
4574 user's sessions and generally a graphical session is
4575 preferred over a text one.
4577 * A minimal systemd-resolved daemon has been added. It
4578 currently simply acts as a companion to systemd-networkd and
4579 manages resolv.conf based on per-interface DNS
4580 configuration, possibly supplied via DHCP. In the long run
4581 we hope to extend this into a local DNSSEC enabled DNS and
4584 * The systemd-networkd-wait-online tool is now enabled by
4585 default. It will delay network-online.target until a network
4586 connection has been configured. The tool primarily integrates
4587 with networkd, but will also make a best effort to make sense
4588 of network configuration performed in some other way.
4590 * Two new service options StartupCPUShares= and
4591 StartupBlockIOWeight= have been added that work similarly to
4592 CPUShares= and BlockIOWeight= however only apply during
4593 system startup. This is useful to prioritize certain services
4594 differently during bootup than during normal runtime.
4596 * hostnamed has been changed to prefer the statically
4597 configured hostname in /etc/hostname (unless set to
4598 'localhost' or empty) over any dynamic one supplied by
4599 dhcp. With this change, the rules for picking the hostname
4600 match more closely the rules of other configuration settings
4601 where the local administrator's configuration in /etc always
4602 overrides any other settings.
4604 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4605 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4606 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4607 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4608 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4609 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4610 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4611 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4612 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4613 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4614 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4615 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4616 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4617 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4618 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4619 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4622 — Beijing, 2014-05-28
4626 * When restoring the screen brightness at boot, stay away from
4627 the darkest setting or from the lowest 5% of the available
4628 range, depending on which is the larger value of both. This
4629 should effectively protect the user from rebooting into a
4630 black screen, should the brightness have been set to minimum
4633 * sd-login gained a new sd_machine_get_class() call to
4634 determine the class ("vm" or "container") of a machine
4635 registered with machined.
4637 * sd-login gained new calls
4638 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4639 to query the identity of the peer of a local AF_UNIX
4640 connection. They operate similarly to their sd_pid_get_xyz()
4643 * PID 1 will now maintain a system-wide system state engine
4644 with the states "starting", "running", "degraded",
4645 "maintenance", "stopping". These states are bound to system
4646 startup, normal runtime, runtime with at least one failed
4647 service, rescue/emergency mode and system shutdown. This
4648 state is shown in the "systemctl status" output when no unit
4649 name is passed. It is useful to determine system state, in
4650 particularly when doing so for many systems or containers at
4653 * A new command "list-machines" has been added to "systemctl"
4654 that lists all local OS containers and shows their system
4655 state (see above), if systemd runs inside of them.
4657 * systemctl gained a new "-r" switch to recursively enumerate
4658 units on all local containers, when used with the
4659 "list-unit" command (which is the default one that is
4660 executed when no parameters are specified).
4662 * The GPT automatic partition discovery logic will now honour
4663 two GPT partition flags: one may be set on a partition to
4664 cause it to be mounted read-only, and the other may be set
4665 on a partition to ignore it during automatic discovery.
4667 * Two new GPT type UUIDs have been added for automatic root
4668 partition discovery, for 32-bit and 64-bit ARM. This is not
4669 particularly useful for discovering the root directory on
4670 these architectures during bare-metal boots (since UEFI is
4671 not common there), but still very useful to allow booting of
4672 ARM disk images in nspawn with the -i option.
4674 * MAC addresses of interfaces created with nspawn's
4675 --network-interface= switch will now be generated from the
4676 machine name, and thus be stable between multiple invocations
4679 * logind will now automatically remove all IPC objects owned
4680 by a user if she or he fully logs out. This makes sure that
4681 users who are logged out cannot continue to consume IPC
4682 resources. This covers SysV memory, semaphores and message
4683 queues as well as POSIX shared memory and message
4684 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4685 limits. With this functionality, that is corrected. This may
4686 be turned off by using the RemoveIPC= switch of logind.conf.
4688 * The systemd-machine-id-setup and tmpfiles tools gained a
4689 --root= switch to operate on a specific root directory,
4692 * journald can now forward logged messages to the TTYs of all
4693 logged in users ("wall"). This is the default for all
4694 emergency messages now.
4696 * A new tool systemd-journal-remote has been added to stream
4697 journal log messages across the network.
4699 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4700 controller trees are mounted into it. Note that the
4701 directories mounted beneath it are not read-only. This is a
4702 security measure and is particularly useful because glibc
4703 actually includes a search logic to pick any tmpfs it can
4704 find to implement shm_open() if /dev/shm is not available
4705 (which it might very well be in namespaced setups).
4707 * machinectl gained a new "poweroff" command to cleanly power
4708 down a local OS container.
4710 * The PrivateDevices= unit file setting will now also drop the
4711 CAP_MKNOD capability from the capability bound set, and
4712 imply DevicePolicy=closed.
4714 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4715 comprehensively on all long-running systemd services where
4716 this is appropriate.
4718 * systemd-udevd will now run in a disassociated mount
4719 namespace. To mount directories from udev rules, make sure to
4720 pull in mount units via SYSTEMD_WANTS properties.
4722 * The kdbus support gained support for uploading policy into
4723 the kernel. sd-bus gained support for creating "monitoring"
4724 connections that can eavesdrop into all bus communication
4725 for debugging purposes.
4727 * Timestamps may now be specified in seconds since the UNIX
4728 epoch Jan 1st, 1970 by specifying "@" followed by the value
4731 * Native tcpwrap support in systemd has been removed. tcpwrap
4732 is old code, not really maintained anymore and has serious
4733 shortcomings, and better options such as firewalls
4734 exist. For setups that require tcpwrap usage, please
4735 consider invoking your socket-activated service via tcpd,
4736 like on traditional inetd.
4738 * A new system.conf configuration option
4739 DefaultTimerAccuracySec= has been added that controls the
4740 default AccuracySec= setting of .timer units.
4742 * Timer units gained a new WakeSystem= switch. If enabled,
4743 timers configured this way will cause the system to resume
4744 from system suspend (if the system supports that, which most
4747 * Timer units gained a new Persistent= switch. If enabled,
4748 timers configured this way will save to disk when they have
4749 been last triggered. This information is then used on next
4750 reboot to possible execute overdue timer events, that
4751 could not take place because the system was powered off.
4752 This enables simple anacron-like behaviour for timer units.
4754 * systemctl's "list-timers" will now also list the time a
4755 timer unit was last triggered in addition to the next time
4756 it will be triggered.
4758 * systemd-networkd will now assign predictable IPv4LL
4759 addresses to its local interfaces.
4761 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4762 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4763 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4764 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4765 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4766 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4767 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4768 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4771 — Berlin, 2014-03-25
4775 * A new unit file setting RestrictAddressFamilies= has been
4776 added to restrict which socket address families unit
4777 processes gain access to. This takes address family names
4778 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4779 attack surface of services via exotic protocol stacks. This
4780 is built on seccomp system call filters.
4782 * Two new unit file settings RuntimeDirectory= and
4783 RuntimeDirectoryMode= have been added that may be used to
4784 manage a per-daemon runtime directories below /run. This is
4785 an alternative for setting up directory permissions with
4786 tmpfiles snippets, and has the advantage that the runtime
4787 directory's lifetime is bound to the daemon runtime and that
4788 the daemon starts up with an empty directory each time. This
4789 is particularly useful when writing services that drop
4790 privileges using the User= or Group= setting.
4792 * The DeviceAllow= unit setting now supports globbing for
4793 matching against device group names.
4795 * The systemd configuration file system.conf gained new
4796 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4797 DefaultMemoryAccounting= to globally turn on/off accounting
4798 for specific resources (cgroups) for all units. These
4799 settings may still be overridden individually in each unit
4802 * systemd-gpt-auto-generator is now able to discover /srv and
4803 root partitions in addition to /home and swap partitions. It
4804 also supports LUKS-encrypted partitions now. With this in
4805 place, automatic discovery of partitions to mount following
4806 the Discoverable Partitions Specification
4807 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4808 is now a lot more complete. This allows booting without
4809 /etc/fstab and without root= on the kernel command line on
4810 systems prepared appropriately.
4812 * systemd-nspawn gained a new --image= switch which allows
4813 booting up disk images and Linux installations on any block
4814 device that follow the Discoverable Partitions Specification
4815 (see above). This means that installations made with
4816 appropriately updated installers may now be started and
4817 deployed using container managers, completely
4818 unmodified. (We hope that libvirt-lxc will add support for
4819 this feature soon, too.)
4821 * systemd-nspawn gained a new --network-macvlan= setting to
4822 set up a private macvlan interface for the
4823 container. Similarly, systemd-networkd gained a new
4824 Kind=macvlan setting in .netdev files.
4826 * systemd-networkd now supports configuring local addresses
4829 * A new tool systemd-network-wait-online has been added to
4830 synchronously wait for network connectivity using
4833 * The sd-bus.h bus API gained a new sd_bus_track object for
4834 tracking the life-cycle of bus peers. Note that sd-bus.h is
4835 still not a public API though (unless you specify
4836 --enable-kdbus on the configure command line, which however
4837 voids your warranty and you get no API stability guarantee).
4839 * The $XDG_RUNTIME_DIR runtime directories for each user are
4840 now individual tmpfs instances, which has the benefit of
4841 introducing separate pools for each user, with individual
4842 size limits, and thus making sure that unprivileged clients
4843 can no longer negatively impact the system or other users by
4844 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4845 RuntimeDirectorySize= has been introduced that allows
4846 controlling the default size limit for all users. It
4847 defaults to 10% of the available physical memory. This is no
4848 replacement for quotas on tmpfs though (which the kernel
4849 still does not support), as /dev/shm and /tmp are still
4850 shared resources used by both the system and unprivileged
4853 * logind will now automatically turn off automatic suspending
4854 on laptop lid close when more than one display is
4855 connected. This was previously expected to be implemented
4856 individually in desktop environments (such as GNOME),
4857 however has been added to logind now, in order to fix a
4858 boot-time race where a desktop environment might not have
4859 been started yet and thus not been able to take an inhibitor
4860 lock at the time where logind already suspends the system
4861 due to a closed lid.
4863 * logind will now wait at least 30s after each system
4864 suspend/resume cycle, and 3min after system boot before
4865 suspending the system due to a closed laptop lid. This
4866 should give USB docking stations and similar enough time to
4867 be probed and configured after system resume and boot in
4868 order to then act as suspend blocker.
4870 * systemd-run gained a new --property= setting which allows
4871 initialization of resource control properties (and others)
4872 for the created scope or service unit. Example: "systemd-run
4873 --property=BlockIOWeight=10 updatedb" may be used to run
4874 updatedb at a low block IO scheduling weight.
4876 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4877 now also work in --scope mode.
4879 * When systemd is compiled with kdbus support, basic support
4880 for enforced policies is now in place. (Note that enabling
4881 kdbus still voids your warranty and no API compatibility
4884 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4885 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4886 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4887 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4888 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4889 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4890 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4891 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4892 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4893 Zbigniew Jędrzejewski-Szmek
4895 — Berlin, 2014-03-12
4899 * systemd will now relabel /dev after loading the SMACK policy
4900 according to SMACK rules.
4902 * A new unit file option AppArmorProfile= has been added to
4903 set the AppArmor profile for the processes of a unit.
4905 * A new condition check ConditionArchitecture= has been added
4906 to conditionalize units based on the system architecture, as
4907 reported by uname()'s "machine" field.
4909 * systemd-networkd now supports matching on the system
4910 virtualization, architecture, kernel command line, host name
4913 * logind is now a lot more aggressive when suspending the
4914 machine due to a closed laptop lid. Instead of acting only
4915 on the lid close action, it will continuously watch the lid
4916 status and act on it. This is useful for laptops where the
4917 power button is on the outside of the chassis so that it can
4918 be reached without opening the lid (such as the Lenovo
4919 Yoga). On those machines, logind will now immediately
4920 re-suspend the machine if the power button has been
4921 accidentally pressed while the laptop was suspended and in a
4922 backpack or similar.
4924 * logind will now watch SW_DOCK switches and inhibit reaction
4925 to the lid switch if it is pressed. This means that logind
4926 will not suspend the machine anymore if the lid is closed
4927 and the system is docked, if the laptop supports SW_DOCK
4928 notifications via the input layer. Note that ACPI docking
4929 stations do not generate this currently. Also note that this
4930 logic is usually not fully sufficient and Desktop
4931 Environments should take a lid switch inhibitor lock when an
4932 external display is connected, as systemd will not watch
4935 * nspawn will now make use of the devices cgroup controller by
4936 default, and only permit creation of and access to the usual
4937 API device nodes like /dev/null or /dev/random, as well as
4938 access to (but not creation of) the pty devices.
4940 * We will now ship a default .network file for
4941 systemd-networkd that automatically configures DHCP for
4942 network interfaces created by nspawn's --network-veth or
4943 --network-bridge= switches.
4945 * systemd will now understand the usual M, K, G, T suffixes
4946 according to SI conventions (i.e. to the base 1000) when
4947 referring to throughput and hardware metrics. It will stay
4948 with IEC conventions (i.e. to the base 1024) for software
4949 metrics, according to what is customary according to
4950 Wikipedia. We explicitly document which base applies for
4951 each configuration option.
4953 * The DeviceAllow= setting in unit files now supports a syntax
4954 to whitelist an entire group of devices node majors at once,
4955 based on the /proc/devices listing. For example, with the
4956 string "char-pts", it is now possible to whitelist all
4957 current and future pseudo-TTYs at once.
4959 * sd-event learned a new "post" event source. Event sources of
4960 this type are triggered by the dispatching of any event
4961 source of a type that is not "post". This is useful for
4962 implementing clean-up and check event sources that are
4963 triggered by other work being done in the program.
4965 * systemd-networkd is no longer statically enabled, but uses
4966 the usual [Install] sections so that it can be
4967 enabled/disabled using systemctl. It still is enabled by
4970 * When creating a veth interface pair with systemd-nspawn, the
4971 host side will now be prefixed with "vb-" if
4972 --network-bridge= is used, and with "ve-" if --network-veth
4973 is used. This way, it is easy to distinguish these cases on
4974 the host, for example to apply different configuration to
4975 them with systemd-networkd.
4977 * The compatibility libraries for libsystemd-journal.so,
4978 libsystem-id128.so, libsystemd-login.so and
4979 libsystemd-daemon.so do not make use of IFUNC
4980 anymore. Instead, we now build libsystemd.so multiple times
4981 under these alternative names. This means that the footprint
4982 is drastically increased, but given that these are
4983 transitional compatibility libraries, this should not matter
4984 much. This change has been made necessary to support the ARM
4985 platform for these compatibility libraries, as the ARM
4986 toolchain is not really at the same level as the toolchain
4987 for other architectures like x86 and does not support
4988 IFUNC. Please make sure to use --enable-compat-libs only
4989 during a transitional period!
4991 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4992 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4993 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4994 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4995 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4996 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4997 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4998 Zbigniew Jędrzejewski-Szmek
5000 — Berlin, 2014-02-24
5004 * A new component "systemd-networkd" has been added that can
5005 be used to configure local network interfaces statically or
5006 via DHCP. It is capable of bringing up bridges, VLANs, and
5007 bonding. Currently, no hook-ups for interactive network
5008 configuration are provided. Use this for your initrd,
5009 container, embedded, or server setup if you need a simple,
5010 yet powerful, network configuration solution. This
5011 configuration subsystem is quite nifty, as it allows wildcard
5012 hotplug matching in interfaces. For example, with a single
5013 configuration snippet, you can configure that all Ethernet
5014 interfaces showing up are automatically added to a bridge,
5015 or similar. It supports link-sensing and more.
5017 * A new tool "systemd-socket-proxyd" has been added which can
5018 act as a bidirectional proxy for TCP sockets. This is
5019 useful for adding socket activation support to services that
5020 do not actually support socket activation, including virtual
5021 machines and the like.
5023 * Add a new tool to save/restore rfkill state on
5026 * Save/restore state of keyboard backlights in addition to
5027 display backlights on shutdown/boot.
5029 * udev learned a new SECLABEL{} construct to label device
5030 nodes with a specific security label when they appear. For
5031 now, only SECLABEL{selinux} is supported, but the syntax is
5032 prepared for additional security frameworks.
5034 * udev gained a new scheme to configure link-level attributes
5035 from files in /etc/systemd/network/*.link. These files can
5036 match against MAC address, device path, driver name and type,
5037 and will apply attributes like the naming policy, link speed,
5038 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5039 address assignment policy (randomized, ...).
5041 * The configuration of network interface naming rules for
5042 "permanent interface names" has changed: a new NamePolicy=
5043 setting in the [Link] section of .link files determines the
5044 priority of possible naming schemes (onboard, slot, MAC,
5045 path). The default value of this setting is determined by
5046 /usr/lib/net/links/99-default.link. Old
5047 80-net-name-slot.rules udev configuration file has been
5048 removed, so local configuration overriding this file should
5049 be adapted to override 99-default.link instead.
5051 * When the User= switch is used in a unit file, also
5052 initialize $SHELL= based on the user database entry.
5054 * systemd no longer depends on libdbus. All communication is
5055 now done with sd-bus, systemd's low-level bus library
5058 * kdbus support has been added to PID 1 itself. When kdbus is
5059 enabled, this causes PID 1 to set up the system bus and
5060 enable support for a new ".busname" unit type that
5061 encapsulates bus name activation on kdbus. It works a little
5062 bit like ".socket" units, except for bus names. A new
5063 generator has been added that converts classic dbus1 service
5064 activation files automatically into native systemd .busname
5067 * sd-bus: add a light-weight vtable implementation that allows
5068 defining objects on the bus with a simple static const
5069 vtable array of its methods, signals and properties.
5071 * systemd will not generate or install static dbus
5072 introspection data anymore to /usr/share/dbus-1/interfaces,
5073 as the precise format of these files is unclear, and
5074 nothing makes use of it.
5076 * A proxy daemon is now provided to proxy clients connecting
5077 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5078 compatibility with classic D-Bus.
5080 * A bus driver implementation has been added that supports the
5081 classic D-Bus bus driver calls on kdbus, also for
5082 compatibility purposes.
5084 * A new API "sd-event.h" has been added that implements a
5085 minimal event loop API built around epoll. It provides a
5086 couple of features that direct epoll usage is lacking:
5087 prioritization of events, scales to large numbers of timer
5088 events, per-event timer slack (accuracy), system-wide
5089 coalescing of timer events, exit handlers, watchdog
5090 supervision support using systemd's sd_notify() API, child
5093 * A new API "sd-rntl.h" has been added that provides an API
5094 around the route netlink interface of the kernel, similar in
5095 style to "sd-bus.h".
5097 * A new API "sd-dhcp-client.h" has been added that provides a
5098 small DHCPv4 client-side implementation. This is used by
5101 * There is a new kernel command line option
5102 "systemd.restore_state=0|1". When set to "0", none of the
5103 systemd tools will restore saved runtime state to hardware
5104 devices. More specifically, the rfkill and backlight states
5107 * The FsckPassNo= compatibility option in mount/service units
5108 has been removed. The fstab generator will now add the
5109 necessary dependencies automatically, and does not require
5110 PID1's support for that anymore.
5112 * journalctl gained a new switch, --list-boots, that lists
5113 recent boots with their times and boot IDs.
5115 * The various tools like systemctl, loginctl, timedatectl,
5116 busctl, systemd-run, ... have gained a new switch "-M" to
5117 connect to a specific, local OS container (as direct
5118 connection, without requiring SSH). This works on any
5119 container that is registered with machined, such as those
5120 created by libvirt-lxc or nspawn.
5122 * systemd-run and systemd-analyze also gained support for "-H"
5123 to connect to remote hosts via SSH. This is particularly
5124 useful for systemd-run because it enables queuing of jobs
5125 onto remote systems.
5127 * machinectl gained a new command "login" to open a getty
5128 login in any local container. This works with any container
5129 that is registered with machined (such as those created by
5130 libvirt-lxc or nspawn), and which runs systemd inside.
5132 * machinectl gained a new "reboot" command that may be used to
5133 trigger a reboot on a specific container that is registered
5134 with machined. This works on any container that runs an init
5135 system of some kind.
5137 * systemctl gained a new "list-timers" command to print a nice
5138 listing of installed timer units with the times they elapse
5141 * Alternative reboot() parameters may now be specified on the
5142 "systemctl reboot" command line and are passed to the
5143 reboot() system call.
5145 * systemctl gained a new --job-mode= switch to configure the
5146 mode to queue a job with. This is a more generic version of
5147 --fail, --irreversible, and --ignore-dependencies, which are
5148 still available but not advertised anymore.
5150 * /etc/systemd/system.conf gained new settings to configure
5151 various default timeouts of units, as well as the default
5152 start limit interval and burst. These may still be overridden
5155 * PID1 will now export on the bus profile data of the security
5156 policy upload process (such as the SELinux policy upload to
5159 * journald: when forwarding logs to the console, include
5160 timestamps (following the setting in
5161 /sys/module/printk/parameters/time).
5163 * OnCalendar= in timer units now understands the special
5164 strings "yearly" and "annually". (Both are equivalent)
5166 * The accuracy of timer units is now configurable with the new
5167 AccuracySec= setting. It defaults to 1min.
5169 * A new dependency type JoinsNamespaceOf= has been added that
5170 allows running two services within the same /tmp and network
5171 namespace, if PrivateNetwork= or PrivateTmp= are used.
5173 * A new command "cat" has been added to systemctl. It outputs
5174 the original unit file of a unit, and concatenates the
5175 contents of additional "drop-in" unit file snippets, so that
5176 the full configuration is shown.
5178 * systemctl now supports globbing on the various "list-xyz"
5179 commands, like "list-units" or "list-sockets", as well as on
5180 those commands which take multiple unit names.
5182 * journalctl's --unit= switch gained support for globbing.
5184 * All systemd daemons now make use of the watchdog logic so
5185 that systemd automatically notices when they hang.
5187 * If the $container_ttys environment variable is set,
5188 getty-generator will automatically spawn a getty for each
5189 listed tty. This is useful for container managers to request
5190 login gettys to be spawned on as many ttys as needed.
5192 * %h, %s, %U specifier support is not available anymore when
5193 used in unit files for PID 1. This is because NSS calls are
5194 not safe from PID 1. They stay available for --user
5195 instances of systemd, and as special case for the root user.
5197 * loginctl gained a new "--no-legend" switch to turn off output
5200 * The "sd-login.h" API gained three new calls:
5201 sd_session_is_remote(), sd_session_get_remote_user(),
5202 sd_session_get_remote_host() to query information about
5205 * The udev hardware database now also carries vendor/product
5206 information of SDIO devices.
5208 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5209 determine whether watchdog notifications are requested by
5212 * Socket-activated per-connection services now include a
5213 short description of the connection parameters in the
5216 * tmpfiles gained a new "--boot" option. When this is not used,
5217 only lines where the command character is not suffixed with
5218 "!" are executed. When this option is specified, those
5219 options are executed too. This partitions tmpfiles
5220 directives into those that can be safely executed at any
5221 time, and those which should be run only at boot (for
5222 example, a line that creates /run/nologin).
5224 * A new API "sd-resolve.h" has been added which provides a simple
5225 asynchronous wrapper around glibc NSS host name resolution
5226 calls, such as getaddrinfo(). In contrast to glibc's
5227 getaddrinfo_a(), it does not use signals. In contrast to most
5228 other asynchronous name resolution libraries, this one does
5229 not reimplement DNS, but reuses NSS, so that alternate
5230 host name resolution systems continue to work, such as mDNS,
5231 LDAP, etc. This API is based on libasyncns, but it has been
5232 cleaned up for inclusion in systemd.
5234 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5235 "sd-daemon.h" are no longer found in individual libraries
5236 libsystemd-journal.so, libsystemd-login.so,
5237 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5238 merged them into a single library, libsystemd.so, which
5239 provides all symbols. The reason for this is cyclic
5240 dependencies, as these libraries tend to use each other's
5241 symbols. So far, we have managed to workaround that by linking
5242 a copy of a good part of our code into each of these
5243 libraries again and again, which, however, makes certain
5244 things hard to do, like sharing static variables. Also, it
5245 substantially increases footprint. With this change, there
5246 is only one library for the basic APIs systemd
5247 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5248 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5249 library as well, however are subject to the --enable-kdbus
5250 switch (see below). Note that "sd-dhcp-client.h" is not part
5251 of this library (this is because it only consumes, never
5252 provides, services of/to other APIs). To make the transition
5253 easy from the separate libraries to the unified one, we
5254 provide the --enable-compat-libs compile-time switch which
5255 will generate stub libraries that are compatible with the
5256 old ones but redirect all calls to the new one.
5258 * All of the kdbus logic and the new APIs "sd-bus.h",
5259 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5260 and "sd-utf8.h" are compile-time optional via the
5261 "--enable-kdbus" switch, and they are not compiled in by
5262 default. To make use of kdbus, you have to explicitly enable
5263 the switch. Note however, that neither the kernel nor the
5264 userspace API for all of this is considered stable yet. We
5265 want to maintain the freedom to still change the APIs for
5266 now. By specifying this build-time switch, you acknowledge
5267 that you are aware of the instability of the current
5270 * Also, note that while kdbus is pretty much complete,
5271 it lacks one thing: proper policy support. This means you
5272 can build a fully working system with all features; however,
5273 it will be highly insecure. Policy support will be added in
5274 one of the next releases, at the same time that we will
5275 declare the APIs stable.
5277 * When the kernel command line argument "kdbus" is specified,
5278 systemd will automatically load the kdbus.ko kernel module. At
5279 this stage of development, it is only useful for testing kdbus
5280 and should not be used in production. Note: if "--enable-kdbus"
5281 is specified, and the kdbus.ko kernel module is available, and
5282 "kdbus" is added to the kernel command line, the entire system
5283 runs with kdbus instead of dbus-daemon, with the above mentioned
5284 problem of missing the system policy enforcement. Also a future
5285 version of kdbus.ko or a newer systemd will not be compatible with
5286 each other, and will unlikely be able to boot the machine if only
5287 one of them is updated.
5289 * systemctl gained a new "import-environment" command which
5290 uploads the caller's environment (or parts thereof) into the
5291 service manager so that it is inherited by services started
5292 by the manager. This is useful to upload variables like
5293 $DISPLAY into the user service manager.
5295 * A new PrivateDevices= switch has been added to service units
5296 which allows running a service with a namespaced /dev
5297 directory that does not contain any device nodes for
5298 physical devices. More specifically, it only includes devices
5299 such as /dev/null, /dev/urandom, and /dev/zero which are API
5302 * logind has been extended to support behaviour like VT
5303 switching on seats that do not support a VT. This makes
5304 multi-session available on seats that are not the first seat
5305 (seat0), and on systems where kernel support for VTs has
5306 been disabled at compile-time.
5308 * If a process holds a delay lock for system sleep or shutdown
5309 and fails to release it in time, we will now log its
5310 identity. This makes it easier to identify processes that
5311 cause slow suspends or power-offs.
5313 * When parsing /etc/crypttab, support for a new key-slot=
5314 option as supported by Debian is added. It allows indicating
5315 which LUKS slot to use on disk, speeding up key loading.
5317 * The sd_journald_sendv() API call has been checked and
5318 officially declared to be async-signal-safe so that it may
5319 be invoked from signal handlers for logging purposes.
5321 * Boot-time status output is now enabled automatically after a
5322 short timeout if boot does not progress, in order to give
5323 the user an indication what she or he is waiting for.
5325 * The boot-time output has been improved to show how much time
5326 remains until jobs expire.
5328 * The KillMode= switch in service units gained a new possible
5329 value "mixed". If set, and the unit is shut down, then the
5330 initial SIGTERM signal is sent only to the main daemon
5331 process, while the following SIGKILL signal is sent to
5332 all remaining processes of the service.
5334 * When a scope unit is registered, a new property "Controller"
5335 may be set. If set to a valid bus name, systemd will send a
5336 RequestStop() signal to this name when it would like to shut
5337 down the scope. This may be used to hook manager logic into
5338 the shutdown logic of scope units. Also, scope units may now
5339 be put in a special "abandoned" state, in which case the
5340 manager process which created them takes no further
5341 responsibilities for it.
5343 * When reading unit files, systemd will now verify
5344 the access mode of these files, and warn about certain
5345 suspicious combinations. This has been added to make it
5346 easier to track down packaging bugs where unit files are
5347 marked executable or world-writable.
5349 * systemd-nspawn gained a new "--setenv=" switch to set
5350 container-wide environment variables. The similar option in
5351 systemd-activate was renamed from "--environment=" to
5352 "--setenv=" for consistency.
5354 * systemd-nspawn has been updated to create a new kdbus domain
5355 for each container that is invoked, thus allowing each
5356 container to have its own set of system and user buses,
5357 independent of the host.
5359 * systemd-nspawn gained a new --drop-capability= switch to run
5360 the container with less capabilities than the default. Both
5361 --drop-capability= and --capability= now take the special
5362 string "all" for dropping or keeping all capabilities.
5364 * systemd-nspawn gained new switches for executing containers
5365 with specific SELinux labels set.
5367 * systemd-nspawn gained a new --quiet switch to not generate
5368 any additional output but the container's own console
5371 * systemd-nspawn gained a new --share-system switch to run a
5372 container without PID namespacing enabled.
5374 * systemd-nspawn gained a new --register= switch to control
5375 whether the container is registered with systemd-machined or
5376 not. This is useful for containers that do not run full
5377 OS images, but only specific apps.
5379 * systemd-nspawn gained a new --keep-unit which may be used
5380 when invoked as the only program from a service unit, and
5381 results in registration of the unit service itself in
5382 systemd-machined, instead of a newly opened scope unit.
5384 * systemd-nspawn gained a new --network-interface= switch for
5385 moving arbitrary interfaces to the container. The new
5386 --network-veth switch creates a virtual Ethernet connection
5387 between host and container. The new --network-bridge=
5388 switch then allows assigning the host side of this virtual
5389 Ethernet connection to a bridge device.
5391 * systemd-nspawn gained a new --personality= switch for
5392 setting the kernel personality for the container. This is
5393 useful when running a 32-bit container on a 64-bit host. A
5394 similar option Personality= is now also available for service
5397 * logind will now also track a "Desktop" identifier for each
5398 session which encodes the desktop environment of it. This is
5399 useful for desktop environments that want to identify
5400 multiple running sessions of itself easily.
5402 * A new SELinuxContext= setting for service units has been
5403 added that allows setting a specific SELinux execution
5404 context for a service.
5406 * Most systemd client tools will now honour $SYSTEMD_LESS for
5407 settings of the "less" pager. By default, these tools will
5408 override $LESS to allow certain operations to work, such as
5409 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5410 influence this logic.
5412 * systemd's "seccomp" hook-up has been changed to make use of
5413 the libseccomp library instead of using its own
5414 implementation. This has benefits for portability among
5417 * For usage together with SystemCallFilter=, a new
5418 SystemCallErrorNumber= setting has been introduced that
5419 allows configuration of a system error number to be returned
5420 on filtered system calls, instead of immediately killing the
5421 process. Also, SystemCallArchitectures= has been added to
5422 limit access to system calls of a particular architecture
5423 (in order to turn off support for unused secondary
5424 architectures). There is also a global
5425 SystemCallArchitectures= setting in system.conf now to turn
5426 off support for non-native system calls system-wide.
5428 * systemd requires a kernel with a working name_to_handle_at(),
5429 please see the kernel config requirements in the README file.
5431 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5432 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5433 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5434 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5435 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5436 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5437 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5438 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5439 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5440 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5441 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5442 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5443 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5444 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5445 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5446 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5447 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5448 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5449 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5450 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5451 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5452 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5453 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5454 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5456 — Berlin, 2014-02-20
5460 * logind has gained support for facilitating privileged input
5461 and drm device access for unprivileged clients. This work is
5462 useful to allow Wayland display servers (and similar
5463 programs, such as kmscon) to run under the user's ID and
5464 access input and drm devices which are normally
5465 protected. When this is used (and the kernel is new enough)
5466 logind will "mute" IO on the file descriptors passed to
5467 Wayland as long as it is in the background and "unmute" it
5468 if it returns into the foreground. This allows secure
5469 session switching without allowing background sessions to
5470 eavesdrop on input and display data. This also introduces
5471 session switching support if VT support is turned off in the
5472 kernel, and on seats that are not seat0.
5474 * A new kernel command line option luks.options= is understood
5475 now which allows specifying LUKS options for usage for LUKS
5476 encrypted partitions specified with luks.uuid=.
5478 * tmpfiles.d(5) snippets may now use specifier expansion in
5479 path names. More specifically %m, %b, %H, %v, are now
5480 replaced by the local machine id, boot id, hostname, and
5481 kernel version number.
5483 * A new tmpfiles.d(5) command "m" has been introduced which
5484 may be used to change the owner/group/access mode of a file
5485 or directory if it exists, but do nothing if it does not.
5487 * This release removes high-level support for the
5488 MemorySoftLimit= cgroup setting. The underlying kernel
5489 cgroup attribute memory.soft_limit= is currently badly
5490 designed and likely to be removed from the kernel API in its
5491 current form, hence we should not expose it for now.
5493 * The memory.use_hierarchy cgroup attribute is now enabled for
5494 all cgroups systemd creates in the memory cgroup
5495 hierarchy. This option is likely to be come the built-in
5496 default in the kernel anyway, and the non-hierarchical mode
5497 never made much sense in the intrinsically hierarchical
5500 * A new field _SYSTEMD_SLICE= is logged along with all journal
5501 messages containing the slice a message was generated
5502 from. This is useful to allow easy per-customer filtering of
5503 logs among other things.
5505 * systemd-journald will no longer adjust the group of journal
5506 files it creates to the "systemd-journal" group. Instead we
5507 rely on the journal directory to be owned by the
5508 "systemd-journal" group, and its setgid bit set, so that the
5509 kernel file system layer will automatically enforce that
5510 journal files inherit this group assignment. The reason for
5511 this change is that we cannot allow NSS look-ups from
5512 journald which would be necessary to resolve
5513 "systemd-journal" to a numeric GID, because this might
5514 create deadlocks if NSS involves synchronous queries to
5515 other daemons (such as nscd, or sssd) which in turn are
5516 logging clients of journald and might block on it, which
5517 would then dead lock. A tmpfiles.d(5) snippet included in
5518 systemd will make sure the setgid bit and group are
5519 properly set on the journal directory if it exists on every
5520 boot. However, we recommend adjusting it manually after
5521 upgrades too (or from RPM scriptlets), so that the change is
5522 not delayed until next reboot.
5524 * Backlight and random seed files in /var/lib/ have moved into
5525 the /var/lib/systemd/ directory, in order to centralize all
5526 systemd generated files in one directory.
5528 * Boot time performance measurements (as displayed by
5529 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5530 performance information if that's available to determine how
5531 much time BIOS and boot loader initialization required. With
5532 a sufficiently new BIOS you hence no longer need to boot
5533 with Gummiboot to get access to such information.
5535 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5536 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5537 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5538 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5539 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5540 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5541 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5543 — Berlin, 2013-10-02
5547 * The Restart= option for services now understands a new
5548 on-watchdog setting, which will restart the service
5549 automatically if the service stops sending out watchdog keep
5550 alive messages (as configured with WatchdogSec=).
5552 * The getty generator (which is responsible for bringing up a
5553 getty on configured serial consoles) will no longer only
5554 start a getty on the primary kernel console but on all
5555 others, too. This makes the order in which console= is
5556 specified on the kernel command line less important.
5558 * libsystemd-logind gained a new sd_session_get_vt() call to
5559 retrieve the VT number of a session.
5561 * If the option "tries=0" is set for an entry of /etc/crypttab
5562 its passphrase is queried indefinitely instead of any
5563 maximum number of tries.
5565 * If a service with a configure PID file terminates its PID
5566 file will now be removed automatically if it still exists
5567 afterwards. This should put an end to stale PID files.
5569 * systemd-run will now also take relative binary path names
5570 for execution and no longer insists on absolute paths.
5572 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5573 paths that are optionally prefixed with "-" to indicate that
5574 it should not be considered a failure if they do not exist.
5576 * journalctl -o (and similar commands) now understands a new
5577 output mode "short-precise", it is similar to "short" but
5578 shows timestamps with usec accuracy.
5580 * The option "discard" (as known from Debian) is now
5581 synonymous to "allow-discards" in /etc/crypttab. In fact,
5582 "discard" is preferred now (since it is easier to remember
5585 * Some licensing clean-ups were made, so that more code is now
5586 LGPL-2.1 licensed than before.
5588 * A minimal tool to save/restore the display backlight
5589 brightness across reboots has been added. It will store the
5590 backlight setting as late as possible at shutdown, and
5591 restore it as early as possible during reboot.
5593 * A logic to automatically discover and enable home and swap
5594 partitions on GPT disks has been added. With this in place
5595 /etc/fstab becomes optional for many setups as systemd can
5596 discover certain partitions located on the root disk
5597 automatically. Home partitions are recognized under their
5598 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5599 partitions are recognized under their GPT type ID
5600 0657fd6da4ab43c484e50933c84b4f4f.
5602 * systemd will no longer pass any environment from the kernel
5603 or initrd to system services. If you want to set an
5604 environment for all services, do so via the kernel command
5605 line systemd.setenv= assignment.
5607 * The systemd-sysctl tool no longer natively reads the file
5608 /etc/sysctl.conf. If desired, the file should be symlinked
5609 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5610 legacy support by a symlink rather than built-in code, it
5611 also makes the otherwise hidden order of application of the
5612 different files visible. (Note that this partly reverts to a
5613 pre-198 application order of sysctl knobs!)
5615 * The "systemctl set-log-level" and "systemctl dump" commands
5616 have been moved to systemd-analyze.
5618 * systemd-run learned the new --remain-after-exit switch,
5619 which causes the scope unit not to be cleaned up
5620 automatically after the process terminated.
5622 * tmpfiles learned a new --exclude-prefix= switch to exclude
5623 certain paths from operation.
5625 * journald will now automatically flush all messages to disk
5626 as soon as a message at the log level CRIT, ALERT or EMERG
5629 Contributions from: Andrew Cook, Brandon Philips, Christian
5630 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5631 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5632 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5633 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5634 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5635 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5636 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5637 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5638 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5639 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5640 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5641 William Giokas, Zbigniew Jędrzejewski-Szmek
5643 — Berlin, 2013-09-13
5647 * The documentation has been updated to cover the various new
5648 concepts introduced with 205.
5650 * Unit files now understand the new %v specifier which
5651 resolves to the kernel version string as returned by "uname
5654 * systemctl now supports filtering the unit list output by
5655 load state, active state and sub state, using the new
5658 * "systemctl status" will now show the results of the
5659 condition checks (like ConditionPathExists= and similar) of
5660 the last start attempts of the unit. They are also logged to
5663 * "journalctl -b" may now be used to look for boot output of a
5664 specific boot. Try "journalctl -b -1" for the previous boot,
5665 but the syntax is substantially more powerful.
5667 * "journalctl --show-cursor" has been added which prints the
5668 cursor string the last shown log line. This may then be used
5669 with the new "journalctl --after-cursor=" switch to continue
5670 browsing logs from that point on.
5672 * "journalctl --force" may now be used to force regeneration
5675 * Creation of "dead" device nodes has been moved from udev
5676 into kmod and tmpfiles. Previously, udev would read the kmod
5677 databases to pre-generate dead device nodes based on meta
5678 information contained in kernel modules, so that these would
5679 be auto-loaded on access rather then at boot. As this
5680 does not really have much to do with the exposing actual
5681 kernel devices to userspace this has always been slightly
5682 alien in the udev codebase. Following the new scheme kmod
5683 will now generate a runtime snippet for tmpfiles from the
5684 module meta information and it now is tmpfiles' job to the
5685 create the nodes. This also allows overriding access and
5686 other parameters for the nodes using the usual tmpfiles
5687 facilities. As side effect this allows us to remove the
5688 CAP_SYS_MKNOD capability bit from udevd entirely.
5690 * logind's device ACLs may now be applied to these "dead"
5691 devices nodes too, thus finally allowing managed access to
5692 devices such as /dev/snd/sequencer without loading the
5693 backing module right-away.
5695 * A new RPM macro has been added that may be used to apply
5696 tmpfiles configuration during package installation.
5698 * systemd-detect-virt and ConditionVirtualization= now can
5699 detect User-Mode-Linux machines (UML).
5701 * journald will now implicitly log the effective capabilities
5702 set of processes in the message metadata.
5704 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5706 * The initrd interface has been simplified (more specifically,
5707 support for passing performance data via environment
5708 variables and fsck results via files in /run has been
5709 removed). These features were non-essential, and are
5710 nowadays available in a much nicer way by having systemd in
5711 the initrd serialize its state and have the hosts systemd
5712 deserialize it again.
5714 * The udev "keymap" data files and tools to apply keyboard
5715 specific mappings of scan to key codes, and force-release
5716 scan code lists have been entirely replaced by a udev
5717 "keyboard" builtin and a hwdb data file.
5719 * systemd will now honour the kernel's "quiet" command line
5720 argument also during late shutdown, resulting in a
5721 completely silent shutdown when used.
5723 * There's now an option to control the SO_REUSEPORT socket
5724 option in .socket units.
5726 * Instance units will now automatically get a per-template
5727 subslice of system.slice unless something else is explicitly
5728 configured. For example, instances of sshd@.service will now
5729 implicitly be placed in system-sshd.slice rather than
5730 system.slice as before.
5732 * Test coverage support may now be enabled at build time.
5734 Contributions from: Dave Reisner, Frederic Crozat, Harald
5735 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5736 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5737 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5738 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5739 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5740 Giokas, Zbigniew Jędrzejewski-Szmek
5742 — Berlin, 2013-07-23
5746 * Two new unit types have been introduced:
5748 Scope units are very similar to service units, however, are
5749 created out of pre-existing processes — instead of PID 1
5750 forking off the processes. By using scope units it is
5751 possible for system services and applications to group their
5752 own child processes (worker processes) in a powerful way
5753 which then maybe used to organize them, or kill them
5754 together, or apply resource limits on them.
5756 Slice units may be used to partition system resources in an
5757 hierarchical fashion and then assign other units to them. By
5758 default there are now three slices: system.slice (for all
5759 system services), user.slice (for all user sessions),
5760 machine.slice (for VMs and containers).
5762 Slices and scopes have been introduced primarily in
5763 context of the work to move cgroup handling to a
5764 single-writer scheme, where only PID 1
5765 creates/removes/manages cgroups.
5767 * There's a new concept of "transient" units. In contrast to
5768 normal units these units are created via an API at runtime,
5769 not from configuration from disk. More specifically this
5770 means it is now possible to run arbitrary programs as
5771 independent services, with all execution parameters passed
5772 in via bus APIs rather than read from disk. Transient units
5773 make systemd substantially more dynamic then it ever was,
5774 and useful as a general batch manager.
5776 * logind has been updated to make use of scope and slice units
5777 for managing user sessions. As a user logs in he will get
5778 his own private slice unit, to which all sessions are added
5779 as scope units. We also added support for automatically
5780 adding an instance of user@.service for the user into the
5781 slice. Effectively logind will no longer create cgroup
5782 hierarchies on its own now, it will defer entirely to PID 1
5783 for this by means of scope, service and slice units. Since
5784 user sessions this way become entities managed by PID 1
5785 the output of "systemctl" is now a lot more comprehensive.
5787 * A new mini-daemon "systemd-machined" has been added which
5788 may be used by virtualization managers to register local
5789 VMs/containers. nspawn has been updated accordingly, and
5790 libvirt will be updated shortly. machined will collect a bit
5791 of meta information about the VMs/containers, and assign
5792 them their own scope unit (see above). The collected
5793 meta-data is then made available via the "machinectl" tool,
5794 and exposed in "ps" and similar tools. machined/machinectl
5795 is compile-time optional.
5797 * As discussed earlier, the low-level cgroup configuration
5798 options ControlGroup=, ControlGroupModify=,
5799 ControlGroupPersistent=, ControlGroupAttribute= have been
5800 removed. Please use high-level attribute settings instead as
5801 well as slice units.
5803 * A new bus call SetUnitProperties() has been added to alter
5804 various runtime parameters of a unit. This is primarily
5805 useful to alter cgroup parameters dynamically in a nice way,
5806 but will be extended later on to make more properties
5807 modifiable at runtime. systemctl gained a new set-properties
5808 command that wraps this call.
5810 * A new tool "systemd-run" has been added which can be used to
5811 run arbitrary command lines as transient services or scopes,
5812 while configuring a number of settings via the command
5813 line. This tool is currently very basic, however already
5814 very useful. We plan to extend this tool to even allow
5815 queuing of execution jobs with time triggers from the
5816 command line, similar in fashion to "at".
5818 * nspawn will now inform the user explicitly that kernels with
5819 audit enabled break containers, and suggest the user to turn
5822 * Support for detecting the IMA and AppArmor security
5823 frameworks with ConditionSecurity= has been added.
5825 * journalctl gained a new "-k" switch for showing only kernel
5826 messages, mimicking dmesg output; in addition to "--user"
5827 and "--system" switches for showing only user's own logs
5830 * systemd-delta can now show information about drop-in
5831 snippets extending unit files.
5833 * libsystemd-bus has been substantially updated but is still
5834 not available as public API.
5836 * systemd will now look for the "debug" argument on the kernel
5837 command line and enable debug logging, similar to what
5838 "systemd.log_level=debug" already did before.
5840 * "systemctl set-default", "systemctl get-default" has been
5841 added to configure the default.target symlink, which
5842 controls what to boot into by default.
5844 * "systemctl set-log-level" has been added as a convenient
5845 way to raise and lower systemd logging threshold.
5847 * "systemd-analyze plot" will now show the time the various
5848 generators needed for execution, as well as information
5849 about the unit file loading.
5851 * libsystemd-journal gained a new sd_journal_open_files() call
5852 for opening specific journal files. journactl also gained a
5853 new switch to expose this new functionality. Previously we
5854 only supported opening all files from a directory, or all
5855 files from the system, as opening individual files only is
5856 racy due to journal file rotation.
5858 * systemd gained the new DefaultEnvironment= setting in
5859 /etc/systemd/system.conf to set environment variables for
5862 * If a privileged process logs a journal message with the
5863 OBJECT_PID= field set, then journald will automatically
5864 augment this with additional OBJECT_UID=, OBJECT_GID=,
5865 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5866 system services want to log events about specific client
5867 processes. journactl/systemctl has been updated to make use
5868 of this information if all log messages regarding a specific
5871 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5872 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5873 Reisner, David Coppa, David King, David Strauss, Eelco
5874 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5875 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5876 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5877 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5878 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5879 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5880 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5881 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5882 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5883 Łukasz Stelmach, 장동준
5887 * The Python bindings gained some minimal support for the APIs
5888 exposed by libsystemd-logind.
5890 * ConditionSecurity= gained support for detecting SMACK. Since
5891 this condition already supports SELinux and AppArmor we only
5892 miss IMA for this. Patches welcome!
5894 Contributions from: Karol Lewandowski, Lennart Poettering,
5895 Zbigniew Jędrzejewski-Szmek
5899 * systemd-nspawn will now create /etc/resolv.conf if
5900 necessary, before bind-mounting the host's file onto it.
5902 * systemd-nspawn will now store meta information about a
5903 container on the container's cgroup as extended attribute
5904 fields, including the root directory.
5906 * The cgroup hierarchy has been reworked in many ways. All
5907 objects any of the components systemd creates in the cgroup
5908 tree are now suffixed. More specifically, user sessions are
5909 now placed in cgroups suffixed with ".session", users in
5910 cgroups suffixed with ".user", and nspawn containers in
5911 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5912 names are now escaped in a simple scheme to avoid collision
5913 of userspace object names with kernel filenames. This work
5914 is preparation for making these objects relocatable in the
5915 cgroup tree, in order to allow easy resource partitioning of
5916 these objects without causing naming conflicts.
5918 * systemctl list-dependencies gained the new switches
5919 --plain, --reverse, --after and --before.
5921 * systemd-inhibit now shows the process name of processes that
5922 have taken an inhibitor lock.
5924 * nss-myhostname will now also resolve "localhost"
5925 implicitly. This makes /etc/hosts an optional file and
5926 nicely handles that on IPv6 ::1 maps to both "localhost" and
5929 * libsystemd-logind.so gained a new call
5930 sd_get_machine_names() to enumerate running containers and
5931 VMs (currently only supported by very new libvirt and
5932 nspawn). sd_login_monitor can now be used to watch
5933 VMs/containers coming and going.
5935 * .include is not allowed recursively anymore, and only in
5936 unit files. Usually it is better to use drop-in snippets in
5937 .d/*.conf anyway, as introduced with systemd 198.
5939 * systemd-analyze gained a new "critical-chain" command that
5940 determines the slowest chain of units run during system
5941 boot-up. It is very useful for tracking down where
5942 optimizing boot time is the most beneficial.
5944 * systemd will no longer allow manipulating service paths in
5945 the name=systemd:/system cgroup tree using ControlGroup= in
5946 units. (But is still fine with it in all other dirs.)
5948 * There's a new systemd-nspawn@.service service file that may
5949 be used to easily run nspawn containers as system
5950 services. With the container's root directory in
5951 /var/lib/container/foobar it is now sufficient to run
5952 "systemctl start systemd-nspawn@foobar.service" to boot it.
5954 * systemd-cgls gained a new parameter "--machine" to list only
5955 the processes within a certain container.
5957 * ConditionSecurity= now can check for "apparmor". We still
5958 are lacking checks for SMACK and IMA for this condition
5959 check though. Patches welcome!
5961 * A new configuration file /etc/systemd/sleep.conf has been
5962 added that may be used to configure which kernel operation
5963 systemd is supposed to execute when "suspend", "hibernate"
5964 or "hybrid-sleep" is requested. This makes the new kernel
5965 "freeze" state accessible to the user.
5967 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5968 the passed argument if applicable.
5970 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5971 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5972 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5973 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5974 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5975 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5976 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5981 * The output of 'systemctl list-jobs' got some polishing. The
5982 '--type=' argument may now be passed more than once. A new
5983 command 'systemctl list-sockets' has been added which shows
5984 a list of kernel sockets systemd is listening on with the
5985 socket units they belong to, plus the units these socket
5988 * The experimental libsystemd-bus library got substantial
5989 updates to work in conjunction with the (also experimental)
5990 kdbus kernel project. It works well enough to exchange
5991 messages with some sophistication. Note that kdbus is not
5992 ready yet, and the library is mostly an elaborate test case
5993 for now, and not installable.
5995 * systemd gained a new unit 'systemd-static-nodes.service'
5996 that generates static device nodes earlier during boot, and
5997 can run in conjunction with udev.
5999 * libsystemd-login gained a new call sd_pid_get_user_unit()
6000 to retrieve the user systemd unit a process is running
6001 in. This is useful for systems where systemd is used as
6004 * systemd-nspawn now places all containers in the new /machine
6005 top-level cgroup directory in the name=systemd
6006 hierarchy. libvirt will soon do the same, so that we get a
6007 uniform separation of /system, /user and /machine for system
6008 services, user processes and containers/virtual
6009 machines. This new cgroup hierarchy is also useful to stick
6010 stable names to specific container instances, which can be
6011 recognized later this way (this name may be controlled
6012 via systemd-nspawn's new -M switch). libsystemd-login also
6013 gained a new call sd_pid_get_machine_name() to retrieve the
6014 name of the container/VM a specific process belongs to.
6016 * bootchart can now store its data in the journal.
6018 * libsystemd-journal gained a new call
6019 sd_journal_add_conjunction() for AND expressions to the
6020 matching logic. This can be used to express more complex
6021 logical expressions.
6023 * journactl can now take multiple --unit= and --user-unit=
6026 * The cryptsetup logic now understands the "luks.key=" kernel
6027 command line switch for specifying a file to read the
6028 decryption key from. Also, if a configured key file is not
6029 found the tool will now automatically fall back to prompting
6032 * Python systemd.journal module was updated to wrap recently
6033 added functions from libsystemd-journal. The interface was
6034 changed to bring the low level interface in s.j._Reader
6035 closer to the C API, and the high level interface in
6036 s.j.Reader was updated to wrap and convert all data about
6039 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6040 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6041 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6042 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6043 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6044 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6048 * journalctl --update-catalog now understands a new --root=
6049 option to operate on catalogs found in a different root
6052 * During shutdown after systemd has terminated all running
6053 services a final killing loop kills all remaining left-over
6054 processes. We will now print the name of these processes
6055 when we send SIGKILL to them, since this usually indicates a
6058 * If /etc/crypttab refers to password files stored on
6059 configured mount points automatic dependencies will now be
6060 generated to ensure the specific mount is established first
6061 before the key file is attempted to be read.
6063 * 'systemctl status' will now show information about the
6064 network sockets a socket unit is listening on.
6066 * 'systemctl status' will also shown information about any
6067 drop-in configuration file for units. (Drop-In configuration
6068 files in this context are files such as
6069 /etc/systemd/systemd/foobar.service.d/*.conf)
6071 * systemd-cgtop now optionally shows summed up CPU times of
6072 cgroups. Press '%' while running cgtop to switch between
6073 percentage and absolute mode. This is useful to determine
6074 which cgroups use up the most CPU time over the entire
6075 runtime of the system. systemd-cgtop has also been updated
6076 to be 'pipeable' for processing with further shell tools.
6078 * 'hostnamectl set-hostname' will now allow setting of FQDN
6081 * The formatting and parsing of time span values has been
6082 changed. The parser now understands fractional expressions
6083 such as "5.5h". The formatter will now output fractional
6084 expressions for all time spans under 1min, i.e. "5.123456s"
6085 rather than "5s 123ms 456us". For time spans under 1s
6086 millisecond values are shown, for those under 1ms
6087 microsecond values are shown. This should greatly improve
6088 all time-related output of systemd.
6090 * libsystemd-login and libsystemd-journal gained new
6091 functions for querying the poll() events mask and poll()
6092 timeout value for integration into arbitrary event
6095 * localectl gained the ability to list available X11 keymaps
6096 (models, layouts, variants, options).
6098 * 'systemd-analyze dot' gained the ability to filter for
6099 specific units via shell-style globs, to create smaller,
6100 more useful graphs. I.e. it is now possible to create simple
6101 graphs of all the dependencies between only target units, or
6102 of all units that Avahi has dependencies with.
6104 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6105 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6106 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6107 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6108 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6109 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6110 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6114 * The boot-time readahead implementation for rotating media
6115 will now read the read-ahead data in multiple passes which
6116 consist of all read requests made in equidistant time
6117 intervals. This means instead of strictly reading read-ahead
6118 data in its physical order on disk we now try to find a
6119 middle ground between physical and access time order.
6121 * /etc/os-release files gained a new BUILD_ID= field for usage
6122 on operating systems that provide continuous builds of OS
6125 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6126 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6127 William Douglas, Zbigniew Jędrzejewski-Szmek
6131 * systemd-python gained an API exposing libsystemd-daemon.
6133 * The SMACK setup logic gained support for uploading CIPSO
6136 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6137 ReadOnlyDirectories= and InaccessibleDirectories= has
6138 changed. The private /tmp and /var/tmp directories are now
6139 shared by all processes of a service (which means
6140 ExecStartPre= may now leave data in /tmp that ExecStart= of
6141 the same service can still access). When a service is
6142 stopped its temporary directories are immediately deleted
6143 (normal clean-up with tmpfiles is still done in addition to
6146 * By default, systemd will now set a couple of sysctl
6147 variables in the kernel: the safe sysrq options are turned
6148 on, IP route verification is turned on, and source routing
6149 disabled. The recently added hardlink and softlink
6150 protection of the kernel is turned on. These settings should
6151 be reasonably safe, and good defaults for all new systems.
6153 * The predictable network naming logic may now be turned off
6154 with a new kernel command line switch: net.ifnames=0.
6156 * A new libsystemd-bus module has been added that implements a
6157 pretty complete D-Bus client library. For details see:
6159 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6161 * journald will now explicitly flush the journal files to disk
6162 at the latest 5min after each write. The file will then also
6163 be marked offline until the next write. This should increase
6164 reliability in case of a crash. The synchronization delay
6165 can be configured via SyncIntervalSec= in journald.conf.
6167 * There's a new remote-fs-setup.target unit that can be used
6168 to pull in specific services when at least one remote file
6169 system is to be mounted.
6171 * There are new targets timers.target and paths.target as
6172 canonical targets to pull user timer and path units in
6173 from. This complements sockets.target with a similar
6174 purpose for socket units.
6176 * libudev gained a new call udev_device_set_attribute_value()
6177 to set sysfs attributes of a device.
6179 * The udev daemon now sets the default number of worker
6180 processes executed in parallel based on the number of available
6181 CPUs instead of the amount of available RAM. This is supposed
6182 to provide a more reliable default and limit a too aggressive
6183 parallelism for setups with 1000s of devices connected.
6185 Contributions from: Auke Kok, Colin Walters, Cristian
6186 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6187 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6188 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6189 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6190 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6191 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6192 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6193 Zbigniew Jędrzejewski-Szmek
6197 * Configuration of unit files may now be extended via drop-in
6198 files without having to edit/override the unit files
6199 themselves. More specifically, if the administrator wants to
6200 change one value for a service file foobar.service he can
6201 now do so by dropping in a configuration snippet into
6202 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6203 will load all these snippets and apply them on top of the
6204 main unit configuration file, possibly extending or
6205 overriding its settings. Using these drop-in snippets is
6206 generally nicer than the two earlier options for changing
6207 unit files locally: copying the files from
6208 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6209 them there; or creating a new file in /etc/systemd/system/
6210 that incorporates the original one via ".include". Drop-in
6211 snippets into these .d/ directories can be placed in any
6212 directory systemd looks for units in, and the usual
6213 overriding semantics between /usr/lib, /etc and /run apply
6216 * Most unit file settings which take lists of items can now be
6217 reset by assigning the empty string to them. For example,
6218 normally, settings such as Environment=FOO=BAR append a new
6219 environment variable assignment to the environment block,
6220 each time they are used. By assigning Environment= the empty
6221 string the environment block can be reset to empty. This is
6222 particularly useful with the .d/*.conf drop-in snippets
6223 mentioned above, since this adds the ability to reset list
6224 settings from vendor unit files via these drop-ins.
6226 * systemctl gained a new "list-dependencies" command for
6227 listing the dependencies of a unit recursively.
6229 * Inhibitors are now honored and listed by "systemctl
6230 suspend", "systemctl poweroff" (and similar) too, not only
6231 GNOME. These commands will also list active sessions by
6234 * Resource limits (as exposed by the various control group
6235 controllers) can now be controlled dynamically at runtime
6236 for all units. More specifically, you can now use a command
6237 like "systemctl set-cgroup-attr foobar.service cpu.shares
6238 2000" to alter the CPU shares a specific service gets. These
6239 settings are stored persistently on disk, and thus allow the
6240 administrator to easily adjust the resource usage of
6241 services with a few simple commands. This dynamic resource
6242 management logic is also available to other programs via the
6243 bus. Almost any kernel cgroup attribute and controller is
6246 * systemd-vconsole-setup will now copy all font settings to
6247 all allocated VTs, where it previously applied them only to
6250 * libsystemd-login gained the new sd_session_get_tty() API
6253 * This release drops support for a few legacy or
6254 distribution-specific LSB facility names when parsing init
6255 scripts: $x-display-manager, $mail-transfer-agent,
6256 $mail-transport-agent, $mail-transfer-agent, $smtp,
6257 $null. Also, the mail-transfer-agent.target unit backing
6258 this has been removed. Distributions which want to retain
6259 compatibility with this should carry the burden for
6260 supporting this themselves and patch support for these back
6261 in, if they really need to. Also, the facilities $syslog and
6262 $local_fs are now ignored, since systemd does not support
6263 early-boot LSB init scripts anymore, and these facilities
6264 are implied anyway for normal services. syslog.target has
6267 * There are new bus calls on PID1's Manager object for
6268 cancelling jobs, and removing snapshot units. Previously,
6269 both calls were only available on the Job and Snapshot
6272 * systemd-journal-gatewayd gained SSL support.
6274 * The various "environment" files, such as /etc/locale.conf
6275 now support continuation lines with a backslash ("\") as
6276 last character in the line, similarly in style (but different)
6277 to how this is supported in shells.
6279 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6280 now implicitly appended to every log entry logged. systemctl
6281 has been updated to filter by this field when operating on a
6282 user systemd instance.
6284 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6285 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6286 the container. This makes it easier to boot unmodified
6287 Fedora systems in a container, which however still requires
6288 audit=0 to be passed on the kernel command line. Auditing in
6289 kernel and userspace is unfortunately still too broken in
6290 context of containers, hence we recommend compiling it out
6291 of the kernel or using audit=0. Hopefully this will be fixed
6292 one day for good in the kernel.
6294 * nspawn gained the new --bind= and --bind-ro= parameters to
6295 bind mount specific directories from the host into the
6298 * nspawn will now mount its own devpts file system instance
6299 into the container, in order not to leak pty devices from
6300 the host into the container.
6302 * systemd will now read the firmware boot time performance
6303 information from the EFI variables, if the used boot loader
6304 supports this, and takes it into account for boot performance
6305 analysis via "systemd-analyze". This is currently supported
6306 only in conjunction with Gummiboot, but could be supported
6307 by other boot loaders too. For details see:
6309 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6311 * A new generator has been added that automatically mounts the
6312 EFI System Partition (ESP) to /boot, if that directory
6313 exists, is empty, and no other file system has been
6314 configured to be mounted there.
6316 * logind will now send out PrepareForSleep(false) out
6317 unconditionally, after coming back from suspend. This may be
6318 used by applications as asynchronous notification for
6319 system resume events.
6321 * "systemctl unlock-sessions" has been added, that allows
6322 unlocking the screens of all user sessions at once, similar
6323 to how "systemctl lock-sessions" already locked all users
6324 sessions. This is backed by a new D-Bus call UnlockSessions().
6326 * "loginctl seat-status" will now show the master device of a
6327 seat. (i.e. the device of a seat that needs to be around for
6328 the seat to be considered available, usually the graphics
6331 * tmpfiles gained a new "X" line type, that allows
6332 configuration of files and directories (with wildcards) that
6333 shall be excluded from automatic cleanup ("aging").
6335 * udev default rules set the device node permissions now only
6336 at "add" events, and do not change them any longer with a
6337 later "change" event.
6339 * The log messages for lid events and power/sleep keypresses
6340 now carry a message ID.
6342 * We now have a substantially larger unit test suite, but this
6343 continues to be work in progress.
6345 * udevadm hwdb gained a new --root= parameter to change the
6346 root directory to operate relative to.
6348 * logind will now issue a background sync() request to the kernel
6349 early at shutdown, so that dirty buffers are flushed to disk early
6350 instead of at the last moment, in order to optimize shutdown
6353 * A new bootctl tool has been added that is an interface for
6354 certain boot loader operations. This is currently a preview
6355 and is likely to be extended into a small mechanism daemon
6356 like timedated, localed, hostnamed, and can be used by
6357 graphical UIs to enumerate available boot options, and
6358 request boot into firmware operations.
6360 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6361 the rest of the package. It also has been updated to work
6362 correctly in initrds.
6364 * Policykit previously has been runtime optional, and is now
6365 also compile time optional via a configure switch.
6367 * systemd-analyze has been reimplemented in C. Also "systemctl
6368 dot" has moved into systemd-analyze.
6370 * "systemctl status" with no further parameters will now print
6371 the status of all active or failed units.
6373 * Operations such as "systemctl start" can now be executed
6374 with a new mode "--irreversible" which may be used to queue
6375 operations that cannot accidentally be reversed by a later
6376 job queuing. This is by default used to make shutdown
6377 requests more robust.
6379 * The Python API of systemd now gained a new module for
6380 reading journal files.
6382 * A new tool kernel-install has been added that can install
6383 kernel images according to the Boot Loader Specification:
6385 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6387 * Boot time console output has been improved to provide
6388 animated boot time output for hanging jobs.
6390 * A new tool systemd-activate has been added which can be used
6391 to test socket activation with, directly from the command
6392 line. This should make it much easier to test and debug
6393 socket activation in daemons.
6395 * journalctl gained a new "--reverse" (or -r) option to show
6396 journal output in reverse order (i.e. newest line first).
6398 * journalctl gained a new "--pager-end" (or -e) option to jump
6399 to immediately jump to the end of the journal in the
6400 pager. This is only supported in conjunction with "less".
6402 * journalctl gained a new "--user-unit=" option, that works
6403 similarly to "--unit=" but filters for user units rather than
6406 * A number of unit files to ease adoption of systemd in
6407 initrds has been added. This moves some minimal logic from
6408 the various initrd implementations into systemd proper.
6410 * The journal files are now owned by a new group
6411 "systemd-journal", which exists specifically to allow access
6412 to the journal, and nothing else. Previously, we used the
6413 "adm" group for that, which however possibly covers more
6414 than just journal/log file access. This new group is now
6415 already used by systemd-journal-gatewayd to ensure this
6416 daemon gets access to the journal files and as little else
6417 as possible. Note that "make install" will also set FS ACLs
6418 up for /var/log/journal to give "adm" and "wheel" read
6419 access to it, in addition to "systemd-journal" which owns
6420 the journal files. We recommend that packaging scripts also
6421 add read access to "adm" + "wheel" to /var/log/journal, and
6422 all existing/future journal files. To normal users and
6423 administrators little changes, however packagers need to
6424 ensure to create the "systemd-journal" system group at
6425 package installation time.
6427 * The systemd-journal-gatewayd now runs as unprivileged user
6428 systemd-journal-gateway:systemd-journal-gateway. Packaging
6429 scripts need to create these system user/group at
6432 * timedated now exposes a new boolean property CanNTP that
6433 indicates whether a local NTP service is available or not.
6435 * systemd-detect-virt will now also detect xen PVs
6437 * The pstore file system is now mounted by default, if it is
6440 * In addition to the SELinux and IMA policies we will now also
6441 load SMACK policies at early boot.
6443 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6444 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6445 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6446 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6447 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6448 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6449 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6450 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6451 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6452 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6453 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6454 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6455 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6456 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6460 * Timer units now support calendar time events in addition to
6461 monotonic time events. That means you can now trigger a unit
6462 based on a calendar time specification such as "Thu,Fri
6463 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6464 or fifth day of any month of the year 2013, given that it is
6465 a thursday or friday. This brings timer event support
6466 considerably closer to cron's capabilities. For details on
6467 the supported calendar time specification language see
6470 * udev now supports a number of different naming policies for
6471 network interfaces for predictable names, and a combination
6472 of these policies is now the default. Please see this wiki
6473 document for details:
6475 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6477 * Auke Kok's bootchart implementation has been added to the
6478 systemd tree. It is an optional component that can graph the
6479 boot in quite some detail. It is one of the best bootchart
6480 implementations around and minimal in its code and
6483 * nss-myhostname has been integrated into the systemd source
6484 tree. nss-myhostname guarantees that the local hostname
6485 always stays resolvable via NSS. It has been a weak
6486 requirement of systemd-hostnamed since a long time, and
6487 since its code is actually trivial we decided to just
6488 include it in systemd's source tree. It can be turned off
6489 with a configure switch.
6491 * The read-ahead logic is now capable of properly detecting
6492 whether a btrfs file system is on SSD or rotating media, in
6493 order to optimize the read-ahead scheme. Previously, it was
6494 only capable of detecting this on traditional file systems
6497 * In udev, additional device properties are now read from the
6498 IAB in addition to the OUI database. Also, Bluetooth company
6499 identities are attached to the devices as well.
6501 * In service files %U may be used as specifier that is
6502 replaced by the configured user name of the service.
6504 * nspawn may now be invoked without a controlling TTY. This
6505 makes it suitable for invocation as its own service. This
6506 may be used to set up a simple containerized server system
6507 using only core OS tools.
6509 * systemd and nspawn can now accept socket file descriptors
6510 when they are started for socket activation. This enables
6511 implementation of socket activated nspawn
6512 containers. i.e. think about autospawning an entire OS image
6513 when the first SSH or HTTP connection is received. We expect
6514 that similar functionality will also be added to libvirt-lxc
6517 * journalctl will now suppress ANSI color codes when
6518 presenting log data.
6520 * systemctl will no longer show control group information for
6521 a unit if the control group is empty anyway.
6523 * logind can now automatically suspend/hibernate/shutdown the
6526 * /etc/machine-info and hostnamed now also expose the chassis
6527 type of the system. This can be used to determine whether
6528 the local system is a laptop, desktop, handset or
6529 tablet. This information may either be configured by the
6530 user/vendor or is automatically determined from ACPI and DMI
6531 information if possible.
6533 * A number of PolicyKit actions are now bound together with
6534 "imply" rules. This should simplify creating UIs because
6535 many actions will now authenticate similar ones as well.
6537 * Unit files learnt a new condition ConditionACPower= which
6538 may be used to conditionalize a unit depending on whether an
6539 AC power source is connected or not, of whether the system
6540 is running on battery power.
6542 * systemctl gained a new "is-failed" verb that may be used in
6543 shell scripts and suchlike to check whether a specific unit
6544 is in the "failed" state.
6546 * The EnvironmentFile= setting in unit files now supports file
6547 globbing, and can hence be used to easily read a number of
6548 environment files at once.
6550 * systemd will no longer detect and recognize specific
6551 distributions. All distribution-specific #ifdeffery has been
6552 removed, systemd is now fully generic and
6553 distribution-agnostic. Effectively, not too much is lost as
6554 a lot of the code is still accessible via explicit configure
6555 switches. However, support for some distribution specific
6556 legacy configuration file formats has been dropped. We
6557 recommend distributions to simply adopt the configuration
6558 files everybody else uses now and convert the old
6559 configuration from packaging scripts. Most distributions
6560 already did that. If that's not possible or desirable,
6561 distributions are welcome to forward port the specific
6562 pieces of code locally from the git history.
6564 * When logging a message about a unit systemd will now always
6565 log the unit name in the message meta data.
6567 * localectl will now also discover system locale data that is
6568 not stored in locale archives, but directly unpacked.
6570 * logind will no longer unconditionally use framebuffer
6571 devices as seat masters, i.e. as devices that are required
6572 to be existing before a seat is considered preset. Instead,
6573 it will now look for all devices that are tagged as
6574 "seat-master" in udev. By default, framebuffer devices will
6575 be marked as such, but depending on local systems, other
6576 devices might be marked as well. This may be used to
6577 integrate graphics cards using closed source drivers (such
6578 as NVidia ones) more nicely into logind. Note however, that
6579 we recommend using the open source NVidia drivers instead,
6580 and no udev rules for the closed-source drivers will be
6581 shipped from us upstream.
6583 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6584 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6585 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6586 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6587 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6588 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6589 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6590 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6591 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6592 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6593 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6594 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6599 * udev gained support for loading additional device properties
6600 from an indexed database that is keyed by vendor/product IDs
6601 and similar device identifiers. For the beginning this
6602 "hwdb" is populated with data from the well-known PCI and
6603 USB database, but also includes PNP, ACPI and OID data. In
6604 the longer run this indexed database shall grow into
6605 becoming the one central database for non-essential
6606 userspace device metadata. Previously, data from the PCI/USB
6607 database was only attached to select devices, since the
6608 lookup was a relatively expensive operation due to O(n) time
6609 complexity (with n being the number of entries in the
6610 database). Since this is now O(1), we decided to add in this
6611 data for all devices where this is available, by
6612 default. Note that the indexed database needs to be rebuilt
6613 when new data files are installed. To achieve this you need
6614 to update your packaging scripts to invoke "udevadm hwdb
6615 --update" after installation of hwdb data files. For
6616 RPM-based distributions we introduced the new
6617 %udev_hwdb_update macro for this purpose.
6619 * The Journal gained support for the "Message Catalog", an
6620 indexed database to link up additional information with
6621 journal entries. For further details please check:
6623 https://www.freedesktop.org/wiki/Software/systemd/catalog
6625 The indexed message catalog database also needs to be
6626 rebuilt after installation of message catalog files. Use
6627 "journalctl --update-catalog" for this. For RPM-based
6628 distributions we introduced the %journal_catalog_update
6629 macro for this purpose.
6631 * The Python Journal bindings gained support for the standard
6632 Python logging framework.
6634 * The Journal API gained new functions for checking whether
6635 the underlying file system of a journal file is capable of
6636 properly reporting file change notifications, or whether
6637 applications that want to reflect journal changes "live"
6638 need to recheck journal files continuously in appropriate
6641 * It is now possible to set the "age" field for tmpfiles
6642 entries to 0, indicating that files matching this entry
6643 shall always be removed when the directories are cleaned up.
6645 * coredumpctl gained a new "gdb" verb which invokes gdb
6646 right-away on the selected coredump.
6648 * There's now support for "hybrid sleep" on kernels that
6649 support this, in addition to "suspend" and "hibernate". Use
6650 "systemctl hybrid-sleep" to make use of this.
6652 * logind's HandleSuspendKey= setting (and related settings)
6653 now gained support for a new "lock" setting to simply
6654 request the screen lock on all local sessions, instead of
6655 actually executing a suspend or hibernation.
6657 * systemd will now mount the EFI variables file system by
6660 * Socket units now gained support for configuration of the
6661 SMACK security label.
6663 * timedatectl will now output the time of the last and next
6664 daylight saving change.
6666 * We dropped support for various legacy and distro-specific
6667 concepts, such as insserv, early-boot SysV services
6668 (i.e. those for non-standard runlevels such as 'b' or 'S')
6669 or ArchLinux /etc/rc.conf support. We recommend the
6670 distributions who still need support this to either continue
6671 to maintain the necessary patches downstream, or find a
6672 different solution. (Talk to us if you have questions!)
6674 * Various systemd components will now bypass PolicyKit checks
6675 for root and otherwise handle properly if PolicyKit is not
6676 found to be around. This should fix most issues for
6677 PolicyKit-less systems. Quite frankly this should have been
6678 this way since day one. It is absolutely our intention to
6679 make systemd work fine on PolicyKit-less systems, and we
6680 consider it a bug if something does not work as it should if
6681 PolicyKit is not around.
6683 * For embedded systems it is now possible to build udev and
6684 systemd without blkid and/or kmod support.
6686 * "systemctl switch-root" is now capable of switching root
6687 more than once. I.e. in addition to transitions from the
6688 initrd to the host OS it is now possible to transition to
6689 further OS images from the host. This is useful to implement
6690 offline updating tools.
6692 * Various other additions have been made to the RPM macros
6693 shipped with systemd. Use %udev_rules_update() after
6694 installing new udev rules files. %_udevhwdbdir,
6695 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6696 %_sysctldir are now available which resolve to the right
6697 directories for packages to place various data files in.
6699 * journalctl gained the new --full switch (in addition to
6700 --all, to disable ellipsation for long messages.
6702 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6703 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6704 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6705 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6706 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6707 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6708 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6709 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6710 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6714 * journalctl gained new --since= and --until= switches to
6715 filter by time. It also now supports nice filtering for
6716 units via --unit=/-u.
6718 * Type=oneshot services may use ExecReload= and do the
6721 * The journal daemon now supports time-based rotation and
6722 vacuuming, in addition to the usual disk-space based
6725 * The journal will now index the available field values for
6726 each field name. This enables clients to show pretty drop
6727 downs of available match values when filtering. The bash
6728 completion of journalctl has been updated
6729 accordingly. journalctl gained a new switch -F to list all
6730 values a certain field takes in the journal database.
6732 * More service events are now written as structured messages
6733 to the journal, and made recognizable via message IDs.
6735 * The timedated, localed and hostnamed mini-services which
6736 previously only provided support for changing time, locale
6737 and hostname settings from graphical DEs such as GNOME now
6738 also have a minimal (but very useful) text-based client
6739 utility each. This is probably the nicest way to changing
6740 these settings from the command line now, especially since
6741 it lists available options and is fully integrated with bash
6744 * There's now a new tool "systemd-coredumpctl" to list and
6745 extract coredumps from the journal.
6747 * We now install a README each in /var/log/ and
6748 /etc/rc.d/init.d explaining where the system logs and init
6749 scripts went. This hopefully should help folks who go to
6750 that dirs and look into the otherwise now empty void and
6751 scratch their heads.
6753 * When user-services are invoked (by systemd --user) the
6754 $MANAGERPID env var is set to the PID of systemd.
6756 * SIGRTMIN+24 when sent to a --user instance will now result
6757 in immediate termination of systemd.
6759 * gatewayd received numerous feature additions such as a
6760 "follow" mode, for live syncing and filtering.
6762 * browse.html now allows filtering and showing detailed
6763 information on specific entries. Keyboard navigation and
6764 mouse screen support has been added.
6766 * gatewayd/journalctl now supports HTML5/JSON
6767 Server-Sent-Events as output.
6769 * The SysV init script compatibility logic will now
6770 heuristically determine whether a script supports the
6771 "reload" verb, and only then make this available as
6774 * "systemctl status --follow" has been removed, use "journalctl
6777 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6778 have been removed since they are hardly useful to be
6781 * And I'd like to take the opportunity to specifically mention
6782 Zbigniew for his great contributions. Zbigniew, you rock!
6784 Contributions from: Andrew Eikum, Christian Hesse, Colin
6785 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6786 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6787 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6788 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6789 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6790 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6794 * If /etc/vconsole.conf is non-existent or empty we will no
6795 longer load any console font or key map at boot by
6796 default. Instead the kernel defaults will be left
6797 intact. This is definitely the right thing to do, as no
6798 configuration should mean no configuration, and hard-coding
6799 font names that are different on all archs is probably a bad
6800 idea. Also, the kernel default key map and font should be
6801 good enough for most cases anyway, and mostly identical to
6802 the userspace fonts/key maps we previously overloaded them
6803 with. If distributions want to continue to default to a
6804 non-kernel font or key map they should ship a default
6805 /etc/vconsole.conf with the appropriate contents.
6807 Contributions from: Colin Walters, Daniel J Walsh, Dave
6808 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6809 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6813 * journalctl gained a new --cursor= switch to show entries
6814 starting from the specified location in the journal.
6816 * We now enforce a size limit on journal entry fields exported
6817 with "-o json" in journalctl. Fields larger than 4K will be
6818 assigned null. This can be turned off with --all.
6820 * An (optional) journal gateway daemon is now available as
6821 "systemd-journal-gatewayd.service". This service provides
6822 access to the journal via HTTP and JSON. This functionality
6823 will be used to implement live log synchronization in both
6824 pull and push modes, but has various other users too, such
6825 as easy log access for debugging of embedded devices. Right
6826 now it is already useful to retrieve the journal via HTTP:
6828 # systemctl start systemd-journal-gatewayd.service
6829 # wget http://localhost:19531/entries
6831 This will download the journal contents in a
6832 /var/log/messages compatible format. The same as JSON:
6834 # curl -H"Accept: application/json" http://localhost:19531/entries
6836 This service is also accessible via a web browser where a
6837 single static HTML5 app is served that uses the JSON logic
6838 to enable the user to do some basic browsing of the
6839 journal. This will be extended later on. Here's an example
6840 screenshot of this app in its current state:
6842 http://0pointer.de/public/journal-gatewayd
6844 Contributions from: Kay Sievers, Lennart Poettering, Robert
6845 Milasan, Tom Gundersen
6849 * The bash completion logic is now available for journalctl
6852 * We do not mount the "cpuset" controller anymore together with
6853 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6854 started if no parameters are assigned to it. "cpuset" hence
6855 broke code that assumed it could create "cpu" groups and
6858 * journalctl -f will now subscribe to terminal size changes,
6859 and line break accordingly.
6861 Contributions from: Dave Reisner, Kay Sievers, Lennart
6862 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6866 * nspawn will now create a symlink /etc/localtime in the
6867 container environment, copying the host's timezone
6868 setting. Previously this has been done via a bind mount, but
6869 since symlinks cannot be bind mounted this has now been
6870 changed to create/update the appropriate symlink.
6872 * journalctl -n's line number argument is now optional, and
6873 will default to 10 if omitted.
6875 * journald will now log the maximum size the journal files may
6876 take up on disk. This is particularly useful if the default
6877 built-in logic of determining this parameter from the file
6878 system size is used. Use "systemctl status
6879 systemd-journald.service" to see this information.
6881 * The multi-seat X wrapper tool has been stripped down. As X
6882 is now capable of enumerating graphics devices via udev in a
6883 seat-aware way the wrapper is not strictly necessary
6884 anymore. A stripped down temporary stop-gap is still shipped
6885 until the upstream display managers have been updated to
6886 fully support the new X logic. Expect this wrapper to be
6887 removed entirely in one of the next releases.
6889 * HandleSleepKey= in logind.conf has been split up into
6890 HandleSuspendKey= and HandleHibernateKey=. The old setting
6891 is not available anymore. X11 and the kernel are
6892 distinguishing between these keys and we should too. This
6893 also means the inhibition lock for these keys has been split
6896 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6897 Poettering, Lukas Nykryn, Václav Pavlín
6901 * Whenever a unit changes state we will now log this to the
6902 journal and show along the unit's own log output in
6905 * ConditionPathIsMountPoint= can now properly detect bind
6906 mount points too. (Previously, a bind mount of one file
6907 system to another place in the same file system could not be
6908 detected as mount, since they shared struct stat's st_dev
6911 * We will now mount the cgroup controllers cpu, cpuacct,
6912 cpuset and the controllers net_cls, net_prio together by
6915 * nspawn containers will now have a virtualized boot
6916 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6917 over with a randomized ID at container initialization). This
6918 has the effect of making "journalctl -b" do the right thing
6921 * The JSON output journal serialization has been updated not
6922 to generate "endless" list objects anymore, but rather one
6923 JSON object per line. This is more in line how most JSON
6924 parsers expect JSON objects. The new output mode
6925 "json-pretty" has been added to provide similar output, but
6926 neatly aligned for readability by humans.
6928 * We dropped all explicit sync() invocations in the shutdown
6929 code. The kernel does this implicitly anyway in the kernel
6930 reboot() syscall. halt(8)'s -n option is now a compatibility
6933 * We now support virtualized reboot() in containers, as
6934 supported by newer kernels. We will fall back to exit() if
6935 CAP_SYS_REBOOT is not available to the container. Also,
6936 nspawn makes use of this now and will actually reboot the
6937 container if the containerized OS asks for that.
6939 * journalctl will only show local log output by default
6940 now. Use --merge (-m) to show remote log output, too.
6942 * libsystemd-journal gained the new sd_journal_get_usage()
6943 call to determine the current disk usage of all journal
6944 files. This is exposed in the new "journalctl --disk-usage"
6947 * journald gained a new configuration setting SplitMode= in
6948 journald.conf which may be used to control how user journals
6949 are split off. See journald.conf(5) for details.
6951 * A new condition type ConditionFileNotEmpty= has been added.
6953 * tmpfiles' "w" lines now support file globbing, to write
6954 multiple files at once.
6956 * We added Python bindings for the journal submission
6957 APIs. More Python APIs for a number of selected APIs will
6958 likely follow. Note that we intend to add native bindings
6959 only for the Python language, as we consider it common
6960 enough to deserve bindings shipped within systemd. There are
6961 various projects outside of systemd that provide bindings
6962 for languages such as PHP or Lua.
6964 * Many conditions will now resolve specifiers such as %i. In
6965 addition, PathChanged= and related directives of .path units
6966 now support specifiers as well.
6968 * There's now a new RPM macro definition for the system preset
6971 * journald will now warn if it ca not forward a message to the
6972 syslog daemon because its socket is full.
6974 * timedated will no longer write or process /etc/timezone,
6975 except on Debian. As we do not support late mounted /usr
6976 anymore /etc/localtime always being a symlink is now safe,
6977 and hence the information in /etc/timezone is not necessary
6980 * logind will now always reserve one VT for a text getty (VT6
6981 by default). Previously if more than 6 X sessions where
6982 started they took up all the VTs with auto-spawned gettys,
6983 so that no text gettys were available anymore.
6985 * udev will now automatically inform the btrfs kernel logic
6986 about btrfs RAID components showing up. This should make
6987 simple hotplug based btrfs RAID assembly work.
6989 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6990 (but not for its children which will stay at the kernel
6991 default). This should allow setups with a lot more listening
6994 * systemd will now always pass the configured timezone to the
6995 kernel at boot. timedated will do the same when the timezone
6998 * logind's inhibition logic has been updated. By default,
6999 logind will now handle the lid switch, the power and sleep
7000 keys all the time, even in graphical sessions. If DEs want
7001 to handle these events on their own they should take the new
7002 handle-power-key, handle-sleep-key and handle-lid-switch
7003 inhibitors during their runtime. A simple way to achieve
7004 that is to invoke the DE wrapped in an invocation of:
7006 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7008 * Access to unit operations is now checked via SELinux taking
7009 the unit file label and client process label into account.
7011 * systemd will now notify the administrator in the journal
7012 when he over-mounts a non-empty directory.
7014 * There are new specifiers that are resolved in unit files,
7015 for the host name (%H), the machine ID (%m) and the boot ID
7018 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7019 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7020 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7021 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7022 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7023 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7024 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7028 * Support for reading structured kernel messages from
7029 /dev/kmsg has now been added and is enabled by default.
7031 * Support for reading kernel messages from /proc/kmsg has now
7032 been removed. If you want kernel messages in the journal
7033 make sure to run a recent kernel (>= 3.5) that supports
7034 reading structured messages from /dev/kmsg (see
7035 above). /proc/kmsg is now exclusive property of classic
7036 syslog daemons again.
7038 * The libudev API gained the new
7039 udev_device_new_from_device_id() call.
7041 * The logic for file system namespace (ReadOnlyDirectory=,
7042 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7043 require pivot_root() anymore. This means fewer temporary
7044 directories are created below /tmp for this feature.
7046 * nspawn containers will now see and receive all submounts
7047 made on the host OS below the root file system of the
7050 * Forward Secure Sealing is now supported for Journal files,
7051 which provide cryptographical sealing of journal files so
7052 that attackers cannot alter log history anymore without this
7053 being detectable. Lennart will soon post a blog story about
7054 this explaining it in more detail.
7056 * There are two new service settings RestartPreventExitStatus=
7057 and SuccessExitStatus= which allow configuration of exit
7058 status (exit code or signal) which will be excepted from the
7059 restart logic, resp. consider successful.
7061 * journalctl gained the new --verify switch that can be used
7062 to check the integrity of the structure of journal files and
7063 (if Forward Secure Sealing is enabled) the contents of
7066 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7067 and similar symlinks pre-created. This makes running shells
7068 as container init process a lot more fun.
7070 * The fstab support can now handle PARTUUID= and PARTLABEL=
7073 * A new ConditionHost= condition has been added to match
7074 against the hostname (with globs) and machine ID. This is
7075 useful for clusters where a single OS image is used to
7076 provision a large number of hosts which shall run slightly
7077 different sets of services.
7079 * Services which hit the restart limit will now be placed in a
7082 Contributions from: Bertram Poettering, Dave Reisner, Huang
7083 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7084 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7088 * When running in --user mode systemd will now become a
7089 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7090 tree a lot more organized.
7092 * A new PartOf= unit dependency type has been introduced that
7093 may be used to group services in a natural way.
7095 * "systemctl enable" may now be used to enable instances of
7098 * journalctl now prints error log levels in red, and
7099 warning/notice log levels in bright white. It also supports
7100 filtering by log level now.
7102 * cgtop gained a new -n switch (similar to top), to configure
7103 the maximum number of iterations to run for. It also gained
7104 -b, to run in batch mode (accepting no input).
7106 * The suffix ".service" may now be omitted on most systemctl
7107 command lines involving service unit names.
7109 * There's a new bus call in logind to lock all sessions, as
7110 well as a loginctl verb for it "lock-sessions".
7112 * libsystemd-logind.so gained a new call sd_journal_perror()
7113 that works similar to libc perror() but logs to the journal
7114 and encodes structured information about the error number.
7116 * /etc/crypttab entries now understand the new keyfile-size=
7119 * shutdown(8) now can send a (configurable) wall message when
7120 a shutdown is cancelled.
7122 * The mount propagation mode for the root file system will now
7123 default to "shared", which is useful to make containers work
7124 nicely out-of-the-box so that they receive new mounts from
7125 the host. This can be undone locally by running "mount
7126 --make-rprivate /" if needed.
7128 * The prefdm.service file has been removed. Distributions
7129 should maintain this unit downstream if they intend to keep
7130 it around. However, we recommend writing normal unit files
7131 for display managers instead.
7133 * Since systemd is a crucial part of the OS we will now
7134 default to a number of compiler switches that improve
7135 security (hardening) such as read-only relocations, stack
7136 protection, and suchlike.
7138 * The TimeoutSec= setting for services is now split into
7139 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7140 of individual time outs for the start and the stop phase of
7143 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7144 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7145 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7146 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7147 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7148 Gundersen, Zbigniew Jędrzejewski-Szmek
7152 * The journal and id128 C APIs are now fully documented as man
7155 * Extra safety checks have been added when transitioning from
7156 the initial RAM disk to the main system to avoid accidental
7159 * /etc/crypttab entries now understand the new keyfile-offset=
7162 * systemctl -t can now be used to filter by unit load state.
7164 * The journal C API gained the new sd_journal_wait() call to
7165 make writing synchronous journal clients easier.
7167 * journalctl gained the new -D switch to show journals from a
7170 * journalctl now displays a special marker between log
7171 messages of two different boots.
7173 * The journal is now explicitly flushed to /var via a service
7174 systemd-journal-flush.service, rather than implicitly simply
7175 by seeing /var/log/journal to be writable.
7177 * journalctl (and the journal C APIs) can now match for much
7178 more complex expressions, with alternatives and
7181 * When transitioning from the initial RAM disk to the main
7182 system we will now kill all processes in a killing spree to
7183 ensure no processes stay around by accident.
7185 * Three new specifiers may be used in unit files: %u, %h, %s
7186 resolve to the user name, user home directory resp. user
7187 shell. This is useful for running systemd user instances.
7189 * We now automatically rotate journal files if their data
7190 object hash table gets a fill level > 75%. We also size the
7191 hash table based on the configured maximum file size. This
7192 together should lower hash collisions drastically and thus
7193 speed things up a bit.
7195 * journalctl gained the new "--header" switch to introspect
7196 header data of journal files.
7198 * A new setting SystemCallFilters= has been added to services
7199 which may be used to apply blacklists or whitelists to
7200 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7202 * nspawn gained a new --link-journal= switch (and quicker: -j)
7203 to link the container journal with the host. This makes it
7204 very easy to centralize log viewing on the host for all
7205 guests while still keeping the journal files separated.
7207 * Many bugfixes and optimizations
7209 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7210 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7211 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7216 * Several tools now understand kernel command line arguments,
7217 which are only read when run in an initial RAM disk. They
7218 usually follow closely their normal counterparts, but are
7221 * There's a new tool to analyze the readahead files that are
7222 automatically generated at boot. Use:
7224 /usr/lib/systemd/systemd-readahead analyze /.readahead
7226 * We now provide an early debug shell on tty9 if this enabled. Use:
7228 systemctl enable debug-shell.service
7230 * All plymouth related units have been moved into the Plymouth
7231 package. Please make sure to upgrade your Plymouth version
7234 * systemd-tmpfiles now supports getting passed the basename of
7235 a configuration file only, in which case it will look for it
7236 in all appropriate directories automatically.
7238 * udevadm info now takes a /dev or /sys path as argument, and
7239 does the right thing. Example:
7241 udevadm info /dev/sda
7242 udevadm info /sys/class/block/sda
7244 * systemctl now prints a warning if a unit is stopped but a
7245 unit that might trigger it continues to run. Example: a
7246 service is stopped but the socket that activates it is left
7249 * "systemctl status" will now mention if the log output was
7250 shortened due to rotation since a service has been started.
7252 * The journal API now exposes functions to determine the
7253 "cutoff" times due to rotation.
7255 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7256 immediately flushing of runtime logs to /var if possible,
7257 resp. for triggering immediate rotation of the journal
7260 * It is now considered an error if a service is attempted to
7261 be stopped that is not loaded.
7263 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7265 * systemd-analyze now supports Python 3
7267 * tmpfiles now supports cleaning up directories via aging
7268 where the first level dirs are always kept around but
7269 directories beneath it automatically aged. This is enabled
7270 by prefixing the age field with '~'.
7272 * Seat objects now expose CanGraphical, CanTTY properties
7273 which is required to deal with very fast bootups where the
7274 display manager might be running before the graphics drivers
7275 completed initialization.
7277 * Seat objects now expose a State property.
7279 * We now include RPM macros for service enabling/disabling
7280 based on the preset logic. We recommend RPM based
7281 distributions to make use of these macros if possible. This
7282 makes it simpler to reuse RPM spec files across
7285 * We now make sure that the collected systemd unit name is
7286 always valid when services log to the journal via
7289 * There's a new man page kernel-command-line(7) detailing all
7290 command line options we understand.
7292 * The fstab generator may now be disabled at boot by passing
7293 fstab=0 on the kernel command line.
7295 * A new kernel command line option modules-load= is now understood
7296 to load a specific kernel module statically, early at boot.
7298 * Unit names specified on the systemctl command line are now
7299 automatically escaped as needed. Also, if file system or
7300 device paths are specified they are automatically turned
7301 into the appropriate mount or device unit names. Example:
7303 systemctl status /home
7304 systemctl status /dev/sda
7306 * The SysVConsole= configuration option has been removed from
7307 system.conf parsing.
7309 * The SysV search path is no longer exported on the D-Bus
7312 * The Names= option has been removed from unit file parsing.
7314 * There's a new man page bootup(7) detailing the boot process.
7316 * Every unit and every generator we ship with systemd now
7317 comes with full documentation. The self-explanatory boot is
7320 * A couple of services gained "systemd-" prefixes in their
7321 name if they wrap systemd code, rather than only external
7322 code. Among them fsck@.service which is now
7323 systemd-fsck@.service.
7325 * The HaveWatchdog property has been removed from the D-Bus
7328 * systemd.confirm_spawn= on the kernel command line should now
7331 * There's a new man page crypttab(5) which details all options
7332 we actually understand.
7334 * systemd-nspawn gained a new --capability= switch to pass
7335 additional capabilities to the container.
7337 * timedated will now read known NTP implementation unit names
7338 from /usr/lib/systemd/ntp-units.d/*.list,
7339 systemd-timedated-ntp.target has been removed.
7341 * journalctl gained a new switch "-b" that lists log data of
7342 the current boot only.
7344 * The notify socket is in the abstract namespace again, in
7345 order to support daemons which chroot() at start-up.
7347 * There is a new Storage= configuration option for journald
7348 which allows configuration of where log data should go. This
7349 also provides a way to disable journal logging entirely, so
7350 that data collected is only forwarded to the console, the
7351 kernel log buffer or another syslog implementation.
7353 * Many bugfixes and optimizations
7355 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7356 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7357 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7358 Shawn Landden, Tom Gundersen
7362 * "systemctl help <unit>" now shows the man page if one is
7365 * Several new man pages have been added.
7367 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7368 MaxLevelConsole= can now be specified in
7369 journald.conf. These options allow reducing the amount of
7370 data stored on disk or forwarded by the log level.
7372 * TimerSlackNSec= can now be specified in system.conf for
7373 PID1. This allows system-wide power savings.
7375 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7376 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7381 * logind is now capable of (optionally) handling power and
7382 sleep keys as well as the lid switch.
7384 * journalctl now understands the syntax "journalctl
7385 /usr/bin/avahi-daemon" to get all log output of a specific
7388 * CapabilityBoundingSet= in system.conf now also influences
7389 the capability bound set of usermode helpers of the kernel.
7391 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7392 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7393 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7394 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7398 * Note that we skipped 139 releases here in order to set the
7399 new version to something that is greater than both udev's
7400 and systemd's most recent version number.
7402 * udev: all udev sources are merged into the systemd source tree now.
7403 All future udev development will happen in the systemd tree. It
7404 is still fully supported to use the udev daemon and tools without
7405 systemd running, like in initramfs or other init systems. Building
7406 udev though, will require the *build* of the systemd tree, but
7407 udev can be properly *run* without systemd.
7409 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7410 should be used to create dead device nodes as workarounds for broken
7413 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7414 no longer supported. udev_monitor_new_from_netlink() needs to be
7415 used to subscribe to events.
7417 * udev: when udevd is started by systemd, processes which are left
7418 behind by forking them off of udev rules, are unconditionally cleaned
7419 up and killed now after the event handling has finished. Services or
7420 daemons must be started as systemd services. Services can be
7421 pulled-in by udev to get started, but they can no longer be directly
7422 forked by udev rules.
7424 * udev: the daemon binary is called systemd-udevd now and installed
7425 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7426 to adapt to that, create symlink, or rename the binary after building
7429 * libudev no longer provides these symbols:
7430 udev_monitor_from_socket()
7431 udev_queue_get_failed_list_entry()
7432 udev_get_{dev,sys,run}_path()
7433 The versions number was bumped and symbol versioning introduced.
7435 * systemd-loginctl and systemd-journalctl have been renamed
7436 to loginctl and journalctl to match systemctl.
7438 * The config files: /etc/systemd/systemd-logind.conf and
7439 /etc/systemd/systemd-journald.conf have been renamed to
7440 logind.conf and journald.conf. Package updates should rename
7441 the files to the new names on upgrade.
7443 * For almost all files the license is now LGPL2.1+, changed
7444 from the previous GPL2.0+. Exceptions are some minor stuff
7445 of udev (which will be changed to LGPL2.1 eventually, too),
7446 and the MIT licensed sd-daemon.[ch] library that is suitable
7447 to be used as drop-in files.
7449 * systemd and logind now handle system sleep states, in
7450 particular suspending and hibernating.
7452 * logind now implements a sleep/shutdown/idle inhibiting logic
7453 suitable for a variety of uses. Soonishly Lennart will blog
7454 about this in more detail.
7456 * var-run.mount and var-lock.mount are no longer provided
7457 (which previously bind mounted these directories to their new
7458 places). Distributions which have not converted these
7459 directories to symlinks should consider stealing these files
7460 from git history and add them downstream.
7462 * We introduced the Documentation= field for units and added
7463 this to all our shipped units. This is useful to make it
7464 easier to explore the boot and the purpose of the various
7467 * All smaller setup units (such as
7468 systemd-vconsole-setup.service) now detect properly if they
7469 are run in a container and are skipped when
7470 appropriate. This guarantees an entirely noise-free boot in
7471 Linux container environments such as systemd-nspawn.
7473 * A framework for implementing offline system updates is now
7474 integrated, for details see:
7475 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7477 * A new service type Type=idle is available now which helps us
7478 avoiding ugly interleaving of getty output and boot status
7481 * There's now a system-wide CapabilityBoundingSet= option to
7482 globally reduce the set of capabilities for the
7483 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7484 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7485 even CAP_NET_ADMIN system-wide for secure systems.
7487 * There are now system-wide DefaultLimitXXX= options to
7488 globally change the defaults of the various resource limits
7489 for all units started by PID 1.
7491 * Harald Hoyer's systemd test suite has been integrated into
7492 systemd which allows easy testing of systemd builds in qemu
7493 and nspawn. (This is really awesome! Ask us for details!)
7495 * The fstab parser is now implemented as generator, not inside
7498 * systemctl will now warn you if .mount units generated from
7499 /etc/fstab are out of date due to changes in fstab that
7500 have not been read by systemd yet.
7502 * systemd is now suitable for usage in initrds. Dracut has
7503 already been updated to make use of this. With this in place
7504 initrds get a slight bit faster but primarily are much
7505 easier to introspect and debug since "systemctl status" in
7506 the host system can be used to introspect initrd services,
7507 and the journal from the initrd is kept around too.
7509 * systemd-delta has been added, a tool to explore differences
7510 between user/admin configuration and vendor defaults.
7512 * PrivateTmp= now affects both /tmp and /var/tmp.
7514 * Boot time status messages are now much prettier and feature
7515 proper english language. Booting up systemd has never been
7518 * Read-ahead pack files now include the inode number of all
7519 files to pre-cache. When the inode changes the pre-caching
7520 is not attempted. This should be nicer to deal with updated
7521 packages which might result in changes of read-ahead
7524 * We now temporaritly lower the kernel's read_ahead_kb variable
7525 when collecting read-ahead data to ensure the kernel's
7526 built-in read-ahead does not add noise to our measurements
7527 of necessary blocks to pre-cache.
7529 * There's now RequiresMountsFor= to add automatic dependencies
7530 for all mounts necessary for a specific file system path.
7532 * MountAuto= and SwapAuto= have been removed from
7533 system.conf. Mounting file systems at boot has to take place
7536 * nspawn now learned a new switch --uuid= to set the machine
7537 ID on the command line.
7539 * nspawn now learned the -b switch to automatically search
7542 * vt102 is now the default TERM for serial TTYs, upgraded from
7545 * systemd-logind now works on VT-less systems.
7547 * The build tree has been reorganized. The individual
7548 components now have directories of their own.
7550 * A new condition type ConditionPathIsReadWrite= is now available.
7552 * nspawn learned the new -C switch to create cgroups for the
7553 container in other hierarchies.
7555 * We now have support for hardware watchdogs, configurable in
7558 * The scheduled shutdown logic now has a public API.
7560 * We now mount /tmp as tmpfs by default, but this can be
7561 masked and /etc/fstab can override it.
7563 * Since udisks does not make use of /media anymore we are not
7564 mounting a tmpfs on it anymore.
7566 * journalctl gained a new --local switch to only interleave
7567 locally generated journal files.
7569 * We can now load the IMA policy at boot automatically.
7571 * The GTK tools have been split off into a systemd-ui.
7573 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7574 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7575 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7576 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7577 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7578 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7579 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7580 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7581 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7586 * This is mostly a bugfix release
7588 * Support optional initialization of the machine ID from the
7589 KVM or container configured UUID.
7591 * Support immediate reboots with "systemctl reboot -ff"
7593 * Show /etc/os-release data in systemd-analyze output
7595 * Many bugfixes for the journal, including endianness fixes and
7596 ensuring that disk space enforcement works
7598 * sd-login.h is C++ compatible again
7600 * Extend the /etc/os-release format on request of the Debian
7603 * We now refuse non-UTF8 strings used in various configuration
7604 and unit files. This is done to ensure we do not pass invalid
7605 data over D-Bus or expose it elsewhere.
7607 * Register Mimo USB Screens as suitable for automatic seat
7610 * Read SELinux client context from journal clients in a race
7613 * Reorder configuration file lookup order. /etc now always
7614 overrides /run in order to allow the administrator to always
7615 and unconditionally override vendor-supplied or
7616 automatically generated data.
7618 * The various user visible bits of the journal now have man
7619 pages. We still lack man pages for the journal API calls
7622 * We now ship all man pages in HTML format again in the
7625 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7626 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7627 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7632 * This is mostly a bugfix release
7634 * systems lacking /etc/os-release are no longer supported.
7636 * Various functionality updates to libsystemd-login.so
7638 * Track class of PAM logins to distinguish greeters from
7641 Contributions from: Kay Sievers, Lennart Poettering, Michael
7646 * This is an important bugfix release for v41.
7648 * Building man pages is now optional which should be useful
7649 for those building systemd from git but unwilling to install
7652 * Watchdog support for supervising services is now usable. In
7653 a future release support for hardware watchdogs
7654 (i.e. /dev/watchdog) will be added building on this.
7656 * Service start rate limiting is now configurable and can be
7657 turned off per service. When a start rate limit is hit a
7658 reboot can automatically be triggered.
7660 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7662 Contributions from: Benjamin Franzke, Bill Nottingham,
7663 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7664 Schmidt, Michał Górny, Piotr Drąg
7668 * The systemd binary is installed /usr/lib/systemd/systemd now;
7669 An existing /sbin/init symlink needs to be adapted with the
7672 * The code that loads kernel modules has been ported to invoke
7673 libkmod directly, instead of modprobe. This means we do not
7674 support systems with module-init-tools anymore.
7676 * Watchdog support is now already useful, but still not
7679 * A new kernel command line option systemd.setenv= is
7680 understood to set system wide environment variables
7681 dynamically at boot.
7683 * We now limit the set of capabilities of systemd-journald.
7685 * We now set SIGPIPE to ignore by default, since it only is
7686 useful in shell pipelines, and has little use in general
7687 code. This can be disabled with IgnoreSIPIPE=no in unit
7690 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7691 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7696 * This is mostly a bugfix release
7698 * We now expose the reason why a service failed in the
7699 "Result" D-Bus property.
7701 * Rudimentary service watchdog support (will be completed over
7702 the next few releases.)
7704 * When systemd forks off in order execute some service we will
7705 now immediately changes its argv[0] to reflect which process
7706 it will execute. This is useful to minimize the time window
7707 with a generic argv[0], which makes bootcharts more useful
7709 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7710 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7711 Mike Kazantsev, Ray Strode
7715 * This is mostly a test release, but incorporates many
7718 * New systemd-cgtop tool to show control groups by their
7721 * Linking against libacl for ACLs is optional again. If
7722 disabled, support tracking device access for active logins
7723 goes becomes unavailable, and so does access to the user
7724 journals by the respective users.
7726 * If a group "adm" exists, journal files are automatically
7727 owned by them, thus allow members of this group full access
7728 to the system journal as well as all user journals.
7730 * The journal now stores the SELinux context of the logging
7731 client for all entries.
7733 * Add C++ inclusion guards to all public headers
7735 * New output mode "cat" in the journal to print only text
7736 messages, without any meta data like date or time.
7738 * Include tiny X server wrapper as a temporary stop-gap to
7739 teach XOrg udev display enumeration. This is used by display
7740 managers such as gdm, and will go away as soon as XOrg
7741 learned native udev hotplugging for display devices.
7743 * Add new systemd-cat tool for executing arbitrary programs
7744 with STDERR/STDOUT connected to the journal. Can also act as
7745 BSD logger replacement, and does so by default.
7747 * Optionally store all locally generated coredumps in the
7748 journal along with meta data.
7750 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7751 writing short strings to files (for usage for /sys), and for
7752 creating symlinks, character and block device nodes.
7754 * New unit file option ControlGroupPersistent= to make cgroups
7755 persistent, following the mechanisms outlined in
7756 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7758 * Support multiple local RTCs in a sane way
7760 * No longer monopolize IO when replaying readahead data on
7761 rotating disks, since we might starve non-file-system IO to
7762 death, since fanotify() will not see accesses done by blkid,
7765 * Do not show kernel threads in systemd-cgls anymore, unless
7766 requested with new -k switch.
7768 Contributions from: Dan Horák, Kay Sievers, Lennart
7769 Poettering, Michal Schmidt
7773 * This is mostly a test release, but incorporates many
7776 * The git repository moved to:
7777 git://anongit.freedesktop.org/systemd/systemd
7778 ssh://git.freedesktop.org/git/systemd/systemd
7780 * First release with the journal
7781 http://0pointer.de/blog/projects/the-journal.html
7783 * The journal replaces both systemd-kmsg-syslogd and
7784 systemd-stdout-bridge.
7786 * New sd_pid_get_unit() API call in libsystemd-logind
7788 * Many systemadm clean-ups
7790 * Introduce remote-fs-pre.target which is ordered before all
7791 remote mounts and may be used to start services before all
7794 * Added Mageia support
7796 * Add bash completion for systemd-loginctl
7798 * Actively monitor PID file creation for daemons which exit in
7799 the parent process before having finished writing the PID
7800 file in the daemon process. Daemons which do this need to be
7801 fixed (i.e. PID file creation must have finished before the
7802 parent exits), but we now react a bit more gracefully to them.
7804 * Add colourful boot output, mimicking the well-known output
7805 of existing distributions.
7807 * New option PassCredentials= for socket units, for
7808 compatibility with a recent kernel ABI breakage.
7810 * /etc/rc.local is now hooked in via a generator binary, and
7811 thus will no longer act as synchronization point during
7814 * systemctl list-unit-files now supports --root=.
7816 * systemd-tmpfiles now understands two new commands: z, Z for
7817 relabelling files according to the SELinux database. This is
7818 useful to apply SELinux labels to specific files in /sys,
7821 * Output of SysV services is now forwarded to both the console
7822 and the journal by default, not only just the console.
7824 * New man pages for all APIs from libsystemd-login.
7826 * The build tree got reorganized and the build system is a
7827 lot more modular allowing embedded setups to specifically
7828 select the components of systemd they are interested in.
7830 * Support for Linux systems lacking the kernel VT subsystem is
7833 * configure's --with-rootdir= got renamed to
7834 --with-rootprefix= to follow the naming used by udev and
7837 * Unless specified otherwise we will now install to /usr instead
7838 of /usr/local by default.
7840 * Processes with '@' in argv[0][0] are now excluded from the
7841 final shut-down killing spree, following the logic explained
7843 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7845 * All processes remaining in a service cgroup when we enter
7846 the START or START_PRE states are now killed with
7847 SIGKILL. That means it is no longer possible to spawn
7848 background processes from ExecStart= lines (which was never
7849 supported anyway, and bad style).
7851 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7852 reloading of units together.
7854 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7855 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7856 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7857 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7858 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek