1 systemd System and Service Manager
3 CHANGES WITH 236 in spe:
5 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
6 in v235 has been extended to also set the dummy.ko module option
7 numdummies=0, preventing the kernel from automatically creating
8 dummy0. All dummy interfaces must now be explicitly created.
10 * systemd-resolved now maintains a new dynamic
11 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
12 recommended to make /etc/resolv.conf a symlink to it. This file
13 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
14 includes dynamically acquired search domains, achieving more correct
15 DNS resolution by software that bypasses local DNS APIs such as NSS.
17 * The "uaccess" udev tag has been dropped from /dev/kvm and
18 /dev/dri/renderD*. These devices now have the 0666 permissions by
19 default (but this may be changed at build-time). /dev/dri/renderD*
20 will now be owned by the "render" group along with /dev/kfd.
22 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
23 systemd-journal-gatewayd.service and
24 systemd-journal-upload.service. This means "nss-systemd" must be
25 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
26 services are resolved properly.
28 * In systemd-networkd, the IPv6 RA logic now optionally may announce
29 DNS server and domain information.
31 * Support for the LUKS2 on-disk format for encrypted partitions has
32 been added. This requires libcryptsetup2 during compilation and
35 * The systemd --user instance will now signal "readiness" when its
36 basic.target unit has been reached, instead of when the run queue ran
37 empty for the first time.
39 * Unit files learnt three new % specifiers that are expanded during
40 loading: %S resolves to the top-level state directory (/var/lib for
41 the system instance, $XDG_CONFIG_HOME for the user instance), %C
42 resolves to the top-level cache directory (/var/cache for the system
43 instance, $XDG_CACHE_HOME for the user instance), %L resolves to the
44 top-level logs directory (/var/log for the system instance,
45 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
46 existing %t specifier, that resolves to the top-level runtime
47 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
50 * journalctl learnt a new parameter --output-fields= for limiting the
51 set of journal fields to output in verbose and JSON output modes.
53 * systemd-timesyncd's configuration file gained a new option
54 RootDistanceMaxSec= for setting the maximum root distance of servers
55 it'll use, as well as the new options PollIntervalMinSec= and
56 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
58 * bootctl gained a new command "list" for listing all available boot
59 menu items on systems that follow the boot loader specification.
61 * systemctl gained a new --dry-run switch that shows what would be done
62 instead of doing it, and is currently supported by the shutdown and
65 * ConditionSecurity= can now detect the TOMOYO security module.
67 * Unit file [Install] sections are now also respected in unit drop-in
68 files. This is intended to be used by drop-ins under /usr/lib/.
70 * systemd-firstboot may now also set the initial keyboard mapping.
72 * Udev "changed" events for devices which are exposed as systemd
73 .device units are now propagated to units specified in
74 ReloadPropagatedFrom= as reload requests.
76 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
77 unit template name (i.e. a name in the form of 'foobar@.service',
78 without the instance component between the '@' and - the '.'), then
79 the escaped sysfs path of the device is automatically used as the
82 * SystemCallFilter= in unit files has been extended so that an "errno"
83 can be specified individually for each system call. Example:
84 SystemCallFilter=~uname:EILSEQ.
86 * The cgroup delegation logic has been substantially updated. Delegate=
87 now optionally takes a list of controllers (instead of a boolean, as
88 before), which lists the controllers to delegate at least.
90 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
92 * A new LogLevelMax= setting configures the maximum log level any
93 process of the service may log at (i.e. anything with a lesser
94 priority than what is specified is automatically dropped). A new
95 LogExtraFields= setting allows configuration of additional journal
96 fields to attach to all log records generated by any of the unit's
99 * New StandardInputData= and StandardInputText= settings along with the
100 new option StandardInput=data may be used to configure textual or
101 binary data that shall be passed to the executed service process via
102 standard input, encoded in-line in the unit file.
104 * StandardInput=, StandardOutput= and StandardError= may now be used to
105 connect stdin/stdout/stderr of executed processes directly with a
106 file or AF_UNIX socket in the file system, using the new "file:" option.
108 * A new unit file option CollectMode= has been added, that allows
109 tweaking the garbage collection logic for units. It may be used to
110 tell systemd to garbage collect units that have failed automatically
111 (normally it only GCs units that exited successfully). systemd-run
112 and systemd-mount expose this new functionality with a new -G option.
114 * "machinectl bind" may now be used to bind mount non-directories
115 (i.e. regularfiles, devices, fifos, sockets).
117 * systemd-analyze gained a new verb "calendar" for validating and
118 testing calendar time specifications to use for OnCalendar= in timer
119 units. Besides validating the expression it will calculate the next
120 time the specified expression would elapse.
122 * In addition to the pre-existing FailureAction= unit file setting
123 there's now SuccessAction=, for configuring a shutdown action to
124 execute when a unit completes successfully. This is useful in
125 particular inside containers that shall terminate after some workload
126 has been completed. Also, both options are now supported for all unit
127 types, not just services.
129 * networkds's IP rule support gained two new options
130 IncomingInterface=and OutgoingInterface= for configuring the incoming
131 and outgoing interfaces of configured rules. systemd-networkd also
132 gained support for "vxcan" network devices.
134 * networkd gained a new setting RequiredForOnline=, taking a
135 boolean. If set, systemd-wait-online will take it into consideration
136 when determining that the system is up, otherwise it will ignore the
137 interface for this purpose.
139 * The sd_notify() protocol gained support for a new operation: with
140 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
141 store again, ahead of POLLHUP or POLLERR when they are removed
144 Contributions from: aeywalee, Alan Jenkins, Alessandro Ghedini, Andrew
145 Jeddeloh, Antonio Rojas, Ari, bleep_blop, Carsten Strotmann, Christian
146 Brauner, Christian Hesse, Collin Eggert, Daniel Lockyer, Daniel Rusek,
147 Dimitri John Ledkov, Evgeny Vereshchagin, Florian Klink, Franck Bui,
148 gwendalcr, Hans de Goede, Jakub Wilk, Jérémy Rosen, jobol, John Lin,
149 juga0, Krzysztof Nowicki, Lars Karlitski, Lars Kellogg-Stedman, Lauri
150 Tirkkonen, Lennart Poettering, longersson, Lubomir Rintel, Lucas
151 Werkmeister, lukas, Lukáš Nykrýn, Lukasz Rubaszewski, Maciej
152 S. Szmigiero, macrothian, Mantas Mikulėnas, martingh, Mathieu
153 Trudel-Lapierre, Matija Skala, Michael Biebl, Michael Vogt, Michal
154 Sekletar, Mike Gilbert, Muhammet Kara, myrkr, Neil Brown, Ondrej
155 Kozina, Patrik Flykt, Peter Hutterer, Piotr Drąg, Razvan Cojocaru,
156 Robin McCorkell, Roland Hieber, Sergey Ptashnick, Shawn Landden, Shuang
157 Liu, Simon Arlott, Simon Peeters, Stefan Agner, Susant Sahani, Sylvain
158 Plantefève, Thomas Blume, Tom Stellard, Topi Miettinen, Vito Caputo,
159 Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
160 Jędrzejewski-Szmek, Zeal Jagannatha
166 * INCOMPATIBILITY: systemd-logind.service and other long-running
167 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
168 communication with the outside. This generally improves security of
169 the system, and is in almost all cases a safe and good choice, as
170 these services do not and should not provide any network-facing
171 functionality. However, systemd-logind uses the glibc NSS API to
172 query the user database. This creates problems on systems where NSS
173 is set up to directly consult network services for user database
174 lookups. In particular, this creates incompatibilities with the
175 "nss-nis" module, which attempts to directly contact the NIS/YP
176 network servers it is configured for, and will now consistently
177 fail. In such cases, it is possible to turn off IP sandboxing for
178 systemd-logind.service (set IPAddressDeny= in its [Service] section
179 to the empty string, via a .d/ unit file drop-in). Downstream
180 distributions might want to update their nss-nis packaging to include
181 such a drop-in snippet, accordingly, to hide this incompatibility
182 from the user. Another option is to make use of glibc's nscd service
183 to proxy such network requests through a privilege-separated, minimal
184 local caching daemon, or to switch to more modern technologies such
185 sssd, whose NSS hook-ups generally do not involve direct network
186 access. In general, we think it's definitely time to question the
187 implementation choices of nss-nis, i.e. whether it's a good idea
188 today to embed a network-facing loadable module into all local
189 processes that need to query the user database, including the most
190 trivial and benign ones, such as "ls". For more details about
191 IPAddressDeny= see below.
193 * A new modprobe.d drop-in is now shipped by default that sets the
194 bonding module option max_bonds=0. This overrides the kernel default,
195 to avoid conflicts and ambiguity as to whether or not bond0 should be
196 managed by systemd-networkd or not. This resolves multiple issues
197 with bond0 properties not being applied, when bond0 is configured
198 with systemd-networkd. Distributors may choose to not package this,
199 however in that case users will be prevented from correctly managing
200 bond0 interface using systemd-networkd.
202 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
203 which print the logging level and target of the system manager. They
204 complement the existing "set-log-level" and "set-log-target" verbs
205 used to change those values.
207 * journald.conf gained a new boolean setting ReadKMsg= which defaults
208 to on. If turned off kernel log messages will not be read by
209 systemd-journald or included in the logs. It also gained a new
210 setting LineMax= for configuring the maximum line length in
211 STDOUT/STDERR log streams. The new default for this value is 48K, up
212 from the previous hardcoded 2048.
214 * A new unit setting RuntimeDirectoryPreserve= has been added, which
215 allows more detailed control of what to do with a runtime directory
216 configured with RuntimeDirectory= (i.e. a directory below /run or
217 $XDG_RUNTIME_DIR) after a unit is stopped.
219 * The RuntimeDirectory= setting for units gained support for creating
220 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
221 one top-level directory.
223 * Units gained new options StateDirectory=, CacheDirectory=,
224 LogsDirectory= and ConfigurationDirectory= which are closely related
225 to RuntimeDirectory= but manage per-service directories below
226 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
227 possible to write unit files which when activated automatically gain
228 properly owned service specific directories in these locations, thus
229 making unit files self-contained and increasing compatibility with
230 stateless systems and factory reset where /etc or /var are
231 unpopulated at boot. Matching these new settings there's also
232 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
233 ConfigurationDirectoryMode= for configuring the access mode of these
234 directories. These settings are particularly useful in combination
235 with DynamicUser=yes as they provide secure, properly-owned,
236 writable, and stateful locations for storage, excluded from the
237 sandbox that such services live in otherwise.
239 * Automake support has been removed from this release. systemd is now
242 * systemd-journald will now aggressively cache client metadata during
243 runtime, speeding up log write performance under pressure. This comes
244 at a small price though: as much of the metadata is read
245 asynchronously from /proc/ (and isn't implicitly attached to log
246 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
247 metadata stored alongside a log entry might be slightly
248 out-of-date. Previously it could only be slightly newer than the log
249 message. The time window is small however, and given that the kernel
250 is unlikely to be improved anytime soon in this regard, this appears
253 * nss-myhostname/systemd-resolved will now by default synthesize an
254 A/AAAA resource record for the "_gateway" hostname, pointing to the
255 current default IP gateway. Previously it did that for the "gateway"
256 name, hampering adoption, as some distributions wanted to leave that
257 host name open for local use. The old behaviour may still be
258 requested at build time.
260 * systemd-networkd's [Address] section in .network files gained a new
261 Scope= setting for configuring the IP address scope. The [Network]
262 section gained a new boolean setting ConfigureWithoutCarrier= that
263 tells systemd-networkd to ignore link sensing when configuring the
264 device. The [DHCP] section gained a new Anonymize= boolean option for
265 turning on a number of options suggested in RFC 7844. A new
266 [RoutingPolicyRule] section has been added for configuring the IP
267 routing policy. The [Route] section has gained support for a new
268 Type= setting which permits configuring
269 blackhole/unreachable/prohibit routes.
271 * The [VRF] section in .netdev files gained a new Table= setting for
272 configuring the routing table to use. The [Tunnel] section gained a
273 new Independent= boolean field for configuring tunnels independent of
274 an underlying network interface. The [Bridge] section gained a new
275 GroupForwardMask= option for configuration of propagation of link
276 local frames between bridge ports.
278 * The WakeOnLan= setting in .link files gained support for a number of
279 new modes. A new TCP6SegmentationOffload= setting has been added for
280 configuring TCP/IPv6 hardware segmentation offload.
282 * The IPv6 RA sender implementation may now optionally send out RDNSS
283 and RDNSSL records to supply DNS configuration to peers.
285 * systemd-nspawn gained support for a new --system-call-filter= command
286 line option for adding and removing entries in the default system
287 call filter it applies. Moreover systemd-nspawn has been changed to
288 implement a system call whitelist instead of a blacklist.
290 * systemd-run gained support for a new --pipe command line option. If
291 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
292 are directly passed on to the activated transient service
293 executable. This allows invoking arbitrary processes as systemd
294 services (for example to take benefit of dependency management,
295 accounting management, resource management or log management that is
296 done automatically for services) — while still allowing them to be
297 integrated in a classic UNIX shell pipeline.
299 * When a service sends RELOAD=1 via sd_notify() and reload propagation
300 using ReloadPropagationTo= is configured, a reload is now propagated
301 to configured units. (Previously this was only done on explicitly
302 requested reloads, using "systemctl reload" or an equivalent
305 * For each service unit a restart counter is now kept: it is increased
306 each time the service is restarted due to Restart=, and may be
307 queried using "systemctl show -p NRestarts …".
309 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
310 @signal and @timer have been added, for usage with SystemCallFilter=
311 in unit files and the new --system-call-filter= command line option
312 of systemd-nspawn (see above).
314 * ExecStart= lines in unit files gained two new modifiers: when a
315 command line is prefixed with "!" the command will be executed as
316 configured, except for the credentials applied by
317 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
318 "+", but does still apply namespacing options unlike "+". There's
319 also "!!" now, which is mostly identical, but becomes a NOP on
320 systems that support ambient capabilities. This is useful to write
321 unit files that work with ambient capabilities where possible but
322 automatically fall back to traditional privilege dropping mechanisms
323 on systems where this is not supported.
325 * ListenNetlink= settings in socket units now support RDMA netlink
328 * A new unit file setting LockPersonality= has been added which permits
329 locking down the chosen execution domain ("personality") of a service
332 * A new special target "getty-pre.target" has been added, which is
333 ordered before all text logins, and may be used to order services
334 before textual logins acquire access to the console.
336 * systemd will now attempt to load the virtio-rng.ko kernel module very
337 early on if a VM environment supporting this is detected. This should
338 improve entropy during early boot in virtualized environments.
340 * A _netdev option is now supported in /etc/crypttab that operates in a
341 similar way as the same option in /etc/fstab: it permits configuring
342 encrypted devices that need to be ordered after the network is up.
343 Following this logic, two new special targets
344 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
345 added that are to cryptsetup.target what remote-fs.target and
346 remote-fs-pre.target are to local-fs.target.
348 * Service units gained a new UnsetEnvironment= setting which permits
349 unsetting specific environment variables for services that are
350 normally passed to it (for example in order to mask out locale
351 settings for specific services that can't deal with it).
353 * Units acquired a new boolean option IPAccounting=. When turned on, IP
354 traffic accounting (packet count as well as byte count) is done for
355 the service, and shown as part of "systemctl status" or "systemd-run
358 * Service units acquired two new options IPAddressAllow= and
359 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
360 for configuring a simple IP access control list for all sockets of
361 the unit. These options are available also on .slice and .socket
362 units, permitting flexible access list configuration for individual
363 services as well as groups of services (as defined by a slice unit),
364 including system-wide. Note that IP ACLs configured this way are
365 enforced on every single IPv4 and IPv6 socket created by any process
366 of the service unit, and apply to ingress as well as egress traffic.
368 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
369 structured log message is generated each time the unit is stopped,
370 containing information about the consumed resources of this
373 * A new setting KeyringMode= has been added to unit files, which may be
374 used to control how the kernel keyring is set up for executed
377 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
378 "systemctl kexec" and "systemctl exit" are now always asynchronous in
379 behaviour (that is: these commands return immediately after the
380 operation was enqueued instead of waiting for the operation to
381 complete). Previously, "systemctl poweroff" and "systemctl reboot"
382 were asynchronous on systems using systemd-logind (i.e. almost
383 always, and like they were on sysvinit), and the other three commands
384 were unconditionally synchronous. With this release this is cleaned
385 up, and callers will see the same asynchronous behaviour on all
386 systems for all five operations.
388 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
391 * .timer units now accept calendar specifications in other timezones
392 than UTC or the local timezone.
394 * The tmpfiles snippet var.conf has been changed to create
395 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
396 the "utmp" group already, and it appears to be generally understood
397 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
398 databases. Previously this was implemented correctly for all these
399 databases excepts btmp, which has been opened up like this now
400 too. Note that while the other databases are world-readable
401 (i.e. 0644), btmp is not and remains more restrictive.
403 * The systemd-resolve tool gained a new --reset-server-features
404 switch. When invoked like this systemd-resolved will forget
405 everything it learnt about the features supported by the configured
406 upstream DNS servers, and restarts the feature probing logic on the
407 next resolver look-up for them at the highest feature level
410 * The status dump systemd-resolved sends to the logs upon receiving
411 SIGUSR1 now also includes information about all DNS servers it is
412 configured to use, and the features levels it probed for them.
414 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
415 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
416 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
417 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
418 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
419 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
420 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
421 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
422 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
423 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
424 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
425 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
426 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
427 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
428 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
429 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
430 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
431 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
432 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
433 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
439 * Meson is now supported as build system in addition to Automake. It is
440 our plan to remove Automake in one of our next releases, so that
441 Meson becomes our exclusive build system. Hence, please start using
442 the Meson build system in your downstream packaging. There's plenty
443 of documentation around how to use Meson, the extremely brief
446 ./autogen.sh && ./configure && make && sudo make install
450 meson build && ninja -C build && sudo ninja -C build install
452 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
453 which permits configuring a timeout on the time a job is
454 running. This is particularly useful for setting timeouts on jobs for
457 * Unit files gained two new options ConditionUser= and ConditionGroup=
458 for conditionalizing units based on the identity of the user/group
459 running a systemd user instance.
461 * systemd-networkd now understands a new FlowLabel= setting in the
462 [VXLAN] section of .network files, as well as a Priority= in
463 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
464 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
465 gained support for configuration of GENEVE links, and IPv6 address
466 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
468 * .link files now understand a new Port= setting.
470 * systemd-networkd's DHCP support gained support for DHCP option 119
471 (domain search list).
473 * systemd-networkd gained support for serving IPv6 address ranges using
474 the Router Advertisment protocol. The new .network configuration
475 section [IPv6Prefix] may be used to configure the ranges to
476 serve. This is implemented based on a new, minimal, native server
477 implementation of RA.
479 * journalctl's --output= switch gained support for a new parameter
480 "short-iso-precise" for a mode where timestamps are shown as precise
483 * systemd-udevd's "net_id" builtin may now generate stable network
484 interface names from IBM PowerVM VIO devices as well as ACPI platform
487 * MulticastDNS support in systemd-resolved may now be explicitly
488 enabled/disabled using the new MulticastDNS= configuration file
491 * systemd-resolved may now optionally use libidn2 instead of the libidn
492 for processing internationalized domain names. Support for libidn2
493 should be considered experimental and should not be enabled by
496 * "machinectl pull-tar" and related call may now do verification of
497 downloaded images using SUSE-style .sha256 checksum files in addition
498 to the already existing support for validating using Ubuntu-style
501 * sd-bus gained support for a new sd_bus_message_appendv() call which
502 is va_list equivalent of sd_bus_message_append().
504 * sd-boot gained support for validating images using SHIM/MOK.
506 * The SMACK code learnt support for "onlycap".
508 * systemd-mount --umount is now much smarter in figuring out how to
509 properly unmount a device given its mount or device path.
511 * The code to call libnss_dns as a fallback from libnss_resolve when
512 the communication with systemd-resolved fails was removed. This
513 fallback was redundant and interfered with the [!UNAVAIL=return]
514 suffix. See nss-resolve(8) for the recommended configuration.
516 * systemd-logind may now be restarted without losing state. It stores
517 the file descriptors for devices it manages in the system manager
518 using the FDSTORE= mechanism. Please note that further changes in
519 other components may be required to make use of this (for example
520 Xorg has code to listen for stops of systemd-logind and terminate
521 itself when logind is stopped or restarted, in order to avoid using
522 stale file descriptors for graphical devices, which is now
523 counterproductive and must be reverted in order for restarts of
524 systemd-logind to be safe. See
525 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
527 * All kernel install plugins are called with the environment variable
528 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
529 /etc/machine-id. If the file is missing or empty, the variable is
530 empty and BOOT_DIR_ABS is the path of a temporary directory which is
531 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
532 is empty, all plugins should not put anything in BOOT_DIR_ABS.
534 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
535 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
536 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
537 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
538 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
539 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
540 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
541 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
542 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
543 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
544 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
545 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
546 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
547 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
548 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
549 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
550 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
551 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
552 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
553 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
554 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
555 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
556 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
557 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
558 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
559 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
560 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
561 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
562 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
569 * The "hybrid" control group mode has been modified to improve
570 compatibility with "legacy" cgroups-v1 setups. Specifically, the
571 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
572 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
573 cgroups-v1 hierarchy), the only externally visible change being that
574 the cgroups-v2 hierarchy is also mounted, to
575 /sys/fs/cgroup/unified. This should provide a large degree of
576 compatibility with "legacy" cgroups-v1, while taking benefit of the
577 better management capabilities of cgroups-v2.
579 * The default control group setup mode may be selected both a boot-time
580 via a set of kernel command line parameters (specifically:
581 systemd.unified_cgroup_hierarchy= and
582 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
583 default selected on the configure command line
584 (--with-default-hierarchy=). The upstream default is "hybrid"
585 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
586 this will change in a future systemd version to be "unified" (pure
587 cgroups-v2 mode). The third option for the compile time option is
588 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
589 distributions to default to "hybrid" mode for release distributions,
590 starting with v233. We recommend "unified" for development
591 distributions (specifically: distributions such as Fedora's rawhide)
592 as that's where things are headed in the long run. Use "legacy" for
593 greatest stability and compatibility only.
595 * Note one current limitation of "unified" and "hybrid" control group
596 setup modes: the kernel currently does not permit the systemd --user
597 instance (i.e. unprivileged code) to migrate processes between two
598 disconnected cgroup subtrees, even if both are managed and owned by
599 the user. This effectively means "systemd-run --user --scope" doesn't
600 work when invoked from outside of any "systemd --user" service or
601 scope. Specifically, it is not supported from session scopes. We are
602 working on fixing this in a future systemd version. (See #3388 for
603 further details about this.)
605 * DBus policy files are now installed into /usr rather than /etc. Make
606 sure your system has dbus >= 1.9.18 running before upgrading to this
607 version, or override the install path with --with-dbuspolicydir= .
609 * All python scripts shipped with systemd (specifically: the various
610 tests written in Python) now require Python 3.
612 * systemd unit tests can now run standalone (without the source or
613 build directories), and can be installed into /usr/lib/systemd/tests/
614 with 'make install-tests'.
616 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
617 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
620 * Support for the %c, %r, %R specifiers in unit files has been
621 removed. Specifiers are not supposed to be dependent on configuration
622 in the unit file itself (so that they resolve the same regardless
623 where used in the unit files), but these specifiers were influenced
624 by the Slice= option.
626 * The shell invoked by debug-shell.service now defaults to /bin/sh in
627 all cases. If distributions want to use a different shell for this
628 purpose (for example Fedora's /sbin/sushell) they need to specify
629 this explicitly at configure time using --with-debug-shell=.
631 * The confirmation spawn prompt has been reworked to offer the
634 (c)ontinue, proceed without asking anymore
635 (D)ump, show the state of the unit
636 (f)ail, don't execute the command and pretend it failed
638 (i)nfo, show a short summary of the unit
639 (j)obs, show jobs that are in progress
640 (s)kip, don't execute the command and pretend it succeeded
641 (y)es, execute the command
643 The 'n' choice for the confirmation spawn prompt has been removed,
644 because its meaning was confusing.
646 The prompt may now also be redirected to an alternative console by
647 specifying the console as parameter to systemd.confirm_spawn=.
649 * Services of Type=notify require a READY=1 notification to be sent
650 during startup. If no such message is sent, the service now fails,
651 even if the main process exited with a successful exit code.
653 * Services that fail to start up correctly now always have their
654 ExecStopPost= commands executed. Previously, they'd enter "failed"
655 state directly, without executing these commands.
657 * The option MulticastDNS= of network configuration files has acquired
658 an actual implementation. With MulticastDNS=yes a host can resolve
659 names of remote hosts and reply to mDNS A and AAAA requests.
661 * When units are about to be started an additional check is now done to
662 ensure that all dependencies of type BindsTo= (when used in
663 combination with After=) have been started.
665 * systemd-analyze gained a new verb "syscall-filter" which shows which
666 system call groups are defined for the SystemCallFilter= unit file
667 setting, and which system calls they contain.
669 * A new system call filter group "@filesystem" has been added,
670 consisting of various file system related system calls. Group
671 "@reboot" has been added, covering reboot, kexec and shutdown related
672 calls. Finally, group "@swap" has been added covering swap
673 configuration related calls.
675 * A new unit file option RestrictNamespaces= has been added that may be
676 used to restrict access to the various process namespace types the
677 Linux kernel provides. Specifically, it may be used to take away the
678 right for a service unit to create additional file system, network,
679 user, and other namespaces. This sandboxing option is particularly
680 relevant due to the high amount of recently discovered namespacing
681 related vulnerabilities in the kernel.
683 * systemd-udev's .link files gained support for a new AutoNegotiation=
684 setting for configuring Ethernet auto-negotiation.
686 * systemd-networkd's .network files gained support for a new
687 ListenPort= setting in the [DHCP] section to explicitly configure the
688 UDP client port the DHCP client shall listen on.
690 * .network files gained a new Unmanaged= boolean setting for explicitly
691 excluding one or more interfaces from management by systemd-networkd.
693 * The systemd-networkd ProxyARP= option has been renamed to
694 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
695 renamed to ReduceARPProxy=. The old names continue to be available
698 * systemd-networkd gained support for configuring IPv6 Proxy NDP
699 addresses via the new IPv6ProxyNDPAddress= .network file setting.
701 * systemd-networkd's bonding device support gained support for two new
702 configuration options ActiveSlave= and PrimarySlave=.
704 * The various options in the [Match] section of .network files gained
705 support for negative matching.
707 * New systemd-specific mount options are now understood in /etc/fstab:
709 x-systemd.mount-timeout= may be used to configure the maximum
710 permitted runtime of the mount command.
712 x-systemd.device-bound may be set to bind a mount point to its
713 backing device unit, in order to automatically remove a mount point
714 if its backing device is unplugged. This option may also be
715 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
716 on the block device, which is now automatically set for all CDROM
717 drives, so that mounted CDs are automatically unmounted when they are
718 removed from the drive.
720 x-systemd.after= and x-systemd.before= may be used to explicitly
721 order a mount after or before another unit or mount point.
723 * Enqueued start jobs for device units are now automatically garbage
724 collected if there are no jobs waiting for them anymore.
726 * systemctl list-jobs gained two new switches: with --after, for every
727 queued job the jobs it's waiting for are shown; with --before the
728 jobs which it's blocking are shown.
730 * systemd-nspawn gained support for ephemeral boots from disk images
731 (or in other words: --ephemeral and --image= may now be
732 combined). Moreover, ephemeral boots are now supported for normal
733 directories, even if the backing file system is not btrfs. Of course,
734 if the file system does not support file system snapshots or
735 reflinks, the initial copy operation will be relatively expensive, but
736 this should still be suitable for many use cases.
738 * Calendar time specifications in .timer units now support
739 specifications relative to the end of a month by using "~" instead of
740 "-" as separator between month and day. For example, "*-02~03" means
741 "the third last day in February". In addition a new syntax for
742 repeated events has been added using the "/" character. For example,
743 "9..17/2:00" means "every two hours from 9am to 5pm".
745 * systemd-socket-proxyd gained a new parameter --connections-max= for
746 configuring the maximum number of concurrent connections.
748 * sd-id128 gained a new API for generating unique IDs for the host in a
749 way that does not leak the machine ID. Specifically,
750 sd_id128_get_machine_app_specific() derives an ID based on the
751 machine ID a in well-defined, non-reversible, stable way. This is
752 useful whenever an identifier for the host is needed but where the
753 identifier shall not be useful to identify the system beyond the
754 scope of the application itself. (Internally this uses HMAC-SHA256 as
755 keyed hash function using the machine ID as input.)
757 * NotifyAccess= gained a new supported value "exec". When set
758 notifications are accepted from all processes systemd itself invoked,
759 including all control processes.
761 * .nspawn files gained support for defining overlay mounts using the
762 Overlay= and OverlayReadOnly= options. Previously this functionality
763 was only available on the systemd-nspawn command line.
765 * systemd-nspawn's --bind= and --overlay= options gained support for
766 bind/overlay mounts whose source lies within the container tree by
767 prefixing the source path with "+".
769 * systemd-nspawn's --bind= and --overlay= options gained support for
770 automatically allocating a temporary source directory in /var/tmp
771 that is removed when the container dies. Specifically, if the source
772 directory is specified as empty string this mechanism is selected. An
773 example usage is --overlay=+/var::/var, which creates an overlay
774 mount based on the original /var contained in the image, overlayed
775 with a temporary directory in the host's /var/tmp. This way changes
776 to /var are automatically flushed when the container shuts down.
778 * systemd-nspawn --image= option does now permit raw file system block
779 devices (in addition to images containing partition tables, as
782 * The disk image dissection logic in systemd-nspawn gained support for
783 automatically setting up LUKS encrypted as well as Verity protected
784 partitions. When a container is booted from an encrypted image the
785 passphrase is queried at start-up time. When a container with Verity
786 data is started, the root hash is search in a ".roothash" file
787 accompanying the disk image (alternatively, pass the root hash via
788 the new --root-hash= command line option).
790 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
791 be used to dissect disk images the same way as systemd-nspawn does
792 it, following the Bootable Partition Specification. It may even be
793 used to mount disk images with complex partition setups (including
794 LUKS and Verity partitions) to a local host directory, in order to
795 inspect them. This tool is not considered public API (yet), and is
796 thus not installed into /usr/bin. Please do not rely on its
797 existence, since it might go away or be changed in later systemd
800 * A new generator "systemd-verity-generator" has been added, similar in
801 style to "systemd-cryptsetup-generator", permitting automatic setup of
802 Verity root partitions when systemd boots up. In order to make use of
803 this your partition setup should follow the Discoverable Partitions
804 Specification, and the GPT partition ID of the root file system
805 partition should be identical to the upper 128bit of the Verity root
806 hash. The GPT partition ID of the Verity partition protecting it
807 should be the lower 128bit of the Verity root hash. If the partition
808 image follows this model it is sufficient to specify a single
809 "roothash=" kernel command line argument to both configure which root
810 image and verity partition to use as well as the root hash for
811 it. Note that systemd-nspawn's Verity support follows the same
812 semantics, meaning that disk images with proper Verity data in place
813 may be booted in containers with systemd-nspawn as well as on
814 physical systems via the verity generator. Also note that the "mkosi"
815 tool available at https://github.com/systemd/mkosi has been updated
816 to generate Verity protected disk images following this scheme. In
817 fact, it has been updated to generate disk images that optionally
818 implement a complete UEFI SecureBoot trust chain, involving a signed
819 kernel and initrd image that incorporates such a root hash as well as
820 a Verity-enabled root partition.
822 * The hardware database (hwdb) udev supports has been updated to carry
823 accelerometer quirks.
825 * All system services are now run with a fresh kernel keyring set up
826 for them. The invocation ID is stored by default in it, thus
827 providing a safe, non-overridable way to determine the invocation
830 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
831 options for bind mounting arbitrary paths in a service-specific
832 way. When these options are used, arbitrary host or service files and
833 directories may be mounted to arbitrary locations in the service's
836 * Documentation has been added that lists all of systemd's low-level
837 environment variables:
839 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
841 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
842 whether a specific socket file descriptor matches a specified socket
845 * systemd-firstboot has been updated to check for the
846 systemd.firstboot= kernel command line option. It accepts a boolean
847 and when set to false the first boot questions are skipped.
849 * systemd-fstab-generator has been updated to check for the
850 systemd.volatile= kernel command line option, which either takes an
851 optional boolean parameter or the special value "state". If used the
852 system may be booted in a "volatile" boot mode. Specifically,
853 "systemd.volatile" is used, the root directory will be mounted as
854 tmpfs, and only /usr is mounted from the actual root file system. If
855 "systemd.volatile=state" is used, the root directory will be mounted
856 as usual, but /var is mounted as tmpfs. This concept provides similar
857 functionality as systemd-nspawn's --volatile= option, but provides it
858 on physical boots. Use this option for implementing stateless
859 systems, or testing systems with all state and/or configuration reset
860 to the defaults. (Note though that many distributions are not
861 prepared to boot up without a populated /etc or /var, though.)
863 * systemd-gpt-auto-generator gained support for LUKS encrypted root
864 partitions. Previously it only supported LUKS encrypted partitions
865 for all other uses, except for the root partition itself.
867 * Socket units gained support for listening on AF_VSOCK sockets for
868 communication in virtualized QEMU environments.
870 * The "configure" script gained a new option --with-fallback-hostname=
871 for specifying the fallback hostname to use if none is configured in
872 /etc/hostname. For example, by specifying
873 --with-fallback-hostname=fedora it is possible to default to a
874 hostname of "fedora" on pristine installations.
876 * systemd-cgls gained support for a new --unit= switch for listing only
877 the control groups of a specific unit. Similar --user-unit= has been
878 added for listing only the control groups of a specific user unit.
880 * systemd-mount gained a new --umount switch for unmounting a mount or
881 automount point (and all mount/automount points below it).
883 * systemd will now refuse full configuration reloads (via systemctl
884 daemon-reload and related calls) unless at least 16MiB of free space
885 are available in /run. This is a safety precaution in order to ensure
886 that generators can safely operate after the reload completed.
888 * A new unit file option RootImage= has been added, which has a similar
889 effect as RootDirectory= but mounts the service's root directory from
890 a disk image instead of plain directory. This logic reuses the same
891 image dissection and mount logic that systemd-nspawn already uses,
892 and hence supports any disk images systemd-nspawn supports, including
893 those following the Discoverable Partition Specification, as well as
894 Verity enabled images. This option enables systemd to run system
895 services directly off disk images acting as resource bundles,
896 possibly even including full integrity data.
898 * A new MountAPIVFS= unit file option has been added, taking a boolean
899 argument. If enabled /proc, /sys and /dev (collectively called the
900 "API VFS") will be mounted for the service. This is only relevant if
901 RootDirectory= or RootImage= is used for the service, as these mounts
902 are of course in place in the host mount namespace anyway.
904 * systemd-nspawn gained support for a new --pivot-root= switch. If
905 specified the root directory within the container image is pivoted to
906 the specified mount point, while the original root disk is moved to a
907 different place. This option enables booting of ostree images
908 directly with systemd-nspawn.
910 * The systemd build scripts will no longer complain if the NTP server
911 addresses are not changed from the defaults. Google now supports
912 these NTP servers officially. We still recommend downstreams to
913 properly register an NTP pool with the NTP pool project though.
915 * coredumpctl gained a new "--reverse" option for printing the list
916 of coredumps in reverse order.
918 * coredumpctl will now show additional information about truncated and
919 inaccessible coredumps, as well as coredumps that are still being
920 processed. It also gained a new --quiet switch for suppressing
921 additional informational message in its output.
923 * coredumpctl gained support for only showing coredumps newer and/or
924 older than specific timestamps, using the new --since= and --until=
925 options, reminiscent of journalctl's options by the same name.
927 * The systemd-coredump logic has been improved so that it may be reused
928 to collect backtraces in non-compiled languages, for example in
929 scripting languages such as Python.
931 * machinectl will now show the UID shift of local containers, if user
932 namespacing is enabled for them.
934 * systemd will now optionally run "environment generator" binaries at
935 configuration load time. They may be used to add environment
936 variables to the environment block passed to services invoked. One
937 user environment generator is shipped by default that sets up
938 environment variables based on files dropped into /etc/environment.d
939 and ~/.config/environment.d/.
941 * systemd-resolved now includes the new, recently published 2017 DNSSEC
944 * hostnamed has been updated to report a new chassis type of
945 "convertible" to cover "foldable" laptops that can both act as a
946 tablet and as a laptop, such as various Lenovo Yoga devices.
948 Contributions from: Adrián López, Alexander Galanin, Alexander
949 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
950 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
951 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
952 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
953 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
954 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
955 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
956 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
957 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
958 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
959 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
960 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
961 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
962 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
963 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
964 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
965 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
966 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
967 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
968 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
969 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
970 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
971 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
972 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
973 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
974 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
975 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
982 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
983 RestrictAddressFamilies= enabled. These sandboxing options should
984 generally be compatible with the various external udev call-out
985 binaries we are aware of, however there may be exceptions, in
986 particular when exotic languages for these call-outs are used. In
987 this case, consider turning off these settings locally.
989 * The new RemoveIPC= option can be used to remove IPC objects owned by
990 the user or group of a service when that service exits.
992 * The new ProtectKernelModules= option can be used to disable explicit
993 load and unload operations of kernel modules by a service. In
994 addition access to /usr/lib/modules is removed if this option is set.
996 * ProtectSystem= option gained a new value "strict", which causes the
997 whole file system tree with the exception of /dev, /proc, and /sys,
998 to be remounted read-only for a service.
1000 * The new ProtectKernelTunables= option can be used to disable
1001 modification of configuration files in /sys and /proc by a service.
1002 Various directories and files are remounted read-only, so access is
1003 restricted even if the file permissions would allow it.
1005 * The new ProtectControlGroups= option can be used to disable write
1006 access by a service to /sys/fs/cgroup.
1008 * Various systemd services have been hardened with
1009 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1010 RestrictAddressFamilies=.
1012 * Support for dynamically creating users for the lifetime of a service
1013 has been added. If DynamicUser=yes is specified, user and group IDs
1014 will be allocated from the range 61184..65519 for the lifetime of the
1015 service. They can be resolved using the new nss-systemd.so NSS
1016 module. The module must be enabled in /etc/nsswitch.conf. Services
1017 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1018 any resources allocated by the service will be cleaned up when the
1019 service exits. They also have ProtectHome=read-only and
1020 ProtectSystem=strict enabled, so they are not able to make any
1021 permanent modifications to the system.
1023 * The nss-systemd module also always resolves root and nobody, making
1024 it possible to have no /etc/passwd or /etc/group files in minimal
1025 container or chroot environments.
1027 * Services may be started with their own user namespace using the new
1028 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1029 under which the service is running are mapped. All other users are
1032 * Support for the cgroup namespace has been added to systemd-nspawn. If
1033 supported by kernel, the container system started by systemd-nspawn
1034 will have its own view of the cgroup hierarchy. This new behaviour
1035 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1037 * The new MemorySwapMax= option can be used to limit the maximum swap
1038 usage under the unified cgroup hierarchy.
1040 * Support for the CPU controller in the unified cgroup hierarchy has
1041 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1042 options. This controller requires out-of-tree patches for the kernel
1043 and the support is provisional.
1045 * Mount and automount units may now be created transiently
1046 (i.e. dynamically at runtime via the bus API, instead of requiring
1047 unit files in the file system).
1049 * systemd-mount is a new tool which may mount file systems – much like
1050 mount(8), optionally pulling in additional dependencies through
1051 transient .mount and .automount units. For example, this tool
1052 automatically runs fsck on a backing block device before mounting,
1053 and allows the automount logic to be used dynamically from the
1054 command line for establishing mount points. This tool is particularly
1055 useful when dealing with removable media, as it will ensure fsck is
1056 run – if necessary – before the first access and that the file system
1057 is quickly unmounted after each access by utilizing the automount
1058 logic. This maximizes the chance that the file system on the
1059 removable media stays in a clean state, and if it isn't in a clean
1060 state is fixed automatically.
1062 * LazyUnmount=yes option for mount units has been added to expose the
1063 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1066 * /efi will be used as the mount point of the EFI boot partition, if
1067 the directory is present, and the mount point was not configured
1068 through other means (e.g. fstab). If /efi directory does not exist,
1069 /boot will be used as before. This makes it easier to automatically
1070 mount the EFI partition on systems where /boot is used for something
1073 * When operating on GPT disk images for containers, systemd-nspawn will
1074 now mount the ESP to /boot or /efi according to the same rules as PID
1075 1 running on a host. This allows tools like "bootctl" to operate
1076 correctly within such containers, in order to make container images
1077 bootable on physical systems.
1079 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1081 * Two new user session targets have been added to support running
1082 graphical sessions under the systemd --user instance:
1083 graphical-session.target and graphical-session-pre.target. See
1084 systemd.special(7) for a description of how those targets should be
1087 * The vconsole initialization code has been significantly reworked to
1088 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1089 support unicode keymaps. Font and keymap configuration will now be
1090 copied to all allocated virtual consoles.
1092 * FreeBSD's bhyve virtualization is now detected.
1094 * Information recorded in the journal for core dumps now includes the
1095 contents of /proc/mountinfo and the command line of the process at
1096 the top of the process hierarchy (which is usually the init process
1099 * systemd-journal-gatewayd learned the --directory= option to serve
1100 files from the specified location.
1102 * journalctl --root=… can be used to peruse the journal in the
1103 /var/log/ directories inside of a container tree. This is similar to
1104 the existing --machine= option, but does not require the container to
1107 * The hardware database has been extended to support
1108 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1111 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1112 specify the click rate for mice which include a horizontal wheel with
1113 a click rate that is different than the one for the vertical wheel.
1115 * systemd-run gained a new --wait option that makes service execution
1116 synchronous. (Specifically, the command will not return until the
1117 specified service binary exited.)
1119 * systemctl gained a new --wait option that causes the start command to
1120 wait until the units being started have terminated again.
1122 * A new journal output mode "short-full" has been added which displays
1123 timestamps with abbreviated English day names and adds a timezone
1124 suffix. Those timestamps include more information than the default
1125 "short" output mode, and can be passed directly to journalctl's
1126 --since= and --until= options.
1128 * /etc/resolv.conf will be bind-mounted into containers started by
1129 systemd-nspawn, if possible, so any changes to resolv.conf contents
1130 are automatically propagated to the container.
1132 * The number of instances for socket-activated services originating
1133 from a single IP address can be limited with
1134 MaxConnectionsPerSource=, extending the existing setting of
1137 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1140 * .netdev and .network configuration can now be extended through
1143 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1144 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1145 can be enabled and disabled using the new UDPSegmentationOffload=,
1146 TCPSegmentationOffload=, GenericSegmentationOffload=,
1147 GenericReceiveOffload=, LargeReceiveOffload= options in the
1148 [Link] section of .link files.
1150 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1151 Port VLAN ID can be configured for bridge devices using the new STP=,
1152 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1153 section of .netdev files.
1155 * The route table to which routes received over DHCP or RA should be
1156 added can be configured with the new RouteTable= option in the [DHCP]
1157 and [IPv6AcceptRA] sections of .network files.
1159 * The Address Resolution Protocol can be disabled on links managed by
1160 systemd-networkd using the ARP=no setting in the [Link] section of
1163 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1164 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1165 encode information about the result and exit codes of the current
1166 service runtime cycle.
1168 * systemd-sysctl will now configure kernel parameters in the order
1169 they occur in the configuration files. This matches what sysctl
1170 has been traditionally doing.
1172 * kernel-install "plugins" that are executed to perform various
1173 tasks after a new kernel is added and before an old one is removed
1174 can now return a special value to terminate the procedure and
1175 prevent any later plugins from running.
1177 * Journald's SplitMode=login setting has been deprecated. It has been
1178 removed from documentation, and its use is discouraged. In a future
1179 release it will be completely removed, and made equivalent to current
1180 default of SplitMode=uid.
1182 * Storage=both option setting in /etc/systemd/coredump.conf has been
1183 removed. With fast LZ4 compression storing the core dump twice is not
1186 * The --share-system systemd-nspawn option has been replaced with an
1187 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1188 this functionality is discouraged. In addition the variables
1189 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1190 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1191 individual namespaces.
1193 * "machinectl list" now shows the IP address of running containers in
1194 the output, as well as OS release information.
1196 * "loginctl list" now shows the TTY of each session in the output.
1198 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1199 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1200 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1201 tracking objects in a "recursive" mode, where a single client can be
1202 counted multiple times, if it takes multiple references.
1204 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1205 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1206 process using sd-bus automatically exit if the bus connection is
1209 * Bus clients of the service manager may now "pin" loaded units into
1210 memory, by taking an explicit reference on them. This is useful to
1211 ensure the client can retrieve runtime data about the service even
1212 after the service completed execution. Taking such a reference is
1213 available only for privileged clients and should be helpful to watch
1214 running services in a race-free manner, and in particular collect
1215 information about exit statuses and results.
1217 * The nss-resolve module has been changed to strictly return UNAVAIL
1218 when communication via D-Bus with resolved failed, and NOTFOUND when
1219 a lookup completed but was negative. This means it is now possible to
1220 neatly configure fallbacks using nsswitch.conf result checking
1221 expressions. Taking benefit of this, the new recommended
1222 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1224 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1226 * A new setting CtrlAltDelBurstAction= has been added to
1227 /etc/systemd/system.conf which may be used to configure the precise
1228 behaviour if the user on the console presses Ctrl-Alt-Del more often
1229 than 7 times in 2s. Previously this would unconditionally result in
1230 an expedited, immediate reboot. With this new setting the precise
1231 operation may be configured in more detail, and also turned off
1234 * In .netdev files two new settings RemoteChecksumTx= and
1235 RemoteChecksumRx= are now understood that permit configuring the
1236 remote checksumming logic for VXLAN networks.
1238 * The service manager learnt a new "invocation ID" concept for invoked
1239 services. Each runtime cycle of a service will get a new invocation
1240 ID (a 128bit random UUID) assigned that identifies the current
1241 run of the service uniquely and globally. A new invocation ID
1242 is generated each time a service starts up. The journal will store
1243 the invocation ID of a service along with any logged messages, thus
1244 making the invocation ID useful for matching the online runtime of a
1245 service with the offline log data it generated in a safe way without
1246 relying on synchronized timestamps. In many ways this new service
1247 invocation ID concept is similar to the kernel's boot ID concept that
1248 uniquely and globally identifies the runtime of each boot. The
1249 invocation ID of a service is passed to the service itself via an
1250 environment variable ($INVOCATION_ID). A new bus call
1251 GetUnitByInvocationID() has been added that is similar to GetUnit()
1252 but instead of retrieving the bus path for a unit by its name
1253 retrieves it by its invocation ID. The returned path is valid only as
1254 long as the passed invocation ID is current.
1256 * systemd-resolved gained a new "DNSStubListener" setting in
1257 resolved.conf. It either takes a boolean value or the special values
1258 "udp" and "tcp", and configures whether to enable the stub DNS
1259 listener on 127.0.0.53:53.
1261 * IP addresses configured via networkd may now carry additional
1262 configuration settings supported by the kernel. New options include:
1263 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1264 PrefixRoute=, AutoJoin=.
1266 * The PAM configuration fragment file for "user@.service" shipped with
1267 systemd (i.e. the --user instance of systemd) has been stripped to
1268 the minimum necessary to make the system boot. Previously, it
1269 contained Fedora-specific stanzas that did not apply to other
1270 distributions. It is expected that downstream distributions add
1271 additional configuration lines, matching their needs to this file,
1272 using it only as rough template of what systemd itself needs. Note
1273 that this reduced fragment does not even include an invocation of
1274 pam_limits which most distributions probably want to add, even though
1275 systemd itself does not need it. (There's also the new build time
1276 option --with-pamconfdir=no to disable installation of the PAM
1279 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1280 capability is now also dropped from its set (in addition to
1281 CAP_SYS_MKNOD as before).
1283 * In service unit files it is now possible to connect a specific named
1284 file descriptor with stdin/stdout/stdout of an executed service. The
1285 name may be specified in matching .socket units using the
1286 FileDescriptorName= setting.
1288 * A number of journal settings may now be configured on the kernel
1289 command line. Specifically, the following options are now understood:
1290 systemd.journald.max_level_console=,
1291 systemd.journald.max_level_store=,
1292 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1293 systemd.journald.max_level_wall=.
1295 * "systemctl is-enabled --full" will now show by which symlinks a unit
1296 file is enabled in the unit dependency tree.
1298 * Support for VeraCrypt encrypted partitions has been added to the
1299 "cryptsetup" logic and /etc/crypttab.
1301 * systemd-detect-virt gained support for a new --private-users switch
1302 that checks whether the invoking processes are running inside a user
1303 namespace. Similar, a new special value "private-users" for the
1304 existing ConditionVirtualization= setting has been added, permitting
1305 skipping of specific units in user namespace environments.
1307 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1308 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1309 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1310 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1311 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1312 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1313 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1314 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1315 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1316 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1317 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1318 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1319 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1320 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1321 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1322 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1323 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1324 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1325 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1326 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1327 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1328 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1329 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1330 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1331 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1332 Jędrzejewski-Szmek, Zeal Jagannatha
1334 — Santa Fe, 2016-11-03
1338 * In service units the various ExecXYZ= settings have been extended
1339 with an additional special character as first argument of the
1340 assigned value: if the character '+' is used the specified command
1341 line it will be run with full privileges, regardless of User=,
1342 Group=, CapabilityBoundingSet= and similar options. The effect is
1343 similar to the existing PermissionsStartOnly= option, but allows
1344 configuration of this concept for each executed command line
1347 * Services may now alter the service watchdog timeout at runtime by
1348 sending a WATCHDOG_USEC= message via sd_notify().
1350 * MemoryLimit= and related unit settings now optionally take percentage
1351 specifications. The percentage is taken relative to the amount of
1352 physical memory in the system (or in case of containers, the assigned
1353 amount of memory). This allows scaling service resources neatly with
1354 the amount of RAM available on the system. Similarly, systemd-logind's
1355 RuntimeDirectorySize= option now also optionally takes percentage
1358 * In similar fashion TasksMax= takes percentage values now, too. The
1359 value is taken relative to the configured maximum number of processes
1360 on the system. The per-service task maximum has been changed to 15%
1361 using this functionality. (Effectively this is an increase of 512 →
1362 4915 for service units, given the kernel's default pid_max setting.)
1364 * Calendar time specifications in .timer units now understand a ".."
1365 syntax for time ranges. Example: "4..7:10" may now be used for
1366 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1369 * The InaccessableDirectories=, ReadOnlyDirectories= and
1370 ReadWriteDirectories= unit file settings have been renamed to
1371 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1372 applied to all kinds of file nodes, and not just directories, with
1373 the exception of symlinks. Specifically these settings may now be
1374 used on block and character device nodes, UNIX sockets and FIFOS as
1375 well as regular files. The old names of these settings remain
1376 available for compatibility.
1378 * systemd will now log about all service processes it kills forcibly
1379 (using SIGKILL) because they remained after the clean shutdown phase
1380 of the service completed. This should help identifying services that
1381 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1382 systemd-logind's configuration a similar log message is generated for
1383 processes killed at the end of each session due to this setting.
1385 * systemd will now set the $JOURNAL_STREAM environment variable for all
1386 services whose stdout/stderr are connected to the Journal (which
1387 effectively means by default: all services). The variable contains
1388 the device and inode number of the file descriptor used for
1389 stdout/stderr. This may be used by invoked programs to detect whether
1390 their stdout/stderr is connected to the Journal, in which case they
1391 can switch over to direct Journal communication, thus being able to
1392 pass extended, structured metadata along with their log messages. As
1393 one example, this is now used by glib's logging primitives.
1395 * When using systemd's default tmp.mount unit for /tmp, the mount point
1396 will now be established with the "nosuid" and "nodev" options. This
1397 avoids privilege escalation attacks that put traps and exploits into
1398 /tmp. However, this might cause problems if you e. g. put container
1399 images or overlays into /tmp; if you need this, override tmp.mount's
1400 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1403 * systemd now supports the "memory" cgroup controller also on
1406 * The systemd-cgtop tool now optionally takes a control group path as
1407 command line argument. If specified, the control group list shown is
1408 limited to subgroups of that group.
1410 * The SystemCallFilter= unit file setting gained support for
1411 pre-defined, named system call filter sets. For example
1412 SystemCallFilter=@clock is now an effective way to make all clock
1413 changing-related system calls unavailable to a service. A number of
1414 similar pre-defined groups are defined. Writing system call filters
1415 for system services is simplified substantially with this new
1416 concept. Accordingly, all of systemd's own, long-running services now
1417 enable system call filtering based on this, by default.
1419 * A new service setting MemoryDenyWriteExecute= has been added, taking
1420 a boolean value. If turned on, a service may no longer create memory
1421 mappings that are writable and executable at the same time. This
1422 enhances security for services where this is enabled as it becomes
1423 harder to dynamically write and then execute memory in exploited
1424 service processes. This option has been enabled for all of systemd's
1425 own long-running services.
1427 * A new RestrictRealtime= service setting has been added, taking a
1428 boolean argument. If set the service's processes may no longer
1429 acquire realtime scheduling. This improves security as realtime
1430 scheduling may otherwise be used to easily freeze the system.
1432 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1433 value. This may be used for requesting that the system manager inside
1434 of the container reports start-up completion to nspawn which then
1435 propagates this notification further to the service manager
1436 supervising nspawn itself. A related option NotifyReady= in .nspawn
1437 files has been added too. This functionality allows ordering of the
1438 start-up of multiple containers using the usual systemd ordering
1441 * machinectl gained a new command "stop" that is an alias for
1444 * systemd-resolved gained support for contacting DNS servers on
1445 link-local IPv6 addresses.
1447 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1448 its caches. A method call for requesting the same operation has been
1449 added to the bus API too, and is made available via "systemd-resolve
1452 * systemd-resolve gained a new --status switch. If passed a brief
1453 summary of the used DNS configuration with per-interface information
1456 * resolved.conf gained a new Cache= boolean option, defaulting to
1457 on. If turned off local DNS caching is disabled. This comes with a
1458 performance penalty in particular when DNSSEC is enabled. Note that
1459 resolved disables its internal caching implicitly anyway, when the
1460 configured DNS server is on a host-local IP address such as ::1 or
1461 127.0.0.1, thus automatically avoiding double local caching.
1463 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1464 for DNS requests. This improves compatibility with local programs
1465 that do not use the libc NSS or systemd-resolved's bus APIs for name
1466 resolution. This minimal DNS service is only available to local
1467 programs and does not implement the full DNS protocol, but enough to
1468 cover local DNS clients. A new, static resolv.conf file, listing just
1469 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1470 now recommended to make /etc/resolv.conf a symlink to this file in
1471 order to route all DNS lookups to systemd-resolved, regardless if
1472 done via NSS, the bus API or raw DNS packets. Note that this local
1473 DNS service is not as fully featured as the libc NSS or
1474 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1475 used to deliver link-local address information (as this implies
1476 sending a local interface index along), LLMNR/mDNS support via this
1477 interface is severely restricted. It is thus strongly recommended for
1478 all applications to use the libc NSS API or native systemd-resolved
1481 * systemd-networkd's bridge support learned a new setting
1482 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1483 in .network files has been added for configuring VLAN bridging in
1484 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1486 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1487 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1488 now be acquired without relying on DHCPv6. Two new options
1489 UseDomains= and UseDNS= have been added to configure this behaviour.
1491 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1492 renamed IPv6AcceptRA=, without altering its behaviour. The old
1493 setting name remains available for compatibility reasons.
1495 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1496 Key=, InputKey= and OutputKey=.
1498 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1499 interface configuration.
1501 * "systemctl edit" may now be used to create new unit files by
1502 specifying the --force switch.
1504 * sd-event gained a new function sd_event_get_iteration() for
1505 requesting the current iteration counter of the event loop. It starts
1506 at zero and is increased by one with each event loop iteration.
1508 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1509 file. It can be used in lieu of %systemd_requires in packages which
1510 don't use any systemd functionality and are intended to be installed
1511 in minimal containers without systemd present. This macro provides
1512 ordering dependencies to ensure that if the package is installed in
1513 the same rpm transaction as systemd, systemd will be installed before
1514 the scriptlets for the package are executed, allowing unit presets
1517 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1518 been added to simplify packaging of generators.
1520 * The os-release file gained VERSION_CODENAME field for the
1521 distribution nickname (e.g. VERSION_CODENAME=woody).
1523 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1524 can be set to disable parsing of metadata and the creation
1525 of persistent symlinks for that device.
1527 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1528 to make them available to logged-in users has been reverted.
1530 * Much of the common code of the various systemd components is now
1531 built into an internal shared library libsystemd-shared-231.so
1532 (incorporating the systemd version number in the name, to be updated
1533 with future releases) that the components link to. This should
1534 decrease systemd footprint both in memory during runtime and on
1535 disk. Note that the shared library is not for public use, and is
1536 neither API not ABI stable, but is likely to change with every new
1537 released update. Packagers need to make sure that binaries
1538 linking to libsystemd-shared.so are updated in step with the
1541 * Configuration for "mkosi" is now part of the systemd
1542 repository. mkosi is a tool to easily build legacy-free OS images,
1543 and is available on github: https://github.com/systemd/mkosi. If
1544 "mkosi" is invoked in the build tree a new raw OS image is generated
1545 incorporating the systemd sources currently being worked on and a
1546 clean, fresh distribution installation. The generated OS image may be
1547 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1548 UEFI PC. This functionality is particularly useful to easily test
1549 local changes made to systemd in a pristine, defined environment. See
1550 HACKING for details.
1552 * configure learned the --with-support-url= option to specify the
1553 distribution's bugtracker.
1555 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1556 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1557 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1558 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1559 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1560 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1561 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1562 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1563 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1564 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1565 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1566 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1567 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1568 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1569 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1570 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1571 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1572 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1573 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1575 — Berlin, 2016-07-25
1579 * DNSSEC is now turned on by default in systemd-resolved (in
1580 "allow-downgrade" mode), but may be turned off during compile time by
1581 passing "--with-default-dnssec=no" to "configure" (and of course,
1582 during runtime with DNSSEC= in resolved.conf). We recommend
1583 downstreams to leave this on at least during development cycles and
1584 report any issues with the DNSSEC logic upstream. We are very
1585 interested in collecting feedback about the DNSSEC validator and its
1586 limitations in the wild. Note however, that DNSSEC support is
1587 probably nothing downstreams should turn on in stable distros just
1588 yet, as it might create incompatibilities with a few DNS servers and
1589 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1590 automatically whenever we detect such incompatible setups, but there
1591 might be systems we do not cover yet. Hence: please help us testing
1592 the DNSSEC code, leave this on where you can, report back, but then
1593 again don't consider turning this on in your stable, LTS or
1594 production release just yet. (Note that you have to enable
1595 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1596 and its DNSSEC mode for host name resolution from local
1599 * systemd-resolve conveniently resolves DANE records with the --tlsa
1600 option and OPENPGPKEY records with the --openpgp option. It also
1601 supports dumping raw DNS record data via the new --raw= switch.
1603 * systemd-logind will now by default terminate user processes that are
1604 part of the user session scope unit (session-XX.scope) when the user
1605 logs out. This behavior is controlled by the KillUserProcesses=
1606 setting in logind.conf, and the previous default of "no" is now
1607 changed to "yes". This means that user sessions will be properly
1608 cleaned up after, but additional steps are necessary to allow
1609 intentionally long-running processes to survive logout.
1611 While the user is logged in at least once, user@.service is running,
1612 and any service that should survive the end of any individual login
1613 session can be started at a user service or scope using systemd-run.
1614 systemd-run(1) man page has been extended with an example which shows
1615 how to run screen in a scope unit underneath user@.service. The same
1616 command works for tmux.
1618 After the user logs out of all sessions, user@.service will be
1619 terminated too, by default, unless the user has "lingering" enabled.
1620 To effectively allow users to run long-term tasks even if they are
1621 logged out, lingering must be enabled for them. See loginctl(1) for
1622 details. The default polkit policy was modified to allow users to
1623 set lingering for themselves without authentication.
1625 Previous defaults can be restored at compile time by the
1626 --without-kill-user-processes option to "configure".
1628 * systemd-logind gained new configuration settings SessionsMax= and
1629 InhibitorsMax=, both with a default of 8192. It will not register new
1630 user sessions or inhibitors above this limit.
1632 * systemd-logind will now reload configuration on SIGHUP.
1634 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1635 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1636 enable. Also, support for the "io" cgroup controller in the unified
1637 hierarchy has been added, so that the "memory", "pids" and "io" are
1638 now the controllers that are supported on the unified hierarchy.
1640 WARNING: it is not possible to use previous systemd versions with
1641 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1642 is necessary to also update systemd in the initramfs if using the
1643 unified hierarchy. An updated SELinux policy is also required.
1645 * LLDP support has been extended, and both passive (receive-only) and
1646 active (sender) modes are supported. Passive mode ("routers-only") is
1647 enabled by default in systemd-networkd. Active LLDP mode is enabled
1648 by default for containers on the internal network. The "networkctl
1649 lldp" command may be used to list information gathered. "networkctl
1650 status" will also show basic LLDP information on connected peers now.
1652 * The IAID and DUID unique identifier sent in DHCP requests may now be
1653 configured for the system and each .network file managed by
1654 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1656 * systemd-networkd gained support for configuring proxy ARP support for
1657 each interface, via the ProxyArp= setting in .network files. It also
1658 gained support for configuring the multicast querier feature of
1659 bridge devices, via the new MulticastQuerier= setting in .netdev
1660 files. Similarly, snooping on the IGMP traffic can be controlled
1661 via the new setting MulticastSnooping=.
1663 A new setting PreferredLifetime= has been added for addresses
1664 configured in .network file to configure the lifetime intended for an
1667 The systemd-networkd DHCP server gained the option EmitRouter=, which
1668 defaults to yes, to configure whether the DHCP Option 3 (Router)
1671 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1672 systemd-socket-activate and installed into /usr/bin. It is now fully
1675 * systemd-journald now uses separate threads to flush changes to disk
1676 when closing journal files, thus reducing impact of slow disk I/O on
1677 logging performance.
1679 * The sd-journal API gained two new calls
1680 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1681 can be used to open journal files using file descriptors instead of
1682 file or directory paths. sd_journal_open_container() has been
1683 deprecated, sd_journal_open_directory_fd() should be used instead
1684 with the flag SD_JOURNAL_OS_ROOT.
1686 * journalctl learned a new output mode "-o short-unix" that outputs log
1687 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1688 UTC). It also gained support for a new --no-hostname setting to
1689 suppress the hostname column in the family of "short" output modes.
1691 * systemd-ask-password now optionally skips printing of the password to
1692 stdout with --no-output which can be useful in scripts.
1694 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1695 (devices tagged with ID_MAKER_TOOL) are now tagged with
1696 "uaccess" and are available to logged in users.
1698 * The DeviceAllow= unit setting now supports specifiers (with "%").
1700 * "systemctl show" gained a new --value switch, which allows print a
1701 only the contents of a specific unit property, without also printing
1702 the property's name. Similar support was added to "show*" verbs
1703 of loginctl and machinectl that output "key=value" lists.
1705 * A new unit type "generated" was added for files dynamically generated
1706 by generator tools. Similarly, a new unit type "transient" is used
1707 for unit files created using the runtime API. "systemctl enable" will
1708 refuse to operate on such files.
1710 * A new command "systemctl revert" has been added that may be used to
1711 revert to the vendor version of a unit file, in case local changes
1712 have been made by adding drop-ins or overriding the unit file.
1714 * "machinectl clean" gained a new verb to automatically remove all or
1715 just hidden container images.
1717 * systemd-tmpfiles gained support for a new line type "e" for emptying
1718 directories, if they exist, without creating them if they don't.
1720 * systemd-nspawn gained support for automatically patching the UID/GIDs
1721 of the owners and the ACLs of all files and directories in a
1722 container tree to match the UID/GID user namespacing range selected
1723 for the container invocation. This mode is enabled via the new
1724 --private-users-chown switch. It also gained support for
1725 automatically choosing a free, previously unused UID/GID range when
1726 starting a container, via the new --private-users=pick setting (which
1727 implies --private-users-chown). Together, these options for the first
1728 time make user namespacing for nspawn containers fully automatic and
1729 thus deployable. The systemd-nspawn@.service template unit file has
1730 been changed to use this functionality by default.
1732 * systemd-nspawn gained a new --network-zone= switch, that allows
1733 creating ad-hoc virtual Ethernet links between multiple containers,
1734 that only exist as long as at least one container referencing them is
1735 running. This allows easy connecting of multiple containers with a
1736 common link that implements an Ethernet broadcast domain. Each of
1737 these network "zones" may be named relatively freely by the user, and
1738 may be referenced by any number of containers, but each container may
1739 only reference one of these "zones". On the lower level, this is
1740 implemented by an automatically managed bridge network interface for
1741 each zone, that is created when the first container referencing its
1742 zone is created and removed when the last one referencing its zone
1745 * The default start timeout may now be configured on the kernel command
1746 line via systemd.default_timeout_start_sec=. It was already
1747 configurable via the DefaultTimeoutStartSec= option in
1748 /etc/systemd/system.conf.
1750 * Socket units gained a new TriggerLimitIntervalSec= and
1751 TriggerLimitBurst= setting to configure a limit on the activation
1752 rate of the socket unit.
1754 * The LimitNICE= setting now optionally takes normal UNIX nice values
1755 in addition to the raw integer limit value. If the specified
1756 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1757 value is understood as UNIX nice value. If not prefixed like this it
1758 is understood as raw RLIMIT_NICE limit.
1760 * Note that the effect of the PrivateDevices= unit file setting changed
1761 slightly with this release: the per-device /dev file system will be
1762 mounted read-only from this version on, and will have "noexec"
1763 set. This (minor) change of behavior might cause some (exceptional)
1764 legacy software to break, when PrivateDevices=yes is set for its
1765 service. Please leave PrivateDevices= off if you run into problems
1768 * systemd-bootchart has been split out to a separate repository:
1769 https://github.com/systemd/systemd-bootchart
1771 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1772 merged into the kernel in its current form.
1774 * The compatibility libraries libsystemd-daemon.so,
1775 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1776 which have been deprecated since systemd-209 have been removed along
1777 with the corresponding pkg-config files. All symbols provided by
1778 those libraries are provided by libsystemd.so.
1780 * The Capabilities= unit file setting has been removed (it is ignored
1781 for backwards compatibility). AmbientCapabilities= and
1782 CapabilityBoundingSet= should be used instead.
1784 * A new special target has been added, initrd-root-device.target,
1785 which creates a synchronization point for dependencies of the root
1786 device in early userspace. Initramfs builders must ensure that this
1787 target is now included in early userspace.
1789 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1790 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1791 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1792 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1793 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1794 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1795 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1796 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1797 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1798 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1799 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1800 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1801 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1802 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1803 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1804 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1805 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1806 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1807 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1808 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1809 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1810 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1811 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1812 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1813 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1816 — Fairfax, 2016-05-21
1820 * The systemd-resolved DNS resolver service has gained a substantial
1821 set of new features, most prominently it may now act as a DNSSEC
1822 validating stub resolver. DNSSEC mode is currently turned off by
1823 default, but is expected to be turned on by default in one of the
1824 next releases. For now, we invite everybody to test the DNSSEC logic
1825 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1826 service also gained a full set of D-Bus interfaces, including calls
1827 to configure DNS and DNSSEC settings per link (for use by external
1828 network management software). systemd-resolved and systemd-networkd
1829 now distinguish between "search" and "routing" domains. The former
1830 are used to qualify single-label names, the latter are used purely
1831 for routing lookups within certain domains to specific links.
1832 resolved now also synthesizes RRs for all entries from /etc/hosts.
1834 * The systemd-resolve tool (which is a client utility for
1835 systemd-resolved) has been improved considerably and is now fully
1836 supported and documented. Hence it has moved from /usr/lib/systemd to
1839 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1842 * The coredump collection logic has been reworked: when a coredump is
1843 collected it is now written to disk, compressed and processed
1844 (including stacktrace extraction) from a new instantiated service
1845 systemd-coredump@.service, instead of directly from the
1846 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1847 processing large coredumps can take up a substantial amount of
1848 resources and time, and this previously happened entirely outside of
1849 systemd's service supervision. With the new logic the core_pattern
1850 hook only does minimal metadata collection before passing off control
1851 to the new instantiated service, which is configured with a time
1852 limit, a nice level and other settings to minimize negative impact on
1853 the rest of the system. Also note that the new logic will honour the
1854 RLIMIT_CORE setting of the crashed process, which now allows users
1855 and processes to turn off coredumping for their processes by setting
1858 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1859 and all forked processes by default. Previously, PID 1 would leave
1860 the setting at "0" for all processes, as set by the kernel. Note that
1861 the resource limit traditionally has no effect on the generated
1862 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1863 logic is used. Since the limit is now honoured (see above) its
1864 default has been changed so that the coredumping logic is enabled by
1865 default for all processes, while allowing specific opt-out.
1867 * When the stacktrace is extracted from processes of system users, this
1868 is now done as "systemd-coredump" user, in order to sandbox this
1869 potentially security sensitive parsing operation. (Note that when
1870 processing coredumps of normal users this is done under the user ID
1871 of process that crashed, as before.) Packagers should take notice
1872 that it is now necessary to create the "systemd-coredump" system user
1873 and group at package installation time.
1875 * The systemd-activate socket activation testing tool gained support
1876 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1877 and --seqpacket switches. It also has been extended to support both
1878 new-style and inetd-style file descriptor passing. Use the new
1879 --inetd switch to request inetd-style file descriptor passing.
1881 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1882 variable, which takes a boolean value. If set to false, ANSI color
1883 output is disabled in the tools even when run on a terminal that
1886 * The VXLAN support in networkd now supports two new settings
1887 DestinationPort= and PortRange=.
1889 * A new systemd.machine_id= kernel command line switch has been added,
1890 that may be used to set the machine ID in /etc/machine-id if it is
1891 not initialized yet. This command line option has no effect if the
1892 file is already initialized.
1894 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1895 specified command line as PID 2 rather than PID 1 in the
1896 container. In this mode PID 1 is a minimal stub init process that
1897 implements the special POSIX and Linux semantics of PID 1 regarding
1898 signal and child process management. Note that this stub init process
1899 is implemented in nspawn itself and requires no support from the
1900 container image. This new logic is useful to support running
1901 arbitrary commands in the container, as normal processes are
1902 generally not prepared to run as PID 1.
1904 * systemd-nspawn gained a new --chdir= switch for setting the current
1905 working directory for the process started in the container.
1907 * "journalctl /dev/sda" will now output all kernel log messages for
1908 specified device from the current boot, in addition to all devices
1909 that are parents of it. This should make log output about devices
1910 pretty useful, as long as kernel drivers attach enough metadata to
1911 the log messages. (The usual SATA drivers do.)
1913 * The sd-journal API gained two new calls
1914 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1915 that report whether log data from /run or /var has been found.
1917 * journalctl gained a new switch "--fields" that prints all journal
1918 record field names currently in use in the journal. This is backed
1919 by two new sd-journal API calls sd_journal_enumerate_fields() and
1920 sd_journal_restart_fields().
1922 * Most configurable timeouts in systemd now expect an argument of
1923 "infinity" to turn them off, instead of "0" as before. The semantics
1924 from now on is that a timeout of "0" means "now", and "infinity"
1925 means "never". To maintain backwards compatibility, "0" continues to
1926 turn off previously existing timeout settings.
1928 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1929 try-reload-or-restart" to clarify what it actually does: the "try"
1930 logic applies to both reloading and restarting, not just restarting.
1931 The old name continues to be accepted for compatibility.
1933 * On boot-up, when PID 1 detects that the system clock is behind the
1934 release date of the systemd version in use, the clock is now set
1935 to the latter. Previously, this was already done in timesyncd, in order
1936 to avoid running with clocks set to the various clock epochs such as
1937 1902, 1938 or 1970. With this change the logic is now done in PID 1
1938 in addition to timesyncd during early boot-up, so that it is enforced
1939 before the first process is spawned by systemd. Note that the logic
1940 in timesyncd remains, as it is more comprehensive and ensures
1941 clock monotonicity by maintaining a persistent timestamp file in
1942 /var. Since /var is generally not available in earliest boot or the
1943 initrd, this part of the logic remains in timesyncd, and is not done
1946 * Support for tweaking details in net_cls.class_id through the
1947 NetClass= configuration directive has been removed, as the kernel
1948 people have decided to deprecate that controller in cgroup v2.
1949 Userspace tools such as nftables are moving over to setting rules
1950 that are specific to the full cgroup path of a task, which obsoletes
1951 these controllers anyway. The NetClass= directive is kept around for
1952 legacy compatibility reasons. For a more in-depth description of the
1953 kernel change, please refer to the respective upstream commit:
1955 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1957 * A new service setting RuntimeMaxSec= has been added that may be used
1958 to specify a maximum runtime for a service. If the timeout is hit, the
1959 service is terminated and put into a failure state.
1961 * A new service setting AmbientCapabilities= has been added. It allows
1962 configuration of additional Linux process capabilities that are
1963 passed to the activated processes. This is only available on very
1966 * The process resource limit settings in service units may now be used
1967 to configure hard and soft limits individually.
1969 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1970 expose support for gcc's __attribute__((cleanup())) C extension.
1971 Specifically, for many object destructor functions alternative
1972 versions have been added that have names suffixed with "p" and take a
1973 pointer to a pointer to the object to destroy, instead of just a
1974 pointer to the object itself. This is useful because these destructor
1975 functions may be used directly as parameters to the cleanup
1976 construct. Internally, systemd has been a heavy user of this GCC
1977 extension for a long time, and with this change similar support is
1978 now available to consumers of the library outside of systemd. Note
1979 that by using this extension in your sources compatibility with old
1980 and strictly ANSI compatible C compilers is lost. However, all gcc or
1981 LLVM versions of recent years support this extension.
1983 * Timer units gained support for a new setting RandomizedDelaySec= that
1984 allows configuring some additional randomized delay to the configured
1985 time. This is useful to spread out timer events to avoid load peaks in
1986 clusters or larger setups.
1988 * Calendar time specifications now support sub-second accuracy.
1990 * Socket units now support listening on SCTP and UDP-lite protocol
1993 * The sd-event API now comes with a full set of man pages.
1995 * Older versions of systemd contained experimental support for
1996 compressing journal files and coredumps with the LZ4 compressor that
1997 was not compatible with the lz4 binary (due to API limitations of the
1998 lz4 library). This support has been removed; only support for files
1999 compatible with the lz4 binary remains. This LZ4 logic is now
2000 officially supported and no longer considered experimental.
2002 * The dkr image import logic has been removed again from importd. dkr's
2003 micro-services focus doesn't fit into the machine image focus of
2004 importd, and quickly got out of date with the upstream dkr API.
2006 * Creation of the /run/lock/lockdev/ directory was dropped from
2007 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2008 been available for many years. If you still need this, you need to
2009 create your own tmpfiles.d config file with:
2011 d /run/lock/lockdev 0775 root lock -
2013 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2014 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2015 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2016 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2017 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2018 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2019 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2020 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2021 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2022 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2023 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2024 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2025 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2026 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2027 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2028 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2029 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2030 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2031 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2033 — Berlin, 2016-02-11
2037 * A number of properties previously only settable in unit
2038 files are now also available as properties to set when
2039 creating transient units programmatically via the bus, as it
2040 is exposed with systemd-run's --property=
2041 setting. Specifically, these are: SyslogIdentifier=,
2042 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2043 EnvironmentFile=, ReadWriteDirectories=,
2044 ReadOnlyDirectories=, InaccessibleDirectories=,
2045 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2047 * When creating transient services via the bus API it is now
2048 possible to pass in a set of file descriptors to use as
2049 STDIN/STDOUT/STDERR for the invoked process.
2051 * Slice units may now be created transiently via the bus APIs,
2052 similar to the way service and scope units may already be
2053 created transiently.
2055 * Wherever systemd expects a calendar timestamp specification
2056 (like in journalctl's --since= and --until= switches) UTC
2057 timestamps are now supported. Timestamps suffixed with "UTC"
2058 are now considered to be in Universal Time Coordinated
2059 instead of the local timezone. Also, timestamps may now
2060 optionally be specified with sub-second accuracy. Both of
2061 these additions also apply to recurring calendar event
2062 specification, such as OnCalendar= in timer units.
2064 * journalctl gained a new "--sync" switch that asks the
2065 journal daemon to write all so far unwritten log messages to
2066 disk and sync the files, before returning.
2068 * systemd-tmpfiles learned two new line types "q" and "Q" that
2069 operate like "v", but also set up a basic btrfs quota
2070 hierarchy when used on a btrfs file system with quota
2073 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2074 instead of a subvolume (even on a btrfs file system) if the
2075 root directory is a plain directory, and not a
2076 subvolume. This should simplify things with certain chroot()
2077 environments which are not aware of the concept of btrfs
2080 * systemd-detect-virt gained a new --chroot switch to detect
2081 whether execution takes place in a chroot() environment.
2083 * CPUAffinity= now takes CPU index ranges in addition to
2086 * The various memory-related resource limit settings (such as
2087 LimitAS=) now understand the usual K, M, G, ... suffixes to
2088 the base of 1024 (IEC). Similar, the time-related resource
2089 limit settings understand the usual min, h, day, ...
2092 * There's a new system.conf setting DefaultTasksMax= to
2093 control the default TasksMax= setting for services and
2094 scopes running on the system. (TasksMax= is the primary
2095 setting that exposes the "pids" cgroup controller on systemd
2096 and was introduced in the previous systemd release.) The
2097 setting now defaults to 512, which means services that are
2098 not explicitly configured otherwise will only be able to
2099 create 512 processes or threads at maximum, from this
2100 version on. Note that this means that thread- or
2101 process-heavy services might need to be reconfigured to set
2102 TasksMax= to a higher value. It is sufficient to set
2103 TasksMax= in these specific unit files to a higher value, or
2104 even "infinity". Similar, there's now a logind.conf setting
2105 UserTasksMax= that defaults to 4096 and limits the total
2106 number of processes or tasks each user may own
2107 concurrently. nspawn containers also have the TasksMax=
2108 value set by default now, to 8192. Note that all of this
2109 only has an effect if the "pids" cgroup controller is
2110 enabled in the kernel. The general benefit of these changes
2111 should be a more robust and safer system, that provides a
2112 certain amount of per-service fork() bomb protection.
2114 * systemd-nspawn gained the new --network-veth-extra= switch
2115 to define additional and arbitrarily-named virtual Ethernet
2116 links between the host and the container.
2118 * A new service execution setting PassEnvironment= has been
2119 added that allows importing select environment variables
2120 from PID1's environment block into the environment block of
2123 * Timer units gained support for a new RemainAfterElapse=
2124 setting which takes a boolean argument. It defaults to on,
2125 exposing behaviour unchanged to previous releases. If set to
2126 off, timer units are unloaded after they elapsed if they
2127 cannot elapse again. This is particularly useful for
2128 transient timer units, which shall not stay around longer
2129 than until they first elapse.
2131 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2132 default now (the kernel default is 16). This is beneficial
2133 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2134 allows substantially larger numbers of queued
2135 datagrams. This should increase the capability of systemd to
2136 parallelize boot-up, as logging and sd_notify() are unlikely
2137 to stall execution anymore. If you need to change the value
2138 from the new defaults, use the usual sysctl.d/ snippets.
2140 * The compression framing format used by the journal or
2141 coredump processing has changed to be in line with what the
2142 official LZ4 tools generate. LZ4 compression support in
2143 systemd was considered unsupported previously, as the format
2144 was not compatible with the normal tools. With this release
2145 this has changed now, and it is hence safe for downstream
2146 distributions to turn it on. While not compressing as well
2147 as the XZ, LZ4 is substantially faster, which makes
2148 it a good default choice for the compression logic in the
2149 journal and in coredump handling.
2151 * Any reference to /etc/mtab has been dropped from
2152 systemd. The file has been obsolete since a while, but
2153 systemd refused to work on systems where it was incorrectly
2154 set up (it should be a symlink or non-existent). Please make
2155 sure to update to util-linux 2.27.1 or newer in conjunction
2156 with this systemd release, which also drops any reference to
2157 /etc/mtab. If you maintain a distribution make sure that no
2158 software you package still references it, as this is a
2159 likely source of bugs. There's also a glibc bug pending,
2160 asking for removal of any reference to this obsolete file:
2162 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2164 Note that only util-linux versions built with
2165 --enable-libmount-force-mountinfo are supported.
2167 * Support for the ".snapshot" unit type has been removed. This
2168 feature turned out to be little useful and little used, and
2169 has now been removed from the core and from systemctl.
2171 * The dependency types RequiresOverridable= and
2172 RequisiteOverridable= have been removed from systemd. They
2173 have been used only very sparingly to our knowledge and
2174 other options that provide a similar effect (such as
2175 systemctl --mode=ignore-dependencies) are much more useful
2176 and commonly used. Moreover, they were only half-way
2177 implemented as the option to control behaviour regarding
2178 these dependencies was never added to systemctl. By removing
2179 these dependency types the execution engine becomes a bit
2180 simpler. Unit files that use these dependencies should be
2181 changed to use the non-Overridable dependency types
2182 instead. In fact, when parsing unit files with these
2183 options, that's what systemd will automatically convert them
2184 too, but it will also warn, asking users to fix the unit
2185 files accordingly. Removal of these dependency types should
2186 only affect a negligible number of unit files in the wild.
2188 * Behaviour of networkd's IPForward= option changed
2189 (again). It will no longer maintain a per-interface setting,
2190 but propagate one way from interfaces where this is enabled
2191 to the global kernel setting. The global setting will be
2192 enabled when requested by a network that is set up, but
2193 never be disabled again. This change was made to make sure
2194 IPv4 and IPv6 behaviour regarding packet forwarding is
2195 similar (as the Linux IPv6 stack does not support
2196 per-interface control of this setting) and to minimize
2199 * In unit files the behaviour of %u, %U, %h, %s has
2200 changed. These specifiers will now unconditionally resolve
2201 to the various user database fields of the user that the
2202 systemd instance is running as, instead of the user
2203 configured in the specific unit via User=. Note that this
2204 effectively doesn't change much, as resolving of these
2205 specifiers was already turned off in the --system instance
2206 of systemd, as we cannot do NSS lookups from PID 1. In the
2207 --user instance of systemd these specifiers where correctly
2208 resolved, but hardly made any sense, since the user instance
2209 lacks privileges to do user switches anyway, and User= is
2210 hence useless. Moreover, even in the --user instance of
2211 systemd behaviour was awkward as it would only take settings
2212 from User= assignment placed before the specifier into
2213 account. In order to unify and simplify the logic around
2214 this the specifiers will now always resolve to the
2215 credentials of the user invoking the manager (which in case
2216 of PID 1 is the root user).
2218 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2219 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2220 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2221 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2222 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2223 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2224 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2225 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2226 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2227 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2228 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2229 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2230 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2231 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2234 — Berlin, 2015-11-18
2238 * systemd now depends on util-linux v2.27. More specifically,
2239 the newly added mount monitor feature in libmount now
2240 replaces systemd's former own implementation.
2242 * libmount mandates /etc/mtab not to be regular file, and
2243 systemd now enforces this condition at early boot.
2244 /etc/mtab has been deprecated and warned about for a very
2245 long time, so systems running systemd should already have
2246 stopped having this file around as anything else than a
2247 symlink to /proc/self/mounts.
2249 * Support for the "pids" cgroup controller has been added. It
2250 allows accounting the number of tasks in a cgroup and
2251 enforcing limits on it. This adds two new setting
2252 TasksAccounting= and TasksMax= to each unit, as well as a
2253 global option DefaultTasksAccounting=.
2255 * Support for the "net_cls" cgroup controller has been added.
2256 It allows assigning a net class ID to each task in the
2257 cgroup, which can then be used in firewall rules and traffic
2258 shaping configurations. Note that the kernel netfilter net
2259 class code does not currently work reliably for ingress
2260 packets on unestablished sockets.
2262 This adds a new config directive called NetClass= to CGroup
2263 enabled units. Allowed values are positive numbers for fixed
2264 assignments and "auto" for picking a free value
2267 * 'systemctl is-system-running' now returns 'offline' if the
2268 system is not booted with systemd. This command can now be
2269 used as a substitute for 'systemd-notify --booted'.
2271 * Watchdog timeouts have been increased to 3 minutes for all
2272 in-tree service files. Apparently, disk IO issues are more
2273 frequent than we hoped, and user reported >1 minute waiting
2276 * 'machine-id-commit' functionality has been merged into
2277 'machine-id-setup --commit'. The separate binary has been
2280 * The WorkingDirectory= directive in unit files may now be set
2281 to the special value '~'. In this case, the working
2282 directory is set to the home directory of the user
2283 configured in User=.
2285 * "machinectl shell" will now open the shell in the home
2286 directory of the selected user by default.
2288 * The CrashChVT= configuration file setting is renamed to
2289 CrashChangeVT=, following our usual logic of not
2290 abbreviating unnecessarily. The old directive is still
2291 supported for compat reasons. Also, this directive now takes
2292 an integer value between 1 and 63, or a boolean value. The
2293 formerly supported '-1' value for disabling stays around for
2296 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2297 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2298 RootDirectory= properties can now be set for transient
2301 * The systemd-analyze tool gained a new "set-log-target" verb
2302 to change the logging target the system manager logs to
2303 dynamically during runtime. This is similar to how
2304 "systemd-analyze set-log-level" already changes the log
2307 * In nspawn /sys is now mounted as tmpfs, with only a selected
2308 set of subdirectories mounted in from the real sysfs. This
2309 enhances security slightly, and is useful for ensuring user
2310 namespaces work correctly.
2312 * Support for USB FunctionFS activation has been added. This
2313 allows implementation of USB gadget services that are
2314 activated as soon as they are requested, so that they don't
2315 have to run continuously, similar to classic socket
2318 * The "systemctl exit" command now optionally takes an
2319 additional parameter that sets the exit code to return from
2320 the systemd manager when exiting. This is only relevant when
2321 running the systemd user instance, or when running the
2322 system instance in a container.
2324 * sd-bus gained the new API calls sd_bus_path_encode_many()
2325 and sd_bus_path_decode_many() that allow easy encoding and
2326 decoding of multiple identifier strings inside a D-Bus
2327 object path. Another new call sd_bus_default_flush_close()
2328 has been added to flush and close per-thread default
2331 * systemd-cgtop gained support for a -M/--machine= switch to
2332 show the control groups within a certain container only.
2334 * "systemctl kill" gained support for an optional --fail
2335 switch. If specified the requested operation will fail of no
2336 processes have been killed, because the unit had no
2337 processes attached, or similar.
2339 * A new systemd.crash_reboot=1 kernel command line option has
2340 been added that triggers a reboot after crashing. This can
2341 also be set through CrashReboot= in systemd.conf.
2343 * The RuntimeDirectory= setting now understands unit
2344 specifiers like %i or %f.
2346 * A new (still internal) library API sd-ipv4acd has been added,
2347 that implements address conflict detection for IPv4. It's
2348 based on code from sd-ipv4ll, and will be useful for
2349 detecting DHCP address conflicts.
2351 * File descriptors passed during socket activation may now be
2352 named. A new API sd_listen_fds_with_names() is added to
2353 access the names. The default names may be overridden,
2354 either in the .socket file using the FileDescriptorName=
2355 parameter, or by passing FDNAME= when storing the file
2356 descriptors using sd_notify().
2358 * systemd-networkd gained support for:
2360 - Setting the IPv6 Router Advertisement settings via
2361 IPv6AcceptRouterAdvertisements= in .network files.
2363 - Configuring the HelloTimeSec=, MaxAgeSec= and
2364 ForwardDelaySec= bridge parameters in .netdev files.
2366 - Configuring PreferredSource= for static routes in
2369 * The "ask-password" framework used to query for LUKS harddisk
2370 passwords or SSL passwords during boot gained support for
2371 caching passwords in the kernel keyring, if it is
2372 available. This makes sure that the user only has to type in
2373 a passphrase once if there are multiple objects to unlock
2374 with the same one. Previously, such password caching was
2375 available only when Plymouth was used; this moves the
2376 caching logic into the systemd codebase itself. The
2377 "systemd-ask-password" utility gained a new --keyname=
2378 switch to control which kernel keyring key to use for
2379 caching a password in. This functionality is also useful for
2380 enabling display managers such as gdm to automatically
2381 unlock the user's GNOME keyring if its passphrase, the
2382 user's password and the harddisk password are the same, if
2383 gdm-autologin is used.
2385 * When downloading tar or raw images using "machinectl
2386 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2387 file is now also downloaded, if it is available and stored
2388 next to the image file.
2390 * Units of type ".socket" gained a new boolean setting
2391 Writable= which is only useful in conjunction with
2392 ListenSpecial=. If true, enables opening the specified
2393 special file in O_RDWR mode rather than O_RDONLY mode.
2395 * systemd-rfkill has been reworked to become a singleton
2396 service that is activated through /dev/rfkill on each rfkill
2397 state change and saves the settings to disk. This way,
2398 systemd-rfkill is now compatible with devices that exist
2399 only intermittendly, and even restores state if the previous
2400 system shutdown was abrupt rather than clean.
2402 * The journal daemon gained support for vacuuming old journal
2403 files controlled by the number of files that shall remain,
2404 in addition to the already existing control by size and by
2405 date. This is useful as journal interleaving performance
2406 degrades with too many separate journal files, and allows
2407 putting an effective limit on them. The new setting defaults
2408 to 100, but this may be changed by setting SystemMaxFiles=
2409 and RuntimeMaxFiles= in journald.conf. Also, the
2410 "journalctl" tool gained the new --vacuum-files= switch to
2411 manually vacuum journal files to leave only the specified
2412 number of files in place.
2414 * udev will now create /dev/disk/by-path links for ATA devices
2415 on kernels where that is supported.
2417 * Galician, Serbian, Turkish and Korean translations were added.
2419 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2420 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2421 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2422 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2423 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2424 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2425 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2426 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2427 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2428 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2429 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2430 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2431 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2432 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2433 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2434 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2435 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2436 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2438 — Berlin, 2015-10-07
2442 * The DHCP implementation of systemd-networkd gained a set of
2445 - The DHCP server now supports emitting DNS and NTP
2446 information. It may be enabled and configured via
2447 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2448 and NTP information is enabled, but no servers are
2449 configured, the corresponding uplink information (if there
2450 is any) is propagated.
2452 - Server and client now support transmission and reception
2453 of timezone information. It can be configured via the
2454 newly introduced network options UseTimezone=,
2455 EmitTimezone=, and Timezone=. Transmission of timezone
2456 information is enabled between host and containers by
2457 default now: the container will change its local timezone
2458 to what the host has set.
2460 - Lease timeouts can now be configured via
2461 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2463 - The DHCP server improved on the stability of
2464 leases. Clients are more likely to get the same lease
2465 information back, even if the server loses state.
2467 - The DHCP server supports two new configuration options to
2468 control the lease address pool metrics, PoolOffset= and
2471 * The encapsulation limit of tunnels in systemd-networkd may
2472 now be configured via 'EncapsulationLimit='. It allows
2473 modifying the maximum additional levels of encapsulation
2474 that are permitted to be prepended to a packet.
2476 * systemd now supports the concept of user buses replacing
2477 session buses, if used with dbus-1.10 (and enabled via dbus
2478 --enable-user-session). It previously only supported this on
2479 kdbus-enabled systems, and this release expands this to
2480 'dbus-daemon' systems.
2482 * systemd-networkd now supports predictable interface names
2485 * systemd now optionally supports the new Linux kernel
2486 "unified" control group hierarchy. If enabled via the kernel
2487 command-line option 'systemd.unified_cgroup_hierarchy=1',
2488 systemd will try to mount the unified cgroup hierarchy
2489 directly on /sys/fs/cgroup. If not enabled, or not
2490 available, systemd will fall back to the legacy cgroup
2491 hierarchy setup, as before. Host system and containers can
2492 mix and match legacy and unified hierarchies as they
2493 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2494 environment variable to individually select the hierarchy to
2495 use for executed containers. By default, nspawn will use the
2496 unified hierarchy for the containers if the host uses the
2497 unified hierarchy, and the legacy hierarchy otherwise.
2498 Please note that at this point the unified hierarchy is an
2499 experimental kernel feature and is likely to change in one
2500 of the next kernel releases. Therefore, it should not be
2501 enabled by default in downstream distributions yet. The
2502 minimum required kernel version for the unified hierarchy to
2503 work is 4.2. Note that when the unified hierarchy is used
2504 for the first time delegated access to controllers is
2505 safe. Because of this systemd-nspawn containers will get
2506 access to controllers now, as will systemd user
2507 sessions. This means containers and user sessions may now
2508 manage their own resources, partitioning up what the system
2511 * A new special scope unit "init.scope" has been introduced
2512 that encapsulates PID 1 of the system. It may be used to
2513 determine resource usage and enforce resource limits on PID
2514 1 itself. PID 1 hence moved out of the root of the control
2517 * The cgtop tool gained support for filtering out kernel
2518 threads when counting tasks in a control group. Also, the
2519 count of processes is now recursively summed up by
2520 default. Two options -k and --recursive= have been added to
2521 revert to old behaviour. The tool has also been updated to
2522 work correctly in containers now.
2524 * systemd-nspawn's --bind= and --bind-ro= options have been
2525 extended to allow creation of non-recursive bind mounts.
2527 * libsystemd gained two new calls sd_pid_get_cgroup() and
2528 sd_peer_get_cgroup() which return the control group path of
2529 a process or peer of a connected AF_UNIX socket. This
2530 function call is particularly useful when implementing
2531 delegated subtrees support in the control group hierarchy.
2533 * The "sd-event" event loop API of libsystemd now supports
2534 correct dequeuing of real-time signals, without losing
2537 * When systemd requests a PolicyKit decision when managing
2538 units it will now add additional fields to the request,
2539 including unit name and desired operation. This enables more
2540 powerful PolicyKit policies, that make decisions depending
2541 on these parameters.
2543 * nspawn learnt support for .nspawn settings files, that may
2544 accompany the image files or directories of containers, and
2545 may contain additional settings for the container. This is
2546 an alternative to configuring container parameters via the
2547 nspawn command line.
2549 Contributions from: Cristian Rodríguez, Daniel Mack, David
2550 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2551 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2552 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2553 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2554 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2555 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2556 Andersen, Tom Gundersen, Torstein Husebø
2558 — Berlin, 2015-09-08
2562 * machinectl gained a new verb 'shell' which opens a fresh
2563 shell on the target container or the host. It is similar to
2564 the existing 'login' command of machinectl, but spawns the
2565 shell directly without prompting for username or
2566 password. The pseudo machine '.host' now refers to the local
2567 host and is used by default. Hence, 'machinectl shell' can
2568 be used as replacement for 'su -' which spawns a session as
2569 a fresh systemd unit in a way that is fully isolated from
2570 the originating session.
2572 * systemd-networkd learned to cope with private-zone DHCP
2573 options and allows other programs to query the values.
2575 * SELinux access control when enabling/disabling units is no
2576 longer enforced with this release. The previous
2577 implementation was incorrect, and a new corrected
2578 implementation is not yet available. As unit file operations
2579 are still protected via PolicyKit and D-Bus policy this is
2580 not a security problem. Yet, distributions which care about
2581 optimal SELinux support should probably not stabilize on
2584 * sd-bus gained support for matches of type "arg0has=", that
2585 test for membership of strings in string arrays sent in bus
2588 * systemd-resolved now dumps the contents of its DNS and LLMNR
2589 caches to the logs on reception of the SIGUSR1 signal. This
2590 is useful to debug DNS behaviour.
2592 * The coredumpctl tool gained a new --directory= option to
2593 operate on journal files in a specific directory.
2595 * "systemctl reboot" and related commands gained a new
2596 "--message=" option which may be used to set a free-text
2597 wall message when shutting down or rebooting the
2598 system. This message is also logged, which is useful for
2599 figuring out the reason for a reboot or shutdown a
2602 * The "systemd-resolve-host" tool's -i switch now takes
2603 network interface numbers as alternative to interface names.
2605 * A new unit file setting for services has been introduced:
2606 UtmpMode= allows configuration of how precisely systemd
2607 handles utmp and wtmp entries for the service if this is
2608 enabled. This allows writing services that appear similar to
2609 user sessions in the output of the "w", "who", "last" and
2612 * systemd-resolved will now locally synthesize DNS resource
2613 records for the "localhost" and "gateway" domains as well as
2614 the local hostname. This should ensure that clients querying
2615 RRs via resolved will get similar results as those going via
2616 NSS, if nss-myhostname is enabled.
2618 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2619 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2620 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2621 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2622 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2623 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2624 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2625 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2626 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2627 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2628 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2629 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2631 — Berlin, 2015-08-27
2635 * The systemd-efi-boot-generator functionality was merged into
2636 systemd-gpt-auto-generator.
2638 * systemd-networkd now supports Group Policy for vxlan
2639 devices. It can be enabled via the new boolean configuration
2640 option called 'GroupPolicyExtension='.
2642 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2643 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2644 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2646 — Berlin, 2015-07-31
2650 * The python-systemd code has been removed from the systemd repository.
2651 A new repository has been created which accommodates the code from
2652 now on, and we kindly ask distributions to create a separate package
2653 for this: https://github.com/systemd/python-systemd
2655 * The systemd daemon will now reload its main configuration
2656 (/etc/systemd/system.conf) on daemon-reload.
2658 * sd-dhcp now exposes vendor specific extensions via
2659 sd_dhcp_lease_get_vendor_specific().
2661 * systemd-networkd gained a number of new configuration options.
2663 - A new boolean configuration option for TAP devices called
2664 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2665 device, thus allowing to send and receive GSO packets.
2667 - A new tunnel configuration option called 'CopyDSCP='.
2668 If enabled, the DSCP field of ip6 tunnels is copied into the
2669 decapsulated packet.
2671 - A set of boolean bridge configuration options were added.
2672 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2673 and 'UnicastFlood=' are now parsed by networkd and applied to the
2674 respective bridge link device via the respective IFLA_BRPORT_*
2677 - A new string configuration option to override the hostname sent
2678 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2679 is true, networkd will use the configured hostname instead of the
2680 system hostname when sending DHCP requests.
2682 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2683 networkd will configure the IPv6 flow-label of the tunnel device
2684 according to RFC2460.
2686 - The 'macvtap' virtual network devices are now supported, similar to
2687 the already supported 'macvlan' devices.
2689 * systemd-resolved now implements RFC5452 to improve resilience against
2690 cache poisoning. Additionally, source port randomization is enabled
2691 by default to further protect against DNS spoofing attacks.
2693 * nss-mymachines now supports translating UIDs and GIDs of running
2694 containers with user-namespaces enabled. If a container 'foo'
2695 translates a host uid 'UID' to the container uid 'TUID', then
2696 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2697 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2698 mapped as 'vg-foo-TGID'.
2700 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2701 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2702 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2703 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2704 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2705 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2706 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2707 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2708 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2709 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2711 — Berlin, 2015-07-29
2715 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2716 There are no known issues with current sysfs, and udev does not need
2717 or should be used to work around such bugs.
2719 * udev does no longer enable USB HID power management. Several reports
2720 indicate, that some devices cannot handle that setting.
2722 * The udev accelerometer helper was removed. The functionality
2723 is now fully included in iio-sensor-proxy. But this means,
2724 older iio-sensor-proxy versions will no longer provide
2725 accelerometer/orientation data with this systemd version.
2726 Please upgrade iio-sensor-proxy to version 1.0.
2728 * networkd gained a new configuration option IPv6PrivacyExtensions=
2729 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2730 for Stateless Address") on selected networks.
2732 * For the sake of fewer build-time dependencies and less code in the
2733 main repository, the python bindings are about to be removed in the
2734 next release. A new repository has been created which accommodates
2735 the code from now on, and we kindly ask distributions to create a
2736 separate package for this. The removal will take place in v223.
2738 https://github.com/systemd/python-systemd
2740 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2741 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2742 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2743 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2744 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2745 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2746 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2747 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2748 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2749 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2751 — Berlin, 2015-07-07
2755 * The sd-bus.h and sd-event.h APIs have now been declared
2756 stable and have been added to the official interface of
2757 libsystemd.so. sd-bus implements an alternative D-Bus client
2758 library, that is relatively easy to use, very efficient and
2759 supports both classic D-Bus as well as kdbus as transport
2760 backend. sd-event is a generic event loop abstraction that
2761 is built around Linux epoll, but adds features such as event
2762 prioritization or efficient timer handling. Both APIs are good
2763 choices for C programs looking for a bus and/or event loop
2764 implementation that is minimal and does not have to be
2765 portable to other kernels.
2767 * kdbus support is no longer compile-time optional. It is now
2768 always built-in. However, it can still be disabled at
2769 runtime using the kdbus=0 kernel command line setting, and
2770 that setting may be changed to default to off, by specifying
2771 --disable-kdbus at build-time. Note though that the kernel
2772 command line setting has no effect if the kdbus.ko kernel
2773 module is not installed, in which case kdbus is (obviously)
2774 also disabled. We encourage all downstream distributions to
2775 begin testing kdbus by adding it to the kernel images in the
2776 development distributions, and leaving kdbus support in
2779 * The minimal required util-linux version has been bumped to
2782 * Support for chkconfig (--enable-chkconfig) was removed in
2783 favor of calling an abstraction tool
2784 /lib/systemd/systemd-sysv-install. This needs to be
2785 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2786 in README for details.
2788 * If there's a systemd unit and a SysV init script for the
2789 same service name, and the user executes "systemctl enable"
2790 for it (or a related call), then this will now enable both
2791 (or execute the related operation on both), not just the
2794 * The libudev API documentation has been converted from gtkdoc
2797 * gudev has been removed from the systemd tree, it is now an
2800 * The systemd-cgtop tool learnt a new --raw switch to generate
2801 "raw" (machine parsable) output.
2803 * networkd's IPForwarding= .network file setting learnt the
2804 new setting "kernel", which ensures that networkd does not
2805 change the IP forwarding sysctl from the default kernel
2808 * The systemd-logind bus API now exposes a new boolean
2809 property "Docked" that reports whether logind considers the
2810 system "docked", i.e. connected to a docking station or not.
2812 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2813 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2814 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2815 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2816 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2817 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2818 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2819 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2820 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2821 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2822 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2823 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2824 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2825 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2826 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2827 Fink, Zbigniew Jędrzejewski-Szmek
2829 — Berlin, 2015-06-19
2833 * The gudev library has been extracted into a separate repository
2834 available at: https://git.gnome.org/browse/libgudev/
2835 It is now managed as part of the Gnome project. Distributions
2836 are recommended to pass --disable-gudev to systemd and use
2837 gudev from the Gnome project instead. gudev is still included
2838 in systemd, for now. It will be removed soon, though. Please
2839 also see the announcement-thread on systemd-devel:
2840 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2842 * systemd now exposes a CPUUsageNSec= property for each
2843 service unit on the bus, that contains the overall consumed
2844 CPU time of a service (the sum of what each process of the
2845 service consumed). This value is only available if
2846 CPUAccounting= is turned on for a service, and is then shown
2847 in the "systemctl status" output.
2849 * Support for configuring alternative mappings of the old SysV
2850 runlevels to systemd targets has been removed. They are now
2851 hardcoded in a way that runlevels 2, 3, 4 all map to
2852 multi-user.target and 5 to graphical.target (which
2853 previously was already the default behaviour).
2855 * The auto-mounter logic gained support for mount point
2856 expiry, using a new TimeoutIdleSec= setting in .automount
2857 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2859 * The EFI System Partition (ESP) as mounted to /boot by
2860 systemd-efi-boot-generator will now be unmounted
2861 automatically after 2 minutes of not being used. This should
2862 minimize the risk of ESP corruptions.
2864 * New /etc/fstab options x-systemd.requires= and
2865 x-systemd.requires-mounts-for= are now supported to express
2866 additional dependencies for mounts. This is useful for
2867 journalling file systems that support external journal
2868 devices or overlay file systems that require underlying file
2869 systems to be mounted.
2871 * systemd does not support direct live-upgrades (via systemctl
2872 daemon-reexec) from versions older than v44 anymore. As no
2873 distribution we are aware of shipped such old versions in a
2874 stable release this should not be problematic.
2876 * When systemd forks off a new per-connection service instance
2877 it will now set the $REMOTE_ADDR environment variable to the
2878 remote IP address, and $REMOTE_PORT environment variable to
2879 the remote IP port. This behaviour is similar to the
2880 corresponding environment variables defined by CGI.
2882 * systemd-networkd gained support for uplink failure
2883 detection. The BindCarrier= option allows binding interface
2884 configuration dynamically to the link sense of other
2885 interfaces. This is useful to achieve behaviour like in
2888 * systemd-networkd gained support for configuring the DHCP
2889 client identifier to use when requesting leases.
2891 * systemd-networkd now has a per-network UseNTP= option to
2892 configure whether NTP server information acquired via DHCP
2893 is passed on to services like systemd-timesyncd.
2895 * systemd-networkd gained support for vti6 tunnels.
2897 * Note that systemd-networkd manages the sysctl variable
2898 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2899 it is configured for since v219. The variable controls IP
2900 forwarding, and is a per-interface alternative to the global
2901 /proc/sys/net/ipv[46]/ip_forward. This setting is
2902 configurable in the IPForward= option, which defaults to
2903 "no". This means if networkd is used for an interface it is
2904 no longer sufficient to set the global sysctl option to turn
2905 on IP forwarding! Instead, the .network file option
2906 IPForward= needs to be turned on! Note that the
2907 implementation of this behaviour was broken in v219 and has
2910 * Many bonding and vxlan options are now configurable in
2913 * systemd-nspawn gained a new --property= setting to set unit
2914 properties for the container scope. This is useful for
2915 setting resource parameters (e.g. "CPUShares=500") on
2916 containers started from the command line.
2918 * systemd-nspawn gained a new --private-users= switch to make
2919 use of user namespacing available on recent Linux kernels.
2921 * systemd-nspawn may now be called as part of a shell pipeline
2922 in which case the pipes used for stdin and stdout are passed
2923 directly to the process invoked in the container, without
2924 indirection via a pseudo tty.
2926 * systemd-nspawn gained a new switch to control the UNIX
2927 signal to use when killing the init process of the container
2930 * systemd-nspawn gained a new --overlay= switch for mounting
2931 overlay file systems into the container using the new kernel
2934 * When a container image is imported via systemd-importd and
2935 the host file system is not btrfs, a loopback block device
2936 file is created in /var/lib/machines.raw with a btrfs file
2937 system inside. It is then mounted to /var/lib/machines to
2938 enable btrfs features for container management. The loopback
2939 file and btrfs file system is grown as needed when container
2940 images are imported via systemd-importd.
2942 * systemd-machined/systemd-importd gained support for btrfs
2943 quota, to enforce container disk space limits on disk. This
2944 is exposed in "machinectl set-limit".
2946 * systemd-importd now can import containers from local .tar,
2947 .raw and .qcow2 images, and export them to .tar and .raw. It
2948 can also import dkr v2 images now from the network (on top
2951 * systemd-importd gained support for verifying downloaded
2952 images with gpg2 (previously only gpg1 was supported).
2954 * systemd-machined, systemd-logind, systemd: most bus calls
2955 are now accessible to unprivileged processes via
2956 PolicyKit. Also, systemd-logind will now allow users to kill
2957 their own sessions without further privileges or
2960 * systemd-shutdownd has been removed. This service was
2961 previously responsible for implementing scheduled shutdowns
2962 as exposed in /usr/bin/shutdown's time parameter. This
2963 functionality has now been moved into systemd-logind and is
2964 accessible via a bus interface.
2966 * "systemctl reboot" gained a new switch --firmware-setup that
2967 can be used to reboot into the EFI firmware setup, if that
2968 is available. systemd-logind now exposes an API on the bus
2969 to trigger such reboots, in case graphical desktop UIs want
2970 to cover this functionality.
2972 * "systemctl enable", "systemctl disable" and "systemctl mask"
2973 now support a new "--now" switch. If specified the units
2974 that are enabled will also be started, and the ones
2975 disabled/masked also stopped.
2977 * The Gummiboot EFI boot loader tool has been merged into
2978 systemd, and renamed to "systemd-boot". The bootctl tool has been
2979 updated to support systemd-boot.
2981 * An EFI kernel stub has been added that may be used to create
2982 kernel EFI binaries that contain not only the actual kernel,
2983 but also an initrd, boot splash, command line and OS release
2984 information. This combined binary can then be signed as a
2985 single image, so that the firmware can verify it all in one
2986 step. systemd-boot has special support for EFI binaries created
2987 like this and can extract OS release information from them
2988 and show them in the boot menu. This functionality is useful
2989 to implement cryptographically verified boot schemes.
2991 * Optional support has been added to systemd-fsck to pass
2992 fsck's progress report to an AF_UNIX socket in the file
2995 * udev will no longer create device symlinks for all block
2996 devices by default. A blacklist for excluding special block
2997 devices from this logic has been turned into a whitelist
2998 that requires picking block devices explicitly that require
3001 * A new (currently still internal) API sd-device.h has been
3002 added to libsystemd. This modernized API is supposed to
3003 replace libudev eventually. In fact, already much of libudev
3004 is now just a wrapper around sd-device.h.
3006 * A new hwdb database for storing metadata about pointing
3007 stick devices has been added.
3009 * systemd-tmpfiles gained support for setting file attributes
3010 similar to the "chattr" tool with new 'h' and 'H' lines.
3012 * systemd-journald will no longer unconditionally set the
3013 btrfs NOCOW flag on new journal files. This is instead done
3014 with tmpfiles snippet using the new 'h' line type. This
3015 allows easy disabling of this logic, by masking the
3016 journal-nocow.conf tmpfiles file.
3018 * systemd-journald will now translate audit message types to
3019 human readable identifiers when writing them to the
3020 journal. This should improve readability of audit messages.
3022 * The LUKS logic gained support for the offset= and skip=
3023 options in /etc/crypttab, as previously implemented by
3026 * /usr/lib/os-release gained a new optional field VARIANT= for
3027 distributions that support multiple variants (such as a
3028 desktop edition, a server edition, ...)
3030 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3031 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3032 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3033 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3034 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3035 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3036 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3037 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3038 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3039 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3040 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3041 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3042 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3043 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3044 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3045 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3046 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3047 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3048 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3049 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3050 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3051 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3052 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3053 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3054 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3055 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3056 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3058 — Berlin, 2015-05-22
3062 * Introduce a new API "sd-hwdb.h" for querying the hardware
3063 metadata database. With this minimal interface one can query
3064 and enumerate the udev hwdb, decoupled from the old libudev
3065 library. libudev's interface for this is now only a wrapper
3066 around sd-hwdb. A new tool systemd-hwdb has been added to
3067 interface with and update the database.
3069 * When any of systemd's tools copies files (for example due to
3070 tmpfiles' C lines) a btrfs reflink will attempted first,
3071 before bytewise copying is done.
3073 * systemd-nspawn gained a new --ephemeral switch. When
3074 specified a btrfs snapshot is taken of the container's root
3075 directory, and immediately removed when the container
3076 terminates again. Thus, a container can be started whose
3077 changes never alter the container's root directory, and are
3078 lost on container termination. This switch can also be used
3079 for starting a container off the root file system of the
3080 host without affecting the host OS. This switch is only
3081 available on btrfs file systems.
3083 * systemd-nspawn gained a new --template= switch. It takes the
3084 path to a container tree to use as template for the tree
3085 specified via --directory=, should that directory be
3086 missing. This allows instantiating containers dynamically,
3087 on first run. This switch is only available on btrfs file
3090 * When a .mount unit refers to a mount point on which multiple
3091 mounts are stacked, and the .mount unit is stopped all of
3092 the stacked mount points will now be unmounted until no
3093 mount point remains.
3095 * systemd now has an explicit notion of supported and
3096 unsupported unit types. Jobs enqueued for unsupported unit
3097 types will now fail with an "unsupported" error code. More
3098 specifically .swap, .automount and .device units are not
3099 supported in containers, .busname units are not supported on
3100 non-kdbus systems. .swap and .automount are also not
3101 supported if their respective kernel compile time options
3104 * machinectl gained support for two new "copy-from" and
3105 "copy-to" commands for copying files from a running
3106 container to the host or vice versa.
3108 * machinectl gained support for a new "bind" command to bind
3109 mount host directories into local containers. This is
3110 currently only supported for nspawn containers.
3112 * networkd gained support for configuring bridge forwarding
3113 database entries (fdb) from .network files.
3115 * A new tiny daemon "systemd-importd" has been added that can
3116 download container images in tar, raw, qcow2 or dkr formats,
3117 and make them available locally in /var/lib/machines, so
3118 that they can run as nspawn containers. The daemon can GPG
3119 verify the downloads (not supported for dkr, since it has no
3120 provisions for verifying downloads). It will transparently
3121 decompress bz2, xz, gzip compressed downloads if necessary,
3122 and restore sparse files on disk. The daemon uses privilege
3123 separation to ensure the actual download logic runs with
3124 fewer privileges than the daemon itself. machinectl has
3125 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3126 make the functionality of importd available to the
3127 user. With this in place the Fedora and Ubuntu "Cloud"
3128 images can be downloaded and booted as containers unmodified
3129 (the Fedora images lack the appropriate GPG signature files
3130 currently, so they cannot be verified, but this will change
3131 soon, hopefully). Note that downloading images is currently
3132 only fully supported on btrfs.
3134 * machinectl is now able to list container images found in
3135 /var/lib/machines, along with some metadata about sizes of
3136 disk and similar. If the directory is located on btrfs and
3137 quota is enabled, this includes quota display. A new command
3138 "image-status" has been added that shows additional
3139 information about images.
3141 * machinectl is now able to clone container images
3142 efficiently, if the underlying file system (btrfs) supports
3143 it, with the new "machinectl clone" command. It also
3144 gained commands for renaming and removing images, as well as
3145 marking them read-only or read-write (supported also on
3146 legacy file systems).
3148 * networkd gained support for collecting LLDP network
3149 announcements, from hardware that supports this. This is
3150 shown in networkctl output.
3152 * systemd-run gained support for a new -t (--pty) switch for
3153 invoking a binary on a pty whose input and output is
3154 connected to the invoking terminal. This allows executing
3155 processes as system services while interactively
3156 communicating with them via the terminal. Most interestingly
3157 this is supported across container boundaries. Invoking
3158 "systemd-run -t /bin/bash" is an alternative to running a
3159 full login session, the difference being that the former
3160 will not register a session, nor go through the PAM session
3163 * tmpfiles gained support for a new "v" line type for creating
3164 btrfs subvolumes. If the underlying file system is a legacy
3165 file system, this automatically degrades to creating a
3166 normal directory. Among others /var/lib/machines is now
3167 created like this at boot, should it be missing.
3169 * The directory /var/lib/containers/ has been deprecated and
3170 been replaced by /var/lib/machines. The term "machines" has
3171 been used in the systemd context as generic term for both
3172 VMs and containers, and hence appears more appropriate for
3173 this, as the directory can also contain raw images bootable
3176 * systemd-nspawn when invoked with -M but without --directory=
3177 or --image= is now capable of searching for the container
3178 root directory, subvolume or disk image automatically, in
3179 /var/lib/machines. systemd-nspawn@.service has been updated
3180 to make use of this, thus allowing it to be used for raw
3183 * A new machines.target unit has been introduced that is
3184 supposed to group all containers/VMs invoked as services on
3185 the system. systemd-nspawn@.service has been updated to
3186 integrate with that.
3188 * machinectl gained a new "start" command, for invoking a
3189 container as a service. "machinectl start foo" is mostly
3190 equivalent to "systemctl start systemd-nspawn@foo.service",
3191 but handles escaping in a nicer way.
3193 * systemd-nspawn will now mount most of the cgroupfs tree
3194 read-only into each container, with the exception of the
3195 container's own subtree in the name=systemd hierarchy.
3197 * journald now sets the special FS_NOCOW file flag for its
3198 journal files. This should improve performance on btrfs, by
3199 avoiding heavy fragmentation when journald's write-pattern
3200 is used on COW file systems. It degrades btrfs' data
3201 integrity guarantees for the files to the same levels as for
3202 ext3/ext4 however. This should be OK though as journald does
3203 its own data integrity checks and all its objects are
3204 checksummed on disk. Also, journald should handle btrfs disk
3205 full events a lot more gracefully now, by processing SIGBUS
3206 errors, and not relying on fallocate() anymore.
3208 * When journald detects that journal files it is writing to
3209 have been deleted it will immediately start new journal
3212 * systemd now provides a way to store file descriptors
3213 per-service in PID 1. This is useful for daemons to ensure
3214 that fds they require are not lost during a daemon
3215 restart. The fds are passed to the daemon on the next
3216 invocation in the same way socket activation fds are
3217 passed. This is now used by journald to ensure that the
3218 various sockets connected to all the system's stdout/stderr
3219 are not lost when journald is restarted. File descriptors
3220 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3221 an extension to sd_notify(). Note that a limit is enforced
3222 on the number of fds a service can store in PID 1, and it
3223 defaults to 0, so that no fds may be stored, unless this is
3224 explicitly turned on.
3226 * The default TERM variable to use for units connected to a
3227 terminal, when no other value is explicitly is set is now
3228 vt220 rather than vt102. This should be fairly safe still,
3229 but allows PgUp/PgDn work.
3231 * The /etc/crypttab option header= as known from Debian is now
3234 * "loginctl user-status" and "loginctl session-status" will
3235 now show the last 10 lines of log messages of the
3236 user/session following the status output. Similar,
3237 "machinectl status" will show the last 10 log lines
3238 associated with a virtual machine or container
3239 service. (Note that this is usually not the log messages
3240 done in the VM/container itself, but simply what the
3241 container manager logs. For nspawn this includes all console
3244 * "loginctl session-status" without further argument will now
3245 show the status of the session of the caller. Similar,
3246 "lock-session", "unlock-session", "activate",
3247 "enable-linger", "disable-linger" may now be called without
3248 session/user parameter in which case they apply to the
3249 caller's session/user.
3251 * An X11 session scriptlet is now shipped that uploads
3252 $DISPLAY and $XAUTHORITY into the environment of the systemd
3253 --user daemon if a session begins. This should improve
3254 compatibility with X11 enabled applications run as systemd
3257 * Generators are now subject to masking via /etc and /run, the
3258 same way as unit files.
3260 * networkd .network files gained support for configuring
3261 per-link IPv4/IPv6 packet forwarding as well as IPv4
3262 masquerading. This is by default turned on for veth links to
3263 containers, as registered by systemd-nspawn. This means that
3264 nspawn containers run with --network-veth will now get
3265 automatic routed access to the host's networks without any
3266 further configuration or setup, as long as networkd runs on
3269 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3270 or UDP posts of a container on the host. With this in place
3271 it is possible to run containers with private veth links
3272 (--network-veth), and have their functionality exposed on
3273 the host as if their services were running directly on the
3276 * systemd-nspawn's --network-veth switch now gained a short
3277 version "-n", since with the changes above it is now truly
3278 useful out-of-the-box. The systemd-nspawn@.service has been
3279 updated to make use of it too by default.
3281 * systemd-nspawn will now maintain a per-image R/W lock, to
3282 ensure that the same image is not started more than once
3283 writable. (It's OK to run an image multiple times
3284 simultaneously in read-only mode.)
3286 * systemd-nspawn's --image= option is now capable of
3287 dissecting and booting MBR and GPT disk images that contain
3288 only a single active Linux partition. Previously it
3289 supported only GPT disk images with proper GPT type
3290 IDs. This allows running cloud images from major
3291 distributions directly with systemd-nspawn, without
3294 * In addition to collecting mouse dpi data in the udev
3295 hardware database, there's now support for collecting angle
3296 information for mouse scroll wheels. The database is
3297 supposed to guarantee similar scrolling behavior on mice
3298 that it knows about. There's also support for collecting
3299 information about Touchpad types.
3301 * udev's input_id built-in will now also collect touch screen
3302 dimension data and attach it to probed devices.
3304 * /etc/os-release gained support for a Distribution Privacy
3307 * networkd gained support for creating "ipvlan", "gretap",
3308 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3310 * systemd-tmpfiles gained support for "a" lines for setting
3313 * systemd-nspawn will now mount /tmp in the container to
3314 tmpfs, automatically.
3316 * systemd now exposes the memory.usage_in_bytes cgroup
3317 attribute and shows it for each service in the "systemctl
3318 status" output, if available.
3320 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3321 immediate reboot is triggered. This useful if shutdown is
3322 hung and is unable to complete, to expedite the
3323 operation. Note that this kind of reboot will still unmount
3324 all file systems, and hence should not result in fsck being
3327 * A .device unit for an optical block device will now be
3328 considered active only when a medium is in the drive. Also,
3329 mount units are now bound to their backing devices thus
3330 triggering automatic unmounting when devices become
3331 unavailable. With this in place systemd will now
3332 automatically unmount left-over mounts when a CD-ROM is
3333 ejected or an USB stick is yanked from the system.
3335 * networkd-wait-online now has support for waiting for
3336 specific interfaces only (with globbing), and for giving up
3337 after a configurable timeout.
3339 * networkd now exits when idle. It will be automatically
3340 restarted as soon as interfaces show up, are removed or
3341 change state. networkd will stay around as long as there is
3342 at least one DHCP state machine or similar around, that keep
3345 * networkd may now configure IPv6 link-local addressing in
3346 addition to IPv4 link-local addressing.
3348 * The IPv6 "token" for use in SLAAC may now be configured for
3349 each .network interface in networkd.
3351 * Routes configured with networkd may now be assigned a scope
3354 * networkd's [Match] sections now support globbing and lists
3355 of multiple space-separated matches per item.
3357 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3358 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3359 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3360 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3361 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3362 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3363 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3364 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3365 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3366 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3367 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3368 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3369 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3370 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3371 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3372 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3373 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3374 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3375 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3376 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3377 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3378 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3379 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3380 Hoffmann, Zbigniew Jędrzejewski-Szmek
3382 — Berlin, 2015-02-16
3386 * When querying unit file enablement status (for example via
3387 "systemctl is-enabled"), a new state "indirect" is now known
3388 which indicates that a unit might not be enabled itself, but
3389 another unit listed in its Also= setting might be.
3391 * Similar to the various existing ConditionXYZ= settings for
3392 units, there are now matching AssertXYZ= settings. While
3393 failing conditions cause a unit to be skipped, but its job
3394 to succeed, failing assertions declared like this will cause
3395 a unit start operation and its job to fail.
3397 * hostnamed now knows a new chassis type "embedded".
3399 * systemctl gained a new "edit" command. When used on a unit
3400 file, this allows extending unit files with .d/ drop-in
3401 configuration snippets or editing the full file (after
3402 copying it from /usr/lib to /etc). This will invoke the
3403 user's editor (as configured with $EDITOR), and reload the
3404 modified configuration after editing.
3406 * "systemctl status" now shows the suggested enablement state
3407 for a unit, as declared in the (usually vendor-supplied)
3408 system preset files.
3410 * nss-myhostname will now resolve the single-label host name
3411 "gateway" to the locally configured default IP routing
3412 gateways, ordered by their metrics. This assigns a stable
3413 name to the used gateways, regardless which ones are
3414 currently configured. Note that the name will only be
3415 resolved after all other name sources (if nss-myhostname is
3416 configured properly) and should hence not negatively impact
3417 systems that use the single-label host name "gateway" in
3420 * systemd-inhibit now allows filtering by mode when listing
3423 * Scope and service units gained a new "Delegate" boolean
3424 property, which, when set, allows processes running inside the
3425 unit to further partition resources. This is primarily
3426 useful for systemd user instances as well as container
3429 * journald will now pick up audit messages directly from
3430 the kernel, and log them like any other log message. The
3431 audit fields are split up and fully indexed. This means that
3432 journalctl in many ways is now a (nicer!) alternative to
3433 ausearch, the traditional audit client. Note that this
3434 implements only a minimal audit client. If you want the
3435 special audit modes like reboot-on-log-overflow, please use
3436 the traditional auditd instead, which can be used in
3437 parallel to journald.
3439 * The ConditionSecurity= unit file option now understands the
3440 special string "audit" to check whether auditing is
3443 * journalctl gained two new commands --vacuum-size= and
3444 --vacuum-time= to delete old journal files until the
3445 remaining ones take up no more than the specified size on disk,
3446 or are not older than the specified time.
3448 * A new, native PPPoE library has been added to sd-network,
3449 systemd's library of light-weight networking protocols. This
3450 library will be used in a future version of networkd to
3451 enable PPPoE communication without an external pppd daemon.
3453 * The busctl tool now understands a new "capture" verb that
3454 works similar to "monitor", but writes a packet capture
3455 trace to STDOUT that can be redirected to a file which is
3456 compatible with libcap's capture file format. This can then
3457 be loaded in Wireshark and similar tools to inspect bus
3460 * The busctl tool now understands a new "tree" verb that shows
3461 the object trees of a specific service on the bus, or of all
3464 * The busctl tool now understands a new "introspect" verb that
3465 shows all interfaces and members of objects on the bus,
3466 including their signature and values. This is particularly
3467 useful to get more information about bus objects shown by
3468 the new "busctl tree" command.
3470 * The busctl tool now understands new verbs "call",
3471 "set-property" and "get-property" for invoking bus method
3472 calls, setting and getting bus object properties in a
3475 * busctl gained a new --augment-creds= argument that controls
3476 whether the tool shall augment credential information it
3477 gets from the bus with data from /proc, in a possibly
3480 * nspawn's --link-journal= switch gained two new values
3481 "try-guest" and "try-host" that work like "guest" and
3482 "host", but do not fail if the host has no persistent
3483 journalling enabled. -j is now equivalent to
3484 --link-journal=try-guest.
3486 * macvlan network devices created by nspawn will now have
3487 stable MAC addresses.
3489 * A new SmackProcessLabel= unit setting has been added, which
3490 controls the SMACK security label processes forked off by
3491 the respective unit shall use.
3493 * If compiled with --enable-xkbcommon, systemd-localed will
3494 verify x11 keymap settings by compiling the given keymap. It
3495 will spew out warnings if the compilation fails. This
3496 requires libxkbcommon to be installed.
3498 * When a coredump is collected, a larger number of metadata
3499 fields is now collected and included in the journal records
3500 created for it. More specifically, control group membership,
3501 environment variables, memory maps, working directory,
3502 chroot directory, /proc/$PID/status, and a list of open file
3503 descriptors is now stored in the log entry.
3505 * The udev hwdb now contains DPI information for mice. For
3508 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3510 * All systemd programs that read standalone configuration
3511 files in /etc now also support a corresponding series of
3512 .conf.d configuration directories in /etc/, /run/,
3513 /usr/local/lib/, /usr/lib/, and (if configured with
3514 --enable-split-usr) /lib/. In particular, the following
3515 configuration files now have corresponding configuration
3516 directories: system.conf user.conf, logind.conf,
3517 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3518 resolved.conf, timesyncd.conf, journal-remote.conf, and
3519 journal-upload.conf. Note that distributions should use the
3520 configuration directories in /usr/lib/; the directories in
3521 /etc/ are reserved for the system administrator.
3523 * systemd-rfkill will no longer take the rfkill device name
3524 into account when storing rfkill state on disk, as the name
3525 might be dynamically assigned and not stable. Instead, the
3526 ID_PATH udev variable combined with the rfkill type (wlan,
3527 bluetooth, ...) is used.
3529 * A new service systemd-machine-id-commit.service has been
3530 added. When used on systems where /etc is read-only during
3531 boot, and /etc/machine-id is not initialized (but an empty
3532 file), this service will copy the temporary machine ID
3533 created as replacement into /etc after the system is fully
3534 booted up. This is useful for systems that are freshly
3535 installed with a non-initialized machine ID, but should get
3536 a fixed machine ID for subsequent boots.
3538 * networkd's .netdev files now provide a large set of
3539 configuration parameters for VXLAN devices. Similarly, the
3540 bridge port cost parameter is now configurable in .network
3541 files. There's also new support for configuring IP source
3542 routing. networkd .link files gained support for a new
3543 OriginalName= match that is useful to match against the
3544 original interface name the kernel assigned. .network files
3545 may include MTU= and MACAddress= fields for altering the MTU
3546 and MAC address while being connected to a specific network
3549 * The LUKS logic gained supported for configuring
3550 UUID-specific key files. There's also new support for naming
3551 LUKS device from the kernel command line, using the new
3552 luks.name= argument.
3554 * Timer units may now be transiently created via the bus API
3555 (this was previously already available for scope and service
3556 units). In addition it is now possible to create multiple
3557 transient units at the same time with a single bus call. The
3558 "systemd-run" tool has been updated to make use of this for
3559 running commands on a specified time, in at(1)-style.
3561 * tmpfiles gained support for "t" lines, for assigning
3562 extended attributes to files. Among other uses this may be
3563 used to assign SMACK labels to files.
3565 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3566 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3567 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3568 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3569 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3570 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3571 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3572 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3573 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3574 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3575 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3576 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3577 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3578 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3579 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3580 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3581 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3582 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3584 — Berlin, 2014-12-10
3588 * journalctl gained the new options -t/--identifier= to match
3589 on the syslog identifier (aka "tag"), as well as --utc to
3590 show log timestamps in the UTC timezone. journalctl now also
3591 accepts -n/--lines=all to disable line capping in a pager.
3593 * journalctl gained a new switch, --flush, that synchronously
3594 flushes logs from /run/log/journal to /var/log/journal if
3595 persistent storage is enabled. systemd-journal-flush.service
3596 now waits until the operation is complete.
3598 * Services can notify the manager before they start a reload
3599 (by sending RELOADING=1) or shutdown (by sending
3600 STOPPING=1). This allows the manager to track and show the
3601 internal state of daemons and closes a race condition when
3602 the process is still running but has closed its D-Bus
3605 * Services with Type=oneshot do not have to have any ExecStart
3608 * User units are now loaded also from
3609 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3610 /run/systemd/user directory that was already previously
3611 supported, but is under the control of the user.
3613 * Job timeouts (i.e. time-outs on the time a job that is
3614 queued stays in the run queue) can now optionally result in
3615 immediate reboot or power-off actions (JobTimeoutAction= and
3616 JobTimeoutRebootArgument=). This is useful on ".target"
3617 units, to limit the maximum time a target remains
3618 undispatched in the run queue, and to trigger an emergency
3619 operation in such a case. This is now used by default to
3620 turn off the system if boot-up (as defined by everything in
3621 basic.target) hangs and does not complete for at least
3622 15min. Also, if power-off or reboot hang for at least 30min
3623 an immediate power-off/reboot operation is triggered. This
3624 functionality is particularly useful to increase reliability
3625 on embedded devices, but also on laptops which might
3626 accidentally get powered on when carried in a backpack and
3627 whose boot stays stuck in a hard disk encryption passphrase
3630 * systemd-logind can be configured to also handle lid switch
3631 events even when the machine is docked or multiple displays
3632 are attached (HandleLidSwitchDocked= option).
3634 * A helper binary and a service have been added which can be
3635 used to resume from hibernation in the initramfs. A
3636 generator will parse the resume= option on the kernel
3637 command line to trigger resume.
3639 * A user console daemon systemd-consoled has been
3640 added. Currently, it is a preview, and will so far open a
3641 single terminal on each session of the user marked as
3642 Desktop=systemd-console.
3644 * Route metrics can be specified for DHCP routes added by
3647 * The SELinux context of socket-activated services can be set
3648 from the information provided by the networking stack
3649 (SELinuxContextFromNet= option).
3651 * Userspace firmware loading support has been removed and
3652 the minimum supported kernel version is thus bumped to 3.7.
3654 * Timeout for udev workers has been increased from 1 to 3
3655 minutes, but a warning will be printed after 1 minute to
3656 help diagnose kernel modules that take a long time to load.
3658 * Udev rules can now remove tags on devices with TAG-="foobar".
3660 * systemd's readahead implementation has been removed. In many
3661 circumstances it didn't give expected benefits even for
3662 rotational disk drives and was becoming less relevant in the
3663 age of SSDs. As none of the developers has been using
3664 rotating media anymore, and nobody stepped up to actively
3665 maintain this component of systemd it has now been removed.
3667 * Swap units can use Options= to specify discard options.
3668 Discard options specified for swaps in /etc/fstab are now
3671 * Docker containers are now detected as a separate type of
3674 * The Password Agent protocol gained support for queries where
3675 the user input is shown, useful e.g. for user names.
3676 systemd-ask-password gained a new --echo option to turn that
3679 * The default sysctl.d/ snippets will now set:
3681 net.core.default_qdisc = fq_codel
3683 This selects Fair Queuing Controlled Delay as the default
3684 queuing discipline for network interfaces. fq_codel helps
3685 fight the network bufferbloat problem. It is believed to be
3686 a good default with no tuning required for most workloads.
3687 Downstream distributions may override this choice. On 10Gbit
3688 servers that do not do forwarding, "fq" may perform better.
3689 Systems without a good clocksource should use "pfifo_fast".
3691 * If kdbus is enabled during build a new option BusPolicy= is
3692 available for service units, that allows locking all service
3693 processes into a stricter bus policy, in order to limit
3694 access to various bus services, or even hide most of them
3695 from the service's view entirely.
3697 * networkctl will now show the .network and .link file
3698 networkd has applied to a specific interface.
3700 * sd-login gained a new API call sd_session_get_desktop() to
3701 query which desktop environment has been selected for a
3704 * UNIX utmp support is now compile-time optional to support
3705 legacy-free systems.
3707 * systemctl gained two new commands "add-wants" and
3708 "add-requires" for pulling in units from specific targets
3711 * If the word "rescue" is specified on the kernel command line
3712 the system will now boot into rescue mode (aka
3713 rescue.target), which was previously available only by
3714 specifying "1" or "systemd.unit=rescue.target" on the kernel
3715 command line. This new kernel command line option nicely
3716 mirrors the already existing "emergency" kernel command line
3719 * New kernel command line options mount.usr=, mount.usrflags=,
3720 mount.usrfstype= have been added that match root=, rootflags=,
3721 rootfstype= but allow mounting a specific file system to
3724 * The $NOTIFY_SOCKET is now also passed to control processes of
3725 services, not only the main process.
3727 * This version reenables support for fsck's -l switch. This
3728 means at least version v2.25 of util-linux is required for
3729 operation, otherwise dead-locks on device nodes may
3730 occur. Again: you need to update util-linux to at least
3731 v2.25 when updating systemd to v217.
3733 * The "multi-seat-x" tool has been removed from systemd, as
3734 its functionality has been integrated into X servers 1.16,
3735 and the tool is hence redundant. It is recommended to update
3736 display managers invoking this tool to simply invoke X
3737 directly from now on, again.
3739 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3740 message flag has been added for all of systemd's PolicyKit
3741 authenticated method calls has been added. In particular
3742 this now allows optional interactive authorization via
3743 PolicyKit for many of PID1's privileged operations such as
3744 unit file enabling and disabling.
3746 * "udevadm hwdb --update" learnt a new switch "--usr" for
3747 placing the rebuilt hardware database in /usr instead of
3748 /etc. When used only hardware database entries stored in
3749 /usr will be used, and any user database entries in /etc are
3750 ignored. This functionality is useful for vendors to ship a
3751 pre-built database on systems where local configuration is
3752 unnecessary or unlikely.
3754 * Calendar time specifications in .timer units now also
3755 understand the strings "semi-annually", "quarterly" and
3756 "minutely" as shortcuts (in addition to the preexisting
3757 "anually", "hourly", ...).
3759 * systemd-tmpfiles will now correctly create files in /dev
3760 at boot which are marked for creation only at boot. It is
3761 recommended to always create static device nodes with 'c!'
3762 and 'b!', so that they are created only at boot and not
3763 overwritten at runtime.
3765 * When the watchdog logic is used for a service (WatchdogSec=)
3766 and the watchdog timeout is hit the service will now be
3767 terminated with SIGABRT (instead of just SIGTERM), in order
3768 to make sure a proper coredump and backtrace is
3769 generated. This ensures that hanging services will result in
3770 similar coredump/backtrace behaviour as services that hit a
3773 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3774 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3775 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3776 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3777 Herrmann, David Sommerseth, David Strauss, Emil Renner
3778 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3779 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3780 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3781 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3782 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3783 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3784 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3785 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3786 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3787 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3788 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3789 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3790 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3791 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3792 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3793 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3796 — Berlin, 2014-10-28
3800 * timedated no longer reads NTP implementation unit names from
3801 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3802 implementations should add a
3804 Conflicts=systemd-timesyncd.service
3806 to their unit files to take over and replace systemd's NTP
3807 default functionality.
3809 * systemd-sysusers gained a new line type "r" for configuring
3810 which UID/GID ranges to allocate system users/groups
3811 from. Lines of type "u" may now add an additional column
3812 that specifies the home directory for the system user to be
3813 created. Also, systemd-sysusers may now optionally read user
3814 information from STDIN instead of a file. This is useful for
3815 invoking it from RPM preinst scriptlets that need to create
3816 users before the first RPM file is installed since these
3817 files might need to be owned by them. A new
3818 %sysusers_create_inline RPM macro has been introduced to do
3819 just that. systemd-sysusers now updates the shadow files as
3820 well as the user/group databases, which should enhance
3821 compatibility with certain tools like grpck.
3823 * A number of bus APIs of PID 1 now optionally consult
3824 PolicyKit to permit access for otherwise unprivileged
3825 clients under certain conditions. Note that this currently
3826 doesn't support interactive authentication yet, but this is
3827 expected to be added eventually, too.
3829 * /etc/machine-info now has new fields for configuring the
3830 deployment environment of the machine, as well as the
3831 location of the machine. hostnamectl has been updated with
3832 new command to update these fields.
3834 * systemd-timesyncd has been updated to automatically acquire
3835 NTP server information from systemd-networkd, which might
3836 have been discovered via DHCP.
3838 * systemd-resolved now includes a caching DNS stub resolver
3839 and a complete LLMNR name resolution implementation. A new
3840 NSS module "nss-resolve" has been added which can be used
3841 instead of glibc's own "nss-dns" to resolve hostnames via
3842 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3843 be resolved via systemd-resolved D-Bus APIs. In contrast to
3844 the glibc internal resolver systemd-resolved is aware of
3845 multi-homed system, and keeps DNS server and caches separate
3846 and per-interface. Queries are sent simultaneously on all
3847 interfaces that have DNS servers configured, in order to
3848 properly handle VPNs and local LANs which might resolve
3849 separate sets of domain names. systemd-resolved may acquire
3850 DNS server information from systemd-networkd automatically,
3851 which in turn might have discovered them via DHCP. A tool
3852 "systemd-resolve-host" has been added that may be used to
3853 query the DNS logic in resolved. systemd-resolved implements
3854 IDNA and automatically uses IDNA or UTF-8 encoding depending
3855 on whether classic DNS or LLMNR is used as transport. In the
3856 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3857 implementation to systemd-resolved.
3859 * A new NSS module nss-mymachines has been added, that
3860 automatically resolves the names of all local registered
3861 containers to their respective IP addresses.
3863 * A new client tool "networkctl" for systemd-networkd has been
3864 added. It currently is entirely passive and will query
3865 networking configuration from udev, rtnetlink and networkd,
3866 and present it to the user in a very friendly
3867 way. Eventually, we hope to extend it to become a full
3868 control utility for networkd.
3870 * .socket units gained a new DeferAcceptSec= setting that
3871 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3872 TCP. Similarly, support for controlling TCP keep-alive
3873 settings has been added (KeepAliveTimeSec=,
3874 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3875 turning off Nagle's algorithm on TCP has been added
3878 * logind learned a new session type "web", for use in projects
3879 like Cockpit which register web clients as PAM sessions.
3881 * timer units with at least one OnCalendar= setting will now
3882 be started only after time-sync.target has been
3883 reached. This way they will not elapse before the system
3884 clock has been corrected by a local NTP client or
3885 similar. This is particular useful on RTC-less embedded
3886 machines, that come up with an invalid system clock.
3888 * systemd-nspawn's --network-veth= switch should now result in
3889 stable MAC addresses for both the outer and the inner side
3892 * systemd-nspawn gained a new --volatile= switch for running
3893 container instances with /etc or /var unpopulated.
3895 * The kdbus client code has been updated to use the new Linux
3896 3.17 memfd subsystem instead of the old kdbus-specific one.
3898 * systemd-networkd's DHCP client and server now support
3899 FORCERENEW. There are also new configuration options to
3900 configure the vendor client identifier and broadcast mode
3903 * systemd will no longer inform the kernel about the current
3904 timezone, as this is necessarily incorrect and racy as the
3905 kernel has no understanding of DST and similar
3906 concepts. This hence means FAT timestamps will be always
3907 considered UTC, similar to what Android is already
3908 doing. Also, when the RTC is configured to the local time
3909 (rather than UTC) systemd will never synchronize back to it,
3910 as this might confuse Windows at a later boot.
3912 * systemd-analyze gained a new command "verify" for offline
3913 validation of unit files.
3915 * systemd-networkd gained support for a couple of additional
3916 settings for bonding networking setups. Also, the metric for
3917 statically configured routes may now be configured. For
3918 network interfaces where this is appropriate the peer IP
3919 address may now be configured.
3921 * systemd-networkd's DHCP client will no longer request
3922 broadcasting by default, as this tripped up some networks.
3923 For hardware where broadcast is required the feature should
3924 be switched back on using RequestBroadcast=yes.
3926 * systemd-networkd will now set up IPv4LL addresses (when
3927 enabled) even if DHCP is configured successfully.
3929 * udev will now default to respect network device names given
3930 by the kernel when the kernel indicates that these are
3931 predictable. This behavior can be tweaked by changing
3932 NamePolicy= in the relevant .link file.
3934 * A new library systemd-terminal has been added that
3935 implements full TTY stream parsing and rendering. This
3936 library is supposed to be used later on for implementing a
3937 full userspace VT subsystem, replacing the current kernel
3940 * A new tool systemd-journal-upload has been added to push
3941 journal data to a remote system running
3942 systemd-journal-remote.
3944 * journald will no longer forward all local data to another
3945 running syslog daemon. This change has been made because
3946 rsyslog (which appears to be the most commonly used syslog
3947 implementation these days) no longer makes use of this, and
3948 instead pulls the data out of the journal on its own. Since
3949 forwarding the messages to a non-existent syslog server is
3950 more expensive than we assumed we have now turned this
3951 off. If you run a syslog server that is not a recent rsyslog
3952 version, you have to turn this option on again
3953 (ForwardToSyslog= in journald.conf).
3955 * journald now optionally supports the LZ4 compressor for
3956 larger journal fields. This compressor should perform much
3957 better than XZ which was the previous default.
3959 * machinectl now shows the IP addresses of local containers,
3960 if it knows them, plus the interface name of the container.
3962 * A new tool "systemd-escape" has been added that makes it
3963 easy to escape strings to build unit names and similar.
3965 * sd_notify() messages may now include a new ERRNO= field
3966 which is parsed and collected by systemd and shown among the
3967 "systemctl status" output for a service.
3969 * A new component "systemd-firstboot" has been added that
3970 queries the most basic systemd information (timezone,
3971 hostname, root password) interactively on first
3972 boot. Alternatively it may also be used to provision these
3973 things offline on OS images installed into directories.
3975 * The default sysctl.d/ snippets will now set
3977 net.ipv4.conf.default.promote_secondaries=1
3979 This has the benefit of no flushing secondary IP addresses
3980 when primary addresses are removed.
3982 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3983 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3984 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3985 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3986 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3987 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3988 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3989 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3990 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3991 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3992 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3993 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3994 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3995 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3996 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3998 — Berlin, 2014-08-19
4002 * A new tool systemd-sysusers has been added. This tool
4003 creates system users and groups in /etc/passwd and
4004 /etc/group, based on static declarative system user/group
4005 definitions in /usr/lib/sysusers.d/. This is useful to
4006 enable factory resets and volatile systems that boot up with
4007 an empty /etc directory, and thus need system users and
4008 groups created during early boot. systemd now also ships
4009 with two default sysusers.d/ files for the most basic
4010 users and groups systemd and the core operating system
4013 * A new tmpfiles snippet has been added that rebuilds the
4014 essential files in /etc on boot, should they be missing.
4016 * A directive for ensuring automatic clean-up of
4017 /var/cache/man/ has been removed from the default
4018 configuration. This line should now be shipped by the man
4019 implementation. The necessary change has been made to the
4020 man-db implementation. Note that you need to update your man
4021 implementation to one that ships this line, otherwise no
4022 automatic clean-up of /var/cache/man will take place.
4024 * A new condition ConditionNeedsUpdate= has been added that
4025 may conditionalize services to only run when /etc or /var
4026 are "older" than the vendor operating system resources in
4027 /usr. This is useful for reconstructing or updating /etc
4028 after an offline update of /usr or a factory reset, on the
4029 next reboot. Services that want to run once after such an
4030 update or reset should use this condition and order
4031 themselves before the new systemd-update-done.service, which
4032 will mark the two directories as fully updated. A number of
4033 service files have been added making use of this, to rebuild
4034 the udev hardware database, the journald message catalog and
4035 dynamic loader cache (ldconfig). The systemd-sysusers tool
4036 described above also makes use of this now. With this in
4037 place it is now possible to start up a minimal operating
4038 system with /etc empty cleanly. For more information on the
4039 concepts involved see this recent blog story:
4041 http://0pointer.de/blog/projects/stateless.html
4043 * A new system group "input" has been introduced, and all
4044 input device nodes get this group assigned. This is useful
4045 for system-level software to get access to input devices. It
4046 complements what is already done for "audio" and "video".
4048 * systemd-networkd learnt minimal DHCPv4 server support in
4049 addition to the existing DHCPv4 client support. It also
4050 learnt DHCPv6 client and IPv6 Router Solicitation client
4051 support. The DHCPv4 client gained support for static routes
4052 passed in from the server. Note that the [DHCPv4] section
4053 known in older systemd-networkd versions has been renamed to
4054 [DHCP] and is now also used by the DHCPv6 client. Existing
4055 .network files using settings of this section should be
4056 updated, though compatibility is maintained. Optionally, the
4057 client hostname may now be sent to the DHCP server.
4059 * networkd gained support for vxlan virtual networks as well
4060 as tun/tap and dummy devices.
4062 * networkd gained support for automatic allocation of address
4063 ranges for interfaces from a system-wide pool of
4064 addresses. This is useful for dynamically managing a large
4065 number of interfaces with a single network configuration
4066 file. In particular this is useful to easily assign
4067 appropriate IP addresses to the veth links of a large number
4068 of nspawn instances.
4070 * RPM macros for processing sysusers, sysctl and binfmt
4071 drop-in snippets at package installation time have been
4074 * The /etc/os-release file should now be placed in
4075 /usr/lib/os-release. The old location is automatically
4076 created as symlink. /usr/lib is the more appropriate
4077 location of this file, since it shall actually describe the
4078 vendor operating system shipped in /usr, and not the
4079 configuration stored in /etc.
4081 * .mount units gained a new boolean SloppyOptions= setting
4082 that maps to mount(8)'s -s option which enables permissive
4083 parsing of unknown mount options.
4085 * tmpfiles learnt a new "L+" directive which creates a symlink
4086 but (unlike "L") deletes a pre-existing file first, should
4087 it already exist and not already be the correct
4088 symlink. Similarly, "b+", "c+" and "p+" directives have been
4089 added as well, which create block and character devices, as
4090 well as fifos in the filesystem, possibly removing any
4091 pre-existing files of different types.
4093 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4094 'argument' field (which so far specified the source to
4095 symlink/copy the files from) is now optional. If omitted the
4096 same file os copied from /usr/share/factory/ suffixed by the
4097 full destination path. This is useful for populating /etc
4098 with essential files, by copying them from vendor defaults
4099 shipped in /usr/share/factory/etc.
4101 * A new command "systemctl preset-all" has been added that
4102 applies the service preset settings to all installed unit
4103 files. A new switch --preset-mode= has been added that
4104 controls whether only enable or only disable operations
4107 * A new command "systemctl is-system-running" has been added
4108 that allows checking the overall state of the system, for
4109 example whether it is fully up and running.
4111 * When the system boots up with an empty /etc, the equivalent
4112 to "systemctl preset-all" is executed during early boot, to
4113 make sure all default services are enabled after a factory
4116 * systemd now contains a minimal preset file that enables the
4117 most basic services systemd ships by default.
4119 * Unit files' [Install] section gained a new DefaultInstance=
4120 field for defining the default instance to create if a
4121 template unit is enabled with no instance specified.
4123 * A new passive target cryptsetup-pre.target has been added
4124 that may be used by services that need to make they run and
4125 finish before the first LUKS cryptographic device is set up.
4127 * The /dev/loop-control and /dev/btrfs-control device nodes
4128 are now owned by the "disk" group by default, opening up
4129 access to this group.
4131 * systemd-coredump will now automatically generate a
4132 stack trace of all core dumps taking place on the system,
4133 based on elfutils' libdw library. This stack trace is logged
4136 * systemd-coredump may now optionally store coredumps directly
4137 on disk (in /var/lib/systemd/coredump, possibly compressed),
4138 instead of storing them unconditionally in the journal. This
4139 mode is the new default. A new configuration file
4140 /etc/systemd/coredump.conf has been added to configure this
4141 and other parameters of systemd-coredump.
4143 * coredumpctl gained a new "info" verb to show details about a
4144 specific coredump. A new switch "-1" has also been added
4145 that makes sure to only show information about the most
4146 recent entry instead of all entries. Also, as the tool is
4147 generally useful now the "systemd-" prefix of the binary
4148 name has been removed. Distributions that want to maintain
4149 compatibility with the old name should add a symlink from
4150 the old name to the new name.
4152 * journald's SplitMode= now defaults to "uid". This makes sure
4153 that unprivileged users can access their own coredumps with
4154 coredumpctl without restrictions.
4156 * New kernel command line options "systemd.wants=" (for
4157 pulling an additional unit during boot), "systemd.mask="
4158 (for masking a specific unit for the boot), and
4159 "systemd.debug-shell" (for enabling the debug shell on tty9)
4160 have been added. This is implemented in the new generator
4161 "systemd-debug-generator".
4163 * systemd-nspawn will now by default filter a couple of
4164 syscalls for containers, among them those required for
4165 kernel module loading, direct x86 IO port access, swap
4166 management, and kexec. Most importantly though
4167 open_by_handle_at() is now prohibited for containers,
4168 closing a hole similar to a recently discussed vulnerability
4169 in docker regarding access to files on file hierarchies the
4170 container should normally not have access to. Note that, for
4171 nspawn, we generally make no security claims anyway (and
4172 this is explicitly documented in the man page), so this is
4173 just a fix for one of the most obvious problems.
4175 * A new man page file-hierarchy(7) has been added that
4176 contains a minimized, modernized version of the file system
4177 layout systemd expects, similar in style to the FHS
4178 specification or hier(5). A new tool systemd-path(1) has
4179 been added to query many of these paths for the local
4182 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4183 longer done. Since the directory now has a per-user size
4184 limit, and is cleaned on logout this appears unnecessary,
4185 in particular since this now brings the lifecycle of this
4186 directory closer in line with how IPC objects are handled.
4188 * systemd.pc now exports a number of additional directories,
4189 including $libdir (which is useful to identify the library
4190 path for the primary architecture of the system), and a
4191 couple of drop-in directories.
4193 * udev's predictable network interface names now use the dev_port
4194 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4195 distinguish between ports of the same PCI function. dev_id should
4196 only be used for ports using the same HW address, hence the need
4199 * machined has been updated to export the OS version of a
4200 container (read from /etc/os-release and
4201 /usr/lib/os-release) on the bus. This is now shown in
4202 "machinectl status" for a machine.
4204 * A new service setting RestartForceExitStatus= has been
4205 added. If configured to a set of exit signals or process
4206 return values, the service will be restarted when the main
4207 daemon process exits with any of them, regardless of the
4210 * systemctl's -H switch for connecting to remote systemd
4211 machines has been extended so that it may be used to
4212 directly connect to a specific container on the
4213 host. "systemctl -H root@foobar:waldi" will now connect as
4214 user "root" to host "foobar", and then proceed directly to
4215 the container named "waldi". Note that currently you have to
4216 authenticate as user "root" for this to work, as entering
4217 containers is a privileged operation.
4219 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4220 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4221 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4222 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4223 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4224 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4225 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4226 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4227 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4228 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4229 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4230 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4232 — Berlin, 2014-07-03
4236 * As an experimental feature, udev now tries to lock the
4237 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4238 executes events for the disk or any of its partitions.
4239 Applications like partitioning programs can lock the
4240 disk device node (flock(LOCK_EX)) and claim temporary
4241 device ownership that way; udev will entirely skip all event
4242 handling for this disk and its partitions. If the disk
4243 was opened for writing, the close will trigger a partition
4244 table rescan in udev's "watch" facility, and if needed
4245 synthesize "change" events for the disk and all its partitions.
4246 This is now unconditionally enabled, and if it turns out to
4247 cause major problems, we might turn it on only for specific
4248 devices, or might need to disable it entirely. Device Mapper
4249 devices are excluded from this logic.
4251 * We temporarily dropped the "-l" switch for fsck invocations,
4252 since they collide with the flock() logic above. util-linux
4253 upstream has been changed already to avoid this conflict,
4254 and we will readd "-l" as soon as util-linux with this
4255 change has been released.
4257 * The dependency on libattr has been removed. Since a long
4258 time, the extended attribute calls have moved to glibc, and
4259 libattr is thus unnecessary.
4261 * Virtualization detection works without privileges now. This
4262 means the systemd-detect-virt binary no longer requires
4263 CAP_SYS_PTRACE file capabilities, and our daemons can run
4264 with fewer privileges.
4266 * systemd-networkd now runs under its own "systemd-network"
4267 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4268 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4269 loses the ability to write to files owned by root this way.
4271 * Similarly, systemd-resolved now runs under its own
4272 "systemd-resolve" user with no capabilities remaining.
4274 * Similarly, systemd-bus-proxyd now runs under its own
4275 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4277 * systemd-networkd gained support for setting up "veth"
4278 virtual Ethernet devices for container connectivity, as well
4279 as GRE and VTI tunnels.
4281 * systemd-networkd will no longer automatically attempt to
4282 manually load kernel modules necessary for certain tunnel
4283 transports. Instead, it is assumed the kernel loads them
4284 automatically when required. This only works correctly on
4285 very new kernels. On older kernels, please consider adding
4286 the kernel modules to /etc/modules-load.d/ as a work-around.
4288 * The resolv.conf file systemd-resolved generates has been
4289 moved to /run/systemd/resolve/. If you have a symlink from
4290 /etc/resolv.conf, it might be necessary to correct it.
4292 * Two new service settings, ProtectHome= and ProtectSystem=,
4293 have been added. When enabled, they will make the user data
4294 (such as /home) inaccessible or read-only and the system
4295 (such as /usr) read-only, for specific services. This allows
4296 very light-weight per-service sandboxing to avoid
4297 modifications of user data or system files from
4298 services. These two new switches have been enabled for all
4299 of systemd's long-running services, where appropriate.
4301 * Socket units gained new SocketUser= and SocketGroup=
4302 settings to set the owner user and group of AF_UNIX sockets
4303 and FIFOs in the file system.
4305 * Socket units gained a new RemoveOnStop= setting. If enabled,
4306 all FIFOS and sockets in the file system will be removed
4307 when the specific socket unit is stopped.
4309 * Socket units gained a new Symlinks= setting. It takes a list
4310 of symlinks to create to file system sockets or FIFOs
4311 created by the specific Unix sockets. This is useful to
4312 manage symlinks to socket nodes with the same life-cycle as
4315 * The /dev/log socket and /dev/initctl FIFO have been moved to
4316 /run, and have been replaced by symlinks. This allows
4317 connecting to these facilities even if PrivateDevices=yes is
4318 used for a service (which makes /dev/log itself unavailable,
4319 but /run is left). This also has the benefit of ensuring
4320 that /dev only contains device nodes, directories and
4321 symlinks, and nothing else.
4323 * sd-daemon gained two new calls sd_pid_notify() and
4324 sd_pid_notifyf(). They are similar to sd_notify() and
4325 sd_notifyf(), but allow overriding of the source PID of
4326 notification messages if permissions permit this. This is
4327 useful to send notify messages on behalf of a different
4328 process (for example, the parent process). The
4329 systemd-notify tool has been updated to make use of this
4330 when sending messages (so that notification messages now
4331 originate from the shell script invoking systemd-notify and
4332 not the systemd-notify process itself. This should minimize
4333 a race where systemd fails to associate notification
4334 messages to services when the originating process already
4337 * A new "on-abnormal" setting for Restart= has been added. If
4338 set, it will result in automatic restarts on all "abnormal"
4339 reasons for a process to exit, which includes unclean
4340 signals, core dumps, timeouts and watchdog timeouts, but
4341 does not include clean and unclean exit codes or clean
4342 signals. Restart=on-abnormal is an alternative for
4343 Restart=on-failure for services that shall be able to
4344 terminate and avoid restarts on certain errors, by
4345 indicating so with an unclean exit code. Restart=on-failure
4346 or Restart=on-abnormal is now the recommended setting for
4347 all long-running services.
4349 * If the InaccessibleDirectories= service setting points to a
4350 mount point (or if there are any submounts contained within
4351 it), it is now attempted to completely unmount it, to make
4352 the file systems truly unavailable for the respective
4355 * The ReadOnlyDirectories= service setting and
4356 systemd-nspawn's --read-only parameter are now recursively
4357 applied to all submounts, too.
4359 * Mount units may now be created transiently via the bus APIs.
4361 * The support for SysV and LSB init scripts has been removed
4362 from the systemd daemon itself. Instead, it is now
4363 implemented as a generator that creates native systemd units
4364 from these scripts when needed. This enables us to remove a
4365 substantial amount of legacy code from PID 1, following the
4366 fact that many distributions only ship a very small number
4367 of LSB/SysV init scripts nowadays.
4369 * Privileged Xen (dom0) domains are not considered
4370 virtualization anymore by the virtualization detection
4371 logic. After all, they generally have unrestricted access to
4372 the hardware and usually are used to manage the unprivileged
4375 * systemd-tmpfiles gained a new "C" line type, for copying
4376 files or entire directories.
4378 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4379 lines. So far, they have been non-globbing versions of the
4380 latter, and have thus been redundant. In future, it is
4381 recommended to only use "z". "m" has hence been removed
4382 from the documentation, even though it stays supported.
4384 * A tmpfiles snippet to recreate the most basic structure in
4385 /var has been added. This is enough to create the /var/run →
4386 /run symlink and create a couple of structural
4387 directories. This allows systems to boot up with an empty or
4388 volatile /var. Of course, while with this change, the core OS
4389 now is capable with dealing with a volatile /var, not all
4390 user services are ready for it. However, we hope that sooner
4391 or later, many service daemons will be changed upstream so
4392 that they are able to automatically create their necessary
4393 directories in /var at boot, should they be missing. This is
4394 the first step to allow state-less systems that only require
4395 the vendor image for /usr to boot.
4397 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4398 empty tmpfs instance to a specific directory. This is
4399 particularly useful for making use of the automatic
4400 reconstruction of /var (see above), by passing --tmpfs=/var.
4402 * Access modes specified in tmpfiles snippets may now be
4403 prefixed with "~", which indicates that they shall be masked
4404 by whether the existing file or directory is currently
4405 writable, readable or executable at all. Also, if specified,
4406 the sgid/suid/sticky bits will be masked for all
4409 * A new passive target unit "network-pre.target" has been
4410 added which is useful for services that shall run before any
4411 network is configured, for example firewall scripts.
4413 * The "floppy" group that previously owned the /dev/fd*
4414 devices is no longer used. The "disk" group is now used
4415 instead. Distributions should probably deprecate usage of
4418 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4419 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4420 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4421 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4422 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4423 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4426 — Berlin, 2014-06-11
4430 * A new "systemd-timesyncd" daemon has been added for
4431 synchronizing the system clock across the network. It
4432 implements an SNTP client. In contrast to NTP
4433 implementations such as chrony or the NTP reference server,
4434 this only implements a client side, and does not bother with
4435 the full NTP complexity, focusing only on querying time from
4436 one remote server and synchronizing the local clock to
4437 it. Unless you intend to serve NTP to networked clients or
4438 want to connect to local hardware clocks, this simple NTP
4439 client should be more than appropriate for most
4440 installations. The daemon runs with minimal privileges, and
4441 has been hooked up with networkd to only operate when
4442 network connectivity is available. The daemon saves the
4443 current clock to disk every time a new NTP sync has been
4444 acquired, and uses this to possibly correct the system clock
4445 early at bootup, in order to accommodate for systems that
4446 lack an RTC such as the Raspberry Pi and embedded devices,
4447 and to make sure that time monotonically progresses on these
4448 systems, even if it is not always correct. To make use of
4449 this daemon, a new system user and group "systemd-timesync"
4450 needs to be created on installation of systemd.
4452 * The queue "seqnum" interface of libudev has been disabled, as
4453 it was generally incompatible with device namespacing as
4454 sequence numbers of devices go "missing" if the devices are
4455 part of a different namespace.
4457 * "systemctl list-timers" and "systemctl list-sockets" gained
4458 a --recursive switch for showing units of these types also
4459 for all local containers, similar in style to the already
4460 supported --recursive switch for "systemctl list-units".
4462 * A new RebootArgument= setting has been added for service
4463 units, which may be used to specify a kernel reboot argument
4464 to use when triggering reboots with StartLimitAction=.
4466 * A new FailureAction= setting has been added for service
4467 units which may be used to specify an operation to trigger
4468 when a service fails. This works similarly to
4469 StartLimitAction=, but unlike it, controls what is done
4470 immediately rather than only after several attempts to
4471 restart the service in question.
4473 * hostnamed got updated to also expose the kernel name,
4474 release, and version on the bus. This is useful for
4475 executing commands like hostnamectl with the -H switch.
4476 systemd-analyze makes use of this to properly display
4477 details when running non-locally.
4479 * The bootchart tool can now show cgroup information in the
4480 graphs it generates.
4482 * The CFS CPU quota cgroup attribute is now exposed for
4483 services. The new CPUQuota= switch has been added for this
4484 which takes a percentage value. Setting this will have the
4485 result that a service may never get more CPU time than the
4486 specified percentage, even if the machine is otherwise idle.
4488 * systemd-networkd learned IPIP and SIT tunnel support.
4490 * LSB init scripts exposing a dependency on $network will now
4491 get a dependency on network-online.target rather than simply
4492 network.target. This should bring LSB handling closer to
4493 what it was on SysV systems.
4495 * A new fsck.repair= kernel option has been added to control
4496 how fsck shall deal with unclean file systems at boot.
4498 * The (.ini) configuration file parser will now silently
4499 ignore sections whose name begins with "X-". This may be
4500 used to maintain application-specific extension sections in unit
4503 * machined gained a new API to query the IP addresses of
4504 registered containers. "machinectl status" has been updated
4505 to show these addresses in its output.
4507 * A new call sd_uid_get_display() has been added to the
4508 sd-login APIs for querying the "primary" session of a
4509 user. The "primary" session of the user is elected from the
4510 user's sessions and generally a graphical session is
4511 preferred over a text one.
4513 * A minimal systemd-resolved daemon has been added. It
4514 currently simply acts as a companion to systemd-networkd and
4515 manages resolv.conf based on per-interface DNS
4516 configuration, possibly supplied via DHCP. In the long run
4517 we hope to extend this into a local DNSSEC enabled DNS and
4520 * The systemd-networkd-wait-online tool is now enabled by
4521 default. It will delay network-online.target until a network
4522 connection has been configured. The tool primarily integrates
4523 with networkd, but will also make a best effort to make sense
4524 of network configuration performed in some other way.
4526 * Two new service options StartupCPUShares= and
4527 StartupBlockIOWeight= have been added that work similarly to
4528 CPUShares= and BlockIOWeight= however only apply during
4529 system startup. This is useful to prioritize certain services
4530 differently during bootup than during normal runtime.
4532 * hostnamed has been changed to prefer the statically
4533 configured hostname in /etc/hostname (unless set to
4534 'localhost' or empty) over any dynamic one supplied by
4535 dhcp. With this change, the rules for picking the hostname
4536 match more closely the rules of other configuration settings
4537 where the local administrator's configuration in /etc always
4538 overrides any other settings.
4540 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4541 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4542 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4543 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4544 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4545 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4546 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4547 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4548 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4549 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4550 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4551 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4552 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4553 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4554 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4555 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4558 — Beijing, 2014-05-28
4562 * When restoring the screen brightness at boot, stay away from
4563 the darkest setting or from the lowest 5% of the available
4564 range, depending on which is the larger value of both. This
4565 should effectively protect the user from rebooting into a
4566 black screen, should the brightness have been set to minimum
4569 * sd-login gained a new sd_machine_get_class() call to
4570 determine the class ("vm" or "container") of a machine
4571 registered with machined.
4573 * sd-login gained new calls
4574 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4575 to query the identity of the peer of a local AF_UNIX
4576 connection. They operate similarly to their sd_pid_get_xyz()
4579 * PID 1 will now maintain a system-wide system state engine
4580 with the states "starting", "running", "degraded",
4581 "maintenance", "stopping". These states are bound to system
4582 startup, normal runtime, runtime with at least one failed
4583 service, rescue/emergency mode and system shutdown. This
4584 state is shown in the "systemctl status" output when no unit
4585 name is passed. It is useful to determine system state, in
4586 particularly when doing so for many systems or containers at
4589 * A new command "list-machines" has been added to "systemctl"
4590 that lists all local OS containers and shows their system
4591 state (see above), if systemd runs inside of them.
4593 * systemctl gained a new "-r" switch to recursively enumerate
4594 units on all local containers, when used with the
4595 "list-unit" command (which is the default one that is
4596 executed when no parameters are specified).
4598 * The GPT automatic partition discovery logic will now honour
4599 two GPT partition flags: one may be set on a partition to
4600 cause it to be mounted read-only, and the other may be set
4601 on a partition to ignore it during automatic discovery.
4603 * Two new GPT type UUIDs have been added for automatic root
4604 partition discovery, for 32-bit and 64-bit ARM. This is not
4605 particularly useful for discovering the root directory on
4606 these architectures during bare-metal boots (since UEFI is
4607 not common there), but still very useful to allow booting of
4608 ARM disk images in nspawn with the -i option.
4610 * MAC addresses of interfaces created with nspawn's
4611 --network-interface= switch will now be generated from the
4612 machine name, and thus be stable between multiple invocations
4615 * logind will now automatically remove all IPC objects owned
4616 by a user if she or he fully logs out. This makes sure that
4617 users who are logged out cannot continue to consume IPC
4618 resources. This covers SysV memory, semaphores and message
4619 queues as well as POSIX shared memory and message
4620 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4621 limits. With this functionality, that is corrected. This may
4622 be turned off by using the RemoveIPC= switch of logind.conf.
4624 * The systemd-machine-id-setup and tmpfiles tools gained a
4625 --root= switch to operate on a specific root directory,
4628 * journald can now forward logged messages to the TTYs of all
4629 logged in users ("wall"). This is the default for all
4630 emergency messages now.
4632 * A new tool systemd-journal-remote has been added to stream
4633 journal log messages across the network.
4635 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4636 controller trees are mounted into it. Note that the
4637 directories mounted beneath it are not read-only. This is a
4638 security measure and is particularly useful because glibc
4639 actually includes a search logic to pick any tmpfs it can
4640 find to implement shm_open() if /dev/shm is not available
4641 (which it might very well be in namespaced setups).
4643 * machinectl gained a new "poweroff" command to cleanly power
4644 down a local OS container.
4646 * The PrivateDevices= unit file setting will now also drop the
4647 CAP_MKNOD capability from the capability bound set, and
4648 imply DevicePolicy=closed.
4650 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4651 comprehensively on all long-running systemd services where
4652 this is appropriate.
4654 * systemd-udevd will now run in a disassociated mount
4655 namespace. To mount directories from udev rules, make sure to
4656 pull in mount units via SYSTEMD_WANTS properties.
4658 * The kdbus support gained support for uploading policy into
4659 the kernel. sd-bus gained support for creating "monitoring"
4660 connections that can eavesdrop into all bus communication
4661 for debugging purposes.
4663 * Timestamps may now be specified in seconds since the UNIX
4664 epoch Jan 1st, 1970 by specifying "@" followed by the value
4667 * Native tcpwrap support in systemd has been removed. tcpwrap
4668 is old code, not really maintained anymore and has serious
4669 shortcomings, and better options such as firewalls
4670 exist. For setups that require tcpwrap usage, please
4671 consider invoking your socket-activated service via tcpd,
4672 like on traditional inetd.
4674 * A new system.conf configuration option
4675 DefaultTimerAccuracySec= has been added that controls the
4676 default AccuracySec= setting of .timer units.
4678 * Timer units gained a new WakeSystem= switch. If enabled,
4679 timers configured this way will cause the system to resume
4680 from system suspend (if the system supports that, which most
4683 * Timer units gained a new Persistent= switch. If enabled,
4684 timers configured this way will save to disk when they have
4685 been last triggered. This information is then used on next
4686 reboot to possible execute overdue timer events, that
4687 could not take place because the system was powered off.
4688 This enables simple anacron-like behaviour for timer units.
4690 * systemctl's "list-timers" will now also list the time a
4691 timer unit was last triggered in addition to the next time
4692 it will be triggered.
4694 * systemd-networkd will now assign predictable IPv4LL
4695 addresses to its local interfaces.
4697 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4698 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4699 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4700 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4701 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4702 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4703 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4704 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4707 — Berlin, 2014-03-25
4711 * A new unit file setting RestrictAddressFamilies= has been
4712 added to restrict which socket address families unit
4713 processes gain access to. This takes address family names
4714 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4715 attack surface of services via exotic protocol stacks. This
4716 is built on seccomp system call filters.
4718 * Two new unit file settings RuntimeDirectory= and
4719 RuntimeDirectoryMode= have been added that may be used to
4720 manage a per-daemon runtime directories below /run. This is
4721 an alternative for setting up directory permissions with
4722 tmpfiles snippets, and has the advantage that the runtime
4723 directory's lifetime is bound to the daemon runtime and that
4724 the daemon starts up with an empty directory each time. This
4725 is particularly useful when writing services that drop
4726 privileges using the User= or Group= setting.
4728 * The DeviceAllow= unit setting now supports globbing for
4729 matching against device group names.
4731 * The systemd configuration file system.conf gained new
4732 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4733 DefaultMemoryAccounting= to globally turn on/off accounting
4734 for specific resources (cgroups) for all units. These
4735 settings may still be overridden individually in each unit
4738 * systemd-gpt-auto-generator is now able to discover /srv and
4739 root partitions in addition to /home and swap partitions. It
4740 also supports LUKS-encrypted partitions now. With this in
4741 place, automatic discovery of partitions to mount following
4742 the Discoverable Partitions Specification
4743 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4744 is now a lot more complete. This allows booting without
4745 /etc/fstab and without root= on the kernel command line on
4746 systems prepared appropriately.
4748 * systemd-nspawn gained a new --image= switch which allows
4749 booting up disk images and Linux installations on any block
4750 device that follow the Discoverable Partitions Specification
4751 (see above). This means that installations made with
4752 appropriately updated installers may now be started and
4753 deployed using container managers, completely
4754 unmodified. (We hope that libvirt-lxc will add support for
4755 this feature soon, too.)
4757 * systemd-nspawn gained a new --network-macvlan= setting to
4758 set up a private macvlan interface for the
4759 container. Similarly, systemd-networkd gained a new
4760 Kind=macvlan setting in .netdev files.
4762 * systemd-networkd now supports configuring local addresses
4765 * A new tool systemd-network-wait-online has been added to
4766 synchronously wait for network connectivity using
4769 * The sd-bus.h bus API gained a new sd_bus_track object for
4770 tracking the life-cycle of bus peers. Note that sd-bus.h is
4771 still not a public API though (unless you specify
4772 --enable-kdbus on the configure command line, which however
4773 voids your warranty and you get no API stability guarantee).
4775 * The $XDG_RUNTIME_DIR runtime directories for each user are
4776 now individual tmpfs instances, which has the benefit of
4777 introducing separate pools for each user, with individual
4778 size limits, and thus making sure that unprivileged clients
4779 can no longer negatively impact the system or other users by
4780 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4781 RuntimeDirectorySize= has been introduced that allows
4782 controlling the default size limit for all users. It
4783 defaults to 10% of the available physical memory. This is no
4784 replacement for quotas on tmpfs though (which the kernel
4785 still does not support), as /dev/shm and /tmp are still
4786 shared resources used by both the system and unprivileged
4789 * logind will now automatically turn off automatic suspending
4790 on laptop lid close when more than one display is
4791 connected. This was previously expected to be implemented
4792 individually in desktop environments (such as GNOME),
4793 however has been added to logind now, in order to fix a
4794 boot-time race where a desktop environment might not have
4795 been started yet and thus not been able to take an inhibitor
4796 lock at the time where logind already suspends the system
4797 due to a closed lid.
4799 * logind will now wait at least 30s after each system
4800 suspend/resume cycle, and 3min after system boot before
4801 suspending the system due to a closed laptop lid. This
4802 should give USB docking stations and similar enough time to
4803 be probed and configured after system resume and boot in
4804 order to then act as suspend blocker.
4806 * systemd-run gained a new --property= setting which allows
4807 initialization of resource control properties (and others)
4808 for the created scope or service unit. Example: "systemd-run
4809 --property=BlockIOWeight=10 updatedb" may be used to run
4810 updatedb at a low block IO scheduling weight.
4812 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4813 now also work in --scope mode.
4815 * When systemd is compiled with kdbus support, basic support
4816 for enforced policies is now in place. (Note that enabling
4817 kdbus still voids your warranty and no API compatibility
4820 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4821 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4822 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4823 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4824 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4825 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4826 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4827 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4828 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4829 Zbigniew Jędrzejewski-Szmek
4831 — Berlin, 2014-03-12
4835 * systemd will now relabel /dev after loading the SMACK policy
4836 according to SMACK rules.
4838 * A new unit file option AppArmorProfile= has been added to
4839 set the AppArmor profile for the processes of a unit.
4841 * A new condition check ConditionArchitecture= has been added
4842 to conditionalize units based on the system architecture, as
4843 reported by uname()'s "machine" field.
4845 * systemd-networkd now supports matching on the system
4846 virtualization, architecture, kernel command line, host name
4849 * logind is now a lot more aggressive when suspending the
4850 machine due to a closed laptop lid. Instead of acting only
4851 on the lid close action, it will continuously watch the lid
4852 status and act on it. This is useful for laptops where the
4853 power button is on the outside of the chassis so that it can
4854 be reached without opening the lid (such as the Lenovo
4855 Yoga). On those machines, logind will now immediately
4856 re-suspend the machine if the power button has been
4857 accidentally pressed while the laptop was suspended and in a
4858 backpack or similar.
4860 * logind will now watch SW_DOCK switches and inhibit reaction
4861 to the lid switch if it is pressed. This means that logind
4862 will not suspend the machine anymore if the lid is closed
4863 and the system is docked, if the laptop supports SW_DOCK
4864 notifications via the input layer. Note that ACPI docking
4865 stations do not generate this currently. Also note that this
4866 logic is usually not fully sufficient and Desktop
4867 Environments should take a lid switch inhibitor lock when an
4868 external display is connected, as systemd will not watch
4871 * nspawn will now make use of the devices cgroup controller by
4872 default, and only permit creation of and access to the usual
4873 API device nodes like /dev/null or /dev/random, as well as
4874 access to (but not creation of) the pty devices.
4876 * We will now ship a default .network file for
4877 systemd-networkd that automatically configures DHCP for
4878 network interfaces created by nspawn's --network-veth or
4879 --network-bridge= switches.
4881 * systemd will now understand the usual M, K, G, T suffixes
4882 according to SI conventions (i.e. to the base 1000) when
4883 referring to throughput and hardware metrics. It will stay
4884 with IEC conventions (i.e. to the base 1024) for software
4885 metrics, according to what is customary according to
4886 Wikipedia. We explicitly document which base applies for
4887 each configuration option.
4889 * The DeviceAllow= setting in unit files now supports a syntax
4890 to whitelist an entire group of devices node majors at once,
4891 based on the /proc/devices listing. For example, with the
4892 string "char-pts", it is now possible to whitelist all
4893 current and future pseudo-TTYs at once.
4895 * sd-event learned a new "post" event source. Event sources of
4896 this type are triggered by the dispatching of any event
4897 source of a type that is not "post". This is useful for
4898 implementing clean-up and check event sources that are
4899 triggered by other work being done in the program.
4901 * systemd-networkd is no longer statically enabled, but uses
4902 the usual [Install] sections so that it can be
4903 enabled/disabled using systemctl. It still is enabled by
4906 * When creating a veth interface pair with systemd-nspawn, the
4907 host side will now be prefixed with "vb-" if
4908 --network-bridge= is used, and with "ve-" if --network-veth
4909 is used. This way, it is easy to distinguish these cases on
4910 the host, for example to apply different configuration to
4911 them with systemd-networkd.
4913 * The compatibility libraries for libsystemd-journal.so,
4914 libsystem-id128.so, libsystemd-login.so and
4915 libsystemd-daemon.so do not make use of IFUNC
4916 anymore. Instead, we now build libsystemd.so multiple times
4917 under these alternative names. This means that the footprint
4918 is drastically increased, but given that these are
4919 transitional compatibility libraries, this should not matter
4920 much. This change has been made necessary to support the ARM
4921 platform for these compatibility libraries, as the ARM
4922 toolchain is not really at the same level as the toolchain
4923 for other architectures like x86 and does not support
4924 IFUNC. Please make sure to use --enable-compat-libs only
4925 during a transitional period!
4927 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4928 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4929 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4930 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4931 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4932 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4933 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4934 Zbigniew Jędrzejewski-Szmek
4936 — Berlin, 2014-02-24
4940 * A new component "systemd-networkd" has been added that can
4941 be used to configure local network interfaces statically or
4942 via DHCP. It is capable of bringing up bridges, VLANs, and
4943 bonding. Currently, no hook-ups for interactive network
4944 configuration are provided. Use this for your initrd,
4945 container, embedded, or server setup if you need a simple,
4946 yet powerful, network configuration solution. This
4947 configuration subsystem is quite nifty, as it allows wildcard
4948 hotplug matching in interfaces. For example, with a single
4949 configuration snippet, you can configure that all Ethernet
4950 interfaces showing up are automatically added to a bridge,
4951 or similar. It supports link-sensing and more.
4953 * A new tool "systemd-socket-proxyd" has been added which can
4954 act as a bidirectional proxy for TCP sockets. This is
4955 useful for adding socket activation support to services that
4956 do not actually support socket activation, including virtual
4957 machines and the like.
4959 * Add a new tool to save/restore rfkill state on
4962 * Save/restore state of keyboard backlights in addition to
4963 display backlights on shutdown/boot.
4965 * udev learned a new SECLABEL{} construct to label device
4966 nodes with a specific security label when they appear. For
4967 now, only SECLABEL{selinux} is supported, but the syntax is
4968 prepared for additional security frameworks.
4970 * udev gained a new scheme to configure link-level attributes
4971 from files in /etc/systemd/network/*.link. These files can
4972 match against MAC address, device path, driver name and type,
4973 and will apply attributes like the naming policy, link speed,
4974 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4975 address assignment policy (randomized, ...).
4977 * The configuration of network interface naming rules for
4978 "permanent interface names" has changed: a new NamePolicy=
4979 setting in the [Link] section of .link files determines the
4980 priority of possible naming schemes (onboard, slot, MAC,
4981 path). The default value of this setting is determined by
4982 /usr/lib/net/links/99-default.link. Old
4983 80-net-name-slot.rules udev configuration file has been
4984 removed, so local configuration overriding this file should
4985 be adapted to override 99-default.link instead.
4987 * When the User= switch is used in a unit file, also
4988 initialize $SHELL= based on the user database entry.
4990 * systemd no longer depends on libdbus. All communication is
4991 now done with sd-bus, systemd's low-level bus library
4994 * kdbus support has been added to PID 1 itself. When kdbus is
4995 enabled, this causes PID 1 to set up the system bus and
4996 enable support for a new ".busname" unit type that
4997 encapsulates bus name activation on kdbus. It works a little
4998 bit like ".socket" units, except for bus names. A new
4999 generator has been added that converts classic dbus1 service
5000 activation files automatically into native systemd .busname
5003 * sd-bus: add a light-weight vtable implementation that allows
5004 defining objects on the bus with a simple static const
5005 vtable array of its methods, signals and properties.
5007 * systemd will not generate or install static dbus
5008 introspection data anymore to /usr/share/dbus-1/interfaces,
5009 as the precise format of these files is unclear, and
5010 nothing makes use of it.
5012 * A proxy daemon is now provided to proxy clients connecting
5013 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5014 compatibility with classic D-Bus.
5016 * A bus driver implementation has been added that supports the
5017 classic D-Bus bus driver calls on kdbus, also for
5018 compatibility purposes.
5020 * A new API "sd-event.h" has been added that implements a
5021 minimal event loop API built around epoll. It provides a
5022 couple of features that direct epoll usage is lacking:
5023 prioritization of events, scales to large numbers of timer
5024 events, per-event timer slack (accuracy), system-wide
5025 coalescing of timer events, exit handlers, watchdog
5026 supervision support using systemd's sd_notify() API, child
5029 * A new API "sd-rntl.h" has been added that provides an API
5030 around the route netlink interface of the kernel, similar in
5031 style to "sd-bus.h".
5033 * A new API "sd-dhcp-client.h" has been added that provides a
5034 small DHCPv4 client-side implementation. This is used by
5037 * There is a new kernel command line option
5038 "systemd.restore_state=0|1". When set to "0", none of the
5039 systemd tools will restore saved runtime state to hardware
5040 devices. More specifically, the rfkill and backlight states
5043 * The FsckPassNo= compatibility option in mount/service units
5044 has been removed. The fstab generator will now add the
5045 necessary dependencies automatically, and does not require
5046 PID1's support for that anymore.
5048 * journalctl gained a new switch, --list-boots, that lists
5049 recent boots with their times and boot IDs.
5051 * The various tools like systemctl, loginctl, timedatectl,
5052 busctl, systemd-run, ... have gained a new switch "-M" to
5053 connect to a specific, local OS container (as direct
5054 connection, without requiring SSH). This works on any
5055 container that is registered with machined, such as those
5056 created by libvirt-lxc or nspawn.
5058 * systemd-run and systemd-analyze also gained support for "-H"
5059 to connect to remote hosts via SSH. This is particularly
5060 useful for systemd-run because it enables queuing of jobs
5061 onto remote systems.
5063 * machinectl gained a new command "login" to open a getty
5064 login in any local container. This works with any container
5065 that is registered with machined (such as those created by
5066 libvirt-lxc or nspawn), and which runs systemd inside.
5068 * machinectl gained a new "reboot" command that may be used to
5069 trigger a reboot on a specific container that is registered
5070 with machined. This works on any container that runs an init
5071 system of some kind.
5073 * systemctl gained a new "list-timers" command to print a nice
5074 listing of installed timer units with the times they elapse
5077 * Alternative reboot() parameters may now be specified on the
5078 "systemctl reboot" command line and are passed to the
5079 reboot() system call.
5081 * systemctl gained a new --job-mode= switch to configure the
5082 mode to queue a job with. This is a more generic version of
5083 --fail, --irreversible, and --ignore-dependencies, which are
5084 still available but not advertised anymore.
5086 * /etc/systemd/system.conf gained new settings to configure
5087 various default timeouts of units, as well as the default
5088 start limit interval and burst. These may still be overridden
5091 * PID1 will now export on the bus profile data of the security
5092 policy upload process (such as the SELinux policy upload to
5095 * journald: when forwarding logs to the console, include
5096 timestamps (following the setting in
5097 /sys/module/printk/parameters/time).
5099 * OnCalendar= in timer units now understands the special
5100 strings "yearly" and "annually". (Both are equivalent)
5102 * The accuracy of timer units is now configurable with the new
5103 AccuracySec= setting. It defaults to 1min.
5105 * A new dependency type JoinsNamespaceOf= has been added that
5106 allows running two services within the same /tmp and network
5107 namespace, if PrivateNetwork= or PrivateTmp= are used.
5109 * A new command "cat" has been added to systemctl. It outputs
5110 the original unit file of a unit, and concatenates the
5111 contents of additional "drop-in" unit file snippets, so that
5112 the full configuration is shown.
5114 * systemctl now supports globbing on the various "list-xyz"
5115 commands, like "list-units" or "list-sockets", as well as on
5116 those commands which take multiple unit names.
5118 * journalctl's --unit= switch gained support for globbing.
5120 * All systemd daemons now make use of the watchdog logic so
5121 that systemd automatically notices when they hang.
5123 * If the $container_ttys environment variable is set,
5124 getty-generator will automatically spawn a getty for each
5125 listed tty. This is useful for container managers to request
5126 login gettys to be spawned on as many ttys as needed.
5128 * %h, %s, %U specifier support is not available anymore when
5129 used in unit files for PID 1. This is because NSS calls are
5130 not safe from PID 1. They stay available for --user
5131 instances of systemd, and as special case for the root user.
5133 * loginctl gained a new "--no-legend" switch to turn off output
5136 * The "sd-login.h" API gained three new calls:
5137 sd_session_is_remote(), sd_session_get_remote_user(),
5138 sd_session_get_remote_host() to query information about
5141 * The udev hardware database now also carries vendor/product
5142 information of SDIO devices.
5144 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5145 determine whether watchdog notifications are requested by
5148 * Socket-activated per-connection services now include a
5149 short description of the connection parameters in the
5152 * tmpfiles gained a new "--boot" option. When this is not used,
5153 only lines where the command character is not suffixed with
5154 "!" are executed. When this option is specified, those
5155 options are executed too. This partitions tmpfiles
5156 directives into those that can be safely executed at any
5157 time, and those which should be run only at boot (for
5158 example, a line that creates /run/nologin).
5160 * A new API "sd-resolve.h" has been added which provides a simple
5161 asynchronous wrapper around glibc NSS host name resolution
5162 calls, such as getaddrinfo(). In contrast to glibc's
5163 getaddrinfo_a(), it does not use signals. In contrast to most
5164 other asynchronous name resolution libraries, this one does
5165 not reimplement DNS, but reuses NSS, so that alternate
5166 host name resolution systems continue to work, such as mDNS,
5167 LDAP, etc. This API is based on libasyncns, but it has been
5168 cleaned up for inclusion in systemd.
5170 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5171 "sd-daemon.h" are no longer found in individual libraries
5172 libsystemd-journal.so, libsystemd-login.so,
5173 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5174 merged them into a single library, libsystemd.so, which
5175 provides all symbols. The reason for this is cyclic
5176 dependencies, as these libraries tend to use each other's
5177 symbols. So far, we have managed to workaround that by linking
5178 a copy of a good part of our code into each of these
5179 libraries again and again, which, however, makes certain
5180 things hard to do, like sharing static variables. Also, it
5181 substantially increases footprint. With this change, there
5182 is only one library for the basic APIs systemd
5183 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5184 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5185 library as well, however are subject to the --enable-kdbus
5186 switch (see below). Note that "sd-dhcp-client.h" is not part
5187 of this library (this is because it only consumes, never
5188 provides, services of/to other APIs). To make the transition
5189 easy from the separate libraries to the unified one, we
5190 provide the --enable-compat-libs compile-time switch which
5191 will generate stub libraries that are compatible with the
5192 old ones but redirect all calls to the new one.
5194 * All of the kdbus logic and the new APIs "sd-bus.h",
5195 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5196 and "sd-utf8.h" are compile-time optional via the
5197 "--enable-kdbus" switch, and they are not compiled in by
5198 default. To make use of kdbus, you have to explicitly enable
5199 the switch. Note however, that neither the kernel nor the
5200 userspace API for all of this is considered stable yet. We
5201 want to maintain the freedom to still change the APIs for
5202 now. By specifying this build-time switch, you acknowledge
5203 that you are aware of the instability of the current
5206 * Also, note that while kdbus is pretty much complete,
5207 it lacks one thing: proper policy support. This means you
5208 can build a fully working system with all features; however,
5209 it will be highly insecure. Policy support will be added in
5210 one of the next releases, at the same time that we will
5211 declare the APIs stable.
5213 * When the kernel command line argument "kdbus" is specified,
5214 systemd will automatically load the kdbus.ko kernel module. At
5215 this stage of development, it is only useful for testing kdbus
5216 and should not be used in production. Note: if "--enable-kdbus"
5217 is specified, and the kdbus.ko kernel module is available, and
5218 "kdbus" is added to the kernel command line, the entire system
5219 runs with kdbus instead of dbus-daemon, with the above mentioned
5220 problem of missing the system policy enforcement. Also a future
5221 version of kdbus.ko or a newer systemd will not be compatible with
5222 each other, and will unlikely be able to boot the machine if only
5223 one of them is updated.
5225 * systemctl gained a new "import-environment" command which
5226 uploads the caller's environment (or parts thereof) into the
5227 service manager so that it is inherited by services started
5228 by the manager. This is useful to upload variables like
5229 $DISPLAY into the user service manager.
5231 * A new PrivateDevices= switch has been added to service units
5232 which allows running a service with a namespaced /dev
5233 directory that does not contain any device nodes for
5234 physical devices. More specifically, it only includes devices
5235 such as /dev/null, /dev/urandom, and /dev/zero which are API
5238 * logind has been extended to support behaviour like VT
5239 switching on seats that do not support a VT. This makes
5240 multi-session available on seats that are not the first seat
5241 (seat0), and on systems where kernel support for VTs has
5242 been disabled at compile-time.
5244 * If a process holds a delay lock for system sleep or shutdown
5245 and fails to release it in time, we will now log its
5246 identity. This makes it easier to identify processes that
5247 cause slow suspends or power-offs.
5249 * When parsing /etc/crypttab, support for a new key-slot=
5250 option as supported by Debian is added. It allows indicating
5251 which LUKS slot to use on disk, speeding up key loading.
5253 * The sd_journald_sendv() API call has been checked and
5254 officially declared to be async-signal-safe so that it may
5255 be invoked from signal handlers for logging purposes.
5257 * Boot-time status output is now enabled automatically after a
5258 short timeout if boot does not progress, in order to give
5259 the user an indication what she or he is waiting for.
5261 * The boot-time output has been improved to show how much time
5262 remains until jobs expire.
5264 * The KillMode= switch in service units gained a new possible
5265 value "mixed". If set, and the unit is shut down, then the
5266 initial SIGTERM signal is sent only to the main daemon
5267 process, while the following SIGKILL signal is sent to
5268 all remaining processes of the service.
5270 * When a scope unit is registered, a new property "Controller"
5271 may be set. If set to a valid bus name, systemd will send a
5272 RequestStop() signal to this name when it would like to shut
5273 down the scope. This may be used to hook manager logic into
5274 the shutdown logic of scope units. Also, scope units may now
5275 be put in a special "abandoned" state, in which case the
5276 manager process which created them takes no further
5277 responsibilities for it.
5279 * When reading unit files, systemd will now verify
5280 the access mode of these files, and warn about certain
5281 suspicious combinations. This has been added to make it
5282 easier to track down packaging bugs where unit files are
5283 marked executable or world-writable.
5285 * systemd-nspawn gained a new "--setenv=" switch to set
5286 container-wide environment variables. The similar option in
5287 systemd-activate was renamed from "--environment=" to
5288 "--setenv=" for consistency.
5290 * systemd-nspawn has been updated to create a new kdbus domain
5291 for each container that is invoked, thus allowing each
5292 container to have its own set of system and user buses,
5293 independent of the host.
5295 * systemd-nspawn gained a new --drop-capability= switch to run
5296 the container with less capabilities than the default. Both
5297 --drop-capability= and --capability= now take the special
5298 string "all" for dropping or keeping all capabilities.
5300 * systemd-nspawn gained new switches for executing containers
5301 with specific SELinux labels set.
5303 * systemd-nspawn gained a new --quiet switch to not generate
5304 any additional output but the container's own console
5307 * systemd-nspawn gained a new --share-system switch to run a
5308 container without PID namespacing enabled.
5310 * systemd-nspawn gained a new --register= switch to control
5311 whether the container is registered with systemd-machined or
5312 not. This is useful for containers that do not run full
5313 OS images, but only specific apps.
5315 * systemd-nspawn gained a new --keep-unit which may be used
5316 when invoked as the only program from a service unit, and
5317 results in registration of the unit service itself in
5318 systemd-machined, instead of a newly opened scope unit.
5320 * systemd-nspawn gained a new --network-interface= switch for
5321 moving arbitrary interfaces to the container. The new
5322 --network-veth switch creates a virtual Ethernet connection
5323 between host and container. The new --network-bridge=
5324 switch then allows assigning the host side of this virtual
5325 Ethernet connection to a bridge device.
5327 * systemd-nspawn gained a new --personality= switch for
5328 setting the kernel personality for the container. This is
5329 useful when running a 32-bit container on a 64-bit host. A
5330 similar option Personality= is now also available for service
5333 * logind will now also track a "Desktop" identifier for each
5334 session which encodes the desktop environment of it. This is
5335 useful for desktop environments that want to identify
5336 multiple running sessions of itself easily.
5338 * A new SELinuxContext= setting for service units has been
5339 added that allows setting a specific SELinux execution
5340 context for a service.
5342 * Most systemd client tools will now honour $SYSTEMD_LESS for
5343 settings of the "less" pager. By default, these tools will
5344 override $LESS to allow certain operations to work, such as
5345 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5346 influence this logic.
5348 * systemd's "seccomp" hook-up has been changed to make use of
5349 the libseccomp library instead of using its own
5350 implementation. This has benefits for portability among
5353 * For usage together with SystemCallFilter=, a new
5354 SystemCallErrorNumber= setting has been introduced that
5355 allows configuration of a system error number to be returned
5356 on filtered system calls, instead of immediately killing the
5357 process. Also, SystemCallArchitectures= has been added to
5358 limit access to system calls of a particular architecture
5359 (in order to turn off support for unused secondary
5360 architectures). There is also a global
5361 SystemCallArchitectures= setting in system.conf now to turn
5362 off support for non-native system calls system-wide.
5364 * systemd requires a kernel with a working name_to_handle_at(),
5365 please see the kernel config requirements in the README file.
5367 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5368 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5369 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5370 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5371 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5372 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5373 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5374 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5375 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5376 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5377 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5378 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5379 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5380 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5381 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5382 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5383 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5384 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5385 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5386 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5387 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5388 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5389 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5390 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5392 — Berlin, 2014-02-20
5396 * logind has gained support for facilitating privileged input
5397 and drm device access for unprivileged clients. This work is
5398 useful to allow Wayland display servers (and similar
5399 programs, such as kmscon) to run under the user's ID and
5400 access input and drm devices which are normally
5401 protected. When this is used (and the kernel is new enough)
5402 logind will "mute" IO on the file descriptors passed to
5403 Wayland as long as it is in the background and "unmute" it
5404 if it returns into the foreground. This allows secure
5405 session switching without allowing background sessions to
5406 eavesdrop on input and display data. This also introduces
5407 session switching support if VT support is turned off in the
5408 kernel, and on seats that are not seat0.
5410 * A new kernel command line option luks.options= is understood
5411 now which allows specifying LUKS options for usage for LUKS
5412 encrypted partitions specified with luks.uuid=.
5414 * tmpfiles.d(5) snippets may now use specifier expansion in
5415 path names. More specifically %m, %b, %H, %v, are now
5416 replaced by the local machine id, boot id, hostname, and
5417 kernel version number.
5419 * A new tmpfiles.d(5) command "m" has been introduced which
5420 may be used to change the owner/group/access mode of a file
5421 or directory if it exists, but do nothing if it does not.
5423 * This release removes high-level support for the
5424 MemorySoftLimit= cgroup setting. The underlying kernel
5425 cgroup attribute memory.soft_limit= is currently badly
5426 designed and likely to be removed from the kernel API in its
5427 current form, hence we should not expose it for now.
5429 * The memory.use_hierarchy cgroup attribute is now enabled for
5430 all cgroups systemd creates in the memory cgroup
5431 hierarchy. This option is likely to be come the built-in
5432 default in the kernel anyway, and the non-hierarchical mode
5433 never made much sense in the intrinsically hierarchical
5436 * A new field _SYSTEMD_SLICE= is logged along with all journal
5437 messages containing the slice a message was generated
5438 from. This is useful to allow easy per-customer filtering of
5439 logs among other things.
5441 * systemd-journald will no longer adjust the group of journal
5442 files it creates to the "systemd-journal" group. Instead we
5443 rely on the journal directory to be owned by the
5444 "systemd-journal" group, and its setgid bit set, so that the
5445 kernel file system layer will automatically enforce that
5446 journal files inherit this group assignment. The reason for
5447 this change is that we cannot allow NSS look-ups from
5448 journald which would be necessary to resolve
5449 "systemd-journal" to a numeric GID, because this might
5450 create deadlocks if NSS involves synchronous queries to
5451 other daemons (such as nscd, or sssd) which in turn are
5452 logging clients of journald and might block on it, which
5453 would then dead lock. A tmpfiles.d(5) snippet included in
5454 systemd will make sure the setgid bit and group are
5455 properly set on the journal directory if it exists on every
5456 boot. However, we recommend adjusting it manually after
5457 upgrades too (or from RPM scriptlets), so that the change is
5458 not delayed until next reboot.
5460 * Backlight and random seed files in /var/lib/ have moved into
5461 the /var/lib/systemd/ directory, in order to centralize all
5462 systemd generated files in one directory.
5464 * Boot time performance measurements (as displayed by
5465 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5466 performance information if that's available to determine how
5467 much time BIOS and boot loader initialization required. With
5468 a sufficiently new BIOS you hence no longer need to boot
5469 with Gummiboot to get access to such information.
5471 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5472 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5473 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5474 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5475 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5476 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5477 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5479 — Berlin, 2013-10-02
5483 * The Restart= option for services now understands a new
5484 on-watchdog setting, which will restart the service
5485 automatically if the service stops sending out watchdog keep
5486 alive messages (as configured with WatchdogSec=).
5488 * The getty generator (which is responsible for bringing up a
5489 getty on configured serial consoles) will no longer only
5490 start a getty on the primary kernel console but on all
5491 others, too. This makes the order in which console= is
5492 specified on the kernel command line less important.
5494 * libsystemd-logind gained a new sd_session_get_vt() call to
5495 retrieve the VT number of a session.
5497 * If the option "tries=0" is set for an entry of /etc/crypttab
5498 its passphrase is queried indefinitely instead of any
5499 maximum number of tries.
5501 * If a service with a configure PID file terminates its PID
5502 file will now be removed automatically if it still exists
5503 afterwards. This should put an end to stale PID files.
5505 * systemd-run will now also take relative binary path names
5506 for execution and no longer insists on absolute paths.
5508 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5509 paths that are optionally prefixed with "-" to indicate that
5510 it should not be considered a failure if they do not exist.
5512 * journalctl -o (and similar commands) now understands a new
5513 output mode "short-precise", it is similar to "short" but
5514 shows timestamps with usec accuracy.
5516 * The option "discard" (as known from Debian) is now
5517 synonymous to "allow-discards" in /etc/crypttab. In fact,
5518 "discard" is preferred now (since it is easier to remember
5521 * Some licensing clean-ups were made, so that more code is now
5522 LGPL-2.1 licensed than before.
5524 * A minimal tool to save/restore the display backlight
5525 brightness across reboots has been added. It will store the
5526 backlight setting as late as possible at shutdown, and
5527 restore it as early as possible during reboot.
5529 * A logic to automatically discover and enable home and swap
5530 partitions on GPT disks has been added. With this in place
5531 /etc/fstab becomes optional for many setups as systemd can
5532 discover certain partitions located on the root disk
5533 automatically. Home partitions are recognized under their
5534 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5535 partitions are recognized under their GPT type ID
5536 0657fd6da4ab43c484e50933c84b4f4f.
5538 * systemd will no longer pass any environment from the kernel
5539 or initrd to system services. If you want to set an
5540 environment for all services, do so via the kernel command
5541 line systemd.setenv= assignment.
5543 * The systemd-sysctl tool no longer natively reads the file
5544 /etc/sysctl.conf. If desired, the file should be symlinked
5545 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5546 legacy support by a symlink rather than built-in code, it
5547 also makes the otherwise hidden order of application of the
5548 different files visible. (Note that this partly reverts to a
5549 pre-198 application order of sysctl knobs!)
5551 * The "systemctl set-log-level" and "systemctl dump" commands
5552 have been moved to systemd-analyze.
5554 * systemd-run learned the new --remain-after-exit switch,
5555 which causes the scope unit not to be cleaned up
5556 automatically after the process terminated.
5558 * tmpfiles learned a new --exclude-prefix= switch to exclude
5559 certain paths from operation.
5561 * journald will now automatically flush all messages to disk
5562 as soon as a message at the log level CRIT, ALERT or EMERG
5565 Contributions from: Andrew Cook, Brandon Philips, Christian
5566 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5567 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5568 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5569 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5570 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5571 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5572 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5573 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5574 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5575 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5576 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5577 William Giokas, Zbigniew Jędrzejewski-Szmek
5579 — Berlin, 2013-09-13
5583 * The documentation has been updated to cover the various new
5584 concepts introduced with 205.
5586 * Unit files now understand the new %v specifier which
5587 resolves to the kernel version string as returned by "uname
5590 * systemctl now supports filtering the unit list output by
5591 load state, active state and sub state, using the new
5594 * "systemctl status" will now show the results of the
5595 condition checks (like ConditionPathExists= and similar) of
5596 the last start attempts of the unit. They are also logged to
5599 * "journalctl -b" may now be used to look for boot output of a
5600 specific boot. Try "journalctl -b -1" for the previous boot,
5601 but the syntax is substantially more powerful.
5603 * "journalctl --show-cursor" has been added which prints the
5604 cursor string the last shown log line. This may then be used
5605 with the new "journalctl --after-cursor=" switch to continue
5606 browsing logs from that point on.
5608 * "journalctl --force" may now be used to force regeneration
5611 * Creation of "dead" device nodes has been moved from udev
5612 into kmod and tmpfiles. Previously, udev would read the kmod
5613 databases to pre-generate dead device nodes based on meta
5614 information contained in kernel modules, so that these would
5615 be auto-loaded on access rather then at boot. As this
5616 does not really have much to do with the exposing actual
5617 kernel devices to userspace this has always been slightly
5618 alien in the udev codebase. Following the new scheme kmod
5619 will now generate a runtime snippet for tmpfiles from the
5620 module meta information and it now is tmpfiles' job to the
5621 create the nodes. This also allows overriding access and
5622 other parameters for the nodes using the usual tmpfiles
5623 facilities. As side effect this allows us to remove the
5624 CAP_SYS_MKNOD capability bit from udevd entirely.
5626 * logind's device ACLs may now be applied to these "dead"
5627 devices nodes too, thus finally allowing managed access to
5628 devices such as /dev/snd/sequencer without loading the
5629 backing module right-away.
5631 * A new RPM macro has been added that may be used to apply
5632 tmpfiles configuration during package installation.
5634 * systemd-detect-virt and ConditionVirtualization= now can
5635 detect User-Mode-Linux machines (UML).
5637 * journald will now implicitly log the effective capabilities
5638 set of processes in the message metadata.
5640 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5642 * The initrd interface has been simplified (more specifically,
5643 support for passing performance data via environment
5644 variables and fsck results via files in /run has been
5645 removed). These features were non-essential, and are
5646 nowadays available in a much nicer way by having systemd in
5647 the initrd serialize its state and have the hosts systemd
5648 deserialize it again.
5650 * The udev "keymap" data files and tools to apply keyboard
5651 specific mappings of scan to key codes, and force-release
5652 scan code lists have been entirely replaced by a udev
5653 "keyboard" builtin and a hwdb data file.
5655 * systemd will now honour the kernel's "quiet" command line
5656 argument also during late shutdown, resulting in a
5657 completely silent shutdown when used.
5659 * There's now an option to control the SO_REUSEPORT socket
5660 option in .socket units.
5662 * Instance units will now automatically get a per-template
5663 subslice of system.slice unless something else is explicitly
5664 configured. For example, instances of sshd@.service will now
5665 implicitly be placed in system-sshd.slice rather than
5666 system.slice as before.
5668 * Test coverage support may now be enabled at build time.
5670 Contributions from: Dave Reisner, Frederic Crozat, Harald
5671 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5672 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5673 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5674 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5675 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5676 Giokas, Zbigniew Jędrzejewski-Szmek
5678 — Berlin, 2013-07-23
5682 * Two new unit types have been introduced:
5684 Scope units are very similar to service units, however, are
5685 created out of pre-existing processes — instead of PID 1
5686 forking off the processes. By using scope units it is
5687 possible for system services and applications to group their
5688 own child processes (worker processes) in a powerful way
5689 which then maybe used to organize them, or kill them
5690 together, or apply resource limits on them.
5692 Slice units may be used to partition system resources in an
5693 hierarchical fashion and then assign other units to them. By
5694 default there are now three slices: system.slice (for all
5695 system services), user.slice (for all user sessions),
5696 machine.slice (for VMs and containers).
5698 Slices and scopes have been introduced primarily in
5699 context of the work to move cgroup handling to a
5700 single-writer scheme, where only PID 1
5701 creates/removes/manages cgroups.
5703 * There's a new concept of "transient" units. In contrast to
5704 normal units these units are created via an API at runtime,
5705 not from configuration from disk. More specifically this
5706 means it is now possible to run arbitrary programs as
5707 independent services, with all execution parameters passed
5708 in via bus APIs rather than read from disk. Transient units
5709 make systemd substantially more dynamic then it ever was,
5710 and useful as a general batch manager.
5712 * logind has been updated to make use of scope and slice units
5713 for managing user sessions. As a user logs in he will get
5714 his own private slice unit, to which all sessions are added
5715 as scope units. We also added support for automatically
5716 adding an instance of user@.service for the user into the
5717 slice. Effectively logind will no longer create cgroup
5718 hierarchies on its own now, it will defer entirely to PID 1
5719 for this by means of scope, service and slice units. Since
5720 user sessions this way become entities managed by PID 1
5721 the output of "systemctl" is now a lot more comprehensive.
5723 * A new mini-daemon "systemd-machined" has been added which
5724 may be used by virtualization managers to register local
5725 VMs/containers. nspawn has been updated accordingly, and
5726 libvirt will be updated shortly. machined will collect a bit
5727 of meta information about the VMs/containers, and assign
5728 them their own scope unit (see above). The collected
5729 meta-data is then made available via the "machinectl" tool,
5730 and exposed in "ps" and similar tools. machined/machinectl
5731 is compile-time optional.
5733 * As discussed earlier, the low-level cgroup configuration
5734 options ControlGroup=, ControlGroupModify=,
5735 ControlGroupPersistent=, ControlGroupAttribute= have been
5736 removed. Please use high-level attribute settings instead as
5737 well as slice units.
5739 * A new bus call SetUnitProperties() has been added to alter
5740 various runtime parameters of a unit. This is primarily
5741 useful to alter cgroup parameters dynamically in a nice way,
5742 but will be extended later on to make more properties
5743 modifiable at runtime. systemctl gained a new set-properties
5744 command that wraps this call.
5746 * A new tool "systemd-run" has been added which can be used to
5747 run arbitrary command lines as transient services or scopes,
5748 while configuring a number of settings via the command
5749 line. This tool is currently very basic, however already
5750 very useful. We plan to extend this tool to even allow
5751 queuing of execution jobs with time triggers from the
5752 command line, similar in fashion to "at".
5754 * nspawn will now inform the user explicitly that kernels with
5755 audit enabled break containers, and suggest the user to turn
5758 * Support for detecting the IMA and AppArmor security
5759 frameworks with ConditionSecurity= has been added.
5761 * journalctl gained a new "-k" switch for showing only kernel
5762 messages, mimicking dmesg output; in addition to "--user"
5763 and "--system" switches for showing only user's own logs
5766 * systemd-delta can now show information about drop-in
5767 snippets extending unit files.
5769 * libsystemd-bus has been substantially updated but is still
5770 not available as public API.
5772 * systemd will now look for the "debug" argument on the kernel
5773 command line and enable debug logging, similar to what
5774 "systemd.log_level=debug" already did before.
5776 * "systemctl set-default", "systemctl get-default" has been
5777 added to configure the default.target symlink, which
5778 controls what to boot into by default.
5780 * "systemctl set-log-level" has been added as a convenient
5781 way to raise and lower systemd logging threshold.
5783 * "systemd-analyze plot" will now show the time the various
5784 generators needed for execution, as well as information
5785 about the unit file loading.
5787 * libsystemd-journal gained a new sd_journal_open_files() call
5788 for opening specific journal files. journactl also gained a
5789 new switch to expose this new functionality. Previously we
5790 only supported opening all files from a directory, or all
5791 files from the system, as opening individual files only is
5792 racy due to journal file rotation.
5794 * systemd gained the new DefaultEnvironment= setting in
5795 /etc/systemd/system.conf to set environment variables for
5798 * If a privileged process logs a journal message with the
5799 OBJECT_PID= field set, then journald will automatically
5800 augment this with additional OBJECT_UID=, OBJECT_GID=,
5801 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5802 system services want to log events about specific client
5803 processes. journactl/systemctl has been updated to make use
5804 of this information if all log messages regarding a specific
5807 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5808 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5809 Reisner, David Coppa, David King, David Strauss, Eelco
5810 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5811 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5812 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5813 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5814 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5815 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5816 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5817 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5818 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5819 Łukasz Stelmach, 장동준
5823 * The Python bindings gained some minimal support for the APIs
5824 exposed by libsystemd-logind.
5826 * ConditionSecurity= gained support for detecting SMACK. Since
5827 this condition already supports SELinux and AppArmor we only
5828 miss IMA for this. Patches welcome!
5830 Contributions from: Karol Lewandowski, Lennart Poettering,
5831 Zbigniew Jędrzejewski-Szmek
5835 * systemd-nspawn will now create /etc/resolv.conf if
5836 necessary, before bind-mounting the host's file onto it.
5838 * systemd-nspawn will now store meta information about a
5839 container on the container's cgroup as extended attribute
5840 fields, including the root directory.
5842 * The cgroup hierarchy has been reworked in many ways. All
5843 objects any of the components systemd creates in the cgroup
5844 tree are now suffixed. More specifically, user sessions are
5845 now placed in cgroups suffixed with ".session", users in
5846 cgroups suffixed with ".user", and nspawn containers in
5847 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5848 names are now escaped in a simple scheme to avoid collision
5849 of userspace object names with kernel filenames. This work
5850 is preparation for making these objects relocatable in the
5851 cgroup tree, in order to allow easy resource partitioning of
5852 these objects without causing naming conflicts.
5854 * systemctl list-dependencies gained the new switches
5855 --plain, --reverse, --after and --before.
5857 * systemd-inhibit now shows the process name of processes that
5858 have taken an inhibitor lock.
5860 * nss-myhostname will now also resolve "localhost"
5861 implicitly. This makes /etc/hosts an optional file and
5862 nicely handles that on IPv6 ::1 maps to both "localhost" and
5865 * libsystemd-logind.so gained a new call
5866 sd_get_machine_names() to enumerate running containers and
5867 VMs (currently only supported by very new libvirt and
5868 nspawn). sd_login_monitor can now be used to watch
5869 VMs/containers coming and going.
5871 * .include is not allowed recursively anymore, and only in
5872 unit files. Usually it is better to use drop-in snippets in
5873 .d/*.conf anyway, as introduced with systemd 198.
5875 * systemd-analyze gained a new "critical-chain" command that
5876 determines the slowest chain of units run during system
5877 boot-up. It is very useful for tracking down where
5878 optimizing boot time is the most beneficial.
5880 * systemd will no longer allow manipulating service paths in
5881 the name=systemd:/system cgroup tree using ControlGroup= in
5882 units. (But is still fine with it in all other dirs.)
5884 * There's a new systemd-nspawn@.service service file that may
5885 be used to easily run nspawn containers as system
5886 services. With the container's root directory in
5887 /var/lib/container/foobar it is now sufficient to run
5888 "systemctl start systemd-nspawn@foobar.service" to boot it.
5890 * systemd-cgls gained a new parameter "--machine" to list only
5891 the processes within a certain container.
5893 * ConditionSecurity= now can check for "apparmor". We still
5894 are lacking checks for SMACK and IMA for this condition
5895 check though. Patches welcome!
5897 * A new configuration file /etc/systemd/sleep.conf has been
5898 added that may be used to configure which kernel operation
5899 systemd is supposed to execute when "suspend", "hibernate"
5900 or "hybrid-sleep" is requested. This makes the new kernel
5901 "freeze" state accessible to the user.
5903 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5904 the passed argument if applicable.
5906 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5907 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5908 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5909 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5910 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5911 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5912 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5917 * The output of 'systemctl list-jobs' got some polishing. The
5918 '--type=' argument may now be passed more than once. A new
5919 command 'systemctl list-sockets' has been added which shows
5920 a list of kernel sockets systemd is listening on with the
5921 socket units they belong to, plus the units these socket
5924 * The experimental libsystemd-bus library got substantial
5925 updates to work in conjunction with the (also experimental)
5926 kdbus kernel project. It works well enough to exchange
5927 messages with some sophistication. Note that kdbus is not
5928 ready yet, and the library is mostly an elaborate test case
5929 for now, and not installable.
5931 * systemd gained a new unit 'systemd-static-nodes.service'
5932 that generates static device nodes earlier during boot, and
5933 can run in conjunction with udev.
5935 * libsystemd-login gained a new call sd_pid_get_user_unit()
5936 to retrieve the user systemd unit a process is running
5937 in. This is useful for systems where systemd is used as
5940 * systemd-nspawn now places all containers in the new /machine
5941 top-level cgroup directory in the name=systemd
5942 hierarchy. libvirt will soon do the same, so that we get a
5943 uniform separation of /system, /user and /machine for system
5944 services, user processes and containers/virtual
5945 machines. This new cgroup hierarchy is also useful to stick
5946 stable names to specific container instances, which can be
5947 recognized later this way (this name may be controlled
5948 via systemd-nspawn's new -M switch). libsystemd-login also
5949 gained a new call sd_pid_get_machine_name() to retrieve the
5950 name of the container/VM a specific process belongs to.
5952 * bootchart can now store its data in the journal.
5954 * libsystemd-journal gained a new call
5955 sd_journal_add_conjunction() for AND expressions to the
5956 matching logic. This can be used to express more complex
5957 logical expressions.
5959 * journactl can now take multiple --unit= and --user-unit=
5962 * The cryptsetup logic now understands the "luks.key=" kernel
5963 command line switch for specifying a file to read the
5964 decryption key from. Also, if a configured key file is not
5965 found the tool will now automatically fall back to prompting
5968 * Python systemd.journal module was updated to wrap recently
5969 added functions from libsystemd-journal. The interface was
5970 changed to bring the low level interface in s.j._Reader
5971 closer to the C API, and the high level interface in
5972 s.j.Reader was updated to wrap and convert all data about
5975 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5976 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5977 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5978 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5979 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5980 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5984 * journalctl --update-catalog now understands a new --root=
5985 option to operate on catalogs found in a different root
5988 * During shutdown after systemd has terminated all running
5989 services a final killing loop kills all remaining left-over
5990 processes. We will now print the name of these processes
5991 when we send SIGKILL to them, since this usually indicates a
5994 * If /etc/crypttab refers to password files stored on
5995 configured mount points automatic dependencies will now be
5996 generated to ensure the specific mount is established first
5997 before the key file is attempted to be read.
5999 * 'systemctl status' will now show information about the
6000 network sockets a socket unit is listening on.
6002 * 'systemctl status' will also shown information about any
6003 drop-in configuration file for units. (Drop-In configuration
6004 files in this context are files such as
6005 /etc/systemd/systemd/foobar.service.d/*.conf)
6007 * systemd-cgtop now optionally shows summed up CPU times of
6008 cgroups. Press '%' while running cgtop to switch between
6009 percentage and absolute mode. This is useful to determine
6010 which cgroups use up the most CPU time over the entire
6011 runtime of the system. systemd-cgtop has also been updated
6012 to be 'pipeable' for processing with further shell tools.
6014 * 'hostnamectl set-hostname' will now allow setting of FQDN
6017 * The formatting and parsing of time span values has been
6018 changed. The parser now understands fractional expressions
6019 such as "5.5h". The formatter will now output fractional
6020 expressions for all time spans under 1min, i.e. "5.123456s"
6021 rather than "5s 123ms 456us". For time spans under 1s
6022 millisecond values are shown, for those under 1ms
6023 microsecond values are shown. This should greatly improve
6024 all time-related output of systemd.
6026 * libsystemd-login and libsystemd-journal gained new
6027 functions for querying the poll() events mask and poll()
6028 timeout value for integration into arbitrary event
6031 * localectl gained the ability to list available X11 keymaps
6032 (models, layouts, variants, options).
6034 * 'systemd-analyze dot' gained the ability to filter for
6035 specific units via shell-style globs, to create smaller,
6036 more useful graphs. I.e. it is now possible to create simple
6037 graphs of all the dependencies between only target units, or
6038 of all units that Avahi has dependencies with.
6040 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6041 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6042 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6043 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6044 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6045 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6046 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6050 * The boot-time readahead implementation for rotating media
6051 will now read the read-ahead data in multiple passes which
6052 consist of all read requests made in equidistant time
6053 intervals. This means instead of strictly reading read-ahead
6054 data in its physical order on disk we now try to find a
6055 middle ground between physical and access time order.
6057 * /etc/os-release files gained a new BUILD_ID= field for usage
6058 on operating systems that provide continuous builds of OS
6061 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6062 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6063 William Douglas, Zbigniew Jędrzejewski-Szmek
6067 * systemd-python gained an API exposing libsystemd-daemon.
6069 * The SMACK setup logic gained support for uploading CIPSO
6072 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6073 ReadOnlyDirectories= and InaccessibleDirectories= has
6074 changed. The private /tmp and /var/tmp directories are now
6075 shared by all processes of a service (which means
6076 ExecStartPre= may now leave data in /tmp that ExecStart= of
6077 the same service can still access). When a service is
6078 stopped its temporary directories are immediately deleted
6079 (normal clean-up with tmpfiles is still done in addition to
6082 * By default, systemd will now set a couple of sysctl
6083 variables in the kernel: the safe sysrq options are turned
6084 on, IP route verification is turned on, and source routing
6085 disabled. The recently added hardlink and softlink
6086 protection of the kernel is turned on. These settings should
6087 be reasonably safe, and good defaults for all new systems.
6089 * The predictable network naming logic may now be turned off
6090 with a new kernel command line switch: net.ifnames=0.
6092 * A new libsystemd-bus module has been added that implements a
6093 pretty complete D-Bus client library. For details see:
6095 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6097 * journald will now explicitly flush the journal files to disk
6098 at the latest 5min after each write. The file will then also
6099 be marked offline until the next write. This should increase
6100 reliability in case of a crash. The synchronization delay
6101 can be configured via SyncIntervalSec= in journald.conf.
6103 * There's a new remote-fs-setup.target unit that can be used
6104 to pull in specific services when at least one remote file
6105 system is to be mounted.
6107 * There are new targets timers.target and paths.target as
6108 canonical targets to pull user timer and path units in
6109 from. This complements sockets.target with a similar
6110 purpose for socket units.
6112 * libudev gained a new call udev_device_set_attribute_value()
6113 to set sysfs attributes of a device.
6115 * The udev daemon now sets the default number of worker
6116 processes executed in parallel based on the number of available
6117 CPUs instead of the amount of available RAM. This is supposed
6118 to provide a more reliable default and limit a too aggressive
6119 parallelism for setups with 1000s of devices connected.
6121 Contributions from: Auke Kok, Colin Walters, Cristian
6122 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6123 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6124 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6125 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6126 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6127 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6128 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6129 Zbigniew Jędrzejewski-Szmek
6133 * Configuration of unit files may now be extended via drop-in
6134 files without having to edit/override the unit files
6135 themselves. More specifically, if the administrator wants to
6136 change one value for a service file foobar.service he can
6137 now do so by dropping in a configuration snippet into
6138 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6139 will load all these snippets and apply them on top of the
6140 main unit configuration file, possibly extending or
6141 overriding its settings. Using these drop-in snippets is
6142 generally nicer than the two earlier options for changing
6143 unit files locally: copying the files from
6144 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6145 them there; or creating a new file in /etc/systemd/system/
6146 that incorporates the original one via ".include". Drop-in
6147 snippets into these .d/ directories can be placed in any
6148 directory systemd looks for units in, and the usual
6149 overriding semantics between /usr/lib, /etc and /run apply
6152 * Most unit file settings which take lists of items can now be
6153 reset by assigning the empty string to them. For example,
6154 normally, settings such as Environment=FOO=BAR append a new
6155 environment variable assignment to the environment block,
6156 each time they are used. By assigning Environment= the empty
6157 string the environment block can be reset to empty. This is
6158 particularly useful with the .d/*.conf drop-in snippets
6159 mentioned above, since this adds the ability to reset list
6160 settings from vendor unit files via these drop-ins.
6162 * systemctl gained a new "list-dependencies" command for
6163 listing the dependencies of a unit recursively.
6165 * Inhibitors are now honored and listed by "systemctl
6166 suspend", "systemctl poweroff" (and similar) too, not only
6167 GNOME. These commands will also list active sessions by
6170 * Resource limits (as exposed by the various control group
6171 controllers) can now be controlled dynamically at runtime
6172 for all units. More specifically, you can now use a command
6173 like "systemctl set-cgroup-attr foobar.service cpu.shares
6174 2000" to alter the CPU shares a specific service gets. These
6175 settings are stored persistently on disk, and thus allow the
6176 administrator to easily adjust the resource usage of
6177 services with a few simple commands. This dynamic resource
6178 management logic is also available to other programs via the
6179 bus. Almost any kernel cgroup attribute and controller is
6182 * systemd-vconsole-setup will now copy all font settings to
6183 all allocated VTs, where it previously applied them only to
6186 * libsystemd-login gained the new sd_session_get_tty() API
6189 * This release drops support for a few legacy or
6190 distribution-specific LSB facility names when parsing init
6191 scripts: $x-display-manager, $mail-transfer-agent,
6192 $mail-transport-agent, $mail-transfer-agent, $smtp,
6193 $null. Also, the mail-transfer-agent.target unit backing
6194 this has been removed. Distributions which want to retain
6195 compatibility with this should carry the burden for
6196 supporting this themselves and patch support for these back
6197 in, if they really need to. Also, the facilities $syslog and
6198 $local_fs are now ignored, since systemd does not support
6199 early-boot LSB init scripts anymore, and these facilities
6200 are implied anyway for normal services. syslog.target has
6203 * There are new bus calls on PID1's Manager object for
6204 cancelling jobs, and removing snapshot units. Previously,
6205 both calls were only available on the Job and Snapshot
6208 * systemd-journal-gatewayd gained SSL support.
6210 * The various "environment" files, such as /etc/locale.conf
6211 now support continuation lines with a backslash ("\") as
6212 last character in the line, similarly in style (but different)
6213 to how this is supported in shells.
6215 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6216 now implicitly appended to every log entry logged. systemctl
6217 has been updated to filter by this field when operating on a
6218 user systemd instance.
6220 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6221 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6222 the container. This makes it easier to boot unmodified
6223 Fedora systems in a container, which however still requires
6224 audit=0 to be passed on the kernel command line. Auditing in
6225 kernel and userspace is unfortunately still too broken in
6226 context of containers, hence we recommend compiling it out
6227 of the kernel or using audit=0. Hopefully this will be fixed
6228 one day for good in the kernel.
6230 * nspawn gained the new --bind= and --bind-ro= parameters to
6231 bind mount specific directories from the host into the
6234 * nspawn will now mount its own devpts file system instance
6235 into the container, in order not to leak pty devices from
6236 the host into the container.
6238 * systemd will now read the firmware boot time performance
6239 information from the EFI variables, if the used boot loader
6240 supports this, and takes it into account for boot performance
6241 analysis via "systemd-analyze". This is currently supported
6242 only in conjunction with Gummiboot, but could be supported
6243 by other boot loaders too. For details see:
6245 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6247 * A new generator has been added that automatically mounts the
6248 EFI System Partition (ESP) to /boot, if that directory
6249 exists, is empty, and no other file system has been
6250 configured to be mounted there.
6252 * logind will now send out PrepareForSleep(false) out
6253 unconditionally, after coming back from suspend. This may be
6254 used by applications as asynchronous notification for
6255 system resume events.
6257 * "systemctl unlock-sessions" has been added, that allows
6258 unlocking the screens of all user sessions at once, similar
6259 to how "systemctl lock-sessions" already locked all users
6260 sessions. This is backed by a new D-Bus call UnlockSessions().
6262 * "loginctl seat-status" will now show the master device of a
6263 seat. (i.e. the device of a seat that needs to be around for
6264 the seat to be considered available, usually the graphics
6267 * tmpfiles gained a new "X" line type, that allows
6268 configuration of files and directories (with wildcards) that
6269 shall be excluded from automatic cleanup ("aging").
6271 * udev default rules set the device node permissions now only
6272 at "add" events, and do not change them any longer with a
6273 later "change" event.
6275 * The log messages for lid events and power/sleep keypresses
6276 now carry a message ID.
6278 * We now have a substantially larger unit test suite, but this
6279 continues to be work in progress.
6281 * udevadm hwdb gained a new --root= parameter to change the
6282 root directory to operate relative to.
6284 * logind will now issue a background sync() request to the kernel
6285 early at shutdown, so that dirty buffers are flushed to disk early
6286 instead of at the last moment, in order to optimize shutdown
6289 * A new bootctl tool has been added that is an interface for
6290 certain boot loader operations. This is currently a preview
6291 and is likely to be extended into a small mechanism daemon
6292 like timedated, localed, hostnamed, and can be used by
6293 graphical UIs to enumerate available boot options, and
6294 request boot into firmware operations.
6296 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6297 the rest of the package. It also has been updated to work
6298 correctly in initrds.
6300 * Policykit previously has been runtime optional, and is now
6301 also compile time optional via a configure switch.
6303 * systemd-analyze has been reimplemented in C. Also "systemctl
6304 dot" has moved into systemd-analyze.
6306 * "systemctl status" with no further parameters will now print
6307 the status of all active or failed units.
6309 * Operations such as "systemctl start" can now be executed
6310 with a new mode "--irreversible" which may be used to queue
6311 operations that cannot accidentally be reversed by a later
6312 job queuing. This is by default used to make shutdown
6313 requests more robust.
6315 * The Python API of systemd now gained a new module for
6316 reading journal files.
6318 * A new tool kernel-install has been added that can install
6319 kernel images according to the Boot Loader Specification:
6321 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6323 * Boot time console output has been improved to provide
6324 animated boot time output for hanging jobs.
6326 * A new tool systemd-activate has been added which can be used
6327 to test socket activation with, directly from the command
6328 line. This should make it much easier to test and debug
6329 socket activation in daemons.
6331 * journalctl gained a new "--reverse" (or -r) option to show
6332 journal output in reverse order (i.e. newest line first).
6334 * journalctl gained a new "--pager-end" (or -e) option to jump
6335 to immediately jump to the end of the journal in the
6336 pager. This is only supported in conjunction with "less".
6338 * journalctl gained a new "--user-unit=" option, that works
6339 similarly to "--unit=" but filters for user units rather than
6342 * A number of unit files to ease adoption of systemd in
6343 initrds has been added. This moves some minimal logic from
6344 the various initrd implementations into systemd proper.
6346 * The journal files are now owned by a new group
6347 "systemd-journal", which exists specifically to allow access
6348 to the journal, and nothing else. Previously, we used the
6349 "adm" group for that, which however possibly covers more
6350 than just journal/log file access. This new group is now
6351 already used by systemd-journal-gatewayd to ensure this
6352 daemon gets access to the journal files and as little else
6353 as possible. Note that "make install" will also set FS ACLs
6354 up for /var/log/journal to give "adm" and "wheel" read
6355 access to it, in addition to "systemd-journal" which owns
6356 the journal files. We recommend that packaging scripts also
6357 add read access to "adm" + "wheel" to /var/log/journal, and
6358 all existing/future journal files. To normal users and
6359 administrators little changes, however packagers need to
6360 ensure to create the "systemd-journal" system group at
6361 package installation time.
6363 * The systemd-journal-gatewayd now runs as unprivileged user
6364 systemd-journal-gateway:systemd-journal-gateway. Packaging
6365 scripts need to create these system user/group at
6368 * timedated now exposes a new boolean property CanNTP that
6369 indicates whether a local NTP service is available or not.
6371 * systemd-detect-virt will now also detect xen PVs
6373 * The pstore file system is now mounted by default, if it is
6376 * In addition to the SELinux and IMA policies we will now also
6377 load SMACK policies at early boot.
6379 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6380 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6381 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6382 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6383 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6384 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6385 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6386 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6387 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6388 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6389 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6390 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6391 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6392 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6396 * Timer units now support calendar time events in addition to
6397 monotonic time events. That means you can now trigger a unit
6398 based on a calendar time specification such as "Thu,Fri
6399 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6400 or fifth day of any month of the year 2013, given that it is
6401 a thursday or friday. This brings timer event support
6402 considerably closer to cron's capabilities. For details on
6403 the supported calendar time specification language see
6406 * udev now supports a number of different naming policies for
6407 network interfaces for predictable names, and a combination
6408 of these policies is now the default. Please see this wiki
6409 document for details:
6411 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6413 * Auke Kok's bootchart implementation has been added to the
6414 systemd tree. It is an optional component that can graph the
6415 boot in quite some detail. It is one of the best bootchart
6416 implementations around and minimal in its code and
6419 * nss-myhostname has been integrated into the systemd source
6420 tree. nss-myhostname guarantees that the local hostname
6421 always stays resolvable via NSS. It has been a weak
6422 requirement of systemd-hostnamed since a long time, and
6423 since its code is actually trivial we decided to just
6424 include it in systemd's source tree. It can be turned off
6425 with a configure switch.
6427 * The read-ahead logic is now capable of properly detecting
6428 whether a btrfs file system is on SSD or rotating media, in
6429 order to optimize the read-ahead scheme. Previously, it was
6430 only capable of detecting this on traditional file systems
6433 * In udev, additional device properties are now read from the
6434 IAB in addition to the OUI database. Also, Bluetooth company
6435 identities are attached to the devices as well.
6437 * In service files %U may be used as specifier that is
6438 replaced by the configured user name of the service.
6440 * nspawn may now be invoked without a controlling TTY. This
6441 makes it suitable for invocation as its own service. This
6442 may be used to set up a simple containerized server system
6443 using only core OS tools.
6445 * systemd and nspawn can now accept socket file descriptors
6446 when they are started for socket activation. This enables
6447 implementation of socket activated nspawn
6448 containers. i.e. think about autospawning an entire OS image
6449 when the first SSH or HTTP connection is received. We expect
6450 that similar functionality will also be added to libvirt-lxc
6453 * journalctl will now suppress ANSI color codes when
6454 presenting log data.
6456 * systemctl will no longer show control group information for
6457 a unit if the control group is empty anyway.
6459 * logind can now automatically suspend/hibernate/shutdown the
6462 * /etc/machine-info and hostnamed now also expose the chassis
6463 type of the system. This can be used to determine whether
6464 the local system is a laptop, desktop, handset or
6465 tablet. This information may either be configured by the
6466 user/vendor or is automatically determined from ACPI and DMI
6467 information if possible.
6469 * A number of PolicyKit actions are now bound together with
6470 "imply" rules. This should simplify creating UIs because
6471 many actions will now authenticate similar ones as well.
6473 * Unit files learnt a new condition ConditionACPower= which
6474 may be used to conditionalize a unit depending on whether an
6475 AC power source is connected or not, of whether the system
6476 is running on battery power.
6478 * systemctl gained a new "is-failed" verb that may be used in
6479 shell scripts and suchlike to check whether a specific unit
6480 is in the "failed" state.
6482 * The EnvironmentFile= setting in unit files now supports file
6483 globbing, and can hence be used to easily read a number of
6484 environment files at once.
6486 * systemd will no longer detect and recognize specific
6487 distributions. All distribution-specific #ifdeffery has been
6488 removed, systemd is now fully generic and
6489 distribution-agnostic. Effectively, not too much is lost as
6490 a lot of the code is still accessible via explicit configure
6491 switches. However, support for some distribution specific
6492 legacy configuration file formats has been dropped. We
6493 recommend distributions to simply adopt the configuration
6494 files everybody else uses now and convert the old
6495 configuration from packaging scripts. Most distributions
6496 already did that. If that's not possible or desirable,
6497 distributions are welcome to forward port the specific
6498 pieces of code locally from the git history.
6500 * When logging a message about a unit systemd will now always
6501 log the unit name in the message meta data.
6503 * localectl will now also discover system locale data that is
6504 not stored in locale archives, but directly unpacked.
6506 * logind will no longer unconditionally use framebuffer
6507 devices as seat masters, i.e. as devices that are required
6508 to be existing before a seat is considered preset. Instead,
6509 it will now look for all devices that are tagged as
6510 "seat-master" in udev. By default, framebuffer devices will
6511 be marked as such, but depending on local systems, other
6512 devices might be marked as well. This may be used to
6513 integrate graphics cards using closed source drivers (such
6514 as NVidia ones) more nicely into logind. Note however, that
6515 we recommend using the open source NVidia drivers instead,
6516 and no udev rules for the closed-source drivers will be
6517 shipped from us upstream.
6519 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6520 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6521 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6522 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6523 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6524 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6525 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6526 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6527 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6528 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6529 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6530 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6535 * udev gained support for loading additional device properties
6536 from an indexed database that is keyed by vendor/product IDs
6537 and similar device identifiers. For the beginning this
6538 "hwdb" is populated with data from the well-known PCI and
6539 USB database, but also includes PNP, ACPI and OID data. In
6540 the longer run this indexed database shall grow into
6541 becoming the one central database for non-essential
6542 userspace device metadata. Previously, data from the PCI/USB
6543 database was only attached to select devices, since the
6544 lookup was a relatively expensive operation due to O(n) time
6545 complexity (with n being the number of entries in the
6546 database). Since this is now O(1), we decided to add in this
6547 data for all devices where this is available, by
6548 default. Note that the indexed database needs to be rebuilt
6549 when new data files are installed. To achieve this you need
6550 to update your packaging scripts to invoke "udevadm hwdb
6551 --update" after installation of hwdb data files. For
6552 RPM-based distributions we introduced the new
6553 %udev_hwdb_update macro for this purpose.
6555 * The Journal gained support for the "Message Catalog", an
6556 indexed database to link up additional information with
6557 journal entries. For further details please check:
6559 https://www.freedesktop.org/wiki/Software/systemd/catalog
6561 The indexed message catalog database also needs to be
6562 rebuilt after installation of message catalog files. Use
6563 "journalctl --update-catalog" for this. For RPM-based
6564 distributions we introduced the %journal_catalog_update
6565 macro for this purpose.
6567 * The Python Journal bindings gained support for the standard
6568 Python logging framework.
6570 * The Journal API gained new functions for checking whether
6571 the underlying file system of a journal file is capable of
6572 properly reporting file change notifications, or whether
6573 applications that want to reflect journal changes "live"
6574 need to recheck journal files continuously in appropriate
6577 * It is now possible to set the "age" field for tmpfiles
6578 entries to 0, indicating that files matching this entry
6579 shall always be removed when the directories are cleaned up.
6581 * coredumpctl gained a new "gdb" verb which invokes gdb
6582 right-away on the selected coredump.
6584 * There's now support for "hybrid sleep" on kernels that
6585 support this, in addition to "suspend" and "hibernate". Use
6586 "systemctl hybrid-sleep" to make use of this.
6588 * logind's HandleSuspendKey= setting (and related settings)
6589 now gained support for a new "lock" setting to simply
6590 request the screen lock on all local sessions, instead of
6591 actually executing a suspend or hibernation.
6593 * systemd will now mount the EFI variables file system by
6596 * Socket units now gained support for configuration of the
6597 SMACK security label.
6599 * timedatectl will now output the time of the last and next
6600 daylight saving change.
6602 * We dropped support for various legacy and distro-specific
6603 concepts, such as insserv, early-boot SysV services
6604 (i.e. those for non-standard runlevels such as 'b' or 'S')
6605 or ArchLinux /etc/rc.conf support. We recommend the
6606 distributions who still need support this to either continue
6607 to maintain the necessary patches downstream, or find a
6608 different solution. (Talk to us if you have questions!)
6610 * Various systemd components will now bypass PolicyKit checks
6611 for root and otherwise handle properly if PolicyKit is not
6612 found to be around. This should fix most issues for
6613 PolicyKit-less systems. Quite frankly this should have been
6614 this way since day one. It is absolutely our intention to
6615 make systemd work fine on PolicyKit-less systems, and we
6616 consider it a bug if something does not work as it should if
6617 PolicyKit is not around.
6619 * For embedded systems it is now possible to build udev and
6620 systemd without blkid and/or kmod support.
6622 * "systemctl switch-root" is now capable of switching root
6623 more than once. I.e. in addition to transitions from the
6624 initrd to the host OS it is now possible to transition to
6625 further OS images from the host. This is useful to implement
6626 offline updating tools.
6628 * Various other additions have been made to the RPM macros
6629 shipped with systemd. Use %udev_rules_update() after
6630 installing new udev rules files. %_udevhwdbdir,
6631 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6632 %_sysctldir are now available which resolve to the right
6633 directories for packages to place various data files in.
6635 * journalctl gained the new --full switch (in addition to
6636 --all, to disable ellipsation for long messages.
6638 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6639 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6640 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6641 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6642 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6643 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6644 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6645 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6646 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6650 * journalctl gained new --since= and --until= switches to
6651 filter by time. It also now supports nice filtering for
6652 units via --unit=/-u.
6654 * Type=oneshot services may use ExecReload= and do the
6657 * The journal daemon now supports time-based rotation and
6658 vacuuming, in addition to the usual disk-space based
6661 * The journal will now index the available field values for
6662 each field name. This enables clients to show pretty drop
6663 downs of available match values when filtering. The bash
6664 completion of journalctl has been updated
6665 accordingly. journalctl gained a new switch -F to list all
6666 values a certain field takes in the journal database.
6668 * More service events are now written as structured messages
6669 to the journal, and made recognizable via message IDs.
6671 * The timedated, localed and hostnamed mini-services which
6672 previously only provided support for changing time, locale
6673 and hostname settings from graphical DEs such as GNOME now
6674 also have a minimal (but very useful) text-based client
6675 utility each. This is probably the nicest way to changing
6676 these settings from the command line now, especially since
6677 it lists available options and is fully integrated with bash
6680 * There's now a new tool "systemd-coredumpctl" to list and
6681 extract coredumps from the journal.
6683 * We now install a README each in /var/log/ and
6684 /etc/rc.d/init.d explaining where the system logs and init
6685 scripts went. This hopefully should help folks who go to
6686 that dirs and look into the otherwise now empty void and
6687 scratch their heads.
6689 * When user-services are invoked (by systemd --user) the
6690 $MANAGERPID env var is set to the PID of systemd.
6692 * SIGRTMIN+24 when sent to a --user instance will now result
6693 in immediate termination of systemd.
6695 * gatewayd received numerous feature additions such as a
6696 "follow" mode, for live syncing and filtering.
6698 * browse.html now allows filtering and showing detailed
6699 information on specific entries. Keyboard navigation and
6700 mouse screen support has been added.
6702 * gatewayd/journalctl now supports HTML5/JSON
6703 Server-Sent-Events as output.
6705 * The SysV init script compatibility logic will now
6706 heuristically determine whether a script supports the
6707 "reload" verb, and only then make this available as
6710 * "systemctl status --follow" has been removed, use "journalctl
6713 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6714 have been removed since they are hardly useful to be
6717 * And I'd like to take the opportunity to specifically mention
6718 Zbigniew for his great contributions. Zbigniew, you rock!
6720 Contributions from: Andrew Eikum, Christian Hesse, Colin
6721 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6722 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6723 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6724 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6725 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6726 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6730 * If /etc/vconsole.conf is non-existent or empty we will no
6731 longer load any console font or key map at boot by
6732 default. Instead the kernel defaults will be left
6733 intact. This is definitely the right thing to do, as no
6734 configuration should mean no configuration, and hard-coding
6735 font names that are different on all archs is probably a bad
6736 idea. Also, the kernel default key map and font should be
6737 good enough for most cases anyway, and mostly identical to
6738 the userspace fonts/key maps we previously overloaded them
6739 with. If distributions want to continue to default to a
6740 non-kernel font or key map they should ship a default
6741 /etc/vconsole.conf with the appropriate contents.
6743 Contributions from: Colin Walters, Daniel J Walsh, Dave
6744 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6745 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6749 * journalctl gained a new --cursor= switch to show entries
6750 starting from the specified location in the journal.
6752 * We now enforce a size limit on journal entry fields exported
6753 with "-o json" in journalctl. Fields larger than 4K will be
6754 assigned null. This can be turned off with --all.
6756 * An (optional) journal gateway daemon is now available as
6757 "systemd-journal-gatewayd.service". This service provides
6758 access to the journal via HTTP and JSON. This functionality
6759 will be used to implement live log synchronization in both
6760 pull and push modes, but has various other users too, such
6761 as easy log access for debugging of embedded devices. Right
6762 now it is already useful to retrieve the journal via HTTP:
6764 # systemctl start systemd-journal-gatewayd.service
6765 # wget http://localhost:19531/entries
6767 This will download the journal contents in a
6768 /var/log/messages compatible format. The same as JSON:
6770 # curl -H"Accept: application/json" http://localhost:19531/entries
6772 This service is also accessible via a web browser where a
6773 single static HTML5 app is served that uses the JSON logic
6774 to enable the user to do some basic browsing of the
6775 journal. This will be extended later on. Here's an example
6776 screenshot of this app in its current state:
6778 http://0pointer.de/public/journal-gatewayd
6780 Contributions from: Kay Sievers, Lennart Poettering, Robert
6781 Milasan, Tom Gundersen
6785 * The bash completion logic is now available for journalctl
6788 * We do not mount the "cpuset" controller anymore together with
6789 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6790 started if no parameters are assigned to it. "cpuset" hence
6791 broke code that assumed it could create "cpu" groups and
6794 * journalctl -f will now subscribe to terminal size changes,
6795 and line break accordingly.
6797 Contributions from: Dave Reisner, Kay Sievers, Lennart
6798 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6802 * nspawn will now create a symlink /etc/localtime in the
6803 container environment, copying the host's timezone
6804 setting. Previously this has been done via a bind mount, but
6805 since symlinks cannot be bind mounted this has now been
6806 changed to create/update the appropriate symlink.
6808 * journalctl -n's line number argument is now optional, and
6809 will default to 10 if omitted.
6811 * journald will now log the maximum size the journal files may
6812 take up on disk. This is particularly useful if the default
6813 built-in logic of determining this parameter from the file
6814 system size is used. Use "systemctl status
6815 systemd-journald.service" to see this information.
6817 * The multi-seat X wrapper tool has been stripped down. As X
6818 is now capable of enumerating graphics devices via udev in a
6819 seat-aware way the wrapper is not strictly necessary
6820 anymore. A stripped down temporary stop-gap is still shipped
6821 until the upstream display managers have been updated to
6822 fully support the new X logic. Expect this wrapper to be
6823 removed entirely in one of the next releases.
6825 * HandleSleepKey= in logind.conf has been split up into
6826 HandleSuspendKey= and HandleHibernateKey=. The old setting
6827 is not available anymore. X11 and the kernel are
6828 distinguishing between these keys and we should too. This
6829 also means the inhibition lock for these keys has been split
6832 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6833 Poettering, Lukas Nykryn, Václav Pavlín
6837 * Whenever a unit changes state we will now log this to the
6838 journal and show along the unit's own log output in
6841 * ConditionPathIsMountPoint= can now properly detect bind
6842 mount points too. (Previously, a bind mount of one file
6843 system to another place in the same file system could not be
6844 detected as mount, since they shared struct stat's st_dev
6847 * We will now mount the cgroup controllers cpu, cpuacct,
6848 cpuset and the controllers net_cls, net_prio together by
6851 * nspawn containers will now have a virtualized boot
6852 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6853 over with a randomized ID at container initialization). This
6854 has the effect of making "journalctl -b" do the right thing
6857 * The JSON output journal serialization has been updated not
6858 to generate "endless" list objects anymore, but rather one
6859 JSON object per line. This is more in line how most JSON
6860 parsers expect JSON objects. The new output mode
6861 "json-pretty" has been added to provide similar output, but
6862 neatly aligned for readability by humans.
6864 * We dropped all explicit sync() invocations in the shutdown
6865 code. The kernel does this implicitly anyway in the kernel
6866 reboot() syscall. halt(8)'s -n option is now a compatibility
6869 * We now support virtualized reboot() in containers, as
6870 supported by newer kernels. We will fall back to exit() if
6871 CAP_SYS_REBOOT is not available to the container. Also,
6872 nspawn makes use of this now and will actually reboot the
6873 container if the containerized OS asks for that.
6875 * journalctl will only show local log output by default
6876 now. Use --merge (-m) to show remote log output, too.
6878 * libsystemd-journal gained the new sd_journal_get_usage()
6879 call to determine the current disk usage of all journal
6880 files. This is exposed in the new "journalctl --disk-usage"
6883 * journald gained a new configuration setting SplitMode= in
6884 journald.conf which may be used to control how user journals
6885 are split off. See journald.conf(5) for details.
6887 * A new condition type ConditionFileNotEmpty= has been added.
6889 * tmpfiles' "w" lines now support file globbing, to write
6890 multiple files at once.
6892 * We added Python bindings for the journal submission
6893 APIs. More Python APIs for a number of selected APIs will
6894 likely follow. Note that we intend to add native bindings
6895 only for the Python language, as we consider it common
6896 enough to deserve bindings shipped within systemd. There are
6897 various projects outside of systemd that provide bindings
6898 for languages such as PHP or Lua.
6900 * Many conditions will now resolve specifiers such as %i. In
6901 addition, PathChanged= and related directives of .path units
6902 now support specifiers as well.
6904 * There's now a new RPM macro definition for the system preset
6907 * journald will now warn if it ca not forward a message to the
6908 syslog daemon because its socket is full.
6910 * timedated will no longer write or process /etc/timezone,
6911 except on Debian. As we do not support late mounted /usr
6912 anymore /etc/localtime always being a symlink is now safe,
6913 and hence the information in /etc/timezone is not necessary
6916 * logind will now always reserve one VT for a text getty (VT6
6917 by default). Previously if more than 6 X sessions where
6918 started they took up all the VTs with auto-spawned gettys,
6919 so that no text gettys were available anymore.
6921 * udev will now automatically inform the btrfs kernel logic
6922 about btrfs RAID components showing up. This should make
6923 simple hotplug based btrfs RAID assembly work.
6925 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6926 (but not for its children which will stay at the kernel
6927 default). This should allow setups with a lot more listening
6930 * systemd will now always pass the configured timezone to the
6931 kernel at boot. timedated will do the same when the timezone
6934 * logind's inhibition logic has been updated. By default,
6935 logind will now handle the lid switch, the power and sleep
6936 keys all the time, even in graphical sessions. If DEs want
6937 to handle these events on their own they should take the new
6938 handle-power-key, handle-sleep-key and handle-lid-switch
6939 inhibitors during their runtime. A simple way to achieve
6940 that is to invoke the DE wrapped in an invocation of:
6942 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6944 * Access to unit operations is now checked via SELinux taking
6945 the unit file label and client process label into account.
6947 * systemd will now notify the administrator in the journal
6948 when he over-mounts a non-empty directory.
6950 * There are new specifiers that are resolved in unit files,
6951 for the host name (%H), the machine ID (%m) and the boot ID
6954 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6955 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6956 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6957 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6958 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6959 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6960 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6964 * Support for reading structured kernel messages from
6965 /dev/kmsg has now been added and is enabled by default.
6967 * Support for reading kernel messages from /proc/kmsg has now
6968 been removed. If you want kernel messages in the journal
6969 make sure to run a recent kernel (>= 3.5) that supports
6970 reading structured messages from /dev/kmsg (see
6971 above). /proc/kmsg is now exclusive property of classic
6972 syslog daemons again.
6974 * The libudev API gained the new
6975 udev_device_new_from_device_id() call.
6977 * The logic for file system namespace (ReadOnlyDirectory=,
6978 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6979 require pivot_root() anymore. This means fewer temporary
6980 directories are created below /tmp for this feature.
6982 * nspawn containers will now see and receive all submounts
6983 made on the host OS below the root file system of the
6986 * Forward Secure Sealing is now supported for Journal files,
6987 which provide cryptographical sealing of journal files so
6988 that attackers cannot alter log history anymore without this
6989 being detectable. Lennart will soon post a blog story about
6990 this explaining it in more detail.
6992 * There are two new service settings RestartPreventExitStatus=
6993 and SuccessExitStatus= which allow configuration of exit
6994 status (exit code or signal) which will be excepted from the
6995 restart logic, resp. consider successful.
6997 * journalctl gained the new --verify switch that can be used
6998 to check the integrity of the structure of journal files and
6999 (if Forward Secure Sealing is enabled) the contents of
7002 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7003 and similar symlinks pre-created. This makes running shells
7004 as container init process a lot more fun.
7006 * The fstab support can now handle PARTUUID= and PARTLABEL=
7009 * A new ConditionHost= condition has been added to match
7010 against the hostname (with globs) and machine ID. This is
7011 useful for clusters where a single OS image is used to
7012 provision a large number of hosts which shall run slightly
7013 different sets of services.
7015 * Services which hit the restart limit will now be placed in a
7018 Contributions from: Bertram Poettering, Dave Reisner, Huang
7019 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7020 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7024 * When running in --user mode systemd will now become a
7025 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7026 tree a lot more organized.
7028 * A new PartOf= unit dependency type has been introduced that
7029 may be used to group services in a natural way.
7031 * "systemctl enable" may now be used to enable instances of
7034 * journalctl now prints error log levels in red, and
7035 warning/notice log levels in bright white. It also supports
7036 filtering by log level now.
7038 * cgtop gained a new -n switch (similar to top), to configure
7039 the maximum number of iterations to run for. It also gained
7040 -b, to run in batch mode (accepting no input).
7042 * The suffix ".service" may now be omitted on most systemctl
7043 command lines involving service unit names.
7045 * There's a new bus call in logind to lock all sessions, as
7046 well as a loginctl verb for it "lock-sessions".
7048 * libsystemd-logind.so gained a new call sd_journal_perror()
7049 that works similar to libc perror() but logs to the journal
7050 and encodes structured information about the error number.
7052 * /etc/crypttab entries now understand the new keyfile-size=
7055 * shutdown(8) now can send a (configurable) wall message when
7056 a shutdown is cancelled.
7058 * The mount propagation mode for the root file system will now
7059 default to "shared", which is useful to make containers work
7060 nicely out-of-the-box so that they receive new mounts from
7061 the host. This can be undone locally by running "mount
7062 --make-rprivate /" if needed.
7064 * The prefdm.service file has been removed. Distributions
7065 should maintain this unit downstream if they intend to keep
7066 it around. However, we recommend writing normal unit files
7067 for display managers instead.
7069 * Since systemd is a crucial part of the OS we will now
7070 default to a number of compiler switches that improve
7071 security (hardening) such as read-only relocations, stack
7072 protection, and suchlike.
7074 * The TimeoutSec= setting for services is now split into
7075 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7076 of individual time outs for the start and the stop phase of
7079 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7080 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7081 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7082 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7083 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7084 Gundersen, Zbigniew Jędrzejewski-Szmek
7088 * The journal and id128 C APIs are now fully documented as man
7091 * Extra safety checks have been added when transitioning from
7092 the initial RAM disk to the main system to avoid accidental
7095 * /etc/crypttab entries now understand the new keyfile-offset=
7098 * systemctl -t can now be used to filter by unit load state.
7100 * The journal C API gained the new sd_journal_wait() call to
7101 make writing synchronous journal clients easier.
7103 * journalctl gained the new -D switch to show journals from a
7106 * journalctl now displays a special marker between log
7107 messages of two different boots.
7109 * The journal is now explicitly flushed to /var via a service
7110 systemd-journal-flush.service, rather than implicitly simply
7111 by seeing /var/log/journal to be writable.
7113 * journalctl (and the journal C APIs) can now match for much
7114 more complex expressions, with alternatives and
7117 * When transitioning from the initial RAM disk to the main
7118 system we will now kill all processes in a killing spree to
7119 ensure no processes stay around by accident.
7121 * Three new specifiers may be used in unit files: %u, %h, %s
7122 resolve to the user name, user home directory resp. user
7123 shell. This is useful for running systemd user instances.
7125 * We now automatically rotate journal files if their data
7126 object hash table gets a fill level > 75%. We also size the
7127 hash table based on the configured maximum file size. This
7128 together should lower hash collisions drastically and thus
7129 speed things up a bit.
7131 * journalctl gained the new "--header" switch to introspect
7132 header data of journal files.
7134 * A new setting SystemCallFilters= has been added to services
7135 which may be used to apply blacklists or whitelists to
7136 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7138 * nspawn gained a new --link-journal= switch (and quicker: -j)
7139 to link the container journal with the host. This makes it
7140 very easy to centralize log viewing on the host for all
7141 guests while still keeping the journal files separated.
7143 * Many bugfixes and optimizations
7145 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7146 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7147 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7152 * Several tools now understand kernel command line arguments,
7153 which are only read when run in an initial RAM disk. They
7154 usually follow closely their normal counterparts, but are
7157 * There's a new tool to analyze the readahead files that are
7158 automatically generated at boot. Use:
7160 /usr/lib/systemd/systemd-readahead analyze /.readahead
7162 * We now provide an early debug shell on tty9 if this enabled. Use:
7164 systemctl enable debug-shell.service
7166 * All plymouth related units have been moved into the Plymouth
7167 package. Please make sure to upgrade your Plymouth version
7170 * systemd-tmpfiles now supports getting passed the basename of
7171 a configuration file only, in which case it will look for it
7172 in all appropriate directories automatically.
7174 * udevadm info now takes a /dev or /sys path as argument, and
7175 does the right thing. Example:
7177 udevadm info /dev/sda
7178 udevadm info /sys/class/block/sda
7180 * systemctl now prints a warning if a unit is stopped but a
7181 unit that might trigger it continues to run. Example: a
7182 service is stopped but the socket that activates it is left
7185 * "systemctl status" will now mention if the log output was
7186 shortened due to rotation since a service has been started.
7188 * The journal API now exposes functions to determine the
7189 "cutoff" times due to rotation.
7191 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7192 immediately flushing of runtime logs to /var if possible,
7193 resp. for triggering immediate rotation of the journal
7196 * It is now considered an error if a service is attempted to
7197 be stopped that is not loaded.
7199 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7201 * systemd-analyze now supports Python 3
7203 * tmpfiles now supports cleaning up directories via aging
7204 where the first level dirs are always kept around but
7205 directories beneath it automatically aged. This is enabled
7206 by prefixing the age field with '~'.
7208 * Seat objects now expose CanGraphical, CanTTY properties
7209 which is required to deal with very fast bootups where the
7210 display manager might be running before the graphics drivers
7211 completed initialization.
7213 * Seat objects now expose a State property.
7215 * We now include RPM macros for service enabling/disabling
7216 based on the preset logic. We recommend RPM based
7217 distributions to make use of these macros if possible. This
7218 makes it simpler to reuse RPM spec files across
7221 * We now make sure that the collected systemd unit name is
7222 always valid when services log to the journal via
7225 * There's a new man page kernel-command-line(7) detailing all
7226 command line options we understand.
7228 * The fstab generator may now be disabled at boot by passing
7229 fstab=0 on the kernel command line.
7231 * A new kernel command line option modules-load= is now understood
7232 to load a specific kernel module statically, early at boot.
7234 * Unit names specified on the systemctl command line are now
7235 automatically escaped as needed. Also, if file system or
7236 device paths are specified they are automatically turned
7237 into the appropriate mount or device unit names. Example:
7239 systemctl status /home
7240 systemctl status /dev/sda
7242 * The SysVConsole= configuration option has been removed from
7243 system.conf parsing.
7245 * The SysV search path is no longer exported on the D-Bus
7248 * The Names= option has been removed from unit file parsing.
7250 * There's a new man page bootup(7) detailing the boot process.
7252 * Every unit and every generator we ship with systemd now
7253 comes with full documentation. The self-explanatory boot is
7256 * A couple of services gained "systemd-" prefixes in their
7257 name if they wrap systemd code, rather than only external
7258 code. Among them fsck@.service which is now
7259 systemd-fsck@.service.
7261 * The HaveWatchdog property has been removed from the D-Bus
7264 * systemd.confirm_spawn= on the kernel command line should now
7267 * There's a new man page crypttab(5) which details all options
7268 we actually understand.
7270 * systemd-nspawn gained a new --capability= switch to pass
7271 additional capabilities to the container.
7273 * timedated will now read known NTP implementation unit names
7274 from /usr/lib/systemd/ntp-units.d/*.list,
7275 systemd-timedated-ntp.target has been removed.
7277 * journalctl gained a new switch "-b" that lists log data of
7278 the current boot only.
7280 * The notify socket is in the abstract namespace again, in
7281 order to support daemons which chroot() at start-up.
7283 * There is a new Storage= configuration option for journald
7284 which allows configuration of where log data should go. This
7285 also provides a way to disable journal logging entirely, so
7286 that data collected is only forwarded to the console, the
7287 kernel log buffer or another syslog implementation.
7289 * Many bugfixes and optimizations
7291 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7292 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7293 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7294 Shawn Landden, Tom Gundersen
7298 * "systemctl help <unit>" now shows the man page if one is
7301 * Several new man pages have been added.
7303 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7304 MaxLevelConsole= can now be specified in
7305 journald.conf. These options allow reducing the amount of
7306 data stored on disk or forwarded by the log level.
7308 * TimerSlackNSec= can now be specified in system.conf for
7309 PID1. This allows system-wide power savings.
7311 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7312 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7317 * logind is now capable of (optionally) handling power and
7318 sleep keys as well as the lid switch.
7320 * journalctl now understands the syntax "journalctl
7321 /usr/bin/avahi-daemon" to get all log output of a specific
7324 * CapabilityBoundingSet= in system.conf now also influences
7325 the capability bound set of usermode helpers of the kernel.
7327 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7328 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7329 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7330 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7334 * Note that we skipped 139 releases here in order to set the
7335 new version to something that is greater than both udev's
7336 and systemd's most recent version number.
7338 * udev: all udev sources are merged into the systemd source tree now.
7339 All future udev development will happen in the systemd tree. It
7340 is still fully supported to use the udev daemon and tools without
7341 systemd running, like in initramfs or other init systems. Building
7342 udev though, will require the *build* of the systemd tree, but
7343 udev can be properly *run* without systemd.
7345 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7346 should be used to create dead device nodes as workarounds for broken
7349 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7350 no longer supported. udev_monitor_new_from_netlink() needs to be
7351 used to subscribe to events.
7353 * udev: when udevd is started by systemd, processes which are left
7354 behind by forking them off of udev rules, are unconditionally cleaned
7355 up and killed now after the event handling has finished. Services or
7356 daemons must be started as systemd services. Services can be
7357 pulled-in by udev to get started, but they can no longer be directly
7358 forked by udev rules.
7360 * udev: the daemon binary is called systemd-udevd now and installed
7361 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7362 to adapt to that, create symlink, or rename the binary after building
7365 * libudev no longer provides these symbols:
7366 udev_monitor_from_socket()
7367 udev_queue_get_failed_list_entry()
7368 udev_get_{dev,sys,run}_path()
7369 The versions number was bumped and symbol versioning introduced.
7371 * systemd-loginctl and systemd-journalctl have been renamed
7372 to loginctl and journalctl to match systemctl.
7374 * The config files: /etc/systemd/systemd-logind.conf and
7375 /etc/systemd/systemd-journald.conf have been renamed to
7376 logind.conf and journald.conf. Package updates should rename
7377 the files to the new names on upgrade.
7379 * For almost all files the license is now LGPL2.1+, changed
7380 from the previous GPL2.0+. Exceptions are some minor stuff
7381 of udev (which will be changed to LGPL2.1 eventually, too),
7382 and the MIT licensed sd-daemon.[ch] library that is suitable
7383 to be used as drop-in files.
7385 * systemd and logind now handle system sleep states, in
7386 particular suspending and hibernating.
7388 * logind now implements a sleep/shutdown/idle inhibiting logic
7389 suitable for a variety of uses. Soonishly Lennart will blog
7390 about this in more detail.
7392 * var-run.mount and var-lock.mount are no longer provided
7393 (which previously bind mounted these directories to their new
7394 places). Distributions which have not converted these
7395 directories to symlinks should consider stealing these files
7396 from git history and add them downstream.
7398 * We introduced the Documentation= field for units and added
7399 this to all our shipped units. This is useful to make it
7400 easier to explore the boot and the purpose of the various
7403 * All smaller setup units (such as
7404 systemd-vconsole-setup.service) now detect properly if they
7405 are run in a container and are skipped when
7406 appropriate. This guarantees an entirely noise-free boot in
7407 Linux container environments such as systemd-nspawn.
7409 * A framework for implementing offline system updates is now
7410 integrated, for details see:
7411 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7413 * A new service type Type=idle is available now which helps us
7414 avoiding ugly interleaving of getty output and boot status
7417 * There's now a system-wide CapabilityBoundingSet= option to
7418 globally reduce the set of capabilities for the
7419 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7420 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7421 even CAP_NET_ADMIN system-wide for secure systems.
7423 * There are now system-wide DefaultLimitXXX= options to
7424 globally change the defaults of the various resource limits
7425 for all units started by PID 1.
7427 * Harald Hoyer's systemd test suite has been integrated into
7428 systemd which allows easy testing of systemd builds in qemu
7429 and nspawn. (This is really awesome! Ask us for details!)
7431 * The fstab parser is now implemented as generator, not inside
7434 * systemctl will now warn you if .mount units generated from
7435 /etc/fstab are out of date due to changes in fstab that
7436 have not been read by systemd yet.
7438 * systemd is now suitable for usage in initrds. Dracut has
7439 already been updated to make use of this. With this in place
7440 initrds get a slight bit faster but primarily are much
7441 easier to introspect and debug since "systemctl status" in
7442 the host system can be used to introspect initrd services,
7443 and the journal from the initrd is kept around too.
7445 * systemd-delta has been added, a tool to explore differences
7446 between user/admin configuration and vendor defaults.
7448 * PrivateTmp= now affects both /tmp and /var/tmp.
7450 * Boot time status messages are now much prettier and feature
7451 proper english language. Booting up systemd has never been
7454 * Read-ahead pack files now include the inode number of all
7455 files to pre-cache. When the inode changes the pre-caching
7456 is not attempted. This should be nicer to deal with updated
7457 packages which might result in changes of read-ahead
7460 * We now temporaritly lower the kernel's read_ahead_kb variable
7461 when collecting read-ahead data to ensure the kernel's
7462 built-in read-ahead does not add noise to our measurements
7463 of necessary blocks to pre-cache.
7465 * There's now RequiresMountsFor= to add automatic dependencies
7466 for all mounts necessary for a specific file system path.
7468 * MountAuto= and SwapAuto= have been removed from
7469 system.conf. Mounting file systems at boot has to take place
7472 * nspawn now learned a new switch --uuid= to set the machine
7473 ID on the command line.
7475 * nspawn now learned the -b switch to automatically search
7478 * vt102 is now the default TERM for serial TTYs, upgraded from
7481 * systemd-logind now works on VT-less systems.
7483 * The build tree has been reorganized. The individual
7484 components now have directories of their own.
7486 * A new condition type ConditionPathIsReadWrite= is now available.
7488 * nspawn learned the new -C switch to create cgroups for the
7489 container in other hierarchies.
7491 * We now have support for hardware watchdogs, configurable in
7494 * The scheduled shutdown logic now has a public API.
7496 * We now mount /tmp as tmpfs by default, but this can be
7497 masked and /etc/fstab can override it.
7499 * Since udisks does not make use of /media anymore we are not
7500 mounting a tmpfs on it anymore.
7502 * journalctl gained a new --local switch to only interleave
7503 locally generated journal files.
7505 * We can now load the IMA policy at boot automatically.
7507 * The GTK tools have been split off into a systemd-ui.
7509 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7510 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7511 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7512 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7513 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7514 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7515 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7516 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7517 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7522 * This is mostly a bugfix release
7524 * Support optional initialization of the machine ID from the
7525 KVM or container configured UUID.
7527 * Support immediate reboots with "systemctl reboot -ff"
7529 * Show /etc/os-release data in systemd-analyze output
7531 * Many bugfixes for the journal, including endianness fixes and
7532 ensuring that disk space enforcement works
7534 * sd-login.h is C++ compatible again
7536 * Extend the /etc/os-release format on request of the Debian
7539 * We now refuse non-UTF8 strings used in various configuration
7540 and unit files. This is done to ensure we do not pass invalid
7541 data over D-Bus or expose it elsewhere.
7543 * Register Mimo USB Screens as suitable for automatic seat
7546 * Read SELinux client context from journal clients in a race
7549 * Reorder configuration file lookup order. /etc now always
7550 overrides /run in order to allow the administrator to always
7551 and unconditionally override vendor-supplied or
7552 automatically generated data.
7554 * The various user visible bits of the journal now have man
7555 pages. We still lack man pages for the journal API calls
7558 * We now ship all man pages in HTML format again in the
7561 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7562 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7563 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7568 * This is mostly a bugfix release
7570 * systems lacking /etc/os-release are no longer supported.
7572 * Various functionality updates to libsystemd-login.so
7574 * Track class of PAM logins to distinguish greeters from
7577 Contributions from: Kay Sievers, Lennart Poettering, Michael
7582 * This is an important bugfix release for v41.
7584 * Building man pages is now optional which should be useful
7585 for those building systemd from git but unwilling to install
7588 * Watchdog support for supervising services is now usable. In
7589 a future release support for hardware watchdogs
7590 (i.e. /dev/watchdog) will be added building on this.
7592 * Service start rate limiting is now configurable and can be
7593 turned off per service. When a start rate limit is hit a
7594 reboot can automatically be triggered.
7596 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7598 Contributions from: Benjamin Franzke, Bill Nottingham,
7599 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7600 Schmidt, Michał Górny, Piotr Drąg
7604 * The systemd binary is installed /usr/lib/systemd/systemd now;
7605 An existing /sbin/init symlink needs to be adapted with the
7608 * The code that loads kernel modules has been ported to invoke
7609 libkmod directly, instead of modprobe. This means we do not
7610 support systems with module-init-tools anymore.
7612 * Watchdog support is now already useful, but still not
7615 * A new kernel command line option systemd.setenv= is
7616 understood to set system wide environment variables
7617 dynamically at boot.
7619 * We now limit the set of capabilities of systemd-journald.
7621 * We now set SIGPIPE to ignore by default, since it only is
7622 useful in shell pipelines, and has little use in general
7623 code. This can be disabled with IgnoreSIPIPE=no in unit
7626 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7627 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7632 * This is mostly a bugfix release
7634 * We now expose the reason why a service failed in the
7635 "Result" D-Bus property.
7637 * Rudimentary service watchdog support (will be completed over
7638 the next few releases.)
7640 * When systemd forks off in order execute some service we will
7641 now immediately changes its argv[0] to reflect which process
7642 it will execute. This is useful to minimize the time window
7643 with a generic argv[0], which makes bootcharts more useful
7645 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7646 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7647 Mike Kazantsev, Ray Strode
7651 * This is mostly a test release, but incorporates many
7654 * New systemd-cgtop tool to show control groups by their
7657 * Linking against libacl for ACLs is optional again. If
7658 disabled, support tracking device access for active logins
7659 goes becomes unavailable, and so does access to the user
7660 journals by the respective users.
7662 * If a group "adm" exists, journal files are automatically
7663 owned by them, thus allow members of this group full access
7664 to the system journal as well as all user journals.
7666 * The journal now stores the SELinux context of the logging
7667 client for all entries.
7669 * Add C++ inclusion guards to all public headers
7671 * New output mode "cat" in the journal to print only text
7672 messages, without any meta data like date or time.
7674 * Include tiny X server wrapper as a temporary stop-gap to
7675 teach XOrg udev display enumeration. This is used by display
7676 managers such as gdm, and will go away as soon as XOrg
7677 learned native udev hotplugging for display devices.
7679 * Add new systemd-cat tool for executing arbitrary programs
7680 with STDERR/STDOUT connected to the journal. Can also act as
7681 BSD logger replacement, and does so by default.
7683 * Optionally store all locally generated coredumps in the
7684 journal along with meta data.
7686 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7687 writing short strings to files (for usage for /sys), and for
7688 creating symlinks, character and block device nodes.
7690 * New unit file option ControlGroupPersistent= to make cgroups
7691 persistent, following the mechanisms outlined in
7692 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7694 * Support multiple local RTCs in a sane way
7696 * No longer monopolize IO when replaying readahead data on
7697 rotating disks, since we might starve non-file-system IO to
7698 death, since fanotify() will not see accesses done by blkid,
7701 * Do not show kernel threads in systemd-cgls anymore, unless
7702 requested with new -k switch.
7704 Contributions from: Dan Horák, Kay Sievers, Lennart
7705 Poettering, Michal Schmidt
7709 * This is mostly a test release, but incorporates many
7712 * The git repository moved to:
7713 git://anongit.freedesktop.org/systemd/systemd
7714 ssh://git.freedesktop.org/git/systemd/systemd
7716 * First release with the journal
7717 http://0pointer.de/blog/projects/the-journal.html
7719 * The journal replaces both systemd-kmsg-syslogd and
7720 systemd-stdout-bridge.
7722 * New sd_pid_get_unit() API call in libsystemd-logind
7724 * Many systemadm clean-ups
7726 * Introduce remote-fs-pre.target which is ordered before all
7727 remote mounts and may be used to start services before all
7730 * Added Mageia support
7732 * Add bash completion for systemd-loginctl
7734 * Actively monitor PID file creation for daemons which exit in
7735 the parent process before having finished writing the PID
7736 file in the daemon process. Daemons which do this need to be
7737 fixed (i.e. PID file creation must have finished before the
7738 parent exits), but we now react a bit more gracefully to them.
7740 * Add colourful boot output, mimicking the well-known output
7741 of existing distributions.
7743 * New option PassCredentials= for socket units, for
7744 compatibility with a recent kernel ABI breakage.
7746 * /etc/rc.local is now hooked in via a generator binary, and
7747 thus will no longer act as synchronization point during
7750 * systemctl list-unit-files now supports --root=.
7752 * systemd-tmpfiles now understands two new commands: z, Z for
7753 relabelling files according to the SELinux database. This is
7754 useful to apply SELinux labels to specific files in /sys,
7757 * Output of SysV services is now forwarded to both the console
7758 and the journal by default, not only just the console.
7760 * New man pages for all APIs from libsystemd-login.
7762 * The build tree got reorganized and the build system is a
7763 lot more modular allowing embedded setups to specifically
7764 select the components of systemd they are interested in.
7766 * Support for Linux systems lacking the kernel VT subsystem is
7769 * configure's --with-rootdir= got renamed to
7770 --with-rootprefix= to follow the naming used by udev and
7773 * Unless specified otherwise we will now install to /usr instead
7774 of /usr/local by default.
7776 * Processes with '@' in argv[0][0] are now excluded from the
7777 final shut-down killing spree, following the logic explained
7779 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7781 * All processes remaining in a service cgroup when we enter
7782 the START or START_PRE states are now killed with
7783 SIGKILL. That means it is no longer possible to spawn
7784 background processes from ExecStart= lines (which was never
7785 supported anyway, and bad style).
7787 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7788 reloading of units together.
7790 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7791 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7792 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7793 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7794 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek