1 systemd System and Service Manager
5 * A new modprobe.d drop-in is now shipped by default that sets the
6 bonding module option max_bonds=0. This overrides the kernel default,
7 to avoid conflicts and ambiguity as to whether or not bond0 should be
8 managed by systemd-networkd or not. This resolves multiple issues
9 with bond0 properties not being applied, when bond0 is configured
10 with systemd-networkd. Distributors may choose to not package this,
11 however in that case users will be prevented from correctly managing
12 bond0 interface using systemd-networkd.
14 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
15 which print the logging level and target of the system manager. They
16 complement the existing "set-log-level" and "set-log-target" verbs
17 used to change those values.
19 * journald.conf gained a new boolean setting ReadKMsg= which defaults
20 to on. If turned off kernel log messages will not be read by
21 systemd-journald or included in the logs. It also gained a new
22 setting LineMax= for configuring the maximum line length in
23 STDOUT/STDERR log streams. The new default for this value is 48K, up
24 from the previous hardcoded 2048.
26 * A new unit setting RuntimeDirectoryPreserve= has been added, which
27 allows more detailed control of what to do with a runtime directory
28 configured with RuntimeDirectory= (i.e. a directory below /run or
29 $XDG_RUNTIME_DIR) after a unit is stopped.
31 * The RuntimeDirectory= setting for units gained support for creating
32 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
33 one top-level directory.
35 * Units gained new options StateDirectory=, CacheDirectory=,
36 LogsDirectory= and ConfigurationDirectory= which are closely related
37 to RuntimeDirectory= but manage per-service directories below
38 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
39 possible to write unit files which when activated automatically gain
40 properly owned service specific directories in these locations, thus
41 making unit files self-contained and increasing compatibility with
42 stateless systems and factory reset where /etc or /var are
43 unpopulated at boot. Matching these new settings there's also
44 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
45 ConfigurationDirectoryMode= for configuring the access mode of these
46 directories. These settings are particularly useful in combination
47 with DynamicUser=yes as they provide secure, properly-owned,
48 writable, and stateful locations for storage, excluded from the
49 sandbox that such services live in otherwise.
51 * Automake support has been removed from this release. systemd is now
54 * systemd-journald will now aggressively cache client metadata during
55 runtime, speeding up log write performance under pressure. This comes
56 at a small price though: as much of the metadata is read
57 asynchronously from /proc/ (and isn't implicitly attached to log
58 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
59 metadata stored alongside a log entry might be slightly
60 out-of-date. Previously it could only be slightly newer than the log
61 message. The time window is small however, and given that the kernel
62 is unlikely to be improved anytime soon in this regard, this appears
65 * nss-myhostname/systemd-resolved will now by default synthesize an
66 A/AAAA resource record for the "_gateway" hostname, pointing to the
67 current default IP gateway. Previously it did that for the "gateway"
68 name, hampering adoption, as some distributions wanted to leave that
69 host name open for local use. The old behaviour may still be
70 requested at build time.
72 * systemd-networkd's [Address] section in .network files gained a new
73 Scope= setting for configuring the IP address scope. The [Network]
74 section gained a new boolean setting ConfigureWithoutCarrier= that
75 tells systemd-networkd to ignore link sensing when configuring the
76 device. The [DHCP] section gained a new Anonymize= boolean option for
77 turning on a number of options suggested in RFC 7844. A new
78 [RoutingPolicyRule] section has been added for configuring the IP
79 routing policy. The [Route] section has gained support for a new
80 Type= setting which permits configuring
81 blackhole/unreachable/prohibit routes.
83 * The [VRF] section in .netdev files gained a new Table= setting for
84 configuring the routing table to use. The [Tunnel] section gained a
85 new Independent= boolean field for configuring tunnels independent of
86 an underlying network interface. The [Bridge] section gained a new
87 GroupForwardMask= option for configuration of propagation of link
88 local frames between bridge ports.
90 * The WakeOnLan= setting in .link files gained support for a number of
91 new modes. A new TCP6SegmentationOffload= setting has been added for
92 configuring TCP/IPv6 hardware segmentation offload.
94 * The IPv6 RA sender implementation may now optionally send out RDNSS
95 and RDNSSL records to supply DNS configuration to peers.
97 * systemd-nspawn gained support for a new --system-call-filter= command
98 line option for adding and removing entries in the default system
99 call filter it applies. Moreover systemd-nspawn has been changed to
100 implement a system call whitelist instead of a blacklist.
102 * systemd-run gained support for a new --pipe command line option. If
103 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
104 are directly passed on to the activated transient service
105 executable. This allows invoking arbitrary processes as systemd
106 services (for example to take benefit of dependency management,
107 accounting management, resource management or log management that is
108 done automatically for services) — while still allowing them to be
109 integrated in a classic UNIX shell pipeline.
111 * When a service sends RELOAD=1 via sd_notify() and reload propagation
112 using ReloadPropagationTo= is configured, a reload is now propagated
113 to configured units. (Previously this was only done on explicitly
114 requested reloads, using "systemctl reload" or an equivalent
117 * For each service unit a restart counter is now kept: it is increased
118 each time the service is restarted due to Restart=, and may be
119 queried using "systemctl show -p NRestarts …".
121 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
122 @signal and @timer have been added, for usage with SystemCallFilter=
123 in unit files and the new --system-call-filter= command line option
124 of systemd-nspawn (see above).
126 * ExecStart= lines in unit files gained two new modifiers: when a
127 command line is prefixed with "!" the command will be executed as
128 configured, except for the credentials applied by
129 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
130 "+", but does still apply namespacing options unlike "+". There's
131 also "!!" now, which is mostly identical, but becomes a NOP on
132 systems that support ambient capabilities. This is useful to write
133 unit files that work with ambient capabilities where possible but
134 automatically fall back to traditional privilege dropping mechanisms
135 on systems where this is not supported.
137 * ListenNetlink= settings in socket units now support RDMA netlink
140 * A new unit file setting LockPersonality= has been added which permits
141 locking down the chosen execution domain ("personality") of a service
144 * A new special target "getty-pre.target" has been added, which is
145 ordered before all text logins, and may be used to order services
146 before textual logins acquire access to the console.
148 * systemd will now attempt to load the virtio-rng.ko kernel module very
149 early on if a VM environment supporting this is detected. This should
150 improve entropy during early boot in virtualized environments.
152 * A _netdev option is now supported in /etc/crypttab that operates in a
153 similar way as the same option in /etc/fstab: it permits configuring
154 encrypted devices that need to be ordered after the network is up.
155 Following this logic, two new special targets
156 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
157 added that are to cryptsetup.target what remote-fs.target and
158 remote-fs-pre.target are to local-fs.target.
160 * Service units gained a new UnsetEnvironment= setting which permits
161 unsetting specific environment variables for services that are
162 normally passed to it (for example in order to mask out locale
163 settings for specific services that can't deal with it).
165 * Units acquired a new boolean option IPAccounting=. When turned on, IP
166 traffic accounting (packet count as well as byte count) is done for
167 the service, and shown as part of "systemctl status" or "systemd-run
170 * Service units acquired two new options IPAddressAllow= and
171 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
172 for configuring a simple IP access control list for all sockets of
173 the unit. These options are available also on .slice and .socket
174 units, permitting flexible access list configuration for individual
175 services as well as groups of services (as defined by a slice unit),
176 including system-wide. Note that IP ACLs configured this way are
177 enforced on every single IPv4 and IPv6 socket created by any process
178 of the service unit, and apply to ingress as well as egress traffic.
180 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
181 structured log message is generated each time the unit is stopped,
182 containing information about the consumed resources of this
185 * A new setting KeyringMode= has been added to unit files, which may be
186 used to control how the kernel keyring is set up for executed
189 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
190 "systemctl kexec" and "systemctl exit" are now always asynchronous in
191 behaviour (that is: these commands return immediately after the
192 operation was enqueued instead of waiting for the operation to
193 complete). Previously, "systemctl poweroff" and "systemctl reboot"
194 were asynchronous on systems using systemd-logind (i.e. almost
195 always, and like they were on sysvinit), and the other three commands
196 were unconditionally synchronous. With this release this is cleaned
197 up, and callers will see the same asynchronous behaviour on all
198 systems for all five operations.
200 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
203 * .timer units now accept calendar specifications in other timezones
204 than UTC or the local timezone.
206 * The tmpfiles snippet var.conf has been changed to create
207 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
208 the "utmp" group already, and it appears to be generally understood
209 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
210 databases. Previously this was implemented correctly for all these
211 databases excepts btmp, which has been opened up like this now
212 too. Note that while the other databases are world-readable
213 (i.e. 0644), btmp is not and remains more restrictive.
215 * The systemd-resolve tool gained a new --reset-server-features
216 switch. When invoked like this systemd-resolved will forget
217 everything it learnt about the features supported by the configured
218 upstream DNS servers, and restarts the feature probing logic on the
219 next resolver look-up for them at the highest feature level
222 * The status dump systemd-resolved sends to the logs upon receiving
223 SIGUSR1 now also includes information about all DNS servers it is
224 configured to use, and the features levels it probed for them.
226 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
227 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
228 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
229 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
230 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
231 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
232 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
233 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
234 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
235 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
236 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
237 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
238 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
239 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
240 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
241 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
242 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
243 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
244 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
245 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
251 * Meson is now supported as build system in addition to Automake. It is
252 our plan to remove Automake in one of our next releases, so that
253 Meson becomes our exclusive build system. Hence, please start using
254 the Meson build system in your downstream packaging. There's plenty
255 of documentation around how to use Meson, the extremely brief
258 ./autogen.sh && ./configure && make && sudo make install
262 meson build && ninja -C build && sudo ninja -C build install
264 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
265 which permits configuring a timeout on the time a job is
266 running. This is particularly useful for setting timeouts on jobs for
269 * Unit files gained two new options ConditionUser= and ConditionGroup=
270 for conditionalizing units based on the identity of the user/group
271 running a systemd user instance.
273 * systemd-networkd now understands a new FlowLabel= setting in the
274 [VXLAN] section of .network files, as well as a Priority= in
275 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
276 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
277 gained support for configuration of GENEVE links, and IPv6 address
278 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
280 * .link files now understand a new Port= setting.
282 * systemd-networkd's DHCP support gained support for DHCP option 119
283 (domain search list).
285 * systemd-networkd gained support for serving IPv6 address ranges using
286 the Router Advertisment protocol. The new .network configuration
287 section [IPv6Prefix] may be used to configure the ranges to
288 serve. This is implemented based on a new, minimal, native server
289 implementation of RA.
291 * journalctl's --output= switch gained support for a new parameter
292 "short-iso-precise" for a mode where timestamps are shown as precise
295 * systemd-udevd's "net_id" builtin may now generate stable network
296 interface names from IBM PowerVM VIO devices as well as ACPI platform
299 * MulticastDNS support in systemd-resolved may now be explicitly
300 enabled/disabled using the new MulticastDNS= configuration file
303 * systemd-resolved may now optionally use libidn2 instead of the libidn
304 for processing internationalized domain names. Support for libidn2
305 should be considered experimental and should not be enabled by
308 * "machinectl pull-tar" and related call may now do verification of
309 downloaded images using SUSE-style .sha256 checksum files in addition
310 to the already existing support for validating using Ubuntu-style
313 * sd-bus gained support for a new sd_bus_message_appendv() call which
314 is va_list equivalent of sd_bus_message_append().
316 * sd-boot gained support for validating images using SHIM/MOK.
318 * The SMACK code learnt support for "onlycap".
320 * systemd-mount --umount is now much smarter in figuring out how to
321 properly unmount a device given its mount or device path.
323 * The code to call libnss_dns as a fallback from libnss_resolve when
324 the communication with systemd-resolved fails was removed. This
325 fallback was redundant and interfered with the [!UNAVAIL=return]
326 suffix. See nss-resolve(8) for the recommended configuration.
328 * systemd-logind may now be restarted without losing state. It stores
329 the file descriptors for devices it manages in the system manager
330 using the FDSTORE= mechanism. Please note that further changes in
331 other components may be required to make use of this (for example
332 Xorg has code to listen for stops of systemd-logind and terminate
333 itself when logind is stopped or restarted, in order to avoid using
334 stale file descriptors for graphical devices, which is now
335 counterproductive and must be reverted in order for restarts of
336 systemd-logind to be safe. See
337 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
339 * All kernel install plugins are called with the environment variable
340 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
341 /etc/machine-id. If the file is missing or empty, the variable is
342 empty and BOOT_DIR_ABS is the path of a temporary directory which is
343 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
344 is empty, all plugins should not put anything in BOOT_DIR_ABS.
346 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
347 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
348 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
349 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
350 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
351 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
352 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
353 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
354 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
355 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
356 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
357 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
358 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
359 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
360 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
361 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
362 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
363 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
364 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
365 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
366 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
367 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
368 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
369 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
370 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
371 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
372 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
373 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
374 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
381 * The "hybrid" control group mode has been modified to improve
382 compatibility with "legacy" cgroups-v1 setups. Specifically, the
383 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
384 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
385 cgroups-v1 hierarchy), the only externally visible change being that
386 the cgroups-v2 hierarchy is also mounted, to
387 /sys/fs/cgroup/unified. This should provide a large degree of
388 compatibility with "legacy" cgroups-v1, while taking benefit of the
389 better management capabilities of cgroups-v2.
391 * The default control group setup mode may be selected both a boot-time
392 via a set of kernel command line parameters (specifically:
393 systemd.unified_cgroup_hierarchy= and
394 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
395 default selected on the configure command line
396 (--with-default-hierarchy=). The upstream default is "hybrid"
397 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
398 this will change in a future systemd version to be "unified" (pure
399 cgroups-v2 mode). The third option for the compile time option is
400 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
401 distributions to default to "hybrid" mode for release distributions,
402 starting with v233. We recommend "unified" for development
403 distributions (specifically: distributions such as Fedora's rawhide)
404 as that's where things are headed in the long run. Use "legacy" for
405 greatest stability and compatibility only.
407 * Note one current limitation of "unified" and "hybrid" control group
408 setup modes: the kernel currently does not permit the systemd --user
409 instance (i.e. unprivileged code) to migrate processes between two
410 disconnected cgroup subtrees, even if both are managed and owned by
411 the user. This effectively means "systemd-run --user --scope" doesn't
412 work when invoked from outside of any "systemd --user" service or
413 scope. Specifically, it is not supported from session scopes. We are
414 working on fixing this in a future systemd version. (See #3388 for
415 further details about this.)
417 * DBus policy files are now installed into /usr rather than /etc. Make
418 sure your system has dbus >= 1.9.18 running before upgrading to this
419 version, or override the install path with --with-dbuspolicydir= .
421 * All python scripts shipped with systemd (specifically: the various
422 tests written in Python) now require Python 3.
424 * systemd unit tests can now run standalone (without the source or
425 build directories), and can be installed into /usr/lib/systemd/tests/
426 with 'make install-tests'.
428 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
429 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
432 * Support for the %c, %r, %R specifiers in unit files has been
433 removed. Specifiers are not supposed to be dependent on configuration
434 in the unit file itself (so that they resolve the same regardless
435 where used in the unit files), but these specifiers were influenced
436 by the Slice= option.
438 * The shell invoked by debug-shell.service now defaults to /bin/sh in
439 all cases. If distributions want to use a different shell for this
440 purpose (for example Fedora's /sbin/sushell) they need to specify
441 this explicitly at configure time using --with-debug-shell=.
443 * The confirmation spawn prompt has been reworked to offer the
446 (c)ontinue, proceed without asking anymore
447 (D)ump, show the state of the unit
448 (f)ail, don't execute the command and pretend it failed
450 (i)nfo, show a short summary of the unit
451 (j)obs, show jobs that are in progress
452 (s)kip, don't execute the command and pretend it succeeded
453 (y)es, execute the command
455 The 'n' choice for the confirmation spawn prompt has been removed,
456 because its meaning was confusing.
458 The prompt may now also be redirected to an alternative console by
459 specifying the console as parameter to systemd.confirm_spawn=.
461 * Services of Type=notify require a READY=1 notification to be sent
462 during startup. If no such message is sent, the service now fails,
463 even if the main process exited with a successful exit code.
465 * Services that fail to start up correctly now always have their
466 ExecStopPost= commands executed. Previously, they'd enter "failed"
467 state directly, without executing these commands.
469 * The option MulticastDNS= of network configuration files has acquired
470 an actual implementation. With MulticastDNS=yes a host can resolve
471 names of remote hosts and reply to mDNS A and AAAA requests.
473 * When units are about to be started an additional check is now done to
474 ensure that all dependencies of type BindsTo= (when used in
475 combination with After=) have been started.
477 * systemd-analyze gained a new verb "syscall-filter" which shows which
478 system call groups are defined for the SystemCallFilter= unit file
479 setting, and which system calls they contain.
481 * A new system call filter group "@filesystem" has been added,
482 consisting of various file system related system calls. Group
483 "@reboot" has been added, covering reboot, kexec and shutdown related
484 calls. Finally, group "@swap" has been added covering swap
485 configuration related calls.
487 * A new unit file option RestrictNamespaces= has been added that may be
488 used to restrict access to the various process namespace types the
489 Linux kernel provides. Specifically, it may be used to take away the
490 right for a service unit to create additional file system, network,
491 user, and other namespaces. This sandboxing option is particularly
492 relevant due to the high amount of recently discovered namespacing
493 related vulnerabilities in the kernel.
495 * systemd-udev's .link files gained support for a new AutoNegotiation=
496 setting for configuring Ethernet auto-negotiation.
498 * systemd-networkd's .network files gained support for a new
499 ListenPort= setting in the [DHCP] section to explicitly configure the
500 UDP client port the DHCP client shall listen on.
502 * .network files gained a new Unmanaged= boolean setting for explicitly
503 excluding one or more interfaces from management by systemd-networkd.
505 * The systemd-networkd ProxyARP= option has been renamed to
506 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
507 renamed to ReduceARPProxy=. The old names continue to be available
510 * systemd-networkd gained support for configuring IPv6 Proxy NDP
511 addresses via the new IPv6ProxyNDPAddress= .network file setting.
513 * systemd-networkd's bonding device support gained support for two new
514 configuration options ActiveSlave= and PrimarySlave=.
516 * The various options in the [Match] section of .network files gained
517 support for negative matching.
519 * New systemd-specific mount options are now understood in /etc/fstab:
521 x-systemd.mount-timeout= may be used to configure the maximum
522 permitted runtime of the mount command.
524 x-systemd.device-bound may be set to bind a mount point to its
525 backing device unit, in order to automatically remove a mount point
526 if its backing device is unplugged. This option may also be
527 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
528 on the block device, which is now automatically set for all CDROM
529 drives, so that mounted CDs are automatically unmounted when they are
530 removed from the drive.
532 x-systemd.after= and x-systemd.before= may be used to explicitly
533 order a mount after or before another unit or mount point.
535 * Enqueued start jobs for device units are now automatically garbage
536 collected if there are no jobs waiting for them anymore.
538 * systemctl list-jobs gained two new switches: with --after, for every
539 queued job the jobs it's waiting for are shown; with --before the
540 jobs which it's blocking are shown.
542 * systemd-nspawn gained support for ephemeral boots from disk images
543 (or in other words: --ephemeral and --image= may now be
544 combined). Moreover, ephemeral boots are now supported for normal
545 directories, even if the backing file system is not btrfs. Of course,
546 if the file system does not support file system snapshots or
547 reflinks, the initial copy operation will be relatively expensive, but
548 this should still be suitable for many use cases.
550 * Calendar time specifications in .timer units now support
551 specifications relative to the end of a month by using "~" instead of
552 "-" as separator between month and day. For example, "*-02~03" means
553 "the third last day in February". In addition a new syntax for
554 repeated events has been added using the "/" character. For example,
555 "9..17/2:00" means "every two hours from 9am to 5pm".
557 * systemd-socket-proxyd gained a new parameter --connections-max= for
558 configuring the maximum number of concurrent connections.
560 * sd-id128 gained a new API for generating unique IDs for the host in a
561 way that does not leak the machine ID. Specifically,
562 sd_id128_get_machine_app_specific() derives an ID based on the
563 machine ID a in well-defined, non-reversible, stable way. This is
564 useful whenever an identifier for the host is needed but where the
565 identifier shall not be useful to identify the system beyond the
566 scope of the application itself. (Internally this uses HMAC-SHA256 as
567 keyed hash function using the machine ID as input.)
569 * NotifyAccess= gained a new supported value "exec". When set
570 notifications are accepted from all processes systemd itself invoked,
571 including all control processes.
573 * .nspawn files gained support for defining overlay mounts using the
574 Overlay= and OverlayReadOnly= options. Previously this functionality
575 was only available on the systemd-nspawn command line.
577 * systemd-nspawn's --bind= and --overlay= options gained support for
578 bind/overlay mounts whose source lies within the container tree by
579 prefixing the source path with "+".
581 * systemd-nspawn's --bind= and --overlay= options gained support for
582 automatically allocating a temporary source directory in /var/tmp
583 that is removed when the container dies. Specifically, if the source
584 directory is specified as empty string this mechanism is selected. An
585 example usage is --overlay=+/var::/var, which creates an overlay
586 mount based on the original /var contained in the image, overlayed
587 with a temporary directory in the host's /var/tmp. This way changes
588 to /var are automatically flushed when the container shuts down.
590 * systemd-nspawn --image= option does now permit raw file system block
591 devices (in addition to images containing partition tables, as
594 * The disk image dissection logic in systemd-nspawn gained support for
595 automatically setting up LUKS encrypted as well as Verity protected
596 partitions. When a container is booted from an encrypted image the
597 passphrase is queried at start-up time. When a container with Verity
598 data is started, the root hash is search in a ".roothash" file
599 accompanying the disk image (alternatively, pass the root hash via
600 the new --root-hash= command line option).
602 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
603 be used to dissect disk images the same way as systemd-nspawn does
604 it, following the Bootable Partition Specification. It may even be
605 used to mount disk images with complex partition setups (including
606 LUKS and Verity partitions) to a local host directory, in order to
607 inspect them. This tool is not considered public API (yet), and is
608 thus not installed into /usr/bin. Please do not rely on its
609 existence, since it might go away or be changed in later systemd
612 * A new generator "systemd-verity-generator" has been added, similar in
613 style to "systemd-cryptsetup-generator", permitting automatic setup of
614 Verity root partitions when systemd boots up. In order to make use of
615 this your partition setup should follow the Discoverable Partitions
616 Specification, and the GPT partition ID of the root file system
617 partition should be identical to the upper 128bit of the Verity root
618 hash. The GPT partition ID of the Verity partition protecting it
619 should be the lower 128bit of the Verity root hash. If the partition
620 image follows this model it is sufficient to specify a single
621 "roothash=" kernel command line argument to both configure which root
622 image and verity partition to use as well as the root hash for
623 it. Note that systemd-nspawn's Verity support follows the same
624 semantics, meaning that disk images with proper Verity data in place
625 may be booted in containers with systemd-nspawn as well as on
626 physical systems via the verity generator. Also note that the "mkosi"
627 tool available at https://github.com/systemd/mkosi has been updated
628 to generate Verity protected disk images following this scheme. In
629 fact, it has been updated to generate disk images that optionally
630 implement a complete UEFI SecureBoot trust chain, involving a signed
631 kernel and initrd image that incorporates such a root hash as well as
632 a Verity-enabled root partition.
634 * The hardware database (hwdb) udev supports has been updated to carry
635 accelerometer quirks.
637 * All system services are now run with a fresh kernel keyring set up
638 for them. The invocation ID is stored by default in it, thus
639 providing a safe, non-overridable way to determine the invocation
642 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
643 options for bind mounting arbitrary paths in a service-specific
644 way. When these options are used, arbitrary host or service files and
645 directories may be mounted to arbitrary locations in the service's
648 * Documentation has been added that lists all of systemd's low-level
649 environment variables:
651 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
653 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
654 whether a specific socket file descriptor matches a specified socket
657 * systemd-firstboot has been updated to check for the
658 systemd.firstboot= kernel command line option. It accepts a boolean
659 and when set to false the first boot questions are skipped.
661 * systemd-fstab-generator has been updated to check for the
662 systemd.volatile= kernel command line option, which either takes an
663 optional boolean parameter or the special value "state". If used the
664 system may be booted in a "volatile" boot mode. Specifically,
665 "systemd.volatile" is used, the root directory will be mounted as
666 tmpfs, and only /usr is mounted from the actual root file system. If
667 "systemd.volatile=state" is used, the root directory will be mounted
668 as usual, but /var is mounted as tmpfs. This concept provides similar
669 functionality as systemd-nspawn's --volatile= option, but provides it
670 on physical boots. Use this option for implementing stateless
671 systems, or testing systems with all state and/or configuration reset
672 to the defaults. (Note though that many distributions are not
673 prepared to boot up without a populated /etc or /var, though.)
675 * systemd-gpt-auto-generator gained support for LUKS encrypted root
676 partitions. Previously it only supported LUKS encrypted partitions
677 for all other uses, except for the root partition itself.
679 * Socket units gained support for listening on AF_VSOCK sockets for
680 communication in virtualized QEMU environments.
682 * The "configure" script gained a new option --with-fallback-hostname=
683 for specifying the fallback hostname to use if none is configured in
684 /etc/hostname. For example, by specifying
685 --with-fallback-hostname=fedora it is possible to default to a
686 hostname of "fedora" on pristine installations.
688 * systemd-cgls gained support for a new --unit= switch for listing only
689 the control groups of a specific unit. Similar --user-unit= has been
690 added for listing only the control groups of a specific user unit.
692 * systemd-mount gained a new --umount switch for unmounting a mount or
693 automount point (and all mount/automount points below it).
695 * systemd will now refuse full configuration reloads (via systemctl
696 daemon-reload and related calls) unless at least 16MiB of free space
697 are available in /run. This is a safety precaution in order to ensure
698 that generators can safely operate after the reload completed.
700 * A new unit file option RootImage= has been added, which has a similar
701 effect as RootDirectory= but mounts the service's root directory from
702 a disk image instead of plain directory. This logic reuses the same
703 image dissection and mount logic that systemd-nspawn already uses,
704 and hence supports any disk images systemd-nspawn supports, including
705 those following the Discoverable Partition Specification, as well as
706 Verity enabled images. This option enables systemd to run system
707 services directly off disk images acting as resource bundles,
708 possibly even including full integrity data.
710 * A new MountAPIVFS= unit file option has been added, taking a boolean
711 argument. If enabled /proc, /sys and /dev (collectively called the
712 "API VFS") will be mounted for the service. This is only relevant if
713 RootDirectory= or RootImage= is used for the service, as these mounts
714 are of course in place in the host mount namespace anyway.
716 * systemd-nspawn gained support for a new --pivot-root= switch. If
717 specified the root directory within the container image is pivoted to
718 the specified mount point, while the original root disk is moved to a
719 different place. This option enables booting of ostree images
720 directly with systemd-nspawn.
722 * The systemd build scripts will no longer complain if the NTP server
723 addresses are not changed from the defaults. Google now supports
724 these NTP servers officially. We still recommend downstreams to
725 properly register an NTP pool with the NTP pool project though.
727 * coredumpctl gained a new "--reverse" option for printing the list
728 of coredumps in reverse order.
730 * coredumpctl will now show additional information about truncated and
731 inaccessible coredumps, as well as coredumps that are still being
732 processed. It also gained a new --quiet switch for suppressing
733 additional informational message in its output.
735 * coredumpctl gained support for only showing coredumps newer and/or
736 older than specific timestamps, using the new --since= and --until=
737 options, reminiscent of journalctl's options by the same name.
739 * The systemd-coredump logic has been improved so that it may be reused
740 to collect backtraces in non-compiled languages, for example in
741 scripting languages such as Python.
743 * machinectl will now show the UID shift of local containers, if user
744 namespacing is enabled for them.
746 * systemd will now optionally run "environment generator" binaries at
747 configuration load time. They may be used to add environment
748 variables to the environment block passed to services invoked. One
749 user environment generator is shipped by default that sets up
750 environment variables based on files dropped into /etc/environment.d
751 and ~/.config/environment.d/.
753 * systemd-resolved now includes the new, recently published 2017 DNSSEC
756 * hostnamed has been updated to report a new chassis type of
757 "convertible" to cover "foldable" laptops that can both act as a
758 tablet and as a laptop, such as various Lenovo Yoga devices.
760 Contributions from: Adrián López, Alexander Galanin, Alexander
761 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
762 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
763 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
764 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
765 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
766 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
767 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
768 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
769 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
770 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
771 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
772 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
773 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
774 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
775 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
776 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
777 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
778 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
779 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
780 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
781 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
782 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
783 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
784 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
785 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
786 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
787 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
794 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
795 RestrictAddressFamilies= enabled. These sandboxing options should
796 generally be compatible with the various external udev call-out
797 binaries we are aware of, however there may be exceptions, in
798 particular when exotic languages for these call-outs are used. In
799 this case, consider turning off these settings locally.
801 * The new RemoveIPC= option can be used to remove IPC objects owned by
802 the user or group of a service when that service exits.
804 * The new ProtectKernelModules= option can be used to disable explicit
805 load and unload operations of kernel modules by a service. In
806 addition access to /usr/lib/modules is removed if this option is set.
808 * ProtectSystem= option gained a new value "strict", which causes the
809 whole file system tree with the exception of /dev, /proc, and /sys,
810 to be remounted read-only for a service.
812 * The new ProtectKernelTunables= option can be used to disable
813 modification of configuration files in /sys and /proc by a service.
814 Various directories and files are remounted read-only, so access is
815 restricted even if the file permissions would allow it.
817 * The new ProtectControlGroups= option can be used to disable write
818 access by a service to /sys/fs/cgroup.
820 * Various systemd services have been hardened with
821 ProtectKernelTunables=yes, ProtectControlGroups=yes,
822 RestrictAddressFamilies=.
824 * Support for dynamically creating users for the lifetime of a service
825 has been added. If DynamicUser=yes is specified, user and group IDs
826 will be allocated from the range 61184..65519 for the lifetime of the
827 service. They can be resolved using the new nss-systemd.so NSS
828 module. The module must be enabled in /etc/nsswitch.conf. Services
829 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
830 any resources allocated by the service will be cleaned up when the
831 service exits. They also have ProtectHome=read-only and
832 ProtectSystem=strict enabled, so they are not able to make any
833 permanent modifications to the system.
835 * The nss-systemd module also always resolves root and nobody, making
836 it possible to have no /etc/passwd or /etc/group files in minimal
837 container or chroot environments.
839 * Services may be started with their own user namespace using the new
840 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
841 under which the service is running are mapped. All other users are
844 * Support for the cgroup namespace has been added to systemd-nspawn. If
845 supported by kernel, the container system started by systemd-nspawn
846 will have its own view of the cgroup hierarchy. This new behaviour
847 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
849 * The new MemorySwapMax= option can be used to limit the maximum swap
850 usage under the unified cgroup hierarchy.
852 * Support for the CPU controller in the unified cgroup hierarchy has
853 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
854 options. This controller requires out-of-tree patches for the kernel
855 and the support is provisional.
857 * Mount and automount units may now be created transiently
858 (i.e. dynamically at runtime via the bus API, instead of requiring
859 unit files in the file system).
861 * systemd-mount is a new tool which may mount file systems – much like
862 mount(8), optionally pulling in additional dependencies through
863 transient .mount and .automount units. For example, this tool
864 automatically runs fsck on a backing block device before mounting,
865 and allows the automount logic to be used dynamically from the
866 command line for establishing mount points. This tool is particularly
867 useful when dealing with removable media, as it will ensure fsck is
868 run – if necessary – before the first access and that the file system
869 is quickly unmounted after each access by utilizing the automount
870 logic. This maximizes the chance that the file system on the
871 removable media stays in a clean state, and if it isn't in a clean
872 state is fixed automatically.
874 * LazyUnmount=yes option for mount units has been added to expose the
875 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
878 * /efi will be used as the mount point of the EFI boot partition, if
879 the directory is present, and the mount point was not configured
880 through other means (e.g. fstab). If /efi directory does not exist,
881 /boot will be used as before. This makes it easier to automatically
882 mount the EFI partition on systems where /boot is used for something
885 * When operating on GPT disk images for containers, systemd-nspawn will
886 now mount the ESP to /boot or /efi according to the same rules as PID
887 1 running on a host. This allows tools like "bootctl" to operate
888 correctly within such containers, in order to make container images
889 bootable on physical systems.
891 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
893 * Two new user session targets have been added to support running
894 graphical sessions under the systemd --user instance:
895 graphical-session.target and graphical-session-pre.target. See
896 systemd.special(7) for a description of how those targets should be
899 * The vconsole initialization code has been significantly reworked to
900 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
901 support unicode keymaps. Font and keymap configuration will now be
902 copied to all allocated virtual consoles.
904 * FreeBSD's bhyve virtualization is now detected.
906 * Information recorded in the journal for core dumps now includes the
907 contents of /proc/mountinfo and the command line of the process at
908 the top of the process hierarchy (which is usually the init process
911 * systemd-journal-gatewayd learned the --directory= option to serve
912 files from the specified location.
914 * journalctl --root=… can be used to peruse the journal in the
915 /var/log/ directories inside of a container tree. This is similar to
916 the existing --machine= option, but does not require the container to
919 * The hardware database has been extended to support
920 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
923 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
924 specify the click rate for mice which include a horizontal wheel with
925 a click rate that is different than the one for the vertical wheel.
927 * systemd-run gained a new --wait option that makes service execution
928 synchronous. (Specifically, the command will not return until the
929 specified service binary exited.)
931 * systemctl gained a new --wait option that causes the start command to
932 wait until the units being started have terminated again.
934 * A new journal output mode "short-full" has been added which displays
935 timestamps with abbreviated English day names and adds a timezone
936 suffix. Those timestamps include more information than the default
937 "short" output mode, and can be passed directly to journalctl's
938 --since= and --until= options.
940 * /etc/resolv.conf will be bind-mounted into containers started by
941 systemd-nspawn, if possible, so any changes to resolv.conf contents
942 are automatically propagated to the container.
944 * The number of instances for socket-activated services originating
945 from a single IP address can be limited with
946 MaxConnectionsPerSource=, extending the existing setting of
949 * systemd-networkd gained support for vcan ("Virtual CAN") interface
952 * .netdev and .network configuration can now be extended through
955 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
956 Segmentation Offload, Generic Receive Offload, Large Receive Offload
957 can be enabled and disabled using the new UDPSegmentationOffload=,
958 TCPSegmentationOffload=, GenericSegmentationOffload=,
959 GenericReceiveOffload=, LargeReceiveOffload= options in the
960 [Link] section of .link files.
962 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
963 Port VLAN ID can be configured for bridge devices using the new STP=,
964 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
965 section of .netdev files.
967 * The route table to which routes received over DHCP or RA should be
968 added can be configured with the new RouteTable= option in the [DHCP]
969 and [IPv6AcceptRA] sections of .network files.
971 * The Address Resolution Protocol can be disabled on links managed by
972 systemd-networkd using the ARP=no setting in the [Link] section of
975 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
976 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
977 encode information about the result and exit codes of the current
978 service runtime cycle.
980 * systemd-sysctl will now configure kernel parameters in the order
981 they occur in the configuration files. This matches what sysctl
982 has been traditionally doing.
984 * kernel-install "plugins" that are executed to perform various
985 tasks after a new kernel is added and before an old one is removed
986 can now return a special value to terminate the procedure and
987 prevent any later plugins from running.
989 * Journald's SplitMode=login setting has been deprecated. It has been
990 removed from documentation, and its use is discouraged. In a future
991 release it will be completely removed, and made equivalent to current
992 default of SplitMode=uid.
994 * Storage=both option setting in /etc/systemd/coredump.conf has been
995 removed. With fast LZ4 compression storing the core dump twice is not
998 * The --share-system systemd-nspawn option has been replaced with an
999 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1000 this functionality is discouraged. In addition the variables
1001 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1002 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1003 individual namespaces.
1005 * "machinectl list" now shows the IP address of running containers in
1006 the output, as well as OS release information.
1008 * "loginctl list" now shows the TTY of each session in the output.
1010 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1011 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1012 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1013 tracking objects in a "recursive" mode, where a single client can be
1014 counted multiple times, if it takes multiple references.
1016 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1017 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1018 process using sd-bus automatically exit if the bus connection is
1021 * Bus clients of the service manager may now "pin" loaded units into
1022 memory, by taking an explicit reference on them. This is useful to
1023 ensure the client can retrieve runtime data about the service even
1024 after the service completed execution. Taking such a reference is
1025 available only for privileged clients and should be helpful to watch
1026 running services in a race-free manner, and in particular collect
1027 information about exit statuses and results.
1029 * The nss-resolve module has been changed to strictly return UNAVAIL
1030 when communication via D-Bus with resolved failed, and NOTFOUND when
1031 a lookup completed but was negative. This means it is now possible to
1032 neatly configure fallbacks using nsswitch.conf result checking
1033 expressions. Taking benefit of this, the new recommended
1034 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1036 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1038 * A new setting CtrlAltDelBurstAction= has been added to
1039 /etc/systemd/system.conf which may be used to configure the precise
1040 behaviour if the user on the console presses Ctrl-Alt-Del more often
1041 than 7 times in 2s. Previously this would unconditionally result in
1042 an expedited, immediate reboot. With this new setting the precise
1043 operation may be configured in more detail, and also turned off
1046 * In .netdev files two new settings RemoteChecksumTx= and
1047 RemoteChecksumRx= are now understood that permit configuring the
1048 remote checksumming logic for VXLAN networks.
1050 * The service manager learnt a new "invocation ID" concept for invoked
1051 services. Each runtime cycle of a service will get a new invocation
1052 ID (a 128bit random UUID) assigned that identifies the current
1053 run of the service uniquely and globally. A new invocation ID
1054 is generated each time a service starts up. The journal will store
1055 the invocation ID of a service along with any logged messages, thus
1056 making the invocation ID useful for matching the online runtime of a
1057 service with the offline log data it generated in a safe way without
1058 relying on synchronized timestamps. In many ways this new service
1059 invocation ID concept is similar to the kernel's boot ID concept that
1060 uniquely and globally identifies the runtime of each boot. The
1061 invocation ID of a service is passed to the service itself via an
1062 environment variable ($INVOCATION_ID). A new bus call
1063 GetUnitByInvocationID() has been added that is similar to GetUnit()
1064 but instead of retrieving the bus path for a unit by its name
1065 retrieves it by its invocation ID. The returned path is valid only as
1066 long as the passed invocation ID is current.
1068 * systemd-resolved gained a new "DNSStubListener" setting in
1069 resolved.conf. It either takes a boolean value or the special values
1070 "udp" and "tcp", and configures whether to enable the stub DNS
1071 listener on 127.0.0.53:53.
1073 * IP addresses configured via networkd may now carry additional
1074 configuration settings supported by the kernel. New options include:
1075 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1076 PrefixRoute=, AutoJoin=.
1078 * The PAM configuration fragment file for "user@.service" shipped with
1079 systemd (i.e. the --user instance of systemd) has been stripped to
1080 the minimum necessary to make the system boot. Previously, it
1081 contained Fedora-specific stanzas that did not apply to other
1082 distributions. It is expected that downstream distributions add
1083 additional configuration lines, matching their needs to this file,
1084 using it only as rough template of what systemd itself needs. Note
1085 that this reduced fragment does not even include an invocation of
1086 pam_limits which most distributions probably want to add, even though
1087 systemd itself does not need it. (There's also the new build time
1088 option --with-pamconfdir=no to disable installation of the PAM
1091 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1092 capability is now also dropped from its set (in addition to
1093 CAP_SYS_MKNOD as before).
1095 * In service unit files it is now possible to connect a specific named
1096 file descriptor with stdin/stdout/stdout of an executed service. The
1097 name may be specified in matching .socket units using the
1098 FileDescriptorName= setting.
1100 * A number of journal settings may now be configured on the kernel
1101 command line. Specifically, the following options are now understood:
1102 systemd.journald.max_level_console=,
1103 systemd.journald.max_level_store=,
1104 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1105 systemd.journald.max_level_wall=.
1107 * "systemctl is-enabled --full" will now show by which symlinks a unit
1108 file is enabled in the unit dependency tree.
1110 * Support for VeraCrypt encrypted partitions has been added to the
1111 "cryptsetup" logic and /etc/crypttab.
1113 * systemd-detect-virt gained support for a new --private-users switch
1114 that checks whether the invoking processes are running inside a user
1115 namespace. Similar, a new special value "private-users" for the
1116 existing ConditionVirtualization= setting has been added, permitting
1117 skipping of specific units in user namespace environments.
1119 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1120 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1121 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1122 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1123 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1124 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1125 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1126 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1127 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1128 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1129 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1130 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1131 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1132 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1133 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1134 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1135 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1136 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1137 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1138 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1139 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1140 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1141 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1142 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1143 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1144 Jędrzejewski-Szmek, Zeal Jagannatha
1146 — Santa Fe, 2016-11-03
1150 * In service units the various ExecXYZ= settings have been extended
1151 with an additional special character as first argument of the
1152 assigned value: if the character '+' is used the specified command
1153 line it will be run with full privileges, regardless of User=,
1154 Group=, CapabilityBoundingSet= and similar options. The effect is
1155 similar to the existing PermissionsStartOnly= option, but allows
1156 configuration of this concept for each executed command line
1159 * Services may now alter the service watchdog timeout at runtime by
1160 sending a WATCHDOG_USEC= message via sd_notify().
1162 * MemoryLimit= and related unit settings now optionally take percentage
1163 specifications. The percentage is taken relative to the amount of
1164 physical memory in the system (or in case of containers, the assigned
1165 amount of memory). This allows scaling service resources neatly with
1166 the amount of RAM available on the system. Similarly, systemd-logind's
1167 RuntimeDirectorySize= option now also optionally takes percentage
1170 * In similar fashion TasksMax= takes percentage values now, too. The
1171 value is taken relative to the configured maximum number of processes
1172 on the system. The per-service task maximum has been changed to 15%
1173 using this functionality. (Effectively this is an increase of 512 →
1174 4915 for service units, given the kernel's default pid_max setting.)
1176 * Calendar time specifications in .timer units now understand a ".."
1177 syntax for time ranges. Example: "4..7:10" may now be used for
1178 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1181 * The InaccessableDirectories=, ReadOnlyDirectories= and
1182 ReadWriteDirectories= unit file settings have been renamed to
1183 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1184 applied to all kinds of file nodes, and not just directories, with
1185 the exception of symlinks. Specifically these settings may now be
1186 used on block and character device nodes, UNIX sockets and FIFOS as
1187 well as regular files. The old names of these settings remain
1188 available for compatibility.
1190 * systemd will now log about all service processes it kills forcibly
1191 (using SIGKILL) because they remained after the clean shutdown phase
1192 of the service completed. This should help identifying services that
1193 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1194 systemd-logind's configuration a similar log message is generated for
1195 processes killed at the end of each session due to this setting.
1197 * systemd will now set the $JOURNAL_STREAM environment variable for all
1198 services whose stdout/stderr are connected to the Journal (which
1199 effectively means by default: all services). The variable contains
1200 the device and inode number of the file descriptor used for
1201 stdout/stderr. This may be used by invoked programs to detect whether
1202 their stdout/stderr is connected to the Journal, in which case they
1203 can switch over to direct Journal communication, thus being able to
1204 pass extended, structured metadata along with their log messages. As
1205 one example, this is now used by glib's logging primitives.
1207 * When using systemd's default tmp.mount unit for /tmp, the mount point
1208 will now be established with the "nosuid" and "nodev" options. This
1209 avoids privilege escalation attacks that put traps and exploits into
1210 /tmp. However, this might cause problems if you e. g. put container
1211 images or overlays into /tmp; if you need this, override tmp.mount's
1212 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1215 * systemd now supports the "memory" cgroup controller also on
1218 * The systemd-cgtop tool now optionally takes a control group path as
1219 command line argument. If specified, the control group list shown is
1220 limited to subgroups of that group.
1222 * The SystemCallFilter= unit file setting gained support for
1223 pre-defined, named system call filter sets. For example
1224 SystemCallFilter=@clock is now an effective way to make all clock
1225 changing-related system calls unavailable to a service. A number of
1226 similar pre-defined groups are defined. Writing system call filters
1227 for system services is simplified substantially with this new
1228 concept. Accordingly, all of systemd's own, long-running services now
1229 enable system call filtering based on this, by default.
1231 * A new service setting MemoryDenyWriteExecute= has been added, taking
1232 a boolean value. If turned on, a service may no longer create memory
1233 mappings that are writable and executable at the same time. This
1234 enhances security for services where this is enabled as it becomes
1235 harder to dynamically write and then execute memory in exploited
1236 service processes. This option has been enabled for all of systemd's
1237 own long-running services.
1239 * A new RestrictRealtime= service setting has been added, taking a
1240 boolean argument. If set the service's processes may no longer
1241 acquire realtime scheduling. This improves security as realtime
1242 scheduling may otherwise be used to easily freeze the system.
1244 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1245 value. This may be used for requesting that the system manager inside
1246 of the container reports start-up completion to nspawn which then
1247 propagates this notification further to the service manager
1248 supervising nspawn itself. A related option NotifyReady= in .nspawn
1249 files has been added too. This functionality allows ordering of the
1250 start-up of multiple containers using the usual systemd ordering
1253 * machinectl gained a new command "stop" that is an alias for
1256 * systemd-resolved gained support for contacting DNS servers on
1257 link-local IPv6 addresses.
1259 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1260 its caches. A method call for requesting the same operation has been
1261 added to the bus API too, and is made available via "systemd-resolve
1264 * systemd-resolve gained a new --status switch. If passed a brief
1265 summary of the used DNS configuration with per-interface information
1268 * resolved.conf gained a new Cache= boolean option, defaulting to
1269 on. If turned off local DNS caching is disabled. This comes with a
1270 performance penalty in particular when DNSSEC is enabled. Note that
1271 resolved disables its internal caching implicitly anyway, when the
1272 configured DNS server is on a host-local IP address such as ::1 or
1273 127.0.0.1, thus automatically avoiding double local caching.
1275 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1276 for DNS requests. This improves compatibility with local programs
1277 that do not use the libc NSS or systemd-resolved's bus APIs for name
1278 resolution. This minimal DNS service is only available to local
1279 programs and does not implement the full DNS protocol, but enough to
1280 cover local DNS clients. A new, static resolv.conf file, listing just
1281 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1282 now recommended to make /etc/resolv.conf a symlink to this file in
1283 order to route all DNS lookups to systemd-resolved, regardless if
1284 done via NSS, the bus API or raw DNS packets. Note that this local
1285 DNS service is not as fully featured as the libc NSS or
1286 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1287 used to deliver link-local address information (as this implies
1288 sending a local interface index along), LLMNR/mDNS support via this
1289 interface is severely restricted. It is thus strongly recommended for
1290 all applications to use the libc NSS API or native systemd-resolved
1293 * systemd-networkd's bridge support learned a new setting
1294 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1295 in .network files has been added for configuring VLAN bridging in
1296 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1298 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1299 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1300 now be acquired without relying on DHCPv6. Two new options
1301 UseDomains= and UseDNS= have been added to configure this behaviour.
1303 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1304 renamed IPv6AcceptRA=, without altering its behaviour. The old
1305 setting name remains available for compatibility reasons.
1307 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1308 Key=, InputKey= and OutputKey=.
1310 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1311 interface configuration.
1313 * "systemctl edit" may now be used to create new unit files by
1314 specifying the --force switch.
1316 * sd-event gained a new function sd_event_get_iteration() for
1317 requesting the current iteration counter of the event loop. It starts
1318 at zero and is increased by one with each event loop iteration.
1320 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1321 file. It can be used in lieu of %systemd_requires in packages which
1322 don't use any systemd functionality and are intended to be installed
1323 in minimal containers without systemd present. This macro provides
1324 ordering dependencies to ensure that if the package is installed in
1325 the same rpm transaction as systemd, systemd will be installed before
1326 the scriptlets for the package are executed, allowing unit presets
1329 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1330 been added to simplify packaging of generators.
1332 * The os-release file gained VERSION_CODENAME field for the
1333 distribution nickname (e.g. VERSION_CODENAME=woody).
1335 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1336 can be set to disable parsing of metadata and the creation
1337 of persistent symlinks for that device.
1339 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1340 to make them available to logged-in users has been reverted.
1342 * Much of the common code of the various systemd components is now
1343 built into an internal shared library libsystemd-shared-231.so
1344 (incorporating the systemd version number in the name, to be updated
1345 with future releases) that the components link to. This should
1346 decrease systemd footprint both in memory during runtime and on
1347 disk. Note that the shared library is not for public use, and is
1348 neither API not ABI stable, but is likely to change with every new
1349 released update. Packagers need to make sure that binaries
1350 linking to libsystemd-shared.so are updated in step with the
1353 * Configuration for "mkosi" is now part of the systemd
1354 repository. mkosi is a tool to easily build legacy-free OS images,
1355 and is available on github: https://github.com/systemd/mkosi. If
1356 "mkosi" is invoked in the build tree a new raw OS image is generated
1357 incorporating the systemd sources currently being worked on and a
1358 clean, fresh distribution installation. The generated OS image may be
1359 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1360 UEFI PC. This functionality is particularly useful to easily test
1361 local changes made to systemd in a pristine, defined environment. See
1362 HACKING for details.
1364 * configure learned the --with-support-url= option to specify the
1365 distribution's bugtracker.
1367 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1368 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1369 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1370 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1371 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1372 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1373 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1374 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1375 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1376 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1377 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1378 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1379 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1380 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1381 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1382 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1383 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1384 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1385 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1387 — Berlin, 2016-07-25
1391 * DNSSEC is now turned on by default in systemd-resolved (in
1392 "allow-downgrade" mode), but may be turned off during compile time by
1393 passing "--with-default-dnssec=no" to "configure" (and of course,
1394 during runtime with DNSSEC= in resolved.conf). We recommend
1395 downstreams to leave this on at least during development cycles and
1396 report any issues with the DNSSEC logic upstream. We are very
1397 interested in collecting feedback about the DNSSEC validator and its
1398 limitations in the wild. Note however, that DNSSEC support is
1399 probably nothing downstreams should turn on in stable distros just
1400 yet, as it might create incompatibilities with a few DNS servers and
1401 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1402 automatically whenever we detect such incompatible setups, but there
1403 might be systems we do not cover yet. Hence: please help us testing
1404 the DNSSEC code, leave this on where you can, report back, but then
1405 again don't consider turning this on in your stable, LTS or
1406 production release just yet. (Note that you have to enable
1407 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1408 and its DNSSEC mode for host name resolution from local
1411 * systemd-resolve conveniently resolves DANE records with the --tlsa
1412 option and OPENPGPKEY records with the --openpgp option. It also
1413 supports dumping raw DNS record data via the new --raw= switch.
1415 * systemd-logind will now by default terminate user processes that are
1416 part of the user session scope unit (session-XX.scope) when the user
1417 logs out. This behavior is controlled by the KillUserProcesses=
1418 setting in logind.conf, and the previous default of "no" is now
1419 changed to "yes". This means that user sessions will be properly
1420 cleaned up after, but additional steps are necessary to allow
1421 intentionally long-running processes to survive logout.
1423 While the user is logged in at least once, user@.service is running,
1424 and any service that should survive the end of any individual login
1425 session can be started at a user service or scope using systemd-run.
1426 systemd-run(1) man page has been extended with an example which shows
1427 how to run screen in a scope unit underneath user@.service. The same
1428 command works for tmux.
1430 After the user logs out of all sessions, user@.service will be
1431 terminated too, by default, unless the user has "lingering" enabled.
1432 To effectively allow users to run long-term tasks even if they are
1433 logged out, lingering must be enabled for them. See loginctl(1) for
1434 details. The default polkit policy was modified to allow users to
1435 set lingering for themselves without authentication.
1437 Previous defaults can be restored at compile time by the
1438 --without-kill-user-processes option to "configure".
1440 * systemd-logind gained new configuration settings SessionsMax= and
1441 InhibitorsMax=, both with a default of 8192. It will not register new
1442 user sessions or inhibitors above this limit.
1444 * systemd-logind will now reload configuration on SIGHUP.
1446 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1447 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1448 enable. Also, support for the "io" cgroup controller in the unified
1449 hierarchy has been added, so that the "memory", "pids" and "io" are
1450 now the controllers that are supported on the unified hierarchy.
1452 WARNING: it is not possible to use previous systemd versions with
1453 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1454 is necessary to also update systemd in the initramfs if using the
1455 unified hierarchy. An updated SELinux policy is also required.
1457 * LLDP support has been extended, and both passive (receive-only) and
1458 active (sender) modes are supported. Passive mode ("routers-only") is
1459 enabled by default in systemd-networkd. Active LLDP mode is enabled
1460 by default for containers on the internal network. The "networkctl
1461 lldp" command may be used to list information gathered. "networkctl
1462 status" will also show basic LLDP information on connected peers now.
1464 * The IAID and DUID unique identifier sent in DHCP requests may now be
1465 configured for the system and each .network file managed by
1466 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1468 * systemd-networkd gained support for configuring proxy ARP support for
1469 each interface, via the ProxyArp= setting in .network files. It also
1470 gained support for configuring the multicast querier feature of
1471 bridge devices, via the new MulticastQuerier= setting in .netdev
1472 files. Similarly, snooping on the IGMP traffic can be controlled
1473 via the new setting MulticastSnooping=.
1475 A new setting PreferredLifetime= has been added for addresses
1476 configured in .network file to configure the lifetime intended for an
1479 The systemd-networkd DHCP server gained the option EmitRouter=, which
1480 defaults to yes, to configure whether the DHCP Option 3 (Router)
1483 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1484 systemd-socket-activate and installed into /usr/bin. It is now fully
1487 * systemd-journald now uses separate threads to flush changes to disk
1488 when closing journal files, thus reducing impact of slow disk I/O on
1489 logging performance.
1491 * The sd-journal API gained two new calls
1492 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1493 can be used to open journal files using file descriptors instead of
1494 file or directory paths. sd_journal_open_container() has been
1495 deprecated, sd_journal_open_directory_fd() should be used instead
1496 with the flag SD_JOURNAL_OS_ROOT.
1498 * journalctl learned a new output mode "-o short-unix" that outputs log
1499 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1500 UTC). It also gained support for a new --no-hostname setting to
1501 suppress the hostname column in the family of "short" output modes.
1503 * systemd-ask-password now optionally skips printing of the password to
1504 stdout with --no-output which can be useful in scripts.
1506 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1507 (devices tagged with ID_MAKER_TOOL) are now tagged with
1508 "uaccess" and are available to logged in users.
1510 * The DeviceAllow= unit setting now supports specifiers (with "%").
1512 * "systemctl show" gained a new --value switch, which allows print a
1513 only the contents of a specific unit property, without also printing
1514 the property's name. Similar support was added to "show*" verbs
1515 of loginctl and machinectl that output "key=value" lists.
1517 * A new unit type "generated" was added for files dynamically generated
1518 by generator tools. Similarly, a new unit type "transient" is used
1519 for unit files created using the runtime API. "systemctl enable" will
1520 refuse to operate on such files.
1522 * A new command "systemctl revert" has been added that may be used to
1523 revert to the vendor version of a unit file, in case local changes
1524 have been made by adding drop-ins or overriding the unit file.
1526 * "machinectl clean" gained a new verb to automatically remove all or
1527 just hidden container images.
1529 * systemd-tmpfiles gained support for a new line type "e" for emptying
1530 directories, if they exist, without creating them if they don't.
1532 * systemd-nspawn gained support for automatically patching the UID/GIDs
1533 of the owners and the ACLs of all files and directories in a
1534 container tree to match the UID/GID user namespacing range selected
1535 for the container invocation. This mode is enabled via the new
1536 --private-users-chown switch. It also gained support for
1537 automatically choosing a free, previously unused UID/GID range when
1538 starting a container, via the new --private-users=pick setting (which
1539 implies --private-users-chown). Together, these options for the first
1540 time make user namespacing for nspawn containers fully automatic and
1541 thus deployable. The systemd-nspawn@.service template unit file has
1542 been changed to use this functionality by default.
1544 * systemd-nspawn gained a new --network-zone= switch, that allows
1545 creating ad-hoc virtual Ethernet links between multiple containers,
1546 that only exist as long as at least one container referencing them is
1547 running. This allows easy connecting of multiple containers with a
1548 common link that implements an Ethernet broadcast domain. Each of
1549 these network "zones" may be named relatively freely by the user, and
1550 may be referenced by any number of containers, but each container may
1551 only reference one of these "zones". On the lower level, this is
1552 implemented by an automatically managed bridge network interface for
1553 each zone, that is created when the first container referencing its
1554 zone is created and removed when the last one referencing its zone
1557 * The default start timeout may now be configured on the kernel command
1558 line via systemd.default_timeout_start_sec=. It was already
1559 configurable via the DefaultTimeoutStartSec= option in
1560 /etc/systemd/system.conf.
1562 * Socket units gained a new TriggerLimitIntervalSec= and
1563 TriggerLimitBurst= setting to configure a limit on the activation
1564 rate of the socket unit.
1566 * The LimitNICE= setting now optionally takes normal UNIX nice values
1567 in addition to the raw integer limit value. If the specified
1568 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1569 value is understood as UNIX nice value. If not prefixed like this it
1570 is understood as raw RLIMIT_NICE limit.
1572 * Note that the effect of the PrivateDevices= unit file setting changed
1573 slightly with this release: the per-device /dev file system will be
1574 mounted read-only from this version on, and will have "noexec"
1575 set. This (minor) change of behavior might cause some (exceptional)
1576 legacy software to break, when PrivateDevices=yes is set for its
1577 service. Please leave PrivateDevices= off if you run into problems
1580 * systemd-bootchart has been split out to a separate repository:
1581 https://github.com/systemd/systemd-bootchart
1583 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1584 merged into the kernel in its current form.
1586 * The compatibility libraries libsystemd-daemon.so,
1587 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1588 which have been deprecated since systemd-209 have been removed along
1589 with the corresponding pkg-config files. All symbols provided by
1590 those libraries are provided by libsystemd.so.
1592 * The Capabilities= unit file setting has been removed (it is ignored
1593 for backwards compatibility). AmbientCapabilities= and
1594 CapabilityBoundingSet= should be used instead.
1596 * A new special target has been added, initrd-root-device.target,
1597 which creates a synchronization point for dependencies of the root
1598 device in early userspace. Initramfs builders must ensure that this
1599 target is now included in early userspace.
1601 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1602 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1603 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1604 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1605 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1606 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1607 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1608 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1609 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1610 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1611 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1612 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1613 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1614 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1615 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1616 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1617 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1618 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1619 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1620 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1621 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1622 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1623 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1624 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1625 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1628 — Fairfax, 2016-05-21
1632 * The systemd-resolved DNS resolver service has gained a substantial
1633 set of new features, most prominently it may now act as a DNSSEC
1634 validating stub resolver. DNSSEC mode is currently turned off by
1635 default, but is expected to be turned on by default in one of the
1636 next releases. For now, we invite everybody to test the DNSSEC logic
1637 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1638 service also gained a full set of D-Bus interfaces, including calls
1639 to configure DNS and DNSSEC settings per link (for use by external
1640 network management software). systemd-resolved and systemd-networkd
1641 now distinguish between "search" and "routing" domains. The former
1642 are used to qualify single-label names, the latter are used purely
1643 for routing lookups within certain domains to specific links.
1644 resolved now also synthesizes RRs for all entries from /etc/hosts.
1646 * The systemd-resolve tool (which is a client utility for
1647 systemd-resolved) has been improved considerably and is now fully
1648 supported and documented. Hence it has moved from /usr/lib/systemd to
1651 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1654 * The coredump collection logic has been reworked: when a coredump is
1655 collected it is now written to disk, compressed and processed
1656 (including stacktrace extraction) from a new instantiated service
1657 systemd-coredump@.service, instead of directly from the
1658 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1659 processing large coredumps can take up a substantial amount of
1660 resources and time, and this previously happened entirely outside of
1661 systemd's service supervision. With the new logic the core_pattern
1662 hook only does minimal metadata collection before passing off control
1663 to the new instantiated service, which is configured with a time
1664 limit, a nice level and other settings to minimize negative impact on
1665 the rest of the system. Also note that the new logic will honour the
1666 RLIMIT_CORE setting of the crashed process, which now allows users
1667 and processes to turn off coredumping for their processes by setting
1670 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1671 and all forked processes by default. Previously, PID 1 would leave
1672 the setting at "0" for all processes, as set by the kernel. Note that
1673 the resource limit traditionally has no effect on the generated
1674 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1675 logic is used. Since the limit is now honoured (see above) its
1676 default has been changed so that the coredumping logic is enabled by
1677 default for all processes, while allowing specific opt-out.
1679 * When the stacktrace is extracted from processes of system users, this
1680 is now done as "systemd-coredump" user, in order to sandbox this
1681 potentially security sensitive parsing operation. (Note that when
1682 processing coredumps of normal users this is done under the user ID
1683 of process that crashed, as before.) Packagers should take notice
1684 that it is now necessary to create the "systemd-coredump" system user
1685 and group at package installation time.
1687 * The systemd-activate socket activation testing tool gained support
1688 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1689 and --seqpacket switches. It also has been extended to support both
1690 new-style and inetd-style file descriptor passing. Use the new
1691 --inetd switch to request inetd-style file descriptor passing.
1693 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1694 variable, which takes a boolean value. If set to false, ANSI color
1695 output is disabled in the tools even when run on a terminal that
1698 * The VXLAN support in networkd now supports two new settings
1699 DestinationPort= and PortRange=.
1701 * A new systemd.machine_id= kernel command line switch has been added,
1702 that may be used to set the machine ID in /etc/machine-id if it is
1703 not initialized yet. This command line option has no effect if the
1704 file is already initialized.
1706 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1707 specified command line as PID 2 rather than PID 1 in the
1708 container. In this mode PID 1 is a minimal stub init process that
1709 implements the special POSIX and Linux semantics of PID 1 regarding
1710 signal and child process management. Note that this stub init process
1711 is implemented in nspawn itself and requires no support from the
1712 container image. This new logic is useful to support running
1713 arbitrary commands in the container, as normal processes are
1714 generally not prepared to run as PID 1.
1716 * systemd-nspawn gained a new --chdir= switch for setting the current
1717 working directory for the process started in the container.
1719 * "journalctl /dev/sda" will now output all kernel log messages for
1720 specified device from the current boot, in addition to all devices
1721 that are parents of it. This should make log output about devices
1722 pretty useful, as long as kernel drivers attach enough metadata to
1723 the log messages. (The usual SATA drivers do.)
1725 * The sd-journal API gained two new calls
1726 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1727 that report whether log data from /run or /var has been found.
1729 * journalctl gained a new switch "--fields" that prints all journal
1730 record field names currently in use in the journal. This is backed
1731 by two new sd-journal API calls sd_journal_enumerate_fields() and
1732 sd_journal_restart_fields().
1734 * Most configurable timeouts in systemd now expect an argument of
1735 "infinity" to turn them off, instead of "0" as before. The semantics
1736 from now on is that a timeout of "0" means "now", and "infinity"
1737 means "never". To maintain backwards compatibility, "0" continues to
1738 turn off previously existing timeout settings.
1740 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1741 try-reload-or-restart" to clarify what it actually does: the "try"
1742 logic applies to both reloading and restarting, not just restarting.
1743 The old name continues to be accepted for compatibility.
1745 * On boot-up, when PID 1 detects that the system clock is behind the
1746 release date of the systemd version in use, the clock is now set
1747 to the latter. Previously, this was already done in timesyncd, in order
1748 to avoid running with clocks set to the various clock epochs such as
1749 1902, 1938 or 1970. With this change the logic is now done in PID 1
1750 in addition to timesyncd during early boot-up, so that it is enforced
1751 before the first process is spawned by systemd. Note that the logic
1752 in timesyncd remains, as it is more comprehensive and ensures
1753 clock monotonicity by maintaining a persistent timestamp file in
1754 /var. Since /var is generally not available in earliest boot or the
1755 initrd, this part of the logic remains in timesyncd, and is not done
1758 * Support for tweaking details in net_cls.class_id through the
1759 NetClass= configuration directive has been removed, as the kernel
1760 people have decided to deprecate that controller in cgroup v2.
1761 Userspace tools such as nftables are moving over to setting rules
1762 that are specific to the full cgroup path of a task, which obsoletes
1763 these controllers anyway. The NetClass= directive is kept around for
1764 legacy compatibility reasons. For a more in-depth description of the
1765 kernel change, please refer to the respective upstream commit:
1767 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1769 * A new service setting RuntimeMaxSec= has been added that may be used
1770 to specify a maximum runtime for a service. If the timeout is hit, the
1771 service is terminated and put into a failure state.
1773 * A new service setting AmbientCapabilities= has been added. It allows
1774 configuration of additional Linux process capabilities that are
1775 passed to the activated processes. This is only available on very
1778 * The process resource limit settings in service units may now be used
1779 to configure hard and soft limits individually.
1781 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1782 expose support for gcc's __attribute__((cleanup())) C extension.
1783 Specifically, for many object destructor functions alternative
1784 versions have been added that have names suffixed with "p" and take a
1785 pointer to a pointer to the object to destroy, instead of just a
1786 pointer to the object itself. This is useful because these destructor
1787 functions may be used directly as parameters to the cleanup
1788 construct. Internally, systemd has been a heavy user of this GCC
1789 extension for a long time, and with this change similar support is
1790 now available to consumers of the library outside of systemd. Note
1791 that by using this extension in your sources compatibility with old
1792 and strictly ANSI compatible C compilers is lost. However, all gcc or
1793 LLVM versions of recent years support this extension.
1795 * Timer units gained support for a new setting RandomizedDelaySec= that
1796 allows configuring some additional randomized delay to the configured
1797 time. This is useful to spread out timer events to avoid load peaks in
1798 clusters or larger setups.
1800 * Calendar time specifications now support sub-second accuracy.
1802 * Socket units now support listening on SCTP and UDP-lite protocol
1805 * The sd-event API now comes with a full set of man pages.
1807 * Older versions of systemd contained experimental support for
1808 compressing journal files and coredumps with the LZ4 compressor that
1809 was not compatible with the lz4 binary (due to API limitations of the
1810 lz4 library). This support has been removed; only support for files
1811 compatible with the lz4 binary remains. This LZ4 logic is now
1812 officially supported and no longer considered experimental.
1814 * The dkr image import logic has been removed again from importd. dkr's
1815 micro-services focus doesn't fit into the machine image focus of
1816 importd, and quickly got out of date with the upstream dkr API.
1818 * Creation of the /run/lock/lockdev/ directory was dropped from
1819 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1820 been available for many years. If you still need this, you need to
1821 create your own tmpfiles.d config file with:
1823 d /run/lock/lockdev 0775 root lock -
1825 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1826 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1827 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1828 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1829 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1830 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1831 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1832 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1833 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1834 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1835 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1836 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1837 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1838 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1839 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1840 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1841 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1842 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1843 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1845 — Berlin, 2016-02-11
1849 * A number of properties previously only settable in unit
1850 files are now also available as properties to set when
1851 creating transient units programmatically via the bus, as it
1852 is exposed with systemd-run's --property=
1853 setting. Specifically, these are: SyslogIdentifier=,
1854 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1855 EnvironmentFile=, ReadWriteDirectories=,
1856 ReadOnlyDirectories=, InaccessibleDirectories=,
1857 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1859 * When creating transient services via the bus API it is now
1860 possible to pass in a set of file descriptors to use as
1861 STDIN/STDOUT/STDERR for the invoked process.
1863 * Slice units may now be created transiently via the bus APIs,
1864 similar to the way service and scope units may already be
1865 created transiently.
1867 * Wherever systemd expects a calendar timestamp specification
1868 (like in journalctl's --since= and --until= switches) UTC
1869 timestamps are now supported. Timestamps suffixed with "UTC"
1870 are now considered to be in Universal Time Coordinated
1871 instead of the local timezone. Also, timestamps may now
1872 optionally be specified with sub-second accuracy. Both of
1873 these additions also apply to recurring calendar event
1874 specification, such as OnCalendar= in timer units.
1876 * journalctl gained a new "--sync" switch that asks the
1877 journal daemon to write all so far unwritten log messages to
1878 disk and sync the files, before returning.
1880 * systemd-tmpfiles learned two new line types "q" and "Q" that
1881 operate like "v", but also set up a basic btrfs quota
1882 hierarchy when used on a btrfs file system with quota
1885 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1886 instead of a subvolume (even on a btrfs file system) if the
1887 root directory is a plain directory, and not a
1888 subvolume. This should simplify things with certain chroot()
1889 environments which are not aware of the concept of btrfs
1892 * systemd-detect-virt gained a new --chroot switch to detect
1893 whether execution takes place in a chroot() environment.
1895 * CPUAffinity= now takes CPU index ranges in addition to
1898 * The various memory-related resource limit settings (such as
1899 LimitAS=) now understand the usual K, M, G, ... suffixes to
1900 the base of 1024 (IEC). Similar, the time-related resource
1901 limit settings understand the usual min, h, day, ...
1904 * There's a new system.conf setting DefaultTasksMax= to
1905 control the default TasksMax= setting for services and
1906 scopes running on the system. (TasksMax= is the primary
1907 setting that exposes the "pids" cgroup controller on systemd
1908 and was introduced in the previous systemd release.) The
1909 setting now defaults to 512, which means services that are
1910 not explicitly configured otherwise will only be able to
1911 create 512 processes or threads at maximum, from this
1912 version on. Note that this means that thread- or
1913 process-heavy services might need to be reconfigured to set
1914 TasksMax= to a higher value. It is sufficient to set
1915 TasksMax= in these specific unit files to a higher value, or
1916 even "infinity". Similar, there's now a logind.conf setting
1917 UserTasksMax= that defaults to 4096 and limits the total
1918 number of processes or tasks each user may own
1919 concurrently. nspawn containers also have the TasksMax=
1920 value set by default now, to 8192. Note that all of this
1921 only has an effect if the "pids" cgroup controller is
1922 enabled in the kernel. The general benefit of these changes
1923 should be a more robust and safer system, that provides a
1924 certain amount of per-service fork() bomb protection.
1926 * systemd-nspawn gained the new --network-veth-extra= switch
1927 to define additional and arbitrarily-named virtual Ethernet
1928 links between the host and the container.
1930 * A new service execution setting PassEnvironment= has been
1931 added that allows importing select environment variables
1932 from PID1's environment block into the environment block of
1935 * Timer units gained support for a new RemainAfterElapse=
1936 setting which takes a boolean argument. It defaults to on,
1937 exposing behaviour unchanged to previous releases. If set to
1938 off, timer units are unloaded after they elapsed if they
1939 cannot elapse again. This is particularly useful for
1940 transient timer units, which shall not stay around longer
1941 than until they first elapse.
1943 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1944 default now (the kernel default is 16). This is beneficial
1945 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1946 allows substantially larger numbers of queued
1947 datagrams. This should increase the capability of systemd to
1948 parallelize boot-up, as logging and sd_notify() are unlikely
1949 to stall execution anymore. If you need to change the value
1950 from the new defaults, use the usual sysctl.d/ snippets.
1952 * The compression framing format used by the journal or
1953 coredump processing has changed to be in line with what the
1954 official LZ4 tools generate. LZ4 compression support in
1955 systemd was considered unsupported previously, as the format
1956 was not compatible with the normal tools. With this release
1957 this has changed now, and it is hence safe for downstream
1958 distributions to turn it on. While not compressing as well
1959 as the XZ, LZ4 is substantially faster, which makes
1960 it a good default choice for the compression logic in the
1961 journal and in coredump handling.
1963 * Any reference to /etc/mtab has been dropped from
1964 systemd. The file has been obsolete since a while, but
1965 systemd refused to work on systems where it was incorrectly
1966 set up (it should be a symlink or non-existent). Please make
1967 sure to update to util-linux 2.27.1 or newer in conjunction
1968 with this systemd release, which also drops any reference to
1969 /etc/mtab. If you maintain a distribution make sure that no
1970 software you package still references it, as this is a
1971 likely source of bugs. There's also a glibc bug pending,
1972 asking for removal of any reference to this obsolete file:
1974 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1976 Note that only util-linux versions built with
1977 --enable-libmount-force-mountinfo are supported.
1979 * Support for the ".snapshot" unit type has been removed. This
1980 feature turned out to be little useful and little used, and
1981 has now been removed from the core and from systemctl.
1983 * The dependency types RequiresOverridable= and
1984 RequisiteOverridable= have been removed from systemd. They
1985 have been used only very sparingly to our knowledge and
1986 other options that provide a similar effect (such as
1987 systemctl --mode=ignore-dependencies) are much more useful
1988 and commonly used. Moreover, they were only half-way
1989 implemented as the option to control behaviour regarding
1990 these dependencies was never added to systemctl. By removing
1991 these dependency types the execution engine becomes a bit
1992 simpler. Unit files that use these dependencies should be
1993 changed to use the non-Overridable dependency types
1994 instead. In fact, when parsing unit files with these
1995 options, that's what systemd will automatically convert them
1996 too, but it will also warn, asking users to fix the unit
1997 files accordingly. Removal of these dependency types should
1998 only affect a negligible number of unit files in the wild.
2000 * Behaviour of networkd's IPForward= option changed
2001 (again). It will no longer maintain a per-interface setting,
2002 but propagate one way from interfaces where this is enabled
2003 to the global kernel setting. The global setting will be
2004 enabled when requested by a network that is set up, but
2005 never be disabled again. This change was made to make sure
2006 IPv4 and IPv6 behaviour regarding packet forwarding is
2007 similar (as the Linux IPv6 stack does not support
2008 per-interface control of this setting) and to minimize
2011 * In unit files the behaviour of %u, %U, %h, %s has
2012 changed. These specifiers will now unconditionally resolve
2013 to the various user database fields of the user that the
2014 systemd instance is running as, instead of the user
2015 configured in the specific unit via User=. Note that this
2016 effectively doesn't change much, as resolving of these
2017 specifiers was already turned off in the --system instance
2018 of systemd, as we cannot do NSS lookups from PID 1. In the
2019 --user instance of systemd these specifiers where correctly
2020 resolved, but hardly made any sense, since the user instance
2021 lacks privileges to do user switches anyway, and User= is
2022 hence useless. Moreover, even in the --user instance of
2023 systemd behaviour was awkward as it would only take settings
2024 from User= assignment placed before the specifier into
2025 account. In order to unify and simplify the logic around
2026 this the specifiers will now always resolve to the
2027 credentials of the user invoking the manager (which in case
2028 of PID 1 is the root user).
2030 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2031 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2032 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2033 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2034 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2035 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2036 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2037 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2038 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2039 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2040 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2041 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2042 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2043 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2046 — Berlin, 2015-11-18
2050 * systemd now depends on util-linux v2.27. More specifically,
2051 the newly added mount monitor feature in libmount now
2052 replaces systemd's former own implementation.
2054 * libmount mandates /etc/mtab not to be regular file, and
2055 systemd now enforces this condition at early boot.
2056 /etc/mtab has been deprecated and warned about for a very
2057 long time, so systems running systemd should already have
2058 stopped having this file around as anything else than a
2059 symlink to /proc/self/mounts.
2061 * Support for the "pids" cgroup controller has been added. It
2062 allows accounting the number of tasks in a cgroup and
2063 enforcing limits on it. This adds two new setting
2064 TasksAccounting= and TasksMax= to each unit, as well as a
2065 global option DefaultTasksAccounting=.
2067 * Support for the "net_cls" cgroup controller has been added.
2068 It allows assigning a net class ID to each task in the
2069 cgroup, which can then be used in firewall rules and traffic
2070 shaping configurations. Note that the kernel netfilter net
2071 class code does not currently work reliably for ingress
2072 packets on unestablished sockets.
2074 This adds a new config directive called NetClass= to CGroup
2075 enabled units. Allowed values are positive numbers for fixed
2076 assignments and "auto" for picking a free value
2079 * 'systemctl is-system-running' now returns 'offline' if the
2080 system is not booted with systemd. This command can now be
2081 used as a substitute for 'systemd-notify --booted'.
2083 * Watchdog timeouts have been increased to 3 minutes for all
2084 in-tree service files. Apparently, disk IO issues are more
2085 frequent than we hoped, and user reported >1 minute waiting
2088 * 'machine-id-commit' functionality has been merged into
2089 'machine-id-setup --commit'. The separate binary has been
2092 * The WorkingDirectory= directive in unit files may now be set
2093 to the special value '~'. In this case, the working
2094 directory is set to the home directory of the user
2095 configured in User=.
2097 * "machinectl shell" will now open the shell in the home
2098 directory of the selected user by default.
2100 * The CrashChVT= configuration file setting is renamed to
2101 CrashChangeVT=, following our usual logic of not
2102 abbreviating unnecessarily. The old directive is still
2103 supported for compat reasons. Also, this directive now takes
2104 an integer value between 1 and 63, or a boolean value. The
2105 formerly supported '-1' value for disabling stays around for
2108 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2109 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2110 RootDirectory= properties can now be set for transient
2113 * The systemd-analyze tool gained a new "set-log-target" verb
2114 to change the logging target the system manager logs to
2115 dynamically during runtime. This is similar to how
2116 "systemd-analyze set-log-level" already changes the log
2119 * In nspawn /sys is now mounted as tmpfs, with only a selected
2120 set of subdirectories mounted in from the real sysfs. This
2121 enhances security slightly, and is useful for ensuring user
2122 namespaces work correctly.
2124 * Support for USB FunctionFS activation has been added. This
2125 allows implementation of USB gadget services that are
2126 activated as soon as they are requested, so that they don't
2127 have to run continuously, similar to classic socket
2130 * The "systemctl exit" command now optionally takes an
2131 additional parameter that sets the exit code to return from
2132 the systemd manager when exiting. This is only relevant when
2133 running the systemd user instance, or when running the
2134 system instance in a container.
2136 * sd-bus gained the new API calls sd_bus_path_encode_many()
2137 and sd_bus_path_decode_many() that allow easy encoding and
2138 decoding of multiple identifier strings inside a D-Bus
2139 object path. Another new call sd_bus_default_flush_close()
2140 has been added to flush and close per-thread default
2143 * systemd-cgtop gained support for a -M/--machine= switch to
2144 show the control groups within a certain container only.
2146 * "systemctl kill" gained support for an optional --fail
2147 switch. If specified the requested operation will fail of no
2148 processes have been killed, because the unit had no
2149 processes attached, or similar.
2151 * A new systemd.crash_reboot=1 kernel command line option has
2152 been added that triggers a reboot after crashing. This can
2153 also be set through CrashReboot= in systemd.conf.
2155 * The RuntimeDirectory= setting now understands unit
2156 specifiers like %i or %f.
2158 * A new (still internal) library API sd-ipv4acd has been added,
2159 that implements address conflict detection for IPv4. It's
2160 based on code from sd-ipv4ll, and will be useful for
2161 detecting DHCP address conflicts.
2163 * File descriptors passed during socket activation may now be
2164 named. A new API sd_listen_fds_with_names() is added to
2165 access the names. The default names may be overridden,
2166 either in the .socket file using the FileDescriptorName=
2167 parameter, or by passing FDNAME= when storing the file
2168 descriptors using sd_notify().
2170 * systemd-networkd gained support for:
2172 - Setting the IPv6 Router Advertisement settings via
2173 IPv6AcceptRouterAdvertisements= in .network files.
2175 - Configuring the HelloTimeSec=, MaxAgeSec= and
2176 ForwardDelaySec= bridge parameters in .netdev files.
2178 - Configuring PreferredSource= for static routes in
2181 * The "ask-password" framework used to query for LUKS harddisk
2182 passwords or SSL passwords during boot gained support for
2183 caching passwords in the kernel keyring, if it is
2184 available. This makes sure that the user only has to type in
2185 a passphrase once if there are multiple objects to unlock
2186 with the same one. Previously, such password caching was
2187 available only when Plymouth was used; this moves the
2188 caching logic into the systemd codebase itself. The
2189 "systemd-ask-password" utility gained a new --keyname=
2190 switch to control which kernel keyring key to use for
2191 caching a password in. This functionality is also useful for
2192 enabling display managers such as gdm to automatically
2193 unlock the user's GNOME keyring if its passphrase, the
2194 user's password and the harddisk password are the same, if
2195 gdm-autologin is used.
2197 * When downloading tar or raw images using "machinectl
2198 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2199 file is now also downloaded, if it is available and stored
2200 next to the image file.
2202 * Units of type ".socket" gained a new boolean setting
2203 Writable= which is only useful in conjunction with
2204 ListenSpecial=. If true, enables opening the specified
2205 special file in O_RDWR mode rather than O_RDONLY mode.
2207 * systemd-rfkill has been reworked to become a singleton
2208 service that is activated through /dev/rfkill on each rfkill
2209 state change and saves the settings to disk. This way,
2210 systemd-rfkill is now compatible with devices that exist
2211 only intermittendly, and even restores state if the previous
2212 system shutdown was abrupt rather than clean.
2214 * The journal daemon gained support for vacuuming old journal
2215 files controlled by the number of files that shall remain,
2216 in addition to the already existing control by size and by
2217 date. This is useful as journal interleaving performance
2218 degrades with too many separate journal files, and allows
2219 putting an effective limit on them. The new setting defaults
2220 to 100, but this may be changed by setting SystemMaxFiles=
2221 and RuntimeMaxFiles= in journald.conf. Also, the
2222 "journalctl" tool gained the new --vacuum-files= switch to
2223 manually vacuum journal files to leave only the specified
2224 number of files in place.
2226 * udev will now create /dev/disk/by-path links for ATA devices
2227 on kernels where that is supported.
2229 * Galician, Serbian, Turkish and Korean translations were added.
2231 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2232 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2233 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2234 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2235 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2236 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2237 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2238 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2239 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2240 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2241 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2242 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2243 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2244 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2245 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2246 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2247 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2248 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2250 — Berlin, 2015-10-07
2254 * The DHCP implementation of systemd-networkd gained a set of
2257 - The DHCP server now supports emitting DNS and NTP
2258 information. It may be enabled and configured via
2259 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2260 and NTP information is enabled, but no servers are
2261 configured, the corresponding uplink information (if there
2262 is any) is propagated.
2264 - Server and client now support transmission and reception
2265 of timezone information. It can be configured via the
2266 newly introduced network options UseTimezone=,
2267 EmitTimezone=, and Timezone=. Transmission of timezone
2268 information is enabled between host and containers by
2269 default now: the container will change its local timezone
2270 to what the host has set.
2272 - Lease timeouts can now be configured via
2273 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2275 - The DHCP server improved on the stability of
2276 leases. Clients are more likely to get the same lease
2277 information back, even if the server loses state.
2279 - The DHCP server supports two new configuration options to
2280 control the lease address pool metrics, PoolOffset= and
2283 * The encapsulation limit of tunnels in systemd-networkd may
2284 now be configured via 'EncapsulationLimit='. It allows
2285 modifying the maximum additional levels of encapsulation
2286 that are permitted to be prepended to a packet.
2288 * systemd now supports the concept of user buses replacing
2289 session buses, if used with dbus-1.10 (and enabled via dbus
2290 --enable-user-session). It previously only supported this on
2291 kdbus-enabled systems, and this release expands this to
2292 'dbus-daemon' systems.
2294 * systemd-networkd now supports predictable interface names
2297 * systemd now optionally supports the new Linux kernel
2298 "unified" control group hierarchy. If enabled via the kernel
2299 command-line option 'systemd.unified_cgroup_hierarchy=1',
2300 systemd will try to mount the unified cgroup hierarchy
2301 directly on /sys/fs/cgroup. If not enabled, or not
2302 available, systemd will fall back to the legacy cgroup
2303 hierarchy setup, as before. Host system and containers can
2304 mix and match legacy and unified hierarchies as they
2305 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2306 environment variable to individually select the hierarchy to
2307 use for executed containers. By default, nspawn will use the
2308 unified hierarchy for the containers if the host uses the
2309 unified hierarchy, and the legacy hierarchy otherwise.
2310 Please note that at this point the unified hierarchy is an
2311 experimental kernel feature and is likely to change in one
2312 of the next kernel releases. Therefore, it should not be
2313 enabled by default in downstream distributions yet. The
2314 minimum required kernel version for the unified hierarchy to
2315 work is 4.2. Note that when the unified hierarchy is used
2316 for the first time delegated access to controllers is
2317 safe. Because of this systemd-nspawn containers will get
2318 access to controllers now, as will systemd user
2319 sessions. This means containers and user sessions may now
2320 manage their own resources, partitioning up what the system
2323 * A new special scope unit "init.scope" has been introduced
2324 that encapsulates PID 1 of the system. It may be used to
2325 determine resource usage and enforce resource limits on PID
2326 1 itself. PID 1 hence moved out of the root of the control
2329 * The cgtop tool gained support for filtering out kernel
2330 threads when counting tasks in a control group. Also, the
2331 count of processes is now recursively summed up by
2332 default. Two options -k and --recursive= have been added to
2333 revert to old behaviour. The tool has also been updated to
2334 work correctly in containers now.
2336 * systemd-nspawn's --bind= and --bind-ro= options have been
2337 extended to allow creation of non-recursive bind mounts.
2339 * libsystemd gained two new calls sd_pid_get_cgroup() and
2340 sd_peer_get_cgroup() which return the control group path of
2341 a process or peer of a connected AF_UNIX socket. This
2342 function call is particularly useful when implementing
2343 delegated subtrees support in the control group hierarchy.
2345 * The "sd-event" event loop API of libsystemd now supports
2346 correct dequeuing of real-time signals, without losing
2349 * When systemd requests a PolicyKit decision when managing
2350 units it will now add additional fields to the request,
2351 including unit name and desired operation. This enables more
2352 powerful PolicyKit policies, that make decisions depending
2353 on these parameters.
2355 * nspawn learnt support for .nspawn settings files, that may
2356 accompany the image files or directories of containers, and
2357 may contain additional settings for the container. This is
2358 an alternative to configuring container parameters via the
2359 nspawn command line.
2361 Contributions from: Cristian Rodríguez, Daniel Mack, David
2362 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2363 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2364 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2365 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2366 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2367 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2368 Andersen, Tom Gundersen, Torstein Husebø
2370 — Berlin, 2015-09-08
2374 * machinectl gained a new verb 'shell' which opens a fresh
2375 shell on the target container or the host. It is similar to
2376 the existing 'login' command of machinectl, but spawns the
2377 shell directly without prompting for username or
2378 password. The pseudo machine '.host' now refers to the local
2379 host and is used by default. Hence, 'machinectl shell' can
2380 be used as replacement for 'su -' which spawns a session as
2381 a fresh systemd unit in a way that is fully isolated from
2382 the originating session.
2384 * systemd-networkd learned to cope with private-zone DHCP
2385 options and allows other programs to query the values.
2387 * SELinux access control when enabling/disabling units is no
2388 longer enforced with this release. The previous
2389 implementation was incorrect, and a new corrected
2390 implementation is not yet available. As unit file operations
2391 are still protected via PolicyKit and D-Bus policy this is
2392 not a security problem. Yet, distributions which care about
2393 optimal SELinux support should probably not stabilize on
2396 * sd-bus gained support for matches of type "arg0has=", that
2397 test for membership of strings in string arrays sent in bus
2400 * systemd-resolved now dumps the contents of its DNS and LLMNR
2401 caches to the logs on reception of the SIGUSR1 signal. This
2402 is useful to debug DNS behaviour.
2404 * The coredumpctl tool gained a new --directory= option to
2405 operate on journal files in a specific directory.
2407 * "systemctl reboot" and related commands gained a new
2408 "--message=" option which may be used to set a free-text
2409 wall message when shutting down or rebooting the
2410 system. This message is also logged, which is useful for
2411 figuring out the reason for a reboot or shutdown a
2414 * The "systemd-resolve-host" tool's -i switch now takes
2415 network interface numbers as alternative to interface names.
2417 * A new unit file setting for services has been introduced:
2418 UtmpMode= allows configuration of how precisely systemd
2419 handles utmp and wtmp entries for the service if this is
2420 enabled. This allows writing services that appear similar to
2421 user sessions in the output of the "w", "who", "last" and
2424 * systemd-resolved will now locally synthesize DNS resource
2425 records for the "localhost" and "gateway" domains as well as
2426 the local hostname. This should ensure that clients querying
2427 RRs via resolved will get similar results as those going via
2428 NSS, if nss-myhostname is enabled.
2430 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2431 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2432 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2433 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2434 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2435 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2436 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2437 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2438 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2439 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2440 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2441 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2443 — Berlin, 2015-08-27
2447 * The systemd-efi-boot-generator functionality was merged into
2448 systemd-gpt-auto-generator.
2450 * systemd-networkd now supports Group Policy for vxlan
2451 devices. It can be enabled via the new boolean configuration
2452 option called 'GroupPolicyExtension='.
2454 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2455 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2456 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2458 — Berlin, 2015-07-31
2462 * The python-systemd code has been removed from the systemd repository.
2463 A new repository has been created which accommodates the code from
2464 now on, and we kindly ask distributions to create a separate package
2465 for this: https://github.com/systemd/python-systemd
2467 * The systemd daemon will now reload its main configuration
2468 (/etc/systemd/system.conf) on daemon-reload.
2470 * sd-dhcp now exposes vendor specific extensions via
2471 sd_dhcp_lease_get_vendor_specific().
2473 * systemd-networkd gained a number of new configuration options.
2475 - A new boolean configuration option for TAP devices called
2476 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2477 device, thus allowing to send and receive GSO packets.
2479 - A new tunnel configuration option called 'CopyDSCP='.
2480 If enabled, the DSCP field of ip6 tunnels is copied into the
2481 decapsulated packet.
2483 - A set of boolean bridge configuration options were added.
2484 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2485 and 'UnicastFlood=' are now parsed by networkd and applied to the
2486 respective bridge link device via the respective IFLA_BRPORT_*
2489 - A new string configuration option to override the hostname sent
2490 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2491 is true, networkd will use the configured hostname instead of the
2492 system hostname when sending DHCP requests.
2494 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2495 networkd will configure the IPv6 flow-label of the tunnel device
2496 according to RFC2460.
2498 - The 'macvtap' virtual network devices are now supported, similar to
2499 the already supported 'macvlan' devices.
2501 * systemd-resolved now implements RFC5452 to improve resilience against
2502 cache poisoning. Additionally, source port randomization is enabled
2503 by default to further protect against DNS spoofing attacks.
2505 * nss-mymachines now supports translating UIDs and GIDs of running
2506 containers with user-namespaces enabled. If a container 'foo'
2507 translates a host uid 'UID' to the container uid 'TUID', then
2508 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2509 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2510 mapped as 'vg-foo-TGID'.
2512 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2513 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2514 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2515 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2516 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2517 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2518 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2519 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2520 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2521 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2523 — Berlin, 2015-07-29
2527 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2528 There are no known issues with current sysfs, and udev does not need
2529 or should be used to work around such bugs.
2531 * udev does no longer enable USB HID power management. Several reports
2532 indicate, that some devices cannot handle that setting.
2534 * The udev accelerometer helper was removed. The functionality
2535 is now fully included in iio-sensor-proxy. But this means,
2536 older iio-sensor-proxy versions will no longer provide
2537 accelerometer/orientation data with this systemd version.
2538 Please upgrade iio-sensor-proxy to version 1.0.
2540 * networkd gained a new configuration option IPv6PrivacyExtensions=
2541 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2542 for Stateless Address") on selected networks.
2544 * For the sake of fewer build-time dependencies and less code in the
2545 main repository, the python bindings are about to be removed in the
2546 next release. A new repository has been created which accommodates
2547 the code from now on, and we kindly ask distributions to create a
2548 separate package for this. The removal will take place in v223.
2550 https://github.com/systemd/python-systemd
2552 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2553 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2554 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2555 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2556 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2557 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2558 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2559 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2560 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2561 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2563 — Berlin, 2015-07-07
2567 * The sd-bus.h and sd-event.h APIs have now been declared
2568 stable and have been added to the official interface of
2569 libsystemd.so. sd-bus implements an alternative D-Bus client
2570 library, that is relatively easy to use, very efficient and
2571 supports both classic D-Bus as well as kdbus as transport
2572 backend. sd-event is a generic event loop abstraction that
2573 is built around Linux epoll, but adds features such as event
2574 prioritization or efficient timer handling. Both APIs are good
2575 choices for C programs looking for a bus and/or event loop
2576 implementation that is minimal and does not have to be
2577 portable to other kernels.
2579 * kdbus support is no longer compile-time optional. It is now
2580 always built-in. However, it can still be disabled at
2581 runtime using the kdbus=0 kernel command line setting, and
2582 that setting may be changed to default to off, by specifying
2583 --disable-kdbus at build-time. Note though that the kernel
2584 command line setting has no effect if the kdbus.ko kernel
2585 module is not installed, in which case kdbus is (obviously)
2586 also disabled. We encourage all downstream distributions to
2587 begin testing kdbus by adding it to the kernel images in the
2588 development distributions, and leaving kdbus support in
2591 * The minimal required util-linux version has been bumped to
2594 * Support for chkconfig (--enable-chkconfig) was removed in
2595 favor of calling an abstraction tool
2596 /lib/systemd/systemd-sysv-install. This needs to be
2597 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2598 in README for details.
2600 * If there's a systemd unit and a SysV init script for the
2601 same service name, and the user executes "systemctl enable"
2602 for it (or a related call), then this will now enable both
2603 (or execute the related operation on both), not just the
2606 * The libudev API documentation has been converted from gtkdoc
2609 * gudev has been removed from the systemd tree, it is now an
2612 * The systemd-cgtop tool learnt a new --raw switch to generate
2613 "raw" (machine parsable) output.
2615 * networkd's IPForwarding= .network file setting learnt the
2616 new setting "kernel", which ensures that networkd does not
2617 change the IP forwarding sysctl from the default kernel
2620 * The systemd-logind bus API now exposes a new boolean
2621 property "Docked" that reports whether logind considers the
2622 system "docked", i.e. connected to a docking station or not.
2624 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2625 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2626 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2627 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2628 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2629 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2630 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2631 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2632 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2633 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2634 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2635 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2636 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2637 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2638 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2639 Fink, Zbigniew Jędrzejewski-Szmek
2641 — Berlin, 2015-06-19
2645 * The gudev library has been extracted into a separate repository
2646 available at: https://git.gnome.org/browse/libgudev/
2647 It is now managed as part of the Gnome project. Distributions
2648 are recommended to pass --disable-gudev to systemd and use
2649 gudev from the Gnome project instead. gudev is still included
2650 in systemd, for now. It will be removed soon, though. Please
2651 also see the announcement-thread on systemd-devel:
2652 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2654 * systemd now exposes a CPUUsageNSec= property for each
2655 service unit on the bus, that contains the overall consumed
2656 CPU time of a service (the sum of what each process of the
2657 service consumed). This value is only available if
2658 CPUAccounting= is turned on for a service, and is then shown
2659 in the "systemctl status" output.
2661 * Support for configuring alternative mappings of the old SysV
2662 runlevels to systemd targets has been removed. They are now
2663 hardcoded in a way that runlevels 2, 3, 4 all map to
2664 multi-user.target and 5 to graphical.target (which
2665 previously was already the default behaviour).
2667 * The auto-mounter logic gained support for mount point
2668 expiry, using a new TimeoutIdleSec= setting in .automount
2669 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2671 * The EFI System Partition (ESP) as mounted to /boot by
2672 systemd-efi-boot-generator will now be unmounted
2673 automatically after 2 minutes of not being used. This should
2674 minimize the risk of ESP corruptions.
2676 * New /etc/fstab options x-systemd.requires= and
2677 x-systemd.requires-mounts-for= are now supported to express
2678 additional dependencies for mounts. This is useful for
2679 journalling file systems that support external journal
2680 devices or overlay file systems that require underlying file
2681 systems to be mounted.
2683 * systemd does not support direct live-upgrades (via systemctl
2684 daemon-reexec) from versions older than v44 anymore. As no
2685 distribution we are aware of shipped such old versions in a
2686 stable release this should not be problematic.
2688 * When systemd forks off a new per-connection service instance
2689 it will now set the $REMOTE_ADDR environment variable to the
2690 remote IP address, and $REMOTE_PORT environment variable to
2691 the remote IP port. This behaviour is similar to the
2692 corresponding environment variables defined by CGI.
2694 * systemd-networkd gained support for uplink failure
2695 detection. The BindCarrier= option allows binding interface
2696 configuration dynamically to the link sense of other
2697 interfaces. This is useful to achieve behaviour like in
2700 * systemd-networkd gained support for configuring the DHCP
2701 client identifier to use when requesting leases.
2703 * systemd-networkd now has a per-network UseNTP= option to
2704 configure whether NTP server information acquired via DHCP
2705 is passed on to services like systemd-timesyncd.
2707 * systemd-networkd gained support for vti6 tunnels.
2709 * Note that systemd-networkd manages the sysctl variable
2710 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2711 it is configured for since v219. The variable controls IP
2712 forwarding, and is a per-interface alternative to the global
2713 /proc/sys/net/ipv[46]/ip_forward. This setting is
2714 configurable in the IPForward= option, which defaults to
2715 "no". This means if networkd is used for an interface it is
2716 no longer sufficient to set the global sysctl option to turn
2717 on IP forwarding! Instead, the .network file option
2718 IPForward= needs to be turned on! Note that the
2719 implementation of this behaviour was broken in v219 and has
2722 * Many bonding and vxlan options are now configurable in
2725 * systemd-nspawn gained a new --property= setting to set unit
2726 properties for the container scope. This is useful for
2727 setting resource parameters (e.g. "CPUShares=500") on
2728 containers started from the command line.
2730 * systemd-nspawn gained a new --private-users= switch to make
2731 use of user namespacing available on recent Linux kernels.
2733 * systemd-nspawn may now be called as part of a shell pipeline
2734 in which case the pipes used for stdin and stdout are passed
2735 directly to the process invoked in the container, without
2736 indirection via a pseudo tty.
2738 * systemd-nspawn gained a new switch to control the UNIX
2739 signal to use when killing the init process of the container
2742 * systemd-nspawn gained a new --overlay= switch for mounting
2743 overlay file systems into the container using the new kernel
2746 * When a container image is imported via systemd-importd and
2747 the host file system is not btrfs, a loopback block device
2748 file is created in /var/lib/machines.raw with a btrfs file
2749 system inside. It is then mounted to /var/lib/machines to
2750 enable btrfs features for container management. The loopback
2751 file and btrfs file system is grown as needed when container
2752 images are imported via systemd-importd.
2754 * systemd-machined/systemd-importd gained support for btrfs
2755 quota, to enforce container disk space limits on disk. This
2756 is exposed in "machinectl set-limit".
2758 * systemd-importd now can import containers from local .tar,
2759 .raw and .qcow2 images, and export them to .tar and .raw. It
2760 can also import dkr v2 images now from the network (on top
2763 * systemd-importd gained support for verifying downloaded
2764 images with gpg2 (previously only gpg1 was supported).
2766 * systemd-machined, systemd-logind, systemd: most bus calls
2767 are now accessible to unprivileged processes via
2768 PolicyKit. Also, systemd-logind will now allow users to kill
2769 their own sessions without further privileges or
2772 * systemd-shutdownd has been removed. This service was
2773 previously responsible for implementing scheduled shutdowns
2774 as exposed in /usr/bin/shutdown's time parameter. This
2775 functionality has now been moved into systemd-logind and is
2776 accessible via a bus interface.
2778 * "systemctl reboot" gained a new switch --firmware-setup that
2779 can be used to reboot into the EFI firmware setup, if that
2780 is available. systemd-logind now exposes an API on the bus
2781 to trigger such reboots, in case graphical desktop UIs want
2782 to cover this functionality.
2784 * "systemctl enable", "systemctl disable" and "systemctl mask"
2785 now support a new "--now" switch. If specified the units
2786 that are enabled will also be started, and the ones
2787 disabled/masked also stopped.
2789 * The Gummiboot EFI boot loader tool has been merged into
2790 systemd, and renamed to "systemd-boot". The bootctl tool has been
2791 updated to support systemd-boot.
2793 * An EFI kernel stub has been added that may be used to create
2794 kernel EFI binaries that contain not only the actual kernel,
2795 but also an initrd, boot splash, command line and OS release
2796 information. This combined binary can then be signed as a
2797 single image, so that the firmware can verify it all in one
2798 step. systemd-boot has special support for EFI binaries created
2799 like this and can extract OS release information from them
2800 and show them in the boot menu. This functionality is useful
2801 to implement cryptographically verified boot schemes.
2803 * Optional support has been added to systemd-fsck to pass
2804 fsck's progress report to an AF_UNIX socket in the file
2807 * udev will no longer create device symlinks for all block
2808 devices by default. A blacklist for excluding special block
2809 devices from this logic has been turned into a whitelist
2810 that requires picking block devices explicitly that require
2813 * A new (currently still internal) API sd-device.h has been
2814 added to libsystemd. This modernized API is supposed to
2815 replace libudev eventually. In fact, already much of libudev
2816 is now just a wrapper around sd-device.h.
2818 * A new hwdb database for storing metadata about pointing
2819 stick devices has been added.
2821 * systemd-tmpfiles gained support for setting file attributes
2822 similar to the "chattr" tool with new 'h' and 'H' lines.
2824 * systemd-journald will no longer unconditionally set the
2825 btrfs NOCOW flag on new journal files. This is instead done
2826 with tmpfiles snippet using the new 'h' line type. This
2827 allows easy disabling of this logic, by masking the
2828 journal-nocow.conf tmpfiles file.
2830 * systemd-journald will now translate audit message types to
2831 human readable identifiers when writing them to the
2832 journal. This should improve readability of audit messages.
2834 * The LUKS logic gained support for the offset= and skip=
2835 options in /etc/crypttab, as previously implemented by
2838 * /usr/lib/os-release gained a new optional field VARIANT= for
2839 distributions that support multiple variants (such as a
2840 desktop edition, a server edition, ...)
2842 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2843 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2844 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2845 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2846 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2847 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2848 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2849 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2850 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2851 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2852 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2853 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2854 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2855 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2856 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2857 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2858 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2859 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2860 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2861 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2862 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2863 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2864 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2865 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2866 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2867 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2868 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2870 — Berlin, 2015-05-22
2874 * Introduce a new API "sd-hwdb.h" for querying the hardware
2875 metadata database. With this minimal interface one can query
2876 and enumerate the udev hwdb, decoupled from the old libudev
2877 library. libudev's interface for this is now only a wrapper
2878 around sd-hwdb. A new tool systemd-hwdb has been added to
2879 interface with and update the database.
2881 * When any of systemd's tools copies files (for example due to
2882 tmpfiles' C lines) a btrfs reflink will attempted first,
2883 before bytewise copying is done.
2885 * systemd-nspawn gained a new --ephemeral switch. When
2886 specified a btrfs snapshot is taken of the container's root
2887 directory, and immediately removed when the container
2888 terminates again. Thus, a container can be started whose
2889 changes never alter the container's root directory, and are
2890 lost on container termination. This switch can also be used
2891 for starting a container off the root file system of the
2892 host without affecting the host OS. This switch is only
2893 available on btrfs file systems.
2895 * systemd-nspawn gained a new --template= switch. It takes the
2896 path to a container tree to use as template for the tree
2897 specified via --directory=, should that directory be
2898 missing. This allows instantiating containers dynamically,
2899 on first run. This switch is only available on btrfs file
2902 * When a .mount unit refers to a mount point on which multiple
2903 mounts are stacked, and the .mount unit is stopped all of
2904 the stacked mount points will now be unmounted until no
2905 mount point remains.
2907 * systemd now has an explicit notion of supported and
2908 unsupported unit types. Jobs enqueued for unsupported unit
2909 types will now fail with an "unsupported" error code. More
2910 specifically .swap, .automount and .device units are not
2911 supported in containers, .busname units are not supported on
2912 non-kdbus systems. .swap and .automount are also not
2913 supported if their respective kernel compile time options
2916 * machinectl gained support for two new "copy-from" and
2917 "copy-to" commands for copying files from a running
2918 container to the host or vice versa.
2920 * machinectl gained support for a new "bind" command to bind
2921 mount host directories into local containers. This is
2922 currently only supported for nspawn containers.
2924 * networkd gained support for configuring bridge forwarding
2925 database entries (fdb) from .network files.
2927 * A new tiny daemon "systemd-importd" has been added that can
2928 download container images in tar, raw, qcow2 or dkr formats,
2929 and make them available locally in /var/lib/machines, so
2930 that they can run as nspawn containers. The daemon can GPG
2931 verify the downloads (not supported for dkr, since it has no
2932 provisions for verifying downloads). It will transparently
2933 decompress bz2, xz, gzip compressed downloads if necessary,
2934 and restore sparse files on disk. The daemon uses privilege
2935 separation to ensure the actual download logic runs with
2936 fewer privileges than the daemon itself. machinectl has
2937 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2938 make the functionality of importd available to the
2939 user. With this in place the Fedora and Ubuntu "Cloud"
2940 images can be downloaded and booted as containers unmodified
2941 (the Fedora images lack the appropriate GPG signature files
2942 currently, so they cannot be verified, but this will change
2943 soon, hopefully). Note that downloading images is currently
2944 only fully supported on btrfs.
2946 * machinectl is now able to list container images found in
2947 /var/lib/machines, along with some metadata about sizes of
2948 disk and similar. If the directory is located on btrfs and
2949 quota is enabled, this includes quota display. A new command
2950 "image-status" has been added that shows additional
2951 information about images.
2953 * machinectl is now able to clone container images
2954 efficiently, if the underlying file system (btrfs) supports
2955 it, with the new "machinectl clone" command. It also
2956 gained commands for renaming and removing images, as well as
2957 marking them read-only or read-write (supported also on
2958 legacy file systems).
2960 * networkd gained support for collecting LLDP network
2961 announcements, from hardware that supports this. This is
2962 shown in networkctl output.
2964 * systemd-run gained support for a new -t (--pty) switch for
2965 invoking a binary on a pty whose input and output is
2966 connected to the invoking terminal. This allows executing
2967 processes as system services while interactively
2968 communicating with them via the terminal. Most interestingly
2969 this is supported across container boundaries. Invoking
2970 "systemd-run -t /bin/bash" is an alternative to running a
2971 full login session, the difference being that the former
2972 will not register a session, nor go through the PAM session
2975 * tmpfiles gained support for a new "v" line type for creating
2976 btrfs subvolumes. If the underlying file system is a legacy
2977 file system, this automatically degrades to creating a
2978 normal directory. Among others /var/lib/machines is now
2979 created like this at boot, should it be missing.
2981 * The directory /var/lib/containers/ has been deprecated and
2982 been replaced by /var/lib/machines. The term "machines" has
2983 been used in the systemd context as generic term for both
2984 VMs and containers, and hence appears more appropriate for
2985 this, as the directory can also contain raw images bootable
2988 * systemd-nspawn when invoked with -M but without --directory=
2989 or --image= is now capable of searching for the container
2990 root directory, subvolume or disk image automatically, in
2991 /var/lib/machines. systemd-nspawn@.service has been updated
2992 to make use of this, thus allowing it to be used for raw
2995 * A new machines.target unit has been introduced that is
2996 supposed to group all containers/VMs invoked as services on
2997 the system. systemd-nspawn@.service has been updated to
2998 integrate with that.
3000 * machinectl gained a new "start" command, for invoking a
3001 container as a service. "machinectl start foo" is mostly
3002 equivalent to "systemctl start systemd-nspawn@foo.service",
3003 but handles escaping in a nicer way.
3005 * systemd-nspawn will now mount most of the cgroupfs tree
3006 read-only into each container, with the exception of the
3007 container's own subtree in the name=systemd hierarchy.
3009 * journald now sets the special FS_NOCOW file flag for its
3010 journal files. This should improve performance on btrfs, by
3011 avoiding heavy fragmentation when journald's write-pattern
3012 is used on COW file systems. It degrades btrfs' data
3013 integrity guarantees for the files to the same levels as for
3014 ext3/ext4 however. This should be OK though as journald does
3015 its own data integrity checks and all its objects are
3016 checksummed on disk. Also, journald should handle btrfs disk
3017 full events a lot more gracefully now, by processing SIGBUS
3018 errors, and not relying on fallocate() anymore.
3020 * When journald detects that journal files it is writing to
3021 have been deleted it will immediately start new journal
3024 * systemd now provides a way to store file descriptors
3025 per-service in PID 1. This is useful for daemons to ensure
3026 that fds they require are not lost during a daemon
3027 restart. The fds are passed to the daemon on the next
3028 invocation in the same way socket activation fds are
3029 passed. This is now used by journald to ensure that the
3030 various sockets connected to all the system's stdout/stderr
3031 are not lost when journald is restarted. File descriptors
3032 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3033 an extension to sd_notify(). Note that a limit is enforced
3034 on the number of fds a service can store in PID 1, and it
3035 defaults to 0, so that no fds may be stored, unless this is
3036 explicitly turned on.
3038 * The default TERM variable to use for units connected to a
3039 terminal, when no other value is explicitly is set is now
3040 vt220 rather than vt102. This should be fairly safe still,
3041 but allows PgUp/PgDn work.
3043 * The /etc/crypttab option header= as known from Debian is now
3046 * "loginctl user-status" and "loginctl session-status" will
3047 now show the last 10 lines of log messages of the
3048 user/session following the status output. Similar,
3049 "machinectl status" will show the last 10 log lines
3050 associated with a virtual machine or container
3051 service. (Note that this is usually not the log messages
3052 done in the VM/container itself, but simply what the
3053 container manager logs. For nspawn this includes all console
3056 * "loginctl session-status" without further argument will now
3057 show the status of the session of the caller. Similar,
3058 "lock-session", "unlock-session", "activate",
3059 "enable-linger", "disable-linger" may now be called without
3060 session/user parameter in which case they apply to the
3061 caller's session/user.
3063 * An X11 session scriptlet is now shipped that uploads
3064 $DISPLAY and $XAUTHORITY into the environment of the systemd
3065 --user daemon if a session begins. This should improve
3066 compatibility with X11 enabled applications run as systemd
3069 * Generators are now subject to masking via /etc and /run, the
3070 same way as unit files.
3072 * networkd .network files gained support for configuring
3073 per-link IPv4/IPv6 packet forwarding as well as IPv4
3074 masquerading. This is by default turned on for veth links to
3075 containers, as registered by systemd-nspawn. This means that
3076 nspawn containers run with --network-veth will now get
3077 automatic routed access to the host's networks without any
3078 further configuration or setup, as long as networkd runs on
3081 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3082 or UDP posts of a container on the host. With this in place
3083 it is possible to run containers with private veth links
3084 (--network-veth), and have their functionality exposed on
3085 the host as if their services were running directly on the
3088 * systemd-nspawn's --network-veth switch now gained a short
3089 version "-n", since with the changes above it is now truly
3090 useful out-of-the-box. The systemd-nspawn@.service has been
3091 updated to make use of it too by default.
3093 * systemd-nspawn will now maintain a per-image R/W lock, to
3094 ensure that the same image is not started more than once
3095 writable. (It's OK to run an image multiple times
3096 simultaneously in read-only mode.)
3098 * systemd-nspawn's --image= option is now capable of
3099 dissecting and booting MBR and GPT disk images that contain
3100 only a single active Linux partition. Previously it
3101 supported only GPT disk images with proper GPT type
3102 IDs. This allows running cloud images from major
3103 distributions directly with systemd-nspawn, without
3106 * In addition to collecting mouse dpi data in the udev
3107 hardware database, there's now support for collecting angle
3108 information for mouse scroll wheels. The database is
3109 supposed to guarantee similar scrolling behavior on mice
3110 that it knows about. There's also support for collecting
3111 information about Touchpad types.
3113 * udev's input_id built-in will now also collect touch screen
3114 dimension data and attach it to probed devices.
3116 * /etc/os-release gained support for a Distribution Privacy
3119 * networkd gained support for creating "ipvlan", "gretap",
3120 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3122 * systemd-tmpfiles gained support for "a" lines for setting
3125 * systemd-nspawn will now mount /tmp in the container to
3126 tmpfs, automatically.
3128 * systemd now exposes the memory.usage_in_bytes cgroup
3129 attribute and shows it for each service in the "systemctl
3130 status" output, if available.
3132 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3133 immediate reboot is triggered. This useful if shutdown is
3134 hung and is unable to complete, to expedite the
3135 operation. Note that this kind of reboot will still unmount
3136 all file systems, and hence should not result in fsck being
3139 * A .device unit for an optical block device will now be
3140 considered active only when a medium is in the drive. Also,
3141 mount units are now bound to their backing devices thus
3142 triggering automatic unmounting when devices become
3143 unavailable. With this in place systemd will now
3144 automatically unmount left-over mounts when a CD-ROM is
3145 ejected or an USB stick is yanked from the system.
3147 * networkd-wait-online now has support for waiting for
3148 specific interfaces only (with globbing), and for giving up
3149 after a configurable timeout.
3151 * networkd now exits when idle. It will be automatically
3152 restarted as soon as interfaces show up, are removed or
3153 change state. networkd will stay around as long as there is
3154 at least one DHCP state machine or similar around, that keep
3157 * networkd may now configure IPv6 link-local addressing in
3158 addition to IPv4 link-local addressing.
3160 * The IPv6 "token" for use in SLAAC may now be configured for
3161 each .network interface in networkd.
3163 * Routes configured with networkd may now be assigned a scope
3166 * networkd's [Match] sections now support globbing and lists
3167 of multiple space-separated matches per item.
3169 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3170 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3171 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3172 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3173 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3174 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3175 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3176 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3177 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3178 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3179 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3180 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3181 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3182 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3183 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3184 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3185 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3186 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3187 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3188 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3189 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3190 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3191 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3192 Hoffmann, Zbigniew Jędrzejewski-Szmek
3194 — Berlin, 2015-02-16
3198 * When querying unit file enablement status (for example via
3199 "systemctl is-enabled"), a new state "indirect" is now known
3200 which indicates that a unit might not be enabled itself, but
3201 another unit listed in its Also= setting might be.
3203 * Similar to the various existing ConditionXYZ= settings for
3204 units, there are now matching AssertXYZ= settings. While
3205 failing conditions cause a unit to be skipped, but its job
3206 to succeed, failing assertions declared like this will cause
3207 a unit start operation and its job to fail.
3209 * hostnamed now knows a new chassis type "embedded".
3211 * systemctl gained a new "edit" command. When used on a unit
3212 file, this allows extending unit files with .d/ drop-in
3213 configuration snippets or editing the full file (after
3214 copying it from /usr/lib to /etc). This will invoke the
3215 user's editor (as configured with $EDITOR), and reload the
3216 modified configuration after editing.
3218 * "systemctl status" now shows the suggested enablement state
3219 for a unit, as declared in the (usually vendor-supplied)
3220 system preset files.
3222 * nss-myhostname will now resolve the single-label host name
3223 "gateway" to the locally configured default IP routing
3224 gateways, ordered by their metrics. This assigns a stable
3225 name to the used gateways, regardless which ones are
3226 currently configured. Note that the name will only be
3227 resolved after all other name sources (if nss-myhostname is
3228 configured properly) and should hence not negatively impact
3229 systems that use the single-label host name "gateway" in
3232 * systemd-inhibit now allows filtering by mode when listing
3235 * Scope and service units gained a new "Delegate" boolean
3236 property, which, when set, allows processes running inside the
3237 unit to further partition resources. This is primarily
3238 useful for systemd user instances as well as container
3241 * journald will now pick up audit messages directly from
3242 the kernel, and log them like any other log message. The
3243 audit fields are split up and fully indexed. This means that
3244 journalctl in many ways is now a (nicer!) alternative to
3245 ausearch, the traditional audit client. Note that this
3246 implements only a minimal audit client. If you want the
3247 special audit modes like reboot-on-log-overflow, please use
3248 the traditional auditd instead, which can be used in
3249 parallel to journald.
3251 * The ConditionSecurity= unit file option now understands the
3252 special string "audit" to check whether auditing is
3255 * journalctl gained two new commands --vacuum-size= and
3256 --vacuum-time= to delete old journal files until the
3257 remaining ones take up no more than the specified size on disk,
3258 or are not older than the specified time.
3260 * A new, native PPPoE library has been added to sd-network,
3261 systemd's library of light-weight networking protocols. This
3262 library will be used in a future version of networkd to
3263 enable PPPoE communication without an external pppd daemon.
3265 * The busctl tool now understands a new "capture" verb that
3266 works similar to "monitor", but writes a packet capture
3267 trace to STDOUT that can be redirected to a file which is
3268 compatible with libcap's capture file format. This can then
3269 be loaded in Wireshark and similar tools to inspect bus
3272 * The busctl tool now understands a new "tree" verb that shows
3273 the object trees of a specific service on the bus, or of all
3276 * The busctl tool now understands a new "introspect" verb that
3277 shows all interfaces and members of objects on the bus,
3278 including their signature and values. This is particularly
3279 useful to get more information about bus objects shown by
3280 the new "busctl tree" command.
3282 * The busctl tool now understands new verbs "call",
3283 "set-property" and "get-property" for invoking bus method
3284 calls, setting and getting bus object properties in a
3287 * busctl gained a new --augment-creds= argument that controls
3288 whether the tool shall augment credential information it
3289 gets from the bus with data from /proc, in a possibly
3292 * nspawn's --link-journal= switch gained two new values
3293 "try-guest" and "try-host" that work like "guest" and
3294 "host", but do not fail if the host has no persistent
3295 journalling enabled. -j is now equivalent to
3296 --link-journal=try-guest.
3298 * macvlan network devices created by nspawn will now have
3299 stable MAC addresses.
3301 * A new SmackProcessLabel= unit setting has been added, which
3302 controls the SMACK security label processes forked off by
3303 the respective unit shall use.
3305 * If compiled with --enable-xkbcommon, systemd-localed will
3306 verify x11 keymap settings by compiling the given keymap. It
3307 will spew out warnings if the compilation fails. This
3308 requires libxkbcommon to be installed.
3310 * When a coredump is collected, a larger number of metadata
3311 fields is now collected and included in the journal records
3312 created for it. More specifically, control group membership,
3313 environment variables, memory maps, working directory,
3314 chroot directory, /proc/$PID/status, and a list of open file
3315 descriptors is now stored in the log entry.
3317 * The udev hwdb now contains DPI information for mice. For
3320 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3322 * All systemd programs that read standalone configuration
3323 files in /etc now also support a corresponding series of
3324 .conf.d configuration directories in /etc/, /run/,
3325 /usr/local/lib/, /usr/lib/, and (if configured with
3326 --enable-split-usr) /lib/. In particular, the following
3327 configuration files now have corresponding configuration
3328 directories: system.conf user.conf, logind.conf,
3329 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3330 resolved.conf, timesyncd.conf, journal-remote.conf, and
3331 journal-upload.conf. Note that distributions should use the
3332 configuration directories in /usr/lib/; the directories in
3333 /etc/ are reserved for the system administrator.
3335 * systemd-rfkill will no longer take the rfkill device name
3336 into account when storing rfkill state on disk, as the name
3337 might be dynamically assigned and not stable. Instead, the
3338 ID_PATH udev variable combined with the rfkill type (wlan,
3339 bluetooth, ...) is used.
3341 * A new service systemd-machine-id-commit.service has been
3342 added. When used on systems where /etc is read-only during
3343 boot, and /etc/machine-id is not initialized (but an empty
3344 file), this service will copy the temporary machine ID
3345 created as replacement into /etc after the system is fully
3346 booted up. This is useful for systems that are freshly
3347 installed with a non-initialized machine ID, but should get
3348 a fixed machine ID for subsequent boots.
3350 * networkd's .netdev files now provide a large set of
3351 configuration parameters for VXLAN devices. Similarly, the
3352 bridge port cost parameter is now configurable in .network
3353 files. There's also new support for configuring IP source
3354 routing. networkd .link files gained support for a new
3355 OriginalName= match that is useful to match against the
3356 original interface name the kernel assigned. .network files
3357 may include MTU= and MACAddress= fields for altering the MTU
3358 and MAC address while being connected to a specific network
3361 * The LUKS logic gained supported for configuring
3362 UUID-specific key files. There's also new support for naming
3363 LUKS device from the kernel command line, using the new
3364 luks.name= argument.
3366 * Timer units may now be transiently created via the bus API
3367 (this was previously already available for scope and service
3368 units). In addition it is now possible to create multiple
3369 transient units at the same time with a single bus call. The
3370 "systemd-run" tool has been updated to make use of this for
3371 running commands on a specified time, in at(1)-style.
3373 * tmpfiles gained support for "t" lines, for assigning
3374 extended attributes to files. Among other uses this may be
3375 used to assign SMACK labels to files.
3377 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3378 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3379 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3380 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3381 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3382 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3383 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3384 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3385 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3386 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3387 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3388 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3389 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3390 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3391 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3392 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3393 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3394 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3396 — Berlin, 2014-12-10
3400 * journalctl gained the new options -t/--identifier= to match
3401 on the syslog identifier (aka "tag"), as well as --utc to
3402 show log timestamps in the UTC timezone. journalctl now also
3403 accepts -n/--lines=all to disable line capping in a pager.
3405 * journalctl gained a new switch, --flush, that synchronously
3406 flushes logs from /run/log/journal to /var/log/journal if
3407 persistent storage is enabled. systemd-journal-flush.service
3408 now waits until the operation is complete.
3410 * Services can notify the manager before they start a reload
3411 (by sending RELOADING=1) or shutdown (by sending
3412 STOPPING=1). This allows the manager to track and show the
3413 internal state of daemons and closes a race condition when
3414 the process is still running but has closed its D-Bus
3417 * Services with Type=oneshot do not have to have any ExecStart
3420 * User units are now loaded also from
3421 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3422 /run/systemd/user directory that was already previously
3423 supported, but is under the control of the user.
3425 * Job timeouts (i.e. time-outs on the time a job that is
3426 queued stays in the run queue) can now optionally result in
3427 immediate reboot or power-off actions (JobTimeoutAction= and
3428 JobTimeoutRebootArgument=). This is useful on ".target"
3429 units, to limit the maximum time a target remains
3430 undispatched in the run queue, and to trigger an emergency
3431 operation in such a case. This is now used by default to
3432 turn off the system if boot-up (as defined by everything in
3433 basic.target) hangs and does not complete for at least
3434 15min. Also, if power-off or reboot hang for at least 30min
3435 an immediate power-off/reboot operation is triggered. This
3436 functionality is particularly useful to increase reliability
3437 on embedded devices, but also on laptops which might
3438 accidentally get powered on when carried in a backpack and
3439 whose boot stays stuck in a hard disk encryption passphrase
3442 * systemd-logind can be configured to also handle lid switch
3443 events even when the machine is docked or multiple displays
3444 are attached (HandleLidSwitchDocked= option).
3446 * A helper binary and a service have been added which can be
3447 used to resume from hibernation in the initramfs. A
3448 generator will parse the resume= option on the kernel
3449 command line to trigger resume.
3451 * A user console daemon systemd-consoled has been
3452 added. Currently, it is a preview, and will so far open a
3453 single terminal on each session of the user marked as
3454 Desktop=systemd-console.
3456 * Route metrics can be specified for DHCP routes added by
3459 * The SELinux context of socket-activated services can be set
3460 from the information provided by the networking stack
3461 (SELinuxContextFromNet= option).
3463 * Userspace firmware loading support has been removed and
3464 the minimum supported kernel version is thus bumped to 3.7.
3466 * Timeout for udev workers has been increased from 1 to 3
3467 minutes, but a warning will be printed after 1 minute to
3468 help diagnose kernel modules that take a long time to load.
3470 * Udev rules can now remove tags on devices with TAG-="foobar".
3472 * systemd's readahead implementation has been removed. In many
3473 circumstances it didn't give expected benefits even for
3474 rotational disk drives and was becoming less relevant in the
3475 age of SSDs. As none of the developers has been using
3476 rotating media anymore, and nobody stepped up to actively
3477 maintain this component of systemd it has now been removed.
3479 * Swap units can use Options= to specify discard options.
3480 Discard options specified for swaps in /etc/fstab are now
3483 * Docker containers are now detected as a separate type of
3486 * The Password Agent protocol gained support for queries where
3487 the user input is shown, useful e.g. for user names.
3488 systemd-ask-password gained a new --echo option to turn that
3491 * The default sysctl.d/ snippets will now set:
3493 net.core.default_qdisc = fq_codel
3495 This selects Fair Queuing Controlled Delay as the default
3496 queuing discipline for network interfaces. fq_codel helps
3497 fight the network bufferbloat problem. It is believed to be
3498 a good default with no tuning required for most workloads.
3499 Downstream distributions may override this choice. On 10Gbit
3500 servers that do not do forwarding, "fq" may perform better.
3501 Systems without a good clocksource should use "pfifo_fast".
3503 * If kdbus is enabled during build a new option BusPolicy= is
3504 available for service units, that allows locking all service
3505 processes into a stricter bus policy, in order to limit
3506 access to various bus services, or even hide most of them
3507 from the service's view entirely.
3509 * networkctl will now show the .network and .link file
3510 networkd has applied to a specific interface.
3512 * sd-login gained a new API call sd_session_get_desktop() to
3513 query which desktop environment has been selected for a
3516 * UNIX utmp support is now compile-time optional to support
3517 legacy-free systems.
3519 * systemctl gained two new commands "add-wants" and
3520 "add-requires" for pulling in units from specific targets
3523 * If the word "rescue" is specified on the kernel command line
3524 the system will now boot into rescue mode (aka
3525 rescue.target), which was previously available only by
3526 specifying "1" or "systemd.unit=rescue.target" on the kernel
3527 command line. This new kernel command line option nicely
3528 mirrors the already existing "emergency" kernel command line
3531 * New kernel command line options mount.usr=, mount.usrflags=,
3532 mount.usrfstype= have been added that match root=, rootflags=,
3533 rootfstype= but allow mounting a specific file system to
3536 * The $NOTIFY_SOCKET is now also passed to control processes of
3537 services, not only the main process.
3539 * This version reenables support for fsck's -l switch. This
3540 means at least version v2.25 of util-linux is required for
3541 operation, otherwise dead-locks on device nodes may
3542 occur. Again: you need to update util-linux to at least
3543 v2.25 when updating systemd to v217.
3545 * The "multi-seat-x" tool has been removed from systemd, as
3546 its functionality has been integrated into X servers 1.16,
3547 and the tool is hence redundant. It is recommended to update
3548 display managers invoking this tool to simply invoke X
3549 directly from now on, again.
3551 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3552 message flag has been added for all of systemd's PolicyKit
3553 authenticated method calls has been added. In particular
3554 this now allows optional interactive authorization via
3555 PolicyKit for many of PID1's privileged operations such as
3556 unit file enabling and disabling.
3558 * "udevadm hwdb --update" learnt a new switch "--usr" for
3559 placing the rebuilt hardware database in /usr instead of
3560 /etc. When used only hardware database entries stored in
3561 /usr will be used, and any user database entries in /etc are
3562 ignored. This functionality is useful for vendors to ship a
3563 pre-built database on systems where local configuration is
3564 unnecessary or unlikely.
3566 * Calendar time specifications in .timer units now also
3567 understand the strings "semi-annually", "quarterly" and
3568 "minutely" as shortcuts (in addition to the preexisting
3569 "anually", "hourly", ...).
3571 * systemd-tmpfiles will now correctly create files in /dev
3572 at boot which are marked for creation only at boot. It is
3573 recommended to always create static device nodes with 'c!'
3574 and 'b!', so that they are created only at boot and not
3575 overwritten at runtime.
3577 * When the watchdog logic is used for a service (WatchdogSec=)
3578 and the watchdog timeout is hit the service will now be
3579 terminated with SIGABRT (instead of just SIGTERM), in order
3580 to make sure a proper coredump and backtrace is
3581 generated. This ensures that hanging services will result in
3582 similar coredump/backtrace behaviour as services that hit a
3585 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3586 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3587 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3588 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3589 Herrmann, David Sommerseth, David Strauss, Emil Renner
3590 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3591 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3592 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3593 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3594 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3595 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3596 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3597 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3598 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3599 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3600 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3601 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3602 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3603 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3604 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3605 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3608 — Berlin, 2014-10-28
3612 * timedated no longer reads NTP implementation unit names from
3613 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3614 implementations should add a
3616 Conflicts=systemd-timesyncd.service
3618 to their unit files to take over and replace systemd's NTP
3619 default functionality.
3621 * systemd-sysusers gained a new line type "r" for configuring
3622 which UID/GID ranges to allocate system users/groups
3623 from. Lines of type "u" may now add an additional column
3624 that specifies the home directory for the system user to be
3625 created. Also, systemd-sysusers may now optionally read user
3626 information from STDIN instead of a file. This is useful for
3627 invoking it from RPM preinst scriptlets that need to create
3628 users before the first RPM file is installed since these
3629 files might need to be owned by them. A new
3630 %sysusers_create_inline RPM macro has been introduced to do
3631 just that. systemd-sysusers now updates the shadow files as
3632 well as the user/group databases, which should enhance
3633 compatibility with certain tools like grpck.
3635 * A number of bus APIs of PID 1 now optionally consult
3636 PolicyKit to permit access for otherwise unprivileged
3637 clients under certain conditions. Note that this currently
3638 doesn't support interactive authentication yet, but this is
3639 expected to be added eventually, too.
3641 * /etc/machine-info now has new fields for configuring the
3642 deployment environment of the machine, as well as the
3643 location of the machine. hostnamectl has been updated with
3644 new command to update these fields.
3646 * systemd-timesyncd has been updated to automatically acquire
3647 NTP server information from systemd-networkd, which might
3648 have been discovered via DHCP.
3650 * systemd-resolved now includes a caching DNS stub resolver
3651 and a complete LLMNR name resolution implementation. A new
3652 NSS module "nss-resolve" has been added which can be used
3653 instead of glibc's own "nss-dns" to resolve hostnames via
3654 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3655 be resolved via systemd-resolved D-Bus APIs. In contrast to
3656 the glibc internal resolver systemd-resolved is aware of
3657 multi-homed system, and keeps DNS server and caches separate
3658 and per-interface. Queries are sent simultaneously on all
3659 interfaces that have DNS servers configured, in order to
3660 properly handle VPNs and local LANs which might resolve
3661 separate sets of domain names. systemd-resolved may acquire
3662 DNS server information from systemd-networkd automatically,
3663 which in turn might have discovered them via DHCP. A tool
3664 "systemd-resolve-host" has been added that may be used to
3665 query the DNS logic in resolved. systemd-resolved implements
3666 IDNA and automatically uses IDNA or UTF-8 encoding depending
3667 on whether classic DNS or LLMNR is used as transport. In the
3668 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3669 implementation to systemd-resolved.
3671 * A new NSS module nss-mymachines has been added, that
3672 automatically resolves the names of all local registered
3673 containers to their respective IP addresses.
3675 * A new client tool "networkctl" for systemd-networkd has been
3676 added. It currently is entirely passive and will query
3677 networking configuration from udev, rtnetlink and networkd,
3678 and present it to the user in a very friendly
3679 way. Eventually, we hope to extend it to become a full
3680 control utility for networkd.
3682 * .socket units gained a new DeferAcceptSec= setting that
3683 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3684 TCP. Similarly, support for controlling TCP keep-alive
3685 settings has been added (KeepAliveTimeSec=,
3686 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3687 turning off Nagle's algorithm on TCP has been added
3690 * logind learned a new session type "web", for use in projects
3691 like Cockpit which register web clients as PAM sessions.
3693 * timer units with at least one OnCalendar= setting will now
3694 be started only after time-sync.target has been
3695 reached. This way they will not elapse before the system
3696 clock has been corrected by a local NTP client or
3697 similar. This is particular useful on RTC-less embedded
3698 machines, that come up with an invalid system clock.
3700 * systemd-nspawn's --network-veth= switch should now result in
3701 stable MAC addresses for both the outer and the inner side
3704 * systemd-nspawn gained a new --volatile= switch for running
3705 container instances with /etc or /var unpopulated.
3707 * The kdbus client code has been updated to use the new Linux
3708 3.17 memfd subsystem instead of the old kdbus-specific one.
3710 * systemd-networkd's DHCP client and server now support
3711 FORCERENEW. There are also new configuration options to
3712 configure the vendor client identifier and broadcast mode
3715 * systemd will no longer inform the kernel about the current
3716 timezone, as this is necessarily incorrect and racy as the
3717 kernel has no understanding of DST and similar
3718 concepts. This hence means FAT timestamps will be always
3719 considered UTC, similar to what Android is already
3720 doing. Also, when the RTC is configured to the local time
3721 (rather than UTC) systemd will never synchronize back to it,
3722 as this might confuse Windows at a later boot.
3724 * systemd-analyze gained a new command "verify" for offline
3725 validation of unit files.
3727 * systemd-networkd gained support for a couple of additional
3728 settings for bonding networking setups. Also, the metric for
3729 statically configured routes may now be configured. For
3730 network interfaces where this is appropriate the peer IP
3731 address may now be configured.
3733 * systemd-networkd's DHCP client will no longer request
3734 broadcasting by default, as this tripped up some networks.
3735 For hardware where broadcast is required the feature should
3736 be switched back on using RequestBroadcast=yes.
3738 * systemd-networkd will now set up IPv4LL addresses (when
3739 enabled) even if DHCP is configured successfully.
3741 * udev will now default to respect network device names given
3742 by the kernel when the kernel indicates that these are
3743 predictable. This behavior can be tweaked by changing
3744 NamePolicy= in the relevant .link file.
3746 * A new library systemd-terminal has been added that
3747 implements full TTY stream parsing and rendering. This
3748 library is supposed to be used later on for implementing a
3749 full userspace VT subsystem, replacing the current kernel
3752 * A new tool systemd-journal-upload has been added to push
3753 journal data to a remote system running
3754 systemd-journal-remote.
3756 * journald will no longer forward all local data to another
3757 running syslog daemon. This change has been made because
3758 rsyslog (which appears to be the most commonly used syslog
3759 implementation these days) no longer makes use of this, and
3760 instead pulls the data out of the journal on its own. Since
3761 forwarding the messages to a non-existent syslog server is
3762 more expensive than we assumed we have now turned this
3763 off. If you run a syslog server that is not a recent rsyslog
3764 version, you have to turn this option on again
3765 (ForwardToSyslog= in journald.conf).
3767 * journald now optionally supports the LZ4 compressor for
3768 larger journal fields. This compressor should perform much
3769 better than XZ which was the previous default.
3771 * machinectl now shows the IP addresses of local containers,
3772 if it knows them, plus the interface name of the container.
3774 * A new tool "systemd-escape" has been added that makes it
3775 easy to escape strings to build unit names and similar.
3777 * sd_notify() messages may now include a new ERRNO= field
3778 which is parsed and collected by systemd and shown among the
3779 "systemctl status" output for a service.
3781 * A new component "systemd-firstboot" has been added that
3782 queries the most basic systemd information (timezone,
3783 hostname, root password) interactively on first
3784 boot. Alternatively it may also be used to provision these
3785 things offline on OS images installed into directories.
3787 * The default sysctl.d/ snippets will now set
3789 net.ipv4.conf.default.promote_secondaries=1
3791 This has the benefit of no flushing secondary IP addresses
3792 when primary addresses are removed.
3794 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3795 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3796 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3797 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3798 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3799 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3800 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3801 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3802 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3803 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3804 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3805 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3806 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3807 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3808 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3810 — Berlin, 2014-08-19
3814 * A new tool systemd-sysusers has been added. This tool
3815 creates system users and groups in /etc/passwd and
3816 /etc/group, based on static declarative system user/group
3817 definitions in /usr/lib/sysusers.d/. This is useful to
3818 enable factory resets and volatile systems that boot up with
3819 an empty /etc directory, and thus need system users and
3820 groups created during early boot. systemd now also ships
3821 with two default sysusers.d/ files for the most basic
3822 users and groups systemd and the core operating system
3825 * A new tmpfiles snippet has been added that rebuilds the
3826 essential files in /etc on boot, should they be missing.
3828 * A directive for ensuring automatic clean-up of
3829 /var/cache/man/ has been removed from the default
3830 configuration. This line should now be shipped by the man
3831 implementation. The necessary change has been made to the
3832 man-db implementation. Note that you need to update your man
3833 implementation to one that ships this line, otherwise no
3834 automatic clean-up of /var/cache/man will take place.
3836 * A new condition ConditionNeedsUpdate= has been added that
3837 may conditionalize services to only run when /etc or /var
3838 are "older" than the vendor operating system resources in
3839 /usr. This is useful for reconstructing or updating /etc
3840 after an offline update of /usr or a factory reset, on the
3841 next reboot. Services that want to run once after such an
3842 update or reset should use this condition and order
3843 themselves before the new systemd-update-done.service, which
3844 will mark the two directories as fully updated. A number of
3845 service files have been added making use of this, to rebuild
3846 the udev hardware database, the journald message catalog and
3847 dynamic loader cache (ldconfig). The systemd-sysusers tool
3848 described above also makes use of this now. With this in
3849 place it is now possible to start up a minimal operating
3850 system with /etc empty cleanly. For more information on the
3851 concepts involved see this recent blog story:
3853 http://0pointer.de/blog/projects/stateless.html
3855 * A new system group "input" has been introduced, and all
3856 input device nodes get this group assigned. This is useful
3857 for system-level software to get access to input devices. It
3858 complements what is already done for "audio" and "video".
3860 * systemd-networkd learnt minimal DHCPv4 server support in
3861 addition to the existing DHCPv4 client support. It also
3862 learnt DHCPv6 client and IPv6 Router Solicitation client
3863 support. The DHCPv4 client gained support for static routes
3864 passed in from the server. Note that the [DHCPv4] section
3865 known in older systemd-networkd versions has been renamed to
3866 [DHCP] and is now also used by the DHCPv6 client. Existing
3867 .network files using settings of this section should be
3868 updated, though compatibility is maintained. Optionally, the
3869 client hostname may now be sent to the DHCP server.
3871 * networkd gained support for vxlan virtual networks as well
3872 as tun/tap and dummy devices.
3874 * networkd gained support for automatic allocation of address
3875 ranges for interfaces from a system-wide pool of
3876 addresses. This is useful for dynamically managing a large
3877 number of interfaces with a single network configuration
3878 file. In particular this is useful to easily assign
3879 appropriate IP addresses to the veth links of a large number
3880 of nspawn instances.
3882 * RPM macros for processing sysusers, sysctl and binfmt
3883 drop-in snippets at package installation time have been
3886 * The /etc/os-release file should now be placed in
3887 /usr/lib/os-release. The old location is automatically
3888 created as symlink. /usr/lib is the more appropriate
3889 location of this file, since it shall actually describe the
3890 vendor operating system shipped in /usr, and not the
3891 configuration stored in /etc.
3893 * .mount units gained a new boolean SloppyOptions= setting
3894 that maps to mount(8)'s -s option which enables permissive
3895 parsing of unknown mount options.
3897 * tmpfiles learnt a new "L+" directive which creates a symlink
3898 but (unlike "L") deletes a pre-existing file first, should
3899 it already exist and not already be the correct
3900 symlink. Similarly, "b+", "c+" and "p+" directives have been
3901 added as well, which create block and character devices, as
3902 well as fifos in the filesystem, possibly removing any
3903 pre-existing files of different types.
3905 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3906 'argument' field (which so far specified the source to
3907 symlink/copy the files from) is now optional. If omitted the
3908 same file os copied from /usr/share/factory/ suffixed by the
3909 full destination path. This is useful for populating /etc
3910 with essential files, by copying them from vendor defaults
3911 shipped in /usr/share/factory/etc.
3913 * A new command "systemctl preset-all" has been added that
3914 applies the service preset settings to all installed unit
3915 files. A new switch --preset-mode= has been added that
3916 controls whether only enable or only disable operations
3919 * A new command "systemctl is-system-running" has been added
3920 that allows checking the overall state of the system, for
3921 example whether it is fully up and running.
3923 * When the system boots up with an empty /etc, the equivalent
3924 to "systemctl preset-all" is executed during early boot, to
3925 make sure all default services are enabled after a factory
3928 * systemd now contains a minimal preset file that enables the
3929 most basic services systemd ships by default.
3931 * Unit files' [Install] section gained a new DefaultInstance=
3932 field for defining the default instance to create if a
3933 template unit is enabled with no instance specified.
3935 * A new passive target cryptsetup-pre.target has been added
3936 that may be used by services that need to make they run and
3937 finish before the first LUKS cryptographic device is set up.
3939 * The /dev/loop-control and /dev/btrfs-control device nodes
3940 are now owned by the "disk" group by default, opening up
3941 access to this group.
3943 * systemd-coredump will now automatically generate a
3944 stack trace of all core dumps taking place on the system,
3945 based on elfutils' libdw library. This stack trace is logged
3948 * systemd-coredump may now optionally store coredumps directly
3949 on disk (in /var/lib/systemd/coredump, possibly compressed),
3950 instead of storing them unconditionally in the journal. This
3951 mode is the new default. A new configuration file
3952 /etc/systemd/coredump.conf has been added to configure this
3953 and other parameters of systemd-coredump.
3955 * coredumpctl gained a new "info" verb to show details about a
3956 specific coredump. A new switch "-1" has also been added
3957 that makes sure to only show information about the most
3958 recent entry instead of all entries. Also, as the tool is
3959 generally useful now the "systemd-" prefix of the binary
3960 name has been removed. Distributions that want to maintain
3961 compatibility with the old name should add a symlink from
3962 the old name to the new name.
3964 * journald's SplitMode= now defaults to "uid". This makes sure
3965 that unprivileged users can access their own coredumps with
3966 coredumpctl without restrictions.
3968 * New kernel command line options "systemd.wants=" (for
3969 pulling an additional unit during boot), "systemd.mask="
3970 (for masking a specific unit for the boot), and
3971 "systemd.debug-shell" (for enabling the debug shell on tty9)
3972 have been added. This is implemented in the new generator
3973 "systemd-debug-generator".
3975 * systemd-nspawn will now by default filter a couple of
3976 syscalls for containers, among them those required for
3977 kernel module loading, direct x86 IO port access, swap
3978 management, and kexec. Most importantly though
3979 open_by_handle_at() is now prohibited for containers,
3980 closing a hole similar to a recently discussed vulnerability
3981 in docker regarding access to files on file hierarchies the
3982 container should normally not have access to. Note that, for
3983 nspawn, we generally make no security claims anyway (and
3984 this is explicitly documented in the man page), so this is
3985 just a fix for one of the most obvious problems.
3987 * A new man page file-hierarchy(7) has been added that
3988 contains a minimized, modernized version of the file system
3989 layout systemd expects, similar in style to the FHS
3990 specification or hier(5). A new tool systemd-path(1) has
3991 been added to query many of these paths for the local
3994 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3995 longer done. Since the directory now has a per-user size
3996 limit, and is cleaned on logout this appears unnecessary,
3997 in particular since this now brings the lifecycle of this
3998 directory closer in line with how IPC objects are handled.
4000 * systemd.pc now exports a number of additional directories,
4001 including $libdir (which is useful to identify the library
4002 path for the primary architecture of the system), and a
4003 couple of drop-in directories.
4005 * udev's predictable network interface names now use the dev_port
4006 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4007 distinguish between ports of the same PCI function. dev_id should
4008 only be used for ports using the same HW address, hence the need
4011 * machined has been updated to export the OS version of a
4012 container (read from /etc/os-release and
4013 /usr/lib/os-release) on the bus. This is now shown in
4014 "machinectl status" for a machine.
4016 * A new service setting RestartForceExitStatus= has been
4017 added. If configured to a set of exit signals or process
4018 return values, the service will be restarted when the main
4019 daemon process exits with any of them, regardless of the
4022 * systemctl's -H switch for connecting to remote systemd
4023 machines has been extended so that it may be used to
4024 directly connect to a specific container on the
4025 host. "systemctl -H root@foobar:waldi" will now connect as
4026 user "root" to host "foobar", and then proceed directly to
4027 the container named "waldi". Note that currently you have to
4028 authenticate as user "root" for this to work, as entering
4029 containers is a privileged operation.
4031 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4032 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4033 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4034 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4035 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4036 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4037 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4038 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4039 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4040 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4041 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4042 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4044 — Berlin, 2014-07-03
4048 * As an experimental feature, udev now tries to lock the
4049 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4050 executes events for the disk or any of its partitions.
4051 Applications like partitioning programs can lock the
4052 disk device node (flock(LOCK_EX)) and claim temporary
4053 device ownership that way; udev will entirely skip all event
4054 handling for this disk and its partitions. If the disk
4055 was opened for writing, the close will trigger a partition
4056 table rescan in udev's "watch" facility, and if needed
4057 synthesize "change" events for the disk and all its partitions.
4058 This is now unconditionally enabled, and if it turns out to
4059 cause major problems, we might turn it on only for specific
4060 devices, or might need to disable it entirely. Device Mapper
4061 devices are excluded from this logic.
4063 * We temporarily dropped the "-l" switch for fsck invocations,
4064 since they collide with the flock() logic above. util-linux
4065 upstream has been changed already to avoid this conflict,
4066 and we will readd "-l" as soon as util-linux with this
4067 change has been released.
4069 * The dependency on libattr has been removed. Since a long
4070 time, the extended attribute calls have moved to glibc, and
4071 libattr is thus unnecessary.
4073 * Virtualization detection works without privileges now. This
4074 means the systemd-detect-virt binary no longer requires
4075 CAP_SYS_PTRACE file capabilities, and our daemons can run
4076 with fewer privileges.
4078 * systemd-networkd now runs under its own "systemd-network"
4079 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4080 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4081 loses the ability to write to files owned by root this way.
4083 * Similarly, systemd-resolved now runs under its own
4084 "systemd-resolve" user with no capabilities remaining.
4086 * Similarly, systemd-bus-proxyd now runs under its own
4087 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4089 * systemd-networkd gained support for setting up "veth"
4090 virtual Ethernet devices for container connectivity, as well
4091 as GRE and VTI tunnels.
4093 * systemd-networkd will no longer automatically attempt to
4094 manually load kernel modules necessary for certain tunnel
4095 transports. Instead, it is assumed the kernel loads them
4096 automatically when required. This only works correctly on
4097 very new kernels. On older kernels, please consider adding
4098 the kernel modules to /etc/modules-load.d/ as a work-around.
4100 * The resolv.conf file systemd-resolved generates has been
4101 moved to /run/systemd/resolve/. If you have a symlink from
4102 /etc/resolv.conf, it might be necessary to correct it.
4104 * Two new service settings, ProtectHome= and ProtectSystem=,
4105 have been added. When enabled, they will make the user data
4106 (such as /home) inaccessible or read-only and the system
4107 (such as /usr) read-only, for specific services. This allows
4108 very light-weight per-service sandboxing to avoid
4109 modifications of user data or system files from
4110 services. These two new switches have been enabled for all
4111 of systemd's long-running services, where appropriate.
4113 * Socket units gained new SocketUser= and SocketGroup=
4114 settings to set the owner user and group of AF_UNIX sockets
4115 and FIFOs in the file system.
4117 * Socket units gained a new RemoveOnStop= setting. If enabled,
4118 all FIFOS and sockets in the file system will be removed
4119 when the specific socket unit is stopped.
4121 * Socket units gained a new Symlinks= setting. It takes a list
4122 of symlinks to create to file system sockets or FIFOs
4123 created by the specific Unix sockets. This is useful to
4124 manage symlinks to socket nodes with the same life-cycle as
4127 * The /dev/log socket and /dev/initctl FIFO have been moved to
4128 /run, and have been replaced by symlinks. This allows
4129 connecting to these facilities even if PrivateDevices=yes is
4130 used for a service (which makes /dev/log itself unavailable,
4131 but /run is left). This also has the benefit of ensuring
4132 that /dev only contains device nodes, directories and
4133 symlinks, and nothing else.
4135 * sd-daemon gained two new calls sd_pid_notify() and
4136 sd_pid_notifyf(). They are similar to sd_notify() and
4137 sd_notifyf(), but allow overriding of the source PID of
4138 notification messages if permissions permit this. This is
4139 useful to send notify messages on behalf of a different
4140 process (for example, the parent process). The
4141 systemd-notify tool has been updated to make use of this
4142 when sending messages (so that notification messages now
4143 originate from the shell script invoking systemd-notify and
4144 not the systemd-notify process itself. This should minimize
4145 a race where systemd fails to associate notification
4146 messages to services when the originating process already
4149 * A new "on-abnormal" setting for Restart= has been added. If
4150 set, it will result in automatic restarts on all "abnormal"
4151 reasons for a process to exit, which includes unclean
4152 signals, core dumps, timeouts and watchdog timeouts, but
4153 does not include clean and unclean exit codes or clean
4154 signals. Restart=on-abnormal is an alternative for
4155 Restart=on-failure for services that shall be able to
4156 terminate and avoid restarts on certain errors, by
4157 indicating so with an unclean exit code. Restart=on-failure
4158 or Restart=on-abnormal is now the recommended setting for
4159 all long-running services.
4161 * If the InaccessibleDirectories= service setting points to a
4162 mount point (or if there are any submounts contained within
4163 it), it is now attempted to completely unmount it, to make
4164 the file systems truly unavailable for the respective
4167 * The ReadOnlyDirectories= service setting and
4168 systemd-nspawn's --read-only parameter are now recursively
4169 applied to all submounts, too.
4171 * Mount units may now be created transiently via the bus APIs.
4173 * The support for SysV and LSB init scripts has been removed
4174 from the systemd daemon itself. Instead, it is now
4175 implemented as a generator that creates native systemd units
4176 from these scripts when needed. This enables us to remove a
4177 substantial amount of legacy code from PID 1, following the
4178 fact that many distributions only ship a very small number
4179 of LSB/SysV init scripts nowadays.
4181 * Privileged Xen (dom0) domains are not considered
4182 virtualization anymore by the virtualization detection
4183 logic. After all, they generally have unrestricted access to
4184 the hardware and usually are used to manage the unprivileged
4187 * systemd-tmpfiles gained a new "C" line type, for copying
4188 files or entire directories.
4190 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4191 lines. So far, they have been non-globbing versions of the
4192 latter, and have thus been redundant. In future, it is
4193 recommended to only use "z". "m" has hence been removed
4194 from the documentation, even though it stays supported.
4196 * A tmpfiles snippet to recreate the most basic structure in
4197 /var has been added. This is enough to create the /var/run →
4198 /run symlink and create a couple of structural
4199 directories. This allows systems to boot up with an empty or
4200 volatile /var. Of course, while with this change, the core OS
4201 now is capable with dealing with a volatile /var, not all
4202 user services are ready for it. However, we hope that sooner
4203 or later, many service daemons will be changed upstream so
4204 that they are able to automatically create their necessary
4205 directories in /var at boot, should they be missing. This is
4206 the first step to allow state-less systems that only require
4207 the vendor image for /usr to boot.
4209 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4210 empty tmpfs instance to a specific directory. This is
4211 particularly useful for making use of the automatic
4212 reconstruction of /var (see above), by passing --tmpfs=/var.
4214 * Access modes specified in tmpfiles snippets may now be
4215 prefixed with "~", which indicates that they shall be masked
4216 by whether the existing file or directory is currently
4217 writable, readable or executable at all. Also, if specified,
4218 the sgid/suid/sticky bits will be masked for all
4221 * A new passive target unit "network-pre.target" has been
4222 added which is useful for services that shall run before any
4223 network is configured, for example firewall scripts.
4225 * The "floppy" group that previously owned the /dev/fd*
4226 devices is no longer used. The "disk" group is now used
4227 instead. Distributions should probably deprecate usage of
4230 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4231 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4232 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4233 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4234 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4235 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4238 — Berlin, 2014-06-11
4242 * A new "systemd-timesyncd" daemon has been added for
4243 synchronizing the system clock across the network. It
4244 implements an SNTP client. In contrast to NTP
4245 implementations such as chrony or the NTP reference server,
4246 this only implements a client side, and does not bother with
4247 the full NTP complexity, focusing only on querying time from
4248 one remote server and synchronizing the local clock to
4249 it. Unless you intend to serve NTP to networked clients or
4250 want to connect to local hardware clocks, this simple NTP
4251 client should be more than appropriate for most
4252 installations. The daemon runs with minimal privileges, and
4253 has been hooked up with networkd to only operate when
4254 network connectivity is available. The daemon saves the
4255 current clock to disk every time a new NTP sync has been
4256 acquired, and uses this to possibly correct the system clock
4257 early at bootup, in order to accommodate for systems that
4258 lack an RTC such as the Raspberry Pi and embedded devices,
4259 and to make sure that time monotonically progresses on these
4260 systems, even if it is not always correct. To make use of
4261 this daemon, a new system user and group "systemd-timesync"
4262 needs to be created on installation of systemd.
4264 * The queue "seqnum" interface of libudev has been disabled, as
4265 it was generally incompatible with device namespacing as
4266 sequence numbers of devices go "missing" if the devices are
4267 part of a different namespace.
4269 * "systemctl list-timers" and "systemctl list-sockets" gained
4270 a --recursive switch for showing units of these types also
4271 for all local containers, similar in style to the already
4272 supported --recursive switch for "systemctl list-units".
4274 * A new RebootArgument= setting has been added for service
4275 units, which may be used to specify a kernel reboot argument
4276 to use when triggering reboots with StartLimitAction=.
4278 * A new FailureAction= setting has been added for service
4279 units which may be used to specify an operation to trigger
4280 when a service fails. This works similarly to
4281 StartLimitAction=, but unlike it, controls what is done
4282 immediately rather than only after several attempts to
4283 restart the service in question.
4285 * hostnamed got updated to also expose the kernel name,
4286 release, and version on the bus. This is useful for
4287 executing commands like hostnamectl with the -H switch.
4288 systemd-analyze makes use of this to properly display
4289 details when running non-locally.
4291 * The bootchart tool can now show cgroup information in the
4292 graphs it generates.
4294 * The CFS CPU quota cgroup attribute is now exposed for
4295 services. The new CPUQuota= switch has been added for this
4296 which takes a percentage value. Setting this will have the
4297 result that a service may never get more CPU time than the
4298 specified percentage, even if the machine is otherwise idle.
4300 * systemd-networkd learned IPIP and SIT tunnel support.
4302 * LSB init scripts exposing a dependency on $network will now
4303 get a dependency on network-online.target rather than simply
4304 network.target. This should bring LSB handling closer to
4305 what it was on SysV systems.
4307 * A new fsck.repair= kernel option has been added to control
4308 how fsck shall deal with unclean file systems at boot.
4310 * The (.ini) configuration file parser will now silently
4311 ignore sections whose name begins with "X-". This may be
4312 used to maintain application-specific extension sections in unit
4315 * machined gained a new API to query the IP addresses of
4316 registered containers. "machinectl status" has been updated
4317 to show these addresses in its output.
4319 * A new call sd_uid_get_display() has been added to the
4320 sd-login APIs for querying the "primary" session of a
4321 user. The "primary" session of the user is elected from the
4322 user's sessions and generally a graphical session is
4323 preferred over a text one.
4325 * A minimal systemd-resolved daemon has been added. It
4326 currently simply acts as a companion to systemd-networkd and
4327 manages resolv.conf based on per-interface DNS
4328 configuration, possibly supplied via DHCP. In the long run
4329 we hope to extend this into a local DNSSEC enabled DNS and
4332 * The systemd-networkd-wait-online tool is now enabled by
4333 default. It will delay network-online.target until a network
4334 connection has been configured. The tool primarily integrates
4335 with networkd, but will also make a best effort to make sense
4336 of network configuration performed in some other way.
4338 * Two new service options StartupCPUShares= and
4339 StartupBlockIOWeight= have been added that work similarly to
4340 CPUShares= and BlockIOWeight= however only apply during
4341 system startup. This is useful to prioritize certain services
4342 differently during bootup than during normal runtime.
4344 * hostnamed has been changed to prefer the statically
4345 configured hostname in /etc/hostname (unless set to
4346 'localhost' or empty) over any dynamic one supplied by
4347 dhcp. With this change, the rules for picking the hostname
4348 match more closely the rules of other configuration settings
4349 where the local administrator's configuration in /etc always
4350 overrides any other settings.
4352 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4353 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4354 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4355 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4356 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4357 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4358 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4359 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4360 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4361 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4362 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4363 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4364 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4365 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4366 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4367 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4370 — Beijing, 2014-05-28
4374 * When restoring the screen brightness at boot, stay away from
4375 the darkest setting or from the lowest 5% of the available
4376 range, depending on which is the larger value of both. This
4377 should effectively protect the user from rebooting into a
4378 black screen, should the brightness have been set to minimum
4381 * sd-login gained a new sd_machine_get_class() call to
4382 determine the class ("vm" or "container") of a machine
4383 registered with machined.
4385 * sd-login gained new calls
4386 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4387 to query the identity of the peer of a local AF_UNIX
4388 connection. They operate similarly to their sd_pid_get_xyz()
4391 * PID 1 will now maintain a system-wide system state engine
4392 with the states "starting", "running", "degraded",
4393 "maintenance", "stopping". These states are bound to system
4394 startup, normal runtime, runtime with at least one failed
4395 service, rescue/emergency mode and system shutdown. This
4396 state is shown in the "systemctl status" output when no unit
4397 name is passed. It is useful to determine system state, in
4398 particularly when doing so for many systems or containers at
4401 * A new command "list-machines" has been added to "systemctl"
4402 that lists all local OS containers and shows their system
4403 state (see above), if systemd runs inside of them.
4405 * systemctl gained a new "-r" switch to recursively enumerate
4406 units on all local containers, when used with the
4407 "list-unit" command (which is the default one that is
4408 executed when no parameters are specified).
4410 * The GPT automatic partition discovery logic will now honour
4411 two GPT partition flags: one may be set on a partition to
4412 cause it to be mounted read-only, and the other may be set
4413 on a partition to ignore it during automatic discovery.
4415 * Two new GPT type UUIDs have been added for automatic root
4416 partition discovery, for 32-bit and 64-bit ARM. This is not
4417 particularly useful for discovering the root directory on
4418 these architectures during bare-metal boots (since UEFI is
4419 not common there), but still very useful to allow booting of
4420 ARM disk images in nspawn with the -i option.
4422 * MAC addresses of interfaces created with nspawn's
4423 --network-interface= switch will now be generated from the
4424 machine name, and thus be stable between multiple invocations
4427 * logind will now automatically remove all IPC objects owned
4428 by a user if she or he fully logs out. This makes sure that
4429 users who are logged out cannot continue to consume IPC
4430 resources. This covers SysV memory, semaphores and message
4431 queues as well as POSIX shared memory and message
4432 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4433 limits. With this functionality, that is corrected. This may
4434 be turned off by using the RemoveIPC= switch of logind.conf.
4436 * The systemd-machine-id-setup and tmpfiles tools gained a
4437 --root= switch to operate on a specific root directory,
4440 * journald can now forward logged messages to the TTYs of all
4441 logged in users ("wall"). This is the default for all
4442 emergency messages now.
4444 * A new tool systemd-journal-remote has been added to stream
4445 journal log messages across the network.
4447 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4448 controller trees are mounted into it. Note that the
4449 directories mounted beneath it are not read-only. This is a
4450 security measure and is particularly useful because glibc
4451 actually includes a search logic to pick any tmpfs it can
4452 find to implement shm_open() if /dev/shm is not available
4453 (which it might very well be in namespaced setups).
4455 * machinectl gained a new "poweroff" command to cleanly power
4456 down a local OS container.
4458 * The PrivateDevices= unit file setting will now also drop the
4459 CAP_MKNOD capability from the capability bound set, and
4460 imply DevicePolicy=closed.
4462 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4463 comprehensively on all long-running systemd services where
4464 this is appropriate.
4466 * systemd-udevd will now run in a disassociated mount
4467 namespace. To mount directories from udev rules, make sure to
4468 pull in mount units via SYSTEMD_WANTS properties.
4470 * The kdbus support gained support for uploading policy into
4471 the kernel. sd-bus gained support for creating "monitoring"
4472 connections that can eavesdrop into all bus communication
4473 for debugging purposes.
4475 * Timestamps may now be specified in seconds since the UNIX
4476 epoch Jan 1st, 1970 by specifying "@" followed by the value
4479 * Native tcpwrap support in systemd has been removed. tcpwrap
4480 is old code, not really maintained anymore and has serious
4481 shortcomings, and better options such as firewalls
4482 exist. For setups that require tcpwrap usage, please
4483 consider invoking your socket-activated service via tcpd,
4484 like on traditional inetd.
4486 * A new system.conf configuration option
4487 DefaultTimerAccuracySec= has been added that controls the
4488 default AccuracySec= setting of .timer units.
4490 * Timer units gained a new WakeSystem= switch. If enabled,
4491 timers configured this way will cause the system to resume
4492 from system suspend (if the system supports that, which most
4495 * Timer units gained a new Persistent= switch. If enabled,
4496 timers configured this way will save to disk when they have
4497 been last triggered. This information is then used on next
4498 reboot to possible execute overdue timer events, that
4499 could not take place because the system was powered off.
4500 This enables simple anacron-like behaviour for timer units.
4502 * systemctl's "list-timers" will now also list the time a
4503 timer unit was last triggered in addition to the next time
4504 it will be triggered.
4506 * systemd-networkd will now assign predictable IPv4LL
4507 addresses to its local interfaces.
4509 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4510 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4511 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4512 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4513 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4514 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4515 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4516 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4519 — Berlin, 2014-03-25
4523 * A new unit file setting RestrictAddressFamilies= has been
4524 added to restrict which socket address families unit
4525 processes gain access to. This takes address family names
4526 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4527 attack surface of services via exotic protocol stacks. This
4528 is built on seccomp system call filters.
4530 * Two new unit file settings RuntimeDirectory= and
4531 RuntimeDirectoryMode= have been added that may be used to
4532 manage a per-daemon runtime directories below /run. This is
4533 an alternative for setting up directory permissions with
4534 tmpfiles snippets, and has the advantage that the runtime
4535 directory's lifetime is bound to the daemon runtime and that
4536 the daemon starts up with an empty directory each time. This
4537 is particularly useful when writing services that drop
4538 privileges using the User= or Group= setting.
4540 * The DeviceAllow= unit setting now supports globbing for
4541 matching against device group names.
4543 * The systemd configuration file system.conf gained new
4544 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4545 DefaultMemoryAccounting= to globally turn on/off accounting
4546 for specific resources (cgroups) for all units. These
4547 settings may still be overridden individually in each unit
4550 * systemd-gpt-auto-generator is now able to discover /srv and
4551 root partitions in addition to /home and swap partitions. It
4552 also supports LUKS-encrypted partitions now. With this in
4553 place, automatic discovery of partitions to mount following
4554 the Discoverable Partitions Specification
4555 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4556 is now a lot more complete. This allows booting without
4557 /etc/fstab and without root= on the kernel command line on
4558 systems prepared appropriately.
4560 * systemd-nspawn gained a new --image= switch which allows
4561 booting up disk images and Linux installations on any block
4562 device that follow the Discoverable Partitions Specification
4563 (see above). This means that installations made with
4564 appropriately updated installers may now be started and
4565 deployed using container managers, completely
4566 unmodified. (We hope that libvirt-lxc will add support for
4567 this feature soon, too.)
4569 * systemd-nspawn gained a new --network-macvlan= setting to
4570 set up a private macvlan interface for the
4571 container. Similarly, systemd-networkd gained a new
4572 Kind=macvlan setting in .netdev files.
4574 * systemd-networkd now supports configuring local addresses
4577 * A new tool systemd-network-wait-online has been added to
4578 synchronously wait for network connectivity using
4581 * The sd-bus.h bus API gained a new sd_bus_track object for
4582 tracking the life-cycle of bus peers. Note that sd-bus.h is
4583 still not a public API though (unless you specify
4584 --enable-kdbus on the configure command line, which however
4585 voids your warranty and you get no API stability guarantee).
4587 * The $XDG_RUNTIME_DIR runtime directories for each user are
4588 now individual tmpfs instances, which has the benefit of
4589 introducing separate pools for each user, with individual
4590 size limits, and thus making sure that unprivileged clients
4591 can no longer negatively impact the system or other users by
4592 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4593 RuntimeDirectorySize= has been introduced that allows
4594 controlling the default size limit for all users. It
4595 defaults to 10% of the available physical memory. This is no
4596 replacement for quotas on tmpfs though (which the kernel
4597 still does not support), as /dev/shm and /tmp are still
4598 shared resources used by both the system and unprivileged
4601 * logind will now automatically turn off automatic suspending
4602 on laptop lid close when more than one display is
4603 connected. This was previously expected to be implemented
4604 individually in desktop environments (such as GNOME),
4605 however has been added to logind now, in order to fix a
4606 boot-time race where a desktop environment might not have
4607 been started yet and thus not been able to take an inhibitor
4608 lock at the time where logind already suspends the system
4609 due to a closed lid.
4611 * logind will now wait at least 30s after each system
4612 suspend/resume cycle, and 3min after system boot before
4613 suspending the system due to a closed laptop lid. This
4614 should give USB docking stations and similar enough time to
4615 be probed and configured after system resume and boot in
4616 order to then act as suspend blocker.
4618 * systemd-run gained a new --property= setting which allows
4619 initialization of resource control properties (and others)
4620 for the created scope or service unit. Example: "systemd-run
4621 --property=BlockIOWeight=10 updatedb" may be used to run
4622 updatedb at a low block IO scheduling weight.
4624 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4625 now also work in --scope mode.
4627 * When systemd is compiled with kdbus support, basic support
4628 for enforced policies is now in place. (Note that enabling
4629 kdbus still voids your warranty and no API compatibility
4632 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4633 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4634 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4635 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4636 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4637 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4638 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4639 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4640 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4641 Zbigniew Jędrzejewski-Szmek
4643 — Berlin, 2014-03-12
4647 * systemd will now relabel /dev after loading the SMACK policy
4648 according to SMACK rules.
4650 * A new unit file option AppArmorProfile= has been added to
4651 set the AppArmor profile for the processes of a unit.
4653 * A new condition check ConditionArchitecture= has been added
4654 to conditionalize units based on the system architecture, as
4655 reported by uname()'s "machine" field.
4657 * systemd-networkd now supports matching on the system
4658 virtualization, architecture, kernel command line, host name
4661 * logind is now a lot more aggressive when suspending the
4662 machine due to a closed laptop lid. Instead of acting only
4663 on the lid close action, it will continuously watch the lid
4664 status and act on it. This is useful for laptops where the
4665 power button is on the outside of the chassis so that it can
4666 be reached without opening the lid (such as the Lenovo
4667 Yoga). On those machines, logind will now immediately
4668 re-suspend the machine if the power button has been
4669 accidentally pressed while the laptop was suspended and in a
4670 backpack or similar.
4672 * logind will now watch SW_DOCK switches and inhibit reaction
4673 to the lid switch if it is pressed. This means that logind
4674 will not suspend the machine anymore if the lid is closed
4675 and the system is docked, if the laptop supports SW_DOCK
4676 notifications via the input layer. Note that ACPI docking
4677 stations do not generate this currently. Also note that this
4678 logic is usually not fully sufficient and Desktop
4679 Environments should take a lid switch inhibitor lock when an
4680 external display is connected, as systemd will not watch
4683 * nspawn will now make use of the devices cgroup controller by
4684 default, and only permit creation of and access to the usual
4685 API device nodes like /dev/null or /dev/random, as well as
4686 access to (but not creation of) the pty devices.
4688 * We will now ship a default .network file for
4689 systemd-networkd that automatically configures DHCP for
4690 network interfaces created by nspawn's --network-veth or
4691 --network-bridge= switches.
4693 * systemd will now understand the usual M, K, G, T suffixes
4694 according to SI conventions (i.e. to the base 1000) when
4695 referring to throughput and hardware metrics. It will stay
4696 with IEC conventions (i.e. to the base 1024) for software
4697 metrics, according to what is customary according to
4698 Wikipedia. We explicitly document which base applies for
4699 each configuration option.
4701 * The DeviceAllow= setting in unit files now supports a syntax
4702 to whitelist an entire group of devices node majors at once,
4703 based on the /proc/devices listing. For example, with the
4704 string "char-pts", it is now possible to whitelist all
4705 current and future pseudo-TTYs at once.
4707 * sd-event learned a new "post" event source. Event sources of
4708 this type are triggered by the dispatching of any event
4709 source of a type that is not "post". This is useful for
4710 implementing clean-up and check event sources that are
4711 triggered by other work being done in the program.
4713 * systemd-networkd is no longer statically enabled, but uses
4714 the usual [Install] sections so that it can be
4715 enabled/disabled using systemctl. It still is enabled by
4718 * When creating a veth interface pair with systemd-nspawn, the
4719 host side will now be prefixed with "vb-" if
4720 --network-bridge= is used, and with "ve-" if --network-veth
4721 is used. This way, it is easy to distinguish these cases on
4722 the host, for example to apply different configuration to
4723 them with systemd-networkd.
4725 * The compatibility libraries for libsystemd-journal.so,
4726 libsystem-id128.so, libsystemd-login.so and
4727 libsystemd-daemon.so do not make use of IFUNC
4728 anymore. Instead, we now build libsystemd.so multiple times
4729 under these alternative names. This means that the footprint
4730 is drastically increased, but given that these are
4731 transitional compatibility libraries, this should not matter
4732 much. This change has been made necessary to support the ARM
4733 platform for these compatibility libraries, as the ARM
4734 toolchain is not really at the same level as the toolchain
4735 for other architectures like x86 and does not support
4736 IFUNC. Please make sure to use --enable-compat-libs only
4737 during a transitional period!
4739 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4740 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4741 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4742 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4743 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4744 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4745 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4746 Zbigniew Jędrzejewski-Szmek
4748 — Berlin, 2014-02-24
4752 * A new component "systemd-networkd" has been added that can
4753 be used to configure local network interfaces statically or
4754 via DHCP. It is capable of bringing up bridges, VLANs, and
4755 bonding. Currently, no hook-ups for interactive network
4756 configuration are provided. Use this for your initrd,
4757 container, embedded, or server setup if you need a simple,
4758 yet powerful, network configuration solution. This
4759 configuration subsystem is quite nifty, as it allows wildcard
4760 hotplug matching in interfaces. For example, with a single
4761 configuration snippet, you can configure that all Ethernet
4762 interfaces showing up are automatically added to a bridge,
4763 or similar. It supports link-sensing and more.
4765 * A new tool "systemd-socket-proxyd" has been added which can
4766 act as a bidirectional proxy for TCP sockets. This is
4767 useful for adding socket activation support to services that
4768 do not actually support socket activation, including virtual
4769 machines and the like.
4771 * Add a new tool to save/restore rfkill state on
4774 * Save/restore state of keyboard backlights in addition to
4775 display backlights on shutdown/boot.
4777 * udev learned a new SECLABEL{} construct to label device
4778 nodes with a specific security label when they appear. For
4779 now, only SECLABEL{selinux} is supported, but the syntax is
4780 prepared for additional security frameworks.
4782 * udev gained a new scheme to configure link-level attributes
4783 from files in /etc/systemd/network/*.link. These files can
4784 match against MAC address, device path, driver name and type,
4785 and will apply attributes like the naming policy, link speed,
4786 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4787 address assignment policy (randomized, ...).
4789 * The configuration of network interface naming rules for
4790 "permanent interface names" has changed: a new NamePolicy=
4791 setting in the [Link] section of .link files determines the
4792 priority of possible naming schemes (onboard, slot, MAC,
4793 path). The default value of this setting is determined by
4794 /usr/lib/net/links/99-default.link. Old
4795 80-net-name-slot.rules udev configuration file has been
4796 removed, so local configuration overriding this file should
4797 be adapted to override 99-default.link instead.
4799 * When the User= switch is used in a unit file, also
4800 initialize $SHELL= based on the user database entry.
4802 * systemd no longer depends on libdbus. All communication is
4803 now done with sd-bus, systemd's low-level bus library
4806 * kdbus support has been added to PID 1 itself. When kdbus is
4807 enabled, this causes PID 1 to set up the system bus and
4808 enable support for a new ".busname" unit type that
4809 encapsulates bus name activation on kdbus. It works a little
4810 bit like ".socket" units, except for bus names. A new
4811 generator has been added that converts classic dbus1 service
4812 activation files automatically into native systemd .busname
4815 * sd-bus: add a light-weight vtable implementation that allows
4816 defining objects on the bus with a simple static const
4817 vtable array of its methods, signals and properties.
4819 * systemd will not generate or install static dbus
4820 introspection data anymore to /usr/share/dbus-1/interfaces,
4821 as the precise format of these files is unclear, and
4822 nothing makes use of it.
4824 * A proxy daemon is now provided to proxy clients connecting
4825 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4826 compatibility with classic D-Bus.
4828 * A bus driver implementation has been added that supports the
4829 classic D-Bus bus driver calls on kdbus, also for
4830 compatibility purposes.
4832 * A new API "sd-event.h" has been added that implements a
4833 minimal event loop API built around epoll. It provides a
4834 couple of features that direct epoll usage is lacking:
4835 prioritization of events, scales to large numbers of timer
4836 events, per-event timer slack (accuracy), system-wide
4837 coalescing of timer events, exit handlers, watchdog
4838 supervision support using systemd's sd_notify() API, child
4841 * A new API "sd-rntl.h" has been added that provides an API
4842 around the route netlink interface of the kernel, similar in
4843 style to "sd-bus.h".
4845 * A new API "sd-dhcp-client.h" has been added that provides a
4846 small DHCPv4 client-side implementation. This is used by
4849 * There is a new kernel command line option
4850 "systemd.restore_state=0|1". When set to "0", none of the
4851 systemd tools will restore saved runtime state to hardware
4852 devices. More specifically, the rfkill and backlight states
4855 * The FsckPassNo= compatibility option in mount/service units
4856 has been removed. The fstab generator will now add the
4857 necessary dependencies automatically, and does not require
4858 PID1's support for that anymore.
4860 * journalctl gained a new switch, --list-boots, that lists
4861 recent boots with their times and boot IDs.
4863 * The various tools like systemctl, loginctl, timedatectl,
4864 busctl, systemd-run, ... have gained a new switch "-M" to
4865 connect to a specific, local OS container (as direct
4866 connection, without requiring SSH). This works on any
4867 container that is registered with machined, such as those
4868 created by libvirt-lxc or nspawn.
4870 * systemd-run and systemd-analyze also gained support for "-H"
4871 to connect to remote hosts via SSH. This is particularly
4872 useful for systemd-run because it enables queuing of jobs
4873 onto remote systems.
4875 * machinectl gained a new command "login" to open a getty
4876 login in any local container. This works with any container
4877 that is registered with machined (such as those created by
4878 libvirt-lxc or nspawn), and which runs systemd inside.
4880 * machinectl gained a new "reboot" command that may be used to
4881 trigger a reboot on a specific container that is registered
4882 with machined. This works on any container that runs an init
4883 system of some kind.
4885 * systemctl gained a new "list-timers" command to print a nice
4886 listing of installed timer units with the times they elapse
4889 * Alternative reboot() parameters may now be specified on the
4890 "systemctl reboot" command line and are passed to the
4891 reboot() system call.
4893 * systemctl gained a new --job-mode= switch to configure the
4894 mode to queue a job with. This is a more generic version of
4895 --fail, --irreversible, and --ignore-dependencies, which are
4896 still available but not advertised anymore.
4898 * /etc/systemd/system.conf gained new settings to configure
4899 various default timeouts of units, as well as the default
4900 start limit interval and burst. These may still be overridden
4903 * PID1 will now export on the bus profile data of the security
4904 policy upload process (such as the SELinux policy upload to
4907 * journald: when forwarding logs to the console, include
4908 timestamps (following the setting in
4909 /sys/module/printk/parameters/time).
4911 * OnCalendar= in timer units now understands the special
4912 strings "yearly" and "annually". (Both are equivalent)
4914 * The accuracy of timer units is now configurable with the new
4915 AccuracySec= setting. It defaults to 1min.
4917 * A new dependency type JoinsNamespaceOf= has been added that
4918 allows running two services within the same /tmp and network
4919 namespace, if PrivateNetwork= or PrivateTmp= are used.
4921 * A new command "cat" has been added to systemctl. It outputs
4922 the original unit file of a unit, and concatenates the
4923 contents of additional "drop-in" unit file snippets, so that
4924 the full configuration is shown.
4926 * systemctl now supports globbing on the various "list-xyz"
4927 commands, like "list-units" or "list-sockets", as well as on
4928 those commands which take multiple unit names.
4930 * journalctl's --unit= switch gained support for globbing.
4932 * All systemd daemons now make use of the watchdog logic so
4933 that systemd automatically notices when they hang.
4935 * If the $container_ttys environment variable is set,
4936 getty-generator will automatically spawn a getty for each
4937 listed tty. This is useful for container managers to request
4938 login gettys to be spawned on as many ttys as needed.
4940 * %h, %s, %U specifier support is not available anymore when
4941 used in unit files for PID 1. This is because NSS calls are
4942 not safe from PID 1. They stay available for --user
4943 instances of systemd, and as special case for the root user.
4945 * loginctl gained a new "--no-legend" switch to turn off output
4948 * The "sd-login.h" API gained three new calls:
4949 sd_session_is_remote(), sd_session_get_remote_user(),
4950 sd_session_get_remote_host() to query information about
4953 * The udev hardware database now also carries vendor/product
4954 information of SDIO devices.
4956 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4957 determine whether watchdog notifications are requested by
4960 * Socket-activated per-connection services now include a
4961 short description of the connection parameters in the
4964 * tmpfiles gained a new "--boot" option. When this is not used,
4965 only lines where the command character is not suffixed with
4966 "!" are executed. When this option is specified, those
4967 options are executed too. This partitions tmpfiles
4968 directives into those that can be safely executed at any
4969 time, and those which should be run only at boot (for
4970 example, a line that creates /run/nologin).
4972 * A new API "sd-resolve.h" has been added which provides a simple
4973 asynchronous wrapper around glibc NSS host name resolution
4974 calls, such as getaddrinfo(). In contrast to glibc's
4975 getaddrinfo_a(), it does not use signals. In contrast to most
4976 other asynchronous name resolution libraries, this one does
4977 not reimplement DNS, but reuses NSS, so that alternate
4978 host name resolution systems continue to work, such as mDNS,
4979 LDAP, etc. This API is based on libasyncns, but it has been
4980 cleaned up for inclusion in systemd.
4982 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4983 "sd-daemon.h" are no longer found in individual libraries
4984 libsystemd-journal.so, libsystemd-login.so,
4985 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4986 merged them into a single library, libsystemd.so, which
4987 provides all symbols. The reason for this is cyclic
4988 dependencies, as these libraries tend to use each other's
4989 symbols. So far, we have managed to workaround that by linking
4990 a copy of a good part of our code into each of these
4991 libraries again and again, which, however, makes certain
4992 things hard to do, like sharing static variables. Also, it
4993 substantially increases footprint. With this change, there
4994 is only one library for the basic APIs systemd
4995 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4996 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4997 library as well, however are subject to the --enable-kdbus
4998 switch (see below). Note that "sd-dhcp-client.h" is not part
4999 of this library (this is because it only consumes, never
5000 provides, services of/to other APIs). To make the transition
5001 easy from the separate libraries to the unified one, we
5002 provide the --enable-compat-libs compile-time switch which
5003 will generate stub libraries that are compatible with the
5004 old ones but redirect all calls to the new one.
5006 * All of the kdbus logic and the new APIs "sd-bus.h",
5007 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5008 and "sd-utf8.h" are compile-time optional via the
5009 "--enable-kdbus" switch, and they are not compiled in by
5010 default. To make use of kdbus, you have to explicitly enable
5011 the switch. Note however, that neither the kernel nor the
5012 userspace API for all of this is considered stable yet. We
5013 want to maintain the freedom to still change the APIs for
5014 now. By specifying this build-time switch, you acknowledge
5015 that you are aware of the instability of the current
5018 * Also, note that while kdbus is pretty much complete,
5019 it lacks one thing: proper policy support. This means you
5020 can build a fully working system with all features; however,
5021 it will be highly insecure. Policy support will be added in
5022 one of the next releases, at the same time that we will
5023 declare the APIs stable.
5025 * When the kernel command line argument "kdbus" is specified,
5026 systemd will automatically load the kdbus.ko kernel module. At
5027 this stage of development, it is only useful for testing kdbus
5028 and should not be used in production. Note: if "--enable-kdbus"
5029 is specified, and the kdbus.ko kernel module is available, and
5030 "kdbus" is added to the kernel command line, the entire system
5031 runs with kdbus instead of dbus-daemon, with the above mentioned
5032 problem of missing the system policy enforcement. Also a future
5033 version of kdbus.ko or a newer systemd will not be compatible with
5034 each other, and will unlikely be able to boot the machine if only
5035 one of them is updated.
5037 * systemctl gained a new "import-environment" command which
5038 uploads the caller's environment (or parts thereof) into the
5039 service manager so that it is inherited by services started
5040 by the manager. This is useful to upload variables like
5041 $DISPLAY into the user service manager.
5043 * A new PrivateDevices= switch has been added to service units
5044 which allows running a service with a namespaced /dev
5045 directory that does not contain any device nodes for
5046 physical devices. More specifically, it only includes devices
5047 such as /dev/null, /dev/urandom, and /dev/zero which are API
5050 * logind has been extended to support behaviour like VT
5051 switching on seats that do not support a VT. This makes
5052 multi-session available on seats that are not the first seat
5053 (seat0), and on systems where kernel support for VTs has
5054 been disabled at compile-time.
5056 * If a process holds a delay lock for system sleep or shutdown
5057 and fails to release it in time, we will now log its
5058 identity. This makes it easier to identify processes that
5059 cause slow suspends or power-offs.
5061 * When parsing /etc/crypttab, support for a new key-slot=
5062 option as supported by Debian is added. It allows indicating
5063 which LUKS slot to use on disk, speeding up key loading.
5065 * The sd_journald_sendv() API call has been checked and
5066 officially declared to be async-signal-safe so that it may
5067 be invoked from signal handlers for logging purposes.
5069 * Boot-time status output is now enabled automatically after a
5070 short timeout if boot does not progress, in order to give
5071 the user an indication what she or he is waiting for.
5073 * The boot-time output has been improved to show how much time
5074 remains until jobs expire.
5076 * The KillMode= switch in service units gained a new possible
5077 value "mixed". If set, and the unit is shut down, then the
5078 initial SIGTERM signal is sent only to the main daemon
5079 process, while the following SIGKILL signal is sent to
5080 all remaining processes of the service.
5082 * When a scope unit is registered, a new property "Controller"
5083 may be set. If set to a valid bus name, systemd will send a
5084 RequestStop() signal to this name when it would like to shut
5085 down the scope. This may be used to hook manager logic into
5086 the shutdown logic of scope units. Also, scope units may now
5087 be put in a special "abandoned" state, in which case the
5088 manager process which created them takes no further
5089 responsibilities for it.
5091 * When reading unit files, systemd will now verify
5092 the access mode of these files, and warn about certain
5093 suspicious combinations. This has been added to make it
5094 easier to track down packaging bugs where unit files are
5095 marked executable or world-writable.
5097 * systemd-nspawn gained a new "--setenv=" switch to set
5098 container-wide environment variables. The similar option in
5099 systemd-activate was renamed from "--environment=" to
5100 "--setenv=" for consistency.
5102 * systemd-nspawn has been updated to create a new kdbus domain
5103 for each container that is invoked, thus allowing each
5104 container to have its own set of system and user buses,
5105 independent of the host.
5107 * systemd-nspawn gained a new --drop-capability= switch to run
5108 the container with less capabilities than the default. Both
5109 --drop-capability= and --capability= now take the special
5110 string "all" for dropping or keeping all capabilities.
5112 * systemd-nspawn gained new switches for executing containers
5113 with specific SELinux labels set.
5115 * systemd-nspawn gained a new --quiet switch to not generate
5116 any additional output but the container's own console
5119 * systemd-nspawn gained a new --share-system switch to run a
5120 container without PID namespacing enabled.
5122 * systemd-nspawn gained a new --register= switch to control
5123 whether the container is registered with systemd-machined or
5124 not. This is useful for containers that do not run full
5125 OS images, but only specific apps.
5127 * systemd-nspawn gained a new --keep-unit which may be used
5128 when invoked as the only program from a service unit, and
5129 results in registration of the unit service itself in
5130 systemd-machined, instead of a newly opened scope unit.
5132 * systemd-nspawn gained a new --network-interface= switch for
5133 moving arbitrary interfaces to the container. The new
5134 --network-veth switch creates a virtual Ethernet connection
5135 between host and container. The new --network-bridge=
5136 switch then allows assigning the host side of this virtual
5137 Ethernet connection to a bridge device.
5139 * systemd-nspawn gained a new --personality= switch for
5140 setting the kernel personality for the container. This is
5141 useful when running a 32-bit container on a 64-bit host. A
5142 similar option Personality= is now also available for service
5145 * logind will now also track a "Desktop" identifier for each
5146 session which encodes the desktop environment of it. This is
5147 useful for desktop environments that want to identify
5148 multiple running sessions of itself easily.
5150 * A new SELinuxContext= setting for service units has been
5151 added that allows setting a specific SELinux execution
5152 context for a service.
5154 * Most systemd client tools will now honour $SYSTEMD_LESS for
5155 settings of the "less" pager. By default, these tools will
5156 override $LESS to allow certain operations to work, such as
5157 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5158 influence this logic.
5160 * systemd's "seccomp" hook-up has been changed to make use of
5161 the libseccomp library instead of using its own
5162 implementation. This has benefits for portability among
5165 * For usage together with SystemCallFilter=, a new
5166 SystemCallErrorNumber= setting has been introduced that
5167 allows configuration of a system error number to be returned
5168 on filtered system calls, instead of immediately killing the
5169 process. Also, SystemCallArchitectures= has been added to
5170 limit access to system calls of a particular architecture
5171 (in order to turn off support for unused secondary
5172 architectures). There is also a global
5173 SystemCallArchitectures= setting in system.conf now to turn
5174 off support for non-native system calls system-wide.
5176 * systemd requires a kernel with a working name_to_handle_at(),
5177 please see the kernel config requirements in the README file.
5179 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5180 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5181 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5182 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5183 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5184 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5185 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5186 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5187 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5188 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5189 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5190 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5191 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5192 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5193 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5194 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5195 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5196 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5197 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5198 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5199 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5200 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5201 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5202 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5204 — Berlin, 2014-02-20
5208 * logind has gained support for facilitating privileged input
5209 and drm device access for unprivileged clients. This work is
5210 useful to allow Wayland display servers (and similar
5211 programs, such as kmscon) to run under the user's ID and
5212 access input and drm devices which are normally
5213 protected. When this is used (and the kernel is new enough)
5214 logind will "mute" IO on the file descriptors passed to
5215 Wayland as long as it is in the background and "unmute" it
5216 if it returns into the foreground. This allows secure
5217 session switching without allowing background sessions to
5218 eavesdrop on input and display data. This also introduces
5219 session switching support if VT support is turned off in the
5220 kernel, and on seats that are not seat0.
5222 * A new kernel command line option luks.options= is understood
5223 now which allows specifying LUKS options for usage for LUKS
5224 encrypted partitions specified with luks.uuid=.
5226 * tmpfiles.d(5) snippets may now use specifier expansion in
5227 path names. More specifically %m, %b, %H, %v, are now
5228 replaced by the local machine id, boot id, hostname, and
5229 kernel version number.
5231 * A new tmpfiles.d(5) command "m" has been introduced which
5232 may be used to change the owner/group/access mode of a file
5233 or directory if it exists, but do nothing if it does not.
5235 * This release removes high-level support for the
5236 MemorySoftLimit= cgroup setting. The underlying kernel
5237 cgroup attribute memory.soft_limit= is currently badly
5238 designed and likely to be removed from the kernel API in its
5239 current form, hence we should not expose it for now.
5241 * The memory.use_hierarchy cgroup attribute is now enabled for
5242 all cgroups systemd creates in the memory cgroup
5243 hierarchy. This option is likely to be come the built-in
5244 default in the kernel anyway, and the non-hierarchical mode
5245 never made much sense in the intrinsically hierarchical
5248 * A new field _SYSTEMD_SLICE= is logged along with all journal
5249 messages containing the slice a message was generated
5250 from. This is useful to allow easy per-customer filtering of
5251 logs among other things.
5253 * systemd-journald will no longer adjust the group of journal
5254 files it creates to the "systemd-journal" group. Instead we
5255 rely on the journal directory to be owned by the
5256 "systemd-journal" group, and its setgid bit set, so that the
5257 kernel file system layer will automatically enforce that
5258 journal files inherit this group assignment. The reason for
5259 this change is that we cannot allow NSS look-ups from
5260 journald which would be necessary to resolve
5261 "systemd-journal" to a numeric GID, because this might
5262 create deadlocks if NSS involves synchronous queries to
5263 other daemons (such as nscd, or sssd) which in turn are
5264 logging clients of journald and might block on it, which
5265 would then dead lock. A tmpfiles.d(5) snippet included in
5266 systemd will make sure the setgid bit and group are
5267 properly set on the journal directory if it exists on every
5268 boot. However, we recommend adjusting it manually after
5269 upgrades too (or from RPM scriptlets), so that the change is
5270 not delayed until next reboot.
5272 * Backlight and random seed files in /var/lib/ have moved into
5273 the /var/lib/systemd/ directory, in order to centralize all
5274 systemd generated files in one directory.
5276 * Boot time performance measurements (as displayed by
5277 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5278 performance information if that's available to determine how
5279 much time BIOS and boot loader initialization required. With
5280 a sufficiently new BIOS you hence no longer need to boot
5281 with Gummiboot to get access to such information.
5283 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5284 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5285 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5286 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5287 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5288 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5289 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5291 — Berlin, 2013-10-02
5295 * The Restart= option for services now understands a new
5296 on-watchdog setting, which will restart the service
5297 automatically if the service stops sending out watchdog keep
5298 alive messages (as configured with WatchdogSec=).
5300 * The getty generator (which is responsible for bringing up a
5301 getty on configured serial consoles) will no longer only
5302 start a getty on the primary kernel console but on all
5303 others, too. This makes the order in which console= is
5304 specified on the kernel command line less important.
5306 * libsystemd-logind gained a new sd_session_get_vt() call to
5307 retrieve the VT number of a session.
5309 * If the option "tries=0" is set for an entry of /etc/crypttab
5310 its passphrase is queried indefinitely instead of any
5311 maximum number of tries.
5313 * If a service with a configure PID file terminates its PID
5314 file will now be removed automatically if it still exists
5315 afterwards. This should put an end to stale PID files.
5317 * systemd-run will now also take relative binary path names
5318 for execution and no longer insists on absolute paths.
5320 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5321 paths that are optionally prefixed with "-" to indicate that
5322 it should not be considered a failure if they do not exist.
5324 * journalctl -o (and similar commands) now understands a new
5325 output mode "short-precise", it is similar to "short" but
5326 shows timestamps with usec accuracy.
5328 * The option "discard" (as known from Debian) is now
5329 synonymous to "allow-discards" in /etc/crypttab. In fact,
5330 "discard" is preferred now (since it is easier to remember
5333 * Some licensing clean-ups were made, so that more code is now
5334 LGPL-2.1 licensed than before.
5336 * A minimal tool to save/restore the display backlight
5337 brightness across reboots has been added. It will store the
5338 backlight setting as late as possible at shutdown, and
5339 restore it as early as possible during reboot.
5341 * A logic to automatically discover and enable home and swap
5342 partitions on GPT disks has been added. With this in place
5343 /etc/fstab becomes optional for many setups as systemd can
5344 discover certain partitions located on the root disk
5345 automatically. Home partitions are recognized under their
5346 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5347 partitions are recognized under their GPT type ID
5348 0657fd6da4ab43c484e50933c84b4f4f.
5350 * systemd will no longer pass any environment from the kernel
5351 or initrd to system services. If you want to set an
5352 environment for all services, do so via the kernel command
5353 line systemd.setenv= assignment.
5355 * The systemd-sysctl tool no longer natively reads the file
5356 /etc/sysctl.conf. If desired, the file should be symlinked
5357 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5358 legacy support by a symlink rather than built-in code, it
5359 also makes the otherwise hidden order of application of the
5360 different files visible. (Note that this partly reverts to a
5361 pre-198 application order of sysctl knobs!)
5363 * The "systemctl set-log-level" and "systemctl dump" commands
5364 have been moved to systemd-analyze.
5366 * systemd-run learned the new --remain-after-exit switch,
5367 which causes the scope unit not to be cleaned up
5368 automatically after the process terminated.
5370 * tmpfiles learned a new --exclude-prefix= switch to exclude
5371 certain paths from operation.
5373 * journald will now automatically flush all messages to disk
5374 as soon as a message at the log level CRIT, ALERT or EMERG
5377 Contributions from: Andrew Cook, Brandon Philips, Christian
5378 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5379 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5380 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5381 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5382 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5383 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5384 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5385 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5386 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5387 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5388 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5389 William Giokas, Zbigniew Jędrzejewski-Szmek
5391 — Berlin, 2013-09-13
5395 * The documentation has been updated to cover the various new
5396 concepts introduced with 205.
5398 * Unit files now understand the new %v specifier which
5399 resolves to the kernel version string as returned by "uname
5402 * systemctl now supports filtering the unit list output by
5403 load state, active state and sub state, using the new
5406 * "systemctl status" will now show the results of the
5407 condition checks (like ConditionPathExists= and similar) of
5408 the last start attempts of the unit. They are also logged to
5411 * "journalctl -b" may now be used to look for boot output of a
5412 specific boot. Try "journalctl -b -1" for the previous boot,
5413 but the syntax is substantially more powerful.
5415 * "journalctl --show-cursor" has been added which prints the
5416 cursor string the last shown log line. This may then be used
5417 with the new "journalctl --after-cursor=" switch to continue
5418 browsing logs from that point on.
5420 * "journalctl --force" may now be used to force regeneration
5423 * Creation of "dead" device nodes has been moved from udev
5424 into kmod and tmpfiles. Previously, udev would read the kmod
5425 databases to pre-generate dead device nodes based on meta
5426 information contained in kernel modules, so that these would
5427 be auto-loaded on access rather then at boot. As this
5428 does not really have much to do with the exposing actual
5429 kernel devices to userspace this has always been slightly
5430 alien in the udev codebase. Following the new scheme kmod
5431 will now generate a runtime snippet for tmpfiles from the
5432 module meta information and it now is tmpfiles' job to the
5433 create the nodes. This also allows overriding access and
5434 other parameters for the nodes using the usual tmpfiles
5435 facilities. As side effect this allows us to remove the
5436 CAP_SYS_MKNOD capability bit from udevd entirely.
5438 * logind's device ACLs may now be applied to these "dead"
5439 devices nodes too, thus finally allowing managed access to
5440 devices such as /dev/snd/sequencer without loading the
5441 backing module right-away.
5443 * A new RPM macro has been added that may be used to apply
5444 tmpfiles configuration during package installation.
5446 * systemd-detect-virt and ConditionVirtualization= now can
5447 detect User-Mode-Linux machines (UML).
5449 * journald will now implicitly log the effective capabilities
5450 set of processes in the message metadata.
5452 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5454 * The initrd interface has been simplified (more specifically,
5455 support for passing performance data via environment
5456 variables and fsck results via files in /run has been
5457 removed). These features were non-essential, and are
5458 nowadays available in a much nicer way by having systemd in
5459 the initrd serialize its state and have the hosts systemd
5460 deserialize it again.
5462 * The udev "keymap" data files and tools to apply keyboard
5463 specific mappings of scan to key codes, and force-release
5464 scan code lists have been entirely replaced by a udev
5465 "keyboard" builtin and a hwdb data file.
5467 * systemd will now honour the kernel's "quiet" command line
5468 argument also during late shutdown, resulting in a
5469 completely silent shutdown when used.
5471 * There's now an option to control the SO_REUSEPORT socket
5472 option in .socket units.
5474 * Instance units will now automatically get a per-template
5475 subslice of system.slice unless something else is explicitly
5476 configured. For example, instances of sshd@.service will now
5477 implicitly be placed in system-sshd.slice rather than
5478 system.slice as before.
5480 * Test coverage support may now be enabled at build time.
5482 Contributions from: Dave Reisner, Frederic Crozat, Harald
5483 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5484 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5485 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5486 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5487 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5488 Giokas, Zbigniew Jędrzejewski-Szmek
5490 — Berlin, 2013-07-23
5494 * Two new unit types have been introduced:
5496 Scope units are very similar to service units, however, are
5497 created out of pre-existing processes — instead of PID 1
5498 forking off the processes. By using scope units it is
5499 possible for system services and applications to group their
5500 own child processes (worker processes) in a powerful way
5501 which then maybe used to organize them, or kill them
5502 together, or apply resource limits on them.
5504 Slice units may be used to partition system resources in an
5505 hierarchical fashion and then assign other units to them. By
5506 default there are now three slices: system.slice (for all
5507 system services), user.slice (for all user sessions),
5508 machine.slice (for VMs and containers).
5510 Slices and scopes have been introduced primarily in
5511 context of the work to move cgroup handling to a
5512 single-writer scheme, where only PID 1
5513 creates/removes/manages cgroups.
5515 * There's a new concept of "transient" units. In contrast to
5516 normal units these units are created via an API at runtime,
5517 not from configuration from disk. More specifically this
5518 means it is now possible to run arbitrary programs as
5519 independent services, with all execution parameters passed
5520 in via bus APIs rather than read from disk. Transient units
5521 make systemd substantially more dynamic then it ever was,
5522 and useful as a general batch manager.
5524 * logind has been updated to make use of scope and slice units
5525 for managing user sessions. As a user logs in he will get
5526 his own private slice unit, to which all sessions are added
5527 as scope units. We also added support for automatically
5528 adding an instance of user@.service for the user into the
5529 slice. Effectively logind will no longer create cgroup
5530 hierarchies on its own now, it will defer entirely to PID 1
5531 for this by means of scope, service and slice units. Since
5532 user sessions this way become entities managed by PID 1
5533 the output of "systemctl" is now a lot more comprehensive.
5535 * A new mini-daemon "systemd-machined" has been added which
5536 may be used by virtualization managers to register local
5537 VMs/containers. nspawn has been updated accordingly, and
5538 libvirt will be updated shortly. machined will collect a bit
5539 of meta information about the VMs/containers, and assign
5540 them their own scope unit (see above). The collected
5541 meta-data is then made available via the "machinectl" tool,
5542 and exposed in "ps" and similar tools. machined/machinectl
5543 is compile-time optional.
5545 * As discussed earlier, the low-level cgroup configuration
5546 options ControlGroup=, ControlGroupModify=,
5547 ControlGroupPersistent=, ControlGroupAttribute= have been
5548 removed. Please use high-level attribute settings instead as
5549 well as slice units.
5551 * A new bus call SetUnitProperties() has been added to alter
5552 various runtime parameters of a unit. This is primarily
5553 useful to alter cgroup parameters dynamically in a nice way,
5554 but will be extended later on to make more properties
5555 modifiable at runtime. systemctl gained a new set-properties
5556 command that wraps this call.
5558 * A new tool "systemd-run" has been added which can be used to
5559 run arbitrary command lines as transient services or scopes,
5560 while configuring a number of settings via the command
5561 line. This tool is currently very basic, however already
5562 very useful. We plan to extend this tool to even allow
5563 queuing of execution jobs with time triggers from the
5564 command line, similar in fashion to "at".
5566 * nspawn will now inform the user explicitly that kernels with
5567 audit enabled break containers, and suggest the user to turn
5570 * Support for detecting the IMA and AppArmor security
5571 frameworks with ConditionSecurity= has been added.
5573 * journalctl gained a new "-k" switch for showing only kernel
5574 messages, mimicking dmesg output; in addition to "--user"
5575 and "--system" switches for showing only user's own logs
5578 * systemd-delta can now show information about drop-in
5579 snippets extending unit files.
5581 * libsystemd-bus has been substantially updated but is still
5582 not available as public API.
5584 * systemd will now look for the "debug" argument on the kernel
5585 command line and enable debug logging, similar to what
5586 "systemd.log_level=debug" already did before.
5588 * "systemctl set-default", "systemctl get-default" has been
5589 added to configure the default.target symlink, which
5590 controls what to boot into by default.
5592 * "systemctl set-log-level" has been added as a convenient
5593 way to raise and lower systemd logging threshold.
5595 * "systemd-analyze plot" will now show the time the various
5596 generators needed for execution, as well as information
5597 about the unit file loading.
5599 * libsystemd-journal gained a new sd_journal_open_files() call
5600 for opening specific journal files. journactl also gained a
5601 new switch to expose this new functionality. Previously we
5602 only supported opening all files from a directory, or all
5603 files from the system, as opening individual files only is
5604 racy due to journal file rotation.
5606 * systemd gained the new DefaultEnvironment= setting in
5607 /etc/systemd/system.conf to set environment variables for
5610 * If a privileged process logs a journal message with the
5611 OBJECT_PID= field set, then journald will automatically
5612 augment this with additional OBJECT_UID=, OBJECT_GID=,
5613 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5614 system services want to log events about specific client
5615 processes. journactl/systemctl has been updated to make use
5616 of this information if all log messages regarding a specific
5619 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5620 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5621 Reisner, David Coppa, David King, David Strauss, Eelco
5622 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5623 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5624 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5625 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5626 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5627 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5628 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5629 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5630 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5631 Łukasz Stelmach, 장동준
5635 * The Python bindings gained some minimal support for the APIs
5636 exposed by libsystemd-logind.
5638 * ConditionSecurity= gained support for detecting SMACK. Since
5639 this condition already supports SELinux and AppArmor we only
5640 miss IMA for this. Patches welcome!
5642 Contributions from: Karol Lewandowski, Lennart Poettering,
5643 Zbigniew Jędrzejewski-Szmek
5647 * systemd-nspawn will now create /etc/resolv.conf if
5648 necessary, before bind-mounting the host's file onto it.
5650 * systemd-nspawn will now store meta information about a
5651 container on the container's cgroup as extended attribute
5652 fields, including the root directory.
5654 * The cgroup hierarchy has been reworked in many ways. All
5655 objects any of the components systemd creates in the cgroup
5656 tree are now suffixed. More specifically, user sessions are
5657 now placed in cgroups suffixed with ".session", users in
5658 cgroups suffixed with ".user", and nspawn containers in
5659 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5660 names are now escaped in a simple scheme to avoid collision
5661 of userspace object names with kernel filenames. This work
5662 is preparation for making these objects relocatable in the
5663 cgroup tree, in order to allow easy resource partitioning of
5664 these objects without causing naming conflicts.
5666 * systemctl list-dependencies gained the new switches
5667 --plain, --reverse, --after and --before.
5669 * systemd-inhibit now shows the process name of processes that
5670 have taken an inhibitor lock.
5672 * nss-myhostname will now also resolve "localhost"
5673 implicitly. This makes /etc/hosts an optional file and
5674 nicely handles that on IPv6 ::1 maps to both "localhost" and
5677 * libsystemd-logind.so gained a new call
5678 sd_get_machine_names() to enumerate running containers and
5679 VMs (currently only supported by very new libvirt and
5680 nspawn). sd_login_monitor can now be used to watch
5681 VMs/containers coming and going.
5683 * .include is not allowed recursively anymore, and only in
5684 unit files. Usually it is better to use drop-in snippets in
5685 .d/*.conf anyway, as introduced with systemd 198.
5687 * systemd-analyze gained a new "critical-chain" command that
5688 determines the slowest chain of units run during system
5689 boot-up. It is very useful for tracking down where
5690 optimizing boot time is the most beneficial.
5692 * systemd will no longer allow manipulating service paths in
5693 the name=systemd:/system cgroup tree using ControlGroup= in
5694 units. (But is still fine with it in all other dirs.)
5696 * There's a new systemd-nspawn@.service service file that may
5697 be used to easily run nspawn containers as system
5698 services. With the container's root directory in
5699 /var/lib/container/foobar it is now sufficient to run
5700 "systemctl start systemd-nspawn@foobar.service" to boot it.
5702 * systemd-cgls gained a new parameter "--machine" to list only
5703 the processes within a certain container.
5705 * ConditionSecurity= now can check for "apparmor". We still
5706 are lacking checks for SMACK and IMA for this condition
5707 check though. Patches welcome!
5709 * A new configuration file /etc/systemd/sleep.conf has been
5710 added that may be used to configure which kernel operation
5711 systemd is supposed to execute when "suspend", "hibernate"
5712 or "hybrid-sleep" is requested. This makes the new kernel
5713 "freeze" state accessible to the user.
5715 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5716 the passed argument if applicable.
5718 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5719 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5720 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5721 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5722 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5723 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5724 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5729 * The output of 'systemctl list-jobs' got some polishing. The
5730 '--type=' argument may now be passed more than once. A new
5731 command 'systemctl list-sockets' has been added which shows
5732 a list of kernel sockets systemd is listening on with the
5733 socket units they belong to, plus the units these socket
5736 * The experimental libsystemd-bus library got substantial
5737 updates to work in conjunction with the (also experimental)
5738 kdbus kernel project. It works well enough to exchange
5739 messages with some sophistication. Note that kdbus is not
5740 ready yet, and the library is mostly an elaborate test case
5741 for now, and not installable.
5743 * systemd gained a new unit 'systemd-static-nodes.service'
5744 that generates static device nodes earlier during boot, and
5745 can run in conjunction with udev.
5747 * libsystemd-login gained a new call sd_pid_get_user_unit()
5748 to retrieve the user systemd unit a process is running
5749 in. This is useful for systems where systemd is used as
5752 * systemd-nspawn now places all containers in the new /machine
5753 top-level cgroup directory in the name=systemd
5754 hierarchy. libvirt will soon do the same, so that we get a
5755 uniform separation of /system, /user and /machine for system
5756 services, user processes and containers/virtual
5757 machines. This new cgroup hierarchy is also useful to stick
5758 stable names to specific container instances, which can be
5759 recognized later this way (this name may be controlled
5760 via systemd-nspawn's new -M switch). libsystemd-login also
5761 gained a new call sd_pid_get_machine_name() to retrieve the
5762 name of the container/VM a specific process belongs to.
5764 * bootchart can now store its data in the journal.
5766 * libsystemd-journal gained a new call
5767 sd_journal_add_conjunction() for AND expressions to the
5768 matching logic. This can be used to express more complex
5769 logical expressions.
5771 * journactl can now take multiple --unit= and --user-unit=
5774 * The cryptsetup logic now understands the "luks.key=" kernel
5775 command line switch for specifying a file to read the
5776 decryption key from. Also, if a configured key file is not
5777 found the tool will now automatically fall back to prompting
5780 * Python systemd.journal module was updated to wrap recently
5781 added functions from libsystemd-journal. The interface was
5782 changed to bring the low level interface in s.j._Reader
5783 closer to the C API, and the high level interface in
5784 s.j.Reader was updated to wrap and convert all data about
5787 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5788 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5789 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5790 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5791 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5792 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5796 * journalctl --update-catalog now understands a new --root=
5797 option to operate on catalogs found in a different root
5800 * During shutdown after systemd has terminated all running
5801 services a final killing loop kills all remaining left-over
5802 processes. We will now print the name of these processes
5803 when we send SIGKILL to them, since this usually indicates a
5806 * If /etc/crypttab refers to password files stored on
5807 configured mount points automatic dependencies will now be
5808 generated to ensure the specific mount is established first
5809 before the key file is attempted to be read.
5811 * 'systemctl status' will now show information about the
5812 network sockets a socket unit is listening on.
5814 * 'systemctl status' will also shown information about any
5815 drop-in configuration file for units. (Drop-In configuration
5816 files in this context are files such as
5817 /etc/systemd/systemd/foobar.service.d/*.conf)
5819 * systemd-cgtop now optionally shows summed up CPU times of
5820 cgroups. Press '%' while running cgtop to switch between
5821 percentage and absolute mode. This is useful to determine
5822 which cgroups use up the most CPU time over the entire
5823 runtime of the system. systemd-cgtop has also been updated
5824 to be 'pipeable' for processing with further shell tools.
5826 * 'hostnamectl set-hostname' will now allow setting of FQDN
5829 * The formatting and parsing of time span values has been
5830 changed. The parser now understands fractional expressions
5831 such as "5.5h". The formatter will now output fractional
5832 expressions for all time spans under 1min, i.e. "5.123456s"
5833 rather than "5s 123ms 456us". For time spans under 1s
5834 millisecond values are shown, for those under 1ms
5835 microsecond values are shown. This should greatly improve
5836 all time-related output of systemd.
5838 * libsystemd-login and libsystemd-journal gained new
5839 functions for querying the poll() events mask and poll()
5840 timeout value for integration into arbitrary event
5843 * localectl gained the ability to list available X11 keymaps
5844 (models, layouts, variants, options).
5846 * 'systemd-analyze dot' gained the ability to filter for
5847 specific units via shell-style globs, to create smaller,
5848 more useful graphs. I.e. it is now possible to create simple
5849 graphs of all the dependencies between only target units, or
5850 of all units that Avahi has dependencies with.
5852 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5853 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5854 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5855 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5856 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5857 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5858 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5862 * The boot-time readahead implementation for rotating media
5863 will now read the read-ahead data in multiple passes which
5864 consist of all read requests made in equidistant time
5865 intervals. This means instead of strictly reading read-ahead
5866 data in its physical order on disk we now try to find a
5867 middle ground between physical and access time order.
5869 * /etc/os-release files gained a new BUILD_ID= field for usage
5870 on operating systems that provide continuous builds of OS
5873 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5874 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5875 William Douglas, Zbigniew Jędrzejewski-Szmek
5879 * systemd-python gained an API exposing libsystemd-daemon.
5881 * The SMACK setup logic gained support for uploading CIPSO
5884 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5885 ReadOnlyDirectories= and InaccessibleDirectories= has
5886 changed. The private /tmp and /var/tmp directories are now
5887 shared by all processes of a service (which means
5888 ExecStartPre= may now leave data in /tmp that ExecStart= of
5889 the same service can still access). When a service is
5890 stopped its temporary directories are immediately deleted
5891 (normal clean-up with tmpfiles is still done in addition to
5894 * By default, systemd will now set a couple of sysctl
5895 variables in the kernel: the safe sysrq options are turned
5896 on, IP route verification is turned on, and source routing
5897 disabled. The recently added hardlink and softlink
5898 protection of the kernel is turned on. These settings should
5899 be reasonably safe, and good defaults for all new systems.
5901 * The predictable network naming logic may now be turned off
5902 with a new kernel command line switch: net.ifnames=0.
5904 * A new libsystemd-bus module has been added that implements a
5905 pretty complete D-Bus client library. For details see:
5907 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5909 * journald will now explicitly flush the journal files to disk
5910 at the latest 5min after each write. The file will then also
5911 be marked offline until the next write. This should increase
5912 reliability in case of a crash. The synchronization delay
5913 can be configured via SyncIntervalSec= in journald.conf.
5915 * There's a new remote-fs-setup.target unit that can be used
5916 to pull in specific services when at least one remote file
5917 system is to be mounted.
5919 * There are new targets timers.target and paths.target as
5920 canonical targets to pull user timer and path units in
5921 from. This complements sockets.target with a similar
5922 purpose for socket units.
5924 * libudev gained a new call udev_device_set_attribute_value()
5925 to set sysfs attributes of a device.
5927 * The udev daemon now sets the default number of worker
5928 processes executed in parallel based on the number of available
5929 CPUs instead of the amount of available RAM. This is supposed
5930 to provide a more reliable default and limit a too aggressive
5931 parallelism for setups with 1000s of devices connected.
5933 Contributions from: Auke Kok, Colin Walters, Cristian
5934 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5935 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5936 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5937 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5938 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5939 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5940 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5941 Zbigniew Jędrzejewski-Szmek
5945 * Configuration of unit files may now be extended via drop-in
5946 files without having to edit/override the unit files
5947 themselves. More specifically, if the administrator wants to
5948 change one value for a service file foobar.service he can
5949 now do so by dropping in a configuration snippet into
5950 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5951 will load all these snippets and apply them on top of the
5952 main unit configuration file, possibly extending or
5953 overriding its settings. Using these drop-in snippets is
5954 generally nicer than the two earlier options for changing
5955 unit files locally: copying the files from
5956 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5957 them there; or creating a new file in /etc/systemd/system/
5958 that incorporates the original one via ".include". Drop-in
5959 snippets into these .d/ directories can be placed in any
5960 directory systemd looks for units in, and the usual
5961 overriding semantics between /usr/lib, /etc and /run apply
5964 * Most unit file settings which take lists of items can now be
5965 reset by assigning the empty string to them. For example,
5966 normally, settings such as Environment=FOO=BAR append a new
5967 environment variable assignment to the environment block,
5968 each time they are used. By assigning Environment= the empty
5969 string the environment block can be reset to empty. This is
5970 particularly useful with the .d/*.conf drop-in snippets
5971 mentioned above, since this adds the ability to reset list
5972 settings from vendor unit files via these drop-ins.
5974 * systemctl gained a new "list-dependencies" command for
5975 listing the dependencies of a unit recursively.
5977 * Inhibitors are now honored and listed by "systemctl
5978 suspend", "systemctl poweroff" (and similar) too, not only
5979 GNOME. These commands will also list active sessions by
5982 * Resource limits (as exposed by the various control group
5983 controllers) can now be controlled dynamically at runtime
5984 for all units. More specifically, you can now use a command
5985 like "systemctl set-cgroup-attr foobar.service cpu.shares
5986 2000" to alter the CPU shares a specific service gets. These
5987 settings are stored persistently on disk, and thus allow the
5988 administrator to easily adjust the resource usage of
5989 services with a few simple commands. This dynamic resource
5990 management logic is also available to other programs via the
5991 bus. Almost any kernel cgroup attribute and controller is
5994 * systemd-vconsole-setup will now copy all font settings to
5995 all allocated VTs, where it previously applied them only to
5998 * libsystemd-login gained the new sd_session_get_tty() API
6001 * This release drops support for a few legacy or
6002 distribution-specific LSB facility names when parsing init
6003 scripts: $x-display-manager, $mail-transfer-agent,
6004 $mail-transport-agent, $mail-transfer-agent, $smtp,
6005 $null. Also, the mail-transfer-agent.target unit backing
6006 this has been removed. Distributions which want to retain
6007 compatibility with this should carry the burden for
6008 supporting this themselves and patch support for these back
6009 in, if they really need to. Also, the facilities $syslog and
6010 $local_fs are now ignored, since systemd does not support
6011 early-boot LSB init scripts anymore, and these facilities
6012 are implied anyway for normal services. syslog.target has
6015 * There are new bus calls on PID1's Manager object for
6016 cancelling jobs, and removing snapshot units. Previously,
6017 both calls were only available on the Job and Snapshot
6020 * systemd-journal-gatewayd gained SSL support.
6022 * The various "environment" files, such as /etc/locale.conf
6023 now support continuation lines with a backslash ("\") as
6024 last character in the line, similarly in style (but different)
6025 to how this is supported in shells.
6027 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6028 now implicitly appended to every log entry logged. systemctl
6029 has been updated to filter by this field when operating on a
6030 user systemd instance.
6032 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6033 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6034 the container. This makes it easier to boot unmodified
6035 Fedora systems in a container, which however still requires
6036 audit=0 to be passed on the kernel command line. Auditing in
6037 kernel and userspace is unfortunately still too broken in
6038 context of containers, hence we recommend compiling it out
6039 of the kernel or using audit=0. Hopefully this will be fixed
6040 one day for good in the kernel.
6042 * nspawn gained the new --bind= and --bind-ro= parameters to
6043 bind mount specific directories from the host into the
6046 * nspawn will now mount its own devpts file system instance
6047 into the container, in order not to leak pty devices from
6048 the host into the container.
6050 * systemd will now read the firmware boot time performance
6051 information from the EFI variables, if the used boot loader
6052 supports this, and takes it into account for boot performance
6053 analysis via "systemd-analyze". This is currently supported
6054 only in conjunction with Gummiboot, but could be supported
6055 by other boot loaders too. For details see:
6057 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6059 * A new generator has been added that automatically mounts the
6060 EFI System Partition (ESP) to /boot, if that directory
6061 exists, is empty, and no other file system has been
6062 configured to be mounted there.
6064 * logind will now send out PrepareForSleep(false) out
6065 unconditionally, after coming back from suspend. This may be
6066 used by applications as asynchronous notification for
6067 system resume events.
6069 * "systemctl unlock-sessions" has been added, that allows
6070 unlocking the screens of all user sessions at once, similar
6071 to how "systemctl lock-sessions" already locked all users
6072 sessions. This is backed by a new D-Bus call UnlockSessions().
6074 * "loginctl seat-status" will now show the master device of a
6075 seat. (i.e. the device of a seat that needs to be around for
6076 the seat to be considered available, usually the graphics
6079 * tmpfiles gained a new "X" line type, that allows
6080 configuration of files and directories (with wildcards) that
6081 shall be excluded from automatic cleanup ("aging").
6083 * udev default rules set the device node permissions now only
6084 at "add" events, and do not change them any longer with a
6085 later "change" event.
6087 * The log messages for lid events and power/sleep keypresses
6088 now carry a message ID.
6090 * We now have a substantially larger unit test suite, but this
6091 continues to be work in progress.
6093 * udevadm hwdb gained a new --root= parameter to change the
6094 root directory to operate relative to.
6096 * logind will now issue a background sync() request to the kernel
6097 early at shutdown, so that dirty buffers are flushed to disk early
6098 instead of at the last moment, in order to optimize shutdown
6101 * A new bootctl tool has been added that is an interface for
6102 certain boot loader operations. This is currently a preview
6103 and is likely to be extended into a small mechanism daemon
6104 like timedated, localed, hostnamed, and can be used by
6105 graphical UIs to enumerate available boot options, and
6106 request boot into firmware operations.
6108 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6109 the rest of the package. It also has been updated to work
6110 correctly in initrds.
6112 * Policykit previously has been runtime optional, and is now
6113 also compile time optional via a configure switch.
6115 * systemd-analyze has been reimplemented in C. Also "systemctl
6116 dot" has moved into systemd-analyze.
6118 * "systemctl status" with no further parameters will now print
6119 the status of all active or failed units.
6121 * Operations such as "systemctl start" can now be executed
6122 with a new mode "--irreversible" which may be used to queue
6123 operations that cannot accidentally be reversed by a later
6124 job queuing. This is by default used to make shutdown
6125 requests more robust.
6127 * The Python API of systemd now gained a new module for
6128 reading journal files.
6130 * A new tool kernel-install has been added that can install
6131 kernel images according to the Boot Loader Specification:
6133 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6135 * Boot time console output has been improved to provide
6136 animated boot time output for hanging jobs.
6138 * A new tool systemd-activate has been added which can be used
6139 to test socket activation with, directly from the command
6140 line. This should make it much easier to test and debug
6141 socket activation in daemons.
6143 * journalctl gained a new "--reverse" (or -r) option to show
6144 journal output in reverse order (i.e. newest line first).
6146 * journalctl gained a new "--pager-end" (or -e) option to jump
6147 to immediately jump to the end of the journal in the
6148 pager. This is only supported in conjunction with "less".
6150 * journalctl gained a new "--user-unit=" option, that works
6151 similarly to "--unit=" but filters for user units rather than
6154 * A number of unit files to ease adoption of systemd in
6155 initrds has been added. This moves some minimal logic from
6156 the various initrd implementations into systemd proper.
6158 * The journal files are now owned by a new group
6159 "systemd-journal", which exists specifically to allow access
6160 to the journal, and nothing else. Previously, we used the
6161 "adm" group for that, which however possibly covers more
6162 than just journal/log file access. This new group is now
6163 already used by systemd-journal-gatewayd to ensure this
6164 daemon gets access to the journal files and as little else
6165 as possible. Note that "make install" will also set FS ACLs
6166 up for /var/log/journal to give "adm" and "wheel" read
6167 access to it, in addition to "systemd-journal" which owns
6168 the journal files. We recommend that packaging scripts also
6169 add read access to "adm" + "wheel" to /var/log/journal, and
6170 all existing/future journal files. To normal users and
6171 administrators little changes, however packagers need to
6172 ensure to create the "systemd-journal" system group at
6173 package installation time.
6175 * The systemd-journal-gatewayd now runs as unprivileged user
6176 systemd-journal-gateway:systemd-journal-gateway. Packaging
6177 scripts need to create these system user/group at
6180 * timedated now exposes a new boolean property CanNTP that
6181 indicates whether a local NTP service is available or not.
6183 * systemd-detect-virt will now also detect xen PVs
6185 * The pstore file system is now mounted by default, if it is
6188 * In addition to the SELinux and IMA policies we will now also
6189 load SMACK policies at early boot.
6191 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6192 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6193 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6194 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6195 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6196 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6197 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6198 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6199 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6200 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6201 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6202 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6203 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6204 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6208 * Timer units now support calendar time events in addition to
6209 monotonic time events. That means you can now trigger a unit
6210 based on a calendar time specification such as "Thu,Fri
6211 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6212 or fifth day of any month of the year 2013, given that it is
6213 a thursday or friday. This brings timer event support
6214 considerably closer to cron's capabilities. For details on
6215 the supported calendar time specification language see
6218 * udev now supports a number of different naming policies for
6219 network interfaces for predictable names, and a combination
6220 of these policies is now the default. Please see this wiki
6221 document for details:
6223 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6225 * Auke Kok's bootchart implementation has been added to the
6226 systemd tree. It is an optional component that can graph the
6227 boot in quite some detail. It is one of the best bootchart
6228 implementations around and minimal in its code and
6231 * nss-myhostname has been integrated into the systemd source
6232 tree. nss-myhostname guarantees that the local hostname
6233 always stays resolvable via NSS. It has been a weak
6234 requirement of systemd-hostnamed since a long time, and
6235 since its code is actually trivial we decided to just
6236 include it in systemd's source tree. It can be turned off
6237 with a configure switch.
6239 * The read-ahead logic is now capable of properly detecting
6240 whether a btrfs file system is on SSD or rotating media, in
6241 order to optimize the read-ahead scheme. Previously, it was
6242 only capable of detecting this on traditional file systems
6245 * In udev, additional device properties are now read from the
6246 IAB in addition to the OUI database. Also, Bluetooth company
6247 identities are attached to the devices as well.
6249 * In service files %U may be used as specifier that is
6250 replaced by the configured user name of the service.
6252 * nspawn may now be invoked without a controlling TTY. This
6253 makes it suitable for invocation as its own service. This
6254 may be used to set up a simple containerized server system
6255 using only core OS tools.
6257 * systemd and nspawn can now accept socket file descriptors
6258 when they are started for socket activation. This enables
6259 implementation of socket activated nspawn
6260 containers. i.e. think about autospawning an entire OS image
6261 when the first SSH or HTTP connection is received. We expect
6262 that similar functionality will also be added to libvirt-lxc
6265 * journalctl will now suppress ANSI color codes when
6266 presenting log data.
6268 * systemctl will no longer show control group information for
6269 a unit if the control group is empty anyway.
6271 * logind can now automatically suspend/hibernate/shutdown the
6274 * /etc/machine-info and hostnamed now also expose the chassis
6275 type of the system. This can be used to determine whether
6276 the local system is a laptop, desktop, handset or
6277 tablet. This information may either be configured by the
6278 user/vendor or is automatically determined from ACPI and DMI
6279 information if possible.
6281 * A number of PolicyKit actions are now bound together with
6282 "imply" rules. This should simplify creating UIs because
6283 many actions will now authenticate similar ones as well.
6285 * Unit files learnt a new condition ConditionACPower= which
6286 may be used to conditionalize a unit depending on whether an
6287 AC power source is connected or not, of whether the system
6288 is running on battery power.
6290 * systemctl gained a new "is-failed" verb that may be used in
6291 shell scripts and suchlike to check whether a specific unit
6292 is in the "failed" state.
6294 * The EnvironmentFile= setting in unit files now supports file
6295 globbing, and can hence be used to easily read a number of
6296 environment files at once.
6298 * systemd will no longer detect and recognize specific
6299 distributions. All distribution-specific #ifdeffery has been
6300 removed, systemd is now fully generic and
6301 distribution-agnostic. Effectively, not too much is lost as
6302 a lot of the code is still accessible via explicit configure
6303 switches. However, support for some distribution specific
6304 legacy configuration file formats has been dropped. We
6305 recommend distributions to simply adopt the configuration
6306 files everybody else uses now and convert the old
6307 configuration from packaging scripts. Most distributions
6308 already did that. If that's not possible or desirable,
6309 distributions are welcome to forward port the specific
6310 pieces of code locally from the git history.
6312 * When logging a message about a unit systemd will now always
6313 log the unit name in the message meta data.
6315 * localectl will now also discover system locale data that is
6316 not stored in locale archives, but directly unpacked.
6318 * logind will no longer unconditionally use framebuffer
6319 devices as seat masters, i.e. as devices that are required
6320 to be existing before a seat is considered preset. Instead,
6321 it will now look for all devices that are tagged as
6322 "seat-master" in udev. By default, framebuffer devices will
6323 be marked as such, but depending on local systems, other
6324 devices might be marked as well. This may be used to
6325 integrate graphics cards using closed source drivers (such
6326 as NVidia ones) more nicely into logind. Note however, that
6327 we recommend using the open source NVidia drivers instead,
6328 and no udev rules for the closed-source drivers will be
6329 shipped from us upstream.
6331 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6332 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6333 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6334 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6335 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6336 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6337 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6338 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6339 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6340 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6341 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6342 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6347 * udev gained support for loading additional device properties
6348 from an indexed database that is keyed by vendor/product IDs
6349 and similar device identifiers. For the beginning this
6350 "hwdb" is populated with data from the well-known PCI and
6351 USB database, but also includes PNP, ACPI and OID data. In
6352 the longer run this indexed database shall grow into
6353 becoming the one central database for non-essential
6354 userspace device metadata. Previously, data from the PCI/USB
6355 database was only attached to select devices, since the
6356 lookup was a relatively expensive operation due to O(n) time
6357 complexity (with n being the number of entries in the
6358 database). Since this is now O(1), we decided to add in this
6359 data for all devices where this is available, by
6360 default. Note that the indexed database needs to be rebuilt
6361 when new data files are installed. To achieve this you need
6362 to update your packaging scripts to invoke "udevadm hwdb
6363 --update" after installation of hwdb data files. For
6364 RPM-based distributions we introduced the new
6365 %udev_hwdb_update macro for this purpose.
6367 * The Journal gained support for the "Message Catalog", an
6368 indexed database to link up additional information with
6369 journal entries. For further details please check:
6371 https://www.freedesktop.org/wiki/Software/systemd/catalog
6373 The indexed message catalog database also needs to be
6374 rebuilt after installation of message catalog files. Use
6375 "journalctl --update-catalog" for this. For RPM-based
6376 distributions we introduced the %journal_catalog_update
6377 macro for this purpose.
6379 * The Python Journal bindings gained support for the standard
6380 Python logging framework.
6382 * The Journal API gained new functions for checking whether
6383 the underlying file system of a journal file is capable of
6384 properly reporting file change notifications, or whether
6385 applications that want to reflect journal changes "live"
6386 need to recheck journal files continuously in appropriate
6389 * It is now possible to set the "age" field for tmpfiles
6390 entries to 0, indicating that files matching this entry
6391 shall always be removed when the directories are cleaned up.
6393 * coredumpctl gained a new "gdb" verb which invokes gdb
6394 right-away on the selected coredump.
6396 * There's now support for "hybrid sleep" on kernels that
6397 support this, in addition to "suspend" and "hibernate". Use
6398 "systemctl hybrid-sleep" to make use of this.
6400 * logind's HandleSuspendKey= setting (and related settings)
6401 now gained support for a new "lock" setting to simply
6402 request the screen lock on all local sessions, instead of
6403 actually executing a suspend or hibernation.
6405 * systemd will now mount the EFI variables file system by
6408 * Socket units now gained support for configuration of the
6409 SMACK security label.
6411 * timedatectl will now output the time of the last and next
6412 daylight saving change.
6414 * We dropped support for various legacy and distro-specific
6415 concepts, such as insserv, early-boot SysV services
6416 (i.e. those for non-standard runlevels such as 'b' or 'S')
6417 or ArchLinux /etc/rc.conf support. We recommend the
6418 distributions who still need support this to either continue
6419 to maintain the necessary patches downstream, or find a
6420 different solution. (Talk to us if you have questions!)
6422 * Various systemd components will now bypass PolicyKit checks
6423 for root and otherwise handle properly if PolicyKit is not
6424 found to be around. This should fix most issues for
6425 PolicyKit-less systems. Quite frankly this should have been
6426 this way since day one. It is absolutely our intention to
6427 make systemd work fine on PolicyKit-less systems, and we
6428 consider it a bug if something does not work as it should if
6429 PolicyKit is not around.
6431 * For embedded systems it is now possible to build udev and
6432 systemd without blkid and/or kmod support.
6434 * "systemctl switch-root" is now capable of switching root
6435 more than once. I.e. in addition to transitions from the
6436 initrd to the host OS it is now possible to transition to
6437 further OS images from the host. This is useful to implement
6438 offline updating tools.
6440 * Various other additions have been made to the RPM macros
6441 shipped with systemd. Use %udev_rules_update() after
6442 installing new udev rules files. %_udevhwdbdir,
6443 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6444 %_sysctldir are now available which resolve to the right
6445 directories for packages to place various data files in.
6447 * journalctl gained the new --full switch (in addition to
6448 --all, to disable ellipsation for long messages.
6450 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6451 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6452 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6453 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6454 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6455 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6456 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6457 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6458 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6462 * journalctl gained new --since= and --until= switches to
6463 filter by time. It also now supports nice filtering for
6464 units via --unit=/-u.
6466 * Type=oneshot services may use ExecReload= and do the
6469 * The journal daemon now supports time-based rotation and
6470 vacuuming, in addition to the usual disk-space based
6473 * The journal will now index the available field values for
6474 each field name. This enables clients to show pretty drop
6475 downs of available match values when filtering. The bash
6476 completion of journalctl has been updated
6477 accordingly. journalctl gained a new switch -F to list all
6478 values a certain field takes in the journal database.
6480 * More service events are now written as structured messages
6481 to the journal, and made recognizable via message IDs.
6483 * The timedated, localed and hostnamed mini-services which
6484 previously only provided support for changing time, locale
6485 and hostname settings from graphical DEs such as GNOME now
6486 also have a minimal (but very useful) text-based client
6487 utility each. This is probably the nicest way to changing
6488 these settings from the command line now, especially since
6489 it lists available options and is fully integrated with bash
6492 * There's now a new tool "systemd-coredumpctl" to list and
6493 extract coredumps from the journal.
6495 * We now install a README each in /var/log/ and
6496 /etc/rc.d/init.d explaining where the system logs and init
6497 scripts went. This hopefully should help folks who go to
6498 that dirs and look into the otherwise now empty void and
6499 scratch their heads.
6501 * When user-services are invoked (by systemd --user) the
6502 $MANAGERPID env var is set to the PID of systemd.
6504 * SIGRTMIN+24 when sent to a --user instance will now result
6505 in immediate termination of systemd.
6507 * gatewayd received numerous feature additions such as a
6508 "follow" mode, for live syncing and filtering.
6510 * browse.html now allows filtering and showing detailed
6511 information on specific entries. Keyboard navigation and
6512 mouse screen support has been added.
6514 * gatewayd/journalctl now supports HTML5/JSON
6515 Server-Sent-Events as output.
6517 * The SysV init script compatibility logic will now
6518 heuristically determine whether a script supports the
6519 "reload" verb, and only then make this available as
6522 * "systemctl status --follow" has been removed, use "journalctl
6525 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6526 have been removed since they are hardly useful to be
6529 * And I'd like to take the opportunity to specifically mention
6530 Zbigniew for his great contributions. Zbigniew, you rock!
6532 Contributions from: Andrew Eikum, Christian Hesse, Colin
6533 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6534 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6535 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6536 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6537 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6538 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6542 * If /etc/vconsole.conf is non-existent or empty we will no
6543 longer load any console font or key map at boot by
6544 default. Instead the kernel defaults will be left
6545 intact. This is definitely the right thing to do, as no
6546 configuration should mean no configuration, and hard-coding
6547 font names that are different on all archs is probably a bad
6548 idea. Also, the kernel default key map and font should be
6549 good enough for most cases anyway, and mostly identical to
6550 the userspace fonts/key maps we previously overloaded them
6551 with. If distributions want to continue to default to a
6552 non-kernel font or key map they should ship a default
6553 /etc/vconsole.conf with the appropriate contents.
6555 Contributions from: Colin Walters, Daniel J Walsh, Dave
6556 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6557 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6561 * journalctl gained a new --cursor= switch to show entries
6562 starting from the specified location in the journal.
6564 * We now enforce a size limit on journal entry fields exported
6565 with "-o json" in journalctl. Fields larger than 4K will be
6566 assigned null. This can be turned off with --all.
6568 * An (optional) journal gateway daemon is now available as
6569 "systemd-journal-gatewayd.service". This service provides
6570 access to the journal via HTTP and JSON. This functionality
6571 will be used to implement live log synchronization in both
6572 pull and push modes, but has various other users too, such
6573 as easy log access for debugging of embedded devices. Right
6574 now it is already useful to retrieve the journal via HTTP:
6576 # systemctl start systemd-journal-gatewayd.service
6577 # wget http://localhost:19531/entries
6579 This will download the journal contents in a
6580 /var/log/messages compatible format. The same as JSON:
6582 # curl -H"Accept: application/json" http://localhost:19531/entries
6584 This service is also accessible via a web browser where a
6585 single static HTML5 app is served that uses the JSON logic
6586 to enable the user to do some basic browsing of the
6587 journal. This will be extended later on. Here's an example
6588 screenshot of this app in its current state:
6590 http://0pointer.de/public/journal-gatewayd
6592 Contributions from: Kay Sievers, Lennart Poettering, Robert
6593 Milasan, Tom Gundersen
6597 * The bash completion logic is now available for journalctl
6600 * We do not mount the "cpuset" controller anymore together with
6601 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6602 started if no parameters are assigned to it. "cpuset" hence
6603 broke code that assumed it could create "cpu" groups and
6606 * journalctl -f will now subscribe to terminal size changes,
6607 and line break accordingly.
6609 Contributions from: Dave Reisner, Kay Sievers, Lennart
6610 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6614 * nspawn will now create a symlink /etc/localtime in the
6615 container environment, copying the host's timezone
6616 setting. Previously this has been done via a bind mount, but
6617 since symlinks cannot be bind mounted this has now been
6618 changed to create/update the appropriate symlink.
6620 * journalctl -n's line number argument is now optional, and
6621 will default to 10 if omitted.
6623 * journald will now log the maximum size the journal files may
6624 take up on disk. This is particularly useful if the default
6625 built-in logic of determining this parameter from the file
6626 system size is used. Use "systemctl status
6627 systemd-journald.service" to see this information.
6629 * The multi-seat X wrapper tool has been stripped down. As X
6630 is now capable of enumerating graphics devices via udev in a
6631 seat-aware way the wrapper is not strictly necessary
6632 anymore. A stripped down temporary stop-gap is still shipped
6633 until the upstream display managers have been updated to
6634 fully support the new X logic. Expect this wrapper to be
6635 removed entirely in one of the next releases.
6637 * HandleSleepKey= in logind.conf has been split up into
6638 HandleSuspendKey= and HandleHibernateKey=. The old setting
6639 is not available anymore. X11 and the kernel are
6640 distinguishing between these keys and we should too. This
6641 also means the inhibition lock for these keys has been split
6644 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6645 Poettering, Lukas Nykryn, Václav Pavlín
6649 * Whenever a unit changes state we will now log this to the
6650 journal and show along the unit's own log output in
6653 * ConditionPathIsMountPoint= can now properly detect bind
6654 mount points too. (Previously, a bind mount of one file
6655 system to another place in the same file system could not be
6656 detected as mount, since they shared struct stat's st_dev
6659 * We will now mount the cgroup controllers cpu, cpuacct,
6660 cpuset and the controllers net_cls, net_prio together by
6663 * nspawn containers will now have a virtualized boot
6664 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6665 over with a randomized ID at container initialization). This
6666 has the effect of making "journalctl -b" do the right thing
6669 * The JSON output journal serialization has been updated not
6670 to generate "endless" list objects anymore, but rather one
6671 JSON object per line. This is more in line how most JSON
6672 parsers expect JSON objects. The new output mode
6673 "json-pretty" has been added to provide similar output, but
6674 neatly aligned for readability by humans.
6676 * We dropped all explicit sync() invocations in the shutdown
6677 code. The kernel does this implicitly anyway in the kernel
6678 reboot() syscall. halt(8)'s -n option is now a compatibility
6681 * We now support virtualized reboot() in containers, as
6682 supported by newer kernels. We will fall back to exit() if
6683 CAP_SYS_REBOOT is not available to the container. Also,
6684 nspawn makes use of this now and will actually reboot the
6685 container if the containerized OS asks for that.
6687 * journalctl will only show local log output by default
6688 now. Use --merge (-m) to show remote log output, too.
6690 * libsystemd-journal gained the new sd_journal_get_usage()
6691 call to determine the current disk usage of all journal
6692 files. This is exposed in the new "journalctl --disk-usage"
6695 * journald gained a new configuration setting SplitMode= in
6696 journald.conf which may be used to control how user journals
6697 are split off. See journald.conf(5) for details.
6699 * A new condition type ConditionFileNotEmpty= has been added.
6701 * tmpfiles' "w" lines now support file globbing, to write
6702 multiple files at once.
6704 * We added Python bindings for the journal submission
6705 APIs. More Python APIs for a number of selected APIs will
6706 likely follow. Note that we intend to add native bindings
6707 only for the Python language, as we consider it common
6708 enough to deserve bindings shipped within systemd. There are
6709 various projects outside of systemd that provide bindings
6710 for languages such as PHP or Lua.
6712 * Many conditions will now resolve specifiers such as %i. In
6713 addition, PathChanged= and related directives of .path units
6714 now support specifiers as well.
6716 * There's now a new RPM macro definition for the system preset
6719 * journald will now warn if it ca not forward a message to the
6720 syslog daemon because its socket is full.
6722 * timedated will no longer write or process /etc/timezone,
6723 except on Debian. As we do not support late mounted /usr
6724 anymore /etc/localtime always being a symlink is now safe,
6725 and hence the information in /etc/timezone is not necessary
6728 * logind will now always reserve one VT for a text getty (VT6
6729 by default). Previously if more than 6 X sessions where
6730 started they took up all the VTs with auto-spawned gettys,
6731 so that no text gettys were available anymore.
6733 * udev will now automatically inform the btrfs kernel logic
6734 about btrfs RAID components showing up. This should make
6735 simple hotplug based btrfs RAID assembly work.
6737 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6738 (but not for its children which will stay at the kernel
6739 default). This should allow setups with a lot more listening
6742 * systemd will now always pass the configured timezone to the
6743 kernel at boot. timedated will do the same when the timezone
6746 * logind's inhibition logic has been updated. By default,
6747 logind will now handle the lid switch, the power and sleep
6748 keys all the time, even in graphical sessions. If DEs want
6749 to handle these events on their own they should take the new
6750 handle-power-key, handle-sleep-key and handle-lid-switch
6751 inhibitors during their runtime. A simple way to achieve
6752 that is to invoke the DE wrapped in an invocation of:
6754 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6756 * Access to unit operations is now checked via SELinux taking
6757 the unit file label and client process label into account.
6759 * systemd will now notify the administrator in the journal
6760 when he over-mounts a non-empty directory.
6762 * There are new specifiers that are resolved in unit files,
6763 for the host name (%H), the machine ID (%m) and the boot ID
6766 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6767 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6768 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6769 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6770 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6771 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6772 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6776 * Support for reading structured kernel messages from
6777 /dev/kmsg has now been added and is enabled by default.
6779 * Support for reading kernel messages from /proc/kmsg has now
6780 been removed. If you want kernel messages in the journal
6781 make sure to run a recent kernel (>= 3.5) that supports
6782 reading structured messages from /dev/kmsg (see
6783 above). /proc/kmsg is now exclusive property of classic
6784 syslog daemons again.
6786 * The libudev API gained the new
6787 udev_device_new_from_device_id() call.
6789 * The logic for file system namespace (ReadOnlyDirectory=,
6790 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6791 require pivot_root() anymore. This means fewer temporary
6792 directories are created below /tmp for this feature.
6794 * nspawn containers will now see and receive all submounts
6795 made on the host OS below the root file system of the
6798 * Forward Secure Sealing is now supported for Journal files,
6799 which provide cryptographical sealing of journal files so
6800 that attackers cannot alter log history anymore without this
6801 being detectable. Lennart will soon post a blog story about
6802 this explaining it in more detail.
6804 * There are two new service settings RestartPreventExitStatus=
6805 and SuccessExitStatus= which allow configuration of exit
6806 status (exit code or signal) which will be excepted from the
6807 restart logic, resp. consider successful.
6809 * journalctl gained the new --verify switch that can be used
6810 to check the integrity of the structure of journal files and
6811 (if Forward Secure Sealing is enabled) the contents of
6814 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6815 and similar symlinks pre-created. This makes running shells
6816 as container init process a lot more fun.
6818 * The fstab support can now handle PARTUUID= and PARTLABEL=
6821 * A new ConditionHost= condition has been added to match
6822 against the hostname (with globs) and machine ID. This is
6823 useful for clusters where a single OS image is used to
6824 provision a large number of hosts which shall run slightly
6825 different sets of services.
6827 * Services which hit the restart limit will now be placed in a
6830 Contributions from: Bertram Poettering, Dave Reisner, Huang
6831 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6832 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6836 * When running in --user mode systemd will now become a
6837 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6838 tree a lot more organized.
6840 * A new PartOf= unit dependency type has been introduced that
6841 may be used to group services in a natural way.
6843 * "systemctl enable" may now be used to enable instances of
6846 * journalctl now prints error log levels in red, and
6847 warning/notice log levels in bright white. It also supports
6848 filtering by log level now.
6850 * cgtop gained a new -n switch (similar to top), to configure
6851 the maximum number of iterations to run for. It also gained
6852 -b, to run in batch mode (accepting no input).
6854 * The suffix ".service" may now be omitted on most systemctl
6855 command lines involving service unit names.
6857 * There's a new bus call in logind to lock all sessions, as
6858 well as a loginctl verb for it "lock-sessions".
6860 * libsystemd-logind.so gained a new call sd_journal_perror()
6861 that works similar to libc perror() but logs to the journal
6862 and encodes structured information about the error number.
6864 * /etc/crypttab entries now understand the new keyfile-size=
6867 * shutdown(8) now can send a (configurable) wall message when
6868 a shutdown is cancelled.
6870 * The mount propagation mode for the root file system will now
6871 default to "shared", which is useful to make containers work
6872 nicely out-of-the-box so that they receive new mounts from
6873 the host. This can be undone locally by running "mount
6874 --make-rprivate /" if needed.
6876 * The prefdm.service file has been removed. Distributions
6877 should maintain this unit downstream if they intend to keep
6878 it around. However, we recommend writing normal unit files
6879 for display managers instead.
6881 * Since systemd is a crucial part of the OS we will now
6882 default to a number of compiler switches that improve
6883 security (hardening) such as read-only relocations, stack
6884 protection, and suchlike.
6886 * The TimeoutSec= setting for services is now split into
6887 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6888 of individual time outs for the start and the stop phase of
6891 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6892 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6893 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6894 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6895 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6896 Gundersen, Zbigniew Jędrzejewski-Szmek
6900 * The journal and id128 C APIs are now fully documented as man
6903 * Extra safety checks have been added when transitioning from
6904 the initial RAM disk to the main system to avoid accidental
6907 * /etc/crypttab entries now understand the new keyfile-offset=
6910 * systemctl -t can now be used to filter by unit load state.
6912 * The journal C API gained the new sd_journal_wait() call to
6913 make writing synchronous journal clients easier.
6915 * journalctl gained the new -D switch to show journals from a
6918 * journalctl now displays a special marker between log
6919 messages of two different boots.
6921 * The journal is now explicitly flushed to /var via a service
6922 systemd-journal-flush.service, rather than implicitly simply
6923 by seeing /var/log/journal to be writable.
6925 * journalctl (and the journal C APIs) can now match for much
6926 more complex expressions, with alternatives and
6929 * When transitioning from the initial RAM disk to the main
6930 system we will now kill all processes in a killing spree to
6931 ensure no processes stay around by accident.
6933 * Three new specifiers may be used in unit files: %u, %h, %s
6934 resolve to the user name, user home directory resp. user
6935 shell. This is useful for running systemd user instances.
6937 * We now automatically rotate journal files if their data
6938 object hash table gets a fill level > 75%. We also size the
6939 hash table based on the configured maximum file size. This
6940 together should lower hash collisions drastically and thus
6941 speed things up a bit.
6943 * journalctl gained the new "--header" switch to introspect
6944 header data of journal files.
6946 * A new setting SystemCallFilters= has been added to services
6947 which may be used to apply blacklists or whitelists to
6948 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6950 * nspawn gained a new --link-journal= switch (and quicker: -j)
6951 to link the container journal with the host. This makes it
6952 very easy to centralize log viewing on the host for all
6953 guests while still keeping the journal files separated.
6955 * Many bugfixes and optimizations
6957 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6958 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6959 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6964 * Several tools now understand kernel command line arguments,
6965 which are only read when run in an initial RAM disk. They
6966 usually follow closely their normal counterparts, but are
6969 * There's a new tool to analyze the readahead files that are
6970 automatically generated at boot. Use:
6972 /usr/lib/systemd/systemd-readahead analyze /.readahead
6974 * We now provide an early debug shell on tty9 if this enabled. Use:
6976 systemctl enable debug-shell.service
6978 * All plymouth related units have been moved into the Plymouth
6979 package. Please make sure to upgrade your Plymouth version
6982 * systemd-tmpfiles now supports getting passed the basename of
6983 a configuration file only, in which case it will look for it
6984 in all appropriate directories automatically.
6986 * udevadm info now takes a /dev or /sys path as argument, and
6987 does the right thing. Example:
6989 udevadm info /dev/sda
6990 udevadm info /sys/class/block/sda
6992 * systemctl now prints a warning if a unit is stopped but a
6993 unit that might trigger it continues to run. Example: a
6994 service is stopped but the socket that activates it is left
6997 * "systemctl status" will now mention if the log output was
6998 shortened due to rotation since a service has been started.
7000 * The journal API now exposes functions to determine the
7001 "cutoff" times due to rotation.
7003 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7004 immediately flushing of runtime logs to /var if possible,
7005 resp. for triggering immediate rotation of the journal
7008 * It is now considered an error if a service is attempted to
7009 be stopped that is not loaded.
7011 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7013 * systemd-analyze now supports Python 3
7015 * tmpfiles now supports cleaning up directories via aging
7016 where the first level dirs are always kept around but
7017 directories beneath it automatically aged. This is enabled
7018 by prefixing the age field with '~'.
7020 * Seat objects now expose CanGraphical, CanTTY properties
7021 which is required to deal with very fast bootups where the
7022 display manager might be running before the graphics drivers
7023 completed initialization.
7025 * Seat objects now expose a State property.
7027 * We now include RPM macros for service enabling/disabling
7028 based on the preset logic. We recommend RPM based
7029 distributions to make use of these macros if possible. This
7030 makes it simpler to reuse RPM spec files across
7033 * We now make sure that the collected systemd unit name is
7034 always valid when services log to the journal via
7037 * There's a new man page kernel-command-line(7) detailing all
7038 command line options we understand.
7040 * The fstab generator may now be disabled at boot by passing
7041 fstab=0 on the kernel command line.
7043 * A new kernel command line option modules-load= is now understood
7044 to load a specific kernel module statically, early at boot.
7046 * Unit names specified on the systemctl command line are now
7047 automatically escaped as needed. Also, if file system or
7048 device paths are specified they are automatically turned
7049 into the appropriate mount or device unit names. Example:
7051 systemctl status /home
7052 systemctl status /dev/sda
7054 * The SysVConsole= configuration option has been removed from
7055 system.conf parsing.
7057 * The SysV search path is no longer exported on the D-Bus
7060 * The Names= option has been removed from unit file parsing.
7062 * There's a new man page bootup(7) detailing the boot process.
7064 * Every unit and every generator we ship with systemd now
7065 comes with full documentation. The self-explanatory boot is
7068 * A couple of services gained "systemd-" prefixes in their
7069 name if they wrap systemd code, rather than only external
7070 code. Among them fsck@.service which is now
7071 systemd-fsck@.service.
7073 * The HaveWatchdog property has been removed from the D-Bus
7076 * systemd.confirm_spawn= on the kernel command line should now
7079 * There's a new man page crypttab(5) which details all options
7080 we actually understand.
7082 * systemd-nspawn gained a new --capability= switch to pass
7083 additional capabilities to the container.
7085 * timedated will now read known NTP implementation unit names
7086 from /usr/lib/systemd/ntp-units.d/*.list,
7087 systemd-timedated-ntp.target has been removed.
7089 * journalctl gained a new switch "-b" that lists log data of
7090 the current boot only.
7092 * The notify socket is in the abstract namespace again, in
7093 order to support daemons which chroot() at start-up.
7095 * There is a new Storage= configuration option for journald
7096 which allows configuration of where log data should go. This
7097 also provides a way to disable journal logging entirely, so
7098 that data collected is only forwarded to the console, the
7099 kernel log buffer or another syslog implementation.
7101 * Many bugfixes and optimizations
7103 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7104 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7105 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7106 Shawn Landden, Tom Gundersen
7110 * "systemctl help <unit>" now shows the man page if one is
7113 * Several new man pages have been added.
7115 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7116 MaxLevelConsole= can now be specified in
7117 journald.conf. These options allow reducing the amount of
7118 data stored on disk or forwarded by the log level.
7120 * TimerSlackNSec= can now be specified in system.conf for
7121 PID1. This allows system-wide power savings.
7123 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7124 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7129 * logind is now capable of (optionally) handling power and
7130 sleep keys as well as the lid switch.
7132 * journalctl now understands the syntax "journalctl
7133 /usr/bin/avahi-daemon" to get all log output of a specific
7136 * CapabilityBoundingSet= in system.conf now also influences
7137 the capability bound set of usermode helpers of the kernel.
7139 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7140 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7141 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7142 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7146 * Note that we skipped 139 releases here in order to set the
7147 new version to something that is greater than both udev's
7148 and systemd's most recent version number.
7150 * udev: all udev sources are merged into the systemd source tree now.
7151 All future udev development will happen in the systemd tree. It
7152 is still fully supported to use the udev daemon and tools without
7153 systemd running, like in initramfs or other init systems. Building
7154 udev though, will require the *build* of the systemd tree, but
7155 udev can be properly *run* without systemd.
7157 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7158 should be used to create dead device nodes as workarounds for broken
7161 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7162 no longer supported. udev_monitor_new_from_netlink() needs to be
7163 used to subscribe to events.
7165 * udev: when udevd is started by systemd, processes which are left
7166 behind by forking them off of udev rules, are unconditionally cleaned
7167 up and killed now after the event handling has finished. Services or
7168 daemons must be started as systemd services. Services can be
7169 pulled-in by udev to get started, but they can no longer be directly
7170 forked by udev rules.
7172 * udev: the daemon binary is called systemd-udevd now and installed
7173 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7174 to adapt to that, create symlink, or rename the binary after building
7177 * libudev no longer provides these symbols:
7178 udev_monitor_from_socket()
7179 udev_queue_get_failed_list_entry()
7180 udev_get_{dev,sys,run}_path()
7181 The versions number was bumped and symbol versioning introduced.
7183 * systemd-loginctl and systemd-journalctl have been renamed
7184 to loginctl and journalctl to match systemctl.
7186 * The config files: /etc/systemd/systemd-logind.conf and
7187 /etc/systemd/systemd-journald.conf have been renamed to
7188 logind.conf and journald.conf. Package updates should rename
7189 the files to the new names on upgrade.
7191 * For almost all files the license is now LGPL2.1+, changed
7192 from the previous GPL2.0+. Exceptions are some minor stuff
7193 of udev (which will be changed to LGPL2.1 eventually, too),
7194 and the MIT licensed sd-daemon.[ch] library that is suitable
7195 to be used as drop-in files.
7197 * systemd and logind now handle system sleep states, in
7198 particular suspending and hibernating.
7200 * logind now implements a sleep/shutdown/idle inhibiting logic
7201 suitable for a variety of uses. Soonishly Lennart will blog
7202 about this in more detail.
7204 * var-run.mount and var-lock.mount are no longer provided
7205 (which previously bind mounted these directories to their new
7206 places). Distributions which have not converted these
7207 directories to symlinks should consider stealing these files
7208 from git history and add them downstream.
7210 * We introduced the Documentation= field for units and added
7211 this to all our shipped units. This is useful to make it
7212 easier to explore the boot and the purpose of the various
7215 * All smaller setup units (such as
7216 systemd-vconsole-setup.service) now detect properly if they
7217 are run in a container and are skipped when
7218 appropriate. This guarantees an entirely noise-free boot in
7219 Linux container environments such as systemd-nspawn.
7221 * A framework for implementing offline system updates is now
7222 integrated, for details see:
7223 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7225 * A new service type Type=idle is available now which helps us
7226 avoiding ugly interleaving of getty output and boot status
7229 * There's now a system-wide CapabilityBoundingSet= option to
7230 globally reduce the set of capabilities for the
7231 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7232 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7233 even CAP_NET_ADMIN system-wide for secure systems.
7235 * There are now system-wide DefaultLimitXXX= options to
7236 globally change the defaults of the various resource limits
7237 for all units started by PID 1.
7239 * Harald Hoyer's systemd test suite has been integrated into
7240 systemd which allows easy testing of systemd builds in qemu
7241 and nspawn. (This is really awesome! Ask us for details!)
7243 * The fstab parser is now implemented as generator, not inside
7246 * systemctl will now warn you if .mount units generated from
7247 /etc/fstab are out of date due to changes in fstab that
7248 have not been read by systemd yet.
7250 * systemd is now suitable for usage in initrds. Dracut has
7251 already been updated to make use of this. With this in place
7252 initrds get a slight bit faster but primarily are much
7253 easier to introspect and debug since "systemctl status" in
7254 the host system can be used to introspect initrd services,
7255 and the journal from the initrd is kept around too.
7257 * systemd-delta has been added, a tool to explore differences
7258 between user/admin configuration and vendor defaults.
7260 * PrivateTmp= now affects both /tmp and /var/tmp.
7262 * Boot time status messages are now much prettier and feature
7263 proper english language. Booting up systemd has never been
7266 * Read-ahead pack files now include the inode number of all
7267 files to pre-cache. When the inode changes the pre-caching
7268 is not attempted. This should be nicer to deal with updated
7269 packages which might result in changes of read-ahead
7272 * We now temporaritly lower the kernel's read_ahead_kb variable
7273 when collecting read-ahead data to ensure the kernel's
7274 built-in read-ahead does not add noise to our measurements
7275 of necessary blocks to pre-cache.
7277 * There's now RequiresMountsFor= to add automatic dependencies
7278 for all mounts necessary for a specific file system path.
7280 * MountAuto= and SwapAuto= have been removed from
7281 system.conf. Mounting file systems at boot has to take place
7284 * nspawn now learned a new switch --uuid= to set the machine
7285 ID on the command line.
7287 * nspawn now learned the -b switch to automatically search
7290 * vt102 is now the default TERM for serial TTYs, upgraded from
7293 * systemd-logind now works on VT-less systems.
7295 * The build tree has been reorganized. The individual
7296 components now have directories of their own.
7298 * A new condition type ConditionPathIsReadWrite= is now available.
7300 * nspawn learned the new -C switch to create cgroups for the
7301 container in other hierarchies.
7303 * We now have support for hardware watchdogs, configurable in
7306 * The scheduled shutdown logic now has a public API.
7308 * We now mount /tmp as tmpfs by default, but this can be
7309 masked and /etc/fstab can override it.
7311 * Since udisks does not make use of /media anymore we are not
7312 mounting a tmpfs on it anymore.
7314 * journalctl gained a new --local switch to only interleave
7315 locally generated journal files.
7317 * We can now load the IMA policy at boot automatically.
7319 * The GTK tools have been split off into a systemd-ui.
7321 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7322 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7323 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7324 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7325 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7326 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7327 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7328 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7329 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7334 * This is mostly a bugfix release
7336 * Support optional initialization of the machine ID from the
7337 KVM or container configured UUID.
7339 * Support immediate reboots with "systemctl reboot -ff"
7341 * Show /etc/os-release data in systemd-analyze output
7343 * Many bugfixes for the journal, including endianness fixes and
7344 ensuring that disk space enforcement works
7346 * sd-login.h is C++ compatible again
7348 * Extend the /etc/os-release format on request of the Debian
7351 * We now refuse non-UTF8 strings used in various configuration
7352 and unit files. This is done to ensure we do not pass invalid
7353 data over D-Bus or expose it elsewhere.
7355 * Register Mimo USB Screens as suitable for automatic seat
7358 * Read SELinux client context from journal clients in a race
7361 * Reorder configuration file lookup order. /etc now always
7362 overrides /run in order to allow the administrator to always
7363 and unconditionally override vendor-supplied or
7364 automatically generated data.
7366 * The various user visible bits of the journal now have man
7367 pages. We still lack man pages for the journal API calls
7370 * We now ship all man pages in HTML format again in the
7373 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7374 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7375 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7380 * This is mostly a bugfix release
7382 * systems lacking /etc/os-release are no longer supported.
7384 * Various functionality updates to libsystemd-login.so
7386 * Track class of PAM logins to distinguish greeters from
7389 Contributions from: Kay Sievers, Lennart Poettering, Michael
7394 * This is an important bugfix release for v41.
7396 * Building man pages is now optional which should be useful
7397 for those building systemd from git but unwilling to install
7400 * Watchdog support for supervising services is now usable. In
7401 a future release support for hardware watchdogs
7402 (i.e. /dev/watchdog) will be added building on this.
7404 * Service start rate limiting is now configurable and can be
7405 turned off per service. When a start rate limit is hit a
7406 reboot can automatically be triggered.
7408 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7410 Contributions from: Benjamin Franzke, Bill Nottingham,
7411 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7412 Schmidt, Michał Górny, Piotr Drąg
7416 * The systemd binary is installed /usr/lib/systemd/systemd now;
7417 An existing /sbin/init symlink needs to be adapted with the
7420 * The code that loads kernel modules has been ported to invoke
7421 libkmod directly, instead of modprobe. This means we do not
7422 support systems with module-init-tools anymore.
7424 * Watchdog support is now already useful, but still not
7427 * A new kernel command line option systemd.setenv= is
7428 understood to set system wide environment variables
7429 dynamically at boot.
7431 * We now limit the set of capabilities of systemd-journald.
7433 * We now set SIGPIPE to ignore by default, since it only is
7434 useful in shell pipelines, and has little use in general
7435 code. This can be disabled with IgnoreSIPIPE=no in unit
7438 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7439 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7444 * This is mostly a bugfix release
7446 * We now expose the reason why a service failed in the
7447 "Result" D-Bus property.
7449 * Rudimentary service watchdog support (will be completed over
7450 the next few releases.)
7452 * When systemd forks off in order execute some service we will
7453 now immediately changes its argv[0] to reflect which process
7454 it will execute. This is useful to minimize the time window
7455 with a generic argv[0], which makes bootcharts more useful
7457 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7458 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7459 Mike Kazantsev, Ray Strode
7463 * This is mostly a test release, but incorporates many
7466 * New systemd-cgtop tool to show control groups by their
7469 * Linking against libacl for ACLs is optional again. If
7470 disabled, support tracking device access for active logins
7471 goes becomes unavailable, and so does access to the user
7472 journals by the respective users.
7474 * If a group "adm" exists, journal files are automatically
7475 owned by them, thus allow members of this group full access
7476 to the system journal as well as all user journals.
7478 * The journal now stores the SELinux context of the logging
7479 client for all entries.
7481 * Add C++ inclusion guards to all public headers
7483 * New output mode "cat" in the journal to print only text
7484 messages, without any meta data like date or time.
7486 * Include tiny X server wrapper as a temporary stop-gap to
7487 teach XOrg udev display enumeration. This is used by display
7488 managers such as gdm, and will go away as soon as XOrg
7489 learned native udev hotplugging for display devices.
7491 * Add new systemd-cat tool for executing arbitrary programs
7492 with STDERR/STDOUT connected to the journal. Can also act as
7493 BSD logger replacement, and does so by default.
7495 * Optionally store all locally generated coredumps in the
7496 journal along with meta data.
7498 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7499 writing short strings to files (for usage for /sys), and for
7500 creating symlinks, character and block device nodes.
7502 * New unit file option ControlGroupPersistent= to make cgroups
7503 persistent, following the mechanisms outlined in
7504 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7506 * Support multiple local RTCs in a sane way
7508 * No longer monopolize IO when replaying readahead data on
7509 rotating disks, since we might starve non-file-system IO to
7510 death, since fanotify() will not see accesses done by blkid,
7513 * Do not show kernel threads in systemd-cgls anymore, unless
7514 requested with new -k switch.
7516 Contributions from: Dan Horák, Kay Sievers, Lennart
7517 Poettering, Michal Schmidt
7521 * This is mostly a test release, but incorporates many
7524 * The git repository moved to:
7525 git://anongit.freedesktop.org/systemd/systemd
7526 ssh://git.freedesktop.org/git/systemd/systemd
7528 * First release with the journal
7529 http://0pointer.de/blog/projects/the-journal.html
7531 * The journal replaces both systemd-kmsg-syslogd and
7532 systemd-stdout-bridge.
7534 * New sd_pid_get_unit() API call in libsystemd-logind
7536 * Many systemadm clean-ups
7538 * Introduce remote-fs-pre.target which is ordered before all
7539 remote mounts and may be used to start services before all
7542 * Added Mageia support
7544 * Add bash completion for systemd-loginctl
7546 * Actively monitor PID file creation for daemons which exit in
7547 the parent process before having finished writing the PID
7548 file in the daemon process. Daemons which do this need to be
7549 fixed (i.e. PID file creation must have finished before the
7550 parent exits), but we now react a bit more gracefully to them.
7552 * Add colourful boot output, mimicking the well-known output
7553 of existing distributions.
7555 * New option PassCredentials= for socket units, for
7556 compatibility with a recent kernel ABI breakage.
7558 * /etc/rc.local is now hooked in via a generator binary, and
7559 thus will no longer act as synchronization point during
7562 * systemctl list-unit-files now supports --root=.
7564 * systemd-tmpfiles now understands two new commands: z, Z for
7565 relabelling files according to the SELinux database. This is
7566 useful to apply SELinux labels to specific files in /sys,
7569 * Output of SysV services is now forwarded to both the console
7570 and the journal by default, not only just the console.
7572 * New man pages for all APIs from libsystemd-login.
7574 * The build tree got reorganized and the build system is a
7575 lot more modular allowing embedded setups to specifically
7576 select the components of systemd they are interested in.
7578 * Support for Linux systems lacking the kernel VT subsystem is
7581 * configure's --with-rootdir= got renamed to
7582 --with-rootprefix= to follow the naming used by udev and
7585 * Unless specified otherwise we will now install to /usr instead
7586 of /usr/local by default.
7588 * Processes with '@' in argv[0][0] are now excluded from the
7589 final shut-down killing spree, following the logic explained
7591 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7593 * All processes remaining in a service cgroup when we enter
7594 the START or START_PRE states are now killed with
7595 SIGKILL. That means it is no longer possible to spawn
7596 background processes from ExecStart= lines (which was never
7597 supported anyway, and bad style).
7599 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7600 reloading of units together.
7602 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7603 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7604 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7605 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7606 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek