1 systemd System and Service Manager
3 CHANGES WITH 237 in spe:
5 * Some keyboards come with a zoom see-saw or rocker which until now got
6 mapped to the Linux "zoomin/out" keys in hwdb. However, these
7 keycodes are not recognized by any major desktop. They now produce
8 Up/Down key events so that they can be used for scrolling.
14 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
15 in v235 has been extended to also set the dummy.ko module option
16 numdummies=0, preventing the kernel from automatically creating
17 dummy0. All dummy interfaces must now be explicitly created.
19 * Unknown '%' specifiers in configuration files are now rejected. This
20 applies to units and tmpfiles.d configuration. Any percent characters
21 that are followed by a letter or digit that are not supposed to be
22 interpreted as the beginning of a specifier should be escaped by
23 doubling ("%%"). (So "size=5%" is still accepted, as well as
24 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
25 valid specifiers today.)
27 * systemd-resolved now maintains a new dynamic
28 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
29 recommended to make /etc/resolv.conf a symlink to it. This file
30 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
31 includes dynamically acquired search domains, achieving more correct
32 DNS resolution by software that bypasses local DNS APIs such as NSS.
34 * The "uaccess" udev tag has been dropped from /dev/kvm and
35 /dev/dri/renderD*. These devices now have the 0666 permissions by
36 default (but this may be changed at build-time). /dev/dri/renderD*
37 will now be owned by the "render" group along with /dev/kfd.
39 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
40 systemd-journal-gatewayd.service and
41 systemd-journal-upload.service. This means "nss-systemd" must be
42 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
43 services are resolved properly.
45 * In /etc/fstab two new mount options are now understood:
46 x-systemd.makefs and x-systemd.growfs. The former has the effect that
47 the configured file system is formatted before it is mounted, the
48 latter that the file system is resized to the full block device size
49 after it is mounted (i.e. if the file system is smaller than the
50 partition it resides on, it's grown). This is similar to the fsck
51 logic in /etc/fstab, and pulls in systemd-makefs@.service and
52 systemd-growfs@.service as necessary, similar to
53 systemd-fsck@.service. Resizing is currently only supported on ext4
56 * In systemd-networkd, the IPv6 RA logic now optionally may announce
57 DNS server and domain information.
59 * Support for the LUKS2 on-disk format for encrypted partitions has
60 been added. This requires libcryptsetup2 during compilation and
63 * The systemd --user instance will now signal "readiness" when its
64 basic.target unit has been reached, instead of when the run queue ran
65 empty for the first time.
67 * Tmpfiles.d with user configuration are now also supported.
68 systemd-tmpfiles gained a new --user switch, and snippets placed in
69 ~/.config/user-tmpfiles.d/ and corresponding directories will be
70 executed by systemd-tmpfiles --user running in the new
71 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
72 running in the user session.
74 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
75 %S resolves to the top-level state directory (/var/lib for the system
76 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
77 top-level cache directory (/var/cache for the system instance,
78 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
79 logs directory (/var/log for the system instance,
80 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
81 existing %t specifier, that resolves to the top-level runtime
82 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
85 * journalctl learnt a new parameter --output-fields= for limiting the
86 set of journal fields to output in verbose and JSON output modes.
88 * systemd-timesyncd's configuration file gained a new option
89 RootDistanceMaxSec= for setting the maximum root distance of servers
90 it'll use, as well as the new options PollIntervalMinSec= and
91 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
93 * bootctl gained a new command "list" for listing all available boot
94 menu items on systems that follow the boot loader specification.
96 * systemctl gained a new --dry-run switch that shows what would be done
97 instead of doing it, and is currently supported by the shutdown and
100 * ConditionSecurity= can now detect the TOMOYO security module.
102 * Unit file [Install] sections are now also respected in unit drop-in
103 files. This is intended to be used by drop-ins under /usr/lib/.
105 * systemd-firstboot may now also set the initial keyboard mapping.
107 * Udev "changed" events for devices which are exposed as systemd
108 .device units are now propagated to units specified in
109 ReloadPropagatedFrom= as reload requests.
111 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
112 unit template name (i.e. a name in the form of 'foobar@.service',
113 without the instance component between the '@' and - the '.'), then
114 the escaped sysfs path of the device is automatically used as the
117 * SystemCallFilter= in unit files has been extended so that an "errno"
118 can be specified individually for each system call. Example:
119 SystemCallFilter=~uname:EILSEQ.
121 * The cgroup delegation logic has been substantially updated. Delegate=
122 now optionally takes a list of controllers (instead of a boolean, as
123 before), which lists the controllers to delegate at least.
125 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
127 * A new LogLevelMax= setting configures the maximum log level any
128 process of the service may log at (i.e. anything with a lesser
129 priority than what is specified is automatically dropped). A new
130 LogExtraFields= setting allows configuration of additional journal
131 fields to attach to all log records generated by any of the unit's
134 * New StandardInputData= and StandardInputText= settings along with the
135 new option StandardInput=data may be used to configure textual or
136 binary data that shall be passed to the executed service process via
137 standard input, encoded in-line in the unit file.
139 * StandardInput=, StandardOutput= and StandardError= may now be used to
140 connect stdin/stdout/stderr of executed processes directly with a
141 file or AF_UNIX socket in the file system, using the new "file:" option.
143 * A new unit file option CollectMode= has been added, that allows
144 tweaking the garbage collection logic for units. It may be used to
145 tell systemd to garbage collect units that have failed automatically
146 (normally it only GCs units that exited successfully). systemd-run
147 and systemd-mount expose this new functionality with a new -G option.
149 * "machinectl bind" may now be used to bind mount non-directories
150 (i.e. regularfiles, devices, fifos, sockets).
152 * systemd-analyze gained a new verb "calendar" for validating and
153 testing calendar time specifications to use for OnCalendar= in timer
154 units. Besides validating the expression it will calculate the next
155 time the specified expression would elapse.
157 * In addition to the pre-existing FailureAction= unit file setting
158 there's now SuccessAction=, for configuring a shutdown action to
159 execute when a unit completes successfully. This is useful in
160 particular inside containers that shall terminate after some workload
161 has been completed. Also, both options are now supported for all unit
162 types, not just services.
164 * networkds's IP rule support gained two new options
165 IncomingInterface= and OutgoingInterface= for configuring the incoming
166 and outgoing interfaces of configured rules. systemd-networkd also
167 gained support for "vxcan" network devices.
169 * networkd gained a new setting RequiredForOnline=, taking a
170 boolean. If set, systemd-wait-online will take it into consideration
171 when determining that the system is up, otherwise it will ignore the
172 interface for this purpose.
174 * The sd_notify() protocol gained support for a new operation: with
175 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
176 store again, ahead of POLLHUP or POLLERR when they are removed
179 * A new document UIDS-GIDS.md has been added to the source tree, that
180 documents the UID/GID range and assignment assumptions and
181 requirements of systemd.
183 * The watchdog device PID 1 will ping may now be configured through the
184 WatchdogDevice= configuration file setting, or by setting the
185 systemd.watchdog_service= kernel commandline option.
187 * systemd-resolved's gained support for registering DNS-SD services on
188 the local network using MulticastDNS. Services may either be
189 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
190 the same dir below /run, /usr/lib), or through its D-Bus API.
192 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
193 extend the effective start, runtime, and stop time. The service must
194 continue to send EXTEND_TIMEOUT_USEC within the period specified to
195 prevent the service manager from making the service as timedout.
197 * systemd-resolved's DNSSEC support gained support for RFC 8080
198 (Ed25519 keys and signatures).
200 * The systemd-resolve command line tool gained a new set of options
201 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
202 --set-nta= and --revert to configure per-interface DNS configuration
203 dynamically during runtime. It's useful for pushing DNS information
204 into systemd-resolved from DNS hook scripts that various interface
205 managing software supports (such as pppd).
207 * systemd-nspawn gained a new --network-namespace-path= command line
208 option, which may be used to make a container join an existing
209 network namespace, by specifying a path to a "netns" file.
211 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
212 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
213 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
214 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
215 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
216 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
217 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
218 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
219 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
220 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
221 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
222 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
223 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
224 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
225 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
226 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
227 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
228 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
229 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
230 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
231 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
232 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
233 Jędrzejewski-Szmek, Zeal Jagannatha
239 * INCOMPATIBILITY: systemd-logind.service and other long-running
240 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
241 communication with the outside. This generally improves security of
242 the system, and is in almost all cases a safe and good choice, as
243 these services do not and should not provide any network-facing
244 functionality. However, systemd-logind uses the glibc NSS API to
245 query the user database. This creates problems on systems where NSS
246 is set up to directly consult network services for user database
247 lookups. In particular, this creates incompatibilities with the
248 "nss-nis" module, which attempts to directly contact the NIS/YP
249 network servers it is configured for, and will now consistently
250 fail. In such cases, it is possible to turn off IP sandboxing for
251 systemd-logind.service (set IPAddressDeny= in its [Service] section
252 to the empty string, via a .d/ unit file drop-in). Downstream
253 distributions might want to update their nss-nis packaging to include
254 such a drop-in snippet, accordingly, to hide this incompatibility
255 from the user. Another option is to make use of glibc's nscd service
256 to proxy such network requests through a privilege-separated, minimal
257 local caching daemon, or to switch to more modern technologies such
258 sssd, whose NSS hook-ups generally do not involve direct network
259 access. In general, we think it's definitely time to question the
260 implementation choices of nss-nis, i.e. whether it's a good idea
261 today to embed a network-facing loadable module into all local
262 processes that need to query the user database, including the most
263 trivial and benign ones, such as "ls". For more details about
264 IPAddressDeny= see below.
266 * A new modprobe.d drop-in is now shipped by default that sets the
267 bonding module option max_bonds=0. This overrides the kernel default,
268 to avoid conflicts and ambiguity as to whether or not bond0 should be
269 managed by systemd-networkd or not. This resolves multiple issues
270 with bond0 properties not being applied, when bond0 is configured
271 with systemd-networkd. Distributors may choose to not package this,
272 however in that case users will be prevented from correctly managing
273 bond0 interface using systemd-networkd.
275 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
276 which print the logging level and target of the system manager. They
277 complement the existing "set-log-level" and "set-log-target" verbs
278 used to change those values.
280 * journald.conf gained a new boolean setting ReadKMsg= which defaults
281 to on. If turned off kernel log messages will not be read by
282 systemd-journald or included in the logs. It also gained a new
283 setting LineMax= for configuring the maximum line length in
284 STDOUT/STDERR log streams. The new default for this value is 48K, up
285 from the previous hardcoded 2048.
287 * A new unit setting RuntimeDirectoryPreserve= has been added, which
288 allows more detailed control of what to do with a runtime directory
289 configured with RuntimeDirectory= (i.e. a directory below /run or
290 $XDG_RUNTIME_DIR) after a unit is stopped.
292 * The RuntimeDirectory= setting for units gained support for creating
293 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
294 one top-level directory.
296 * Units gained new options StateDirectory=, CacheDirectory=,
297 LogsDirectory= and ConfigurationDirectory= which are closely related
298 to RuntimeDirectory= but manage per-service directories below
299 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
300 possible to write unit files which when activated automatically gain
301 properly owned service specific directories in these locations, thus
302 making unit files self-contained and increasing compatibility with
303 stateless systems and factory reset where /etc or /var are
304 unpopulated at boot. Matching these new settings there's also
305 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
306 ConfigurationDirectoryMode= for configuring the access mode of these
307 directories. These settings are particularly useful in combination
308 with DynamicUser=yes as they provide secure, properly-owned,
309 writable, and stateful locations for storage, excluded from the
310 sandbox that such services live in otherwise.
312 * Automake support has been removed from this release. systemd is now
315 * systemd-journald will now aggressively cache client metadata during
316 runtime, speeding up log write performance under pressure. This comes
317 at a small price though: as much of the metadata is read
318 asynchronously from /proc/ (and isn't implicitly attached to log
319 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
320 metadata stored alongside a log entry might be slightly
321 out-of-date. Previously it could only be slightly newer than the log
322 message. The time window is small however, and given that the kernel
323 is unlikely to be improved anytime soon in this regard, this appears
326 * nss-myhostname/systemd-resolved will now by default synthesize an
327 A/AAAA resource record for the "_gateway" hostname, pointing to the
328 current default IP gateway. Previously it did that for the "gateway"
329 name, hampering adoption, as some distributions wanted to leave that
330 host name open for local use. The old behaviour may still be
331 requested at build time.
333 * systemd-networkd's [Address] section in .network files gained a new
334 Scope= setting for configuring the IP address scope. The [Network]
335 section gained a new boolean setting ConfigureWithoutCarrier= that
336 tells systemd-networkd to ignore link sensing when configuring the
337 device. The [DHCP] section gained a new Anonymize= boolean option for
338 turning on a number of options suggested in RFC 7844. A new
339 [RoutingPolicyRule] section has been added for configuring the IP
340 routing policy. The [Route] section has gained support for a new
341 Type= setting which permits configuring
342 blackhole/unreachable/prohibit routes.
344 * The [VRF] section in .netdev files gained a new Table= setting for
345 configuring the routing table to use. The [Tunnel] section gained a
346 new Independent= boolean field for configuring tunnels independent of
347 an underlying network interface. The [Bridge] section gained a new
348 GroupForwardMask= option for configuration of propagation of link
349 local frames between bridge ports.
351 * The WakeOnLan= setting in .link files gained support for a number of
352 new modes. A new TCP6SegmentationOffload= setting has been added for
353 configuring TCP/IPv6 hardware segmentation offload.
355 * The IPv6 RA sender implementation may now optionally send out RDNSS
356 and RDNSSL records to supply DNS configuration to peers.
358 * systemd-nspawn gained support for a new --system-call-filter= command
359 line option for adding and removing entries in the default system
360 call filter it applies. Moreover systemd-nspawn has been changed to
361 implement a system call whitelist instead of a blacklist.
363 * systemd-run gained support for a new --pipe command line option. If
364 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
365 are directly passed on to the activated transient service
366 executable. This allows invoking arbitrary processes as systemd
367 services (for example to take benefit of dependency management,
368 accounting management, resource management or log management that is
369 done automatically for services) — while still allowing them to be
370 integrated in a classic UNIX shell pipeline.
372 * When a service sends RELOAD=1 via sd_notify() and reload propagation
373 using ReloadPropagationTo= is configured, a reload is now propagated
374 to configured units. (Previously this was only done on explicitly
375 requested reloads, using "systemctl reload" or an equivalent
378 * For each service unit a restart counter is now kept: it is increased
379 each time the service is restarted due to Restart=, and may be
380 queried using "systemctl show -p NRestarts …".
382 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
383 @signal and @timer have been added, for usage with SystemCallFilter=
384 in unit files and the new --system-call-filter= command line option
385 of systemd-nspawn (see above).
387 * ExecStart= lines in unit files gained two new modifiers: when a
388 command line is prefixed with "!" the command will be executed as
389 configured, except for the credentials applied by
390 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
391 "+", but does still apply namespacing options unlike "+". There's
392 also "!!" now, which is mostly identical, but becomes a NOP on
393 systems that support ambient capabilities. This is useful to write
394 unit files that work with ambient capabilities where possible but
395 automatically fall back to traditional privilege dropping mechanisms
396 on systems where this is not supported.
398 * ListenNetlink= settings in socket units now support RDMA netlink
401 * A new unit file setting LockPersonality= has been added which permits
402 locking down the chosen execution domain ("personality") of a service
405 * A new special target "getty-pre.target" has been added, which is
406 ordered before all text logins, and may be used to order services
407 before textual logins acquire access to the console.
409 * systemd will now attempt to load the virtio-rng.ko kernel module very
410 early on if a VM environment supporting this is detected. This should
411 improve entropy during early boot in virtualized environments.
413 * A _netdev option is now supported in /etc/crypttab that operates in a
414 similar way as the same option in /etc/fstab: it permits configuring
415 encrypted devices that need to be ordered after the network is up.
416 Following this logic, two new special targets
417 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
418 added that are to cryptsetup.target what remote-fs.target and
419 remote-fs-pre.target are to local-fs.target.
421 * Service units gained a new UnsetEnvironment= setting which permits
422 unsetting specific environment variables for services that are
423 normally passed to it (for example in order to mask out locale
424 settings for specific services that can't deal with it).
426 * Units acquired a new boolean option IPAccounting=. When turned on, IP
427 traffic accounting (packet count as well as byte count) is done for
428 the service, and shown as part of "systemctl status" or "systemd-run
431 * Service units acquired two new options IPAddressAllow= and
432 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
433 for configuring a simple IP access control list for all sockets of
434 the unit. These options are available also on .slice and .socket
435 units, permitting flexible access list configuration for individual
436 services as well as groups of services (as defined by a slice unit),
437 including system-wide. Note that IP ACLs configured this way are
438 enforced on every single IPv4 and IPv6 socket created by any process
439 of the service unit, and apply to ingress as well as egress traffic.
441 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
442 structured log message is generated each time the unit is stopped,
443 containing information about the consumed resources of this
446 * A new setting KeyringMode= has been added to unit files, which may be
447 used to control how the kernel keyring is set up for executed
450 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
451 "systemctl kexec" and "systemctl exit" are now always asynchronous in
452 behaviour (that is: these commands return immediately after the
453 operation was enqueued instead of waiting for the operation to
454 complete). Previously, "systemctl poweroff" and "systemctl reboot"
455 were asynchronous on systems using systemd-logind (i.e. almost
456 always, and like they were on sysvinit), and the other three commands
457 were unconditionally synchronous. With this release this is cleaned
458 up, and callers will see the same asynchronous behaviour on all
459 systems for all five operations.
461 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
464 * .timer units now accept calendar specifications in other timezones
465 than UTC or the local timezone.
467 * The tmpfiles snippet var.conf has been changed to create
468 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
469 the "utmp" group already, and it appears to be generally understood
470 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
471 databases. Previously this was implemented correctly for all these
472 databases excepts btmp, which has been opened up like this now
473 too. Note that while the other databases are world-readable
474 (i.e. 0644), btmp is not and remains more restrictive.
476 * The systemd-resolve tool gained a new --reset-server-features
477 switch. When invoked like this systemd-resolved will forget
478 everything it learnt about the features supported by the configured
479 upstream DNS servers, and restarts the feature probing logic on the
480 next resolver look-up for them at the highest feature level
483 * The status dump systemd-resolved sends to the logs upon receiving
484 SIGUSR1 now also includes information about all DNS servers it is
485 configured to use, and the features levels it probed for them.
487 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
488 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
489 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
490 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
491 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
492 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
493 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
494 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
495 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
496 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
497 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
498 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
499 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
500 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
501 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
502 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
503 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
504 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
505 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
506 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
512 * Meson is now supported as build system in addition to Automake. It is
513 our plan to remove Automake in one of our next releases, so that
514 Meson becomes our exclusive build system. Hence, please start using
515 the Meson build system in your downstream packaging. There's plenty
516 of documentation around how to use Meson, the extremely brief
519 ./autogen.sh && ./configure && make && sudo make install
523 meson build && ninja -C build && sudo ninja -C build install
525 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
526 which permits configuring a timeout on the time a job is
527 running. This is particularly useful for setting timeouts on jobs for
530 * Unit files gained two new options ConditionUser= and ConditionGroup=
531 for conditionalizing units based on the identity of the user/group
532 running a systemd user instance.
534 * systemd-networkd now understands a new FlowLabel= setting in the
535 [VXLAN] section of .network files, as well as a Priority= in
536 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
537 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
538 gained support for configuration of GENEVE links, and IPv6 address
539 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
541 * .link files now understand a new Port= setting.
543 * systemd-networkd's DHCP support gained support for DHCP option 119
544 (domain search list).
546 * systemd-networkd gained support for serving IPv6 address ranges using
547 the Router Advertisment protocol. The new .network configuration
548 section [IPv6Prefix] may be used to configure the ranges to
549 serve. This is implemented based on a new, minimal, native server
550 implementation of RA.
552 * journalctl's --output= switch gained support for a new parameter
553 "short-iso-precise" for a mode where timestamps are shown as precise
556 * systemd-udevd's "net_id" builtin may now generate stable network
557 interface names from IBM PowerVM VIO devices as well as ACPI platform
560 * MulticastDNS support in systemd-resolved may now be explicitly
561 enabled/disabled using the new MulticastDNS= configuration file
564 * systemd-resolved may now optionally use libidn2 instead of the libidn
565 for processing internationalized domain names. Support for libidn2
566 should be considered experimental and should not be enabled by
569 * "machinectl pull-tar" and related call may now do verification of
570 downloaded images using SUSE-style .sha256 checksum files in addition
571 to the already existing support for validating using Ubuntu-style
574 * sd-bus gained support for a new sd_bus_message_appendv() call which
575 is va_list equivalent of sd_bus_message_append().
577 * sd-boot gained support for validating images using SHIM/MOK.
579 * The SMACK code learnt support for "onlycap".
581 * systemd-mount --umount is now much smarter in figuring out how to
582 properly unmount a device given its mount or device path.
584 * The code to call libnss_dns as a fallback from libnss_resolve when
585 the communication with systemd-resolved fails was removed. This
586 fallback was redundant and interfered with the [!UNAVAIL=return]
587 suffix. See nss-resolve(8) for the recommended configuration.
589 * systemd-logind may now be restarted without losing state. It stores
590 the file descriptors for devices it manages in the system manager
591 using the FDSTORE= mechanism. Please note that further changes in
592 other components may be required to make use of this (for example
593 Xorg has code to listen for stops of systemd-logind and terminate
594 itself when logind is stopped or restarted, in order to avoid using
595 stale file descriptors for graphical devices, which is now
596 counterproductive and must be reverted in order for restarts of
597 systemd-logind to be safe. See
598 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
600 * All kernel install plugins are called with the environment variable
601 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
602 /etc/machine-id. If the file is missing or empty, the variable is
603 empty and BOOT_DIR_ABS is the path of a temporary directory which is
604 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
605 is empty, all plugins should not put anything in BOOT_DIR_ABS.
607 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
608 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
609 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
610 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
611 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
612 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
613 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
614 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
615 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
616 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
617 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
618 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
619 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
620 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
621 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
622 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
623 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
624 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
625 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
626 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
627 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
628 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
629 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
630 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
631 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
632 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
633 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
634 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
635 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
642 * The "hybrid" control group mode has been modified to improve
643 compatibility with "legacy" cgroups-v1 setups. Specifically, the
644 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
645 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
646 cgroups-v1 hierarchy), the only externally visible change being that
647 the cgroups-v2 hierarchy is also mounted, to
648 /sys/fs/cgroup/unified. This should provide a large degree of
649 compatibility with "legacy" cgroups-v1, while taking benefit of the
650 better management capabilities of cgroups-v2.
652 * The default control group setup mode may be selected both a boot-time
653 via a set of kernel command line parameters (specifically:
654 systemd.unified_cgroup_hierarchy= and
655 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
656 default selected on the configure command line
657 (--with-default-hierarchy=). The upstream default is "hybrid"
658 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
659 this will change in a future systemd version to be "unified" (pure
660 cgroups-v2 mode). The third option for the compile time option is
661 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
662 distributions to default to "hybrid" mode for release distributions,
663 starting with v233. We recommend "unified" for development
664 distributions (specifically: distributions such as Fedora's rawhide)
665 as that's where things are headed in the long run. Use "legacy" for
666 greatest stability and compatibility only.
668 * Note one current limitation of "unified" and "hybrid" control group
669 setup modes: the kernel currently does not permit the systemd --user
670 instance (i.e. unprivileged code) to migrate processes between two
671 disconnected cgroup subtrees, even if both are managed and owned by
672 the user. This effectively means "systemd-run --user --scope" doesn't
673 work when invoked from outside of any "systemd --user" service or
674 scope. Specifically, it is not supported from session scopes. We are
675 working on fixing this in a future systemd version. (See #3388 for
676 further details about this.)
678 * DBus policy files are now installed into /usr rather than /etc. Make
679 sure your system has dbus >= 1.9.18 running before upgrading to this
680 version, or override the install path with --with-dbuspolicydir= .
682 * All python scripts shipped with systemd (specifically: the various
683 tests written in Python) now require Python 3.
685 * systemd unit tests can now run standalone (without the source or
686 build directories), and can be installed into /usr/lib/systemd/tests/
687 with 'make install-tests'.
689 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
690 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
693 * Support for the %c, %r, %R specifiers in unit files has been
694 removed. Specifiers are not supposed to be dependent on configuration
695 in the unit file itself (so that they resolve the same regardless
696 where used in the unit files), but these specifiers were influenced
697 by the Slice= option.
699 * The shell invoked by debug-shell.service now defaults to /bin/sh in
700 all cases. If distributions want to use a different shell for this
701 purpose (for example Fedora's /sbin/sushell) they need to specify
702 this explicitly at configure time using --with-debug-shell=.
704 * The confirmation spawn prompt has been reworked to offer the
707 (c)ontinue, proceed without asking anymore
708 (D)ump, show the state of the unit
709 (f)ail, don't execute the command and pretend it failed
711 (i)nfo, show a short summary of the unit
712 (j)obs, show jobs that are in progress
713 (s)kip, don't execute the command and pretend it succeeded
714 (y)es, execute the command
716 The 'n' choice for the confirmation spawn prompt has been removed,
717 because its meaning was confusing.
719 The prompt may now also be redirected to an alternative console by
720 specifying the console as parameter to systemd.confirm_spawn=.
722 * Services of Type=notify require a READY=1 notification to be sent
723 during startup. If no such message is sent, the service now fails,
724 even if the main process exited with a successful exit code.
726 * Services that fail to start up correctly now always have their
727 ExecStopPost= commands executed. Previously, they'd enter "failed"
728 state directly, without executing these commands.
730 * The option MulticastDNS= of network configuration files has acquired
731 an actual implementation. With MulticastDNS=yes a host can resolve
732 names of remote hosts and reply to mDNS A and AAAA requests.
734 * When units are about to be started an additional check is now done to
735 ensure that all dependencies of type BindsTo= (when used in
736 combination with After=) have been started.
738 * systemd-analyze gained a new verb "syscall-filter" which shows which
739 system call groups are defined for the SystemCallFilter= unit file
740 setting, and which system calls they contain.
742 * A new system call filter group "@filesystem" has been added,
743 consisting of various file system related system calls. Group
744 "@reboot" has been added, covering reboot, kexec and shutdown related
745 calls. Finally, group "@swap" has been added covering swap
746 configuration related calls.
748 * A new unit file option RestrictNamespaces= has been added that may be
749 used to restrict access to the various process namespace types the
750 Linux kernel provides. Specifically, it may be used to take away the
751 right for a service unit to create additional file system, network,
752 user, and other namespaces. This sandboxing option is particularly
753 relevant due to the high amount of recently discovered namespacing
754 related vulnerabilities in the kernel.
756 * systemd-udev's .link files gained support for a new AutoNegotiation=
757 setting for configuring Ethernet auto-negotiation.
759 * systemd-networkd's .network files gained support for a new
760 ListenPort= setting in the [DHCP] section to explicitly configure the
761 UDP client port the DHCP client shall listen on.
763 * .network files gained a new Unmanaged= boolean setting for explicitly
764 excluding one or more interfaces from management by systemd-networkd.
766 * The systemd-networkd ProxyARP= option has been renamed to
767 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
768 renamed to ReduceARPProxy=. The old names continue to be available
771 * systemd-networkd gained support for configuring IPv6 Proxy NDP
772 addresses via the new IPv6ProxyNDPAddress= .network file setting.
774 * systemd-networkd's bonding device support gained support for two new
775 configuration options ActiveSlave= and PrimarySlave=.
777 * The various options in the [Match] section of .network files gained
778 support for negative matching.
780 * New systemd-specific mount options are now understood in /etc/fstab:
782 x-systemd.mount-timeout= may be used to configure the maximum
783 permitted runtime of the mount command.
785 x-systemd.device-bound may be set to bind a mount point to its
786 backing device unit, in order to automatically remove a mount point
787 if its backing device is unplugged. This option may also be
788 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
789 on the block device, which is now automatically set for all CDROM
790 drives, so that mounted CDs are automatically unmounted when they are
791 removed from the drive.
793 x-systemd.after= and x-systemd.before= may be used to explicitly
794 order a mount after or before another unit or mount point.
796 * Enqueued start jobs for device units are now automatically garbage
797 collected if there are no jobs waiting for them anymore.
799 * systemctl list-jobs gained two new switches: with --after, for every
800 queued job the jobs it's waiting for are shown; with --before the
801 jobs which it's blocking are shown.
803 * systemd-nspawn gained support for ephemeral boots from disk images
804 (or in other words: --ephemeral and --image= may now be
805 combined). Moreover, ephemeral boots are now supported for normal
806 directories, even if the backing file system is not btrfs. Of course,
807 if the file system does not support file system snapshots or
808 reflinks, the initial copy operation will be relatively expensive, but
809 this should still be suitable for many use cases.
811 * Calendar time specifications in .timer units now support
812 specifications relative to the end of a month by using "~" instead of
813 "-" as separator between month and day. For example, "*-02~03" means
814 "the third last day in February". In addition a new syntax for
815 repeated events has been added using the "/" character. For example,
816 "9..17/2:00" means "every two hours from 9am to 5pm".
818 * systemd-socket-proxyd gained a new parameter --connections-max= for
819 configuring the maximum number of concurrent connections.
821 * sd-id128 gained a new API for generating unique IDs for the host in a
822 way that does not leak the machine ID. Specifically,
823 sd_id128_get_machine_app_specific() derives an ID based on the
824 machine ID a in well-defined, non-reversible, stable way. This is
825 useful whenever an identifier for the host is needed but where the
826 identifier shall not be useful to identify the system beyond the
827 scope of the application itself. (Internally this uses HMAC-SHA256 as
828 keyed hash function using the machine ID as input.)
830 * NotifyAccess= gained a new supported value "exec". When set
831 notifications are accepted from all processes systemd itself invoked,
832 including all control processes.
834 * .nspawn files gained support for defining overlay mounts using the
835 Overlay= and OverlayReadOnly= options. Previously this functionality
836 was only available on the systemd-nspawn command line.
838 * systemd-nspawn's --bind= and --overlay= options gained support for
839 bind/overlay mounts whose source lies within the container tree by
840 prefixing the source path with "+".
842 * systemd-nspawn's --bind= and --overlay= options gained support for
843 automatically allocating a temporary source directory in /var/tmp
844 that is removed when the container dies. Specifically, if the source
845 directory is specified as empty string this mechanism is selected. An
846 example usage is --overlay=+/var::/var, which creates an overlay
847 mount based on the original /var contained in the image, overlayed
848 with a temporary directory in the host's /var/tmp. This way changes
849 to /var are automatically flushed when the container shuts down.
851 * systemd-nspawn --image= option does now permit raw file system block
852 devices (in addition to images containing partition tables, as
855 * The disk image dissection logic in systemd-nspawn gained support for
856 automatically setting up LUKS encrypted as well as Verity protected
857 partitions. When a container is booted from an encrypted image the
858 passphrase is queried at start-up time. When a container with Verity
859 data is started, the root hash is search in a ".roothash" file
860 accompanying the disk image (alternatively, pass the root hash via
861 the new --root-hash= command line option).
863 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
864 be used to dissect disk images the same way as systemd-nspawn does
865 it, following the Bootable Partition Specification. It may even be
866 used to mount disk images with complex partition setups (including
867 LUKS and Verity partitions) to a local host directory, in order to
868 inspect them. This tool is not considered public API (yet), and is
869 thus not installed into /usr/bin. Please do not rely on its
870 existence, since it might go away or be changed in later systemd
873 * A new generator "systemd-verity-generator" has been added, similar in
874 style to "systemd-cryptsetup-generator", permitting automatic setup of
875 Verity root partitions when systemd boots up. In order to make use of
876 this your partition setup should follow the Discoverable Partitions
877 Specification, and the GPT partition ID of the root file system
878 partition should be identical to the upper 128bit of the Verity root
879 hash. The GPT partition ID of the Verity partition protecting it
880 should be the lower 128bit of the Verity root hash. If the partition
881 image follows this model it is sufficient to specify a single
882 "roothash=" kernel command line argument to both configure which root
883 image and verity partition to use as well as the root hash for
884 it. Note that systemd-nspawn's Verity support follows the same
885 semantics, meaning that disk images with proper Verity data in place
886 may be booted in containers with systemd-nspawn as well as on
887 physical systems via the verity generator. Also note that the "mkosi"
888 tool available at https://github.com/systemd/mkosi has been updated
889 to generate Verity protected disk images following this scheme. In
890 fact, it has been updated to generate disk images that optionally
891 implement a complete UEFI SecureBoot trust chain, involving a signed
892 kernel and initrd image that incorporates such a root hash as well as
893 a Verity-enabled root partition.
895 * The hardware database (hwdb) udev supports has been updated to carry
896 accelerometer quirks.
898 * All system services are now run with a fresh kernel keyring set up
899 for them. The invocation ID is stored by default in it, thus
900 providing a safe, non-overridable way to determine the invocation
903 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
904 options for bind mounting arbitrary paths in a service-specific
905 way. When these options are used, arbitrary host or service files and
906 directories may be mounted to arbitrary locations in the service's
909 * Documentation has been added that lists all of systemd's low-level
910 environment variables:
912 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
914 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
915 whether a specific socket file descriptor matches a specified socket
918 * systemd-firstboot has been updated to check for the
919 systemd.firstboot= kernel command line option. It accepts a boolean
920 and when set to false the first boot questions are skipped.
922 * systemd-fstab-generator has been updated to check for the
923 systemd.volatile= kernel command line option, which either takes an
924 optional boolean parameter or the special value "state". If used the
925 system may be booted in a "volatile" boot mode. Specifically,
926 "systemd.volatile" is used, the root directory will be mounted as
927 tmpfs, and only /usr is mounted from the actual root file system. If
928 "systemd.volatile=state" is used, the root directory will be mounted
929 as usual, but /var is mounted as tmpfs. This concept provides similar
930 functionality as systemd-nspawn's --volatile= option, but provides it
931 on physical boots. Use this option for implementing stateless
932 systems, or testing systems with all state and/or configuration reset
933 to the defaults. (Note though that many distributions are not
934 prepared to boot up without a populated /etc or /var, though.)
936 * systemd-gpt-auto-generator gained support for LUKS encrypted root
937 partitions. Previously it only supported LUKS encrypted partitions
938 for all other uses, except for the root partition itself.
940 * Socket units gained support for listening on AF_VSOCK sockets for
941 communication in virtualized QEMU environments.
943 * The "configure" script gained a new option --with-fallback-hostname=
944 for specifying the fallback hostname to use if none is configured in
945 /etc/hostname. For example, by specifying
946 --with-fallback-hostname=fedora it is possible to default to a
947 hostname of "fedora" on pristine installations.
949 * systemd-cgls gained support for a new --unit= switch for listing only
950 the control groups of a specific unit. Similar --user-unit= has been
951 added for listing only the control groups of a specific user unit.
953 * systemd-mount gained a new --umount switch for unmounting a mount or
954 automount point (and all mount/automount points below it).
956 * systemd will now refuse full configuration reloads (via systemctl
957 daemon-reload and related calls) unless at least 16MiB of free space
958 are available in /run. This is a safety precaution in order to ensure
959 that generators can safely operate after the reload completed.
961 * A new unit file option RootImage= has been added, which has a similar
962 effect as RootDirectory= but mounts the service's root directory from
963 a disk image instead of plain directory. This logic reuses the same
964 image dissection and mount logic that systemd-nspawn already uses,
965 and hence supports any disk images systemd-nspawn supports, including
966 those following the Discoverable Partition Specification, as well as
967 Verity enabled images. This option enables systemd to run system
968 services directly off disk images acting as resource bundles,
969 possibly even including full integrity data.
971 * A new MountAPIVFS= unit file option has been added, taking a boolean
972 argument. If enabled /proc, /sys and /dev (collectively called the
973 "API VFS") will be mounted for the service. This is only relevant if
974 RootDirectory= or RootImage= is used for the service, as these mounts
975 are of course in place in the host mount namespace anyway.
977 * systemd-nspawn gained support for a new --pivot-root= switch. If
978 specified the root directory within the container image is pivoted to
979 the specified mount point, while the original root disk is moved to a
980 different place. This option enables booting of ostree images
981 directly with systemd-nspawn.
983 * The systemd build scripts will no longer complain if the NTP server
984 addresses are not changed from the defaults. Google now supports
985 these NTP servers officially. We still recommend downstreams to
986 properly register an NTP pool with the NTP pool project though.
988 * coredumpctl gained a new "--reverse" option for printing the list
989 of coredumps in reverse order.
991 * coredumpctl will now show additional information about truncated and
992 inaccessible coredumps, as well as coredumps that are still being
993 processed. It also gained a new --quiet switch for suppressing
994 additional informational message in its output.
996 * coredumpctl gained support for only showing coredumps newer and/or
997 older than specific timestamps, using the new --since= and --until=
998 options, reminiscent of journalctl's options by the same name.
1000 * The systemd-coredump logic has been improved so that it may be reused
1001 to collect backtraces in non-compiled languages, for example in
1002 scripting languages such as Python.
1004 * machinectl will now show the UID shift of local containers, if user
1005 namespacing is enabled for them.
1007 * systemd will now optionally run "environment generator" binaries at
1008 configuration load time. They may be used to add environment
1009 variables to the environment block passed to services invoked. One
1010 user environment generator is shipped by default that sets up
1011 environment variables based on files dropped into /etc/environment.d
1012 and ~/.config/environment.d/.
1014 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1017 * hostnamed has been updated to report a new chassis type of
1018 "convertible" to cover "foldable" laptops that can both act as a
1019 tablet and as a laptop, such as various Lenovo Yoga devices.
1021 Contributions from: Adrián López, Alexander Galanin, Alexander
1022 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1023 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1024 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1025 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1026 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1027 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1028 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1029 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1030 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1031 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1032 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1033 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1034 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1035 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1036 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1037 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1038 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1039 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1040 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1041 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1042 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1043 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1044 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1045 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1046 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1047 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1048 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1051 — Berlin, 2017-03-01
1055 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1056 RestrictAddressFamilies= enabled. These sandboxing options should
1057 generally be compatible with the various external udev call-out
1058 binaries we are aware of, however there may be exceptions, in
1059 particular when exotic languages for these call-outs are used. In
1060 this case, consider turning off these settings locally.
1062 * The new RemoveIPC= option can be used to remove IPC objects owned by
1063 the user or group of a service when that service exits.
1065 * The new ProtectKernelModules= option can be used to disable explicit
1066 load and unload operations of kernel modules by a service. In
1067 addition access to /usr/lib/modules is removed if this option is set.
1069 * ProtectSystem= option gained a new value "strict", which causes the
1070 whole file system tree with the exception of /dev, /proc, and /sys,
1071 to be remounted read-only for a service.
1073 * The new ProtectKernelTunables= option can be used to disable
1074 modification of configuration files in /sys and /proc by a service.
1075 Various directories and files are remounted read-only, so access is
1076 restricted even if the file permissions would allow it.
1078 * The new ProtectControlGroups= option can be used to disable write
1079 access by a service to /sys/fs/cgroup.
1081 * Various systemd services have been hardened with
1082 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1083 RestrictAddressFamilies=.
1085 * Support for dynamically creating users for the lifetime of a service
1086 has been added. If DynamicUser=yes is specified, user and group IDs
1087 will be allocated from the range 61184..65519 for the lifetime of the
1088 service. They can be resolved using the new nss-systemd.so NSS
1089 module. The module must be enabled in /etc/nsswitch.conf. Services
1090 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1091 any resources allocated by the service will be cleaned up when the
1092 service exits. They also have ProtectHome=read-only and
1093 ProtectSystem=strict enabled, so they are not able to make any
1094 permanent modifications to the system.
1096 * The nss-systemd module also always resolves root and nobody, making
1097 it possible to have no /etc/passwd or /etc/group files in minimal
1098 container or chroot environments.
1100 * Services may be started with their own user namespace using the new
1101 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1102 under which the service is running are mapped. All other users are
1105 * Support for the cgroup namespace has been added to systemd-nspawn. If
1106 supported by kernel, the container system started by systemd-nspawn
1107 will have its own view of the cgroup hierarchy. This new behaviour
1108 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1110 * The new MemorySwapMax= option can be used to limit the maximum swap
1111 usage under the unified cgroup hierarchy.
1113 * Support for the CPU controller in the unified cgroup hierarchy has
1114 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1115 options. This controller requires out-of-tree patches for the kernel
1116 and the support is provisional.
1118 * Mount and automount units may now be created transiently
1119 (i.e. dynamically at runtime via the bus API, instead of requiring
1120 unit files in the file system).
1122 * systemd-mount is a new tool which may mount file systems – much like
1123 mount(8), optionally pulling in additional dependencies through
1124 transient .mount and .automount units. For example, this tool
1125 automatically runs fsck on a backing block device before mounting,
1126 and allows the automount logic to be used dynamically from the
1127 command line for establishing mount points. This tool is particularly
1128 useful when dealing with removable media, as it will ensure fsck is
1129 run – if necessary – before the first access and that the file system
1130 is quickly unmounted after each access by utilizing the automount
1131 logic. This maximizes the chance that the file system on the
1132 removable media stays in a clean state, and if it isn't in a clean
1133 state is fixed automatically.
1135 * LazyUnmount=yes option for mount units has been added to expose the
1136 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1139 * /efi will be used as the mount point of the EFI boot partition, if
1140 the directory is present, and the mount point was not configured
1141 through other means (e.g. fstab). If /efi directory does not exist,
1142 /boot will be used as before. This makes it easier to automatically
1143 mount the EFI partition on systems where /boot is used for something
1146 * When operating on GPT disk images for containers, systemd-nspawn will
1147 now mount the ESP to /boot or /efi according to the same rules as PID
1148 1 running on a host. This allows tools like "bootctl" to operate
1149 correctly within such containers, in order to make container images
1150 bootable on physical systems.
1152 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1154 * Two new user session targets have been added to support running
1155 graphical sessions under the systemd --user instance:
1156 graphical-session.target and graphical-session-pre.target. See
1157 systemd.special(7) for a description of how those targets should be
1160 * The vconsole initialization code has been significantly reworked to
1161 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1162 support unicode keymaps. Font and keymap configuration will now be
1163 copied to all allocated virtual consoles.
1165 * FreeBSD's bhyve virtualization is now detected.
1167 * Information recorded in the journal for core dumps now includes the
1168 contents of /proc/mountinfo and the command line of the process at
1169 the top of the process hierarchy (which is usually the init process
1172 * systemd-journal-gatewayd learned the --directory= option to serve
1173 files from the specified location.
1175 * journalctl --root=… can be used to peruse the journal in the
1176 /var/log/ directories inside of a container tree. This is similar to
1177 the existing --machine= option, but does not require the container to
1180 * The hardware database has been extended to support
1181 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1184 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1185 specify the click rate for mice which include a horizontal wheel with
1186 a click rate that is different than the one for the vertical wheel.
1188 * systemd-run gained a new --wait option that makes service execution
1189 synchronous. (Specifically, the command will not return until the
1190 specified service binary exited.)
1192 * systemctl gained a new --wait option that causes the start command to
1193 wait until the units being started have terminated again.
1195 * A new journal output mode "short-full" has been added which displays
1196 timestamps with abbreviated English day names and adds a timezone
1197 suffix. Those timestamps include more information than the default
1198 "short" output mode, and can be passed directly to journalctl's
1199 --since= and --until= options.
1201 * /etc/resolv.conf will be bind-mounted into containers started by
1202 systemd-nspawn, if possible, so any changes to resolv.conf contents
1203 are automatically propagated to the container.
1205 * The number of instances for socket-activated services originating
1206 from a single IP address can be limited with
1207 MaxConnectionsPerSource=, extending the existing setting of
1210 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1213 * .netdev and .network configuration can now be extended through
1216 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1217 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1218 can be enabled and disabled using the new UDPSegmentationOffload=,
1219 TCPSegmentationOffload=, GenericSegmentationOffload=,
1220 GenericReceiveOffload=, LargeReceiveOffload= options in the
1221 [Link] section of .link files.
1223 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1224 Port VLAN ID can be configured for bridge devices using the new STP=,
1225 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1226 section of .netdev files.
1228 * The route table to which routes received over DHCP or RA should be
1229 added can be configured with the new RouteTable= option in the [DHCP]
1230 and [IPv6AcceptRA] sections of .network files.
1232 * The Address Resolution Protocol can be disabled on links managed by
1233 systemd-networkd using the ARP=no setting in the [Link] section of
1236 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1237 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1238 encode information about the result and exit codes of the current
1239 service runtime cycle.
1241 * systemd-sysctl will now configure kernel parameters in the order
1242 they occur in the configuration files. This matches what sysctl
1243 has been traditionally doing.
1245 * kernel-install "plugins" that are executed to perform various
1246 tasks after a new kernel is added and before an old one is removed
1247 can now return a special value to terminate the procedure and
1248 prevent any later plugins from running.
1250 * Journald's SplitMode=login setting has been deprecated. It has been
1251 removed from documentation, and its use is discouraged. In a future
1252 release it will be completely removed, and made equivalent to current
1253 default of SplitMode=uid.
1255 * Storage=both option setting in /etc/systemd/coredump.conf has been
1256 removed. With fast LZ4 compression storing the core dump twice is not
1259 * The --share-system systemd-nspawn option has been replaced with an
1260 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1261 this functionality is discouraged. In addition the variables
1262 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1263 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1264 individual namespaces.
1266 * "machinectl list" now shows the IP address of running containers in
1267 the output, as well as OS release information.
1269 * "loginctl list" now shows the TTY of each session in the output.
1271 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1272 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1273 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1274 tracking objects in a "recursive" mode, where a single client can be
1275 counted multiple times, if it takes multiple references.
1277 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1278 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1279 process using sd-bus automatically exit if the bus connection is
1282 * Bus clients of the service manager may now "pin" loaded units into
1283 memory, by taking an explicit reference on them. This is useful to
1284 ensure the client can retrieve runtime data about the service even
1285 after the service completed execution. Taking such a reference is
1286 available only for privileged clients and should be helpful to watch
1287 running services in a race-free manner, and in particular collect
1288 information about exit statuses and results.
1290 * The nss-resolve module has been changed to strictly return UNAVAIL
1291 when communication via D-Bus with resolved failed, and NOTFOUND when
1292 a lookup completed but was negative. This means it is now possible to
1293 neatly configure fallbacks using nsswitch.conf result checking
1294 expressions. Taking benefit of this, the new recommended
1295 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1297 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1299 * A new setting CtrlAltDelBurstAction= has been added to
1300 /etc/systemd/system.conf which may be used to configure the precise
1301 behaviour if the user on the console presses Ctrl-Alt-Del more often
1302 than 7 times in 2s. Previously this would unconditionally result in
1303 an expedited, immediate reboot. With this new setting the precise
1304 operation may be configured in more detail, and also turned off
1307 * In .netdev files two new settings RemoteChecksumTx= and
1308 RemoteChecksumRx= are now understood that permit configuring the
1309 remote checksumming logic for VXLAN networks.
1311 * The service manager learnt a new "invocation ID" concept for invoked
1312 services. Each runtime cycle of a service will get a new invocation
1313 ID (a 128bit random UUID) assigned that identifies the current
1314 run of the service uniquely and globally. A new invocation ID
1315 is generated each time a service starts up. The journal will store
1316 the invocation ID of a service along with any logged messages, thus
1317 making the invocation ID useful for matching the online runtime of a
1318 service with the offline log data it generated in a safe way without
1319 relying on synchronized timestamps. In many ways this new service
1320 invocation ID concept is similar to the kernel's boot ID concept that
1321 uniquely and globally identifies the runtime of each boot. The
1322 invocation ID of a service is passed to the service itself via an
1323 environment variable ($INVOCATION_ID). A new bus call
1324 GetUnitByInvocationID() has been added that is similar to GetUnit()
1325 but instead of retrieving the bus path for a unit by its name
1326 retrieves it by its invocation ID. The returned path is valid only as
1327 long as the passed invocation ID is current.
1329 * systemd-resolved gained a new "DNSStubListener" setting in
1330 resolved.conf. It either takes a boolean value or the special values
1331 "udp" and "tcp", and configures whether to enable the stub DNS
1332 listener on 127.0.0.53:53.
1334 * IP addresses configured via networkd may now carry additional
1335 configuration settings supported by the kernel. New options include:
1336 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1337 PrefixRoute=, AutoJoin=.
1339 * The PAM configuration fragment file for "user@.service" shipped with
1340 systemd (i.e. the --user instance of systemd) has been stripped to
1341 the minimum necessary to make the system boot. Previously, it
1342 contained Fedora-specific stanzas that did not apply to other
1343 distributions. It is expected that downstream distributions add
1344 additional configuration lines, matching their needs to this file,
1345 using it only as rough template of what systemd itself needs. Note
1346 that this reduced fragment does not even include an invocation of
1347 pam_limits which most distributions probably want to add, even though
1348 systemd itself does not need it. (There's also the new build time
1349 option --with-pamconfdir=no to disable installation of the PAM
1352 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1353 capability is now also dropped from its set (in addition to
1354 CAP_SYS_MKNOD as before).
1356 * In service unit files it is now possible to connect a specific named
1357 file descriptor with stdin/stdout/stdout of an executed service. The
1358 name may be specified in matching .socket units using the
1359 FileDescriptorName= setting.
1361 * A number of journal settings may now be configured on the kernel
1362 command line. Specifically, the following options are now understood:
1363 systemd.journald.max_level_console=,
1364 systemd.journald.max_level_store=,
1365 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1366 systemd.journald.max_level_wall=.
1368 * "systemctl is-enabled --full" will now show by which symlinks a unit
1369 file is enabled in the unit dependency tree.
1371 * Support for VeraCrypt encrypted partitions has been added to the
1372 "cryptsetup" logic and /etc/crypttab.
1374 * systemd-detect-virt gained support for a new --private-users switch
1375 that checks whether the invoking processes are running inside a user
1376 namespace. Similar, a new special value "private-users" for the
1377 existing ConditionVirtualization= setting has been added, permitting
1378 skipping of specific units in user namespace environments.
1380 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1381 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1382 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1383 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1384 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1385 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1386 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1387 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1388 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1389 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1390 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1391 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1392 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1393 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1394 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1395 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1396 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1397 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1398 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1399 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1400 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1401 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1402 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1403 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1404 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1405 Jędrzejewski-Szmek, Zeal Jagannatha
1407 — Santa Fe, 2016-11-03
1411 * In service units the various ExecXYZ= settings have been extended
1412 with an additional special character as first argument of the
1413 assigned value: if the character '+' is used the specified command
1414 line it will be run with full privileges, regardless of User=,
1415 Group=, CapabilityBoundingSet= and similar options. The effect is
1416 similar to the existing PermissionsStartOnly= option, but allows
1417 configuration of this concept for each executed command line
1420 * Services may now alter the service watchdog timeout at runtime by
1421 sending a WATCHDOG_USEC= message via sd_notify().
1423 * MemoryLimit= and related unit settings now optionally take percentage
1424 specifications. The percentage is taken relative to the amount of
1425 physical memory in the system (or in case of containers, the assigned
1426 amount of memory). This allows scaling service resources neatly with
1427 the amount of RAM available on the system. Similarly, systemd-logind's
1428 RuntimeDirectorySize= option now also optionally takes percentage
1431 * In similar fashion TasksMax= takes percentage values now, too. The
1432 value is taken relative to the configured maximum number of processes
1433 on the system. The per-service task maximum has been changed to 15%
1434 using this functionality. (Effectively this is an increase of 512 →
1435 4915 for service units, given the kernel's default pid_max setting.)
1437 * Calendar time specifications in .timer units now understand a ".."
1438 syntax for time ranges. Example: "4..7:10" may now be used for
1439 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1442 * The InaccessableDirectories=, ReadOnlyDirectories= and
1443 ReadWriteDirectories= unit file settings have been renamed to
1444 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1445 applied to all kinds of file nodes, and not just directories, with
1446 the exception of symlinks. Specifically these settings may now be
1447 used on block and character device nodes, UNIX sockets and FIFOS as
1448 well as regular files. The old names of these settings remain
1449 available for compatibility.
1451 * systemd will now log about all service processes it kills forcibly
1452 (using SIGKILL) because they remained after the clean shutdown phase
1453 of the service completed. This should help identifying services that
1454 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1455 systemd-logind's configuration a similar log message is generated for
1456 processes killed at the end of each session due to this setting.
1458 * systemd will now set the $JOURNAL_STREAM environment variable for all
1459 services whose stdout/stderr are connected to the Journal (which
1460 effectively means by default: all services). The variable contains
1461 the device and inode number of the file descriptor used for
1462 stdout/stderr. This may be used by invoked programs to detect whether
1463 their stdout/stderr is connected to the Journal, in which case they
1464 can switch over to direct Journal communication, thus being able to
1465 pass extended, structured metadata along with their log messages. As
1466 one example, this is now used by glib's logging primitives.
1468 * When using systemd's default tmp.mount unit for /tmp, the mount point
1469 will now be established with the "nosuid" and "nodev" options. This
1470 avoids privilege escalation attacks that put traps and exploits into
1471 /tmp. However, this might cause problems if you e. g. put container
1472 images or overlays into /tmp; if you need this, override tmp.mount's
1473 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1476 * systemd now supports the "memory" cgroup controller also on
1479 * The systemd-cgtop tool now optionally takes a control group path as
1480 command line argument. If specified, the control group list shown is
1481 limited to subgroups of that group.
1483 * The SystemCallFilter= unit file setting gained support for
1484 pre-defined, named system call filter sets. For example
1485 SystemCallFilter=@clock is now an effective way to make all clock
1486 changing-related system calls unavailable to a service. A number of
1487 similar pre-defined groups are defined. Writing system call filters
1488 for system services is simplified substantially with this new
1489 concept. Accordingly, all of systemd's own, long-running services now
1490 enable system call filtering based on this, by default.
1492 * A new service setting MemoryDenyWriteExecute= has been added, taking
1493 a boolean value. If turned on, a service may no longer create memory
1494 mappings that are writable and executable at the same time. This
1495 enhances security for services where this is enabled as it becomes
1496 harder to dynamically write and then execute memory in exploited
1497 service processes. This option has been enabled for all of systemd's
1498 own long-running services.
1500 * A new RestrictRealtime= service setting has been added, taking a
1501 boolean argument. If set the service's processes may no longer
1502 acquire realtime scheduling. This improves security as realtime
1503 scheduling may otherwise be used to easily freeze the system.
1505 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1506 value. This may be used for requesting that the system manager inside
1507 of the container reports start-up completion to nspawn which then
1508 propagates this notification further to the service manager
1509 supervising nspawn itself. A related option NotifyReady= in .nspawn
1510 files has been added too. This functionality allows ordering of the
1511 start-up of multiple containers using the usual systemd ordering
1514 * machinectl gained a new command "stop" that is an alias for
1517 * systemd-resolved gained support for contacting DNS servers on
1518 link-local IPv6 addresses.
1520 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1521 its caches. A method call for requesting the same operation has been
1522 added to the bus API too, and is made available via "systemd-resolve
1525 * systemd-resolve gained a new --status switch. If passed a brief
1526 summary of the used DNS configuration with per-interface information
1529 * resolved.conf gained a new Cache= boolean option, defaulting to
1530 on. If turned off local DNS caching is disabled. This comes with a
1531 performance penalty in particular when DNSSEC is enabled. Note that
1532 resolved disables its internal caching implicitly anyway, when the
1533 configured DNS server is on a host-local IP address such as ::1 or
1534 127.0.0.1, thus automatically avoiding double local caching.
1536 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1537 for DNS requests. This improves compatibility with local programs
1538 that do not use the libc NSS or systemd-resolved's bus APIs for name
1539 resolution. This minimal DNS service is only available to local
1540 programs and does not implement the full DNS protocol, but enough to
1541 cover local DNS clients. A new, static resolv.conf file, listing just
1542 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1543 now recommended to make /etc/resolv.conf a symlink to this file in
1544 order to route all DNS lookups to systemd-resolved, regardless if
1545 done via NSS, the bus API or raw DNS packets. Note that this local
1546 DNS service is not as fully featured as the libc NSS or
1547 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1548 used to deliver link-local address information (as this implies
1549 sending a local interface index along), LLMNR/mDNS support via this
1550 interface is severely restricted. It is thus strongly recommended for
1551 all applications to use the libc NSS API or native systemd-resolved
1554 * systemd-networkd's bridge support learned a new setting
1555 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1556 in .network files has been added for configuring VLAN bridging in
1557 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1559 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1560 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1561 now be acquired without relying on DHCPv6. Two new options
1562 UseDomains= and UseDNS= have been added to configure this behaviour.
1564 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1565 renamed IPv6AcceptRA=, without altering its behaviour. The old
1566 setting name remains available for compatibility reasons.
1568 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1569 Key=, InputKey= and OutputKey=.
1571 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1572 interface configuration.
1574 * "systemctl edit" may now be used to create new unit files by
1575 specifying the --force switch.
1577 * sd-event gained a new function sd_event_get_iteration() for
1578 requesting the current iteration counter of the event loop. It starts
1579 at zero and is increased by one with each event loop iteration.
1581 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1582 file. It can be used in lieu of %systemd_requires in packages which
1583 don't use any systemd functionality and are intended to be installed
1584 in minimal containers without systemd present. This macro provides
1585 ordering dependencies to ensure that if the package is installed in
1586 the same rpm transaction as systemd, systemd will be installed before
1587 the scriptlets for the package are executed, allowing unit presets
1590 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1591 been added to simplify packaging of generators.
1593 * The os-release file gained VERSION_CODENAME field for the
1594 distribution nickname (e.g. VERSION_CODENAME=woody).
1596 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1597 can be set to disable parsing of metadata and the creation
1598 of persistent symlinks for that device.
1600 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1601 to make them available to logged-in users has been reverted.
1603 * Much of the common code of the various systemd components is now
1604 built into an internal shared library libsystemd-shared-231.so
1605 (incorporating the systemd version number in the name, to be updated
1606 with future releases) that the components link to. This should
1607 decrease systemd footprint both in memory during runtime and on
1608 disk. Note that the shared library is not for public use, and is
1609 neither API not ABI stable, but is likely to change with every new
1610 released update. Packagers need to make sure that binaries
1611 linking to libsystemd-shared.so are updated in step with the
1614 * Configuration for "mkosi" is now part of the systemd
1615 repository. mkosi is a tool to easily build legacy-free OS images,
1616 and is available on github: https://github.com/systemd/mkosi. If
1617 "mkosi" is invoked in the build tree a new raw OS image is generated
1618 incorporating the systemd sources currently being worked on and a
1619 clean, fresh distribution installation. The generated OS image may be
1620 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1621 UEFI PC. This functionality is particularly useful to easily test
1622 local changes made to systemd in a pristine, defined environment. See
1623 HACKING for details.
1625 * configure learned the --with-support-url= option to specify the
1626 distribution's bugtracker.
1628 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1629 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1630 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1631 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1632 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1633 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1634 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1635 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1636 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1637 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1638 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1639 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1640 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1641 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1642 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1643 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1644 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1645 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1646 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1648 — Berlin, 2016-07-25
1652 * DNSSEC is now turned on by default in systemd-resolved (in
1653 "allow-downgrade" mode), but may be turned off during compile time by
1654 passing "--with-default-dnssec=no" to "configure" (and of course,
1655 during runtime with DNSSEC= in resolved.conf). We recommend
1656 downstreams to leave this on at least during development cycles and
1657 report any issues with the DNSSEC logic upstream. We are very
1658 interested in collecting feedback about the DNSSEC validator and its
1659 limitations in the wild. Note however, that DNSSEC support is
1660 probably nothing downstreams should turn on in stable distros just
1661 yet, as it might create incompatibilities with a few DNS servers and
1662 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1663 automatically whenever we detect such incompatible setups, but there
1664 might be systems we do not cover yet. Hence: please help us testing
1665 the DNSSEC code, leave this on where you can, report back, but then
1666 again don't consider turning this on in your stable, LTS or
1667 production release just yet. (Note that you have to enable
1668 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1669 and its DNSSEC mode for host name resolution from local
1672 * systemd-resolve conveniently resolves DANE records with the --tlsa
1673 option and OPENPGPKEY records with the --openpgp option. It also
1674 supports dumping raw DNS record data via the new --raw= switch.
1676 * systemd-logind will now by default terminate user processes that are
1677 part of the user session scope unit (session-XX.scope) when the user
1678 logs out. This behavior is controlled by the KillUserProcesses=
1679 setting in logind.conf, and the previous default of "no" is now
1680 changed to "yes". This means that user sessions will be properly
1681 cleaned up after, but additional steps are necessary to allow
1682 intentionally long-running processes to survive logout.
1684 While the user is logged in at least once, user@.service is running,
1685 and any service that should survive the end of any individual login
1686 session can be started at a user service or scope using systemd-run.
1687 systemd-run(1) man page has been extended with an example which shows
1688 how to run screen in a scope unit underneath user@.service. The same
1689 command works for tmux.
1691 After the user logs out of all sessions, user@.service will be
1692 terminated too, by default, unless the user has "lingering" enabled.
1693 To effectively allow users to run long-term tasks even if they are
1694 logged out, lingering must be enabled for them. See loginctl(1) for
1695 details. The default polkit policy was modified to allow users to
1696 set lingering for themselves without authentication.
1698 Previous defaults can be restored at compile time by the
1699 --without-kill-user-processes option to "configure".
1701 * systemd-logind gained new configuration settings SessionsMax= and
1702 InhibitorsMax=, both with a default of 8192. It will not register new
1703 user sessions or inhibitors above this limit.
1705 * systemd-logind will now reload configuration on SIGHUP.
1707 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1708 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1709 enable. Also, support for the "io" cgroup controller in the unified
1710 hierarchy has been added, so that the "memory", "pids" and "io" are
1711 now the controllers that are supported on the unified hierarchy.
1713 WARNING: it is not possible to use previous systemd versions with
1714 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1715 is necessary to also update systemd in the initramfs if using the
1716 unified hierarchy. An updated SELinux policy is also required.
1718 * LLDP support has been extended, and both passive (receive-only) and
1719 active (sender) modes are supported. Passive mode ("routers-only") is
1720 enabled by default in systemd-networkd. Active LLDP mode is enabled
1721 by default for containers on the internal network. The "networkctl
1722 lldp" command may be used to list information gathered. "networkctl
1723 status" will also show basic LLDP information on connected peers now.
1725 * The IAID and DUID unique identifier sent in DHCP requests may now be
1726 configured for the system and each .network file managed by
1727 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1729 * systemd-networkd gained support for configuring proxy ARP support for
1730 each interface, via the ProxyArp= setting in .network files. It also
1731 gained support for configuring the multicast querier feature of
1732 bridge devices, via the new MulticastQuerier= setting in .netdev
1733 files. Similarly, snooping on the IGMP traffic can be controlled
1734 via the new setting MulticastSnooping=.
1736 A new setting PreferredLifetime= has been added for addresses
1737 configured in .network file to configure the lifetime intended for an
1740 The systemd-networkd DHCP server gained the option EmitRouter=, which
1741 defaults to yes, to configure whether the DHCP Option 3 (Router)
1744 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1745 systemd-socket-activate and installed into /usr/bin. It is now fully
1748 * systemd-journald now uses separate threads to flush changes to disk
1749 when closing journal files, thus reducing impact of slow disk I/O on
1750 logging performance.
1752 * The sd-journal API gained two new calls
1753 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1754 can be used to open journal files using file descriptors instead of
1755 file or directory paths. sd_journal_open_container() has been
1756 deprecated, sd_journal_open_directory_fd() should be used instead
1757 with the flag SD_JOURNAL_OS_ROOT.
1759 * journalctl learned a new output mode "-o short-unix" that outputs log
1760 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1761 UTC). It also gained support for a new --no-hostname setting to
1762 suppress the hostname column in the family of "short" output modes.
1764 * systemd-ask-password now optionally skips printing of the password to
1765 stdout with --no-output which can be useful in scripts.
1767 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1768 (devices tagged with ID_MAKER_TOOL) are now tagged with
1769 "uaccess" and are available to logged in users.
1771 * The DeviceAllow= unit setting now supports specifiers (with "%").
1773 * "systemctl show" gained a new --value switch, which allows print a
1774 only the contents of a specific unit property, without also printing
1775 the property's name. Similar support was added to "show*" verbs
1776 of loginctl and machinectl that output "key=value" lists.
1778 * A new unit type "generated" was added for files dynamically generated
1779 by generator tools. Similarly, a new unit type "transient" is used
1780 for unit files created using the runtime API. "systemctl enable" will
1781 refuse to operate on such files.
1783 * A new command "systemctl revert" has been added that may be used to
1784 revert to the vendor version of a unit file, in case local changes
1785 have been made by adding drop-ins or overriding the unit file.
1787 * "machinectl clean" gained a new verb to automatically remove all or
1788 just hidden container images.
1790 * systemd-tmpfiles gained support for a new line type "e" for emptying
1791 directories, if they exist, without creating them if they don't.
1793 * systemd-nspawn gained support for automatically patching the UID/GIDs
1794 of the owners and the ACLs of all files and directories in a
1795 container tree to match the UID/GID user namespacing range selected
1796 for the container invocation. This mode is enabled via the new
1797 --private-users-chown switch. It also gained support for
1798 automatically choosing a free, previously unused UID/GID range when
1799 starting a container, via the new --private-users=pick setting (which
1800 implies --private-users-chown). Together, these options for the first
1801 time make user namespacing for nspawn containers fully automatic and
1802 thus deployable. The systemd-nspawn@.service template unit file has
1803 been changed to use this functionality by default.
1805 * systemd-nspawn gained a new --network-zone= switch, that allows
1806 creating ad-hoc virtual Ethernet links between multiple containers,
1807 that only exist as long as at least one container referencing them is
1808 running. This allows easy connecting of multiple containers with a
1809 common link that implements an Ethernet broadcast domain. Each of
1810 these network "zones" may be named relatively freely by the user, and
1811 may be referenced by any number of containers, but each container may
1812 only reference one of these "zones". On the lower level, this is
1813 implemented by an automatically managed bridge network interface for
1814 each zone, that is created when the first container referencing its
1815 zone is created and removed when the last one referencing its zone
1818 * The default start timeout may now be configured on the kernel command
1819 line via systemd.default_timeout_start_sec=. It was already
1820 configurable via the DefaultTimeoutStartSec= option in
1821 /etc/systemd/system.conf.
1823 * Socket units gained a new TriggerLimitIntervalSec= and
1824 TriggerLimitBurst= setting to configure a limit on the activation
1825 rate of the socket unit.
1827 * The LimitNICE= setting now optionally takes normal UNIX nice values
1828 in addition to the raw integer limit value. If the specified
1829 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1830 value is understood as UNIX nice value. If not prefixed like this it
1831 is understood as raw RLIMIT_NICE limit.
1833 * Note that the effect of the PrivateDevices= unit file setting changed
1834 slightly with this release: the per-device /dev file system will be
1835 mounted read-only from this version on, and will have "noexec"
1836 set. This (minor) change of behavior might cause some (exceptional)
1837 legacy software to break, when PrivateDevices=yes is set for its
1838 service. Please leave PrivateDevices= off if you run into problems
1841 * systemd-bootchart has been split out to a separate repository:
1842 https://github.com/systemd/systemd-bootchart
1844 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1845 merged into the kernel in its current form.
1847 * The compatibility libraries libsystemd-daemon.so,
1848 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1849 which have been deprecated since systemd-209 have been removed along
1850 with the corresponding pkg-config files. All symbols provided by
1851 those libraries are provided by libsystemd.so.
1853 * The Capabilities= unit file setting has been removed (it is ignored
1854 for backwards compatibility). AmbientCapabilities= and
1855 CapabilityBoundingSet= should be used instead.
1857 * A new special target has been added, initrd-root-device.target,
1858 which creates a synchronization point for dependencies of the root
1859 device in early userspace. Initramfs builders must ensure that this
1860 target is now included in early userspace.
1862 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1863 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1864 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1865 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1866 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1867 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1868 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1869 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1870 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1871 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1872 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1873 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1874 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1875 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1876 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1877 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1878 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1879 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1880 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1881 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1882 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1883 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1884 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1885 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1886 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1889 — Fairfax, 2016-05-21
1893 * The systemd-resolved DNS resolver service has gained a substantial
1894 set of new features, most prominently it may now act as a DNSSEC
1895 validating stub resolver. DNSSEC mode is currently turned off by
1896 default, but is expected to be turned on by default in one of the
1897 next releases. For now, we invite everybody to test the DNSSEC logic
1898 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1899 service also gained a full set of D-Bus interfaces, including calls
1900 to configure DNS and DNSSEC settings per link (for use by external
1901 network management software). systemd-resolved and systemd-networkd
1902 now distinguish between "search" and "routing" domains. The former
1903 are used to qualify single-label names, the latter are used purely
1904 for routing lookups within certain domains to specific links.
1905 resolved now also synthesizes RRs for all entries from /etc/hosts.
1907 * The systemd-resolve tool (which is a client utility for
1908 systemd-resolved) has been improved considerably and is now fully
1909 supported and documented. Hence it has moved from /usr/lib/systemd to
1912 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1915 * The coredump collection logic has been reworked: when a coredump is
1916 collected it is now written to disk, compressed and processed
1917 (including stacktrace extraction) from a new instantiated service
1918 systemd-coredump@.service, instead of directly from the
1919 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1920 processing large coredumps can take up a substantial amount of
1921 resources and time, and this previously happened entirely outside of
1922 systemd's service supervision. With the new logic the core_pattern
1923 hook only does minimal metadata collection before passing off control
1924 to the new instantiated service, which is configured with a time
1925 limit, a nice level and other settings to minimize negative impact on
1926 the rest of the system. Also note that the new logic will honour the
1927 RLIMIT_CORE setting of the crashed process, which now allows users
1928 and processes to turn off coredumping for their processes by setting
1931 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1932 and all forked processes by default. Previously, PID 1 would leave
1933 the setting at "0" for all processes, as set by the kernel. Note that
1934 the resource limit traditionally has no effect on the generated
1935 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1936 logic is used. Since the limit is now honoured (see above) its
1937 default has been changed so that the coredumping logic is enabled by
1938 default for all processes, while allowing specific opt-out.
1940 * When the stacktrace is extracted from processes of system users, this
1941 is now done as "systemd-coredump" user, in order to sandbox this
1942 potentially security sensitive parsing operation. (Note that when
1943 processing coredumps of normal users this is done under the user ID
1944 of process that crashed, as before.) Packagers should take notice
1945 that it is now necessary to create the "systemd-coredump" system user
1946 and group at package installation time.
1948 * The systemd-activate socket activation testing tool gained support
1949 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1950 and --seqpacket switches. It also has been extended to support both
1951 new-style and inetd-style file descriptor passing. Use the new
1952 --inetd switch to request inetd-style file descriptor passing.
1954 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1955 variable, which takes a boolean value. If set to false, ANSI color
1956 output is disabled in the tools even when run on a terminal that
1959 * The VXLAN support in networkd now supports two new settings
1960 DestinationPort= and PortRange=.
1962 * A new systemd.machine_id= kernel command line switch has been added,
1963 that may be used to set the machine ID in /etc/machine-id if it is
1964 not initialized yet. This command line option has no effect if the
1965 file is already initialized.
1967 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1968 specified command line as PID 2 rather than PID 1 in the
1969 container. In this mode PID 1 is a minimal stub init process that
1970 implements the special POSIX and Linux semantics of PID 1 regarding
1971 signal and child process management. Note that this stub init process
1972 is implemented in nspawn itself and requires no support from the
1973 container image. This new logic is useful to support running
1974 arbitrary commands in the container, as normal processes are
1975 generally not prepared to run as PID 1.
1977 * systemd-nspawn gained a new --chdir= switch for setting the current
1978 working directory for the process started in the container.
1980 * "journalctl /dev/sda" will now output all kernel log messages for
1981 specified device from the current boot, in addition to all devices
1982 that are parents of it. This should make log output about devices
1983 pretty useful, as long as kernel drivers attach enough metadata to
1984 the log messages. (The usual SATA drivers do.)
1986 * The sd-journal API gained two new calls
1987 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1988 that report whether log data from /run or /var has been found.
1990 * journalctl gained a new switch "--fields" that prints all journal
1991 record field names currently in use in the journal. This is backed
1992 by two new sd-journal API calls sd_journal_enumerate_fields() and
1993 sd_journal_restart_fields().
1995 * Most configurable timeouts in systemd now expect an argument of
1996 "infinity" to turn them off, instead of "0" as before. The semantics
1997 from now on is that a timeout of "0" means "now", and "infinity"
1998 means "never". To maintain backwards compatibility, "0" continues to
1999 turn off previously existing timeout settings.
2001 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2002 try-reload-or-restart" to clarify what it actually does: the "try"
2003 logic applies to both reloading and restarting, not just restarting.
2004 The old name continues to be accepted for compatibility.
2006 * On boot-up, when PID 1 detects that the system clock is behind the
2007 release date of the systemd version in use, the clock is now set
2008 to the latter. Previously, this was already done in timesyncd, in order
2009 to avoid running with clocks set to the various clock epochs such as
2010 1902, 1938 or 1970. With this change the logic is now done in PID 1
2011 in addition to timesyncd during early boot-up, so that it is enforced
2012 before the first process is spawned by systemd. Note that the logic
2013 in timesyncd remains, as it is more comprehensive and ensures
2014 clock monotonicity by maintaining a persistent timestamp file in
2015 /var. Since /var is generally not available in earliest boot or the
2016 initrd, this part of the logic remains in timesyncd, and is not done
2019 * Support for tweaking details in net_cls.class_id through the
2020 NetClass= configuration directive has been removed, as the kernel
2021 people have decided to deprecate that controller in cgroup v2.
2022 Userspace tools such as nftables are moving over to setting rules
2023 that are specific to the full cgroup path of a task, which obsoletes
2024 these controllers anyway. The NetClass= directive is kept around for
2025 legacy compatibility reasons. For a more in-depth description of the
2026 kernel change, please refer to the respective upstream commit:
2028 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2030 * A new service setting RuntimeMaxSec= has been added that may be used
2031 to specify a maximum runtime for a service. If the timeout is hit, the
2032 service is terminated and put into a failure state.
2034 * A new service setting AmbientCapabilities= has been added. It allows
2035 configuration of additional Linux process capabilities that are
2036 passed to the activated processes. This is only available on very
2039 * The process resource limit settings in service units may now be used
2040 to configure hard and soft limits individually.
2042 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2043 expose support for gcc's __attribute__((cleanup())) C extension.
2044 Specifically, for many object destructor functions alternative
2045 versions have been added that have names suffixed with "p" and take a
2046 pointer to a pointer to the object to destroy, instead of just a
2047 pointer to the object itself. This is useful because these destructor
2048 functions may be used directly as parameters to the cleanup
2049 construct. Internally, systemd has been a heavy user of this GCC
2050 extension for a long time, and with this change similar support is
2051 now available to consumers of the library outside of systemd. Note
2052 that by using this extension in your sources compatibility with old
2053 and strictly ANSI compatible C compilers is lost. However, all gcc or
2054 LLVM versions of recent years support this extension.
2056 * Timer units gained support for a new setting RandomizedDelaySec= that
2057 allows configuring some additional randomized delay to the configured
2058 time. This is useful to spread out timer events to avoid load peaks in
2059 clusters or larger setups.
2061 * Calendar time specifications now support sub-second accuracy.
2063 * Socket units now support listening on SCTP and UDP-lite protocol
2066 * The sd-event API now comes with a full set of man pages.
2068 * Older versions of systemd contained experimental support for
2069 compressing journal files and coredumps with the LZ4 compressor that
2070 was not compatible with the lz4 binary (due to API limitations of the
2071 lz4 library). This support has been removed; only support for files
2072 compatible with the lz4 binary remains. This LZ4 logic is now
2073 officially supported and no longer considered experimental.
2075 * The dkr image import logic has been removed again from importd. dkr's
2076 micro-services focus doesn't fit into the machine image focus of
2077 importd, and quickly got out of date with the upstream dkr API.
2079 * Creation of the /run/lock/lockdev/ directory was dropped from
2080 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2081 been available for many years. If you still need this, you need to
2082 create your own tmpfiles.d config file with:
2084 d /run/lock/lockdev 0775 root lock -
2086 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2087 and RebootArgument= have been moved from the [Service] section of
2088 unit files to [Unit], and they are now supported on all unit types,
2089 not just service units. Of course, systemd will continue to
2090 understand these settings also at the old location, in order to
2091 maintain compatibility.
2093 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2094 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2095 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2096 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2097 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2098 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2099 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2100 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2101 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2102 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2103 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2104 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2105 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2106 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2107 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2108 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2109 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2110 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2111 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2113 — Berlin, 2016-02-11
2117 * A number of properties previously only settable in unit
2118 files are now also available as properties to set when
2119 creating transient units programmatically via the bus, as it
2120 is exposed with systemd-run's --property=
2121 setting. Specifically, these are: SyslogIdentifier=,
2122 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2123 EnvironmentFile=, ReadWriteDirectories=,
2124 ReadOnlyDirectories=, InaccessibleDirectories=,
2125 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2127 * When creating transient services via the bus API it is now
2128 possible to pass in a set of file descriptors to use as
2129 STDIN/STDOUT/STDERR for the invoked process.
2131 * Slice units may now be created transiently via the bus APIs,
2132 similar to the way service and scope units may already be
2133 created transiently.
2135 * Wherever systemd expects a calendar timestamp specification
2136 (like in journalctl's --since= and --until= switches) UTC
2137 timestamps are now supported. Timestamps suffixed with "UTC"
2138 are now considered to be in Universal Time Coordinated
2139 instead of the local timezone. Also, timestamps may now
2140 optionally be specified with sub-second accuracy. Both of
2141 these additions also apply to recurring calendar event
2142 specification, such as OnCalendar= in timer units.
2144 * journalctl gained a new "--sync" switch that asks the
2145 journal daemon to write all so far unwritten log messages to
2146 disk and sync the files, before returning.
2148 * systemd-tmpfiles learned two new line types "q" and "Q" that
2149 operate like "v", but also set up a basic btrfs quota
2150 hierarchy when used on a btrfs file system with quota
2153 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2154 instead of a subvolume (even on a btrfs file system) if the
2155 root directory is a plain directory, and not a
2156 subvolume. This should simplify things with certain chroot()
2157 environments which are not aware of the concept of btrfs
2160 * systemd-detect-virt gained a new --chroot switch to detect
2161 whether execution takes place in a chroot() environment.
2163 * CPUAffinity= now takes CPU index ranges in addition to
2166 * The various memory-related resource limit settings (such as
2167 LimitAS=) now understand the usual K, M, G, ... suffixes to
2168 the base of 1024 (IEC). Similar, the time-related resource
2169 limit settings understand the usual min, h, day, ...
2172 * There's a new system.conf setting DefaultTasksMax= to
2173 control the default TasksMax= setting for services and
2174 scopes running on the system. (TasksMax= is the primary
2175 setting that exposes the "pids" cgroup controller on systemd
2176 and was introduced in the previous systemd release.) The
2177 setting now defaults to 512, which means services that are
2178 not explicitly configured otherwise will only be able to
2179 create 512 processes or threads at maximum, from this
2180 version on. Note that this means that thread- or
2181 process-heavy services might need to be reconfigured to set
2182 TasksMax= to a higher value. It is sufficient to set
2183 TasksMax= in these specific unit files to a higher value, or
2184 even "infinity". Similar, there's now a logind.conf setting
2185 UserTasksMax= that defaults to 4096 and limits the total
2186 number of processes or tasks each user may own
2187 concurrently. nspawn containers also have the TasksMax=
2188 value set by default now, to 8192. Note that all of this
2189 only has an effect if the "pids" cgroup controller is
2190 enabled in the kernel. The general benefit of these changes
2191 should be a more robust and safer system, that provides a
2192 certain amount of per-service fork() bomb protection.
2194 * systemd-nspawn gained the new --network-veth-extra= switch
2195 to define additional and arbitrarily-named virtual Ethernet
2196 links between the host and the container.
2198 * A new service execution setting PassEnvironment= has been
2199 added that allows importing select environment variables
2200 from PID1's environment block into the environment block of
2203 * Timer units gained support for a new RemainAfterElapse=
2204 setting which takes a boolean argument. It defaults to on,
2205 exposing behaviour unchanged to previous releases. If set to
2206 off, timer units are unloaded after they elapsed if they
2207 cannot elapse again. This is particularly useful for
2208 transient timer units, which shall not stay around longer
2209 than until they first elapse.
2211 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2212 default now (the kernel default is 16). This is beneficial
2213 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2214 allows substantially larger numbers of queued
2215 datagrams. This should increase the capability of systemd to
2216 parallelize boot-up, as logging and sd_notify() are unlikely
2217 to stall execution anymore. If you need to change the value
2218 from the new defaults, use the usual sysctl.d/ snippets.
2220 * The compression framing format used by the journal or
2221 coredump processing has changed to be in line with what the
2222 official LZ4 tools generate. LZ4 compression support in
2223 systemd was considered unsupported previously, as the format
2224 was not compatible with the normal tools. With this release
2225 this has changed now, and it is hence safe for downstream
2226 distributions to turn it on. While not compressing as well
2227 as the XZ, LZ4 is substantially faster, which makes
2228 it a good default choice for the compression logic in the
2229 journal and in coredump handling.
2231 * Any reference to /etc/mtab has been dropped from
2232 systemd. The file has been obsolete since a while, but
2233 systemd refused to work on systems where it was incorrectly
2234 set up (it should be a symlink or non-existent). Please make
2235 sure to update to util-linux 2.27.1 or newer in conjunction
2236 with this systemd release, which also drops any reference to
2237 /etc/mtab. If you maintain a distribution make sure that no
2238 software you package still references it, as this is a
2239 likely source of bugs. There's also a glibc bug pending,
2240 asking for removal of any reference to this obsolete file:
2242 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2244 Note that only util-linux versions built with
2245 --enable-libmount-force-mountinfo are supported.
2247 * Support for the ".snapshot" unit type has been removed. This
2248 feature turned out to be little useful and little used, and
2249 has now been removed from the core and from systemctl.
2251 * The dependency types RequiresOverridable= and
2252 RequisiteOverridable= have been removed from systemd. They
2253 have been used only very sparingly to our knowledge and
2254 other options that provide a similar effect (such as
2255 systemctl --mode=ignore-dependencies) are much more useful
2256 and commonly used. Moreover, they were only half-way
2257 implemented as the option to control behaviour regarding
2258 these dependencies was never added to systemctl. By removing
2259 these dependency types the execution engine becomes a bit
2260 simpler. Unit files that use these dependencies should be
2261 changed to use the non-Overridable dependency types
2262 instead. In fact, when parsing unit files with these
2263 options, that's what systemd will automatically convert them
2264 too, but it will also warn, asking users to fix the unit
2265 files accordingly. Removal of these dependency types should
2266 only affect a negligible number of unit files in the wild.
2268 * Behaviour of networkd's IPForward= option changed
2269 (again). It will no longer maintain a per-interface setting,
2270 but propagate one way from interfaces where this is enabled
2271 to the global kernel setting. The global setting will be
2272 enabled when requested by a network that is set up, but
2273 never be disabled again. This change was made to make sure
2274 IPv4 and IPv6 behaviour regarding packet forwarding is
2275 similar (as the Linux IPv6 stack does not support
2276 per-interface control of this setting) and to minimize
2279 * In unit files the behaviour of %u, %U, %h, %s has
2280 changed. These specifiers will now unconditionally resolve
2281 to the various user database fields of the user that the
2282 systemd instance is running as, instead of the user
2283 configured in the specific unit via User=. Note that this
2284 effectively doesn't change much, as resolving of these
2285 specifiers was already turned off in the --system instance
2286 of systemd, as we cannot do NSS lookups from PID 1. In the
2287 --user instance of systemd these specifiers where correctly
2288 resolved, but hardly made any sense, since the user instance
2289 lacks privileges to do user switches anyway, and User= is
2290 hence useless. Moreover, even in the --user instance of
2291 systemd behaviour was awkward as it would only take settings
2292 from User= assignment placed before the specifier into
2293 account. In order to unify and simplify the logic around
2294 this the specifiers will now always resolve to the
2295 credentials of the user invoking the manager (which in case
2296 of PID 1 is the root user).
2298 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2299 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2300 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2301 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2302 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2303 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2304 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2305 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2306 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2307 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2308 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2309 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2310 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2311 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2314 — Berlin, 2015-11-18
2318 * systemd now depends on util-linux v2.27. More specifically,
2319 the newly added mount monitor feature in libmount now
2320 replaces systemd's former own implementation.
2322 * libmount mandates /etc/mtab not to be regular file, and
2323 systemd now enforces this condition at early boot.
2324 /etc/mtab has been deprecated and warned about for a very
2325 long time, so systems running systemd should already have
2326 stopped having this file around as anything else than a
2327 symlink to /proc/self/mounts.
2329 * Support for the "pids" cgroup controller has been added. It
2330 allows accounting the number of tasks in a cgroup and
2331 enforcing limits on it. This adds two new setting
2332 TasksAccounting= and TasksMax= to each unit, as well as a
2333 global option DefaultTasksAccounting=.
2335 * Support for the "net_cls" cgroup controller has been added.
2336 It allows assigning a net class ID to each task in the
2337 cgroup, which can then be used in firewall rules and traffic
2338 shaping configurations. Note that the kernel netfilter net
2339 class code does not currently work reliably for ingress
2340 packets on unestablished sockets.
2342 This adds a new config directive called NetClass= to CGroup
2343 enabled units. Allowed values are positive numbers for fixed
2344 assignments and "auto" for picking a free value
2347 * 'systemctl is-system-running' now returns 'offline' if the
2348 system is not booted with systemd. This command can now be
2349 used as a substitute for 'systemd-notify --booted'.
2351 * Watchdog timeouts have been increased to 3 minutes for all
2352 in-tree service files. Apparently, disk IO issues are more
2353 frequent than we hoped, and user reported >1 minute waiting
2356 * 'machine-id-commit' functionality has been merged into
2357 'machine-id-setup --commit'. The separate binary has been
2360 * The WorkingDirectory= directive in unit files may now be set
2361 to the special value '~'. In this case, the working
2362 directory is set to the home directory of the user
2363 configured in User=.
2365 * "machinectl shell" will now open the shell in the home
2366 directory of the selected user by default.
2368 * The CrashChVT= configuration file setting is renamed to
2369 CrashChangeVT=, following our usual logic of not
2370 abbreviating unnecessarily. The old directive is still
2371 supported for compat reasons. Also, this directive now takes
2372 an integer value between 1 and 63, or a boolean value. The
2373 formerly supported '-1' value for disabling stays around for
2376 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2377 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2378 RootDirectory= properties can now be set for transient
2381 * The systemd-analyze tool gained a new "set-log-target" verb
2382 to change the logging target the system manager logs to
2383 dynamically during runtime. This is similar to how
2384 "systemd-analyze set-log-level" already changes the log
2387 * In nspawn /sys is now mounted as tmpfs, with only a selected
2388 set of subdirectories mounted in from the real sysfs. This
2389 enhances security slightly, and is useful for ensuring user
2390 namespaces work correctly.
2392 * Support for USB FunctionFS activation has been added. This
2393 allows implementation of USB gadget services that are
2394 activated as soon as they are requested, so that they don't
2395 have to run continuously, similar to classic socket
2398 * The "systemctl exit" command now optionally takes an
2399 additional parameter that sets the exit code to return from
2400 the systemd manager when exiting. This is only relevant when
2401 running the systemd user instance, or when running the
2402 system instance in a container.
2404 * sd-bus gained the new API calls sd_bus_path_encode_many()
2405 and sd_bus_path_decode_many() that allow easy encoding and
2406 decoding of multiple identifier strings inside a D-Bus
2407 object path. Another new call sd_bus_default_flush_close()
2408 has been added to flush and close per-thread default
2411 * systemd-cgtop gained support for a -M/--machine= switch to
2412 show the control groups within a certain container only.
2414 * "systemctl kill" gained support for an optional --fail
2415 switch. If specified the requested operation will fail of no
2416 processes have been killed, because the unit had no
2417 processes attached, or similar.
2419 * A new systemd.crash_reboot=1 kernel command line option has
2420 been added that triggers a reboot after crashing. This can
2421 also be set through CrashReboot= in systemd.conf.
2423 * The RuntimeDirectory= setting now understands unit
2424 specifiers like %i or %f.
2426 * A new (still internal) library API sd-ipv4acd has been added,
2427 that implements address conflict detection for IPv4. It's
2428 based on code from sd-ipv4ll, and will be useful for
2429 detecting DHCP address conflicts.
2431 * File descriptors passed during socket activation may now be
2432 named. A new API sd_listen_fds_with_names() is added to
2433 access the names. The default names may be overridden,
2434 either in the .socket file using the FileDescriptorName=
2435 parameter, or by passing FDNAME= when storing the file
2436 descriptors using sd_notify().
2438 * systemd-networkd gained support for:
2440 - Setting the IPv6 Router Advertisement settings via
2441 IPv6AcceptRouterAdvertisements= in .network files.
2443 - Configuring the HelloTimeSec=, MaxAgeSec= and
2444 ForwardDelaySec= bridge parameters in .netdev files.
2446 - Configuring PreferredSource= for static routes in
2449 * The "ask-password" framework used to query for LUKS harddisk
2450 passwords or SSL passwords during boot gained support for
2451 caching passwords in the kernel keyring, if it is
2452 available. This makes sure that the user only has to type in
2453 a passphrase once if there are multiple objects to unlock
2454 with the same one. Previously, such password caching was
2455 available only when Plymouth was used; this moves the
2456 caching logic into the systemd codebase itself. The
2457 "systemd-ask-password" utility gained a new --keyname=
2458 switch to control which kernel keyring key to use for
2459 caching a password in. This functionality is also useful for
2460 enabling display managers such as gdm to automatically
2461 unlock the user's GNOME keyring if its passphrase, the
2462 user's password and the harddisk password are the same, if
2463 gdm-autologin is used.
2465 * When downloading tar or raw images using "machinectl
2466 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2467 file is now also downloaded, if it is available and stored
2468 next to the image file.
2470 * Units of type ".socket" gained a new boolean setting
2471 Writable= which is only useful in conjunction with
2472 ListenSpecial=. If true, enables opening the specified
2473 special file in O_RDWR mode rather than O_RDONLY mode.
2475 * systemd-rfkill has been reworked to become a singleton
2476 service that is activated through /dev/rfkill on each rfkill
2477 state change and saves the settings to disk. This way,
2478 systemd-rfkill is now compatible with devices that exist
2479 only intermittendly, and even restores state if the previous
2480 system shutdown was abrupt rather than clean.
2482 * The journal daemon gained support for vacuuming old journal
2483 files controlled by the number of files that shall remain,
2484 in addition to the already existing control by size and by
2485 date. This is useful as journal interleaving performance
2486 degrades with too many separate journal files, and allows
2487 putting an effective limit on them. The new setting defaults
2488 to 100, but this may be changed by setting SystemMaxFiles=
2489 and RuntimeMaxFiles= in journald.conf. Also, the
2490 "journalctl" tool gained the new --vacuum-files= switch to
2491 manually vacuum journal files to leave only the specified
2492 number of files in place.
2494 * udev will now create /dev/disk/by-path links for ATA devices
2495 on kernels where that is supported.
2497 * Galician, Serbian, Turkish and Korean translations were added.
2499 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2500 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2501 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2502 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2503 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2504 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2505 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2506 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2507 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2508 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2509 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2510 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2511 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2512 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2513 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2514 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2515 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2516 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2518 — Berlin, 2015-10-07
2522 * The DHCP implementation of systemd-networkd gained a set of
2525 - The DHCP server now supports emitting DNS and NTP
2526 information. It may be enabled and configured via
2527 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2528 and NTP information is enabled, but no servers are
2529 configured, the corresponding uplink information (if there
2530 is any) is propagated.
2532 - Server and client now support transmission and reception
2533 of timezone information. It can be configured via the
2534 newly introduced network options UseTimezone=,
2535 EmitTimezone=, and Timezone=. Transmission of timezone
2536 information is enabled between host and containers by
2537 default now: the container will change its local timezone
2538 to what the host has set.
2540 - Lease timeouts can now be configured via
2541 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2543 - The DHCP server improved on the stability of
2544 leases. Clients are more likely to get the same lease
2545 information back, even if the server loses state.
2547 - The DHCP server supports two new configuration options to
2548 control the lease address pool metrics, PoolOffset= and
2551 * The encapsulation limit of tunnels in systemd-networkd may
2552 now be configured via 'EncapsulationLimit='. It allows
2553 modifying the maximum additional levels of encapsulation
2554 that are permitted to be prepended to a packet.
2556 * systemd now supports the concept of user buses replacing
2557 session buses, if used with dbus-1.10 (and enabled via dbus
2558 --enable-user-session). It previously only supported this on
2559 kdbus-enabled systems, and this release expands this to
2560 'dbus-daemon' systems.
2562 * systemd-networkd now supports predictable interface names
2565 * systemd now optionally supports the new Linux kernel
2566 "unified" control group hierarchy. If enabled via the kernel
2567 command-line option 'systemd.unified_cgroup_hierarchy=1',
2568 systemd will try to mount the unified cgroup hierarchy
2569 directly on /sys/fs/cgroup. If not enabled, or not
2570 available, systemd will fall back to the legacy cgroup
2571 hierarchy setup, as before. Host system and containers can
2572 mix and match legacy and unified hierarchies as they
2573 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2574 environment variable to individually select the hierarchy to
2575 use for executed containers. By default, nspawn will use the
2576 unified hierarchy for the containers if the host uses the
2577 unified hierarchy, and the legacy hierarchy otherwise.
2578 Please note that at this point the unified hierarchy is an
2579 experimental kernel feature and is likely to change in one
2580 of the next kernel releases. Therefore, it should not be
2581 enabled by default in downstream distributions yet. The
2582 minimum required kernel version for the unified hierarchy to
2583 work is 4.2. Note that when the unified hierarchy is used
2584 for the first time delegated access to controllers is
2585 safe. Because of this systemd-nspawn containers will get
2586 access to controllers now, as will systemd user
2587 sessions. This means containers and user sessions may now
2588 manage their own resources, partitioning up what the system
2591 * A new special scope unit "init.scope" has been introduced
2592 that encapsulates PID 1 of the system. It may be used to
2593 determine resource usage and enforce resource limits on PID
2594 1 itself. PID 1 hence moved out of the root of the control
2597 * The cgtop tool gained support for filtering out kernel
2598 threads when counting tasks in a control group. Also, the
2599 count of processes is now recursively summed up by
2600 default. Two options -k and --recursive= have been added to
2601 revert to old behaviour. The tool has also been updated to
2602 work correctly in containers now.
2604 * systemd-nspawn's --bind= and --bind-ro= options have been
2605 extended to allow creation of non-recursive bind mounts.
2607 * libsystemd gained two new calls sd_pid_get_cgroup() and
2608 sd_peer_get_cgroup() which return the control group path of
2609 a process or peer of a connected AF_UNIX socket. This
2610 function call is particularly useful when implementing
2611 delegated subtrees support in the control group hierarchy.
2613 * The "sd-event" event loop API of libsystemd now supports
2614 correct dequeuing of real-time signals, without losing
2617 * When systemd requests a PolicyKit decision when managing
2618 units it will now add additional fields to the request,
2619 including unit name and desired operation. This enables more
2620 powerful PolicyKit policies, that make decisions depending
2621 on these parameters.
2623 * nspawn learnt support for .nspawn settings files, that may
2624 accompany the image files or directories of containers, and
2625 may contain additional settings for the container. This is
2626 an alternative to configuring container parameters via the
2627 nspawn command line.
2629 Contributions from: Cristian Rodríguez, Daniel Mack, David
2630 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2631 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2632 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2633 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2634 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2635 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2636 Andersen, Tom Gundersen, Torstein Husebø
2638 — Berlin, 2015-09-08
2642 * machinectl gained a new verb 'shell' which opens a fresh
2643 shell on the target container or the host. It is similar to
2644 the existing 'login' command of machinectl, but spawns the
2645 shell directly without prompting for username or
2646 password. The pseudo machine '.host' now refers to the local
2647 host and is used by default. Hence, 'machinectl shell' can
2648 be used as replacement for 'su -' which spawns a session as
2649 a fresh systemd unit in a way that is fully isolated from
2650 the originating session.
2652 * systemd-networkd learned to cope with private-zone DHCP
2653 options and allows other programs to query the values.
2655 * SELinux access control when enabling/disabling units is no
2656 longer enforced with this release. The previous
2657 implementation was incorrect, and a new corrected
2658 implementation is not yet available. As unit file operations
2659 are still protected via PolicyKit and D-Bus policy this is
2660 not a security problem. Yet, distributions which care about
2661 optimal SELinux support should probably not stabilize on
2664 * sd-bus gained support for matches of type "arg0has=", that
2665 test for membership of strings in string arrays sent in bus
2668 * systemd-resolved now dumps the contents of its DNS and LLMNR
2669 caches to the logs on reception of the SIGUSR1 signal. This
2670 is useful to debug DNS behaviour.
2672 * The coredumpctl tool gained a new --directory= option to
2673 operate on journal files in a specific directory.
2675 * "systemctl reboot" and related commands gained a new
2676 "--message=" option which may be used to set a free-text
2677 wall message when shutting down or rebooting the
2678 system. This message is also logged, which is useful for
2679 figuring out the reason for a reboot or shutdown a
2682 * The "systemd-resolve-host" tool's -i switch now takes
2683 network interface numbers as alternative to interface names.
2685 * A new unit file setting for services has been introduced:
2686 UtmpMode= allows configuration of how precisely systemd
2687 handles utmp and wtmp entries for the service if this is
2688 enabled. This allows writing services that appear similar to
2689 user sessions in the output of the "w", "who", "last" and
2692 * systemd-resolved will now locally synthesize DNS resource
2693 records for the "localhost" and "gateway" domains as well as
2694 the local hostname. This should ensure that clients querying
2695 RRs via resolved will get similar results as those going via
2696 NSS, if nss-myhostname is enabled.
2698 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2699 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2700 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2701 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2702 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2703 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2704 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2705 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2706 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2707 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2708 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2709 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2711 — Berlin, 2015-08-27
2715 * The systemd-efi-boot-generator functionality was merged into
2716 systemd-gpt-auto-generator.
2718 * systemd-networkd now supports Group Policy for vxlan
2719 devices. It can be enabled via the new boolean configuration
2720 option called 'GroupPolicyExtension='.
2722 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2723 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2724 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2726 — Berlin, 2015-07-31
2730 * The python-systemd code has been removed from the systemd repository.
2731 A new repository has been created which accommodates the code from
2732 now on, and we kindly ask distributions to create a separate package
2733 for this: https://github.com/systemd/python-systemd
2735 * The systemd daemon will now reload its main configuration
2736 (/etc/systemd/system.conf) on daemon-reload.
2738 * sd-dhcp now exposes vendor specific extensions via
2739 sd_dhcp_lease_get_vendor_specific().
2741 * systemd-networkd gained a number of new configuration options.
2743 - A new boolean configuration option for TAP devices called
2744 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2745 device, thus allowing to send and receive GSO packets.
2747 - A new tunnel configuration option called 'CopyDSCP='.
2748 If enabled, the DSCP field of ip6 tunnels is copied into the
2749 decapsulated packet.
2751 - A set of boolean bridge configuration options were added.
2752 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2753 and 'UnicastFlood=' are now parsed by networkd and applied to the
2754 respective bridge link device via the respective IFLA_BRPORT_*
2757 - A new string configuration option to override the hostname sent
2758 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2759 is true, networkd will use the configured hostname instead of the
2760 system hostname when sending DHCP requests.
2762 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2763 networkd will configure the IPv6 flow-label of the tunnel device
2764 according to RFC2460.
2766 - The 'macvtap' virtual network devices are now supported, similar to
2767 the already supported 'macvlan' devices.
2769 * systemd-resolved now implements RFC5452 to improve resilience against
2770 cache poisoning. Additionally, source port randomization is enabled
2771 by default to further protect against DNS spoofing attacks.
2773 * nss-mymachines now supports translating UIDs and GIDs of running
2774 containers with user-namespaces enabled. If a container 'foo'
2775 translates a host uid 'UID' to the container uid 'TUID', then
2776 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2777 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2778 mapped as 'vg-foo-TGID'.
2780 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2781 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2782 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2783 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2784 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2785 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2786 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2787 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2788 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2789 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2791 — Berlin, 2015-07-29
2795 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2796 There are no known issues with current sysfs, and udev does not need
2797 or should be used to work around such bugs.
2799 * udev does no longer enable USB HID power management. Several reports
2800 indicate, that some devices cannot handle that setting.
2802 * The udev accelerometer helper was removed. The functionality
2803 is now fully included in iio-sensor-proxy. But this means,
2804 older iio-sensor-proxy versions will no longer provide
2805 accelerometer/orientation data with this systemd version.
2806 Please upgrade iio-sensor-proxy to version 1.0.
2808 * networkd gained a new configuration option IPv6PrivacyExtensions=
2809 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2810 for Stateless Address") on selected networks.
2812 * For the sake of fewer build-time dependencies and less code in the
2813 main repository, the python bindings are about to be removed in the
2814 next release. A new repository has been created which accommodates
2815 the code from now on, and we kindly ask distributions to create a
2816 separate package for this. The removal will take place in v223.
2818 https://github.com/systemd/python-systemd
2820 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2821 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2822 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2823 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2824 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2825 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2826 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2827 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2828 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2829 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2831 — Berlin, 2015-07-07
2835 * The sd-bus.h and sd-event.h APIs have now been declared
2836 stable and have been added to the official interface of
2837 libsystemd.so. sd-bus implements an alternative D-Bus client
2838 library, that is relatively easy to use, very efficient and
2839 supports both classic D-Bus as well as kdbus as transport
2840 backend. sd-event is a generic event loop abstraction that
2841 is built around Linux epoll, but adds features such as event
2842 prioritization or efficient timer handling. Both APIs are good
2843 choices for C programs looking for a bus and/or event loop
2844 implementation that is minimal and does not have to be
2845 portable to other kernels.
2847 * kdbus support is no longer compile-time optional. It is now
2848 always built-in. However, it can still be disabled at
2849 runtime using the kdbus=0 kernel command line setting, and
2850 that setting may be changed to default to off, by specifying
2851 --disable-kdbus at build-time. Note though that the kernel
2852 command line setting has no effect if the kdbus.ko kernel
2853 module is not installed, in which case kdbus is (obviously)
2854 also disabled. We encourage all downstream distributions to
2855 begin testing kdbus by adding it to the kernel images in the
2856 development distributions, and leaving kdbus support in
2859 * The minimal required util-linux version has been bumped to
2862 * Support for chkconfig (--enable-chkconfig) was removed in
2863 favor of calling an abstraction tool
2864 /lib/systemd/systemd-sysv-install. This needs to be
2865 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2866 in README for details.
2868 * If there's a systemd unit and a SysV init script for the
2869 same service name, and the user executes "systemctl enable"
2870 for it (or a related call), then this will now enable both
2871 (or execute the related operation on both), not just the
2874 * The libudev API documentation has been converted from gtkdoc
2877 * gudev has been removed from the systemd tree, it is now an
2880 * The systemd-cgtop tool learnt a new --raw switch to generate
2881 "raw" (machine parsable) output.
2883 * networkd's IPForwarding= .network file setting learnt the
2884 new setting "kernel", which ensures that networkd does not
2885 change the IP forwarding sysctl from the default kernel
2888 * The systemd-logind bus API now exposes a new boolean
2889 property "Docked" that reports whether logind considers the
2890 system "docked", i.e. connected to a docking station or not.
2892 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2893 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2894 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2895 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2896 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2897 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2898 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2899 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2900 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2901 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2902 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2903 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2904 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2905 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2906 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2907 Fink, Zbigniew Jędrzejewski-Szmek
2909 — Berlin, 2015-06-19
2913 * The gudev library has been extracted into a separate repository
2914 available at: https://git.gnome.org/browse/libgudev/
2915 It is now managed as part of the Gnome project. Distributions
2916 are recommended to pass --disable-gudev to systemd and use
2917 gudev from the Gnome project instead. gudev is still included
2918 in systemd, for now. It will be removed soon, though. Please
2919 also see the announcement-thread on systemd-devel:
2920 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2922 * systemd now exposes a CPUUsageNSec= property for each
2923 service unit on the bus, that contains the overall consumed
2924 CPU time of a service (the sum of what each process of the
2925 service consumed). This value is only available if
2926 CPUAccounting= is turned on for a service, and is then shown
2927 in the "systemctl status" output.
2929 * Support for configuring alternative mappings of the old SysV
2930 runlevels to systemd targets has been removed. They are now
2931 hardcoded in a way that runlevels 2, 3, 4 all map to
2932 multi-user.target and 5 to graphical.target (which
2933 previously was already the default behaviour).
2935 * The auto-mounter logic gained support for mount point
2936 expiry, using a new TimeoutIdleSec= setting in .automount
2937 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2939 * The EFI System Partition (ESP) as mounted to /boot by
2940 systemd-efi-boot-generator will now be unmounted
2941 automatically after 2 minutes of not being used. This should
2942 minimize the risk of ESP corruptions.
2944 * New /etc/fstab options x-systemd.requires= and
2945 x-systemd.requires-mounts-for= are now supported to express
2946 additional dependencies for mounts. This is useful for
2947 journalling file systems that support external journal
2948 devices or overlay file systems that require underlying file
2949 systems to be mounted.
2951 * systemd does not support direct live-upgrades (via systemctl
2952 daemon-reexec) from versions older than v44 anymore. As no
2953 distribution we are aware of shipped such old versions in a
2954 stable release this should not be problematic.
2956 * When systemd forks off a new per-connection service instance
2957 it will now set the $REMOTE_ADDR environment variable to the
2958 remote IP address, and $REMOTE_PORT environment variable to
2959 the remote IP port. This behaviour is similar to the
2960 corresponding environment variables defined by CGI.
2962 * systemd-networkd gained support for uplink failure
2963 detection. The BindCarrier= option allows binding interface
2964 configuration dynamically to the link sense of other
2965 interfaces. This is useful to achieve behaviour like in
2968 * systemd-networkd gained support for configuring the DHCP
2969 client identifier to use when requesting leases.
2971 * systemd-networkd now has a per-network UseNTP= option to
2972 configure whether NTP server information acquired via DHCP
2973 is passed on to services like systemd-timesyncd.
2975 * systemd-networkd gained support for vti6 tunnels.
2977 * Note that systemd-networkd manages the sysctl variable
2978 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2979 it is configured for since v219. The variable controls IP
2980 forwarding, and is a per-interface alternative to the global
2981 /proc/sys/net/ipv[46]/ip_forward. This setting is
2982 configurable in the IPForward= option, which defaults to
2983 "no". This means if networkd is used for an interface it is
2984 no longer sufficient to set the global sysctl option to turn
2985 on IP forwarding! Instead, the .network file option
2986 IPForward= needs to be turned on! Note that the
2987 implementation of this behaviour was broken in v219 and has
2990 * Many bonding and vxlan options are now configurable in
2993 * systemd-nspawn gained a new --property= setting to set unit
2994 properties for the container scope. This is useful for
2995 setting resource parameters (e.g. "CPUShares=500") on
2996 containers started from the command line.
2998 * systemd-nspawn gained a new --private-users= switch to make
2999 use of user namespacing available on recent Linux kernels.
3001 * systemd-nspawn may now be called as part of a shell pipeline
3002 in which case the pipes used for stdin and stdout are passed
3003 directly to the process invoked in the container, without
3004 indirection via a pseudo tty.
3006 * systemd-nspawn gained a new switch to control the UNIX
3007 signal to use when killing the init process of the container
3010 * systemd-nspawn gained a new --overlay= switch for mounting
3011 overlay file systems into the container using the new kernel
3014 * When a container image is imported via systemd-importd and
3015 the host file system is not btrfs, a loopback block device
3016 file is created in /var/lib/machines.raw with a btrfs file
3017 system inside. It is then mounted to /var/lib/machines to
3018 enable btrfs features for container management. The loopback
3019 file and btrfs file system is grown as needed when container
3020 images are imported via systemd-importd.
3022 * systemd-machined/systemd-importd gained support for btrfs
3023 quota, to enforce container disk space limits on disk. This
3024 is exposed in "machinectl set-limit".
3026 * systemd-importd now can import containers from local .tar,
3027 .raw and .qcow2 images, and export them to .tar and .raw. It
3028 can also import dkr v2 images now from the network (on top
3031 * systemd-importd gained support for verifying downloaded
3032 images with gpg2 (previously only gpg1 was supported).
3034 * systemd-machined, systemd-logind, systemd: most bus calls
3035 are now accessible to unprivileged processes via
3036 PolicyKit. Also, systemd-logind will now allow users to kill
3037 their own sessions without further privileges or
3040 * systemd-shutdownd has been removed. This service was
3041 previously responsible for implementing scheduled shutdowns
3042 as exposed in /usr/bin/shutdown's time parameter. This
3043 functionality has now been moved into systemd-logind and is
3044 accessible via a bus interface.
3046 * "systemctl reboot" gained a new switch --firmware-setup that
3047 can be used to reboot into the EFI firmware setup, if that
3048 is available. systemd-logind now exposes an API on the bus
3049 to trigger such reboots, in case graphical desktop UIs want
3050 to cover this functionality.
3052 * "systemctl enable", "systemctl disable" and "systemctl mask"
3053 now support a new "--now" switch. If specified the units
3054 that are enabled will also be started, and the ones
3055 disabled/masked also stopped.
3057 * The Gummiboot EFI boot loader tool has been merged into
3058 systemd, and renamed to "systemd-boot". The bootctl tool has been
3059 updated to support systemd-boot.
3061 * An EFI kernel stub has been added that may be used to create
3062 kernel EFI binaries that contain not only the actual kernel,
3063 but also an initrd, boot splash, command line and OS release
3064 information. This combined binary can then be signed as a
3065 single image, so that the firmware can verify it all in one
3066 step. systemd-boot has special support for EFI binaries created
3067 like this and can extract OS release information from them
3068 and show them in the boot menu. This functionality is useful
3069 to implement cryptographically verified boot schemes.
3071 * Optional support has been added to systemd-fsck to pass
3072 fsck's progress report to an AF_UNIX socket in the file
3075 * udev will no longer create device symlinks for all block
3076 devices by default. A blacklist for excluding special block
3077 devices from this logic has been turned into a whitelist
3078 that requires picking block devices explicitly that require
3081 * A new (currently still internal) API sd-device.h has been
3082 added to libsystemd. This modernized API is supposed to
3083 replace libudev eventually. In fact, already much of libudev
3084 is now just a wrapper around sd-device.h.
3086 * A new hwdb database for storing metadata about pointing
3087 stick devices has been added.
3089 * systemd-tmpfiles gained support for setting file attributes
3090 similar to the "chattr" tool with new 'h' and 'H' lines.
3092 * systemd-journald will no longer unconditionally set the
3093 btrfs NOCOW flag on new journal files. This is instead done
3094 with tmpfiles snippet using the new 'h' line type. This
3095 allows easy disabling of this logic, by masking the
3096 journal-nocow.conf tmpfiles file.
3098 * systemd-journald will now translate audit message types to
3099 human readable identifiers when writing them to the
3100 journal. This should improve readability of audit messages.
3102 * The LUKS logic gained support for the offset= and skip=
3103 options in /etc/crypttab, as previously implemented by
3106 * /usr/lib/os-release gained a new optional field VARIANT= for
3107 distributions that support multiple variants (such as a
3108 desktop edition, a server edition, ...)
3110 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3111 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3112 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3113 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3114 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3115 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3116 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3117 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3118 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3119 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3120 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3121 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3122 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3123 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3124 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3125 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3126 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3127 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3128 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3129 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3130 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3131 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3132 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3133 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3134 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3135 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3136 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3138 — Berlin, 2015-05-22
3142 * Introduce a new API "sd-hwdb.h" for querying the hardware
3143 metadata database. With this minimal interface one can query
3144 and enumerate the udev hwdb, decoupled from the old libudev
3145 library. libudev's interface for this is now only a wrapper
3146 around sd-hwdb. A new tool systemd-hwdb has been added to
3147 interface with and update the database.
3149 * When any of systemd's tools copies files (for example due to
3150 tmpfiles' C lines) a btrfs reflink will attempted first,
3151 before bytewise copying is done.
3153 * systemd-nspawn gained a new --ephemeral switch. When
3154 specified a btrfs snapshot is taken of the container's root
3155 directory, and immediately removed when the container
3156 terminates again. Thus, a container can be started whose
3157 changes never alter the container's root directory, and are
3158 lost on container termination. This switch can also be used
3159 for starting a container off the root file system of the
3160 host without affecting the host OS. This switch is only
3161 available on btrfs file systems.
3163 * systemd-nspawn gained a new --template= switch. It takes the
3164 path to a container tree to use as template for the tree
3165 specified via --directory=, should that directory be
3166 missing. This allows instantiating containers dynamically,
3167 on first run. This switch is only available on btrfs file
3170 * When a .mount unit refers to a mount point on which multiple
3171 mounts are stacked, and the .mount unit is stopped all of
3172 the stacked mount points will now be unmounted until no
3173 mount point remains.
3175 * systemd now has an explicit notion of supported and
3176 unsupported unit types. Jobs enqueued for unsupported unit
3177 types will now fail with an "unsupported" error code. More
3178 specifically .swap, .automount and .device units are not
3179 supported in containers, .busname units are not supported on
3180 non-kdbus systems. .swap and .automount are also not
3181 supported if their respective kernel compile time options
3184 * machinectl gained support for two new "copy-from" and
3185 "copy-to" commands for copying files from a running
3186 container to the host or vice versa.
3188 * machinectl gained support for a new "bind" command to bind
3189 mount host directories into local containers. This is
3190 currently only supported for nspawn containers.
3192 * networkd gained support for configuring bridge forwarding
3193 database entries (fdb) from .network files.
3195 * A new tiny daemon "systemd-importd" has been added that can
3196 download container images in tar, raw, qcow2 or dkr formats,
3197 and make them available locally in /var/lib/machines, so
3198 that they can run as nspawn containers. The daemon can GPG
3199 verify the downloads (not supported for dkr, since it has no
3200 provisions for verifying downloads). It will transparently
3201 decompress bz2, xz, gzip compressed downloads if necessary,
3202 and restore sparse files on disk. The daemon uses privilege
3203 separation to ensure the actual download logic runs with
3204 fewer privileges than the daemon itself. machinectl has
3205 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3206 make the functionality of importd available to the
3207 user. With this in place the Fedora and Ubuntu "Cloud"
3208 images can be downloaded and booted as containers unmodified
3209 (the Fedora images lack the appropriate GPG signature files
3210 currently, so they cannot be verified, but this will change
3211 soon, hopefully). Note that downloading images is currently
3212 only fully supported on btrfs.
3214 * machinectl is now able to list container images found in
3215 /var/lib/machines, along with some metadata about sizes of
3216 disk and similar. If the directory is located on btrfs and
3217 quota is enabled, this includes quota display. A new command
3218 "image-status" has been added that shows additional
3219 information about images.
3221 * machinectl is now able to clone container images
3222 efficiently, if the underlying file system (btrfs) supports
3223 it, with the new "machinectl clone" command. It also
3224 gained commands for renaming and removing images, as well as
3225 marking them read-only or read-write (supported also on
3226 legacy file systems).
3228 * networkd gained support for collecting LLDP network
3229 announcements, from hardware that supports this. This is
3230 shown in networkctl output.
3232 * systemd-run gained support for a new -t (--pty) switch for
3233 invoking a binary on a pty whose input and output is
3234 connected to the invoking terminal. This allows executing
3235 processes as system services while interactively
3236 communicating with them via the terminal. Most interestingly
3237 this is supported across container boundaries. Invoking
3238 "systemd-run -t /bin/bash" is an alternative to running a
3239 full login session, the difference being that the former
3240 will not register a session, nor go through the PAM session
3243 * tmpfiles gained support for a new "v" line type for creating
3244 btrfs subvolumes. If the underlying file system is a legacy
3245 file system, this automatically degrades to creating a
3246 normal directory. Among others /var/lib/machines is now
3247 created like this at boot, should it be missing.
3249 * The directory /var/lib/containers/ has been deprecated and
3250 been replaced by /var/lib/machines. The term "machines" has
3251 been used in the systemd context as generic term for both
3252 VMs and containers, and hence appears more appropriate for
3253 this, as the directory can also contain raw images bootable
3256 * systemd-nspawn when invoked with -M but without --directory=
3257 or --image= is now capable of searching for the container
3258 root directory, subvolume or disk image automatically, in
3259 /var/lib/machines. systemd-nspawn@.service has been updated
3260 to make use of this, thus allowing it to be used for raw
3263 * A new machines.target unit has been introduced that is
3264 supposed to group all containers/VMs invoked as services on
3265 the system. systemd-nspawn@.service has been updated to
3266 integrate with that.
3268 * machinectl gained a new "start" command, for invoking a
3269 container as a service. "machinectl start foo" is mostly
3270 equivalent to "systemctl start systemd-nspawn@foo.service",
3271 but handles escaping in a nicer way.
3273 * systemd-nspawn will now mount most of the cgroupfs tree
3274 read-only into each container, with the exception of the
3275 container's own subtree in the name=systemd hierarchy.
3277 * journald now sets the special FS_NOCOW file flag for its
3278 journal files. This should improve performance on btrfs, by
3279 avoiding heavy fragmentation when journald's write-pattern
3280 is used on COW file systems. It degrades btrfs' data
3281 integrity guarantees for the files to the same levels as for
3282 ext3/ext4 however. This should be OK though as journald does
3283 its own data integrity checks and all its objects are
3284 checksummed on disk. Also, journald should handle btrfs disk
3285 full events a lot more gracefully now, by processing SIGBUS
3286 errors, and not relying on fallocate() anymore.
3288 * When journald detects that journal files it is writing to
3289 have been deleted it will immediately start new journal
3292 * systemd now provides a way to store file descriptors
3293 per-service in PID 1. This is useful for daemons to ensure
3294 that fds they require are not lost during a daemon
3295 restart. The fds are passed to the daemon on the next
3296 invocation in the same way socket activation fds are
3297 passed. This is now used by journald to ensure that the
3298 various sockets connected to all the system's stdout/stderr
3299 are not lost when journald is restarted. File descriptors
3300 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3301 an extension to sd_notify(). Note that a limit is enforced
3302 on the number of fds a service can store in PID 1, and it
3303 defaults to 0, so that no fds may be stored, unless this is
3304 explicitly turned on.
3306 * The default TERM variable to use for units connected to a
3307 terminal, when no other value is explicitly is set is now
3308 vt220 rather than vt102. This should be fairly safe still,
3309 but allows PgUp/PgDn work.
3311 * The /etc/crypttab option header= as known from Debian is now
3314 * "loginctl user-status" and "loginctl session-status" will
3315 now show the last 10 lines of log messages of the
3316 user/session following the status output. Similar,
3317 "machinectl status" will show the last 10 log lines
3318 associated with a virtual machine or container
3319 service. (Note that this is usually not the log messages
3320 done in the VM/container itself, but simply what the
3321 container manager logs. For nspawn this includes all console
3324 * "loginctl session-status" without further argument will now
3325 show the status of the session of the caller. Similar,
3326 "lock-session", "unlock-session", "activate",
3327 "enable-linger", "disable-linger" may now be called without
3328 session/user parameter in which case they apply to the
3329 caller's session/user.
3331 * An X11 session scriptlet is now shipped that uploads
3332 $DISPLAY and $XAUTHORITY into the environment of the systemd
3333 --user daemon if a session begins. This should improve
3334 compatibility with X11 enabled applications run as systemd
3337 * Generators are now subject to masking via /etc and /run, the
3338 same way as unit files.
3340 * networkd .network files gained support for configuring
3341 per-link IPv4/IPv6 packet forwarding as well as IPv4
3342 masquerading. This is by default turned on for veth links to
3343 containers, as registered by systemd-nspawn. This means that
3344 nspawn containers run with --network-veth will now get
3345 automatic routed access to the host's networks without any
3346 further configuration or setup, as long as networkd runs on
3349 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3350 or UDP posts of a container on the host. With this in place
3351 it is possible to run containers with private veth links
3352 (--network-veth), and have their functionality exposed on
3353 the host as if their services were running directly on the
3356 * systemd-nspawn's --network-veth switch now gained a short
3357 version "-n", since with the changes above it is now truly
3358 useful out-of-the-box. The systemd-nspawn@.service has been
3359 updated to make use of it too by default.
3361 * systemd-nspawn will now maintain a per-image R/W lock, to
3362 ensure that the same image is not started more than once
3363 writable. (It's OK to run an image multiple times
3364 simultaneously in read-only mode.)
3366 * systemd-nspawn's --image= option is now capable of
3367 dissecting and booting MBR and GPT disk images that contain
3368 only a single active Linux partition. Previously it
3369 supported only GPT disk images with proper GPT type
3370 IDs. This allows running cloud images from major
3371 distributions directly with systemd-nspawn, without
3374 * In addition to collecting mouse dpi data in the udev
3375 hardware database, there's now support for collecting angle
3376 information for mouse scroll wheels. The database is
3377 supposed to guarantee similar scrolling behavior on mice
3378 that it knows about. There's also support for collecting
3379 information about Touchpad types.
3381 * udev's input_id built-in will now also collect touch screen
3382 dimension data and attach it to probed devices.
3384 * /etc/os-release gained support for a Distribution Privacy
3387 * networkd gained support for creating "ipvlan", "gretap",
3388 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3390 * systemd-tmpfiles gained support for "a" lines for setting
3393 * systemd-nspawn will now mount /tmp in the container to
3394 tmpfs, automatically.
3396 * systemd now exposes the memory.usage_in_bytes cgroup
3397 attribute and shows it for each service in the "systemctl
3398 status" output, if available.
3400 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3401 immediate reboot is triggered. This useful if shutdown is
3402 hung and is unable to complete, to expedite the
3403 operation. Note that this kind of reboot will still unmount
3404 all file systems, and hence should not result in fsck being
3407 * A .device unit for an optical block device will now be
3408 considered active only when a medium is in the drive. Also,
3409 mount units are now bound to their backing devices thus
3410 triggering automatic unmounting when devices become
3411 unavailable. With this in place systemd will now
3412 automatically unmount left-over mounts when a CD-ROM is
3413 ejected or an USB stick is yanked from the system.
3415 * networkd-wait-online now has support for waiting for
3416 specific interfaces only (with globbing), and for giving up
3417 after a configurable timeout.
3419 * networkd now exits when idle. It will be automatically
3420 restarted as soon as interfaces show up, are removed or
3421 change state. networkd will stay around as long as there is
3422 at least one DHCP state machine or similar around, that keep
3425 * networkd may now configure IPv6 link-local addressing in
3426 addition to IPv4 link-local addressing.
3428 * The IPv6 "token" for use in SLAAC may now be configured for
3429 each .network interface in networkd.
3431 * Routes configured with networkd may now be assigned a scope
3434 * networkd's [Match] sections now support globbing and lists
3435 of multiple space-separated matches per item.
3437 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3438 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3439 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3440 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3441 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3442 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3443 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3444 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3445 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3446 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3447 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3448 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3449 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3450 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3451 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3452 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3453 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3454 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3455 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3456 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3457 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3458 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3459 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3460 Hoffmann, Zbigniew Jędrzejewski-Szmek
3462 — Berlin, 2015-02-16
3466 * When querying unit file enablement status (for example via
3467 "systemctl is-enabled"), a new state "indirect" is now known
3468 which indicates that a unit might not be enabled itself, but
3469 another unit listed in its Also= setting might be.
3471 * Similar to the various existing ConditionXYZ= settings for
3472 units, there are now matching AssertXYZ= settings. While
3473 failing conditions cause a unit to be skipped, but its job
3474 to succeed, failing assertions declared like this will cause
3475 a unit start operation and its job to fail.
3477 * hostnamed now knows a new chassis type "embedded".
3479 * systemctl gained a new "edit" command. When used on a unit
3480 file, this allows extending unit files with .d/ drop-in
3481 configuration snippets or editing the full file (after
3482 copying it from /usr/lib to /etc). This will invoke the
3483 user's editor (as configured with $EDITOR), and reload the
3484 modified configuration after editing.
3486 * "systemctl status" now shows the suggested enablement state
3487 for a unit, as declared in the (usually vendor-supplied)
3488 system preset files.
3490 * nss-myhostname will now resolve the single-label host name
3491 "gateway" to the locally configured default IP routing
3492 gateways, ordered by their metrics. This assigns a stable
3493 name to the used gateways, regardless which ones are
3494 currently configured. Note that the name will only be
3495 resolved after all other name sources (if nss-myhostname is
3496 configured properly) and should hence not negatively impact
3497 systems that use the single-label host name "gateway" in
3500 * systemd-inhibit now allows filtering by mode when listing
3503 * Scope and service units gained a new "Delegate" boolean
3504 property, which, when set, allows processes running inside the
3505 unit to further partition resources. This is primarily
3506 useful for systemd user instances as well as container
3509 * journald will now pick up audit messages directly from
3510 the kernel, and log them like any other log message. The
3511 audit fields are split up and fully indexed. This means that
3512 journalctl in many ways is now a (nicer!) alternative to
3513 ausearch, the traditional audit client. Note that this
3514 implements only a minimal audit client. If you want the
3515 special audit modes like reboot-on-log-overflow, please use
3516 the traditional auditd instead, which can be used in
3517 parallel to journald.
3519 * The ConditionSecurity= unit file option now understands the
3520 special string "audit" to check whether auditing is
3523 * journalctl gained two new commands --vacuum-size= and
3524 --vacuum-time= to delete old journal files until the
3525 remaining ones take up no more than the specified size on disk,
3526 or are not older than the specified time.
3528 * A new, native PPPoE library has been added to sd-network,
3529 systemd's library of light-weight networking protocols. This
3530 library will be used in a future version of networkd to
3531 enable PPPoE communication without an external pppd daemon.
3533 * The busctl tool now understands a new "capture" verb that
3534 works similar to "monitor", but writes a packet capture
3535 trace to STDOUT that can be redirected to a file which is
3536 compatible with libcap's capture file format. This can then
3537 be loaded in Wireshark and similar tools to inspect bus
3540 * The busctl tool now understands a new "tree" verb that shows
3541 the object trees of a specific service on the bus, or of all
3544 * The busctl tool now understands a new "introspect" verb that
3545 shows all interfaces and members of objects on the bus,
3546 including their signature and values. This is particularly
3547 useful to get more information about bus objects shown by
3548 the new "busctl tree" command.
3550 * The busctl tool now understands new verbs "call",
3551 "set-property" and "get-property" for invoking bus method
3552 calls, setting and getting bus object properties in a
3555 * busctl gained a new --augment-creds= argument that controls
3556 whether the tool shall augment credential information it
3557 gets from the bus with data from /proc, in a possibly
3560 * nspawn's --link-journal= switch gained two new values
3561 "try-guest" and "try-host" that work like "guest" and
3562 "host", but do not fail if the host has no persistent
3563 journalling enabled. -j is now equivalent to
3564 --link-journal=try-guest.
3566 * macvlan network devices created by nspawn will now have
3567 stable MAC addresses.
3569 * A new SmackProcessLabel= unit setting has been added, which
3570 controls the SMACK security label processes forked off by
3571 the respective unit shall use.
3573 * If compiled with --enable-xkbcommon, systemd-localed will
3574 verify x11 keymap settings by compiling the given keymap. It
3575 will spew out warnings if the compilation fails. This
3576 requires libxkbcommon to be installed.
3578 * When a coredump is collected, a larger number of metadata
3579 fields is now collected and included in the journal records
3580 created for it. More specifically, control group membership,
3581 environment variables, memory maps, working directory,
3582 chroot directory, /proc/$PID/status, and a list of open file
3583 descriptors is now stored in the log entry.
3585 * The udev hwdb now contains DPI information for mice. For
3588 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3590 * All systemd programs that read standalone configuration
3591 files in /etc now also support a corresponding series of
3592 .conf.d configuration directories in /etc/, /run/,
3593 /usr/local/lib/, /usr/lib/, and (if configured with
3594 --enable-split-usr) /lib/. In particular, the following
3595 configuration files now have corresponding configuration
3596 directories: system.conf user.conf, logind.conf,
3597 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3598 resolved.conf, timesyncd.conf, journal-remote.conf, and
3599 journal-upload.conf. Note that distributions should use the
3600 configuration directories in /usr/lib/; the directories in
3601 /etc/ are reserved for the system administrator.
3603 * systemd-rfkill will no longer take the rfkill device name
3604 into account when storing rfkill state on disk, as the name
3605 might be dynamically assigned and not stable. Instead, the
3606 ID_PATH udev variable combined with the rfkill type (wlan,
3607 bluetooth, ...) is used.
3609 * A new service systemd-machine-id-commit.service has been
3610 added. When used on systems where /etc is read-only during
3611 boot, and /etc/machine-id is not initialized (but an empty
3612 file), this service will copy the temporary machine ID
3613 created as replacement into /etc after the system is fully
3614 booted up. This is useful for systems that are freshly
3615 installed with a non-initialized machine ID, but should get
3616 a fixed machine ID for subsequent boots.
3618 * networkd's .netdev files now provide a large set of
3619 configuration parameters for VXLAN devices. Similarly, the
3620 bridge port cost parameter is now configurable in .network
3621 files. There's also new support for configuring IP source
3622 routing. networkd .link files gained support for a new
3623 OriginalName= match that is useful to match against the
3624 original interface name the kernel assigned. .network files
3625 may include MTU= and MACAddress= fields for altering the MTU
3626 and MAC address while being connected to a specific network
3629 * The LUKS logic gained supported for configuring
3630 UUID-specific key files. There's also new support for naming
3631 LUKS device from the kernel command line, using the new
3632 luks.name= argument.
3634 * Timer units may now be transiently created via the bus API
3635 (this was previously already available for scope and service
3636 units). In addition it is now possible to create multiple
3637 transient units at the same time with a single bus call. The
3638 "systemd-run" tool has been updated to make use of this for
3639 running commands on a specified time, in at(1)-style.
3641 * tmpfiles gained support for "t" lines, for assigning
3642 extended attributes to files. Among other uses this may be
3643 used to assign SMACK labels to files.
3645 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3646 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3647 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3648 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3649 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3650 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3651 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3652 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3653 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3654 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3655 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3656 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3657 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3658 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3659 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3660 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3661 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3662 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3664 — Berlin, 2014-12-10
3668 * journalctl gained the new options -t/--identifier= to match
3669 on the syslog identifier (aka "tag"), as well as --utc to
3670 show log timestamps in the UTC timezone. journalctl now also
3671 accepts -n/--lines=all to disable line capping in a pager.
3673 * journalctl gained a new switch, --flush, that synchronously
3674 flushes logs from /run/log/journal to /var/log/journal if
3675 persistent storage is enabled. systemd-journal-flush.service
3676 now waits until the operation is complete.
3678 * Services can notify the manager before they start a reload
3679 (by sending RELOADING=1) or shutdown (by sending
3680 STOPPING=1). This allows the manager to track and show the
3681 internal state of daemons and closes a race condition when
3682 the process is still running but has closed its D-Bus
3685 * Services with Type=oneshot do not have to have any ExecStart
3688 * User units are now loaded also from
3689 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3690 /run/systemd/user directory that was already previously
3691 supported, but is under the control of the user.
3693 * Job timeouts (i.e. time-outs on the time a job that is
3694 queued stays in the run queue) can now optionally result in
3695 immediate reboot or power-off actions (JobTimeoutAction= and
3696 JobTimeoutRebootArgument=). This is useful on ".target"
3697 units, to limit the maximum time a target remains
3698 undispatched in the run queue, and to trigger an emergency
3699 operation in such a case. This is now used by default to
3700 turn off the system if boot-up (as defined by everything in
3701 basic.target) hangs and does not complete for at least
3702 15min. Also, if power-off or reboot hang for at least 30min
3703 an immediate power-off/reboot operation is triggered. This
3704 functionality is particularly useful to increase reliability
3705 on embedded devices, but also on laptops which might
3706 accidentally get powered on when carried in a backpack and
3707 whose boot stays stuck in a hard disk encryption passphrase
3710 * systemd-logind can be configured to also handle lid switch
3711 events even when the machine is docked or multiple displays
3712 are attached (HandleLidSwitchDocked= option).
3714 * A helper binary and a service have been added which can be
3715 used to resume from hibernation in the initramfs. A
3716 generator will parse the resume= option on the kernel
3717 command line to trigger resume.
3719 * A user console daemon systemd-consoled has been
3720 added. Currently, it is a preview, and will so far open a
3721 single terminal on each session of the user marked as
3722 Desktop=systemd-console.
3724 * Route metrics can be specified for DHCP routes added by
3727 * The SELinux context of socket-activated services can be set
3728 from the information provided by the networking stack
3729 (SELinuxContextFromNet= option).
3731 * Userspace firmware loading support has been removed and
3732 the minimum supported kernel version is thus bumped to 3.7.
3734 * Timeout for udev workers has been increased from 1 to 3
3735 minutes, but a warning will be printed after 1 minute to
3736 help diagnose kernel modules that take a long time to load.
3738 * Udev rules can now remove tags on devices with TAG-="foobar".
3740 * systemd's readahead implementation has been removed. In many
3741 circumstances it didn't give expected benefits even for
3742 rotational disk drives and was becoming less relevant in the
3743 age of SSDs. As none of the developers has been using
3744 rotating media anymore, and nobody stepped up to actively
3745 maintain this component of systemd it has now been removed.
3747 * Swap units can use Options= to specify discard options.
3748 Discard options specified for swaps in /etc/fstab are now
3751 * Docker containers are now detected as a separate type of
3754 * The Password Agent protocol gained support for queries where
3755 the user input is shown, useful e.g. for user names.
3756 systemd-ask-password gained a new --echo option to turn that
3759 * The default sysctl.d/ snippets will now set:
3761 net.core.default_qdisc = fq_codel
3763 This selects Fair Queuing Controlled Delay as the default
3764 queuing discipline for network interfaces. fq_codel helps
3765 fight the network bufferbloat problem. It is believed to be
3766 a good default with no tuning required for most workloads.
3767 Downstream distributions may override this choice. On 10Gbit
3768 servers that do not do forwarding, "fq" may perform better.
3769 Systems without a good clocksource should use "pfifo_fast".
3771 * If kdbus is enabled during build a new option BusPolicy= is
3772 available for service units, that allows locking all service
3773 processes into a stricter bus policy, in order to limit
3774 access to various bus services, or even hide most of them
3775 from the service's view entirely.
3777 * networkctl will now show the .network and .link file
3778 networkd has applied to a specific interface.
3780 * sd-login gained a new API call sd_session_get_desktop() to
3781 query which desktop environment has been selected for a
3784 * UNIX utmp support is now compile-time optional to support
3785 legacy-free systems.
3787 * systemctl gained two new commands "add-wants" and
3788 "add-requires" for pulling in units from specific targets
3791 * If the word "rescue" is specified on the kernel command line
3792 the system will now boot into rescue mode (aka
3793 rescue.target), which was previously available only by
3794 specifying "1" or "systemd.unit=rescue.target" on the kernel
3795 command line. This new kernel command line option nicely
3796 mirrors the already existing "emergency" kernel command line
3799 * New kernel command line options mount.usr=, mount.usrflags=,
3800 mount.usrfstype= have been added that match root=, rootflags=,
3801 rootfstype= but allow mounting a specific file system to
3804 * The $NOTIFY_SOCKET is now also passed to control processes of
3805 services, not only the main process.
3807 * This version reenables support for fsck's -l switch. This
3808 means at least version v2.25 of util-linux is required for
3809 operation, otherwise dead-locks on device nodes may
3810 occur. Again: you need to update util-linux to at least
3811 v2.25 when updating systemd to v217.
3813 * The "multi-seat-x" tool has been removed from systemd, as
3814 its functionality has been integrated into X servers 1.16,
3815 and the tool is hence redundant. It is recommended to update
3816 display managers invoking this tool to simply invoke X
3817 directly from now on, again.
3819 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3820 message flag has been added for all of systemd's PolicyKit
3821 authenticated method calls has been added. In particular
3822 this now allows optional interactive authorization via
3823 PolicyKit for many of PID1's privileged operations such as
3824 unit file enabling and disabling.
3826 * "udevadm hwdb --update" learnt a new switch "--usr" for
3827 placing the rebuilt hardware database in /usr instead of
3828 /etc. When used only hardware database entries stored in
3829 /usr will be used, and any user database entries in /etc are
3830 ignored. This functionality is useful for vendors to ship a
3831 pre-built database on systems where local configuration is
3832 unnecessary or unlikely.
3834 * Calendar time specifications in .timer units now also
3835 understand the strings "semi-annually", "quarterly" and
3836 "minutely" as shortcuts (in addition to the preexisting
3837 "anually", "hourly", ...).
3839 * systemd-tmpfiles will now correctly create files in /dev
3840 at boot which are marked for creation only at boot. It is
3841 recommended to always create static device nodes with 'c!'
3842 and 'b!', so that they are created only at boot and not
3843 overwritten at runtime.
3845 * When the watchdog logic is used for a service (WatchdogSec=)
3846 and the watchdog timeout is hit the service will now be
3847 terminated with SIGABRT (instead of just SIGTERM), in order
3848 to make sure a proper coredump and backtrace is
3849 generated. This ensures that hanging services will result in
3850 similar coredump/backtrace behaviour as services that hit a
3853 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3854 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3855 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3856 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3857 Herrmann, David Sommerseth, David Strauss, Emil Renner
3858 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3859 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3860 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3861 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3862 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3863 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3864 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3865 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3866 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3867 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3868 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3869 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3870 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3871 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3872 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3873 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3876 — Berlin, 2014-10-28
3880 * timedated no longer reads NTP implementation unit names from
3881 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3882 implementations should add a
3884 Conflicts=systemd-timesyncd.service
3886 to their unit files to take over and replace systemd's NTP
3887 default functionality.
3889 * systemd-sysusers gained a new line type "r" for configuring
3890 which UID/GID ranges to allocate system users/groups
3891 from. Lines of type "u" may now add an additional column
3892 that specifies the home directory for the system user to be
3893 created. Also, systemd-sysusers may now optionally read user
3894 information from STDIN instead of a file. This is useful for
3895 invoking it from RPM preinst scriptlets that need to create
3896 users before the first RPM file is installed since these
3897 files might need to be owned by them. A new
3898 %sysusers_create_inline RPM macro has been introduced to do
3899 just that. systemd-sysusers now updates the shadow files as
3900 well as the user/group databases, which should enhance
3901 compatibility with certain tools like grpck.
3903 * A number of bus APIs of PID 1 now optionally consult
3904 PolicyKit to permit access for otherwise unprivileged
3905 clients under certain conditions. Note that this currently
3906 doesn't support interactive authentication yet, but this is
3907 expected to be added eventually, too.
3909 * /etc/machine-info now has new fields for configuring the
3910 deployment environment of the machine, as well as the
3911 location of the machine. hostnamectl has been updated with
3912 new command to update these fields.
3914 * systemd-timesyncd has been updated to automatically acquire
3915 NTP server information from systemd-networkd, which might
3916 have been discovered via DHCP.
3918 * systemd-resolved now includes a caching DNS stub resolver
3919 and a complete LLMNR name resolution implementation. A new
3920 NSS module "nss-resolve" has been added which can be used
3921 instead of glibc's own "nss-dns" to resolve hostnames via
3922 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3923 be resolved via systemd-resolved D-Bus APIs. In contrast to
3924 the glibc internal resolver systemd-resolved is aware of
3925 multi-homed system, and keeps DNS server and caches separate
3926 and per-interface. Queries are sent simultaneously on all
3927 interfaces that have DNS servers configured, in order to
3928 properly handle VPNs and local LANs which might resolve
3929 separate sets of domain names. systemd-resolved may acquire
3930 DNS server information from systemd-networkd automatically,
3931 which in turn might have discovered them via DHCP. A tool
3932 "systemd-resolve-host" has been added that may be used to
3933 query the DNS logic in resolved. systemd-resolved implements
3934 IDNA and automatically uses IDNA or UTF-8 encoding depending
3935 on whether classic DNS or LLMNR is used as transport. In the
3936 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3937 implementation to systemd-resolved.
3939 * A new NSS module nss-mymachines has been added, that
3940 automatically resolves the names of all local registered
3941 containers to their respective IP addresses.
3943 * A new client tool "networkctl" for systemd-networkd has been
3944 added. It currently is entirely passive and will query
3945 networking configuration from udev, rtnetlink and networkd,
3946 and present it to the user in a very friendly
3947 way. Eventually, we hope to extend it to become a full
3948 control utility for networkd.
3950 * .socket units gained a new DeferAcceptSec= setting that
3951 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3952 TCP. Similarly, support for controlling TCP keep-alive
3953 settings has been added (KeepAliveTimeSec=,
3954 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3955 turning off Nagle's algorithm on TCP has been added
3958 * logind learned a new session type "web", for use in projects
3959 like Cockpit which register web clients as PAM sessions.
3961 * timer units with at least one OnCalendar= setting will now
3962 be started only after time-sync.target has been
3963 reached. This way they will not elapse before the system
3964 clock has been corrected by a local NTP client or
3965 similar. This is particular useful on RTC-less embedded
3966 machines, that come up with an invalid system clock.
3968 * systemd-nspawn's --network-veth= switch should now result in
3969 stable MAC addresses for both the outer and the inner side
3972 * systemd-nspawn gained a new --volatile= switch for running
3973 container instances with /etc or /var unpopulated.
3975 * The kdbus client code has been updated to use the new Linux
3976 3.17 memfd subsystem instead of the old kdbus-specific one.
3978 * systemd-networkd's DHCP client and server now support
3979 FORCERENEW. There are also new configuration options to
3980 configure the vendor client identifier and broadcast mode
3983 * systemd will no longer inform the kernel about the current
3984 timezone, as this is necessarily incorrect and racy as the
3985 kernel has no understanding of DST and similar
3986 concepts. This hence means FAT timestamps will be always
3987 considered UTC, similar to what Android is already
3988 doing. Also, when the RTC is configured to the local time
3989 (rather than UTC) systemd will never synchronize back to it,
3990 as this might confuse Windows at a later boot.
3992 * systemd-analyze gained a new command "verify" for offline
3993 validation of unit files.
3995 * systemd-networkd gained support for a couple of additional
3996 settings for bonding networking setups. Also, the metric for
3997 statically configured routes may now be configured. For
3998 network interfaces where this is appropriate the peer IP
3999 address may now be configured.
4001 * systemd-networkd's DHCP client will no longer request
4002 broadcasting by default, as this tripped up some networks.
4003 For hardware where broadcast is required the feature should
4004 be switched back on using RequestBroadcast=yes.
4006 * systemd-networkd will now set up IPv4LL addresses (when
4007 enabled) even if DHCP is configured successfully.
4009 * udev will now default to respect network device names given
4010 by the kernel when the kernel indicates that these are
4011 predictable. This behavior can be tweaked by changing
4012 NamePolicy= in the relevant .link file.
4014 * A new library systemd-terminal has been added that
4015 implements full TTY stream parsing and rendering. This
4016 library is supposed to be used later on for implementing a
4017 full userspace VT subsystem, replacing the current kernel
4020 * A new tool systemd-journal-upload has been added to push
4021 journal data to a remote system running
4022 systemd-journal-remote.
4024 * journald will no longer forward all local data to another
4025 running syslog daemon. This change has been made because
4026 rsyslog (which appears to be the most commonly used syslog
4027 implementation these days) no longer makes use of this, and
4028 instead pulls the data out of the journal on its own. Since
4029 forwarding the messages to a non-existent syslog server is
4030 more expensive than we assumed we have now turned this
4031 off. If you run a syslog server that is not a recent rsyslog
4032 version, you have to turn this option on again
4033 (ForwardToSyslog= in journald.conf).
4035 * journald now optionally supports the LZ4 compressor for
4036 larger journal fields. This compressor should perform much
4037 better than XZ which was the previous default.
4039 * machinectl now shows the IP addresses of local containers,
4040 if it knows them, plus the interface name of the container.
4042 * A new tool "systemd-escape" has been added that makes it
4043 easy to escape strings to build unit names and similar.
4045 * sd_notify() messages may now include a new ERRNO= field
4046 which is parsed and collected by systemd and shown among the
4047 "systemctl status" output for a service.
4049 * A new component "systemd-firstboot" has been added that
4050 queries the most basic systemd information (timezone,
4051 hostname, root password) interactively on first
4052 boot. Alternatively it may also be used to provision these
4053 things offline on OS images installed into directories.
4055 * The default sysctl.d/ snippets will now set
4057 net.ipv4.conf.default.promote_secondaries=1
4059 This has the benefit of no flushing secondary IP addresses
4060 when primary addresses are removed.
4062 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4063 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4064 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4065 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4066 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4067 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4068 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4069 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4070 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4071 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4072 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4073 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4074 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4075 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4076 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4078 — Berlin, 2014-08-19
4082 * A new tool systemd-sysusers has been added. This tool
4083 creates system users and groups in /etc/passwd and
4084 /etc/group, based on static declarative system user/group
4085 definitions in /usr/lib/sysusers.d/. This is useful to
4086 enable factory resets and volatile systems that boot up with
4087 an empty /etc directory, and thus need system users and
4088 groups created during early boot. systemd now also ships
4089 with two default sysusers.d/ files for the most basic
4090 users and groups systemd and the core operating system
4093 * A new tmpfiles snippet has been added that rebuilds the
4094 essential files in /etc on boot, should they be missing.
4096 * A directive for ensuring automatic clean-up of
4097 /var/cache/man/ has been removed from the default
4098 configuration. This line should now be shipped by the man
4099 implementation. The necessary change has been made to the
4100 man-db implementation. Note that you need to update your man
4101 implementation to one that ships this line, otherwise no
4102 automatic clean-up of /var/cache/man will take place.
4104 * A new condition ConditionNeedsUpdate= has been added that
4105 may conditionalize services to only run when /etc or /var
4106 are "older" than the vendor operating system resources in
4107 /usr. This is useful for reconstructing or updating /etc
4108 after an offline update of /usr or a factory reset, on the
4109 next reboot. Services that want to run once after such an
4110 update or reset should use this condition and order
4111 themselves before the new systemd-update-done.service, which
4112 will mark the two directories as fully updated. A number of
4113 service files have been added making use of this, to rebuild
4114 the udev hardware database, the journald message catalog and
4115 dynamic loader cache (ldconfig). The systemd-sysusers tool
4116 described above also makes use of this now. With this in
4117 place it is now possible to start up a minimal operating
4118 system with /etc empty cleanly. For more information on the
4119 concepts involved see this recent blog story:
4121 http://0pointer.de/blog/projects/stateless.html
4123 * A new system group "input" has been introduced, and all
4124 input device nodes get this group assigned. This is useful
4125 for system-level software to get access to input devices. It
4126 complements what is already done for "audio" and "video".
4128 * systemd-networkd learnt minimal DHCPv4 server support in
4129 addition to the existing DHCPv4 client support. It also
4130 learnt DHCPv6 client and IPv6 Router Solicitation client
4131 support. The DHCPv4 client gained support for static routes
4132 passed in from the server. Note that the [DHCPv4] section
4133 known in older systemd-networkd versions has been renamed to
4134 [DHCP] and is now also used by the DHCPv6 client. Existing
4135 .network files using settings of this section should be
4136 updated, though compatibility is maintained. Optionally, the
4137 client hostname may now be sent to the DHCP server.
4139 * networkd gained support for vxlan virtual networks as well
4140 as tun/tap and dummy devices.
4142 * networkd gained support for automatic allocation of address
4143 ranges for interfaces from a system-wide pool of
4144 addresses. This is useful for dynamically managing a large
4145 number of interfaces with a single network configuration
4146 file. In particular this is useful to easily assign
4147 appropriate IP addresses to the veth links of a large number
4148 of nspawn instances.
4150 * RPM macros for processing sysusers, sysctl and binfmt
4151 drop-in snippets at package installation time have been
4154 * The /etc/os-release file should now be placed in
4155 /usr/lib/os-release. The old location is automatically
4156 created as symlink. /usr/lib is the more appropriate
4157 location of this file, since it shall actually describe the
4158 vendor operating system shipped in /usr, and not the
4159 configuration stored in /etc.
4161 * .mount units gained a new boolean SloppyOptions= setting
4162 that maps to mount(8)'s -s option which enables permissive
4163 parsing of unknown mount options.
4165 * tmpfiles learnt a new "L+" directive which creates a symlink
4166 but (unlike "L") deletes a pre-existing file first, should
4167 it already exist and not already be the correct
4168 symlink. Similarly, "b+", "c+" and "p+" directives have been
4169 added as well, which create block and character devices, as
4170 well as fifos in the filesystem, possibly removing any
4171 pre-existing files of different types.
4173 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4174 'argument' field (which so far specified the source to
4175 symlink/copy the files from) is now optional. If omitted the
4176 same file os copied from /usr/share/factory/ suffixed by the
4177 full destination path. This is useful for populating /etc
4178 with essential files, by copying them from vendor defaults
4179 shipped in /usr/share/factory/etc.
4181 * A new command "systemctl preset-all" has been added that
4182 applies the service preset settings to all installed unit
4183 files. A new switch --preset-mode= has been added that
4184 controls whether only enable or only disable operations
4187 * A new command "systemctl is-system-running" has been added
4188 that allows checking the overall state of the system, for
4189 example whether it is fully up and running.
4191 * When the system boots up with an empty /etc, the equivalent
4192 to "systemctl preset-all" is executed during early boot, to
4193 make sure all default services are enabled after a factory
4196 * systemd now contains a minimal preset file that enables the
4197 most basic services systemd ships by default.
4199 * Unit files' [Install] section gained a new DefaultInstance=
4200 field for defining the default instance to create if a
4201 template unit is enabled with no instance specified.
4203 * A new passive target cryptsetup-pre.target has been added
4204 that may be used by services that need to make they run and
4205 finish before the first LUKS cryptographic device is set up.
4207 * The /dev/loop-control and /dev/btrfs-control device nodes
4208 are now owned by the "disk" group by default, opening up
4209 access to this group.
4211 * systemd-coredump will now automatically generate a
4212 stack trace of all core dumps taking place on the system,
4213 based on elfutils' libdw library. This stack trace is logged
4216 * systemd-coredump may now optionally store coredumps directly
4217 on disk (in /var/lib/systemd/coredump, possibly compressed),
4218 instead of storing them unconditionally in the journal. This
4219 mode is the new default. A new configuration file
4220 /etc/systemd/coredump.conf has been added to configure this
4221 and other parameters of systemd-coredump.
4223 * coredumpctl gained a new "info" verb to show details about a
4224 specific coredump. A new switch "-1" has also been added
4225 that makes sure to only show information about the most
4226 recent entry instead of all entries. Also, as the tool is
4227 generally useful now the "systemd-" prefix of the binary
4228 name has been removed. Distributions that want to maintain
4229 compatibility with the old name should add a symlink from
4230 the old name to the new name.
4232 * journald's SplitMode= now defaults to "uid". This makes sure
4233 that unprivileged users can access their own coredumps with
4234 coredumpctl without restrictions.
4236 * New kernel command line options "systemd.wants=" (for
4237 pulling an additional unit during boot), "systemd.mask="
4238 (for masking a specific unit for the boot), and
4239 "systemd.debug-shell" (for enabling the debug shell on tty9)
4240 have been added. This is implemented in the new generator
4241 "systemd-debug-generator".
4243 * systemd-nspawn will now by default filter a couple of
4244 syscalls for containers, among them those required for
4245 kernel module loading, direct x86 IO port access, swap
4246 management, and kexec. Most importantly though
4247 open_by_handle_at() is now prohibited for containers,
4248 closing a hole similar to a recently discussed vulnerability
4249 in docker regarding access to files on file hierarchies the
4250 container should normally not have access to. Note that, for
4251 nspawn, we generally make no security claims anyway (and
4252 this is explicitly documented in the man page), so this is
4253 just a fix for one of the most obvious problems.
4255 * A new man page file-hierarchy(7) has been added that
4256 contains a minimized, modernized version of the file system
4257 layout systemd expects, similar in style to the FHS
4258 specification or hier(5). A new tool systemd-path(1) has
4259 been added to query many of these paths for the local
4262 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4263 longer done. Since the directory now has a per-user size
4264 limit, and is cleaned on logout this appears unnecessary,
4265 in particular since this now brings the lifecycle of this
4266 directory closer in line with how IPC objects are handled.
4268 * systemd.pc now exports a number of additional directories,
4269 including $libdir (which is useful to identify the library
4270 path for the primary architecture of the system), and a
4271 couple of drop-in directories.
4273 * udev's predictable network interface names now use the dev_port
4274 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4275 distinguish between ports of the same PCI function. dev_id should
4276 only be used for ports using the same HW address, hence the need
4279 * machined has been updated to export the OS version of a
4280 container (read from /etc/os-release and
4281 /usr/lib/os-release) on the bus. This is now shown in
4282 "machinectl status" for a machine.
4284 * A new service setting RestartForceExitStatus= has been
4285 added. If configured to a set of exit signals or process
4286 return values, the service will be restarted when the main
4287 daemon process exits with any of them, regardless of the
4290 * systemctl's -H switch for connecting to remote systemd
4291 machines has been extended so that it may be used to
4292 directly connect to a specific container on the
4293 host. "systemctl -H root@foobar:waldi" will now connect as
4294 user "root" to host "foobar", and then proceed directly to
4295 the container named "waldi". Note that currently you have to
4296 authenticate as user "root" for this to work, as entering
4297 containers is a privileged operation.
4299 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4300 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4301 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4302 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4303 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4304 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4305 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4306 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4307 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4308 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4309 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4310 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4312 — Berlin, 2014-07-03
4316 * As an experimental feature, udev now tries to lock the
4317 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4318 executes events for the disk or any of its partitions.
4319 Applications like partitioning programs can lock the
4320 disk device node (flock(LOCK_EX)) and claim temporary
4321 device ownership that way; udev will entirely skip all event
4322 handling for this disk and its partitions. If the disk
4323 was opened for writing, the close will trigger a partition
4324 table rescan in udev's "watch" facility, and if needed
4325 synthesize "change" events for the disk and all its partitions.
4326 This is now unconditionally enabled, and if it turns out to
4327 cause major problems, we might turn it on only for specific
4328 devices, or might need to disable it entirely. Device Mapper
4329 devices are excluded from this logic.
4331 * We temporarily dropped the "-l" switch for fsck invocations,
4332 since they collide with the flock() logic above. util-linux
4333 upstream has been changed already to avoid this conflict,
4334 and we will readd "-l" as soon as util-linux with this
4335 change has been released.
4337 * The dependency on libattr has been removed. Since a long
4338 time, the extended attribute calls have moved to glibc, and
4339 libattr is thus unnecessary.
4341 * Virtualization detection works without privileges now. This
4342 means the systemd-detect-virt binary no longer requires
4343 CAP_SYS_PTRACE file capabilities, and our daemons can run
4344 with fewer privileges.
4346 * systemd-networkd now runs under its own "systemd-network"
4347 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4348 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4349 loses the ability to write to files owned by root this way.
4351 * Similarly, systemd-resolved now runs under its own
4352 "systemd-resolve" user with no capabilities remaining.
4354 * Similarly, systemd-bus-proxyd now runs under its own
4355 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4357 * systemd-networkd gained support for setting up "veth"
4358 virtual Ethernet devices for container connectivity, as well
4359 as GRE and VTI tunnels.
4361 * systemd-networkd will no longer automatically attempt to
4362 manually load kernel modules necessary for certain tunnel
4363 transports. Instead, it is assumed the kernel loads them
4364 automatically when required. This only works correctly on
4365 very new kernels. On older kernels, please consider adding
4366 the kernel modules to /etc/modules-load.d/ as a work-around.
4368 * The resolv.conf file systemd-resolved generates has been
4369 moved to /run/systemd/resolve/. If you have a symlink from
4370 /etc/resolv.conf, it might be necessary to correct it.
4372 * Two new service settings, ProtectHome= and ProtectSystem=,
4373 have been added. When enabled, they will make the user data
4374 (such as /home) inaccessible or read-only and the system
4375 (such as /usr) read-only, for specific services. This allows
4376 very light-weight per-service sandboxing to avoid
4377 modifications of user data or system files from
4378 services. These two new switches have been enabled for all
4379 of systemd's long-running services, where appropriate.
4381 * Socket units gained new SocketUser= and SocketGroup=
4382 settings to set the owner user and group of AF_UNIX sockets
4383 and FIFOs in the file system.
4385 * Socket units gained a new RemoveOnStop= setting. If enabled,
4386 all FIFOS and sockets in the file system will be removed
4387 when the specific socket unit is stopped.
4389 * Socket units gained a new Symlinks= setting. It takes a list
4390 of symlinks to create to file system sockets or FIFOs
4391 created by the specific Unix sockets. This is useful to
4392 manage symlinks to socket nodes with the same life-cycle as
4395 * The /dev/log socket and /dev/initctl FIFO have been moved to
4396 /run, and have been replaced by symlinks. This allows
4397 connecting to these facilities even if PrivateDevices=yes is
4398 used for a service (which makes /dev/log itself unavailable,
4399 but /run is left). This also has the benefit of ensuring
4400 that /dev only contains device nodes, directories and
4401 symlinks, and nothing else.
4403 * sd-daemon gained two new calls sd_pid_notify() and
4404 sd_pid_notifyf(). They are similar to sd_notify() and
4405 sd_notifyf(), but allow overriding of the source PID of
4406 notification messages if permissions permit this. This is
4407 useful to send notify messages on behalf of a different
4408 process (for example, the parent process). The
4409 systemd-notify tool has been updated to make use of this
4410 when sending messages (so that notification messages now
4411 originate from the shell script invoking systemd-notify and
4412 not the systemd-notify process itself. This should minimize
4413 a race where systemd fails to associate notification
4414 messages to services when the originating process already
4417 * A new "on-abnormal" setting for Restart= has been added. If
4418 set, it will result in automatic restarts on all "abnormal"
4419 reasons for a process to exit, which includes unclean
4420 signals, core dumps, timeouts and watchdog timeouts, but
4421 does not include clean and unclean exit codes or clean
4422 signals. Restart=on-abnormal is an alternative for
4423 Restart=on-failure for services that shall be able to
4424 terminate and avoid restarts on certain errors, by
4425 indicating so with an unclean exit code. Restart=on-failure
4426 or Restart=on-abnormal is now the recommended setting for
4427 all long-running services.
4429 * If the InaccessibleDirectories= service setting points to a
4430 mount point (or if there are any submounts contained within
4431 it), it is now attempted to completely unmount it, to make
4432 the file systems truly unavailable for the respective
4435 * The ReadOnlyDirectories= service setting and
4436 systemd-nspawn's --read-only parameter are now recursively
4437 applied to all submounts, too.
4439 * Mount units may now be created transiently via the bus APIs.
4441 * The support for SysV and LSB init scripts has been removed
4442 from the systemd daemon itself. Instead, it is now
4443 implemented as a generator that creates native systemd units
4444 from these scripts when needed. This enables us to remove a
4445 substantial amount of legacy code from PID 1, following the
4446 fact that many distributions only ship a very small number
4447 of LSB/SysV init scripts nowadays.
4449 * Privileged Xen (dom0) domains are not considered
4450 virtualization anymore by the virtualization detection
4451 logic. After all, they generally have unrestricted access to
4452 the hardware and usually are used to manage the unprivileged
4455 * systemd-tmpfiles gained a new "C" line type, for copying
4456 files or entire directories.
4458 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4459 lines. So far, they have been non-globbing versions of the
4460 latter, and have thus been redundant. In future, it is
4461 recommended to only use "z". "m" has hence been removed
4462 from the documentation, even though it stays supported.
4464 * A tmpfiles snippet to recreate the most basic structure in
4465 /var has been added. This is enough to create the /var/run →
4466 /run symlink and create a couple of structural
4467 directories. This allows systems to boot up with an empty or
4468 volatile /var. Of course, while with this change, the core OS
4469 now is capable with dealing with a volatile /var, not all
4470 user services are ready for it. However, we hope that sooner
4471 or later, many service daemons will be changed upstream so
4472 that they are able to automatically create their necessary
4473 directories in /var at boot, should they be missing. This is
4474 the first step to allow state-less systems that only require
4475 the vendor image for /usr to boot.
4477 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4478 empty tmpfs instance to a specific directory. This is
4479 particularly useful for making use of the automatic
4480 reconstruction of /var (see above), by passing --tmpfs=/var.
4482 * Access modes specified in tmpfiles snippets may now be
4483 prefixed with "~", which indicates that they shall be masked
4484 by whether the existing file or directory is currently
4485 writable, readable or executable at all. Also, if specified,
4486 the sgid/suid/sticky bits will be masked for all
4489 * A new passive target unit "network-pre.target" has been
4490 added which is useful for services that shall run before any
4491 network is configured, for example firewall scripts.
4493 * The "floppy" group that previously owned the /dev/fd*
4494 devices is no longer used. The "disk" group is now used
4495 instead. Distributions should probably deprecate usage of
4498 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4499 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4500 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4501 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4502 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4503 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4506 — Berlin, 2014-06-11
4510 * A new "systemd-timesyncd" daemon has been added for
4511 synchronizing the system clock across the network. It
4512 implements an SNTP client. In contrast to NTP
4513 implementations such as chrony or the NTP reference server,
4514 this only implements a client side, and does not bother with
4515 the full NTP complexity, focusing only on querying time from
4516 one remote server and synchronizing the local clock to
4517 it. Unless you intend to serve NTP to networked clients or
4518 want to connect to local hardware clocks, this simple NTP
4519 client should be more than appropriate for most
4520 installations. The daemon runs with minimal privileges, and
4521 has been hooked up with networkd to only operate when
4522 network connectivity is available. The daemon saves the
4523 current clock to disk every time a new NTP sync has been
4524 acquired, and uses this to possibly correct the system clock
4525 early at bootup, in order to accommodate for systems that
4526 lack an RTC such as the Raspberry Pi and embedded devices,
4527 and to make sure that time monotonically progresses on these
4528 systems, even if it is not always correct. To make use of
4529 this daemon, a new system user and group "systemd-timesync"
4530 needs to be created on installation of systemd.
4532 * The queue "seqnum" interface of libudev has been disabled, as
4533 it was generally incompatible with device namespacing as
4534 sequence numbers of devices go "missing" if the devices are
4535 part of a different namespace.
4537 * "systemctl list-timers" and "systemctl list-sockets" gained
4538 a --recursive switch for showing units of these types also
4539 for all local containers, similar in style to the already
4540 supported --recursive switch for "systemctl list-units".
4542 * A new RebootArgument= setting has been added for service
4543 units, which may be used to specify a kernel reboot argument
4544 to use when triggering reboots with StartLimitAction=.
4546 * A new FailureAction= setting has been added for service
4547 units which may be used to specify an operation to trigger
4548 when a service fails. This works similarly to
4549 StartLimitAction=, but unlike it, controls what is done
4550 immediately rather than only after several attempts to
4551 restart the service in question.
4553 * hostnamed got updated to also expose the kernel name,
4554 release, and version on the bus. This is useful for
4555 executing commands like hostnamectl with the -H switch.
4556 systemd-analyze makes use of this to properly display
4557 details when running non-locally.
4559 * The bootchart tool can now show cgroup information in the
4560 graphs it generates.
4562 * The CFS CPU quota cgroup attribute is now exposed for
4563 services. The new CPUQuota= switch has been added for this
4564 which takes a percentage value. Setting this will have the
4565 result that a service may never get more CPU time than the
4566 specified percentage, even if the machine is otherwise idle.
4568 * systemd-networkd learned IPIP and SIT tunnel support.
4570 * LSB init scripts exposing a dependency on $network will now
4571 get a dependency on network-online.target rather than simply
4572 network.target. This should bring LSB handling closer to
4573 what it was on SysV systems.
4575 * A new fsck.repair= kernel option has been added to control
4576 how fsck shall deal with unclean file systems at boot.
4578 * The (.ini) configuration file parser will now silently
4579 ignore sections whose name begins with "X-". This may be
4580 used to maintain application-specific extension sections in unit
4583 * machined gained a new API to query the IP addresses of
4584 registered containers. "machinectl status" has been updated
4585 to show these addresses in its output.
4587 * A new call sd_uid_get_display() has been added to the
4588 sd-login APIs for querying the "primary" session of a
4589 user. The "primary" session of the user is elected from the
4590 user's sessions and generally a graphical session is
4591 preferred over a text one.
4593 * A minimal systemd-resolved daemon has been added. It
4594 currently simply acts as a companion to systemd-networkd and
4595 manages resolv.conf based on per-interface DNS
4596 configuration, possibly supplied via DHCP. In the long run
4597 we hope to extend this into a local DNSSEC enabled DNS and
4600 * The systemd-networkd-wait-online tool is now enabled by
4601 default. It will delay network-online.target until a network
4602 connection has been configured. The tool primarily integrates
4603 with networkd, but will also make a best effort to make sense
4604 of network configuration performed in some other way.
4606 * Two new service options StartupCPUShares= and
4607 StartupBlockIOWeight= have been added that work similarly to
4608 CPUShares= and BlockIOWeight= however only apply during
4609 system startup. This is useful to prioritize certain services
4610 differently during bootup than during normal runtime.
4612 * hostnamed has been changed to prefer the statically
4613 configured hostname in /etc/hostname (unless set to
4614 'localhost' or empty) over any dynamic one supplied by
4615 dhcp. With this change, the rules for picking the hostname
4616 match more closely the rules of other configuration settings
4617 where the local administrator's configuration in /etc always
4618 overrides any other settings.
4620 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4621 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4622 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4623 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4624 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4625 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4626 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4627 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4628 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4629 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4630 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4631 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4632 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4633 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4634 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4635 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4638 — Beijing, 2014-05-28
4642 * When restoring the screen brightness at boot, stay away from
4643 the darkest setting or from the lowest 5% of the available
4644 range, depending on which is the larger value of both. This
4645 should effectively protect the user from rebooting into a
4646 black screen, should the brightness have been set to minimum
4649 * sd-login gained a new sd_machine_get_class() call to
4650 determine the class ("vm" or "container") of a machine
4651 registered with machined.
4653 * sd-login gained new calls
4654 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4655 to query the identity of the peer of a local AF_UNIX
4656 connection. They operate similarly to their sd_pid_get_xyz()
4659 * PID 1 will now maintain a system-wide system state engine
4660 with the states "starting", "running", "degraded",
4661 "maintenance", "stopping". These states are bound to system
4662 startup, normal runtime, runtime with at least one failed
4663 service, rescue/emergency mode and system shutdown. This
4664 state is shown in the "systemctl status" output when no unit
4665 name is passed. It is useful to determine system state, in
4666 particularly when doing so for many systems or containers at
4669 * A new command "list-machines" has been added to "systemctl"
4670 that lists all local OS containers and shows their system
4671 state (see above), if systemd runs inside of them.
4673 * systemctl gained a new "-r" switch to recursively enumerate
4674 units on all local containers, when used with the
4675 "list-unit" command (which is the default one that is
4676 executed when no parameters are specified).
4678 * The GPT automatic partition discovery logic will now honour
4679 two GPT partition flags: one may be set on a partition to
4680 cause it to be mounted read-only, and the other may be set
4681 on a partition to ignore it during automatic discovery.
4683 * Two new GPT type UUIDs have been added for automatic root
4684 partition discovery, for 32-bit and 64-bit ARM. This is not
4685 particularly useful for discovering the root directory on
4686 these architectures during bare-metal boots (since UEFI is
4687 not common there), but still very useful to allow booting of
4688 ARM disk images in nspawn with the -i option.
4690 * MAC addresses of interfaces created with nspawn's
4691 --network-interface= switch will now be generated from the
4692 machine name, and thus be stable between multiple invocations
4695 * logind will now automatically remove all IPC objects owned
4696 by a user if she or he fully logs out. This makes sure that
4697 users who are logged out cannot continue to consume IPC
4698 resources. This covers SysV memory, semaphores and message
4699 queues as well as POSIX shared memory and message
4700 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4701 limits. With this functionality, that is corrected. This may
4702 be turned off by using the RemoveIPC= switch of logind.conf.
4704 * The systemd-machine-id-setup and tmpfiles tools gained a
4705 --root= switch to operate on a specific root directory,
4708 * journald can now forward logged messages to the TTYs of all
4709 logged in users ("wall"). This is the default for all
4710 emergency messages now.
4712 * A new tool systemd-journal-remote has been added to stream
4713 journal log messages across the network.
4715 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4716 controller trees are mounted into it. Note that the
4717 directories mounted beneath it are not read-only. This is a
4718 security measure and is particularly useful because glibc
4719 actually includes a search logic to pick any tmpfs it can
4720 find to implement shm_open() if /dev/shm is not available
4721 (which it might very well be in namespaced setups).
4723 * machinectl gained a new "poweroff" command to cleanly power
4724 down a local OS container.
4726 * The PrivateDevices= unit file setting will now also drop the
4727 CAP_MKNOD capability from the capability bound set, and
4728 imply DevicePolicy=closed.
4730 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4731 comprehensively on all long-running systemd services where
4732 this is appropriate.
4734 * systemd-udevd will now run in a disassociated mount
4735 namespace. To mount directories from udev rules, make sure to
4736 pull in mount units via SYSTEMD_WANTS properties.
4738 * The kdbus support gained support for uploading policy into
4739 the kernel. sd-bus gained support for creating "monitoring"
4740 connections that can eavesdrop into all bus communication
4741 for debugging purposes.
4743 * Timestamps may now be specified in seconds since the UNIX
4744 epoch Jan 1st, 1970 by specifying "@" followed by the value
4747 * Native tcpwrap support in systemd has been removed. tcpwrap
4748 is old code, not really maintained anymore and has serious
4749 shortcomings, and better options such as firewalls
4750 exist. For setups that require tcpwrap usage, please
4751 consider invoking your socket-activated service via tcpd,
4752 like on traditional inetd.
4754 * A new system.conf configuration option
4755 DefaultTimerAccuracySec= has been added that controls the
4756 default AccuracySec= setting of .timer units.
4758 * Timer units gained a new WakeSystem= switch. If enabled,
4759 timers configured this way will cause the system to resume
4760 from system suspend (if the system supports that, which most
4763 * Timer units gained a new Persistent= switch. If enabled,
4764 timers configured this way will save to disk when they have
4765 been last triggered. This information is then used on next
4766 reboot to possible execute overdue timer events, that
4767 could not take place because the system was powered off.
4768 This enables simple anacron-like behaviour for timer units.
4770 * systemctl's "list-timers" will now also list the time a
4771 timer unit was last triggered in addition to the next time
4772 it will be triggered.
4774 * systemd-networkd will now assign predictable IPv4LL
4775 addresses to its local interfaces.
4777 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4778 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4779 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4780 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4781 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4782 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4783 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4784 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4787 — Berlin, 2014-03-25
4791 * A new unit file setting RestrictAddressFamilies= has been
4792 added to restrict which socket address families unit
4793 processes gain access to. This takes address family names
4794 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4795 attack surface of services via exotic protocol stacks. This
4796 is built on seccomp system call filters.
4798 * Two new unit file settings RuntimeDirectory= and
4799 RuntimeDirectoryMode= have been added that may be used to
4800 manage a per-daemon runtime directories below /run. This is
4801 an alternative for setting up directory permissions with
4802 tmpfiles snippets, and has the advantage that the runtime
4803 directory's lifetime is bound to the daemon runtime and that
4804 the daemon starts up with an empty directory each time. This
4805 is particularly useful when writing services that drop
4806 privileges using the User= or Group= setting.
4808 * The DeviceAllow= unit setting now supports globbing for
4809 matching against device group names.
4811 * The systemd configuration file system.conf gained new
4812 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4813 DefaultMemoryAccounting= to globally turn on/off accounting
4814 for specific resources (cgroups) for all units. These
4815 settings may still be overridden individually in each unit
4818 * systemd-gpt-auto-generator is now able to discover /srv and
4819 root partitions in addition to /home and swap partitions. It
4820 also supports LUKS-encrypted partitions now. With this in
4821 place, automatic discovery of partitions to mount following
4822 the Discoverable Partitions Specification
4823 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4824 is now a lot more complete. This allows booting without
4825 /etc/fstab and without root= on the kernel command line on
4826 systems prepared appropriately.
4828 * systemd-nspawn gained a new --image= switch which allows
4829 booting up disk images and Linux installations on any block
4830 device that follow the Discoverable Partitions Specification
4831 (see above). This means that installations made with
4832 appropriately updated installers may now be started and
4833 deployed using container managers, completely
4834 unmodified. (We hope that libvirt-lxc will add support for
4835 this feature soon, too.)
4837 * systemd-nspawn gained a new --network-macvlan= setting to
4838 set up a private macvlan interface for the
4839 container. Similarly, systemd-networkd gained a new
4840 Kind=macvlan setting in .netdev files.
4842 * systemd-networkd now supports configuring local addresses
4845 * A new tool systemd-network-wait-online has been added to
4846 synchronously wait for network connectivity using
4849 * The sd-bus.h bus API gained a new sd_bus_track object for
4850 tracking the life-cycle of bus peers. Note that sd-bus.h is
4851 still not a public API though (unless you specify
4852 --enable-kdbus on the configure command line, which however
4853 voids your warranty and you get no API stability guarantee).
4855 * The $XDG_RUNTIME_DIR runtime directories for each user are
4856 now individual tmpfs instances, which has the benefit of
4857 introducing separate pools for each user, with individual
4858 size limits, and thus making sure that unprivileged clients
4859 can no longer negatively impact the system or other users by
4860 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4861 RuntimeDirectorySize= has been introduced that allows
4862 controlling the default size limit for all users. It
4863 defaults to 10% of the available physical memory. This is no
4864 replacement for quotas on tmpfs though (which the kernel
4865 still does not support), as /dev/shm and /tmp are still
4866 shared resources used by both the system and unprivileged
4869 * logind will now automatically turn off automatic suspending
4870 on laptop lid close when more than one display is
4871 connected. This was previously expected to be implemented
4872 individually in desktop environments (such as GNOME),
4873 however has been added to logind now, in order to fix a
4874 boot-time race where a desktop environment might not have
4875 been started yet and thus not been able to take an inhibitor
4876 lock at the time where logind already suspends the system
4877 due to a closed lid.
4879 * logind will now wait at least 30s after each system
4880 suspend/resume cycle, and 3min after system boot before
4881 suspending the system due to a closed laptop lid. This
4882 should give USB docking stations and similar enough time to
4883 be probed and configured after system resume and boot in
4884 order to then act as suspend blocker.
4886 * systemd-run gained a new --property= setting which allows
4887 initialization of resource control properties (and others)
4888 for the created scope or service unit. Example: "systemd-run
4889 --property=BlockIOWeight=10 updatedb" may be used to run
4890 updatedb at a low block IO scheduling weight.
4892 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4893 now also work in --scope mode.
4895 * When systemd is compiled with kdbus support, basic support
4896 for enforced policies is now in place. (Note that enabling
4897 kdbus still voids your warranty and no API compatibility
4900 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4901 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4902 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4903 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4904 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4905 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4906 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4907 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4908 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4909 Zbigniew Jędrzejewski-Szmek
4911 — Berlin, 2014-03-12
4915 * systemd will now relabel /dev after loading the SMACK policy
4916 according to SMACK rules.
4918 * A new unit file option AppArmorProfile= has been added to
4919 set the AppArmor profile for the processes of a unit.
4921 * A new condition check ConditionArchitecture= has been added
4922 to conditionalize units based on the system architecture, as
4923 reported by uname()'s "machine" field.
4925 * systemd-networkd now supports matching on the system
4926 virtualization, architecture, kernel command line, host name
4929 * logind is now a lot more aggressive when suspending the
4930 machine due to a closed laptop lid. Instead of acting only
4931 on the lid close action, it will continuously watch the lid
4932 status and act on it. This is useful for laptops where the
4933 power button is on the outside of the chassis so that it can
4934 be reached without opening the lid (such as the Lenovo
4935 Yoga). On those machines, logind will now immediately
4936 re-suspend the machine if the power button has been
4937 accidentally pressed while the laptop was suspended and in a
4938 backpack or similar.
4940 * logind will now watch SW_DOCK switches and inhibit reaction
4941 to the lid switch if it is pressed. This means that logind
4942 will not suspend the machine anymore if the lid is closed
4943 and the system is docked, if the laptop supports SW_DOCK
4944 notifications via the input layer. Note that ACPI docking
4945 stations do not generate this currently. Also note that this
4946 logic is usually not fully sufficient and Desktop
4947 Environments should take a lid switch inhibitor lock when an
4948 external display is connected, as systemd will not watch
4951 * nspawn will now make use of the devices cgroup controller by
4952 default, and only permit creation of and access to the usual
4953 API device nodes like /dev/null or /dev/random, as well as
4954 access to (but not creation of) the pty devices.
4956 * We will now ship a default .network file for
4957 systemd-networkd that automatically configures DHCP for
4958 network interfaces created by nspawn's --network-veth or
4959 --network-bridge= switches.
4961 * systemd will now understand the usual M, K, G, T suffixes
4962 according to SI conventions (i.e. to the base 1000) when
4963 referring to throughput and hardware metrics. It will stay
4964 with IEC conventions (i.e. to the base 1024) for software
4965 metrics, according to what is customary according to
4966 Wikipedia. We explicitly document which base applies for
4967 each configuration option.
4969 * The DeviceAllow= setting in unit files now supports a syntax
4970 to whitelist an entire group of devices node majors at once,
4971 based on the /proc/devices listing. For example, with the
4972 string "char-pts", it is now possible to whitelist all
4973 current and future pseudo-TTYs at once.
4975 * sd-event learned a new "post" event source. Event sources of
4976 this type are triggered by the dispatching of any event
4977 source of a type that is not "post". This is useful for
4978 implementing clean-up and check event sources that are
4979 triggered by other work being done in the program.
4981 * systemd-networkd is no longer statically enabled, but uses
4982 the usual [Install] sections so that it can be
4983 enabled/disabled using systemctl. It still is enabled by
4986 * When creating a veth interface pair with systemd-nspawn, the
4987 host side will now be prefixed with "vb-" if
4988 --network-bridge= is used, and with "ve-" if --network-veth
4989 is used. This way, it is easy to distinguish these cases on
4990 the host, for example to apply different configuration to
4991 them with systemd-networkd.
4993 * The compatibility libraries for libsystemd-journal.so,
4994 libsystem-id128.so, libsystemd-login.so and
4995 libsystemd-daemon.so do not make use of IFUNC
4996 anymore. Instead, we now build libsystemd.so multiple times
4997 under these alternative names. This means that the footprint
4998 is drastically increased, but given that these are
4999 transitional compatibility libraries, this should not matter
5000 much. This change has been made necessary to support the ARM
5001 platform for these compatibility libraries, as the ARM
5002 toolchain is not really at the same level as the toolchain
5003 for other architectures like x86 and does not support
5004 IFUNC. Please make sure to use --enable-compat-libs only
5005 during a transitional period!
5007 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5008 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5009 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5010 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5011 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5012 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5013 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5014 Zbigniew Jędrzejewski-Szmek
5016 — Berlin, 2014-02-24
5020 * A new component "systemd-networkd" has been added that can
5021 be used to configure local network interfaces statically or
5022 via DHCP. It is capable of bringing up bridges, VLANs, and
5023 bonding. Currently, no hook-ups for interactive network
5024 configuration are provided. Use this for your initrd,
5025 container, embedded, or server setup if you need a simple,
5026 yet powerful, network configuration solution. This
5027 configuration subsystem is quite nifty, as it allows wildcard
5028 hotplug matching in interfaces. For example, with a single
5029 configuration snippet, you can configure that all Ethernet
5030 interfaces showing up are automatically added to a bridge,
5031 or similar. It supports link-sensing and more.
5033 * A new tool "systemd-socket-proxyd" has been added which can
5034 act as a bidirectional proxy for TCP sockets. This is
5035 useful for adding socket activation support to services that
5036 do not actually support socket activation, including virtual
5037 machines and the like.
5039 * Add a new tool to save/restore rfkill state on
5042 * Save/restore state of keyboard backlights in addition to
5043 display backlights on shutdown/boot.
5045 * udev learned a new SECLABEL{} construct to label device
5046 nodes with a specific security label when they appear. For
5047 now, only SECLABEL{selinux} is supported, but the syntax is
5048 prepared for additional security frameworks.
5050 * udev gained a new scheme to configure link-level attributes
5051 from files in /etc/systemd/network/*.link. These files can
5052 match against MAC address, device path, driver name and type,
5053 and will apply attributes like the naming policy, link speed,
5054 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5055 address assignment policy (randomized, ...).
5057 * The configuration of network interface naming rules for
5058 "permanent interface names" has changed: a new NamePolicy=
5059 setting in the [Link] section of .link files determines the
5060 priority of possible naming schemes (onboard, slot, MAC,
5061 path). The default value of this setting is determined by
5062 /usr/lib/net/links/99-default.link. Old
5063 80-net-name-slot.rules udev configuration file has been
5064 removed, so local configuration overriding this file should
5065 be adapted to override 99-default.link instead.
5067 * When the User= switch is used in a unit file, also
5068 initialize $SHELL= based on the user database entry.
5070 * systemd no longer depends on libdbus. All communication is
5071 now done with sd-bus, systemd's low-level bus library
5074 * kdbus support has been added to PID 1 itself. When kdbus is
5075 enabled, this causes PID 1 to set up the system bus and
5076 enable support for a new ".busname" unit type that
5077 encapsulates bus name activation on kdbus. It works a little
5078 bit like ".socket" units, except for bus names. A new
5079 generator has been added that converts classic dbus1 service
5080 activation files automatically into native systemd .busname
5083 * sd-bus: add a light-weight vtable implementation that allows
5084 defining objects on the bus with a simple static const
5085 vtable array of its methods, signals and properties.
5087 * systemd will not generate or install static dbus
5088 introspection data anymore to /usr/share/dbus-1/interfaces,
5089 as the precise format of these files is unclear, and
5090 nothing makes use of it.
5092 * A proxy daemon is now provided to proxy clients connecting
5093 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5094 compatibility with classic D-Bus.
5096 * A bus driver implementation has been added that supports the
5097 classic D-Bus bus driver calls on kdbus, also for
5098 compatibility purposes.
5100 * A new API "sd-event.h" has been added that implements a
5101 minimal event loop API built around epoll. It provides a
5102 couple of features that direct epoll usage is lacking:
5103 prioritization of events, scales to large numbers of timer
5104 events, per-event timer slack (accuracy), system-wide
5105 coalescing of timer events, exit handlers, watchdog
5106 supervision support using systemd's sd_notify() API, child
5109 * A new API "sd-rntl.h" has been added that provides an API
5110 around the route netlink interface of the kernel, similar in
5111 style to "sd-bus.h".
5113 * A new API "sd-dhcp-client.h" has been added that provides a
5114 small DHCPv4 client-side implementation. This is used by
5117 * There is a new kernel command line option
5118 "systemd.restore_state=0|1". When set to "0", none of the
5119 systemd tools will restore saved runtime state to hardware
5120 devices. More specifically, the rfkill and backlight states
5123 * The FsckPassNo= compatibility option in mount/service units
5124 has been removed. The fstab generator will now add the
5125 necessary dependencies automatically, and does not require
5126 PID1's support for that anymore.
5128 * journalctl gained a new switch, --list-boots, that lists
5129 recent boots with their times and boot IDs.
5131 * The various tools like systemctl, loginctl, timedatectl,
5132 busctl, systemd-run, ... have gained a new switch "-M" to
5133 connect to a specific, local OS container (as direct
5134 connection, without requiring SSH). This works on any
5135 container that is registered with machined, such as those
5136 created by libvirt-lxc or nspawn.
5138 * systemd-run and systemd-analyze also gained support for "-H"
5139 to connect to remote hosts via SSH. This is particularly
5140 useful for systemd-run because it enables queuing of jobs
5141 onto remote systems.
5143 * machinectl gained a new command "login" to open a getty
5144 login in any local container. This works with any container
5145 that is registered with machined (such as those created by
5146 libvirt-lxc or nspawn), and which runs systemd inside.
5148 * machinectl gained a new "reboot" command that may be used to
5149 trigger a reboot on a specific container that is registered
5150 with machined. This works on any container that runs an init
5151 system of some kind.
5153 * systemctl gained a new "list-timers" command to print a nice
5154 listing of installed timer units with the times they elapse
5157 * Alternative reboot() parameters may now be specified on the
5158 "systemctl reboot" command line and are passed to the
5159 reboot() system call.
5161 * systemctl gained a new --job-mode= switch to configure the
5162 mode to queue a job with. This is a more generic version of
5163 --fail, --irreversible, and --ignore-dependencies, which are
5164 still available but not advertised anymore.
5166 * /etc/systemd/system.conf gained new settings to configure
5167 various default timeouts of units, as well as the default
5168 start limit interval and burst. These may still be overridden
5171 * PID1 will now export on the bus profile data of the security
5172 policy upload process (such as the SELinux policy upload to
5175 * journald: when forwarding logs to the console, include
5176 timestamps (following the setting in
5177 /sys/module/printk/parameters/time).
5179 * OnCalendar= in timer units now understands the special
5180 strings "yearly" and "annually". (Both are equivalent)
5182 * The accuracy of timer units is now configurable with the new
5183 AccuracySec= setting. It defaults to 1min.
5185 * A new dependency type JoinsNamespaceOf= has been added that
5186 allows running two services within the same /tmp and network
5187 namespace, if PrivateNetwork= or PrivateTmp= are used.
5189 * A new command "cat" has been added to systemctl. It outputs
5190 the original unit file of a unit, and concatenates the
5191 contents of additional "drop-in" unit file snippets, so that
5192 the full configuration is shown.
5194 * systemctl now supports globbing on the various "list-xyz"
5195 commands, like "list-units" or "list-sockets", as well as on
5196 those commands which take multiple unit names.
5198 * journalctl's --unit= switch gained support for globbing.
5200 * All systemd daemons now make use of the watchdog logic so
5201 that systemd automatically notices when they hang.
5203 * If the $container_ttys environment variable is set,
5204 getty-generator will automatically spawn a getty for each
5205 listed tty. This is useful for container managers to request
5206 login gettys to be spawned on as many ttys as needed.
5208 * %h, %s, %U specifier support is not available anymore when
5209 used in unit files for PID 1. This is because NSS calls are
5210 not safe from PID 1. They stay available for --user
5211 instances of systemd, and as special case for the root user.
5213 * loginctl gained a new "--no-legend" switch to turn off output
5216 * The "sd-login.h" API gained three new calls:
5217 sd_session_is_remote(), sd_session_get_remote_user(),
5218 sd_session_get_remote_host() to query information about
5221 * The udev hardware database now also carries vendor/product
5222 information of SDIO devices.
5224 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5225 determine whether watchdog notifications are requested by
5228 * Socket-activated per-connection services now include a
5229 short description of the connection parameters in the
5232 * tmpfiles gained a new "--boot" option. When this is not used,
5233 only lines where the command character is not suffixed with
5234 "!" are executed. When this option is specified, those
5235 options are executed too. This partitions tmpfiles
5236 directives into those that can be safely executed at any
5237 time, and those which should be run only at boot (for
5238 example, a line that creates /run/nologin).
5240 * A new API "sd-resolve.h" has been added which provides a simple
5241 asynchronous wrapper around glibc NSS host name resolution
5242 calls, such as getaddrinfo(). In contrast to glibc's
5243 getaddrinfo_a(), it does not use signals. In contrast to most
5244 other asynchronous name resolution libraries, this one does
5245 not reimplement DNS, but reuses NSS, so that alternate
5246 host name resolution systems continue to work, such as mDNS,
5247 LDAP, etc. This API is based on libasyncns, but it has been
5248 cleaned up for inclusion in systemd.
5250 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5251 "sd-daemon.h" are no longer found in individual libraries
5252 libsystemd-journal.so, libsystemd-login.so,
5253 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5254 merged them into a single library, libsystemd.so, which
5255 provides all symbols. The reason for this is cyclic
5256 dependencies, as these libraries tend to use each other's
5257 symbols. So far, we have managed to workaround that by linking
5258 a copy of a good part of our code into each of these
5259 libraries again and again, which, however, makes certain
5260 things hard to do, like sharing static variables. Also, it
5261 substantially increases footprint. With this change, there
5262 is only one library for the basic APIs systemd
5263 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5264 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5265 library as well, however are subject to the --enable-kdbus
5266 switch (see below). Note that "sd-dhcp-client.h" is not part
5267 of this library (this is because it only consumes, never
5268 provides, services of/to other APIs). To make the transition
5269 easy from the separate libraries to the unified one, we
5270 provide the --enable-compat-libs compile-time switch which
5271 will generate stub libraries that are compatible with the
5272 old ones but redirect all calls to the new one.
5274 * All of the kdbus logic and the new APIs "sd-bus.h",
5275 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5276 and "sd-utf8.h" are compile-time optional via the
5277 "--enable-kdbus" switch, and they are not compiled in by
5278 default. To make use of kdbus, you have to explicitly enable
5279 the switch. Note however, that neither the kernel nor the
5280 userspace API for all of this is considered stable yet. We
5281 want to maintain the freedom to still change the APIs for
5282 now. By specifying this build-time switch, you acknowledge
5283 that you are aware of the instability of the current
5286 * Also, note that while kdbus is pretty much complete,
5287 it lacks one thing: proper policy support. This means you
5288 can build a fully working system with all features; however,
5289 it will be highly insecure. Policy support will be added in
5290 one of the next releases, at the same time that we will
5291 declare the APIs stable.
5293 * When the kernel command line argument "kdbus" is specified,
5294 systemd will automatically load the kdbus.ko kernel module. At
5295 this stage of development, it is only useful for testing kdbus
5296 and should not be used in production. Note: if "--enable-kdbus"
5297 is specified, and the kdbus.ko kernel module is available, and
5298 "kdbus" is added to the kernel command line, the entire system
5299 runs with kdbus instead of dbus-daemon, with the above mentioned
5300 problem of missing the system policy enforcement. Also a future
5301 version of kdbus.ko or a newer systemd will not be compatible with
5302 each other, and will unlikely be able to boot the machine if only
5303 one of them is updated.
5305 * systemctl gained a new "import-environment" command which
5306 uploads the caller's environment (or parts thereof) into the
5307 service manager so that it is inherited by services started
5308 by the manager. This is useful to upload variables like
5309 $DISPLAY into the user service manager.
5311 * A new PrivateDevices= switch has been added to service units
5312 which allows running a service with a namespaced /dev
5313 directory that does not contain any device nodes for
5314 physical devices. More specifically, it only includes devices
5315 such as /dev/null, /dev/urandom, and /dev/zero which are API
5318 * logind has been extended to support behaviour like VT
5319 switching on seats that do not support a VT. This makes
5320 multi-session available on seats that are not the first seat
5321 (seat0), and on systems where kernel support for VTs has
5322 been disabled at compile-time.
5324 * If a process holds a delay lock for system sleep or shutdown
5325 and fails to release it in time, we will now log its
5326 identity. This makes it easier to identify processes that
5327 cause slow suspends or power-offs.
5329 * When parsing /etc/crypttab, support for a new key-slot=
5330 option as supported by Debian is added. It allows indicating
5331 which LUKS slot to use on disk, speeding up key loading.
5333 * The sd_journal_sendv() API call has been checked and
5334 officially declared to be async-signal-safe so that it may
5335 be invoked from signal handlers for logging purposes.
5337 * Boot-time status output is now enabled automatically after a
5338 short timeout if boot does not progress, in order to give
5339 the user an indication what she or he is waiting for.
5341 * The boot-time output has been improved to show how much time
5342 remains until jobs expire.
5344 * The KillMode= switch in service units gained a new possible
5345 value "mixed". If set, and the unit is shut down, then the
5346 initial SIGTERM signal is sent only to the main daemon
5347 process, while the following SIGKILL signal is sent to
5348 all remaining processes of the service.
5350 * When a scope unit is registered, a new property "Controller"
5351 may be set. If set to a valid bus name, systemd will send a
5352 RequestStop() signal to this name when it would like to shut
5353 down the scope. This may be used to hook manager logic into
5354 the shutdown logic of scope units. Also, scope units may now
5355 be put in a special "abandoned" state, in which case the
5356 manager process which created them takes no further
5357 responsibilities for it.
5359 * When reading unit files, systemd will now verify
5360 the access mode of these files, and warn about certain
5361 suspicious combinations. This has been added to make it
5362 easier to track down packaging bugs where unit files are
5363 marked executable or world-writable.
5365 * systemd-nspawn gained a new "--setenv=" switch to set
5366 container-wide environment variables. The similar option in
5367 systemd-activate was renamed from "--environment=" to
5368 "--setenv=" for consistency.
5370 * systemd-nspawn has been updated to create a new kdbus domain
5371 for each container that is invoked, thus allowing each
5372 container to have its own set of system and user buses,
5373 independent of the host.
5375 * systemd-nspawn gained a new --drop-capability= switch to run
5376 the container with less capabilities than the default. Both
5377 --drop-capability= and --capability= now take the special
5378 string "all" for dropping or keeping all capabilities.
5380 * systemd-nspawn gained new switches for executing containers
5381 with specific SELinux labels set.
5383 * systemd-nspawn gained a new --quiet switch to not generate
5384 any additional output but the container's own console
5387 * systemd-nspawn gained a new --share-system switch to run a
5388 container without PID namespacing enabled.
5390 * systemd-nspawn gained a new --register= switch to control
5391 whether the container is registered with systemd-machined or
5392 not. This is useful for containers that do not run full
5393 OS images, but only specific apps.
5395 * systemd-nspawn gained a new --keep-unit which may be used
5396 when invoked as the only program from a service unit, and
5397 results in registration of the unit service itself in
5398 systemd-machined, instead of a newly opened scope unit.
5400 * systemd-nspawn gained a new --network-interface= switch for
5401 moving arbitrary interfaces to the container. The new
5402 --network-veth switch creates a virtual Ethernet connection
5403 between host and container. The new --network-bridge=
5404 switch then allows assigning the host side of this virtual
5405 Ethernet connection to a bridge device.
5407 * systemd-nspawn gained a new --personality= switch for
5408 setting the kernel personality for the container. This is
5409 useful when running a 32-bit container on a 64-bit host. A
5410 similar option Personality= is now also available for service
5413 * logind will now also track a "Desktop" identifier for each
5414 session which encodes the desktop environment of it. This is
5415 useful for desktop environments that want to identify
5416 multiple running sessions of itself easily.
5418 * A new SELinuxContext= setting for service units has been
5419 added that allows setting a specific SELinux execution
5420 context for a service.
5422 * Most systemd client tools will now honour $SYSTEMD_LESS for
5423 settings of the "less" pager. By default, these tools will
5424 override $LESS to allow certain operations to work, such as
5425 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5426 influence this logic.
5428 * systemd's "seccomp" hook-up has been changed to make use of
5429 the libseccomp library instead of using its own
5430 implementation. This has benefits for portability among
5433 * For usage together with SystemCallFilter=, a new
5434 SystemCallErrorNumber= setting has been introduced that
5435 allows configuration of a system error number to be returned
5436 on filtered system calls, instead of immediately killing the
5437 process. Also, SystemCallArchitectures= has been added to
5438 limit access to system calls of a particular architecture
5439 (in order to turn off support for unused secondary
5440 architectures). There is also a global
5441 SystemCallArchitectures= setting in system.conf now to turn
5442 off support for non-native system calls system-wide.
5444 * systemd requires a kernel with a working name_to_handle_at(),
5445 please see the kernel config requirements in the README file.
5447 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5448 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5449 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5450 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5451 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5452 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5453 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5454 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5455 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5456 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5457 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5458 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5459 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5460 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5461 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5462 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5463 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5464 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5465 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5466 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5467 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5468 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5469 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5470 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5472 — Berlin, 2014-02-20
5476 * logind has gained support for facilitating privileged input
5477 and drm device access for unprivileged clients. This work is
5478 useful to allow Wayland display servers (and similar
5479 programs, such as kmscon) to run under the user's ID and
5480 access input and drm devices which are normally
5481 protected. When this is used (and the kernel is new enough)
5482 logind will "mute" IO on the file descriptors passed to
5483 Wayland as long as it is in the background and "unmute" it
5484 if it returns into the foreground. This allows secure
5485 session switching without allowing background sessions to
5486 eavesdrop on input and display data. This also introduces
5487 session switching support if VT support is turned off in the
5488 kernel, and on seats that are not seat0.
5490 * A new kernel command line option luks.options= is understood
5491 now which allows specifying LUKS options for usage for LUKS
5492 encrypted partitions specified with luks.uuid=.
5494 * tmpfiles.d(5) snippets may now use specifier expansion in
5495 path names. More specifically %m, %b, %H, %v, are now
5496 replaced by the local machine id, boot id, hostname, and
5497 kernel version number.
5499 * A new tmpfiles.d(5) command "m" has been introduced which
5500 may be used to change the owner/group/access mode of a file
5501 or directory if it exists, but do nothing if it does not.
5503 * This release removes high-level support for the
5504 MemorySoftLimit= cgroup setting. The underlying kernel
5505 cgroup attribute memory.soft_limit= is currently badly
5506 designed and likely to be removed from the kernel API in its
5507 current form, hence we should not expose it for now.
5509 * The memory.use_hierarchy cgroup attribute is now enabled for
5510 all cgroups systemd creates in the memory cgroup
5511 hierarchy. This option is likely to be come the built-in
5512 default in the kernel anyway, and the non-hierarchical mode
5513 never made much sense in the intrinsically hierarchical
5516 * A new field _SYSTEMD_SLICE= is logged along with all journal
5517 messages containing the slice a message was generated
5518 from. This is useful to allow easy per-customer filtering of
5519 logs among other things.
5521 * systemd-journald will no longer adjust the group of journal
5522 files it creates to the "systemd-journal" group. Instead we
5523 rely on the journal directory to be owned by the
5524 "systemd-journal" group, and its setgid bit set, so that the
5525 kernel file system layer will automatically enforce that
5526 journal files inherit this group assignment. The reason for
5527 this change is that we cannot allow NSS look-ups from
5528 journald which would be necessary to resolve
5529 "systemd-journal" to a numeric GID, because this might
5530 create deadlocks if NSS involves synchronous queries to
5531 other daemons (such as nscd, or sssd) which in turn are
5532 logging clients of journald and might block on it, which
5533 would then dead lock. A tmpfiles.d(5) snippet included in
5534 systemd will make sure the setgid bit and group are
5535 properly set on the journal directory if it exists on every
5536 boot. However, we recommend adjusting it manually after
5537 upgrades too (or from RPM scriptlets), so that the change is
5538 not delayed until next reboot.
5540 * Backlight and random seed files in /var/lib/ have moved into
5541 the /var/lib/systemd/ directory, in order to centralize all
5542 systemd generated files in one directory.
5544 * Boot time performance measurements (as displayed by
5545 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5546 performance information if that's available to determine how
5547 much time BIOS and boot loader initialization required. With
5548 a sufficiently new BIOS you hence no longer need to boot
5549 with Gummiboot to get access to such information.
5551 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5552 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5553 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5554 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5555 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5556 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5557 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5559 — Berlin, 2013-10-02
5563 * The Restart= option for services now understands a new
5564 on-watchdog setting, which will restart the service
5565 automatically if the service stops sending out watchdog keep
5566 alive messages (as configured with WatchdogSec=).
5568 * The getty generator (which is responsible for bringing up a
5569 getty on configured serial consoles) will no longer only
5570 start a getty on the primary kernel console but on all
5571 others, too. This makes the order in which console= is
5572 specified on the kernel command line less important.
5574 * libsystemd-logind gained a new sd_session_get_vt() call to
5575 retrieve the VT number of a session.
5577 * If the option "tries=0" is set for an entry of /etc/crypttab
5578 its passphrase is queried indefinitely instead of any
5579 maximum number of tries.
5581 * If a service with a configure PID file terminates its PID
5582 file will now be removed automatically if it still exists
5583 afterwards. This should put an end to stale PID files.
5585 * systemd-run will now also take relative binary path names
5586 for execution and no longer insists on absolute paths.
5588 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5589 paths that are optionally prefixed with "-" to indicate that
5590 it should not be considered a failure if they do not exist.
5592 * journalctl -o (and similar commands) now understands a new
5593 output mode "short-precise", it is similar to "short" but
5594 shows timestamps with usec accuracy.
5596 * The option "discard" (as known from Debian) is now
5597 synonymous to "allow-discards" in /etc/crypttab. In fact,
5598 "discard" is preferred now (since it is easier to remember
5601 * Some licensing clean-ups were made, so that more code is now
5602 LGPL-2.1 licensed than before.
5604 * A minimal tool to save/restore the display backlight
5605 brightness across reboots has been added. It will store the
5606 backlight setting as late as possible at shutdown, and
5607 restore it as early as possible during reboot.
5609 * A logic to automatically discover and enable home and swap
5610 partitions on GPT disks has been added. With this in place
5611 /etc/fstab becomes optional for many setups as systemd can
5612 discover certain partitions located on the root disk
5613 automatically. Home partitions are recognized under their
5614 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5615 partitions are recognized under their GPT type ID
5616 0657fd6da4ab43c484e50933c84b4f4f.
5618 * systemd will no longer pass any environment from the kernel
5619 or initrd to system services. If you want to set an
5620 environment for all services, do so via the kernel command
5621 line systemd.setenv= assignment.
5623 * The systemd-sysctl tool no longer natively reads the file
5624 /etc/sysctl.conf. If desired, the file should be symlinked
5625 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5626 legacy support by a symlink rather than built-in code, it
5627 also makes the otherwise hidden order of application of the
5628 different files visible. (Note that this partly reverts to a
5629 pre-198 application order of sysctl knobs!)
5631 * The "systemctl set-log-level" and "systemctl dump" commands
5632 have been moved to systemd-analyze.
5634 * systemd-run learned the new --remain-after-exit switch,
5635 which causes the scope unit not to be cleaned up
5636 automatically after the process terminated.
5638 * tmpfiles learned a new --exclude-prefix= switch to exclude
5639 certain paths from operation.
5641 * journald will now automatically flush all messages to disk
5642 as soon as a message at the log level CRIT, ALERT or EMERG
5645 Contributions from: Andrew Cook, Brandon Philips, Christian
5646 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5647 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5648 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5649 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5650 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5651 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5652 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5653 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5654 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5655 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5656 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5657 William Giokas, Zbigniew Jędrzejewski-Szmek
5659 — Berlin, 2013-09-13
5663 * The documentation has been updated to cover the various new
5664 concepts introduced with 205.
5666 * Unit files now understand the new %v specifier which
5667 resolves to the kernel version string as returned by "uname
5670 * systemctl now supports filtering the unit list output by
5671 load state, active state and sub state, using the new
5674 * "systemctl status" will now show the results of the
5675 condition checks (like ConditionPathExists= and similar) of
5676 the last start attempts of the unit. They are also logged to
5679 * "journalctl -b" may now be used to look for boot output of a
5680 specific boot. Try "journalctl -b -1" for the previous boot,
5681 but the syntax is substantially more powerful.
5683 * "journalctl --show-cursor" has been added which prints the
5684 cursor string the last shown log line. This may then be used
5685 with the new "journalctl --after-cursor=" switch to continue
5686 browsing logs from that point on.
5688 * "journalctl --force" may now be used to force regeneration
5691 * Creation of "dead" device nodes has been moved from udev
5692 into kmod and tmpfiles. Previously, udev would read the kmod
5693 databases to pre-generate dead device nodes based on meta
5694 information contained in kernel modules, so that these would
5695 be auto-loaded on access rather then at boot. As this
5696 does not really have much to do with the exposing actual
5697 kernel devices to userspace this has always been slightly
5698 alien in the udev codebase. Following the new scheme kmod
5699 will now generate a runtime snippet for tmpfiles from the
5700 module meta information and it now is tmpfiles' job to the
5701 create the nodes. This also allows overriding access and
5702 other parameters for the nodes using the usual tmpfiles
5703 facilities. As side effect this allows us to remove the
5704 CAP_SYS_MKNOD capability bit from udevd entirely.
5706 * logind's device ACLs may now be applied to these "dead"
5707 devices nodes too, thus finally allowing managed access to
5708 devices such as /dev/snd/sequencer without loading the
5709 backing module right-away.
5711 * A new RPM macro has been added that may be used to apply
5712 tmpfiles configuration during package installation.
5714 * systemd-detect-virt and ConditionVirtualization= now can
5715 detect User-Mode-Linux machines (UML).
5717 * journald will now implicitly log the effective capabilities
5718 set of processes in the message metadata.
5720 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5722 * The initrd interface has been simplified (more specifically,
5723 support for passing performance data via environment
5724 variables and fsck results via files in /run has been
5725 removed). These features were non-essential, and are
5726 nowadays available in a much nicer way by having systemd in
5727 the initrd serialize its state and have the hosts systemd
5728 deserialize it again.
5730 * The udev "keymap" data files and tools to apply keyboard
5731 specific mappings of scan to key codes, and force-release
5732 scan code lists have been entirely replaced by a udev
5733 "keyboard" builtin and a hwdb data file.
5735 * systemd will now honour the kernel's "quiet" command line
5736 argument also during late shutdown, resulting in a
5737 completely silent shutdown when used.
5739 * There's now an option to control the SO_REUSEPORT socket
5740 option in .socket units.
5742 * Instance units will now automatically get a per-template
5743 subslice of system.slice unless something else is explicitly
5744 configured. For example, instances of sshd@.service will now
5745 implicitly be placed in system-sshd.slice rather than
5746 system.slice as before.
5748 * Test coverage support may now be enabled at build time.
5750 Contributions from: Dave Reisner, Frederic Crozat, Harald
5751 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5752 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5753 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5754 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5755 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5756 Giokas, Zbigniew Jędrzejewski-Szmek
5758 — Berlin, 2013-07-23
5762 * Two new unit types have been introduced:
5764 Scope units are very similar to service units, however, are
5765 created out of pre-existing processes — instead of PID 1
5766 forking off the processes. By using scope units it is
5767 possible for system services and applications to group their
5768 own child processes (worker processes) in a powerful way
5769 which then maybe used to organize them, or kill them
5770 together, or apply resource limits on them.
5772 Slice units may be used to partition system resources in an
5773 hierarchical fashion and then assign other units to them. By
5774 default there are now three slices: system.slice (for all
5775 system services), user.slice (for all user sessions),
5776 machine.slice (for VMs and containers).
5778 Slices and scopes have been introduced primarily in
5779 context of the work to move cgroup handling to a
5780 single-writer scheme, where only PID 1
5781 creates/removes/manages cgroups.
5783 * There's a new concept of "transient" units. In contrast to
5784 normal units these units are created via an API at runtime,
5785 not from configuration from disk. More specifically this
5786 means it is now possible to run arbitrary programs as
5787 independent services, with all execution parameters passed
5788 in via bus APIs rather than read from disk. Transient units
5789 make systemd substantially more dynamic then it ever was,
5790 and useful as a general batch manager.
5792 * logind has been updated to make use of scope and slice units
5793 for managing user sessions. As a user logs in he will get
5794 his own private slice unit, to which all sessions are added
5795 as scope units. We also added support for automatically
5796 adding an instance of user@.service for the user into the
5797 slice. Effectively logind will no longer create cgroup
5798 hierarchies on its own now, it will defer entirely to PID 1
5799 for this by means of scope, service and slice units. Since
5800 user sessions this way become entities managed by PID 1
5801 the output of "systemctl" is now a lot more comprehensive.
5803 * A new mini-daemon "systemd-machined" has been added which
5804 may be used by virtualization managers to register local
5805 VMs/containers. nspawn has been updated accordingly, and
5806 libvirt will be updated shortly. machined will collect a bit
5807 of meta information about the VMs/containers, and assign
5808 them their own scope unit (see above). The collected
5809 meta-data is then made available via the "machinectl" tool,
5810 and exposed in "ps" and similar tools. machined/machinectl
5811 is compile-time optional.
5813 * As discussed earlier, the low-level cgroup configuration
5814 options ControlGroup=, ControlGroupModify=,
5815 ControlGroupPersistent=, ControlGroupAttribute= have been
5816 removed. Please use high-level attribute settings instead as
5817 well as slice units.
5819 * A new bus call SetUnitProperties() has been added to alter
5820 various runtime parameters of a unit. This is primarily
5821 useful to alter cgroup parameters dynamically in a nice way,
5822 but will be extended later on to make more properties
5823 modifiable at runtime. systemctl gained a new set-properties
5824 command that wraps this call.
5826 * A new tool "systemd-run" has been added which can be used to
5827 run arbitrary command lines as transient services or scopes,
5828 while configuring a number of settings via the command
5829 line. This tool is currently very basic, however already
5830 very useful. We plan to extend this tool to even allow
5831 queuing of execution jobs with time triggers from the
5832 command line, similar in fashion to "at".
5834 * nspawn will now inform the user explicitly that kernels with
5835 audit enabled break containers, and suggest the user to turn
5838 * Support for detecting the IMA and AppArmor security
5839 frameworks with ConditionSecurity= has been added.
5841 * journalctl gained a new "-k" switch for showing only kernel
5842 messages, mimicking dmesg output; in addition to "--user"
5843 and "--system" switches for showing only user's own logs
5846 * systemd-delta can now show information about drop-in
5847 snippets extending unit files.
5849 * libsystemd-bus has been substantially updated but is still
5850 not available as public API.
5852 * systemd will now look for the "debug" argument on the kernel
5853 command line and enable debug logging, similar to what
5854 "systemd.log_level=debug" already did before.
5856 * "systemctl set-default", "systemctl get-default" has been
5857 added to configure the default.target symlink, which
5858 controls what to boot into by default.
5860 * "systemctl set-log-level" has been added as a convenient
5861 way to raise and lower systemd logging threshold.
5863 * "systemd-analyze plot" will now show the time the various
5864 generators needed for execution, as well as information
5865 about the unit file loading.
5867 * libsystemd-journal gained a new sd_journal_open_files() call
5868 for opening specific journal files. journactl also gained a
5869 new switch to expose this new functionality. Previously we
5870 only supported opening all files from a directory, or all
5871 files from the system, as opening individual files only is
5872 racy due to journal file rotation.
5874 * systemd gained the new DefaultEnvironment= setting in
5875 /etc/systemd/system.conf to set environment variables for
5878 * If a privileged process logs a journal message with the
5879 OBJECT_PID= field set, then journald will automatically
5880 augment this with additional OBJECT_UID=, OBJECT_GID=,
5881 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5882 system services want to log events about specific client
5883 processes. journactl/systemctl has been updated to make use
5884 of this information if all log messages regarding a specific
5887 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5888 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5889 Reisner, David Coppa, David King, David Strauss, Eelco
5890 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5891 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5892 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5893 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5894 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5895 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5896 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5897 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5898 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5899 Łukasz Stelmach, 장동준
5903 * The Python bindings gained some minimal support for the APIs
5904 exposed by libsystemd-logind.
5906 * ConditionSecurity= gained support for detecting SMACK. Since
5907 this condition already supports SELinux and AppArmor we only
5908 miss IMA for this. Patches welcome!
5910 Contributions from: Karol Lewandowski, Lennart Poettering,
5911 Zbigniew Jędrzejewski-Szmek
5915 * systemd-nspawn will now create /etc/resolv.conf if
5916 necessary, before bind-mounting the host's file onto it.
5918 * systemd-nspawn will now store meta information about a
5919 container on the container's cgroup as extended attribute
5920 fields, including the root directory.
5922 * The cgroup hierarchy has been reworked in many ways. All
5923 objects any of the components systemd creates in the cgroup
5924 tree are now suffixed. More specifically, user sessions are
5925 now placed in cgroups suffixed with ".session", users in
5926 cgroups suffixed with ".user", and nspawn containers in
5927 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5928 names are now escaped in a simple scheme to avoid collision
5929 of userspace object names with kernel filenames. This work
5930 is preparation for making these objects relocatable in the
5931 cgroup tree, in order to allow easy resource partitioning of
5932 these objects without causing naming conflicts.
5934 * systemctl list-dependencies gained the new switches
5935 --plain, --reverse, --after and --before.
5937 * systemd-inhibit now shows the process name of processes that
5938 have taken an inhibitor lock.
5940 * nss-myhostname will now also resolve "localhost"
5941 implicitly. This makes /etc/hosts an optional file and
5942 nicely handles that on IPv6 ::1 maps to both "localhost" and
5945 * libsystemd-logind.so gained a new call
5946 sd_get_machine_names() to enumerate running containers and
5947 VMs (currently only supported by very new libvirt and
5948 nspawn). sd_login_monitor can now be used to watch
5949 VMs/containers coming and going.
5951 * .include is not allowed recursively anymore, and only in
5952 unit files. Usually it is better to use drop-in snippets in
5953 .d/*.conf anyway, as introduced with systemd 198.
5955 * systemd-analyze gained a new "critical-chain" command that
5956 determines the slowest chain of units run during system
5957 boot-up. It is very useful for tracking down where
5958 optimizing boot time is the most beneficial.
5960 * systemd will no longer allow manipulating service paths in
5961 the name=systemd:/system cgroup tree using ControlGroup= in
5962 units. (But is still fine with it in all other dirs.)
5964 * There's a new systemd-nspawn@.service service file that may
5965 be used to easily run nspawn containers as system
5966 services. With the container's root directory in
5967 /var/lib/container/foobar it is now sufficient to run
5968 "systemctl start systemd-nspawn@foobar.service" to boot it.
5970 * systemd-cgls gained a new parameter "--machine" to list only
5971 the processes within a certain container.
5973 * ConditionSecurity= now can check for "apparmor". We still
5974 are lacking checks for SMACK and IMA for this condition
5975 check though. Patches welcome!
5977 * A new configuration file /etc/systemd/sleep.conf has been
5978 added that may be used to configure which kernel operation
5979 systemd is supposed to execute when "suspend", "hibernate"
5980 or "hybrid-sleep" is requested. This makes the new kernel
5981 "freeze" state accessible to the user.
5983 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5984 the passed argument if applicable.
5986 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5987 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5988 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5989 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5990 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5991 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5992 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5997 * The output of 'systemctl list-jobs' got some polishing. The
5998 '--type=' argument may now be passed more than once. A new
5999 command 'systemctl list-sockets' has been added which shows
6000 a list of kernel sockets systemd is listening on with the
6001 socket units they belong to, plus the units these socket
6004 * The experimental libsystemd-bus library got substantial
6005 updates to work in conjunction with the (also experimental)
6006 kdbus kernel project. It works well enough to exchange
6007 messages with some sophistication. Note that kdbus is not
6008 ready yet, and the library is mostly an elaborate test case
6009 for now, and not installable.
6011 * systemd gained a new unit 'systemd-static-nodes.service'
6012 that generates static device nodes earlier during boot, and
6013 can run in conjunction with udev.
6015 * libsystemd-login gained a new call sd_pid_get_user_unit()
6016 to retrieve the user systemd unit a process is running
6017 in. This is useful for systems where systemd is used as
6020 * systemd-nspawn now places all containers in the new /machine
6021 top-level cgroup directory in the name=systemd
6022 hierarchy. libvirt will soon do the same, so that we get a
6023 uniform separation of /system, /user and /machine for system
6024 services, user processes and containers/virtual
6025 machines. This new cgroup hierarchy is also useful to stick
6026 stable names to specific container instances, which can be
6027 recognized later this way (this name may be controlled
6028 via systemd-nspawn's new -M switch). libsystemd-login also
6029 gained a new call sd_pid_get_machine_name() to retrieve the
6030 name of the container/VM a specific process belongs to.
6032 * bootchart can now store its data in the journal.
6034 * libsystemd-journal gained a new call
6035 sd_journal_add_conjunction() for AND expressions to the
6036 matching logic. This can be used to express more complex
6037 logical expressions.
6039 * journactl can now take multiple --unit= and --user-unit=
6042 * The cryptsetup logic now understands the "luks.key=" kernel
6043 command line switch for specifying a file to read the
6044 decryption key from. Also, if a configured key file is not
6045 found the tool will now automatically fall back to prompting
6048 * Python systemd.journal module was updated to wrap recently
6049 added functions from libsystemd-journal. The interface was
6050 changed to bring the low level interface in s.j._Reader
6051 closer to the C API, and the high level interface in
6052 s.j.Reader was updated to wrap and convert all data about
6055 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6056 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6057 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6058 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6059 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6060 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6064 * journalctl --update-catalog now understands a new --root=
6065 option to operate on catalogs found in a different root
6068 * During shutdown after systemd has terminated all running
6069 services a final killing loop kills all remaining left-over
6070 processes. We will now print the name of these processes
6071 when we send SIGKILL to them, since this usually indicates a
6074 * If /etc/crypttab refers to password files stored on
6075 configured mount points automatic dependencies will now be
6076 generated to ensure the specific mount is established first
6077 before the key file is attempted to be read.
6079 * 'systemctl status' will now show information about the
6080 network sockets a socket unit is listening on.
6082 * 'systemctl status' will also shown information about any
6083 drop-in configuration file for units. (Drop-In configuration
6084 files in this context are files such as
6085 /etc/systemd/systemd/foobar.service.d/*.conf)
6087 * systemd-cgtop now optionally shows summed up CPU times of
6088 cgroups. Press '%' while running cgtop to switch between
6089 percentage and absolute mode. This is useful to determine
6090 which cgroups use up the most CPU time over the entire
6091 runtime of the system. systemd-cgtop has also been updated
6092 to be 'pipeable' for processing with further shell tools.
6094 * 'hostnamectl set-hostname' will now allow setting of FQDN
6097 * The formatting and parsing of time span values has been
6098 changed. The parser now understands fractional expressions
6099 such as "5.5h". The formatter will now output fractional
6100 expressions for all time spans under 1min, i.e. "5.123456s"
6101 rather than "5s 123ms 456us". For time spans under 1s
6102 millisecond values are shown, for those under 1ms
6103 microsecond values are shown. This should greatly improve
6104 all time-related output of systemd.
6106 * libsystemd-login and libsystemd-journal gained new
6107 functions for querying the poll() events mask and poll()
6108 timeout value for integration into arbitrary event
6111 * localectl gained the ability to list available X11 keymaps
6112 (models, layouts, variants, options).
6114 * 'systemd-analyze dot' gained the ability to filter for
6115 specific units via shell-style globs, to create smaller,
6116 more useful graphs. I.e. it is now possible to create simple
6117 graphs of all the dependencies between only target units, or
6118 of all units that Avahi has dependencies with.
6120 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6121 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6122 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6123 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6124 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6125 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6126 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6130 * The boot-time readahead implementation for rotating media
6131 will now read the read-ahead data in multiple passes which
6132 consist of all read requests made in equidistant time
6133 intervals. This means instead of strictly reading read-ahead
6134 data in its physical order on disk we now try to find a
6135 middle ground between physical and access time order.
6137 * /etc/os-release files gained a new BUILD_ID= field for usage
6138 on operating systems that provide continuous builds of OS
6141 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6142 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6143 William Douglas, Zbigniew Jędrzejewski-Szmek
6147 * systemd-python gained an API exposing libsystemd-daemon.
6149 * The SMACK setup logic gained support for uploading CIPSO
6152 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6153 ReadOnlyDirectories= and InaccessibleDirectories= has
6154 changed. The private /tmp and /var/tmp directories are now
6155 shared by all processes of a service (which means
6156 ExecStartPre= may now leave data in /tmp that ExecStart= of
6157 the same service can still access). When a service is
6158 stopped its temporary directories are immediately deleted
6159 (normal clean-up with tmpfiles is still done in addition to
6162 * By default, systemd will now set a couple of sysctl
6163 variables in the kernel: the safe sysrq options are turned
6164 on, IP route verification is turned on, and source routing
6165 disabled. The recently added hardlink and softlink
6166 protection of the kernel is turned on. These settings should
6167 be reasonably safe, and good defaults for all new systems.
6169 * The predictable network naming logic may now be turned off
6170 with a new kernel command line switch: net.ifnames=0.
6172 * A new libsystemd-bus module has been added that implements a
6173 pretty complete D-Bus client library. For details see:
6175 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6177 * journald will now explicitly flush the journal files to disk
6178 at the latest 5min after each write. The file will then also
6179 be marked offline until the next write. This should increase
6180 reliability in case of a crash. The synchronization delay
6181 can be configured via SyncIntervalSec= in journald.conf.
6183 * There's a new remote-fs-setup.target unit that can be used
6184 to pull in specific services when at least one remote file
6185 system is to be mounted.
6187 * There are new targets timers.target and paths.target as
6188 canonical targets to pull user timer and path units in
6189 from. This complements sockets.target with a similar
6190 purpose for socket units.
6192 * libudev gained a new call udev_device_set_attribute_value()
6193 to set sysfs attributes of a device.
6195 * The udev daemon now sets the default number of worker
6196 processes executed in parallel based on the number of available
6197 CPUs instead of the amount of available RAM. This is supposed
6198 to provide a more reliable default and limit a too aggressive
6199 parallelism for setups with 1000s of devices connected.
6201 Contributions from: Auke Kok, Colin Walters, Cristian
6202 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6203 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6204 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6205 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6206 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6207 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6208 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6209 Zbigniew Jędrzejewski-Szmek
6213 * Configuration of unit files may now be extended via drop-in
6214 files without having to edit/override the unit files
6215 themselves. More specifically, if the administrator wants to
6216 change one value for a service file foobar.service he can
6217 now do so by dropping in a configuration snippet into
6218 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6219 will load all these snippets and apply them on top of the
6220 main unit configuration file, possibly extending or
6221 overriding its settings. Using these drop-in snippets is
6222 generally nicer than the two earlier options for changing
6223 unit files locally: copying the files from
6224 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6225 them there; or creating a new file in /etc/systemd/system/
6226 that incorporates the original one via ".include". Drop-in
6227 snippets into these .d/ directories can be placed in any
6228 directory systemd looks for units in, and the usual
6229 overriding semantics between /usr/lib, /etc and /run apply
6232 * Most unit file settings which take lists of items can now be
6233 reset by assigning the empty string to them. For example,
6234 normally, settings such as Environment=FOO=BAR append a new
6235 environment variable assignment to the environment block,
6236 each time they are used. By assigning Environment= the empty
6237 string the environment block can be reset to empty. This is
6238 particularly useful with the .d/*.conf drop-in snippets
6239 mentioned above, since this adds the ability to reset list
6240 settings from vendor unit files via these drop-ins.
6242 * systemctl gained a new "list-dependencies" command for
6243 listing the dependencies of a unit recursively.
6245 * Inhibitors are now honored and listed by "systemctl
6246 suspend", "systemctl poweroff" (and similar) too, not only
6247 GNOME. These commands will also list active sessions by
6250 * Resource limits (as exposed by the various control group
6251 controllers) can now be controlled dynamically at runtime
6252 for all units. More specifically, you can now use a command
6253 like "systemctl set-cgroup-attr foobar.service cpu.shares
6254 2000" to alter the CPU shares a specific service gets. These
6255 settings are stored persistently on disk, and thus allow the
6256 administrator to easily adjust the resource usage of
6257 services with a few simple commands. This dynamic resource
6258 management logic is also available to other programs via the
6259 bus. Almost any kernel cgroup attribute and controller is
6262 * systemd-vconsole-setup will now copy all font settings to
6263 all allocated VTs, where it previously applied them only to
6266 * libsystemd-login gained the new sd_session_get_tty() API
6269 * This release drops support for a few legacy or
6270 distribution-specific LSB facility names when parsing init
6271 scripts: $x-display-manager, $mail-transfer-agent,
6272 $mail-transport-agent, $mail-transfer-agent, $smtp,
6273 $null. Also, the mail-transfer-agent.target unit backing
6274 this has been removed. Distributions which want to retain
6275 compatibility with this should carry the burden for
6276 supporting this themselves and patch support for these back
6277 in, if they really need to. Also, the facilities $syslog and
6278 $local_fs are now ignored, since systemd does not support
6279 early-boot LSB init scripts anymore, and these facilities
6280 are implied anyway for normal services. syslog.target has
6283 * There are new bus calls on PID1's Manager object for
6284 cancelling jobs, and removing snapshot units. Previously,
6285 both calls were only available on the Job and Snapshot
6288 * systemd-journal-gatewayd gained SSL support.
6290 * The various "environment" files, such as /etc/locale.conf
6291 now support continuation lines with a backslash ("\") as
6292 last character in the line, similarly in style (but different)
6293 to how this is supported in shells.
6295 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6296 now implicitly appended to every log entry logged. systemctl
6297 has been updated to filter by this field when operating on a
6298 user systemd instance.
6300 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6301 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6302 the container. This makes it easier to boot unmodified
6303 Fedora systems in a container, which however still requires
6304 audit=0 to be passed on the kernel command line. Auditing in
6305 kernel and userspace is unfortunately still too broken in
6306 context of containers, hence we recommend compiling it out
6307 of the kernel or using audit=0. Hopefully this will be fixed
6308 one day for good in the kernel.
6310 * nspawn gained the new --bind= and --bind-ro= parameters to
6311 bind mount specific directories from the host into the
6314 * nspawn will now mount its own devpts file system instance
6315 into the container, in order not to leak pty devices from
6316 the host into the container.
6318 * systemd will now read the firmware boot time performance
6319 information from the EFI variables, if the used boot loader
6320 supports this, and takes it into account for boot performance
6321 analysis via "systemd-analyze". This is currently supported
6322 only in conjunction with Gummiboot, but could be supported
6323 by other boot loaders too. For details see:
6325 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6327 * A new generator has been added that automatically mounts the
6328 EFI System Partition (ESP) to /boot, if that directory
6329 exists, is empty, and no other file system has been
6330 configured to be mounted there.
6332 * logind will now send out PrepareForSleep(false) out
6333 unconditionally, after coming back from suspend. This may be
6334 used by applications as asynchronous notification for
6335 system resume events.
6337 * "systemctl unlock-sessions" has been added, that allows
6338 unlocking the screens of all user sessions at once, similar
6339 to how "systemctl lock-sessions" already locked all users
6340 sessions. This is backed by a new D-Bus call UnlockSessions().
6342 * "loginctl seat-status" will now show the master device of a
6343 seat. (i.e. the device of a seat that needs to be around for
6344 the seat to be considered available, usually the graphics
6347 * tmpfiles gained a new "X" line type, that allows
6348 configuration of files and directories (with wildcards) that
6349 shall be excluded from automatic cleanup ("aging").
6351 * udev default rules set the device node permissions now only
6352 at "add" events, and do not change them any longer with a
6353 later "change" event.
6355 * The log messages for lid events and power/sleep keypresses
6356 now carry a message ID.
6358 * We now have a substantially larger unit test suite, but this
6359 continues to be work in progress.
6361 * udevadm hwdb gained a new --root= parameter to change the
6362 root directory to operate relative to.
6364 * logind will now issue a background sync() request to the kernel
6365 early at shutdown, so that dirty buffers are flushed to disk early
6366 instead of at the last moment, in order to optimize shutdown
6369 * A new bootctl tool has been added that is an interface for
6370 certain boot loader operations. This is currently a preview
6371 and is likely to be extended into a small mechanism daemon
6372 like timedated, localed, hostnamed, and can be used by
6373 graphical UIs to enumerate available boot options, and
6374 request boot into firmware operations.
6376 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6377 the rest of the package. It also has been updated to work
6378 correctly in initrds.
6380 * Policykit previously has been runtime optional, and is now
6381 also compile time optional via a configure switch.
6383 * systemd-analyze has been reimplemented in C. Also "systemctl
6384 dot" has moved into systemd-analyze.
6386 * "systemctl status" with no further parameters will now print
6387 the status of all active or failed units.
6389 * Operations such as "systemctl start" can now be executed
6390 with a new mode "--irreversible" which may be used to queue
6391 operations that cannot accidentally be reversed by a later
6392 job queuing. This is by default used to make shutdown
6393 requests more robust.
6395 * The Python API of systemd now gained a new module for
6396 reading journal files.
6398 * A new tool kernel-install has been added that can install
6399 kernel images according to the Boot Loader Specification:
6401 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6403 * Boot time console output has been improved to provide
6404 animated boot time output for hanging jobs.
6406 * A new tool systemd-activate has been added which can be used
6407 to test socket activation with, directly from the command
6408 line. This should make it much easier to test and debug
6409 socket activation in daemons.
6411 * journalctl gained a new "--reverse" (or -r) option to show
6412 journal output in reverse order (i.e. newest line first).
6414 * journalctl gained a new "--pager-end" (or -e) option to jump
6415 to immediately jump to the end of the journal in the
6416 pager. This is only supported in conjunction with "less".
6418 * journalctl gained a new "--user-unit=" option, that works
6419 similarly to "--unit=" but filters for user units rather than
6422 * A number of unit files to ease adoption of systemd in
6423 initrds has been added. This moves some minimal logic from
6424 the various initrd implementations into systemd proper.
6426 * The journal files are now owned by a new group
6427 "systemd-journal", which exists specifically to allow access
6428 to the journal, and nothing else. Previously, we used the
6429 "adm" group for that, which however possibly covers more
6430 than just journal/log file access. This new group is now
6431 already used by systemd-journal-gatewayd to ensure this
6432 daemon gets access to the journal files and as little else
6433 as possible. Note that "make install" will also set FS ACLs
6434 up for /var/log/journal to give "adm" and "wheel" read
6435 access to it, in addition to "systemd-journal" which owns
6436 the journal files. We recommend that packaging scripts also
6437 add read access to "adm" + "wheel" to /var/log/journal, and
6438 all existing/future journal files. To normal users and
6439 administrators little changes, however packagers need to
6440 ensure to create the "systemd-journal" system group at
6441 package installation time.
6443 * The systemd-journal-gatewayd now runs as unprivileged user
6444 systemd-journal-gateway:systemd-journal-gateway. Packaging
6445 scripts need to create these system user/group at
6448 * timedated now exposes a new boolean property CanNTP that
6449 indicates whether a local NTP service is available or not.
6451 * systemd-detect-virt will now also detect xen PVs
6453 * The pstore file system is now mounted by default, if it is
6456 * In addition to the SELinux and IMA policies we will now also
6457 load SMACK policies at early boot.
6459 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6460 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6461 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6462 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6463 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6464 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6465 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6466 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6467 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6468 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6469 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6470 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6471 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6472 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6476 * Timer units now support calendar time events in addition to
6477 monotonic time events. That means you can now trigger a unit
6478 based on a calendar time specification such as "Thu,Fri
6479 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6480 or fifth day of any month of the year 2013, given that it is
6481 a thursday or friday. This brings timer event support
6482 considerably closer to cron's capabilities. For details on
6483 the supported calendar time specification language see
6486 * udev now supports a number of different naming policies for
6487 network interfaces for predictable names, and a combination
6488 of these policies is now the default. Please see this wiki
6489 document for details:
6491 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6493 * Auke Kok's bootchart implementation has been added to the
6494 systemd tree. It is an optional component that can graph the
6495 boot in quite some detail. It is one of the best bootchart
6496 implementations around and minimal in its code and
6499 * nss-myhostname has been integrated into the systemd source
6500 tree. nss-myhostname guarantees that the local hostname
6501 always stays resolvable via NSS. It has been a weak
6502 requirement of systemd-hostnamed since a long time, and
6503 since its code is actually trivial we decided to just
6504 include it in systemd's source tree. It can be turned off
6505 with a configure switch.
6507 * The read-ahead logic is now capable of properly detecting
6508 whether a btrfs file system is on SSD or rotating media, in
6509 order to optimize the read-ahead scheme. Previously, it was
6510 only capable of detecting this on traditional file systems
6513 * In udev, additional device properties are now read from the
6514 IAB in addition to the OUI database. Also, Bluetooth company
6515 identities are attached to the devices as well.
6517 * In service files %U may be used as specifier that is
6518 replaced by the configured user name of the service.
6520 * nspawn may now be invoked without a controlling TTY. This
6521 makes it suitable for invocation as its own service. This
6522 may be used to set up a simple containerized server system
6523 using only core OS tools.
6525 * systemd and nspawn can now accept socket file descriptors
6526 when they are started for socket activation. This enables
6527 implementation of socket activated nspawn
6528 containers. i.e. think about autospawning an entire OS image
6529 when the first SSH or HTTP connection is received. We expect
6530 that similar functionality will also be added to libvirt-lxc
6533 * journalctl will now suppress ANSI color codes when
6534 presenting log data.
6536 * systemctl will no longer show control group information for
6537 a unit if the control group is empty anyway.
6539 * logind can now automatically suspend/hibernate/shutdown the
6542 * /etc/machine-info and hostnamed now also expose the chassis
6543 type of the system. This can be used to determine whether
6544 the local system is a laptop, desktop, handset or
6545 tablet. This information may either be configured by the
6546 user/vendor or is automatically determined from ACPI and DMI
6547 information if possible.
6549 * A number of PolicyKit actions are now bound together with
6550 "imply" rules. This should simplify creating UIs because
6551 many actions will now authenticate similar ones as well.
6553 * Unit files learnt a new condition ConditionACPower= which
6554 may be used to conditionalize a unit depending on whether an
6555 AC power source is connected or not, of whether the system
6556 is running on battery power.
6558 * systemctl gained a new "is-failed" verb that may be used in
6559 shell scripts and suchlike to check whether a specific unit
6560 is in the "failed" state.
6562 * The EnvironmentFile= setting in unit files now supports file
6563 globbing, and can hence be used to easily read a number of
6564 environment files at once.
6566 * systemd will no longer detect and recognize specific
6567 distributions. All distribution-specific #ifdeffery has been
6568 removed, systemd is now fully generic and
6569 distribution-agnostic. Effectively, not too much is lost as
6570 a lot of the code is still accessible via explicit configure
6571 switches. However, support for some distribution specific
6572 legacy configuration file formats has been dropped. We
6573 recommend distributions to simply adopt the configuration
6574 files everybody else uses now and convert the old
6575 configuration from packaging scripts. Most distributions
6576 already did that. If that's not possible or desirable,
6577 distributions are welcome to forward port the specific
6578 pieces of code locally from the git history.
6580 * When logging a message about a unit systemd will now always
6581 log the unit name in the message meta data.
6583 * localectl will now also discover system locale data that is
6584 not stored in locale archives, but directly unpacked.
6586 * logind will no longer unconditionally use framebuffer
6587 devices as seat masters, i.e. as devices that are required
6588 to be existing before a seat is considered preset. Instead,
6589 it will now look for all devices that are tagged as
6590 "seat-master" in udev. By default, framebuffer devices will
6591 be marked as such, but depending on local systems, other
6592 devices might be marked as well. This may be used to
6593 integrate graphics cards using closed source drivers (such
6594 as NVidia ones) more nicely into logind. Note however, that
6595 we recommend using the open source NVidia drivers instead,
6596 and no udev rules for the closed-source drivers will be
6597 shipped from us upstream.
6599 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6600 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6601 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6602 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6603 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6604 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6605 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6606 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6607 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6608 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6609 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6610 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6615 * udev gained support for loading additional device properties
6616 from an indexed database that is keyed by vendor/product IDs
6617 and similar device identifiers. For the beginning this
6618 "hwdb" is populated with data from the well-known PCI and
6619 USB database, but also includes PNP, ACPI and OID data. In
6620 the longer run this indexed database shall grow into
6621 becoming the one central database for non-essential
6622 userspace device metadata. Previously, data from the PCI/USB
6623 database was only attached to select devices, since the
6624 lookup was a relatively expensive operation due to O(n) time
6625 complexity (with n being the number of entries in the
6626 database). Since this is now O(1), we decided to add in this
6627 data for all devices where this is available, by
6628 default. Note that the indexed database needs to be rebuilt
6629 when new data files are installed. To achieve this you need
6630 to update your packaging scripts to invoke "udevadm hwdb
6631 --update" after installation of hwdb data files. For
6632 RPM-based distributions we introduced the new
6633 %udev_hwdb_update macro for this purpose.
6635 * The Journal gained support for the "Message Catalog", an
6636 indexed database to link up additional information with
6637 journal entries. For further details please check:
6639 https://www.freedesktop.org/wiki/Software/systemd/catalog
6641 The indexed message catalog database also needs to be
6642 rebuilt after installation of message catalog files. Use
6643 "journalctl --update-catalog" for this. For RPM-based
6644 distributions we introduced the %journal_catalog_update
6645 macro for this purpose.
6647 * The Python Journal bindings gained support for the standard
6648 Python logging framework.
6650 * The Journal API gained new functions for checking whether
6651 the underlying file system of a journal file is capable of
6652 properly reporting file change notifications, or whether
6653 applications that want to reflect journal changes "live"
6654 need to recheck journal files continuously in appropriate
6657 * It is now possible to set the "age" field for tmpfiles
6658 entries to 0, indicating that files matching this entry
6659 shall always be removed when the directories are cleaned up.
6661 * coredumpctl gained a new "gdb" verb which invokes gdb
6662 right-away on the selected coredump.
6664 * There's now support for "hybrid sleep" on kernels that
6665 support this, in addition to "suspend" and "hibernate". Use
6666 "systemctl hybrid-sleep" to make use of this.
6668 * logind's HandleSuspendKey= setting (and related settings)
6669 now gained support for a new "lock" setting to simply
6670 request the screen lock on all local sessions, instead of
6671 actually executing a suspend or hibernation.
6673 * systemd will now mount the EFI variables file system by
6676 * Socket units now gained support for configuration of the
6677 SMACK security label.
6679 * timedatectl will now output the time of the last and next
6680 daylight saving change.
6682 * We dropped support for various legacy and distro-specific
6683 concepts, such as insserv, early-boot SysV services
6684 (i.e. those for non-standard runlevels such as 'b' or 'S')
6685 or ArchLinux /etc/rc.conf support. We recommend the
6686 distributions who still need support this to either continue
6687 to maintain the necessary patches downstream, or find a
6688 different solution. (Talk to us if you have questions!)
6690 * Various systemd components will now bypass PolicyKit checks
6691 for root and otherwise handle properly if PolicyKit is not
6692 found to be around. This should fix most issues for
6693 PolicyKit-less systems. Quite frankly this should have been
6694 this way since day one. It is absolutely our intention to
6695 make systemd work fine on PolicyKit-less systems, and we
6696 consider it a bug if something does not work as it should if
6697 PolicyKit is not around.
6699 * For embedded systems it is now possible to build udev and
6700 systemd without blkid and/or kmod support.
6702 * "systemctl switch-root" is now capable of switching root
6703 more than once. I.e. in addition to transitions from the
6704 initrd to the host OS it is now possible to transition to
6705 further OS images from the host. This is useful to implement
6706 offline updating tools.
6708 * Various other additions have been made to the RPM macros
6709 shipped with systemd. Use %udev_rules_update() after
6710 installing new udev rules files. %_udevhwdbdir,
6711 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6712 %_sysctldir are now available which resolve to the right
6713 directories for packages to place various data files in.
6715 * journalctl gained the new --full switch (in addition to
6716 --all, to disable ellipsation for long messages.
6718 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6719 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6720 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6721 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6722 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6723 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6724 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6725 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6726 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6730 * journalctl gained new --since= and --until= switches to
6731 filter by time. It also now supports nice filtering for
6732 units via --unit=/-u.
6734 * Type=oneshot services may use ExecReload= and do the
6737 * The journal daemon now supports time-based rotation and
6738 vacuuming, in addition to the usual disk-space based
6741 * The journal will now index the available field values for
6742 each field name. This enables clients to show pretty drop
6743 downs of available match values when filtering. The bash
6744 completion of journalctl has been updated
6745 accordingly. journalctl gained a new switch -F to list all
6746 values a certain field takes in the journal database.
6748 * More service events are now written as structured messages
6749 to the journal, and made recognizable via message IDs.
6751 * The timedated, localed and hostnamed mini-services which
6752 previously only provided support for changing time, locale
6753 and hostname settings from graphical DEs such as GNOME now
6754 also have a minimal (but very useful) text-based client
6755 utility each. This is probably the nicest way to changing
6756 these settings from the command line now, especially since
6757 it lists available options and is fully integrated with bash
6760 * There's now a new tool "systemd-coredumpctl" to list and
6761 extract coredumps from the journal.
6763 * We now install a README each in /var/log/ and
6764 /etc/rc.d/init.d explaining where the system logs and init
6765 scripts went. This hopefully should help folks who go to
6766 that dirs and look into the otherwise now empty void and
6767 scratch their heads.
6769 * When user-services are invoked (by systemd --user) the
6770 $MANAGERPID env var is set to the PID of systemd.
6772 * SIGRTMIN+24 when sent to a --user instance will now result
6773 in immediate termination of systemd.
6775 * gatewayd received numerous feature additions such as a
6776 "follow" mode, for live syncing and filtering.
6778 * browse.html now allows filtering and showing detailed
6779 information on specific entries. Keyboard navigation and
6780 mouse screen support has been added.
6782 * gatewayd/journalctl now supports HTML5/JSON
6783 Server-Sent-Events as output.
6785 * The SysV init script compatibility logic will now
6786 heuristically determine whether a script supports the
6787 "reload" verb, and only then make this available as
6790 * "systemctl status --follow" has been removed, use "journalctl
6793 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6794 have been removed since they are hardly useful to be
6797 * And I'd like to take the opportunity to specifically mention
6798 Zbigniew for his great contributions. Zbigniew, you rock!
6800 Contributions from: Andrew Eikum, Christian Hesse, Colin
6801 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6802 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6803 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6804 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6805 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6806 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6810 * If /etc/vconsole.conf is non-existent or empty we will no
6811 longer load any console font or key map at boot by
6812 default. Instead the kernel defaults will be left
6813 intact. This is definitely the right thing to do, as no
6814 configuration should mean no configuration, and hard-coding
6815 font names that are different on all archs is probably a bad
6816 idea. Also, the kernel default key map and font should be
6817 good enough for most cases anyway, and mostly identical to
6818 the userspace fonts/key maps we previously overloaded them
6819 with. If distributions want to continue to default to a
6820 non-kernel font or key map they should ship a default
6821 /etc/vconsole.conf with the appropriate contents.
6823 Contributions from: Colin Walters, Daniel J Walsh, Dave
6824 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6825 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6829 * journalctl gained a new --cursor= switch to show entries
6830 starting from the specified location in the journal.
6832 * We now enforce a size limit on journal entry fields exported
6833 with "-o json" in journalctl. Fields larger than 4K will be
6834 assigned null. This can be turned off with --all.
6836 * An (optional) journal gateway daemon is now available as
6837 "systemd-journal-gatewayd.service". This service provides
6838 access to the journal via HTTP and JSON. This functionality
6839 will be used to implement live log synchronization in both
6840 pull and push modes, but has various other users too, such
6841 as easy log access for debugging of embedded devices. Right
6842 now it is already useful to retrieve the journal via HTTP:
6844 # systemctl start systemd-journal-gatewayd.service
6845 # wget http://localhost:19531/entries
6847 This will download the journal contents in a
6848 /var/log/messages compatible format. The same as JSON:
6850 # curl -H"Accept: application/json" http://localhost:19531/entries
6852 This service is also accessible via a web browser where a
6853 single static HTML5 app is served that uses the JSON logic
6854 to enable the user to do some basic browsing of the
6855 journal. This will be extended later on. Here's an example
6856 screenshot of this app in its current state:
6858 http://0pointer.de/public/journal-gatewayd
6860 Contributions from: Kay Sievers, Lennart Poettering, Robert
6861 Milasan, Tom Gundersen
6865 * The bash completion logic is now available for journalctl
6868 * We do not mount the "cpuset" controller anymore together with
6869 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6870 started if no parameters are assigned to it. "cpuset" hence
6871 broke code that assumed it could create "cpu" groups and
6874 * journalctl -f will now subscribe to terminal size changes,
6875 and line break accordingly.
6877 Contributions from: Dave Reisner, Kay Sievers, Lennart
6878 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6882 * nspawn will now create a symlink /etc/localtime in the
6883 container environment, copying the host's timezone
6884 setting. Previously this has been done via a bind mount, but
6885 since symlinks cannot be bind mounted this has now been
6886 changed to create/update the appropriate symlink.
6888 * journalctl -n's line number argument is now optional, and
6889 will default to 10 if omitted.
6891 * journald will now log the maximum size the journal files may
6892 take up on disk. This is particularly useful if the default
6893 built-in logic of determining this parameter from the file
6894 system size is used. Use "systemctl status
6895 systemd-journald.service" to see this information.
6897 * The multi-seat X wrapper tool has been stripped down. As X
6898 is now capable of enumerating graphics devices via udev in a
6899 seat-aware way the wrapper is not strictly necessary
6900 anymore. A stripped down temporary stop-gap is still shipped
6901 until the upstream display managers have been updated to
6902 fully support the new X logic. Expect this wrapper to be
6903 removed entirely in one of the next releases.
6905 * HandleSleepKey= in logind.conf has been split up into
6906 HandleSuspendKey= and HandleHibernateKey=. The old setting
6907 is not available anymore. X11 and the kernel are
6908 distinguishing between these keys and we should too. This
6909 also means the inhibition lock for these keys has been split
6912 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6913 Poettering, Lukas Nykryn, Václav Pavlín
6917 * Whenever a unit changes state we will now log this to the
6918 journal and show along the unit's own log output in
6921 * ConditionPathIsMountPoint= can now properly detect bind
6922 mount points too. (Previously, a bind mount of one file
6923 system to another place in the same file system could not be
6924 detected as mount, since they shared struct stat's st_dev
6927 * We will now mount the cgroup controllers cpu, cpuacct,
6928 cpuset and the controllers net_cls, net_prio together by
6931 * nspawn containers will now have a virtualized boot
6932 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6933 over with a randomized ID at container initialization). This
6934 has the effect of making "journalctl -b" do the right thing
6937 * The JSON output journal serialization has been updated not
6938 to generate "endless" list objects anymore, but rather one
6939 JSON object per line. This is more in line how most JSON
6940 parsers expect JSON objects. The new output mode
6941 "json-pretty" has been added to provide similar output, but
6942 neatly aligned for readability by humans.
6944 * We dropped all explicit sync() invocations in the shutdown
6945 code. The kernel does this implicitly anyway in the kernel
6946 reboot() syscall. halt(8)'s -n option is now a compatibility
6949 * We now support virtualized reboot() in containers, as
6950 supported by newer kernels. We will fall back to exit() if
6951 CAP_SYS_REBOOT is not available to the container. Also,
6952 nspawn makes use of this now and will actually reboot the
6953 container if the containerized OS asks for that.
6955 * journalctl will only show local log output by default
6956 now. Use --merge (-m) to show remote log output, too.
6958 * libsystemd-journal gained the new sd_journal_get_usage()
6959 call to determine the current disk usage of all journal
6960 files. This is exposed in the new "journalctl --disk-usage"
6963 * journald gained a new configuration setting SplitMode= in
6964 journald.conf which may be used to control how user journals
6965 are split off. See journald.conf(5) for details.
6967 * A new condition type ConditionFileNotEmpty= has been added.
6969 * tmpfiles' "w" lines now support file globbing, to write
6970 multiple files at once.
6972 * We added Python bindings for the journal submission
6973 APIs. More Python APIs for a number of selected APIs will
6974 likely follow. Note that we intend to add native bindings
6975 only for the Python language, as we consider it common
6976 enough to deserve bindings shipped within systemd. There are
6977 various projects outside of systemd that provide bindings
6978 for languages such as PHP or Lua.
6980 * Many conditions will now resolve specifiers such as %i. In
6981 addition, PathChanged= and related directives of .path units
6982 now support specifiers as well.
6984 * There's now a new RPM macro definition for the system preset
6987 * journald will now warn if it ca not forward a message to the
6988 syslog daemon because its socket is full.
6990 * timedated will no longer write or process /etc/timezone,
6991 except on Debian. As we do not support late mounted /usr
6992 anymore /etc/localtime always being a symlink is now safe,
6993 and hence the information in /etc/timezone is not necessary
6996 * logind will now always reserve one VT for a text getty (VT6
6997 by default). Previously if more than 6 X sessions where
6998 started they took up all the VTs with auto-spawned gettys,
6999 so that no text gettys were available anymore.
7001 * udev will now automatically inform the btrfs kernel logic
7002 about btrfs RAID components showing up. This should make
7003 simple hotplug based btrfs RAID assembly work.
7005 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7006 (but not for its children which will stay at the kernel
7007 default). This should allow setups with a lot more listening
7010 * systemd will now always pass the configured timezone to the
7011 kernel at boot. timedated will do the same when the timezone
7014 * logind's inhibition logic has been updated. By default,
7015 logind will now handle the lid switch, the power and sleep
7016 keys all the time, even in graphical sessions. If DEs want
7017 to handle these events on their own they should take the new
7018 handle-power-key, handle-sleep-key and handle-lid-switch
7019 inhibitors during their runtime. A simple way to achieve
7020 that is to invoke the DE wrapped in an invocation of:
7022 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7024 * Access to unit operations is now checked via SELinux taking
7025 the unit file label and client process label into account.
7027 * systemd will now notify the administrator in the journal
7028 when he over-mounts a non-empty directory.
7030 * There are new specifiers that are resolved in unit files,
7031 for the host name (%H), the machine ID (%m) and the boot ID
7034 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7035 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7036 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7037 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7038 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7039 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7040 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7044 * Support for reading structured kernel messages from
7045 /dev/kmsg has now been added and is enabled by default.
7047 * Support for reading kernel messages from /proc/kmsg has now
7048 been removed. If you want kernel messages in the journal
7049 make sure to run a recent kernel (>= 3.5) that supports
7050 reading structured messages from /dev/kmsg (see
7051 above). /proc/kmsg is now exclusive property of classic
7052 syslog daemons again.
7054 * The libudev API gained the new
7055 udev_device_new_from_device_id() call.
7057 * The logic for file system namespace (ReadOnlyDirectory=,
7058 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7059 require pivot_root() anymore. This means fewer temporary
7060 directories are created below /tmp for this feature.
7062 * nspawn containers will now see and receive all submounts
7063 made on the host OS below the root file system of the
7066 * Forward Secure Sealing is now supported for Journal files,
7067 which provide cryptographical sealing of journal files so
7068 that attackers cannot alter log history anymore without this
7069 being detectable. Lennart will soon post a blog story about
7070 this explaining it in more detail.
7072 * There are two new service settings RestartPreventExitStatus=
7073 and SuccessExitStatus= which allow configuration of exit
7074 status (exit code or signal) which will be excepted from the
7075 restart logic, resp. consider successful.
7077 * journalctl gained the new --verify switch that can be used
7078 to check the integrity of the structure of journal files and
7079 (if Forward Secure Sealing is enabled) the contents of
7082 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7083 and similar symlinks pre-created. This makes running shells
7084 as container init process a lot more fun.
7086 * The fstab support can now handle PARTUUID= and PARTLABEL=
7089 * A new ConditionHost= condition has been added to match
7090 against the hostname (with globs) and machine ID. This is
7091 useful for clusters where a single OS image is used to
7092 provision a large number of hosts which shall run slightly
7093 different sets of services.
7095 * Services which hit the restart limit will now be placed in a
7098 Contributions from: Bertram Poettering, Dave Reisner, Huang
7099 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7100 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7104 * When running in --user mode systemd will now become a
7105 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7106 tree a lot more organized.
7108 * A new PartOf= unit dependency type has been introduced that
7109 may be used to group services in a natural way.
7111 * "systemctl enable" may now be used to enable instances of
7114 * journalctl now prints error log levels in red, and
7115 warning/notice log levels in bright white. It also supports
7116 filtering by log level now.
7118 * cgtop gained a new -n switch (similar to top), to configure
7119 the maximum number of iterations to run for. It also gained
7120 -b, to run in batch mode (accepting no input).
7122 * The suffix ".service" may now be omitted on most systemctl
7123 command lines involving service unit names.
7125 * There's a new bus call in logind to lock all sessions, as
7126 well as a loginctl verb for it "lock-sessions".
7128 * libsystemd-logind.so gained a new call sd_journal_perror()
7129 that works similar to libc perror() but logs to the journal
7130 and encodes structured information about the error number.
7132 * /etc/crypttab entries now understand the new keyfile-size=
7135 * shutdown(8) now can send a (configurable) wall message when
7136 a shutdown is cancelled.
7138 * The mount propagation mode for the root file system will now
7139 default to "shared", which is useful to make containers work
7140 nicely out-of-the-box so that they receive new mounts from
7141 the host. This can be undone locally by running "mount
7142 --make-rprivate /" if needed.
7144 * The prefdm.service file has been removed. Distributions
7145 should maintain this unit downstream if they intend to keep
7146 it around. However, we recommend writing normal unit files
7147 for display managers instead.
7149 * Since systemd is a crucial part of the OS we will now
7150 default to a number of compiler switches that improve
7151 security (hardening) such as read-only relocations, stack
7152 protection, and suchlike.
7154 * The TimeoutSec= setting for services is now split into
7155 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7156 of individual time outs for the start and the stop phase of
7159 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7160 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7161 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7162 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7163 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7164 Gundersen, Zbigniew Jędrzejewski-Szmek
7168 * The journal and id128 C APIs are now fully documented as man
7171 * Extra safety checks have been added when transitioning from
7172 the initial RAM disk to the main system to avoid accidental
7175 * /etc/crypttab entries now understand the new keyfile-offset=
7178 * systemctl -t can now be used to filter by unit load state.
7180 * The journal C API gained the new sd_journal_wait() call to
7181 make writing synchronous journal clients easier.
7183 * journalctl gained the new -D switch to show journals from a
7186 * journalctl now displays a special marker between log
7187 messages of two different boots.
7189 * The journal is now explicitly flushed to /var via a service
7190 systemd-journal-flush.service, rather than implicitly simply
7191 by seeing /var/log/journal to be writable.
7193 * journalctl (and the journal C APIs) can now match for much
7194 more complex expressions, with alternatives and
7197 * When transitioning from the initial RAM disk to the main
7198 system we will now kill all processes in a killing spree to
7199 ensure no processes stay around by accident.
7201 * Three new specifiers may be used in unit files: %u, %h, %s
7202 resolve to the user name, user home directory resp. user
7203 shell. This is useful for running systemd user instances.
7205 * We now automatically rotate journal files if their data
7206 object hash table gets a fill level > 75%. We also size the
7207 hash table based on the configured maximum file size. This
7208 together should lower hash collisions drastically and thus
7209 speed things up a bit.
7211 * journalctl gained the new "--header" switch to introspect
7212 header data of journal files.
7214 * A new setting SystemCallFilters= has been added to services
7215 which may be used to apply blacklists or whitelists to
7216 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7218 * nspawn gained a new --link-journal= switch (and quicker: -j)
7219 to link the container journal with the host. This makes it
7220 very easy to centralize log viewing on the host for all
7221 guests while still keeping the journal files separated.
7223 * Many bugfixes and optimizations
7225 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7226 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7227 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7232 * Several tools now understand kernel command line arguments,
7233 which are only read when run in an initial RAM disk. They
7234 usually follow closely their normal counterparts, but are
7237 * There's a new tool to analyze the readahead files that are
7238 automatically generated at boot. Use:
7240 /usr/lib/systemd/systemd-readahead analyze /.readahead
7242 * We now provide an early debug shell on tty9 if this enabled. Use:
7244 systemctl enable debug-shell.service
7246 * All plymouth related units have been moved into the Plymouth
7247 package. Please make sure to upgrade your Plymouth version
7250 * systemd-tmpfiles now supports getting passed the basename of
7251 a configuration file only, in which case it will look for it
7252 in all appropriate directories automatically.
7254 * udevadm info now takes a /dev or /sys path as argument, and
7255 does the right thing. Example:
7257 udevadm info /dev/sda
7258 udevadm info /sys/class/block/sda
7260 * systemctl now prints a warning if a unit is stopped but a
7261 unit that might trigger it continues to run. Example: a
7262 service is stopped but the socket that activates it is left
7265 * "systemctl status" will now mention if the log output was
7266 shortened due to rotation since a service has been started.
7268 * The journal API now exposes functions to determine the
7269 "cutoff" times due to rotation.
7271 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7272 immediately flushing of runtime logs to /var if possible,
7273 resp. for triggering immediate rotation of the journal
7276 * It is now considered an error if a service is attempted to
7277 be stopped that is not loaded.
7279 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7281 * systemd-analyze now supports Python 3
7283 * tmpfiles now supports cleaning up directories via aging
7284 where the first level dirs are always kept around but
7285 directories beneath it automatically aged. This is enabled
7286 by prefixing the age field with '~'.
7288 * Seat objects now expose CanGraphical, CanTTY properties
7289 which is required to deal with very fast bootups where the
7290 display manager might be running before the graphics drivers
7291 completed initialization.
7293 * Seat objects now expose a State property.
7295 * We now include RPM macros for service enabling/disabling
7296 based on the preset logic. We recommend RPM based
7297 distributions to make use of these macros if possible. This
7298 makes it simpler to reuse RPM spec files across
7301 * We now make sure that the collected systemd unit name is
7302 always valid when services log to the journal via
7305 * There's a new man page kernel-command-line(7) detailing all
7306 command line options we understand.
7308 * The fstab generator may now be disabled at boot by passing
7309 fstab=0 on the kernel command line.
7311 * A new kernel command line option modules-load= is now understood
7312 to load a specific kernel module statically, early at boot.
7314 * Unit names specified on the systemctl command line are now
7315 automatically escaped as needed. Also, if file system or
7316 device paths are specified they are automatically turned
7317 into the appropriate mount or device unit names. Example:
7319 systemctl status /home
7320 systemctl status /dev/sda
7322 * The SysVConsole= configuration option has been removed from
7323 system.conf parsing.
7325 * The SysV search path is no longer exported on the D-Bus
7328 * The Names= option has been removed from unit file parsing.
7330 * There's a new man page bootup(7) detailing the boot process.
7332 * Every unit and every generator we ship with systemd now
7333 comes with full documentation. The self-explanatory boot is
7336 * A couple of services gained "systemd-" prefixes in their
7337 name if they wrap systemd code, rather than only external
7338 code. Among them fsck@.service which is now
7339 systemd-fsck@.service.
7341 * The HaveWatchdog property has been removed from the D-Bus
7344 * systemd.confirm_spawn= on the kernel command line should now
7347 * There's a new man page crypttab(5) which details all options
7348 we actually understand.
7350 * systemd-nspawn gained a new --capability= switch to pass
7351 additional capabilities to the container.
7353 * timedated will now read known NTP implementation unit names
7354 from /usr/lib/systemd/ntp-units.d/*.list,
7355 systemd-timedated-ntp.target has been removed.
7357 * journalctl gained a new switch "-b" that lists log data of
7358 the current boot only.
7360 * The notify socket is in the abstract namespace again, in
7361 order to support daemons which chroot() at start-up.
7363 * There is a new Storage= configuration option for journald
7364 which allows configuration of where log data should go. This
7365 also provides a way to disable journal logging entirely, so
7366 that data collected is only forwarded to the console, the
7367 kernel log buffer or another syslog implementation.
7369 * Many bugfixes and optimizations
7371 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7372 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7373 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7374 Shawn Landden, Tom Gundersen
7378 * "systemctl help <unit>" now shows the man page if one is
7381 * Several new man pages have been added.
7383 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7384 MaxLevelConsole= can now be specified in
7385 journald.conf. These options allow reducing the amount of
7386 data stored on disk or forwarded by the log level.
7388 * TimerSlackNSec= can now be specified in system.conf for
7389 PID1. This allows system-wide power savings.
7391 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7392 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7397 * logind is now capable of (optionally) handling power and
7398 sleep keys as well as the lid switch.
7400 * journalctl now understands the syntax "journalctl
7401 /usr/bin/avahi-daemon" to get all log output of a specific
7404 * CapabilityBoundingSet= in system.conf now also influences
7405 the capability bound set of usermode helpers of the kernel.
7407 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7408 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7409 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7410 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7414 * Note that we skipped 139 releases here in order to set the
7415 new version to something that is greater than both udev's
7416 and systemd's most recent version number.
7418 * udev: all udev sources are merged into the systemd source tree now.
7419 All future udev development will happen in the systemd tree. It
7420 is still fully supported to use the udev daemon and tools without
7421 systemd running, like in initramfs or other init systems. Building
7422 udev though, will require the *build* of the systemd tree, but
7423 udev can be properly *run* without systemd.
7425 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7426 should be used to create dead device nodes as workarounds for broken
7429 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7430 no longer supported. udev_monitor_new_from_netlink() needs to be
7431 used to subscribe to events.
7433 * udev: when udevd is started by systemd, processes which are left
7434 behind by forking them off of udev rules, are unconditionally cleaned
7435 up and killed now after the event handling has finished. Services or
7436 daemons must be started as systemd services. Services can be
7437 pulled-in by udev to get started, but they can no longer be directly
7438 forked by udev rules.
7440 * udev: the daemon binary is called systemd-udevd now and installed
7441 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7442 to adapt to that, create symlink, or rename the binary after building
7445 * libudev no longer provides these symbols:
7446 udev_monitor_from_socket()
7447 udev_queue_get_failed_list_entry()
7448 udev_get_{dev,sys,run}_path()
7449 The versions number was bumped and symbol versioning introduced.
7451 * systemd-loginctl and systemd-journalctl have been renamed
7452 to loginctl and journalctl to match systemctl.
7454 * The config files: /etc/systemd/systemd-logind.conf and
7455 /etc/systemd/systemd-journald.conf have been renamed to
7456 logind.conf and journald.conf. Package updates should rename
7457 the files to the new names on upgrade.
7459 * For almost all files the license is now LGPL2.1+, changed
7460 from the previous GPL2.0+. Exceptions are some minor stuff
7461 of udev (which will be changed to LGPL2.1 eventually, too),
7462 and the MIT licensed sd-daemon.[ch] library that is suitable
7463 to be used as drop-in files.
7465 * systemd and logind now handle system sleep states, in
7466 particular suspending and hibernating.
7468 * logind now implements a sleep/shutdown/idle inhibiting logic
7469 suitable for a variety of uses. Soonishly Lennart will blog
7470 about this in more detail.
7472 * var-run.mount and var-lock.mount are no longer provided
7473 (which previously bind mounted these directories to their new
7474 places). Distributions which have not converted these
7475 directories to symlinks should consider stealing these files
7476 from git history and add them downstream.
7478 * We introduced the Documentation= field for units and added
7479 this to all our shipped units. This is useful to make it
7480 easier to explore the boot and the purpose of the various
7483 * All smaller setup units (such as
7484 systemd-vconsole-setup.service) now detect properly if they
7485 are run in a container and are skipped when
7486 appropriate. This guarantees an entirely noise-free boot in
7487 Linux container environments such as systemd-nspawn.
7489 * A framework for implementing offline system updates is now
7490 integrated, for details see:
7491 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7493 * A new service type Type=idle is available now which helps us
7494 avoiding ugly interleaving of getty output and boot status
7497 * There's now a system-wide CapabilityBoundingSet= option to
7498 globally reduce the set of capabilities for the
7499 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7500 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7501 even CAP_NET_ADMIN system-wide for secure systems.
7503 * There are now system-wide DefaultLimitXXX= options to
7504 globally change the defaults of the various resource limits
7505 for all units started by PID 1.
7507 * Harald Hoyer's systemd test suite has been integrated into
7508 systemd which allows easy testing of systemd builds in qemu
7509 and nspawn. (This is really awesome! Ask us for details!)
7511 * The fstab parser is now implemented as generator, not inside
7514 * systemctl will now warn you if .mount units generated from
7515 /etc/fstab are out of date due to changes in fstab that
7516 have not been read by systemd yet.
7518 * systemd is now suitable for usage in initrds. Dracut has
7519 already been updated to make use of this. With this in place
7520 initrds get a slight bit faster but primarily are much
7521 easier to introspect and debug since "systemctl status" in
7522 the host system can be used to introspect initrd services,
7523 and the journal from the initrd is kept around too.
7525 * systemd-delta has been added, a tool to explore differences
7526 between user/admin configuration and vendor defaults.
7528 * PrivateTmp= now affects both /tmp and /var/tmp.
7530 * Boot time status messages are now much prettier and feature
7531 proper english language. Booting up systemd has never been
7534 * Read-ahead pack files now include the inode number of all
7535 files to pre-cache. When the inode changes the pre-caching
7536 is not attempted. This should be nicer to deal with updated
7537 packages which might result in changes of read-ahead
7540 * We now temporaritly lower the kernel's read_ahead_kb variable
7541 when collecting read-ahead data to ensure the kernel's
7542 built-in read-ahead does not add noise to our measurements
7543 of necessary blocks to pre-cache.
7545 * There's now RequiresMountsFor= to add automatic dependencies
7546 for all mounts necessary for a specific file system path.
7548 * MountAuto= and SwapAuto= have been removed from
7549 system.conf. Mounting file systems at boot has to take place
7552 * nspawn now learned a new switch --uuid= to set the machine
7553 ID on the command line.
7555 * nspawn now learned the -b switch to automatically search
7558 * vt102 is now the default TERM for serial TTYs, upgraded from
7561 * systemd-logind now works on VT-less systems.
7563 * The build tree has been reorganized. The individual
7564 components now have directories of their own.
7566 * A new condition type ConditionPathIsReadWrite= is now available.
7568 * nspawn learned the new -C switch to create cgroups for the
7569 container in other hierarchies.
7571 * We now have support for hardware watchdogs, configurable in
7574 * The scheduled shutdown logic now has a public API.
7576 * We now mount /tmp as tmpfs by default, but this can be
7577 masked and /etc/fstab can override it.
7579 * Since udisks does not make use of /media anymore we are not
7580 mounting a tmpfs on it anymore.
7582 * journalctl gained a new --local switch to only interleave
7583 locally generated journal files.
7585 * We can now load the IMA policy at boot automatically.
7587 * The GTK tools have been split off into a systemd-ui.
7589 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7590 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7591 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7592 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7593 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7594 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7595 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7596 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7597 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7602 * This is mostly a bugfix release
7604 * Support optional initialization of the machine ID from the
7605 KVM or container configured UUID.
7607 * Support immediate reboots with "systemctl reboot -ff"
7609 * Show /etc/os-release data in systemd-analyze output
7611 * Many bugfixes for the journal, including endianness fixes and
7612 ensuring that disk space enforcement works
7614 * sd-login.h is C++ compatible again
7616 * Extend the /etc/os-release format on request of the Debian
7619 * We now refuse non-UTF8 strings used in various configuration
7620 and unit files. This is done to ensure we do not pass invalid
7621 data over D-Bus or expose it elsewhere.
7623 * Register Mimo USB Screens as suitable for automatic seat
7626 * Read SELinux client context from journal clients in a race
7629 * Reorder configuration file lookup order. /etc now always
7630 overrides /run in order to allow the administrator to always
7631 and unconditionally override vendor-supplied or
7632 automatically generated data.
7634 * The various user visible bits of the journal now have man
7635 pages. We still lack man pages for the journal API calls
7638 * We now ship all man pages in HTML format again in the
7641 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7642 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7643 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7648 * This is mostly a bugfix release
7650 * systems lacking /etc/os-release are no longer supported.
7652 * Various functionality updates to libsystemd-login.so
7654 * Track class of PAM logins to distinguish greeters from
7657 Contributions from: Kay Sievers, Lennart Poettering, Michael
7662 * This is an important bugfix release for v41.
7664 * Building man pages is now optional which should be useful
7665 for those building systemd from git but unwilling to install
7668 * Watchdog support for supervising services is now usable. In
7669 a future release support for hardware watchdogs
7670 (i.e. /dev/watchdog) will be added building on this.
7672 * Service start rate limiting is now configurable and can be
7673 turned off per service. When a start rate limit is hit a
7674 reboot can automatically be triggered.
7676 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7678 Contributions from: Benjamin Franzke, Bill Nottingham,
7679 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7680 Schmidt, Michał Górny, Piotr Drąg
7684 * The systemd binary is installed /usr/lib/systemd/systemd now;
7685 An existing /sbin/init symlink needs to be adapted with the
7688 * The code that loads kernel modules has been ported to invoke
7689 libkmod directly, instead of modprobe. This means we do not
7690 support systems with module-init-tools anymore.
7692 * Watchdog support is now already useful, but still not
7695 * A new kernel command line option systemd.setenv= is
7696 understood to set system wide environment variables
7697 dynamically at boot.
7699 * We now limit the set of capabilities of systemd-journald.
7701 * We now set SIGPIPE to ignore by default, since it only is
7702 useful in shell pipelines, and has little use in general
7703 code. This can be disabled with IgnoreSIPIPE=no in unit
7706 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7707 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7712 * This is mostly a bugfix release
7714 * We now expose the reason why a service failed in the
7715 "Result" D-Bus property.
7717 * Rudimentary service watchdog support (will be completed over
7718 the next few releases.)
7720 * When systemd forks off in order execute some service we will
7721 now immediately changes its argv[0] to reflect which process
7722 it will execute. This is useful to minimize the time window
7723 with a generic argv[0], which makes bootcharts more useful
7725 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7726 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7727 Mike Kazantsev, Ray Strode
7731 * This is mostly a test release, but incorporates many
7734 * New systemd-cgtop tool to show control groups by their
7737 * Linking against libacl for ACLs is optional again. If
7738 disabled, support tracking device access for active logins
7739 goes becomes unavailable, and so does access to the user
7740 journals by the respective users.
7742 * If a group "adm" exists, journal files are automatically
7743 owned by them, thus allow members of this group full access
7744 to the system journal as well as all user journals.
7746 * The journal now stores the SELinux context of the logging
7747 client for all entries.
7749 * Add C++ inclusion guards to all public headers
7751 * New output mode "cat" in the journal to print only text
7752 messages, without any meta data like date or time.
7754 * Include tiny X server wrapper as a temporary stop-gap to
7755 teach XOrg udev display enumeration. This is used by display
7756 managers such as gdm, and will go away as soon as XOrg
7757 learned native udev hotplugging for display devices.
7759 * Add new systemd-cat tool for executing arbitrary programs
7760 with STDERR/STDOUT connected to the journal. Can also act as
7761 BSD logger replacement, and does so by default.
7763 * Optionally store all locally generated coredumps in the
7764 journal along with meta data.
7766 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7767 writing short strings to files (for usage for /sys), and for
7768 creating symlinks, character and block device nodes.
7770 * New unit file option ControlGroupPersistent= to make cgroups
7771 persistent, following the mechanisms outlined in
7772 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7774 * Support multiple local RTCs in a sane way
7776 * No longer monopolize IO when replaying readahead data on
7777 rotating disks, since we might starve non-file-system IO to
7778 death, since fanotify() will not see accesses done by blkid,
7781 * Do not show kernel threads in systemd-cgls anymore, unless
7782 requested with new -k switch.
7784 Contributions from: Dan Horák, Kay Sievers, Lennart
7785 Poettering, Michal Schmidt
7789 * This is mostly a test release, but incorporates many
7792 * The git repository moved to:
7793 git://anongit.freedesktop.org/systemd/systemd
7794 ssh://git.freedesktop.org/git/systemd/systemd
7796 * First release with the journal
7797 http://0pointer.de/blog/projects/the-journal.html
7799 * The journal replaces both systemd-kmsg-syslogd and
7800 systemd-stdout-bridge.
7802 * New sd_pid_get_unit() API call in libsystemd-logind
7804 * Many systemadm clean-ups
7806 * Introduce remote-fs-pre.target which is ordered before all
7807 remote mounts and may be used to start services before all
7810 * Added Mageia support
7812 * Add bash completion for systemd-loginctl
7814 * Actively monitor PID file creation for daemons which exit in
7815 the parent process before having finished writing the PID
7816 file in the daemon process. Daemons which do this need to be
7817 fixed (i.e. PID file creation must have finished before the
7818 parent exits), but we now react a bit more gracefully to them.
7820 * Add colourful boot output, mimicking the well-known output
7821 of existing distributions.
7823 * New option PassCredentials= for socket units, for
7824 compatibility with a recent kernel ABI breakage.
7826 * /etc/rc.local is now hooked in via a generator binary, and
7827 thus will no longer act as synchronization point during
7830 * systemctl list-unit-files now supports --root=.
7832 * systemd-tmpfiles now understands two new commands: z, Z for
7833 relabelling files according to the SELinux database. This is
7834 useful to apply SELinux labels to specific files in /sys,
7837 * Output of SysV services is now forwarded to both the console
7838 and the journal by default, not only just the console.
7840 * New man pages for all APIs from libsystemd-login.
7842 * The build tree got reorganized and the build system is a
7843 lot more modular allowing embedded setups to specifically
7844 select the components of systemd they are interested in.
7846 * Support for Linux systems lacking the kernel VT subsystem is
7849 * configure's --with-rootdir= got renamed to
7850 --with-rootprefix= to follow the naming used by udev and
7853 * Unless specified otherwise we will now install to /usr instead
7854 of /usr/local by default.
7856 * Processes with '@' in argv[0][0] are now excluded from the
7857 final shut-down killing spree, following the logic explained
7859 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7861 * All processes remaining in a service cgroup when we enter
7862 the START or START_PRE states are now killed with
7863 SIGKILL. That means it is no longer possible to spawn
7864 background processes from ExecStart= lines (which was never
7865 supported anyway, and bad style).
7867 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7868 reloading of units together.
7870 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7871 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7872 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7873 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7874 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek