1 systemd System and Service Manager
5 * modprobe.d drop-in is now shipped by default that sets bonding module
6 option max_bonds=0. This overrides the kernel default, to avoid
7 conflicts and ambigiuty as to whether or not bond0 should be managed
8 by networkd or not. This resolves multiple bugs of bond0 properties
9 not being applied, when bond0 is configured with
10 networkd. Distributors may choose to not package this, however in
11 that case users will be prevented from correctly managing bond0
12 interface using networkd.
16 * Meson is now supported as build system in addition to Automake. It is
17 our plan to remove Automake in one of our next releases, so that
18 Meson becomes our exclusive build system. Hence, please start using
19 the Meson build system in your downstream packaging. There's plenty
20 of documentation around how to use Meson, the extremely brief
23 ./autogen.sh && ./configure && make && sudo make install
27 meson build && ninja -C build && sudo ninja -C build install
29 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
30 which permits configuring a timeout on the time a job is
31 running. This is particularly useful for setting timeouts on jobs for
34 * Unit files gained two new options ConditionUser= and ConditionGroup=
35 for conditionalizing units based on the identity of the user/group
36 running a systemd user instance.
38 * systemd-networkd now understands a new FlowLabel= setting in the
39 [VXLAN] section of .network files, as well as a Priority= in
40 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
41 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
42 gained support for configuration of GENEVE links, and IPv6 address
43 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
45 * .link files now understand a new Port= setting.
47 * systemd-networkd's DHCP support gained support for DHCP option 119
50 * systemd-networkd gained support for serving IPv6 address ranges using
51 the Router Advertisment protocol. The new .network configuration
52 section [IPv6Prefix] may be used to configure the ranges to
53 serve. This is implemented based on a new, minimal, native server
56 * journalctl's --output= switch gained support for a new parameter
57 "short-iso-precise" for a mode where timestamps are shown as precise
60 * systemd-udevd's "net_id" builtin may now generate stable network
61 interface names from IBM PowerVM VIO devices as well as ACPI platform
64 * MulticastDNS support in systemd-resolved may now be explicitly
65 enabled/disabled using the new MulticastDNS= configuration file
68 * systemd-resolved may now optionally use libidn2 instead of the libidn
69 for processing internationalized domain names. Support for libidn2
70 should be considered experimental and should not be enabled by
73 * "machinectl pull-tar" and related call may now do verification of
74 downloaded images using SUSE-style .sha256 checksum files in addition
75 to the already existing support for validating using Ubuntu-style
78 * sd-bus gained support for a new sd_bus_message_appendv() call which
79 is va_list equivalent of sd_bus_message_append().
81 * sd-boot gained support for validating images using SHIM/MOK.
83 * The SMACK code learnt support for "onlycap".
85 * systemd-mount --umount is now much smarter in figuring out how to
86 properly unmount a device given its mount or device path.
88 * The code to call libnss_dns as a fallback from libnss_resolve when
89 the communication with systemd-resolved fails was removed. This
90 fallback was redundant and interfered with the [!UNAVAIL=return]
91 suffix. See nss-resolve(8) for the recommended configuration.
93 * systemd-logind may now be restarted without losing state. It stores
94 the file descriptors for devices it manages in the system manager
95 using the FDSTORE= mechanism. Please note that further changes in
96 other components may be required to make use of this (for example
97 Xorg has code to listen for stops of systemd-logind and terminate
98 itself when logind is stopped or restarted, in order to avoid using
99 stale file descriptors for graphical devices, which is now
100 counterproductive and must be reverted in order for restarts of
101 systemd-logind to be safe. See
102 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
104 * All kernel install plugins are called with the environment variable
105 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
106 /etc/machine-id. If the file is missing or empty, the variable is
107 empty and BOOT_DIR_ABS is the path of a temporary directory which is
108 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
109 is empty, all plugins should not put anything in BOOT_DIR_ABS.
111 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
112 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
113 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
114 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
115 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
116 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
117 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
118 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
119 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
120 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
121 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
122 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
123 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
124 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
125 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
126 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
127 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
128 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
129 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
130 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
131 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
132 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
133 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
134 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
135 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
136 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
137 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
138 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
139 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
146 * The "hybrid" control group mode has been modified to improve
147 compatibility with "legacy" cgroups-v1 setups. Specifically, the
148 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
149 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
150 cgroups-v1 hierarchy), the only externally visible change being that
151 the cgroups-v2 hierarchy is also mounted, to
152 /sys/fs/cgroup/unified. This should provide a large degree of
153 compatibility with "legacy" cgroups-v1, while taking benefit of the
154 better management capabilities of cgroups-v2.
156 * The default control group setup mode may be selected both a boot-time
157 via a set of kernel command line parameters (specifically:
158 systemd.unified_cgroup_hierarchy= and
159 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
160 default selected on the configure command line
161 (--with-default-hierarchy=). The upstream default is "hybrid"
162 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
163 this will change in a future systemd version to be "unified" (pure
164 cgroups-v2 mode). The third option for the compile time option is
165 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
166 distributions to default to "hybrid" mode for release distributions,
167 starting with v233. We recommend "unified" for development
168 distributions (specifically: distributions such as Fedora's rawhide)
169 as that's where things are headed in the long run. Use "legacy" for
170 greatest stability and compatibility only.
172 * Note one current limitation of "unified" and "hybrid" control group
173 setup modes: the kernel currently does not permit the systemd --user
174 instance (i.e. unprivileged code) to migrate processes between two
175 disconnected cgroup subtrees, even if both are managed and owned by
176 the user. This effectively means "systemd-run --user --scope" doesn't
177 work when invoked from outside of any "systemd --user" service or
178 scope. Specifically, it is not supported from session scopes. We are
179 working on fixing this in a future systemd version. (See #3388 for
180 further details about this.)
182 * DBus policy files are now installed into /usr rather than /etc. Make
183 sure your system has dbus >= 1.9.18 running before upgrading to this
184 version, or override the install path with --with-dbuspolicydir= .
186 * All python scripts shipped with systemd (specifically: the various
187 tests written in Python) now require Python 3.
189 * systemd unit tests can now run standalone (without the source or
190 build directories), and can be installed into /usr/lib/systemd/tests/
191 with 'make install-tests'.
193 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
194 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
197 * Support for the %c, %r, %R specifiers in unit files has been
198 removed. Specifiers are not supposed to be dependent on configuration
199 in the unit file itself (so that they resolve the same regardless
200 where used in the unit files), but these specifiers were influenced
201 by the Slice= option.
203 * The shell invoked by debug-shell.service now defaults to /bin/sh in
204 all cases. If distributions want to use a different shell for this
205 purpose (for example Fedora's /sbin/sushell) they need to specify
206 this explicitly at configure time using --with-debug-shell=.
208 * The confirmation spawn prompt has been reworked to offer the
211 (c)ontinue, proceed without asking anymore
212 (D)ump, show the state of the unit
213 (f)ail, don't execute the command and pretend it failed
215 (i)nfo, show a short summary of the unit
216 (j)obs, show jobs that are in progress
217 (s)kip, don't execute the command and pretend it succeeded
218 (y)es, execute the command
220 The 'n' choice for the confirmation spawn prompt has been removed,
221 because its meaning was confusing.
223 The prompt may now also be redirected to an alternative console by
224 specifying the console as parameter to systemd.confirm_spawn=.
226 * Services of Type=notify require a READY=1 notification to be sent
227 during startup. If no such message is sent, the service now fails,
228 even if the main process exited with a successful exit code.
230 * Services that fail to start up correctly now always have their
231 ExecStopPost= commands executed. Previously, they'd enter "failed"
232 state directly, without executing these commands.
234 * The option MulticastDNS= of network configuration files has acquired
235 an actual implementation. With MulticastDNS=yes a host can resolve
236 names of remote hosts and reply to mDNS A and AAAA requests.
238 * When units are about to be started an additional check is now done to
239 ensure that all dependencies of type BindsTo= (when used in
240 combination with After=) have been started.
242 * systemd-analyze gained a new verb "syscall-filter" which shows which
243 system call groups are defined for the SystemCallFilter= unit file
244 setting, and which system calls they contain.
246 * A new system call filter group "@filesystem" has been added,
247 consisting of various file system related system calls. Group
248 "@reboot" has been added, covering reboot, kexec and shutdown related
249 calls. Finally, group "@swap" has been added covering swap
250 configuration related calls.
252 * A new unit file option RestrictNamespaces= has been added that may be
253 used to restrict access to the various process namespace types the
254 Linux kernel provides. Specifically, it may be used to take away the
255 right for a service unit to create additional file system, network,
256 user, and other namespaces. This sandboxing option is particularly
257 relevant due to the high amount of recently discovered namespacing
258 related vulnerabilities in the kernel.
260 * systemd-udev's .link files gained support for a new AutoNegotiation=
261 setting for configuring Ethernet auto-negotiation.
263 * systemd-networkd's .network files gained support for a new
264 ListenPort= setting in the [DHCP] section to explicitly configure the
265 UDP client port the DHCP client shall listen on.
267 * .network files gained a new Unmanaged= boolean setting for explicitly
268 excluding one or more interfaces from management by systemd-networkd.
270 * The systemd-networkd ProxyARP= option has been renamed to
271 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
272 renamed to ReduceARPProxy=. The old names continue to be available
275 * systemd-networkd gained support for configuring IPv6 Proxy NDP
276 addresses via the new IPv6ProxyNDPAddress= .network file setting.
278 * systemd-networkd's bonding device support gained support for two new
279 configuration options ActiveSlave= and PrimarySlave=.
281 * The various options in the [Match] section of .network files gained
282 support for negative matching.
284 * New systemd-specific mount options are now understood in /etc/fstab:
286 x-systemd.mount-timeout= may be used to configure the maximum
287 permitted runtime of the mount command.
289 x-systemd.device-bound may be set to bind a mount point to its
290 backing device unit, in order to automatically remove a mount point
291 if its backing device is unplugged. This option may also be
292 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
293 on the block device, which is now automatically set for all CDROM
294 drives, so that mounted CDs are automatically unmounted when they are
295 removed from the drive.
297 x-systemd.after= and x-systemd.before= may be used to explicitly
298 order a mount after or before another unit or mount point.
300 * Enqueued start jobs for device units are now automatically garbage
301 collected if there are no jobs waiting for them anymore.
303 * systemctl list-jobs gained two new switches: with --after, for every
304 queued job the jobs it's waiting for are shown; with --before the
305 jobs which it's blocking are shown.
307 * systemd-nspawn gained support for ephemeral boots from disk images
308 (or in other words: --ephemeral and --image= may now be
309 combined). Moreover, ephemeral boots are now supported for normal
310 directories, even if the backing file system is not btrfs. Of course,
311 if the file system does not support file system snapshots or
312 reflinks, the initial copy operation will be relatively expensive, but
313 this should still be suitable for many use cases.
315 * Calendar time specifications in .timer units now support
316 specifications relative to the end of a month by using "~" instead of
317 "-" as separator between month and day. For example, "*-02~03" means
318 "the third last day in February". In addition a new syntax for
319 repeated events has been added using the "/" character. For example,
320 "9..17/2:00" means "every two hours from 9am to 5pm".
322 * systemd-socket-proxyd gained a new parameter --connections-max= for
323 configuring the maximum number of concurrent connections.
325 * sd-id128 gained a new API for generating unique IDs for the host in a
326 way that does not leak the machine ID. Specifically,
327 sd_id128_get_machine_app_specific() derives an ID based on the
328 machine ID a in well-defined, non-reversible, stable way. This is
329 useful whenever an identifier for the host is needed but where the
330 identifier shall not be useful to identify the system beyond the
331 scope of the application itself. (Internally this uses HMAC-SHA256 as
332 keyed hash function using the machine ID as input.)
334 * NotifyAccess= gained a new supported value "exec". When set
335 notifications are accepted from all processes systemd itself invoked,
336 including all control processes.
338 * .nspawn files gained support for defining overlay mounts using the
339 Overlay= and OverlayReadOnly= options. Previously this functionality
340 was only available on the systemd-nspawn command line.
342 * systemd-nspawn's --bind= and --overlay= options gained support for
343 bind/overlay mounts whose source lies within the container tree by
344 prefixing the source path with "+".
346 * systemd-nspawn's --bind= and --overlay= options gained support for
347 automatically allocating a temporary source directory in /var/tmp
348 that is removed when the container dies. Specifically, if the source
349 directory is specified as empty string this mechanism is selected. An
350 example usage is --overlay=+/var::/var, which creates an overlay
351 mount based on the original /var contained in the image, overlayed
352 with a temporary directory in the host's /var/tmp. This way changes
353 to /var are automatically flushed when the container shuts down.
355 * systemd-nspawn --image= option does now permit raw file system block
356 devices (in addition to images containing partition tables, as
359 * The disk image dissection logic in systemd-nspawn gained support for
360 automatically setting up LUKS encrypted as well as Verity protected
361 partitions. When a container is booted from an encrypted image the
362 passphrase is queried at start-up time. When a container with Verity
363 data is started, the root hash is search in a ".roothash" file
364 accompanying the disk image (alternatively, pass the root hash via
365 the new --root-hash= command line option).
367 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
368 be used to dissect disk images the same way as systemd-nspawn does
369 it, following the Bootable Partition Specification. It may even be
370 used to mount disk images with complex partition setups (including
371 LUKS and Verity partitions) to a local host directory, in order to
372 inspect them. This tool is not considered public API (yet), and is
373 thus not installed into /usr/bin. Please do not rely on its
374 existence, since it might go away or be changed in later systemd
377 * A new generator "systemd-verity-generator" has been added, similar in
378 style to "systemd-cryptsetup-generator", permitting automatic setup of
379 Verity root partitions when systemd boots up. In order to make use of
380 this your partition setup should follow the Discoverable Partitions
381 Specification, and the GPT partition ID of the root file system
382 partition should be identical to the upper 128bit of the Verity root
383 hash. The GPT partition ID of the Verity partition protecting it
384 should be the lower 128bit of the Verity root hash. If the partition
385 image follows this model it is sufficient to specify a single
386 "roothash=" kernel command line argument to both configure which root
387 image and verity partition to use as well as the root hash for
388 it. Note that systemd-nspawn's Verity support follows the same
389 semantics, meaning that disk images with proper Verity data in place
390 may be booted in containers with systemd-nspawn as well as on
391 physical systems via the verity generator. Also note that the "mkosi"
392 tool available at https://github.com/systemd/mkosi has been updated
393 to generate Verity protected disk images following this scheme. In
394 fact, it has been updated to generate disk images that optionally
395 implement a complete UEFI SecureBoot trust chain, involving a signed
396 kernel and initrd image that incorporates such a root hash as well as
397 a Verity-enabled root partition.
399 * The hardware database (hwdb) udev supports has been updated to carry
400 accelerometer quirks.
402 * All system services are now run with a fresh kernel keyring set up
403 for them. The invocation ID is stored by default in it, thus
404 providing a safe, non-overridable way to determine the invocation
407 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
408 options for bind mounting arbitrary paths in a service-specific
409 way. When these options are used, arbitrary host or service files and
410 directories may be mounted to arbitrary locations in the service's
413 * Documentation has been added that lists all of systemd's low-level
414 environment variables:
416 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
418 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
419 whether a specific socket file descriptor matches a specified socket
422 * systemd-firstboot has been updated to check for the
423 systemd.firstboot= kernel command line option. It accepts a boolean
424 and when set to false the first boot questions are skipped.
426 * systemd-fstab-generator has been updated to check for the
427 systemd.volatile= kernel command line option, which either takes an
428 optional boolean parameter or the special value "state". If used the
429 system may be booted in a "volatile" boot mode. Specifically,
430 "systemd.volatile" is used, the root directory will be mounted as
431 tmpfs, and only /usr is mounted from the actual root file system. If
432 "systemd.volatile=state" is used, the root directory will be mounted
433 as usual, but /var is mounted as tmpfs. This concept provides similar
434 functionality as systemd-nspawn's --volatile= option, but provides it
435 on physical boots. Use this option for implementing stateless
436 systems, or testing systems with all state and/or configuration reset
437 to the defaults. (Note though that many distributions are not
438 prepared to boot up without a populated /etc or /var, though.)
440 * systemd-gpt-auto-generator gained support for LUKS encrypted root
441 partitions. Previously it only supported LUKS encrypted partitions
442 for all other uses, except for the root partition itself.
444 * Socket units gained support for listening on AF_VSOCK sockets for
445 communication in virtualized QEMU environments.
447 * The "configure" script gained a new option --with-fallback-hostname=
448 for specifying the fallback hostname to use if none is configured in
449 /etc/hostname. For example, by specifying
450 --with-fallback-hostname=fedora it is possible to default to a
451 hostname of "fedora" on pristine installations.
453 * systemd-cgls gained support for a new --unit= switch for listing only
454 the control groups of a specific unit. Similar --user-unit= has been
455 added for listing only the control groups of a specific user unit.
457 * systemd-mount gained a new --umount switch for unmounting a mount or
458 automount point (and all mount/automount points below it).
460 * systemd will now refuse full configuration reloads (via systemctl
461 daemon-reload and related calls) unless at least 16MiB of free space
462 are available in /run. This is a safety precaution in order to ensure
463 that generators can safely operate after the reload completed.
465 * A new unit file option RootImage= has been added, which has a similar
466 effect as RootDirectory= but mounts the service's root directory from
467 a disk image instead of plain directory. This logic reuses the same
468 image dissection and mount logic that systemd-nspawn already uses,
469 and hence supports any disk images systemd-nspawn supports, including
470 those following the Discoverable Partition Specification, as well as
471 Verity enabled images. This option enables systemd to run system
472 services directly off disk images acting as resource bundles,
473 possibly even including full integrity data.
475 * A new MountAPIVFS= unit file option has been added, taking a boolean
476 argument. If enabled /proc, /sys and /dev (collectively called the
477 "API VFS") will be mounted for the service. This is only relevant if
478 RootDirectory= or RootImage= is used for the service, as these mounts
479 are of course in place in the host mount namespace anyway.
481 * systemd-nspawn gained support for a new --pivot-root= switch. If
482 specified the root directory within the container image is pivoted to
483 the specified mount point, while the original root disk is moved to a
484 different place. This option enables booting of ostree images
485 directly with systemd-nspawn.
487 * The systemd build scripts will no longer complain if the NTP server
488 addresses are not changed from the defaults. Google now supports
489 these NTP servers officially. We still recommend downstreams to
490 properly register an NTP pool with the NTP pool project though.
492 * coredumpctl gained a new "--reverse" option for printing the list
493 of coredumps in reverse order.
495 * coredumpctl will now show additional information about truncated and
496 inaccessible coredumps, as well as coredumps that are still being
497 processed. It also gained a new --quiet switch for suppressing
498 additional informational message in its output.
500 * coredumpctl gained support for only showing coredumps newer and/or
501 older than specific timestamps, using the new --since= and --until=
502 options, reminiscent of journalctl's options by the same name.
504 * The systemd-coredump logic has been improved so that it may be reused
505 to collect backtraces in non-compiled languages, for example in
506 scripting languages such as Python.
508 * machinectl will now show the UID shift of local containers, if user
509 namespacing is enabled for them.
511 * systemd will now optionally run "environment generator" binaries at
512 configuration load time. They may be used to add environment
513 variables to the environment block passed to services invoked. One
514 user environment generator is shipped by default that sets up
515 environment variables based on files dropped into /etc/environment.d
516 and ~/.config/environment.d/.
518 * systemd-resolved now includes the new, recently published 2017 DNSSEC
521 * hostnamed has been updated to report a new chassis type of
522 "convertible" to cover "foldable" laptops that can both act as a
523 tablet and as a laptop, such as various Lenovo Yoga devices.
525 Contributions from: Adrián López, Alexander Galanin, Alexander
526 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
527 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
528 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
529 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
530 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
531 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
532 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
533 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
534 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
535 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
536 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
537 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
538 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
539 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
540 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
541 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
542 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
543 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
544 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
545 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
546 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
547 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
548 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
549 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
550 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
551 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
552 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
559 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
560 RestrictAddressFamilies= enabled. These sandboxing options should
561 generally be compatible with the various external udev call-out
562 binaries we are aware of, however there may be exceptions, in
563 particular when exotic languages for these call-outs are used. In
564 this case, consider turning off these settings locally.
566 * The new RemoveIPC= option can be used to remove IPC objects owned by
567 the user or group of a service when that service exits.
569 * The new ProtectKernelModules= option can be used to disable explicit
570 load and unload operations of kernel modules by a service. In
571 addition access to /usr/lib/modules is removed if this option is set.
573 * ProtectSystem= option gained a new value "strict", which causes the
574 whole file system tree with the exception of /dev, /proc, and /sys,
575 to be remounted read-only for a service.
577 * The new ProtectKernelTunables= option can be used to disable
578 modification of configuration files in /sys and /proc by a service.
579 Various directories and files are remounted read-only, so access is
580 restricted even if the file permissions would allow it.
582 * The new ProtectControlGroups= option can be used to disable write
583 access by a service to /sys/fs/cgroup.
585 * Various systemd services have been hardened with
586 ProtectKernelTunables=yes, ProtectControlGroups=yes,
587 RestrictAddressFamilies=.
589 * Support for dynamically creating users for the lifetime of a service
590 has been added. If DynamicUser=yes is specified, user and group IDs
591 will be allocated from the range 61184..65519 for the lifetime of the
592 service. They can be resolved using the new nss-systemd.so NSS
593 module. The module must be enabled in /etc/nsswitch.conf. Services
594 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
595 any resources allocated by the service will be cleaned up when the
596 service exits. They also have ProtectHome=read-only and
597 ProtectSystem=strict enabled, so they are not able to make any
598 permanent modifications to the system.
600 * The nss-systemd module also always resolves root and nobody, making
601 it possible to have no /etc/passwd or /etc/group files in minimal
602 container or chroot environments.
604 * Services may be started with their own user namespace using the new
605 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
606 under which the service is running are mapped. All other users are
609 * Support for the cgroup namespace has been added to systemd-nspawn. If
610 supported by kernel, the container system started by systemd-nspawn
611 will have its own view of the cgroup hierarchy. This new behaviour
612 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
614 * The new MemorySwapMax= option can be used to limit the maximum swap
615 usage under the unified cgroup hierarchy.
617 * Support for the CPU controller in the unified cgroup hierarchy has
618 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
619 options. This controller requires out-of-tree patches for the kernel
620 and the support is provisional.
622 * Mount and automount units may now be created transiently
623 (i.e. dynamically at runtime via the bus API, instead of requiring
624 unit files in the file system).
626 * systemd-mount is a new tool which may mount file systems – much like
627 mount(8), optionally pulling in additional dependencies through
628 transient .mount and .automount units. For example, this tool
629 automatically runs fsck on a backing block device before mounting,
630 and allows the automount logic to be used dynamically from the
631 command line for establishing mount points. This tool is particularly
632 useful when dealing with removable media, as it will ensure fsck is
633 run – if necessary – before the first access and that the file system
634 is quickly unmounted after each access by utilizing the automount
635 logic. This maximizes the chance that the file system on the
636 removable media stays in a clean state, and if it isn't in a clean
637 state is fixed automatically.
639 * LazyUnmount=yes option for mount units has been added to expose the
640 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
643 * /efi will be used as the mount point of the EFI boot partition, if
644 the directory is present, and the mount point was not configured
645 through other means (e.g. fstab). If /efi directory does not exist,
646 /boot will be used as before. This makes it easier to automatically
647 mount the EFI partition on systems where /boot is used for something
650 * When operating on GPT disk images for containers, systemd-nspawn will
651 now mount the ESP to /boot or /efi according to the same rules as PID
652 1 running on a host. This allows tools like "bootctl" to operate
653 correctly within such containers, in order to make container images
654 bootable on physical systems.
656 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
658 * Two new user session targets have been added to support running
659 graphical sessions under the systemd --user instance:
660 graphical-session.target and graphical-session-pre.target. See
661 systemd.special(7) for a description of how those targets should be
664 * The vconsole initialization code has been significantly reworked to
665 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
666 support unicode keymaps. Font and keymap configuration will now be
667 copied to all allocated virtual consoles.
669 * FreeBSD's bhyve virtualization is now detected.
671 * Information recorded in the journal for core dumps now includes the
672 contents of /proc/mountinfo and the command line of the process at
673 the top of the process hierarchy (which is usually the init process
676 * systemd-journal-gatewayd learned the --directory= option to serve
677 files from the specified location.
679 * journalctl --root=… can be used to peruse the journal in the
680 /var/log/ directories inside of a container tree. This is similar to
681 the existing --machine= option, but does not require the container to
684 * The hardware database has been extended to support
685 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
688 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
689 specify the click rate for mice which include a horizontal wheel with
690 a click rate that is different than the one for the vertical wheel.
692 * systemd-run gained a new --wait option that makes service execution
693 synchronous. (Specifically, the command will not return until the
694 specified service binary exited.)
696 * systemctl gained a new --wait option that causes the start command to
697 wait until the units being started have terminated again.
699 * A new journal output mode "short-full" has been added which displays
700 timestamps with abbreviated English day names and adds a timezone
701 suffix. Those timestamps include more information than the default
702 "short" output mode, and can be passed directly to journalctl's
703 --since= and --until= options.
705 * /etc/resolv.conf will be bind-mounted into containers started by
706 systemd-nspawn, if possible, so any changes to resolv.conf contents
707 are automatically propagated to the container.
709 * The number of instances for socket-activated services originating
710 from a single IP address can be limited with
711 MaxConnectionsPerSource=, extending the existing setting of
714 * systemd-networkd gained support for vcan ("Virtual CAN") interface
717 * .netdev and .network configuration can now be extended through
720 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
721 Segmentation Offload, Generic Receive Offload, Large Receive Offload
722 can be enabled and disabled using the new UDPSegmentationOffload=,
723 TCPSegmentationOffload=, GenericSegmentationOffload=,
724 GenericReceiveOffload=, LargeReceiveOffload= options in the
725 [Link] section of .link files.
727 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
728 Port VLAN ID can be configured for bridge devices using the new STP=,
729 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
730 section of .netdev files.
732 * The route table to which routes received over DHCP or RA should be
733 added can be configured with the new RouteTable= option in the [DHCP]
734 and [IPv6AcceptRA] sections of .network files.
736 * The Address Resolution Protocol can be disabled on links managed by
737 systemd-networkd using the ARP=no setting in the [Link] section of
740 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
741 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
742 encode information about the result and exit codes of the current
743 service runtime cycle.
745 * systemd-sysctl will now configure kernel parameters in the order
746 they occur in the configuration files. This matches what sysctl
747 has been traditionally doing.
749 * kernel-install "plugins" that are executed to perform various
750 tasks after a new kernel is added and before an old one is removed
751 can now return a special value to terminate the procedure and
752 prevent any later plugins from running.
754 * Journald's SplitMode=login setting has been deprecated. It has been
755 removed from documentation, and its use is discouraged. In a future
756 release it will be completely removed, and made equivalent to current
757 default of SplitMode=uid.
759 * Storage=both option setting in /etc/systemd/coredump.conf has been
760 removed. With fast LZ4 compression storing the core dump twice is not
763 * The --share-system systemd-nspawn option has been replaced with an
764 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
765 this functionality is discouraged. In addition the variables
766 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
767 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
768 individual namespaces.
770 * "machinectl list" now shows the IP address of running containers in
771 the output, as well as OS release information.
773 * "loginctl list" now shows the TTY of each session in the output.
775 * sd-bus gained new API calls sd_bus_track_set_recursive(),
776 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
777 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
778 tracking objects in a "recursive" mode, where a single client can be
779 counted multiple times, if it takes multiple references.
781 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
782 sd_bus_get_exit_on_disconnect(). They may be used to to make a
783 process using sd-bus automatically exit if the bus connection is
786 * Bus clients of the service manager may now "pin" loaded units into
787 memory, by taking an explicit reference on them. This is useful to
788 ensure the client can retrieve runtime data about the service even
789 after the service completed execution. Taking such a reference is
790 available only for privileged clients and should be helpful to watch
791 running services in a race-free manner, and in particular collect
792 information about exit statuses and results.
794 * The nss-resolve module has been changed to strictly return UNAVAIL
795 when communication via D-Bus with resolved failed, and NOTFOUND when
796 a lookup completed but was negative. This means it is now possible to
797 neatly configure fallbacks using nsswitch.conf result checking
798 expressions. Taking benefit of this, the new recommended
799 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
801 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
803 * A new setting CtrlAltDelBurstAction= has been added to
804 /etc/systemd/system.conf which may be used to configure the precise
805 behaviour if the user on the console presses Ctrl-Alt-Del more often
806 than 7 times in 2s. Previously this would unconditionally result in
807 an expedited, immediate reboot. With this new setting the precise
808 operation may be configured in more detail, and also turned off
811 * In .netdev files two new settings RemoteChecksumTx= and
812 RemoteChecksumRx= are now understood that permit configuring the
813 remote checksumming logic for VXLAN networks.
815 * The service manager learnt a new "invocation ID" concept for invoked
816 services. Each runtime cycle of a service will get a new invocation
817 ID (a 128bit random UUID) assigned that identifies the current
818 run of the service uniquely and globally. A new invocation ID
819 is generated each time a service starts up. The journal will store
820 the invocation ID of a service along with any logged messages, thus
821 making the invocation ID useful for matching the online runtime of a
822 service with the offline log data it generated in a safe way without
823 relying on synchronized timestamps. In many ways this new service
824 invocation ID concept is similar to the kernel's boot ID concept that
825 uniquely and globally identifies the runtime of each boot. The
826 invocation ID of a service is passed to the service itself via an
827 environment variable ($INVOCATION_ID). A new bus call
828 GetUnitByInvocationID() has been added that is similar to GetUnit()
829 but instead of retrieving the bus path for a unit by its name
830 retrieves it by its invocation ID. The returned path is valid only as
831 long as the passed invocation ID is current.
833 * systemd-resolved gained a new "DNSStubListener" setting in
834 resolved.conf. It either takes a boolean value or the special values
835 "udp" and "tcp", and configures whether to enable the stub DNS
836 listener on 127.0.0.53:53.
838 * IP addresses configured via networkd may now carry additional
839 configuration settings supported by the kernel. New options include:
840 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
841 PrefixRoute=, AutoJoin=.
843 * The PAM configuration fragment file for "user@.service" shipped with
844 systemd (i.e. the --user instance of systemd) has been stripped to
845 the minimum necessary to make the system boot. Previously, it
846 contained Fedora-specific stanzas that did not apply to other
847 distributions. It is expected that downstream distributions add
848 additional configuration lines, matching their needs to this file,
849 using it only as rough template of what systemd itself needs. Note
850 that this reduced fragment does not even include an invocation of
851 pam_limits which most distributions probably want to add, even though
852 systemd itself does not need it. (There's also the new build time
853 option --with-pamconfdir=no to disable installation of the PAM
856 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
857 capability is now also dropped from its set (in addition to
858 CAP_SYS_MKNOD as before).
860 * In service unit files it is now possible to connect a specific named
861 file descriptor with stdin/stdout/stdout of an executed service. The
862 name may be specified in matching .socket units using the
863 FileDescriptorName= setting.
865 * A number of journal settings may now be configured on the kernel
866 command line. Specifically, the following options are now understood:
867 systemd.journald.max_level_console=,
868 systemd.journald.max_level_store=,
869 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
870 systemd.journald.max_level_wall=.
872 * "systemctl is-enabled --full" will now show by which symlinks a unit
873 file is enabled in the unit dependency tree.
875 * Support for VeraCrypt encrypted partitions has been added to the
876 "cryptsetup" logic and /etc/crypttab.
878 * systemd-detect-virt gained support for a new --private-users switch
879 that checks whether the invoking processes are running inside a user
880 namespace. Similar, a new special value "private-users" for the
881 existing ConditionVirtualization= setting has been added, permitting
882 skipping of specific units in user namespace environments.
884 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
885 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
886 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
887 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
888 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
889 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
890 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
891 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
892 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
893 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
894 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
895 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
896 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
897 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
898 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
899 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
900 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
901 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
902 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
903 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
904 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
905 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
906 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
907 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
908 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
909 Jędrzejewski-Szmek, Zeal Jagannatha
911 — Santa Fe, 2016-11-03
915 * In service units the various ExecXYZ= settings have been extended
916 with an additional special character as first argument of the
917 assigned value: if the character '+' is used the specified command
918 line it will be run with full privileges, regardless of User=,
919 Group=, CapabilityBoundingSet= and similar options. The effect is
920 similar to the existing PermissionsStartOnly= option, but allows
921 configuration of this concept for each executed command line
924 * Services may now alter the service watchdog timeout at runtime by
925 sending a WATCHDOG_USEC= message via sd_notify().
927 * MemoryLimit= and related unit settings now optionally take percentage
928 specifications. The percentage is taken relative to the amount of
929 physical memory in the system (or in case of containers, the assigned
930 amount of memory). This allows scaling service resources neatly with
931 the amount of RAM available on the system. Similarly, systemd-logind's
932 RuntimeDirectorySize= option now also optionally takes percentage
935 * In similar fashion TasksMax= takes percentage values now, too. The
936 value is taken relative to the configured maximum number of processes
937 on the system. The per-service task maximum has been changed to 15%
938 using this functionality. (Effectively this is an increase of 512 →
939 4915 for service units, given the kernel's default pid_max setting.)
941 * Calendar time specifications in .timer units now understand a ".."
942 syntax for time ranges. Example: "4..7:10" may now be used for
943 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
946 * The InaccessableDirectories=, ReadOnlyDirectories= and
947 ReadWriteDirectories= unit file settings have been renamed to
948 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
949 applied to all kinds of file nodes, and not just directories, with
950 the exception of symlinks. Specifically these settings may now be
951 used on block and character device nodes, UNIX sockets and FIFOS as
952 well as regular files. The old names of these settings remain
953 available for compatibility.
955 * systemd will now log about all service processes it kills forcibly
956 (using SIGKILL) because they remained after the clean shutdown phase
957 of the service completed. This should help identifying services that
958 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
959 systemd-logind's configuration a similar log message is generated for
960 processes killed at the end of each session due to this setting.
962 * systemd will now set the $JOURNAL_STREAM environment variable for all
963 services whose stdout/stderr are connected to the Journal (which
964 effectively means by default: all services). The variable contains
965 the device and inode number of the file descriptor used for
966 stdout/stderr. This may be used by invoked programs to detect whether
967 their stdout/stderr is connected to the Journal, in which case they
968 can switch over to direct Journal communication, thus being able to
969 pass extended, structured metadata along with their log messages. As
970 one example, this is now used by glib's logging primitives.
972 * When using systemd's default tmp.mount unit for /tmp, the mount point
973 will now be established with the "nosuid" and "nodev" options. This
974 avoids privilege escalation attacks that put traps and exploits into
975 /tmp. However, this might cause problems if you e. g. put container
976 images or overlays into /tmp; if you need this, override tmp.mount's
977 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
980 * systemd now supports the "memory" cgroup controller also on
983 * The systemd-cgtop tool now optionally takes a control group path as
984 command line argument. If specified, the control group list shown is
985 limited to subgroups of that group.
987 * The SystemCallFilter= unit file setting gained support for
988 pre-defined, named system call filter sets. For example
989 SystemCallFilter=@clock is now an effective way to make all clock
990 changing-related system calls unavailable to a service. A number of
991 similar pre-defined groups are defined. Writing system call filters
992 for system services is simplified substantially with this new
993 concept. Accordingly, all of systemd's own, long-running services now
994 enable system call filtering based on this, by default.
996 * A new service setting MemoryDenyWriteExecute= has been added, taking
997 a boolean value. If turned on, a service may no longer create memory
998 mappings that are writable and executable at the same time. This
999 enhances security for services where this is enabled as it becomes
1000 harder to dynamically write and then execute memory in exploited
1001 service processes. This option has been enabled for all of systemd's
1002 own long-running services.
1004 * A new RestrictRealtime= service setting has been added, taking a
1005 boolean argument. If set the service's processes may no longer
1006 acquire realtime scheduling. This improves security as realtime
1007 scheduling may otherwise be used to easily freeze the system.
1009 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1010 value. This may be used for requesting that the system manager inside
1011 of the container reports start-up completion to nspawn which then
1012 propagates this notification further to the service manager
1013 supervising nspawn itself. A related option NotifyReady= in .nspawn
1014 files has been added too. This functionality allows ordering of the
1015 start-up of multiple containers using the usual systemd ordering
1018 * machinectl gained a new command "stop" that is an alias for
1021 * systemd-resolved gained support for contacting DNS servers on
1022 link-local IPv6 addresses.
1024 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1025 its caches. A method call for requesting the same operation has been
1026 added to the bus API too, and is made available via "systemd-resolve
1029 * systemd-resolve gained a new --status switch. If passed a brief
1030 summary of the used DNS configuration with per-interface information
1033 * resolved.conf gained a new Cache= boolean option, defaulting to
1034 on. If turned off local DNS caching is disabled. This comes with a
1035 performance penalty in particular when DNSSEC is enabled. Note that
1036 resolved disables its internal caching implicitly anyway, when the
1037 configured DNS server is on a host-local IP address such as ::1 or
1038 127.0.0.1, thus automatically avoiding double local caching.
1040 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1041 for DNS requests. This improves compatibility with local programs
1042 that do not use the libc NSS or systemd-resolved's bus APIs for name
1043 resolution. This minimal DNS service is only available to local
1044 programs and does not implement the full DNS protocol, but enough to
1045 cover local DNS clients. A new, static resolv.conf file, listing just
1046 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1047 now recommended to make /etc/resolv.conf a symlink to this file in
1048 order to route all DNS lookups to systemd-resolved, regardless if
1049 done via NSS, the bus API or raw DNS packets. Note that this local
1050 DNS service is not as fully featured as the libc NSS or
1051 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1052 used to deliver link-local address information (as this implies
1053 sending a local interface index along), LLMNR/mDNS support via this
1054 interface is severely restricted. It is thus strongly recommended for
1055 all applications to use the libc NSS API or native systemd-resolved
1058 * systemd-networkd's bridge support learned a new setting
1059 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1060 in .network files has been added for configuring VLAN bridging in
1061 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1063 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1064 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1065 now be acquired without relying on DHCPv6. Two new options
1066 UseDomains= and UseDNS= have been added to configure this behaviour.
1068 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1069 renamed IPv6AcceptRA=, without altering its behaviour. The old
1070 setting name remains available for compatibility reasons.
1072 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1073 Key=, InputKey= and OutputKey=.
1075 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1076 interface configuration.
1078 * "systemctl edit" may now be used to create new unit files by
1079 specifying the --force switch.
1081 * sd-event gained a new function sd_event_get_iteration() for
1082 requesting the current iteration counter of the event loop. It starts
1083 at zero and is increased by one with each event loop iteration.
1085 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1086 file. It can be used in lieu of %systemd_requires in packages which
1087 don't use any systemd functionality and are intended to be installed
1088 in minimal containers without systemd present. This macro provides
1089 ordering dependencies to ensure that if the package is installed in
1090 the same rpm transaction as systemd, systemd will be installed before
1091 the scriptlets for the package are executed, allowing unit presets
1094 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1095 been added to simplify packaging of generators.
1097 * The os-release file gained VERSION_CODENAME field for the
1098 distribution nickname (e.g. VERSION_CODENAME=woody).
1100 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1101 can be set to disable parsing of metadata and the creation
1102 of persistent symlinks for that device.
1104 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1105 to make them available to logged-in users has been reverted.
1107 * Much of the common code of the various systemd components is now
1108 built into an internal shared library libsystemd-shared-231.so
1109 (incorporating the systemd version number in the name, to be updated
1110 with future releases) that the components link to. This should
1111 decrease systemd footprint both in memory during runtime and on
1112 disk. Note that the shared library is not for public use, and is
1113 neither API not ABI stable, but is likely to change with every new
1114 released update. Packagers need to make sure that binaries
1115 linking to libsystemd-shared.so are updated in step with the
1118 * Configuration for "mkosi" is now part of the systemd
1119 repository. mkosi is a tool to easily build legacy-free OS images,
1120 and is available on github: https://github.com/systemd/mkosi. If
1121 "mkosi" is invoked in the build tree a new raw OS image is generated
1122 incorporating the systemd sources currently being worked on and a
1123 clean, fresh distribution installation. The generated OS image may be
1124 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1125 UEFI PC. This functionality is particularly useful to easily test
1126 local changes made to systemd in a pristine, defined environment. See
1127 HACKING for details.
1129 * configure learned the --with-support-url= option to specify the
1130 distribution's bugtracker.
1132 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1133 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1134 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1135 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1136 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1137 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1138 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1139 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1140 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1141 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1142 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1143 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1144 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1145 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1146 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1147 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1148 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1149 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1150 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1152 — Berlin, 2016-07-25
1156 * DNSSEC is now turned on by default in systemd-resolved (in
1157 "allow-downgrade" mode), but may be turned off during compile time by
1158 passing "--with-default-dnssec=no" to "configure" (and of course,
1159 during runtime with DNSSEC= in resolved.conf). We recommend
1160 downstreams to leave this on at least during development cycles and
1161 report any issues with the DNSSEC logic upstream. We are very
1162 interested in collecting feedback about the DNSSEC validator and its
1163 limitations in the wild. Note however, that DNSSEC support is
1164 probably nothing downstreams should turn on in stable distros just
1165 yet, as it might create incompatibilities with a few DNS servers and
1166 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1167 automatically whenever we detect such incompatible setups, but there
1168 might be systems we do not cover yet. Hence: please help us testing
1169 the DNSSEC code, leave this on where you can, report back, but then
1170 again don't consider turning this on in your stable, LTS or
1171 production release just yet. (Note that you have to enable
1172 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1173 and its DNSSEC mode for host name resolution from local
1176 * systemd-resolve conveniently resolves DANE records with the --tlsa
1177 option and OPENPGPKEY records with the --openpgp option. It also
1178 supports dumping raw DNS record data via the new --raw= switch.
1180 * systemd-logind will now by default terminate user processes that are
1181 part of the user session scope unit (session-XX.scope) when the user
1182 logs out. This behavior is controlled by the KillUserProcesses=
1183 setting in logind.conf, and the previous default of "no" is now
1184 changed to "yes". This means that user sessions will be properly
1185 cleaned up after, but additional steps are necessary to allow
1186 intentionally long-running processes to survive logout.
1188 While the user is logged in at least once, user@.service is running,
1189 and any service that should survive the end of any individual login
1190 session can be started at a user service or scope using systemd-run.
1191 systemd-run(1) man page has been extended with an example which shows
1192 how to run screen in a scope unit underneath user@.service. The same
1193 command works for tmux.
1195 After the user logs out of all sessions, user@.service will be
1196 terminated too, by default, unless the user has "lingering" enabled.
1197 To effectively allow users to run long-term tasks even if they are
1198 logged out, lingering must be enabled for them. See loginctl(1) for
1199 details. The default polkit policy was modified to allow users to
1200 set lingering for themselves without authentication.
1202 Previous defaults can be restored at compile time by the
1203 --without-kill-user-processes option to "configure".
1205 * systemd-logind gained new configuration settings SessionsMax= and
1206 InhibitorsMax=, both with a default of 8192. It will not register new
1207 user sessions or inhibitors above this limit.
1209 * systemd-logind will now reload configuration on SIGHUP.
1211 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1212 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1213 enable. Also, support for the "io" cgroup controller in the unified
1214 hierarchy has been added, so that the "memory", "pids" and "io" are
1215 now the controllers that are supported on the unified hierarchy.
1217 WARNING: it is not possible to use previous systemd versions with
1218 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1219 is necessary to also update systemd in the initramfs if using the
1220 unified hierarchy. An updated SELinux policy is also required.
1222 * LLDP support has been extended, and both passive (receive-only) and
1223 active (sender) modes are supported. Passive mode ("routers-only") is
1224 enabled by default in systemd-networkd. Active LLDP mode is enabled
1225 by default for containers on the internal network. The "networkctl
1226 lldp" command may be used to list information gathered. "networkctl
1227 status" will also show basic LLDP information on connected peers now.
1229 * The IAID and DUID unique identifier sent in DHCP requests may now be
1230 configured for the system and each .network file managed by
1231 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1233 * systemd-networkd gained support for configuring proxy ARP support for
1234 each interface, via the ProxyArp= setting in .network files. It also
1235 gained support for configuring the multicast querier feature of
1236 bridge devices, via the new MulticastQuerier= setting in .netdev
1237 files. Similarly, snooping on the IGMP traffic can be controlled
1238 via the new setting MulticastSnooping=.
1240 A new setting PreferredLifetime= has been added for addresses
1241 configured in .network file to configure the lifetime intended for an
1244 The systemd-networkd DHCP server gained the option EmitRouter=, which
1245 defaults to yes, to configure whether the DHCP Option 3 (Router)
1248 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1249 systemd-socket-activate and installed into /usr/bin. It is now fully
1252 * systemd-journald now uses separate threads to flush changes to disk
1253 when closing journal files, thus reducing impact of slow disk I/O on
1254 logging performance.
1256 * The sd-journal API gained two new calls
1257 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1258 can be used to open journal files using file descriptors instead of
1259 file or directory paths. sd_journal_open_container() has been
1260 deprecated, sd_journal_open_directory_fd() should be used instead
1261 with the flag SD_JOURNAL_OS_ROOT.
1263 * journalctl learned a new output mode "-o short-unix" that outputs log
1264 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1265 UTC). It also gained support for a new --no-hostname setting to
1266 suppress the hostname column in the family of "short" output modes.
1268 * systemd-ask-password now optionally skips printing of the password to
1269 stdout with --no-output which can be useful in scripts.
1271 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1272 (devices tagged with ID_MAKER_TOOL) are now tagged with
1273 "uaccess" and are available to logged in users.
1275 * The DeviceAllow= unit setting now supports specifiers (with "%").
1277 * "systemctl show" gained a new --value switch, which allows print a
1278 only the contents of a specific unit property, without also printing
1279 the property's name. Similar support was added to "show*" verbs
1280 of loginctl and machinectl that output "key=value" lists.
1282 * A new unit type "generated" was added for files dynamically generated
1283 by generator tools. Similarly, a new unit type "transient" is used
1284 for unit files created using the runtime API. "systemctl enable" will
1285 refuse to operate on such files.
1287 * A new command "systemctl revert" has been added that may be used to
1288 revert to the vendor version of a unit file, in case local changes
1289 have been made by adding drop-ins or overriding the unit file.
1291 * "machinectl clean" gained a new verb to automatically remove all or
1292 just hidden container images.
1294 * systemd-tmpfiles gained support for a new line type "e" for emptying
1295 directories, if they exist, without creating them if they don't.
1297 * systemd-nspawn gained support for automatically patching the UID/GIDs
1298 of the owners and the ACLs of all files and directories in a
1299 container tree to match the UID/GID user namespacing range selected
1300 for the container invocation. This mode is enabled via the new
1301 --private-users-chown switch. It also gained support for
1302 automatically choosing a free, previously unused UID/GID range when
1303 starting a container, via the new --private-users=pick setting (which
1304 implies --private-users-chown). Together, these options for the first
1305 time make user namespacing for nspawn containers fully automatic and
1306 thus deployable. The systemd-nspawn@.service template unit file has
1307 been changed to use this functionality by default.
1309 * systemd-nspawn gained a new --network-zone= switch, that allows
1310 creating ad-hoc virtual Ethernet links between multiple containers,
1311 that only exist as long as at least one container referencing them is
1312 running. This allows easy connecting of multiple containers with a
1313 common link that implements an Ethernet broadcast domain. Each of
1314 these network "zones" may be named relatively freely by the user, and
1315 may be referenced by any number of containers, but each container may
1316 only reference one of these "zones". On the lower level, this is
1317 implemented by an automatically managed bridge network interface for
1318 each zone, that is created when the first container referencing its
1319 zone is created and removed when the last one referencing its zone
1322 * The default start timeout may now be configured on the kernel command
1323 line via systemd.default_timeout_start_sec=. It was already
1324 configurable via the DefaultTimeoutStartSec= option in
1325 /etc/systemd/system.conf.
1327 * Socket units gained a new TriggerLimitIntervalSec= and
1328 TriggerLimitBurst= setting to configure a limit on the activation
1329 rate of the socket unit.
1331 * The LimitNICE= setting now optionally takes normal UNIX nice values
1332 in addition to the raw integer limit value. If the specified
1333 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1334 value is understood as UNIX nice value. If not prefixed like this it
1335 is understood as raw RLIMIT_NICE limit.
1337 * Note that the effect of the PrivateDevices= unit file setting changed
1338 slightly with this release: the per-device /dev file system will be
1339 mounted read-only from this version on, and will have "noexec"
1340 set. This (minor) change of behavior might cause some (exceptional)
1341 legacy software to break, when PrivateDevices=yes is set for its
1342 service. Please leave PrivateDevices= off if you run into problems
1345 * systemd-bootchart has been split out to a separate repository:
1346 https://github.com/systemd/systemd-bootchart
1348 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1349 merged into the kernel in its current form.
1351 * The compatibility libraries libsystemd-daemon.so,
1352 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1353 which have been deprecated since systemd-209 have been removed along
1354 with the corresponding pkg-config files. All symbols provided by
1355 those libraries are provided by libsystemd.so.
1357 * The Capabilities= unit file setting has been removed (it is ignored
1358 for backwards compatibility). AmbientCapabilities= and
1359 CapabilityBoundingSet= should be used instead.
1361 * A new special target has been added, initrd-root-device.target,
1362 which creates a synchronization point for dependencies of the root
1363 device in early userspace. Initramfs builders must ensure that this
1364 target is now included in early userspace.
1366 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1367 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1368 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1369 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1370 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1371 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1372 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1373 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1374 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1375 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1376 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1377 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1378 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1379 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1380 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1381 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1382 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1383 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1384 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1385 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1386 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1387 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1388 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1389 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1390 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1393 — Fairfax, 2016-05-21
1397 * The systemd-resolved DNS resolver service has gained a substantial
1398 set of new features, most prominently it may now act as a DNSSEC
1399 validating stub resolver. DNSSEC mode is currently turned off by
1400 default, but is expected to be turned on by default in one of the
1401 next releases. For now, we invite everybody to test the DNSSEC logic
1402 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1403 service also gained a full set of D-Bus interfaces, including calls
1404 to configure DNS and DNSSEC settings per link (for use by external
1405 network management software). systemd-resolved and systemd-networkd
1406 now distinguish between "search" and "routing" domains. The former
1407 are used to qualify single-label names, the latter are used purely
1408 for routing lookups within certain domains to specific links.
1409 resolved now also synthesizes RRs for all entries from /etc/hosts.
1411 * The systemd-resolve tool (which is a client utility for
1412 systemd-resolved) has been improved considerably and is now fully
1413 supported and documented. Hence it has moved from /usr/lib/systemd to
1416 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1419 * The coredump collection logic has been reworked: when a coredump is
1420 collected it is now written to disk, compressed and processed
1421 (including stacktrace extraction) from a new instantiated service
1422 systemd-coredump@.service, instead of directly from the
1423 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1424 processing large coredumps can take up a substantial amount of
1425 resources and time, and this previously happened entirely outside of
1426 systemd's service supervision. With the new logic the core_pattern
1427 hook only does minimal metadata collection before passing off control
1428 to the new instantiated service, which is configured with a time
1429 limit, a nice level and other settings to minimize negative impact on
1430 the rest of the system. Also note that the new logic will honour the
1431 RLIMIT_CORE setting of the crashed process, which now allows users
1432 and processes to turn off coredumping for their processes by setting
1435 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1436 and all forked processes by default. Previously, PID 1 would leave
1437 the setting at "0" for all processes, as set by the kernel. Note that
1438 the resource limit traditionally has no effect on the generated
1439 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1440 logic is used. Since the limit is now honoured (see above) its
1441 default has been changed so that the coredumping logic is enabled by
1442 default for all processes, while allowing specific opt-out.
1444 * When the stacktrace is extracted from processes of system users, this
1445 is now done as "systemd-coredump" user, in order to sandbox this
1446 potentially security sensitive parsing operation. (Note that when
1447 processing coredumps of normal users this is done under the user ID
1448 of process that crashed, as before.) Packagers should take notice
1449 that it is now necessary to create the "systemd-coredump" system user
1450 and group at package installation time.
1452 * The systemd-activate socket activation testing tool gained support
1453 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1454 and --seqpacket switches. It also has been extended to support both
1455 new-style and inetd-style file descriptor passing. Use the new
1456 --inetd switch to request inetd-style file descriptor passing.
1458 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1459 variable, which takes a boolean value. If set to false, ANSI color
1460 output is disabled in the tools even when run on a terminal that
1463 * The VXLAN support in networkd now supports two new settings
1464 DestinationPort= and PortRange=.
1466 * A new systemd.machine_id= kernel command line switch has been added,
1467 that may be used to set the machine ID in /etc/machine-id if it is
1468 not initialized yet. This command line option has no effect if the
1469 file is already initialized.
1471 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1472 specified command line as PID 2 rather than PID 1 in the
1473 container. In this mode PID 1 is a minimal stub init process that
1474 implements the special POSIX and Linux semantics of PID 1 regarding
1475 signal and child process management. Note that this stub init process
1476 is implemented in nspawn itself and requires no support from the
1477 container image. This new logic is useful to support running
1478 arbitrary commands in the container, as normal processes are
1479 generally not prepared to run as PID 1.
1481 * systemd-nspawn gained a new --chdir= switch for setting the current
1482 working directory for the process started in the container.
1484 * "journalctl /dev/sda" will now output all kernel log messages for
1485 specified device from the current boot, in addition to all devices
1486 that are parents of it. This should make log output about devices
1487 pretty useful, as long as kernel drivers attach enough metadata to
1488 the log messages. (The usual SATA drivers do.)
1490 * The sd-journal API gained two new calls
1491 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1492 that report whether log data from /run or /var has been found.
1494 * journalctl gained a new switch "--fields" that prints all journal
1495 record field names currently in use in the journal. This is backed
1496 by two new sd-journal API calls sd_journal_enumerate_fields() and
1497 sd_journal_restart_fields().
1499 * Most configurable timeouts in systemd now expect an argument of
1500 "infinity" to turn them off, instead of "0" as before. The semantics
1501 from now on is that a timeout of "0" means "now", and "infinity"
1502 means "never". To maintain backwards compatibility, "0" continues to
1503 turn off previously existing timeout settings.
1505 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1506 try-reload-or-restart" to clarify what it actually does: the "try"
1507 logic applies to both reloading and restarting, not just restarting.
1508 The old name continues to be accepted for compatibility.
1510 * On boot-up, when PID 1 detects that the system clock is behind the
1511 release date of the systemd version in use, the clock is now set
1512 to the latter. Previously, this was already done in timesyncd, in order
1513 to avoid running with clocks set to the various clock epochs such as
1514 1902, 1938 or 1970. With this change the logic is now done in PID 1
1515 in addition to timesyncd during early boot-up, so that it is enforced
1516 before the first process is spawned by systemd. Note that the logic
1517 in timesyncd remains, as it is more comprehensive and ensures
1518 clock monotonicity by maintaining a persistent timestamp file in
1519 /var. Since /var is generally not available in earliest boot or the
1520 initrd, this part of the logic remains in timesyncd, and is not done
1523 * Support for tweaking details in net_cls.class_id through the
1524 NetClass= configuration directive has been removed, as the kernel
1525 people have decided to deprecate that controller in cgroup v2.
1526 Userspace tools such as nftables are moving over to setting rules
1527 that are specific to the full cgroup path of a task, which obsoletes
1528 these controllers anyway. The NetClass= directive is kept around for
1529 legacy compatibility reasons. For a more in-depth description of the
1530 kernel change, please refer to the respective upstream commit:
1532 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1534 * A new service setting RuntimeMaxSec= has been added that may be used
1535 to specify a maximum runtime for a service. If the timeout is hit, the
1536 service is terminated and put into a failure state.
1538 * A new service setting AmbientCapabilities= has been added. It allows
1539 configuration of additional Linux process capabilities that are
1540 passed to the activated processes. This is only available on very
1543 * The process resource limit settings in service units may now be used
1544 to configure hard and soft limits individually.
1546 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1547 expose support for gcc's __attribute__((cleanup())) C extension.
1548 Specifically, for many object destructor functions alternative
1549 versions have been added that have names suffixed with "p" and take a
1550 pointer to a pointer to the object to destroy, instead of just a
1551 pointer to the object itself. This is useful because these destructor
1552 functions may be used directly as parameters to the cleanup
1553 construct. Internally, systemd has been a heavy user of this GCC
1554 extension for a long time, and with this change similar support is
1555 now available to consumers of the library outside of systemd. Note
1556 that by using this extension in your sources compatibility with old
1557 and strictly ANSI compatible C compilers is lost. However, all gcc or
1558 LLVM versions of recent years support this extension.
1560 * Timer units gained support for a new setting RandomizedDelaySec= that
1561 allows configuring some additional randomized delay to the configured
1562 time. This is useful to spread out timer events to avoid load peaks in
1563 clusters or larger setups.
1565 * Calendar time specifications now support sub-second accuracy.
1567 * Socket units now support listening on SCTP and UDP-lite protocol
1570 * The sd-event API now comes with a full set of man pages.
1572 * Older versions of systemd contained experimental support for
1573 compressing journal files and coredumps with the LZ4 compressor that
1574 was not compatible with the lz4 binary (due to API limitations of the
1575 lz4 library). This support has been removed; only support for files
1576 compatible with the lz4 binary remains. This LZ4 logic is now
1577 officially supported and no longer considered experimental.
1579 * The dkr image import logic has been removed again from importd. dkr's
1580 micro-services focus doesn't fit into the machine image focus of
1581 importd, and quickly got out of date with the upstream dkr API.
1583 * Creation of the /run/lock/lockdev/ directory was dropped from
1584 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1585 been available for many years. If you still need this, you need to
1586 create your own tmpfiles.d config file with:
1588 d /run/lock/lockdev 0775 root lock -
1590 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1591 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1592 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1593 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1594 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1595 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1596 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1597 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1598 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1599 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1600 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1601 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1602 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1603 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1604 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1605 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1606 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1607 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1608 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1610 — Berlin, 2016-02-11
1614 * A number of properties previously only settable in unit
1615 files are now also available as properties to set when
1616 creating transient units programmatically via the bus, as it
1617 is exposed with systemd-run's --property=
1618 setting. Specifically, these are: SyslogIdentifier=,
1619 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1620 EnvironmentFile=, ReadWriteDirectories=,
1621 ReadOnlyDirectories=, InaccessibleDirectories=,
1622 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1624 * When creating transient services via the bus API it is now
1625 possible to pass in a set of file descriptors to use as
1626 STDIN/STDOUT/STDERR for the invoked process.
1628 * Slice units may now be created transiently via the bus APIs,
1629 similar to the way service and scope units may already be
1630 created transiently.
1632 * Wherever systemd expects a calendar timestamp specification
1633 (like in journalctl's --since= and --until= switches) UTC
1634 timestamps are now supported. Timestamps suffixed with "UTC"
1635 are now considered to be in Universal Time Coordinated
1636 instead of the local timezone. Also, timestamps may now
1637 optionally be specified with sub-second accuracy. Both of
1638 these additions also apply to recurring calendar event
1639 specification, such as OnCalendar= in timer units.
1641 * journalctl gained a new "--sync" switch that asks the
1642 journal daemon to write all so far unwritten log messages to
1643 disk and sync the files, before returning.
1645 * systemd-tmpfiles learned two new line types "q" and "Q" that
1646 operate like "v", but also set up a basic btrfs quota
1647 hierarchy when used on a btrfs file system with quota
1650 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1651 instead of a subvolume (even on a btrfs file system) if the
1652 root directory is a plain directory, and not a
1653 subvolume. This should simplify things with certain chroot()
1654 environments which are not aware of the concept of btrfs
1657 * systemd-detect-virt gained a new --chroot switch to detect
1658 whether execution takes place in a chroot() environment.
1660 * CPUAffinity= now takes CPU index ranges in addition to
1663 * The various memory-related resource limit settings (such as
1664 LimitAS=) now understand the usual K, M, G, ... suffixes to
1665 the base of 1024 (IEC). Similar, the time-related resource
1666 limit settings understand the usual min, h, day, ...
1669 * There's a new system.conf setting DefaultTasksMax= to
1670 control the default TasksMax= setting for services and
1671 scopes running on the system. (TasksMax= is the primary
1672 setting that exposes the "pids" cgroup controller on systemd
1673 and was introduced in the previous systemd release.) The
1674 setting now defaults to 512, which means services that are
1675 not explicitly configured otherwise will only be able to
1676 create 512 processes or threads at maximum, from this
1677 version on. Note that this means that thread- or
1678 process-heavy services might need to be reconfigured to set
1679 TasksMax= to a higher value. It is sufficient to set
1680 TasksMax= in these specific unit files to a higher value, or
1681 even "infinity". Similar, there's now a logind.conf setting
1682 UserTasksMax= that defaults to 4096 and limits the total
1683 number of processes or tasks each user may own
1684 concurrently. nspawn containers also have the TasksMax=
1685 value set by default now, to 8192. Note that all of this
1686 only has an effect if the "pids" cgroup controller is
1687 enabled in the kernel. The general benefit of these changes
1688 should be a more robust and safer system, that provides a
1689 certain amount of per-service fork() bomb protection.
1691 * systemd-nspawn gained the new --network-veth-extra= switch
1692 to define additional and arbitrarily-named virtual Ethernet
1693 links between the host and the container.
1695 * A new service execution setting PassEnvironment= has been
1696 added that allows importing select environment variables
1697 from PID1's environment block into the environment block of
1700 * Timer units gained support for a new RemainAfterElapse=
1701 setting which takes a boolean argument. It defaults to on,
1702 exposing behaviour unchanged to previous releases. If set to
1703 off, timer units are unloaded after they elapsed if they
1704 cannot elapse again. This is particularly useful for
1705 transient timer units, which shall not stay around longer
1706 than until they first elapse.
1708 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1709 default now (the kernel default is 16). This is beneficial
1710 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1711 allows substantially larger numbers of queued
1712 datagrams. This should increase the capability of systemd to
1713 parallelize boot-up, as logging and sd_notify() are unlikely
1714 to stall execution anymore. If you need to change the value
1715 from the new defaults, use the usual sysctl.d/ snippets.
1717 * The compression framing format used by the journal or
1718 coredump processing has changed to be in line with what the
1719 official LZ4 tools generate. LZ4 compression support in
1720 systemd was considered unsupported previously, as the format
1721 was not compatible with the normal tools. With this release
1722 this has changed now, and it is hence safe for downstream
1723 distributions to turn it on. While not compressing as well
1724 as the XZ, LZ4 is substantially faster, which makes
1725 it a good default choice for the compression logic in the
1726 journal and in coredump handling.
1728 * Any reference to /etc/mtab has been dropped from
1729 systemd. The file has been obsolete since a while, but
1730 systemd refused to work on systems where it was incorrectly
1731 set up (it should be a symlink or non-existent). Please make
1732 sure to update to util-linux 2.27.1 or newer in conjunction
1733 with this systemd release, which also drops any reference to
1734 /etc/mtab. If you maintain a distribution make sure that no
1735 software you package still references it, as this is a
1736 likely source of bugs. There's also a glibc bug pending,
1737 asking for removal of any reference to this obsolete file:
1739 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1741 Note that only util-linux versions built with
1742 --enable-libmount-force-mountinfo are supported.
1744 * Support for the ".snapshot" unit type has been removed. This
1745 feature turned out to be little useful and little used, and
1746 has now been removed from the core and from systemctl.
1748 * The dependency types RequiresOverridable= and
1749 RequisiteOverridable= have been removed from systemd. They
1750 have been used only very sparingly to our knowledge and
1751 other options that provide a similar effect (such as
1752 systemctl --mode=ignore-dependencies) are much more useful
1753 and commonly used. Moreover, they were only half-way
1754 implemented as the option to control behaviour regarding
1755 these dependencies was never added to systemctl. By removing
1756 these dependency types the execution engine becomes a bit
1757 simpler. Unit files that use these dependencies should be
1758 changed to use the non-Overridable dependency types
1759 instead. In fact, when parsing unit files with these
1760 options, that's what systemd will automatically convert them
1761 too, but it will also warn, asking users to fix the unit
1762 files accordingly. Removal of these dependency types should
1763 only affect a negligible number of unit files in the wild.
1765 * Behaviour of networkd's IPForward= option changed
1766 (again). It will no longer maintain a per-interface setting,
1767 but propagate one way from interfaces where this is enabled
1768 to the global kernel setting. The global setting will be
1769 enabled when requested by a network that is set up, but
1770 never be disabled again. This change was made to make sure
1771 IPv4 and IPv6 behaviour regarding packet forwarding is
1772 similar (as the Linux IPv6 stack does not support
1773 per-interface control of this setting) and to minimize
1776 * In unit files the behaviour of %u, %U, %h, %s has
1777 changed. These specifiers will now unconditionally resolve
1778 to the various user database fields of the user that the
1779 systemd instance is running as, instead of the user
1780 configured in the specific unit via User=. Note that this
1781 effectively doesn't change much, as resolving of these
1782 specifiers was already turned off in the --system instance
1783 of systemd, as we cannot do NSS lookups from PID 1. In the
1784 --user instance of systemd these specifiers where correctly
1785 resolved, but hardly made any sense, since the user instance
1786 lacks privileges to do user switches anyway, and User= is
1787 hence useless. Moreover, even in the --user instance of
1788 systemd behaviour was awkward as it would only take settings
1789 from User= assignment placed before the specifier into
1790 account. In order to unify and simplify the logic around
1791 this the specifiers will now always resolve to the
1792 credentials of the user invoking the manager (which in case
1793 of PID 1 is the root user).
1795 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1796 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1797 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1798 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1799 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1800 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1801 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1802 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1803 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1804 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1805 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1806 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1807 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1808 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1811 — Berlin, 2015-11-18
1815 * systemd now depends on util-linux v2.27. More specifically,
1816 the newly added mount monitor feature in libmount now
1817 replaces systemd's former own implementation.
1819 * libmount mandates /etc/mtab not to be regular file, and
1820 systemd now enforces this condition at early boot.
1821 /etc/mtab has been deprecated and warned about for a very
1822 long time, so systems running systemd should already have
1823 stopped having this file around as anything else than a
1824 symlink to /proc/self/mounts.
1826 * Support for the "pids" cgroup controller has been added. It
1827 allows accounting the number of tasks in a cgroup and
1828 enforcing limits on it. This adds two new setting
1829 TasksAccounting= and TasksMax= to each unit, as well as a
1830 global option DefaultTasksAccounting=.
1832 * Support for the "net_cls" cgroup controller has been added.
1833 It allows assigning a net class ID to each task in the
1834 cgroup, which can then be used in firewall rules and traffic
1835 shaping configurations. Note that the kernel netfilter net
1836 class code does not currently work reliably for ingress
1837 packets on unestablished sockets.
1839 This adds a new config directive called NetClass= to CGroup
1840 enabled units. Allowed values are positive numbers for fixed
1841 assignments and "auto" for picking a free value
1844 * 'systemctl is-system-running' now returns 'offline' if the
1845 system is not booted with systemd. This command can now be
1846 used as a substitute for 'systemd-notify --booted'.
1848 * Watchdog timeouts have been increased to 3 minutes for all
1849 in-tree service files. Apparently, disk IO issues are more
1850 frequent than we hoped, and user reported >1 minute waiting
1853 * 'machine-id-commit' functionality has been merged into
1854 'machine-id-setup --commit'. The separate binary has been
1857 * The WorkingDirectory= directive in unit files may now be set
1858 to the special value '~'. In this case, the working
1859 directory is set to the home directory of the user
1860 configured in User=.
1862 * "machinectl shell" will now open the shell in the home
1863 directory of the selected user by default.
1865 * The CrashChVT= configuration file setting is renamed to
1866 CrashChangeVT=, following our usual logic of not
1867 abbreviating unnecessarily. The old directive is still
1868 supported for compat reasons. Also, this directive now takes
1869 an integer value between 1 and 63, or a boolean value. The
1870 formerly supported '-1' value for disabling stays around for
1873 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1874 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1875 RootDirectory= properties can now be set for transient
1878 * The systemd-analyze tool gained a new "set-log-target" verb
1879 to change the logging target the system manager logs to
1880 dynamically during runtime. This is similar to how
1881 "systemd-analyze set-log-level" already changes the log
1884 * In nspawn /sys is now mounted as tmpfs, with only a selected
1885 set of subdirectories mounted in from the real sysfs. This
1886 enhances security slightly, and is useful for ensuring user
1887 namespaces work correctly.
1889 * Support for USB FunctionFS activation has been added. This
1890 allows implementation of USB gadget services that are
1891 activated as soon as they are requested, so that they don't
1892 have to run continuously, similar to classic socket
1895 * The "systemctl exit" command now optionally takes an
1896 additional parameter that sets the exit code to return from
1897 the systemd manager when exiting. This is only relevant when
1898 running the systemd user instance, or when running the
1899 system instance in a container.
1901 * sd-bus gained the new API calls sd_bus_path_encode_many()
1902 and sd_bus_path_decode_many() that allow easy encoding and
1903 decoding of multiple identifier strings inside a D-Bus
1904 object path. Another new call sd_bus_default_flush_close()
1905 has been added to flush and close per-thread default
1908 * systemd-cgtop gained support for a -M/--machine= switch to
1909 show the control groups within a certain container only.
1911 * "systemctl kill" gained support for an optional --fail
1912 switch. If specified the requested operation will fail of no
1913 processes have been killed, because the unit had no
1914 processes attached, or similar.
1916 * A new systemd.crash_reboot=1 kernel command line option has
1917 been added that triggers a reboot after crashing. This can
1918 also be set through CrashReboot= in systemd.conf.
1920 * The RuntimeDirectory= setting now understands unit
1921 specifiers like %i or %f.
1923 * A new (still internal) library API sd-ipv4acd has been added,
1924 that implements address conflict detection for IPv4. It's
1925 based on code from sd-ipv4ll, and will be useful for
1926 detecting DHCP address conflicts.
1928 * File descriptors passed during socket activation may now be
1929 named. A new API sd_listen_fds_with_names() is added to
1930 access the names. The default names may be overridden,
1931 either in the .socket file using the FileDescriptorName=
1932 parameter, or by passing FDNAME= when storing the file
1933 descriptors using sd_notify().
1935 * systemd-networkd gained support for:
1937 - Setting the IPv6 Router Advertisement settings via
1938 IPv6AcceptRouterAdvertisements= in .network files.
1940 - Configuring the HelloTimeSec=, MaxAgeSec= and
1941 ForwardDelaySec= bridge parameters in .netdev files.
1943 - Configuring PreferredSource= for static routes in
1946 * The "ask-password" framework used to query for LUKS harddisk
1947 passwords or SSL passwords during boot gained support for
1948 caching passwords in the kernel keyring, if it is
1949 available. This makes sure that the user only has to type in
1950 a passphrase once if there are multiple objects to unlock
1951 with the same one. Previously, such password caching was
1952 available only when Plymouth was used; this moves the
1953 caching logic into the systemd codebase itself. The
1954 "systemd-ask-password" utility gained a new --keyname=
1955 switch to control which kernel keyring key to use for
1956 caching a password in. This functionality is also useful for
1957 enabling display managers such as gdm to automatically
1958 unlock the user's GNOME keyring if its passphrase, the
1959 user's password and the harddisk password are the same, if
1960 gdm-autologin is used.
1962 * When downloading tar or raw images using "machinectl
1963 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1964 file is now also downloaded, if it is available and stored
1965 next to the image file.
1967 * Units of type ".socket" gained a new boolean setting
1968 Writable= which is only useful in conjunction with
1969 ListenSpecial=. If true, enables opening the specified
1970 special file in O_RDWR mode rather than O_RDONLY mode.
1972 * systemd-rfkill has been reworked to become a singleton
1973 service that is activated through /dev/rfkill on each rfkill
1974 state change and saves the settings to disk. This way,
1975 systemd-rfkill is now compatible with devices that exist
1976 only intermittendly, and even restores state if the previous
1977 system shutdown was abrupt rather than clean.
1979 * The journal daemon gained support for vacuuming old journal
1980 files controlled by the number of files that shall remain,
1981 in addition to the already existing control by size and by
1982 date. This is useful as journal interleaving performance
1983 degrades with too many separate journal files, and allows
1984 putting an effective limit on them. The new setting defaults
1985 to 100, but this may be changed by setting SystemMaxFiles=
1986 and RuntimeMaxFiles= in journald.conf. Also, the
1987 "journalctl" tool gained the new --vacuum-files= switch to
1988 manually vacuum journal files to leave only the specified
1989 number of files in place.
1991 * udev will now create /dev/disk/by-path links for ATA devices
1992 on kernels where that is supported.
1994 * Galician, Serbian, Turkish and Korean translations were added.
1996 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1997 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1998 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1999 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2000 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2001 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2002 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2003 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2004 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2005 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2006 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2007 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2008 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2009 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2010 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2011 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2012 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2013 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2015 — Berlin, 2015-10-07
2019 * The DHCP implementation of systemd-networkd gained a set of
2022 - The DHCP server now supports emitting DNS and NTP
2023 information. It may be enabled and configured via
2024 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2025 and NTP information is enabled, but no servers are
2026 configured, the corresponding uplink information (if there
2027 is any) is propagated.
2029 - Server and client now support transmission and reception
2030 of timezone information. It can be configured via the
2031 newly introduced network options UseTimezone=,
2032 EmitTimezone=, and Timezone=. Transmission of timezone
2033 information is enabled between host and containers by
2034 default now: the container will change its local timezone
2035 to what the host has set.
2037 - Lease timeouts can now be configured via
2038 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2040 - The DHCP server improved on the stability of
2041 leases. Clients are more likely to get the same lease
2042 information back, even if the server loses state.
2044 - The DHCP server supports two new configuration options to
2045 control the lease address pool metrics, PoolOffset= and
2048 * The encapsulation limit of tunnels in systemd-networkd may
2049 now be configured via 'EncapsulationLimit='. It allows
2050 modifying the maximum additional levels of encapsulation
2051 that are permitted to be prepended to a packet.
2053 * systemd now supports the concept of user buses replacing
2054 session buses, if used with dbus-1.10 (and enabled via dbus
2055 --enable-user-session). It previously only supported this on
2056 kdbus-enabled systems, and this release expands this to
2057 'dbus-daemon' systems.
2059 * systemd-networkd now supports predictable interface names
2062 * systemd now optionally supports the new Linux kernel
2063 "unified" control group hierarchy. If enabled via the kernel
2064 command-line option 'systemd.unified_cgroup_hierarchy=1',
2065 systemd will try to mount the unified cgroup hierarchy
2066 directly on /sys/fs/cgroup. If not enabled, or not
2067 available, systemd will fall back to the legacy cgroup
2068 hierarchy setup, as before. Host system and containers can
2069 mix and match legacy and unified hierarchies as they
2070 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2071 environment variable to individually select the hierarchy to
2072 use for executed containers. By default, nspawn will use the
2073 unified hierarchy for the containers if the host uses the
2074 unified hierarchy, and the legacy hierarchy otherwise.
2075 Please note that at this point the unified hierarchy is an
2076 experimental kernel feature and is likely to change in one
2077 of the next kernel releases. Therefore, it should not be
2078 enabled by default in downstream distributions yet. The
2079 minimum required kernel version for the unified hierarchy to
2080 work is 4.2. Note that when the unified hierarchy is used
2081 for the first time delegated access to controllers is
2082 safe. Because of this systemd-nspawn containers will get
2083 access to controllers now, as will systemd user
2084 sessions. This means containers and user sessions may now
2085 manage their own resources, partitioning up what the system
2088 * A new special scope unit "init.scope" has been introduced
2089 that encapsulates PID 1 of the system. It may be used to
2090 determine resource usage and enforce resource limits on PID
2091 1 itself. PID 1 hence moved out of the root of the control
2094 * The cgtop tool gained support for filtering out kernel
2095 threads when counting tasks in a control group. Also, the
2096 count of processes is now recursively summed up by
2097 default. Two options -k and --recursive= have been added to
2098 revert to old behaviour. The tool has also been updated to
2099 work correctly in containers now.
2101 * systemd-nspawn's --bind= and --bind-ro= options have been
2102 extended to allow creation of non-recursive bind mounts.
2104 * libsystemd gained two new calls sd_pid_get_cgroup() and
2105 sd_peer_get_cgroup() which return the control group path of
2106 a process or peer of a connected AF_UNIX socket. This
2107 function call is particularly useful when implementing
2108 delegated subtrees support in the control group hierarchy.
2110 * The "sd-event" event loop API of libsystemd now supports
2111 correct dequeuing of real-time signals, without losing
2114 * When systemd requests a PolicyKit decision when managing
2115 units it will now add additional fields to the request,
2116 including unit name and desired operation. This enables more
2117 powerful PolicyKit policies, that make decisions depending
2118 on these parameters.
2120 * nspawn learnt support for .nspawn settings files, that may
2121 accompany the image files or directories of containers, and
2122 may contain additional settings for the container. This is
2123 an alternative to configuring container parameters via the
2124 nspawn command line.
2126 Contributions from: Cristian Rodríguez, Daniel Mack, David
2127 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2128 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2129 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2130 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2131 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2132 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2133 Andersen, Tom Gundersen, Torstein Husebø
2135 — Berlin, 2015-09-08
2139 * machinectl gained a new verb 'shell' which opens a fresh
2140 shell on the target container or the host. It is similar to
2141 the existing 'login' command of machinectl, but spawns the
2142 shell directly without prompting for username or
2143 password. The pseudo machine '.host' now refers to the local
2144 host and is used by default. Hence, 'machinectl shell' can
2145 be used as replacement for 'su -' which spawns a session as
2146 a fresh systemd unit in a way that is fully isolated from
2147 the originating session.
2149 * systemd-networkd learned to cope with private-zone DHCP
2150 options and allows other programs to query the values.
2152 * SELinux access control when enabling/disabling units is no
2153 longer enforced with this release. The previous
2154 implementation was incorrect, and a new corrected
2155 implementation is not yet available. As unit file operations
2156 are still protected via PolicyKit and D-Bus policy this is
2157 not a security problem. Yet, distributions which care about
2158 optimal SELinux support should probably not stabilize on
2161 * sd-bus gained support for matches of type "arg0has=", that
2162 test for membership of strings in string arrays sent in bus
2165 * systemd-resolved now dumps the contents of its DNS and LLMNR
2166 caches to the logs on reception of the SIGUSR1 signal. This
2167 is useful to debug DNS behaviour.
2169 * The coredumpctl tool gained a new --directory= option to
2170 operate on journal files in a specific directory.
2172 * "systemctl reboot" and related commands gained a new
2173 "--message=" option which may be used to set a free-text
2174 wall message when shutting down or rebooting the
2175 system. This message is also logged, which is useful for
2176 figuring out the reason for a reboot or shutdown a
2179 * The "systemd-resolve-host" tool's -i switch now takes
2180 network interface numbers as alternative to interface names.
2182 * A new unit file setting for services has been introduced:
2183 UtmpMode= allows configuration of how precisely systemd
2184 handles utmp and wtmp entries for the service if this is
2185 enabled. This allows writing services that appear similar to
2186 user sessions in the output of the "w", "who", "last" and
2189 * systemd-resolved will now locally synthesize DNS resource
2190 records for the "localhost" and "gateway" domains as well as
2191 the local hostname. This should ensure that clients querying
2192 RRs via resolved will get similar results as those going via
2193 NSS, if nss-myhostname is enabled.
2195 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2196 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2197 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2198 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2199 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2200 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2201 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2202 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2203 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2204 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2205 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2206 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2208 — Berlin, 2015-08-27
2212 * The systemd-efi-boot-generator functionality was merged into
2213 systemd-gpt-auto-generator.
2215 * systemd-networkd now supports Group Policy for vxlan
2216 devices. It can be enabled via the new boolean configuration
2217 option called 'GroupPolicyExtension='.
2219 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2220 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2221 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2223 — Berlin, 2015-07-31
2227 * The python-systemd code has been removed from the systemd repository.
2228 A new repository has been created which accommodates the code from
2229 now on, and we kindly ask distributions to create a separate package
2230 for this: https://github.com/systemd/python-systemd
2232 * The systemd daemon will now reload its main configuration
2233 (/etc/systemd/system.conf) on daemon-reload.
2235 * sd-dhcp now exposes vendor specific extensions via
2236 sd_dhcp_lease_get_vendor_specific().
2238 * systemd-networkd gained a number of new configuration options.
2240 - A new boolean configuration option for TAP devices called
2241 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2242 device, thus allowing to send and receive GSO packets.
2244 - A new tunnel configuration option called 'CopyDSCP='.
2245 If enabled, the DSCP field of ip6 tunnels is copied into the
2246 decapsulated packet.
2248 - A set of boolean bridge configuration options were added.
2249 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2250 and 'UnicastFlood=' are now parsed by networkd and applied to the
2251 respective bridge link device via the respective IFLA_BRPORT_*
2254 - A new string configuration option to override the hostname sent
2255 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2256 is true, networkd will use the configured hostname instead of the
2257 system hostname when sending DHCP requests.
2259 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2260 networkd will configure the IPv6 flow-label of the tunnel device
2261 according to RFC2460.
2263 - The 'macvtap' virtual network devices are now supported, similar to
2264 the already supported 'macvlan' devices.
2266 * systemd-resolved now implements RFC5452 to improve resilience against
2267 cache poisoning. Additionally, source port randomization is enabled
2268 by default to further protect against DNS spoofing attacks.
2270 * nss-mymachines now supports translating UIDs and GIDs of running
2271 containers with user-namespaces enabled. If a container 'foo'
2272 translates a host uid 'UID' to the container uid 'TUID', then
2273 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2274 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2275 mapped as 'vg-foo-TGID'.
2277 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2278 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2279 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2280 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2281 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2282 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2283 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2284 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2285 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2286 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2288 — Berlin, 2015-07-29
2292 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2293 There are no known issues with current sysfs, and udev does not need
2294 or should be used to work around such bugs.
2296 * udev does no longer enable USB HID power management. Several reports
2297 indicate, that some devices cannot handle that setting.
2299 * The udev accelerometer helper was removed. The functionality
2300 is now fully included in iio-sensor-proxy. But this means,
2301 older iio-sensor-proxy versions will no longer provide
2302 accelerometer/orientation data with this systemd version.
2303 Please upgrade iio-sensor-proxy to version 1.0.
2305 * networkd gained a new configuration option IPv6PrivacyExtensions=
2306 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2307 for Stateless Address") on selected networks.
2309 * For the sake of fewer build-time dependencies and less code in the
2310 main repository, the python bindings are about to be removed in the
2311 next release. A new repository has been created which accommodates
2312 the code from now on, and we kindly ask distributions to create a
2313 separate package for this. The removal will take place in v223.
2315 https://github.com/systemd/python-systemd
2317 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2318 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2319 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2320 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2321 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2322 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2323 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2324 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2325 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2326 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2328 — Berlin, 2015-07-07
2332 * The sd-bus.h and sd-event.h APIs have now been declared
2333 stable and have been added to the official interface of
2334 libsystemd.so. sd-bus implements an alternative D-Bus client
2335 library, that is relatively easy to use, very efficient and
2336 supports both classic D-Bus as well as kdbus as transport
2337 backend. sd-event is a generic event loop abstraction that
2338 is built around Linux epoll, but adds features such as event
2339 prioritization or efficient timer handling. Both APIs are good
2340 choices for C programs looking for a bus and/or event loop
2341 implementation that is minimal and does not have to be
2342 portable to other kernels.
2344 * kdbus support is no longer compile-time optional. It is now
2345 always built-in. However, it can still be disabled at
2346 runtime using the kdbus=0 kernel command line setting, and
2347 that setting may be changed to default to off, by specifying
2348 --disable-kdbus at build-time. Note though that the kernel
2349 command line setting has no effect if the kdbus.ko kernel
2350 module is not installed, in which case kdbus is (obviously)
2351 also disabled. We encourage all downstream distributions to
2352 begin testing kdbus by adding it to the kernel images in the
2353 development distributions, and leaving kdbus support in
2356 * The minimal required util-linux version has been bumped to
2359 * Support for chkconfig (--enable-chkconfig) was removed in
2360 favor of calling an abstraction tool
2361 /lib/systemd/systemd-sysv-install. This needs to be
2362 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2363 in README for details.
2365 * If there's a systemd unit and a SysV init script for the
2366 same service name, and the user executes "systemctl enable"
2367 for it (or a related call), then this will now enable both
2368 (or execute the related operation on both), not just the
2371 * The libudev API documentation has been converted from gtkdoc
2374 * gudev has been removed from the systemd tree, it is now an
2377 * The systemd-cgtop tool learnt a new --raw switch to generate
2378 "raw" (machine parsable) output.
2380 * networkd's IPForwarding= .network file setting learnt the
2381 new setting "kernel", which ensures that networkd does not
2382 change the IP forwarding sysctl from the default kernel
2385 * The systemd-logind bus API now exposes a new boolean
2386 property "Docked" that reports whether logind considers the
2387 system "docked", i.e. connected to a docking station or not.
2389 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2390 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2391 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2392 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2393 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2394 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2395 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2396 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2397 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2398 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2399 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2400 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2401 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2402 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2403 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2404 Fink, Zbigniew Jędrzejewski-Szmek
2406 — Berlin, 2015-06-19
2410 * The gudev library has been extracted into a separate repository
2411 available at: https://git.gnome.org/browse/libgudev/
2412 It is now managed as part of the Gnome project. Distributions
2413 are recommended to pass --disable-gudev to systemd and use
2414 gudev from the Gnome project instead. gudev is still included
2415 in systemd, for now. It will be removed soon, though. Please
2416 also see the announcement-thread on systemd-devel:
2417 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2419 * systemd now exposes a CPUUsageNSec= property for each
2420 service unit on the bus, that contains the overall consumed
2421 CPU time of a service (the sum of what each process of the
2422 service consumed). This value is only available if
2423 CPUAccounting= is turned on for a service, and is then shown
2424 in the "systemctl status" output.
2426 * Support for configuring alternative mappings of the old SysV
2427 runlevels to systemd targets has been removed. They are now
2428 hardcoded in a way that runlevels 2, 3, 4 all map to
2429 multi-user.target and 5 to graphical.target (which
2430 previously was already the default behaviour).
2432 * The auto-mounter logic gained support for mount point
2433 expiry, using a new TimeoutIdleSec= setting in .automount
2434 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2436 * The EFI System Partition (ESP) as mounted to /boot by
2437 systemd-efi-boot-generator will now be unmounted
2438 automatically after 2 minutes of not being used. This should
2439 minimize the risk of ESP corruptions.
2441 * New /etc/fstab options x-systemd.requires= and
2442 x-systemd.requires-mounts-for= are now supported to express
2443 additional dependencies for mounts. This is useful for
2444 journalling file systems that support external journal
2445 devices or overlay file systems that require underlying file
2446 systems to be mounted.
2448 * systemd does not support direct live-upgrades (via systemctl
2449 daemon-reexec) from versions older than v44 anymore. As no
2450 distribution we are aware of shipped such old versions in a
2451 stable release this should not be problematic.
2453 * When systemd forks off a new per-connection service instance
2454 it will now set the $REMOTE_ADDR environment variable to the
2455 remote IP address, and $REMOTE_PORT environment variable to
2456 the remote IP port. This behaviour is similar to the
2457 corresponding environment variables defined by CGI.
2459 * systemd-networkd gained support for uplink failure
2460 detection. The BindCarrier= option allows binding interface
2461 configuration dynamically to the link sense of other
2462 interfaces. This is useful to achieve behaviour like in
2465 * systemd-networkd gained support for configuring the DHCP
2466 client identifier to use when requesting leases.
2468 * systemd-networkd now has a per-network UseNTP= option to
2469 configure whether NTP server information acquired via DHCP
2470 is passed on to services like systemd-timesyncd.
2472 * systemd-networkd gained support for vti6 tunnels.
2474 * Note that systemd-networkd manages the sysctl variable
2475 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2476 it is configured for since v219. The variable controls IP
2477 forwarding, and is a per-interface alternative to the global
2478 /proc/sys/net/ipv[46]/ip_forward. This setting is
2479 configurable in the IPForward= option, which defaults to
2480 "no". This means if networkd is used for an interface it is
2481 no longer sufficient to set the global sysctl option to turn
2482 on IP forwarding! Instead, the .network file option
2483 IPForward= needs to be turned on! Note that the
2484 implementation of this behaviour was broken in v219 and has
2487 * Many bonding and vxlan options are now configurable in
2490 * systemd-nspawn gained a new --property= setting to set unit
2491 properties for the container scope. This is useful for
2492 setting resource parameters (e.g. "CPUShares=500") on
2493 containers started from the command line.
2495 * systemd-nspawn gained a new --private-users= switch to make
2496 use of user namespacing available on recent Linux kernels.
2498 * systemd-nspawn may now be called as part of a shell pipeline
2499 in which case the pipes used for stdin and stdout are passed
2500 directly to the process invoked in the container, without
2501 indirection via a pseudo tty.
2503 * systemd-nspawn gained a new switch to control the UNIX
2504 signal to use when killing the init process of the container
2507 * systemd-nspawn gained a new --overlay= switch for mounting
2508 overlay file systems into the container using the new kernel
2511 * When a container image is imported via systemd-importd and
2512 the host file system is not btrfs, a loopback block device
2513 file is created in /var/lib/machines.raw with a btrfs file
2514 system inside. It is then mounted to /var/lib/machines to
2515 enable btrfs features for container management. The loopback
2516 file and btrfs file system is grown as needed when container
2517 images are imported via systemd-importd.
2519 * systemd-machined/systemd-importd gained support for btrfs
2520 quota, to enforce container disk space limits on disk. This
2521 is exposed in "machinectl set-limit".
2523 * systemd-importd now can import containers from local .tar,
2524 .raw and .qcow2 images, and export them to .tar and .raw. It
2525 can also import dkr v2 images now from the network (on top
2528 * systemd-importd gained support for verifying downloaded
2529 images with gpg2 (previously only gpg1 was supported).
2531 * systemd-machined, systemd-logind, systemd: most bus calls
2532 are now accessible to unprivileged processes via
2533 PolicyKit. Also, systemd-logind will now allow users to kill
2534 their own sessions without further privileges or
2537 * systemd-shutdownd has been removed. This service was
2538 previously responsible for implementing scheduled shutdowns
2539 as exposed in /usr/bin/shutdown's time parameter. This
2540 functionality has now been moved into systemd-logind and is
2541 accessible via a bus interface.
2543 * "systemctl reboot" gained a new switch --firmware-setup that
2544 can be used to reboot into the EFI firmware setup, if that
2545 is available. systemd-logind now exposes an API on the bus
2546 to trigger such reboots, in case graphical desktop UIs want
2547 to cover this functionality.
2549 * "systemctl enable", "systemctl disable" and "systemctl mask"
2550 now support a new "--now" switch. If specified the units
2551 that are enabled will also be started, and the ones
2552 disabled/masked also stopped.
2554 * The Gummiboot EFI boot loader tool has been merged into
2555 systemd, and renamed to "systemd-boot". The bootctl tool has been
2556 updated to support systemd-boot.
2558 * An EFI kernel stub has been added that may be used to create
2559 kernel EFI binaries that contain not only the actual kernel,
2560 but also an initrd, boot splash, command line and OS release
2561 information. This combined binary can then be signed as a
2562 single image, so that the firmware can verify it all in one
2563 step. systemd-boot has special support for EFI binaries created
2564 like this and can extract OS release information from them
2565 and show them in the boot menu. This functionality is useful
2566 to implement cryptographically verified boot schemes.
2568 * Optional support has been added to systemd-fsck to pass
2569 fsck's progress report to an AF_UNIX socket in the file
2572 * udev will no longer create device symlinks for all block
2573 devices by default. A blacklist for excluding special block
2574 devices from this logic has been turned into a whitelist
2575 that requires picking block devices explicitly that require
2578 * A new (currently still internal) API sd-device.h has been
2579 added to libsystemd. This modernized API is supposed to
2580 replace libudev eventually. In fact, already much of libudev
2581 is now just a wrapper around sd-device.h.
2583 * A new hwdb database for storing metadata about pointing
2584 stick devices has been added.
2586 * systemd-tmpfiles gained support for setting file attributes
2587 similar to the "chattr" tool with new 'h' and 'H' lines.
2589 * systemd-journald will no longer unconditionally set the
2590 btrfs NOCOW flag on new journal files. This is instead done
2591 with tmpfiles snippet using the new 'h' line type. This
2592 allows easy disabling of this logic, by masking the
2593 journal-nocow.conf tmpfiles file.
2595 * systemd-journald will now translate audit message types to
2596 human readable identifiers when writing them to the
2597 journal. This should improve readability of audit messages.
2599 * The LUKS logic gained support for the offset= and skip=
2600 options in /etc/crypttab, as previously implemented by
2603 * /usr/lib/os-release gained a new optional field VARIANT= for
2604 distributions that support multiple variants (such as a
2605 desktop edition, a server edition, ...)
2607 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2608 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2609 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2610 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2611 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2612 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2613 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2614 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2615 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2616 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2617 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2618 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2619 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2620 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2621 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2622 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2623 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2624 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2625 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2626 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2627 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2628 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2629 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2630 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2631 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2632 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2633 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2635 — Berlin, 2015-05-22
2639 * Introduce a new API "sd-hwdb.h" for querying the hardware
2640 metadata database. With this minimal interface one can query
2641 and enumerate the udev hwdb, decoupled from the old libudev
2642 library. libudev's interface for this is now only a wrapper
2643 around sd-hwdb. A new tool systemd-hwdb has been added to
2644 interface with and update the database.
2646 * When any of systemd's tools copies files (for example due to
2647 tmpfiles' C lines) a btrfs reflink will attempted first,
2648 before bytewise copying is done.
2650 * systemd-nspawn gained a new --ephemeral switch. When
2651 specified a btrfs snapshot is taken of the container's root
2652 directory, and immediately removed when the container
2653 terminates again. Thus, a container can be started whose
2654 changes never alter the container's root directory, and are
2655 lost on container termination. This switch can also be used
2656 for starting a container off the root file system of the
2657 host without affecting the host OS. This switch is only
2658 available on btrfs file systems.
2660 * systemd-nspawn gained a new --template= switch. It takes the
2661 path to a container tree to use as template for the tree
2662 specified via --directory=, should that directory be
2663 missing. This allows instantiating containers dynamically,
2664 on first run. This switch is only available on btrfs file
2667 * When a .mount unit refers to a mount point on which multiple
2668 mounts are stacked, and the .mount unit is stopped all of
2669 the stacked mount points will now be unmounted until no
2670 mount point remains.
2672 * systemd now has an explicit notion of supported and
2673 unsupported unit types. Jobs enqueued for unsupported unit
2674 types will now fail with an "unsupported" error code. More
2675 specifically .swap, .automount and .device units are not
2676 supported in containers, .busname units are not supported on
2677 non-kdbus systems. .swap and .automount are also not
2678 supported if their respective kernel compile time options
2681 * machinectl gained support for two new "copy-from" and
2682 "copy-to" commands for copying files from a running
2683 container to the host or vice versa.
2685 * machinectl gained support for a new "bind" command to bind
2686 mount host directories into local containers. This is
2687 currently only supported for nspawn containers.
2689 * networkd gained support for configuring bridge forwarding
2690 database entries (fdb) from .network files.
2692 * A new tiny daemon "systemd-importd" has been added that can
2693 download container images in tar, raw, qcow2 or dkr formats,
2694 and make them available locally in /var/lib/machines, so
2695 that they can run as nspawn containers. The daemon can GPG
2696 verify the downloads (not supported for dkr, since it has no
2697 provisions for verifying downloads). It will transparently
2698 decompress bz2, xz, gzip compressed downloads if necessary,
2699 and restore sparse files on disk. The daemon uses privilege
2700 separation to ensure the actual download logic runs with
2701 fewer privileges than the daemon itself. machinectl has
2702 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2703 make the functionality of importd available to the
2704 user. With this in place the Fedora and Ubuntu "Cloud"
2705 images can be downloaded and booted as containers unmodified
2706 (the Fedora images lack the appropriate GPG signature files
2707 currently, so they cannot be verified, but this will change
2708 soon, hopefully). Note that downloading images is currently
2709 only fully supported on btrfs.
2711 * machinectl is now able to list container images found in
2712 /var/lib/machines, along with some metadata about sizes of
2713 disk and similar. If the directory is located on btrfs and
2714 quota is enabled, this includes quota display. A new command
2715 "image-status" has been added that shows additional
2716 information about images.
2718 * machinectl is now able to clone container images
2719 efficiently, if the underlying file system (btrfs) supports
2720 it, with the new "machinectl clone" command. It also
2721 gained commands for renaming and removing images, as well as
2722 marking them read-only or read-write (supported also on
2723 legacy file systems).
2725 * networkd gained support for collecting LLDP network
2726 announcements, from hardware that supports this. This is
2727 shown in networkctl output.
2729 * systemd-run gained support for a new -t (--pty) switch for
2730 invoking a binary on a pty whose input and output is
2731 connected to the invoking terminal. This allows executing
2732 processes as system services while interactively
2733 communicating with them via the terminal. Most interestingly
2734 this is supported across container boundaries. Invoking
2735 "systemd-run -t /bin/bash" is an alternative to running a
2736 full login session, the difference being that the former
2737 will not register a session, nor go through the PAM session
2740 * tmpfiles gained support for a new "v" line type for creating
2741 btrfs subvolumes. If the underlying file system is a legacy
2742 file system, this automatically degrades to creating a
2743 normal directory. Among others /var/lib/machines is now
2744 created like this at boot, should it be missing.
2746 * The directory /var/lib/containers/ has been deprecated and
2747 been replaced by /var/lib/machines. The term "machines" has
2748 been used in the systemd context as generic term for both
2749 VMs and containers, and hence appears more appropriate for
2750 this, as the directory can also contain raw images bootable
2753 * systemd-nspawn when invoked with -M but without --directory=
2754 or --image= is now capable of searching for the container
2755 root directory, subvolume or disk image automatically, in
2756 /var/lib/machines. systemd-nspawn@.service has been updated
2757 to make use of this, thus allowing it to be used for raw
2760 * A new machines.target unit has been introduced that is
2761 supposed to group all containers/VMs invoked as services on
2762 the system. systemd-nspawn@.service has been updated to
2763 integrate with that.
2765 * machinectl gained a new "start" command, for invoking a
2766 container as a service. "machinectl start foo" is mostly
2767 equivalent to "systemctl start systemd-nspawn@foo.service",
2768 but handles escaping in a nicer way.
2770 * systemd-nspawn will now mount most of the cgroupfs tree
2771 read-only into each container, with the exception of the
2772 container's own subtree in the name=systemd hierarchy.
2774 * journald now sets the special FS_NOCOW file flag for its
2775 journal files. This should improve performance on btrfs, by
2776 avoiding heavy fragmentation when journald's write-pattern
2777 is used on COW file systems. It degrades btrfs' data
2778 integrity guarantees for the files to the same levels as for
2779 ext3/ext4 however. This should be OK though as journald does
2780 its own data integrity checks and all its objects are
2781 checksummed on disk. Also, journald should handle btrfs disk
2782 full events a lot more gracefully now, by processing SIGBUS
2783 errors, and not relying on fallocate() anymore.
2785 * When journald detects that journal files it is writing to
2786 have been deleted it will immediately start new journal
2789 * systemd now provides a way to store file descriptors
2790 per-service in PID 1. This is useful for daemons to ensure
2791 that fds they require are not lost during a daemon
2792 restart. The fds are passed to the daemon on the next
2793 invocation in the same way socket activation fds are
2794 passed. This is now used by journald to ensure that the
2795 various sockets connected to all the system's stdout/stderr
2796 are not lost when journald is restarted. File descriptors
2797 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2798 an extension to sd_notify(). Note that a limit is enforced
2799 on the number of fds a service can store in PID 1, and it
2800 defaults to 0, so that no fds may be stored, unless this is
2801 explicitly turned on.
2803 * The default TERM variable to use for units connected to a
2804 terminal, when no other value is explicitly is set is now
2805 vt220 rather than vt102. This should be fairly safe still,
2806 but allows PgUp/PgDn work.
2808 * The /etc/crypttab option header= as known from Debian is now
2811 * "loginctl user-status" and "loginctl session-status" will
2812 now show the last 10 lines of log messages of the
2813 user/session following the status output. Similar,
2814 "machinectl status" will show the last 10 log lines
2815 associated with a virtual machine or container
2816 service. (Note that this is usually not the log messages
2817 done in the VM/container itself, but simply what the
2818 container manager logs. For nspawn this includes all console
2821 * "loginctl session-status" without further argument will now
2822 show the status of the session of the caller. Similar,
2823 "lock-session", "unlock-session", "activate",
2824 "enable-linger", "disable-linger" may now be called without
2825 session/user parameter in which case they apply to the
2826 caller's session/user.
2828 * An X11 session scriptlet is now shipped that uploads
2829 $DISPLAY and $XAUTHORITY into the environment of the systemd
2830 --user daemon if a session begins. This should improve
2831 compatibility with X11 enabled applications run as systemd
2834 * Generators are now subject to masking via /etc and /run, the
2835 same way as unit files.
2837 * networkd .network files gained support for configuring
2838 per-link IPv4/IPv6 packet forwarding as well as IPv4
2839 masquerading. This is by default turned on for veth links to
2840 containers, as registered by systemd-nspawn. This means that
2841 nspawn containers run with --network-veth will now get
2842 automatic routed access to the host's networks without any
2843 further configuration or setup, as long as networkd runs on
2846 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2847 or UDP posts of a container on the host. With this in place
2848 it is possible to run containers with private veth links
2849 (--network-veth), and have their functionality exposed on
2850 the host as if their services were running directly on the
2853 * systemd-nspawn's --network-veth switch now gained a short
2854 version "-n", since with the changes above it is now truly
2855 useful out-of-the-box. The systemd-nspawn@.service has been
2856 updated to make use of it too by default.
2858 * systemd-nspawn will now maintain a per-image R/W lock, to
2859 ensure that the same image is not started more than once
2860 writable. (It's OK to run an image multiple times
2861 simultaneously in read-only mode.)
2863 * systemd-nspawn's --image= option is now capable of
2864 dissecting and booting MBR and GPT disk images that contain
2865 only a single active Linux partition. Previously it
2866 supported only GPT disk images with proper GPT type
2867 IDs. This allows running cloud images from major
2868 distributions directly with systemd-nspawn, without
2871 * In addition to collecting mouse dpi data in the udev
2872 hardware database, there's now support for collecting angle
2873 information for mouse scroll wheels. The database is
2874 supposed to guarantee similar scrolling behavior on mice
2875 that it knows about. There's also support for collecting
2876 information about Touchpad types.
2878 * udev's input_id built-in will now also collect touch screen
2879 dimension data and attach it to probed devices.
2881 * /etc/os-release gained support for a Distribution Privacy
2884 * networkd gained support for creating "ipvlan", "gretap",
2885 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2887 * systemd-tmpfiles gained support for "a" lines for setting
2890 * systemd-nspawn will now mount /tmp in the container to
2891 tmpfs, automatically.
2893 * systemd now exposes the memory.usage_in_bytes cgroup
2894 attribute and shows it for each service in the "systemctl
2895 status" output, if available.
2897 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2898 immediate reboot is triggered. This useful if shutdown is
2899 hung and is unable to complete, to expedite the
2900 operation. Note that this kind of reboot will still unmount
2901 all file systems, and hence should not result in fsck being
2904 * A .device unit for an optical block device will now be
2905 considered active only when a medium is in the drive. Also,
2906 mount units are now bound to their backing devices thus
2907 triggering automatic unmounting when devices become
2908 unavailable. With this in place systemd will now
2909 automatically unmount left-over mounts when a CD-ROM is
2910 ejected or an USB stick is yanked from the system.
2912 * networkd-wait-online now has support for waiting for
2913 specific interfaces only (with globbing), and for giving up
2914 after a configurable timeout.
2916 * networkd now exits when idle. It will be automatically
2917 restarted as soon as interfaces show up, are removed or
2918 change state. networkd will stay around as long as there is
2919 at least one DHCP state machine or similar around, that keep
2922 * networkd may now configure IPv6 link-local addressing in
2923 addition to IPv4 link-local addressing.
2925 * The IPv6 "token" for use in SLAAC may now be configured for
2926 each .network interface in networkd.
2928 * Routes configured with networkd may now be assigned a scope
2931 * networkd's [Match] sections now support globbing and lists
2932 of multiple space-separated matches per item.
2934 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2935 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2936 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2937 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2938 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2939 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2940 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2941 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2942 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2943 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2944 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2945 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2946 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2947 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2948 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2949 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2950 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2951 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2952 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2953 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2954 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2955 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2956 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2957 Hoffmann, Zbigniew Jędrzejewski-Szmek
2959 — Berlin, 2015-02-16
2963 * When querying unit file enablement status (for example via
2964 "systemctl is-enabled"), a new state "indirect" is now known
2965 which indicates that a unit might not be enabled itself, but
2966 another unit listed in its Also= setting might be.
2968 * Similar to the various existing ConditionXYZ= settings for
2969 units, there are now matching AssertXYZ= settings. While
2970 failing conditions cause a unit to be skipped, but its job
2971 to succeed, failing assertions declared like this will cause
2972 a unit start operation and its job to fail.
2974 * hostnamed now knows a new chassis type "embedded".
2976 * systemctl gained a new "edit" command. When used on a unit
2977 file, this allows extending unit files with .d/ drop-in
2978 configuration snippets or editing the full file (after
2979 copying it from /usr/lib to /etc). This will invoke the
2980 user's editor (as configured with $EDITOR), and reload the
2981 modified configuration after editing.
2983 * "systemctl status" now shows the suggested enablement state
2984 for a unit, as declared in the (usually vendor-supplied)
2985 system preset files.
2987 * nss-myhostname will now resolve the single-label host name
2988 "gateway" to the locally configured default IP routing
2989 gateways, ordered by their metrics. This assigns a stable
2990 name to the used gateways, regardless which ones are
2991 currently configured. Note that the name will only be
2992 resolved after all other name sources (if nss-myhostname is
2993 configured properly) and should hence not negatively impact
2994 systems that use the single-label host name "gateway" in
2997 * systemd-inhibit now allows filtering by mode when listing
3000 * Scope and service units gained a new "Delegate" boolean
3001 property, which, when set, allows processes running inside the
3002 unit to further partition resources. This is primarily
3003 useful for systemd user instances as well as container
3006 * journald will now pick up audit messages directly from
3007 the kernel, and log them like any other log message. The
3008 audit fields are split up and fully indexed. This means that
3009 journalctl in many ways is now a (nicer!) alternative to
3010 ausearch, the traditional audit client. Note that this
3011 implements only a minimal audit client. If you want the
3012 special audit modes like reboot-on-log-overflow, please use
3013 the traditional auditd instead, which can be used in
3014 parallel to journald.
3016 * The ConditionSecurity= unit file option now understands the
3017 special string "audit" to check whether auditing is
3020 * journalctl gained two new commands --vacuum-size= and
3021 --vacuum-time= to delete old journal files until the
3022 remaining ones take up no more than the specified size on disk,
3023 or are not older than the specified time.
3025 * A new, native PPPoE library has been added to sd-network,
3026 systemd's library of light-weight networking protocols. This
3027 library will be used in a future version of networkd to
3028 enable PPPoE communication without an external pppd daemon.
3030 * The busctl tool now understands a new "capture" verb that
3031 works similar to "monitor", but writes a packet capture
3032 trace to STDOUT that can be redirected to a file which is
3033 compatible with libcap's capture file format. This can then
3034 be loaded in Wireshark and similar tools to inspect bus
3037 * The busctl tool now understands a new "tree" verb that shows
3038 the object trees of a specific service on the bus, or of all
3041 * The busctl tool now understands a new "introspect" verb that
3042 shows all interfaces and members of objects on the bus,
3043 including their signature and values. This is particularly
3044 useful to get more information about bus objects shown by
3045 the new "busctl tree" command.
3047 * The busctl tool now understands new verbs "call",
3048 "set-property" and "get-property" for invoking bus method
3049 calls, setting and getting bus object properties in a
3052 * busctl gained a new --augment-creds= argument that controls
3053 whether the tool shall augment credential information it
3054 gets from the bus with data from /proc, in a possibly
3057 * nspawn's --link-journal= switch gained two new values
3058 "try-guest" and "try-host" that work like "guest" and
3059 "host", but do not fail if the host has no persistent
3060 journalling enabled. -j is now equivalent to
3061 --link-journal=try-guest.
3063 * macvlan network devices created by nspawn will now have
3064 stable MAC addresses.
3066 * A new SmackProcessLabel= unit setting has been added, which
3067 controls the SMACK security label processes forked off by
3068 the respective unit shall use.
3070 * If compiled with --enable-xkbcommon, systemd-localed will
3071 verify x11 keymap settings by compiling the given keymap. It
3072 will spew out warnings if the compilation fails. This
3073 requires libxkbcommon to be installed.
3075 * When a coredump is collected, a larger number of metadata
3076 fields is now collected and included in the journal records
3077 created for it. More specifically, control group membership,
3078 environment variables, memory maps, working directory,
3079 chroot directory, /proc/$PID/status, and a list of open file
3080 descriptors is now stored in the log entry.
3082 * The udev hwdb now contains DPI information for mice. For
3085 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3087 * All systemd programs that read standalone configuration
3088 files in /etc now also support a corresponding series of
3089 .conf.d configuration directories in /etc/, /run/,
3090 /usr/local/lib/, /usr/lib/, and (if configured with
3091 --enable-split-usr) /lib/. In particular, the following
3092 configuration files now have corresponding configuration
3093 directories: system.conf user.conf, logind.conf,
3094 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3095 resolved.conf, timesyncd.conf, journal-remote.conf, and
3096 journal-upload.conf. Note that distributions should use the
3097 configuration directories in /usr/lib/; the directories in
3098 /etc/ are reserved for the system administrator.
3100 * systemd-rfkill will no longer take the rfkill device name
3101 into account when storing rfkill state on disk, as the name
3102 might be dynamically assigned and not stable. Instead, the
3103 ID_PATH udev variable combined with the rfkill type (wlan,
3104 bluetooth, ...) is used.
3106 * A new service systemd-machine-id-commit.service has been
3107 added. When used on systems where /etc is read-only during
3108 boot, and /etc/machine-id is not initialized (but an empty
3109 file), this service will copy the temporary machine ID
3110 created as replacement into /etc after the system is fully
3111 booted up. This is useful for systems that are freshly
3112 installed with a non-initialized machine ID, but should get
3113 a fixed machine ID for subsequent boots.
3115 * networkd's .netdev files now provide a large set of
3116 configuration parameters for VXLAN devices. Similarly, the
3117 bridge port cost parameter is now configurable in .network
3118 files. There's also new support for configuring IP source
3119 routing. networkd .link files gained support for a new
3120 OriginalName= match that is useful to match against the
3121 original interface name the kernel assigned. .network files
3122 may include MTU= and MACAddress= fields for altering the MTU
3123 and MAC address while being connected to a specific network
3126 * The LUKS logic gained supported for configuring
3127 UUID-specific key files. There's also new support for naming
3128 LUKS device from the kernel command line, using the new
3129 luks.name= argument.
3131 * Timer units may now be transiently created via the bus API
3132 (this was previously already available for scope and service
3133 units). In addition it is now possible to create multiple
3134 transient units at the same time with a single bus call. The
3135 "systemd-run" tool has been updated to make use of this for
3136 running commands on a specified time, in at(1)-style.
3138 * tmpfiles gained support for "t" lines, for assigning
3139 extended attributes to files. Among other uses this may be
3140 used to assign SMACK labels to files.
3142 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3143 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3144 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3145 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3146 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3147 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3148 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3149 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3150 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3151 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3152 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3153 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3154 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3155 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3156 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3157 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3158 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3159 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3161 — Berlin, 2014-12-10
3165 * journalctl gained the new options -t/--identifier= to match
3166 on the syslog identifier (aka "tag"), as well as --utc to
3167 show log timestamps in the UTC timezone. journalctl now also
3168 accepts -n/--lines=all to disable line capping in a pager.
3170 * journalctl gained a new switch, --flush, that synchronously
3171 flushes logs from /run/log/journal to /var/log/journal if
3172 persistent storage is enabled. systemd-journal-flush.service
3173 now waits until the operation is complete.
3175 * Services can notify the manager before they start a reload
3176 (by sending RELOADING=1) or shutdown (by sending
3177 STOPPING=1). This allows the manager to track and show the
3178 internal state of daemons and closes a race condition when
3179 the process is still running but has closed its D-Bus
3182 * Services with Type=oneshot do not have to have any ExecStart
3185 * User units are now loaded also from
3186 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3187 /run/systemd/user directory that was already previously
3188 supported, but is under the control of the user.
3190 * Job timeouts (i.e. time-outs on the time a job that is
3191 queued stays in the run queue) can now optionally result in
3192 immediate reboot or power-off actions (JobTimeoutAction= and
3193 JobTimeoutRebootArgument=). This is useful on ".target"
3194 units, to limit the maximum time a target remains
3195 undispatched in the run queue, and to trigger an emergency
3196 operation in such a case. This is now used by default to
3197 turn off the system if boot-up (as defined by everything in
3198 basic.target) hangs and does not complete for at least
3199 15min. Also, if power-off or reboot hang for at least 30min
3200 an immediate power-off/reboot operation is triggered. This
3201 functionality is particularly useful to increase reliability
3202 on embedded devices, but also on laptops which might
3203 accidentally get powered on when carried in a backpack and
3204 whose boot stays stuck in a hard disk encryption passphrase
3207 * systemd-logind can be configured to also handle lid switch
3208 events even when the machine is docked or multiple displays
3209 are attached (HandleLidSwitchDocked= option).
3211 * A helper binary and a service have been added which can be
3212 used to resume from hibernation in the initramfs. A
3213 generator will parse the resume= option on the kernel
3214 command line to trigger resume.
3216 * A user console daemon systemd-consoled has been
3217 added. Currently, it is a preview, and will so far open a
3218 single terminal on each session of the user marked as
3219 Desktop=systemd-console.
3221 * Route metrics can be specified for DHCP routes added by
3224 * The SELinux context of socket-activated services can be set
3225 from the information provided by the networking stack
3226 (SELinuxContextFromNet= option).
3228 * Userspace firmware loading support has been removed and
3229 the minimum supported kernel version is thus bumped to 3.7.
3231 * Timeout for udev workers has been increased from 1 to 3
3232 minutes, but a warning will be printed after 1 minute to
3233 help diagnose kernel modules that take a long time to load.
3235 * Udev rules can now remove tags on devices with TAG-="foobar".
3237 * systemd's readahead implementation has been removed. In many
3238 circumstances it didn't give expected benefits even for
3239 rotational disk drives and was becoming less relevant in the
3240 age of SSDs. As none of the developers has been using
3241 rotating media anymore, and nobody stepped up to actively
3242 maintain this component of systemd it has now been removed.
3244 * Swap units can use Options= to specify discard options.
3245 Discard options specified for swaps in /etc/fstab are now
3248 * Docker containers are now detected as a separate type of
3251 * The Password Agent protocol gained support for queries where
3252 the user input is shown, useful e.g. for user names.
3253 systemd-ask-password gained a new --echo option to turn that
3256 * The default sysctl.d/ snippets will now set:
3258 net.core.default_qdisc = fq_codel
3260 This selects Fair Queuing Controlled Delay as the default
3261 queuing discipline for network interfaces. fq_codel helps
3262 fight the network bufferbloat problem. It is believed to be
3263 a good default with no tuning required for most workloads.
3264 Downstream distributions may override this choice. On 10Gbit
3265 servers that do not do forwarding, "fq" may perform better.
3266 Systems without a good clocksource should use "pfifo_fast".
3268 * If kdbus is enabled during build a new option BusPolicy= is
3269 available for service units, that allows locking all service
3270 processes into a stricter bus policy, in order to limit
3271 access to various bus services, or even hide most of them
3272 from the service's view entirely.
3274 * networkctl will now show the .network and .link file
3275 networkd has applied to a specific interface.
3277 * sd-login gained a new API call sd_session_get_desktop() to
3278 query which desktop environment has been selected for a
3281 * UNIX utmp support is now compile-time optional to support
3282 legacy-free systems.
3284 * systemctl gained two new commands "add-wants" and
3285 "add-requires" for pulling in units from specific targets
3288 * If the word "rescue" is specified on the kernel command line
3289 the system will now boot into rescue mode (aka
3290 rescue.target), which was previously available only by
3291 specifying "1" or "systemd.unit=rescue.target" on the kernel
3292 command line. This new kernel command line option nicely
3293 mirrors the already existing "emergency" kernel command line
3296 * New kernel command line options mount.usr=, mount.usrflags=,
3297 mount.usrfstype= have been added that match root=, rootflags=,
3298 rootfstype= but allow mounting a specific file system to
3301 * The $NOTIFY_SOCKET is now also passed to control processes of
3302 services, not only the main process.
3304 * This version reenables support for fsck's -l switch. This
3305 means at least version v2.25 of util-linux is required for
3306 operation, otherwise dead-locks on device nodes may
3307 occur. Again: you need to update util-linux to at least
3308 v2.25 when updating systemd to v217.
3310 * The "multi-seat-x" tool has been removed from systemd, as
3311 its functionality has been integrated into X servers 1.16,
3312 and the tool is hence redundant. It is recommended to update
3313 display managers invoking this tool to simply invoke X
3314 directly from now on, again.
3316 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3317 message flag has been added for all of systemd's PolicyKit
3318 authenticated method calls has been added. In particular
3319 this now allows optional interactive authorization via
3320 PolicyKit for many of PID1's privileged operations such as
3321 unit file enabling and disabling.
3323 * "udevadm hwdb --update" learnt a new switch "--usr" for
3324 placing the rebuilt hardware database in /usr instead of
3325 /etc. When used only hardware database entries stored in
3326 /usr will be used, and any user database entries in /etc are
3327 ignored. This functionality is useful for vendors to ship a
3328 pre-built database on systems where local configuration is
3329 unnecessary or unlikely.
3331 * Calendar time specifications in .timer units now also
3332 understand the strings "semi-annually", "quarterly" and
3333 "minutely" as shortcuts (in addition to the preexisting
3334 "anually", "hourly", ...).
3336 * systemd-tmpfiles will now correctly create files in /dev
3337 at boot which are marked for creation only at boot. It is
3338 recommended to always create static device nodes with 'c!'
3339 and 'b!', so that they are created only at boot and not
3340 overwritten at runtime.
3342 * When the watchdog logic is used for a service (WatchdogSec=)
3343 and the watchdog timeout is hit the service will now be
3344 terminated with SIGABRT (instead of just SIGTERM), in order
3345 to make sure a proper coredump and backtrace is
3346 generated. This ensures that hanging services will result in
3347 similar coredump/backtrace behaviour as services that hit a
3350 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3351 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3352 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3353 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3354 Herrmann, David Sommerseth, David Strauss, Emil Renner
3355 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3356 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3357 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3358 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3359 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3360 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3361 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3362 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3363 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3364 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3365 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3366 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3367 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3368 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3369 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3370 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3373 — Berlin, 2014-10-28
3377 * timedated no longer reads NTP implementation unit names from
3378 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3379 implementations should add a
3381 Conflicts=systemd-timesyncd.service
3383 to their unit files to take over and replace systemd's NTP
3384 default functionality.
3386 * systemd-sysusers gained a new line type "r" for configuring
3387 which UID/GID ranges to allocate system users/groups
3388 from. Lines of type "u" may now add an additional column
3389 that specifies the home directory for the system user to be
3390 created. Also, systemd-sysusers may now optionally read user
3391 information from STDIN instead of a file. This is useful for
3392 invoking it from RPM preinst scriptlets that need to create
3393 users before the first RPM file is installed since these
3394 files might need to be owned by them. A new
3395 %sysusers_create_inline RPM macro has been introduced to do
3396 just that. systemd-sysusers now updates the shadow files as
3397 well as the user/group databases, which should enhance
3398 compatibility with certain tools like grpck.
3400 * A number of bus APIs of PID 1 now optionally consult
3401 PolicyKit to permit access for otherwise unprivileged
3402 clients under certain conditions. Note that this currently
3403 doesn't support interactive authentication yet, but this is
3404 expected to be added eventually, too.
3406 * /etc/machine-info now has new fields for configuring the
3407 deployment environment of the machine, as well as the
3408 location of the machine. hostnamectl has been updated with
3409 new command to update these fields.
3411 * systemd-timesyncd has been updated to automatically acquire
3412 NTP server information from systemd-networkd, which might
3413 have been discovered via DHCP.
3415 * systemd-resolved now includes a caching DNS stub resolver
3416 and a complete LLMNR name resolution implementation. A new
3417 NSS module "nss-resolve" has been added which can be used
3418 instead of glibc's own "nss-dns" to resolve hostnames via
3419 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3420 be resolved via systemd-resolved D-Bus APIs. In contrast to
3421 the glibc internal resolver systemd-resolved is aware of
3422 multi-homed system, and keeps DNS server and caches separate
3423 and per-interface. Queries are sent simultaneously on all
3424 interfaces that have DNS servers configured, in order to
3425 properly handle VPNs and local LANs which might resolve
3426 separate sets of domain names. systemd-resolved may acquire
3427 DNS server information from systemd-networkd automatically,
3428 which in turn might have discovered them via DHCP. A tool
3429 "systemd-resolve-host" has been added that may be used to
3430 query the DNS logic in resolved. systemd-resolved implements
3431 IDNA and automatically uses IDNA or UTF-8 encoding depending
3432 on whether classic DNS or LLMNR is used as transport. In the
3433 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3434 implementation to systemd-resolved.
3436 * A new NSS module nss-mymachines has been added, that
3437 automatically resolves the names of all local registered
3438 containers to their respective IP addresses.
3440 * A new client tool "networkctl" for systemd-networkd has been
3441 added. It currently is entirely passive and will query
3442 networking configuration from udev, rtnetlink and networkd,
3443 and present it to the user in a very friendly
3444 way. Eventually, we hope to extend it to become a full
3445 control utility for networkd.
3447 * .socket units gained a new DeferAcceptSec= setting that
3448 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3449 TCP. Similarly, support for controlling TCP keep-alive
3450 settings has been added (KeepAliveTimeSec=,
3451 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3452 turning off Nagle's algorithm on TCP has been added
3455 * logind learned a new session type "web", for use in projects
3456 like Cockpit which register web clients as PAM sessions.
3458 * timer units with at least one OnCalendar= setting will now
3459 be started only after time-sync.target has been
3460 reached. This way they will not elapse before the system
3461 clock has been corrected by a local NTP client or
3462 similar. This is particular useful on RTC-less embedded
3463 machines, that come up with an invalid system clock.
3465 * systemd-nspawn's --network-veth= switch should now result in
3466 stable MAC addresses for both the outer and the inner side
3469 * systemd-nspawn gained a new --volatile= switch for running
3470 container instances with /etc or /var unpopulated.
3472 * The kdbus client code has been updated to use the new Linux
3473 3.17 memfd subsystem instead of the old kdbus-specific one.
3475 * systemd-networkd's DHCP client and server now support
3476 FORCERENEW. There are also new configuration options to
3477 configure the vendor client identifier and broadcast mode
3480 * systemd will no longer inform the kernel about the current
3481 timezone, as this is necessarily incorrect and racy as the
3482 kernel has no understanding of DST and similar
3483 concepts. This hence means FAT timestamps will be always
3484 considered UTC, similar to what Android is already
3485 doing. Also, when the RTC is configured to the local time
3486 (rather than UTC) systemd will never synchronize back to it,
3487 as this might confuse Windows at a later boot.
3489 * systemd-analyze gained a new command "verify" for offline
3490 validation of unit files.
3492 * systemd-networkd gained support for a couple of additional
3493 settings for bonding networking setups. Also, the metric for
3494 statically configured routes may now be configured. For
3495 network interfaces where this is appropriate the peer IP
3496 address may now be configured.
3498 * systemd-networkd's DHCP client will no longer request
3499 broadcasting by default, as this tripped up some networks.
3500 For hardware where broadcast is required the feature should
3501 be switched back on using RequestBroadcast=yes.
3503 * systemd-networkd will now set up IPv4LL addresses (when
3504 enabled) even if DHCP is configured successfully.
3506 * udev will now default to respect network device names given
3507 by the kernel when the kernel indicates that these are
3508 predictable. This behavior can be tweaked by changing
3509 NamePolicy= in the relevant .link file.
3511 * A new library systemd-terminal has been added that
3512 implements full TTY stream parsing and rendering. This
3513 library is supposed to be used later on for implementing a
3514 full userspace VT subsystem, replacing the current kernel
3517 * A new tool systemd-journal-upload has been added to push
3518 journal data to a remote system running
3519 systemd-journal-remote.
3521 * journald will no longer forward all local data to another
3522 running syslog daemon. This change has been made because
3523 rsyslog (which appears to be the most commonly used syslog
3524 implementation these days) no longer makes use of this, and
3525 instead pulls the data out of the journal on its own. Since
3526 forwarding the messages to a non-existent syslog server is
3527 more expensive than we assumed we have now turned this
3528 off. If you run a syslog server that is not a recent rsyslog
3529 version, you have to turn this option on again
3530 (ForwardToSyslog= in journald.conf).
3532 * journald now optionally supports the LZ4 compressor for
3533 larger journal fields. This compressor should perform much
3534 better than XZ which was the previous default.
3536 * machinectl now shows the IP addresses of local containers,
3537 if it knows them, plus the interface name of the container.
3539 * A new tool "systemd-escape" has been added that makes it
3540 easy to escape strings to build unit names and similar.
3542 * sd_notify() messages may now include a new ERRNO= field
3543 which is parsed and collected by systemd and shown among the
3544 "systemctl status" output for a service.
3546 * A new component "systemd-firstboot" has been added that
3547 queries the most basic systemd information (timezone,
3548 hostname, root password) interactively on first
3549 boot. Alternatively it may also be used to provision these
3550 things offline on OS images installed into directories.
3552 * The default sysctl.d/ snippets will now set
3554 net.ipv4.conf.default.promote_secondaries=1
3556 This has the benefit of no flushing secondary IP addresses
3557 when primary addresses are removed.
3559 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3560 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3561 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3562 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3563 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3564 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3565 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3566 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3567 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3568 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3569 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3570 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3571 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3572 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3573 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3575 — Berlin, 2014-08-19
3579 * A new tool systemd-sysusers has been added. This tool
3580 creates system users and groups in /etc/passwd and
3581 /etc/group, based on static declarative system user/group
3582 definitions in /usr/lib/sysusers.d/. This is useful to
3583 enable factory resets and volatile systems that boot up with
3584 an empty /etc directory, and thus need system users and
3585 groups created during early boot. systemd now also ships
3586 with two default sysusers.d/ files for the most basic
3587 users and groups systemd and the core operating system
3590 * A new tmpfiles snippet has been added that rebuilds the
3591 essential files in /etc on boot, should they be missing.
3593 * A directive for ensuring automatic clean-up of
3594 /var/cache/man/ has been removed from the default
3595 configuration. This line should now be shipped by the man
3596 implementation. The necessary change has been made to the
3597 man-db implementation. Note that you need to update your man
3598 implementation to one that ships this line, otherwise no
3599 automatic clean-up of /var/cache/man will take place.
3601 * A new condition ConditionNeedsUpdate= has been added that
3602 may conditionalize services to only run when /etc or /var
3603 are "older" than the vendor operating system resources in
3604 /usr. This is useful for reconstructing or updating /etc
3605 after an offline update of /usr or a factory reset, on the
3606 next reboot. Services that want to run once after such an
3607 update or reset should use this condition and order
3608 themselves before the new systemd-update-done.service, which
3609 will mark the two directories as fully updated. A number of
3610 service files have been added making use of this, to rebuild
3611 the udev hardware database, the journald message catalog and
3612 dynamic loader cache (ldconfig). The systemd-sysusers tool
3613 described above also makes use of this now. With this in
3614 place it is now possible to start up a minimal operating
3615 system with /etc empty cleanly. For more information on the
3616 concepts involved see this recent blog story:
3618 http://0pointer.de/blog/projects/stateless.html
3620 * A new system group "input" has been introduced, and all
3621 input device nodes get this group assigned. This is useful
3622 for system-level software to get access to input devices. It
3623 complements what is already done for "audio" and "video".
3625 * systemd-networkd learnt minimal DHCPv4 server support in
3626 addition to the existing DHCPv4 client support. It also
3627 learnt DHCPv6 client and IPv6 Router Solicitation client
3628 support. The DHCPv4 client gained support for static routes
3629 passed in from the server. Note that the [DHCPv4] section
3630 known in older systemd-networkd versions has been renamed to
3631 [DHCP] and is now also used by the DHCPv6 client. Existing
3632 .network files using settings of this section should be
3633 updated, though compatibility is maintained. Optionally, the
3634 client hostname may now be sent to the DHCP server.
3636 * networkd gained support for vxlan virtual networks as well
3637 as tun/tap and dummy devices.
3639 * networkd gained support for automatic allocation of address
3640 ranges for interfaces from a system-wide pool of
3641 addresses. This is useful for dynamically managing a large
3642 number of interfaces with a single network configuration
3643 file. In particular this is useful to easily assign
3644 appropriate IP addresses to the veth links of a large number
3645 of nspawn instances.
3647 * RPM macros for processing sysusers, sysctl and binfmt
3648 drop-in snippets at package installation time have been
3651 * The /etc/os-release file should now be placed in
3652 /usr/lib/os-release. The old location is automatically
3653 created as symlink. /usr/lib is the more appropriate
3654 location of this file, since it shall actually describe the
3655 vendor operating system shipped in /usr, and not the
3656 configuration stored in /etc.
3658 * .mount units gained a new boolean SloppyOptions= setting
3659 that maps to mount(8)'s -s option which enables permissive
3660 parsing of unknown mount options.
3662 * tmpfiles learnt a new "L+" directive which creates a symlink
3663 but (unlike "L") deletes a pre-existing file first, should
3664 it already exist and not already be the correct
3665 symlink. Similarly, "b+", "c+" and "p+" directives have been
3666 added as well, which create block and character devices, as
3667 well as fifos in the filesystem, possibly removing any
3668 pre-existing files of different types.
3670 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3671 'argument' field (which so far specified the source to
3672 symlink/copy the files from) is now optional. If omitted the
3673 same file os copied from /usr/share/factory/ suffixed by the
3674 full destination path. This is useful for populating /etc
3675 with essential files, by copying them from vendor defaults
3676 shipped in /usr/share/factory/etc.
3678 * A new command "systemctl preset-all" has been added that
3679 applies the service preset settings to all installed unit
3680 files. A new switch --preset-mode= has been added that
3681 controls whether only enable or only disable operations
3684 * A new command "systemctl is-system-running" has been added
3685 that allows checking the overall state of the system, for
3686 example whether it is fully up and running.
3688 * When the system boots up with an empty /etc, the equivalent
3689 to "systemctl preset-all" is executed during early boot, to
3690 make sure all default services are enabled after a factory
3693 * systemd now contains a minimal preset file that enables the
3694 most basic services systemd ships by default.
3696 * Unit files' [Install] section gained a new DefaultInstance=
3697 field for defining the default instance to create if a
3698 template unit is enabled with no instance specified.
3700 * A new passive target cryptsetup-pre.target has been added
3701 that may be used by services that need to make they run and
3702 finish before the first LUKS cryptographic device is set up.
3704 * The /dev/loop-control and /dev/btrfs-control device nodes
3705 are now owned by the "disk" group by default, opening up
3706 access to this group.
3708 * systemd-coredump will now automatically generate a
3709 stack trace of all core dumps taking place on the system,
3710 based on elfutils' libdw library. This stack trace is logged
3713 * systemd-coredump may now optionally store coredumps directly
3714 on disk (in /var/lib/systemd/coredump, possibly compressed),
3715 instead of storing them unconditionally in the journal. This
3716 mode is the new default. A new configuration file
3717 /etc/systemd/coredump.conf has been added to configure this
3718 and other parameters of systemd-coredump.
3720 * coredumpctl gained a new "info" verb to show details about a
3721 specific coredump. A new switch "-1" has also been added
3722 that makes sure to only show information about the most
3723 recent entry instead of all entries. Also, as the tool is
3724 generally useful now the "systemd-" prefix of the binary
3725 name has been removed. Distributions that want to maintain
3726 compatibility with the old name should add a symlink from
3727 the old name to the new name.
3729 * journald's SplitMode= now defaults to "uid". This makes sure
3730 that unprivileged users can access their own coredumps with
3731 coredumpctl without restrictions.
3733 * New kernel command line options "systemd.wants=" (for
3734 pulling an additional unit during boot), "systemd.mask="
3735 (for masking a specific unit for the boot), and
3736 "systemd.debug-shell" (for enabling the debug shell on tty9)
3737 have been added. This is implemented in the new generator
3738 "systemd-debug-generator".
3740 * systemd-nspawn will now by default filter a couple of
3741 syscalls for containers, among them those required for
3742 kernel module loading, direct x86 IO port access, swap
3743 management, and kexec. Most importantly though
3744 open_by_handle_at() is now prohibited for containers,
3745 closing a hole similar to a recently discussed vulnerability
3746 in docker regarding access to files on file hierarchies the
3747 container should normally not have access to. Note that, for
3748 nspawn, we generally make no security claims anyway (and
3749 this is explicitly documented in the man page), so this is
3750 just a fix for one of the most obvious problems.
3752 * A new man page file-hierarchy(7) has been added that
3753 contains a minimized, modernized version of the file system
3754 layout systemd expects, similar in style to the FHS
3755 specification or hier(5). A new tool systemd-path(1) has
3756 been added to query many of these paths for the local
3759 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3760 longer done. Since the directory now has a per-user size
3761 limit, and is cleaned on logout this appears unnecessary,
3762 in particular since this now brings the lifecycle of this
3763 directory closer in line with how IPC objects are handled.
3765 * systemd.pc now exports a number of additional directories,
3766 including $libdir (which is useful to identify the library
3767 path for the primary architecture of the system), and a
3768 couple of drop-in directories.
3770 * udev's predictable network interface names now use the dev_port
3771 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3772 distinguish between ports of the same PCI function. dev_id should
3773 only be used for ports using the same HW address, hence the need
3776 * machined has been updated to export the OS version of a
3777 container (read from /etc/os-release and
3778 /usr/lib/os-release) on the bus. This is now shown in
3779 "machinectl status" for a machine.
3781 * A new service setting RestartForceExitStatus= has been
3782 added. If configured to a set of exit signals or process
3783 return values, the service will be restarted when the main
3784 daemon process exits with any of them, regardless of the
3787 * systemctl's -H switch for connecting to remote systemd
3788 machines has been extended so that it may be used to
3789 directly connect to a specific container on the
3790 host. "systemctl -H root@foobar:waldi" will now connect as
3791 user "root" to host "foobar", and then proceed directly to
3792 the container named "waldi". Note that currently you have to
3793 authenticate as user "root" for this to work, as entering
3794 containers is a privileged operation.
3796 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3797 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3798 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3799 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3800 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3801 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3802 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3803 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3804 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3805 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3806 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3807 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3809 — Berlin, 2014-07-03
3813 * As an experimental feature, udev now tries to lock the
3814 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3815 executes events for the disk or any of its partitions.
3816 Applications like partitioning programs can lock the
3817 disk device node (flock(LOCK_EX)) and claim temporary
3818 device ownership that way; udev will entirely skip all event
3819 handling for this disk and its partitions. If the disk
3820 was opened for writing, the close will trigger a partition
3821 table rescan in udev's "watch" facility, and if needed
3822 synthesize "change" events for the disk and all its partitions.
3823 This is now unconditionally enabled, and if it turns out to
3824 cause major problems, we might turn it on only for specific
3825 devices, or might need to disable it entirely. Device Mapper
3826 devices are excluded from this logic.
3828 * We temporarily dropped the "-l" switch for fsck invocations,
3829 since they collide with the flock() logic above. util-linux
3830 upstream has been changed already to avoid this conflict,
3831 and we will readd "-l" as soon as util-linux with this
3832 change has been released.
3834 * The dependency on libattr has been removed. Since a long
3835 time, the extended attribute calls have moved to glibc, and
3836 libattr is thus unnecessary.
3838 * Virtualization detection works without privileges now. This
3839 means the systemd-detect-virt binary no longer requires
3840 CAP_SYS_PTRACE file capabilities, and our daemons can run
3841 with fewer privileges.
3843 * systemd-networkd now runs under its own "systemd-network"
3844 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3845 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3846 loses the ability to write to files owned by root this way.
3848 * Similarly, systemd-resolved now runs under its own
3849 "systemd-resolve" user with no capabilities remaining.
3851 * Similarly, systemd-bus-proxyd now runs under its own
3852 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3854 * systemd-networkd gained support for setting up "veth"
3855 virtual Ethernet devices for container connectivity, as well
3856 as GRE and VTI tunnels.
3858 * systemd-networkd will no longer automatically attempt to
3859 manually load kernel modules necessary for certain tunnel
3860 transports. Instead, it is assumed the kernel loads them
3861 automatically when required. This only works correctly on
3862 very new kernels. On older kernels, please consider adding
3863 the kernel modules to /etc/modules-load.d/ as a work-around.
3865 * The resolv.conf file systemd-resolved generates has been
3866 moved to /run/systemd/resolve/. If you have a symlink from
3867 /etc/resolv.conf, it might be necessary to correct it.
3869 * Two new service settings, ProtectHome= and ProtectSystem=,
3870 have been added. When enabled, they will make the user data
3871 (such as /home) inaccessible or read-only and the system
3872 (such as /usr) read-only, for specific services. This allows
3873 very light-weight per-service sandboxing to avoid
3874 modifications of user data or system files from
3875 services. These two new switches have been enabled for all
3876 of systemd's long-running services, where appropriate.
3878 * Socket units gained new SocketUser= and SocketGroup=
3879 settings to set the owner user and group of AF_UNIX sockets
3880 and FIFOs in the file system.
3882 * Socket units gained a new RemoveOnStop= setting. If enabled,
3883 all FIFOS and sockets in the file system will be removed
3884 when the specific socket unit is stopped.
3886 * Socket units gained a new Symlinks= setting. It takes a list
3887 of symlinks to create to file system sockets or FIFOs
3888 created by the specific Unix sockets. This is useful to
3889 manage symlinks to socket nodes with the same life-cycle as
3892 * The /dev/log socket and /dev/initctl FIFO have been moved to
3893 /run, and have been replaced by symlinks. This allows
3894 connecting to these facilities even if PrivateDevices=yes is
3895 used for a service (which makes /dev/log itself unavailable,
3896 but /run is left). This also has the benefit of ensuring
3897 that /dev only contains device nodes, directories and
3898 symlinks, and nothing else.
3900 * sd-daemon gained two new calls sd_pid_notify() and
3901 sd_pid_notifyf(). They are similar to sd_notify() and
3902 sd_notifyf(), but allow overriding of the source PID of
3903 notification messages if permissions permit this. This is
3904 useful to send notify messages on behalf of a different
3905 process (for example, the parent process). The
3906 systemd-notify tool has been updated to make use of this
3907 when sending messages (so that notification messages now
3908 originate from the shell script invoking systemd-notify and
3909 not the systemd-notify process itself. This should minimize
3910 a race where systemd fails to associate notification
3911 messages to services when the originating process already
3914 * A new "on-abnormal" setting for Restart= has been added. If
3915 set, it will result in automatic restarts on all "abnormal"
3916 reasons for a process to exit, which includes unclean
3917 signals, core dumps, timeouts and watchdog timeouts, but
3918 does not include clean and unclean exit codes or clean
3919 signals. Restart=on-abnormal is an alternative for
3920 Restart=on-failure for services that shall be able to
3921 terminate and avoid restarts on certain errors, by
3922 indicating so with an unclean exit code. Restart=on-failure
3923 or Restart=on-abnormal is now the recommended setting for
3924 all long-running services.
3926 * If the InaccessibleDirectories= service setting points to a
3927 mount point (or if there are any submounts contained within
3928 it), it is now attempted to completely unmount it, to make
3929 the file systems truly unavailable for the respective
3932 * The ReadOnlyDirectories= service setting and
3933 systemd-nspawn's --read-only parameter are now recursively
3934 applied to all submounts, too.
3936 * Mount units may now be created transiently via the bus APIs.
3938 * The support for SysV and LSB init scripts has been removed
3939 from the systemd daemon itself. Instead, it is now
3940 implemented as a generator that creates native systemd units
3941 from these scripts when needed. This enables us to remove a
3942 substantial amount of legacy code from PID 1, following the
3943 fact that many distributions only ship a very small number
3944 of LSB/SysV init scripts nowadays.
3946 * Privileged Xen (dom0) domains are not considered
3947 virtualization anymore by the virtualization detection
3948 logic. After all, they generally have unrestricted access to
3949 the hardware and usually are used to manage the unprivileged
3952 * systemd-tmpfiles gained a new "C" line type, for copying
3953 files or entire directories.
3955 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3956 lines. So far, they have been non-globbing versions of the
3957 latter, and have thus been redundant. In future, it is
3958 recommended to only use "z". "m" has hence been removed
3959 from the documentation, even though it stays supported.
3961 * A tmpfiles snippet to recreate the most basic structure in
3962 /var has been added. This is enough to create the /var/run →
3963 /run symlink and create a couple of structural
3964 directories. This allows systems to boot up with an empty or
3965 volatile /var. Of course, while with this change, the core OS
3966 now is capable with dealing with a volatile /var, not all
3967 user services are ready for it. However, we hope that sooner
3968 or later, many service daemons will be changed upstream so
3969 that they are able to automatically create their necessary
3970 directories in /var at boot, should they be missing. This is
3971 the first step to allow state-less systems that only require
3972 the vendor image for /usr to boot.
3974 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3975 empty tmpfs instance to a specific directory. This is
3976 particularly useful for making use of the automatic
3977 reconstruction of /var (see above), by passing --tmpfs=/var.
3979 * Access modes specified in tmpfiles snippets may now be
3980 prefixed with "~", which indicates that they shall be masked
3981 by whether the existing file or directory is currently
3982 writable, readable or executable at all. Also, if specified,
3983 the sgid/suid/sticky bits will be masked for all
3986 * A new passive target unit "network-pre.target" has been
3987 added which is useful for services that shall run before any
3988 network is configured, for example firewall scripts.
3990 * The "floppy" group that previously owned the /dev/fd*
3991 devices is no longer used. The "disk" group is now used
3992 instead. Distributions should probably deprecate usage of
3995 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3996 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3997 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3998 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3999 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4000 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4003 — Berlin, 2014-06-11
4007 * A new "systemd-timesyncd" daemon has been added for
4008 synchronizing the system clock across the network. It
4009 implements an SNTP client. In contrast to NTP
4010 implementations such as chrony or the NTP reference server,
4011 this only implements a client side, and does not bother with
4012 the full NTP complexity, focusing only on querying time from
4013 one remote server and synchronizing the local clock to
4014 it. Unless you intend to serve NTP to networked clients or
4015 want to connect to local hardware clocks, this simple NTP
4016 client should be more than appropriate for most
4017 installations. The daemon runs with minimal privileges, and
4018 has been hooked up with networkd to only operate when
4019 network connectivity is available. The daemon saves the
4020 current clock to disk every time a new NTP sync has been
4021 acquired, and uses this to possibly correct the system clock
4022 early at bootup, in order to accommodate for systems that
4023 lack an RTC such as the Raspberry Pi and embedded devices,
4024 and to make sure that time monotonically progresses on these
4025 systems, even if it is not always correct. To make use of
4026 this daemon, a new system user and group "systemd-timesync"
4027 needs to be created on installation of systemd.
4029 * The queue "seqnum" interface of libudev has been disabled, as
4030 it was generally incompatible with device namespacing as
4031 sequence numbers of devices go "missing" if the devices are
4032 part of a different namespace.
4034 * "systemctl list-timers" and "systemctl list-sockets" gained
4035 a --recursive switch for showing units of these types also
4036 for all local containers, similar in style to the already
4037 supported --recursive switch for "systemctl list-units".
4039 * A new RebootArgument= setting has been added for service
4040 units, which may be used to specify a kernel reboot argument
4041 to use when triggering reboots with StartLimitAction=.
4043 * A new FailureAction= setting has been added for service
4044 units which may be used to specify an operation to trigger
4045 when a service fails. This works similarly to
4046 StartLimitAction=, but unlike it, controls what is done
4047 immediately rather than only after several attempts to
4048 restart the service in question.
4050 * hostnamed got updated to also expose the kernel name,
4051 release, and version on the bus. This is useful for
4052 executing commands like hostnamectl with the -H switch.
4053 systemd-analyze makes use of this to properly display
4054 details when running non-locally.
4056 * The bootchart tool can now show cgroup information in the
4057 graphs it generates.
4059 * The CFS CPU quota cgroup attribute is now exposed for
4060 services. The new CPUQuota= switch has been added for this
4061 which takes a percentage value. Setting this will have the
4062 result that a service may never get more CPU time than the
4063 specified percentage, even if the machine is otherwise idle.
4065 * systemd-networkd learned IPIP and SIT tunnel support.
4067 * LSB init scripts exposing a dependency on $network will now
4068 get a dependency on network-online.target rather than simply
4069 network.target. This should bring LSB handling closer to
4070 what it was on SysV systems.
4072 * A new fsck.repair= kernel option has been added to control
4073 how fsck shall deal with unclean file systems at boot.
4075 * The (.ini) configuration file parser will now silently
4076 ignore sections whose name begins with "X-". This may be
4077 used to maintain application-specific extension sections in unit
4080 * machined gained a new API to query the IP addresses of
4081 registered containers. "machinectl status" has been updated
4082 to show these addresses in its output.
4084 * A new call sd_uid_get_display() has been added to the
4085 sd-login APIs for querying the "primary" session of a
4086 user. The "primary" session of the user is elected from the
4087 user's sessions and generally a graphical session is
4088 preferred over a text one.
4090 * A minimal systemd-resolved daemon has been added. It
4091 currently simply acts as a companion to systemd-networkd and
4092 manages resolv.conf based on per-interface DNS
4093 configuration, possibly supplied via DHCP. In the long run
4094 we hope to extend this into a local DNSSEC enabled DNS and
4097 * The systemd-networkd-wait-online tool is now enabled by
4098 default. It will delay network-online.target until a network
4099 connection has been configured. The tool primarily integrates
4100 with networkd, but will also make a best effort to make sense
4101 of network configuration performed in some other way.
4103 * Two new service options StartupCPUShares= and
4104 StartupBlockIOWeight= have been added that work similarly to
4105 CPUShares= and BlockIOWeight= however only apply during
4106 system startup. This is useful to prioritize certain services
4107 differently during bootup than during normal runtime.
4109 * hostnamed has been changed to prefer the statically
4110 configured hostname in /etc/hostname (unless set to
4111 'localhost' or empty) over any dynamic one supplied by
4112 dhcp. With this change, the rules for picking the hostname
4113 match more closely the rules of other configuration settings
4114 where the local administrator's configuration in /etc always
4115 overrides any other settings.
4117 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4118 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4119 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4120 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4121 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4122 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4123 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4124 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4125 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4126 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4127 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4128 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4129 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4130 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4131 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4132 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4135 — Beijing, 2014-05-28
4139 * When restoring the screen brightness at boot, stay away from
4140 the darkest setting or from the lowest 5% of the available
4141 range, depending on which is the larger value of both. This
4142 should effectively protect the user from rebooting into a
4143 black screen, should the brightness have been set to minimum
4146 * sd-login gained a new sd_machine_get_class() call to
4147 determine the class ("vm" or "container") of a machine
4148 registered with machined.
4150 * sd-login gained new calls
4151 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4152 to query the identity of the peer of a local AF_UNIX
4153 connection. They operate similarly to their sd_pid_get_xyz()
4156 * PID 1 will now maintain a system-wide system state engine
4157 with the states "starting", "running", "degraded",
4158 "maintenance", "stopping". These states are bound to system
4159 startup, normal runtime, runtime with at least one failed
4160 service, rescue/emergency mode and system shutdown. This
4161 state is shown in the "systemctl status" output when no unit
4162 name is passed. It is useful to determine system state, in
4163 particularly when doing so for many systems or containers at
4166 * A new command "list-machines" has been added to "systemctl"
4167 that lists all local OS containers and shows their system
4168 state (see above), if systemd runs inside of them.
4170 * systemctl gained a new "-r" switch to recursively enumerate
4171 units on all local containers, when used with the
4172 "list-unit" command (which is the default one that is
4173 executed when no parameters are specified).
4175 * The GPT automatic partition discovery logic will now honour
4176 two GPT partition flags: one may be set on a partition to
4177 cause it to be mounted read-only, and the other may be set
4178 on a partition to ignore it during automatic discovery.
4180 * Two new GPT type UUIDs have been added for automatic root
4181 partition discovery, for 32-bit and 64-bit ARM. This is not
4182 particularly useful for discovering the root directory on
4183 these architectures during bare-metal boots (since UEFI is
4184 not common there), but still very useful to allow booting of
4185 ARM disk images in nspawn with the -i option.
4187 * MAC addresses of interfaces created with nspawn's
4188 --network-interface= switch will now be generated from the
4189 machine name, and thus be stable between multiple invocations
4192 * logind will now automatically remove all IPC objects owned
4193 by a user if she or he fully logs out. This makes sure that
4194 users who are logged out cannot continue to consume IPC
4195 resources. This covers SysV memory, semaphores and message
4196 queues as well as POSIX shared memory and message
4197 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4198 limits. With this functionality, that is corrected. This may
4199 be turned off by using the RemoveIPC= switch of logind.conf.
4201 * The systemd-machine-id-setup and tmpfiles tools gained a
4202 --root= switch to operate on a specific root directory,
4205 * journald can now forward logged messages to the TTYs of all
4206 logged in users ("wall"). This is the default for all
4207 emergency messages now.
4209 * A new tool systemd-journal-remote has been added to stream
4210 journal log messages across the network.
4212 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4213 controller trees are mounted into it. Note that the
4214 directories mounted beneath it are not read-only. This is a
4215 security measure and is particularly useful because glibc
4216 actually includes a search logic to pick any tmpfs it can
4217 find to implement shm_open() if /dev/shm is not available
4218 (which it might very well be in namespaced setups).
4220 * machinectl gained a new "poweroff" command to cleanly power
4221 down a local OS container.
4223 * The PrivateDevices= unit file setting will now also drop the
4224 CAP_MKNOD capability from the capability bound set, and
4225 imply DevicePolicy=closed.
4227 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4228 comprehensively on all long-running systemd services where
4229 this is appropriate.
4231 * systemd-udevd will now run in a disassociated mount
4232 namespace. To mount directories from udev rules, make sure to
4233 pull in mount units via SYSTEMD_WANTS properties.
4235 * The kdbus support gained support for uploading policy into
4236 the kernel. sd-bus gained support for creating "monitoring"
4237 connections that can eavesdrop into all bus communication
4238 for debugging purposes.
4240 * Timestamps may now be specified in seconds since the UNIX
4241 epoch Jan 1st, 1970 by specifying "@" followed by the value
4244 * Native tcpwrap support in systemd has been removed. tcpwrap
4245 is old code, not really maintained anymore and has serious
4246 shortcomings, and better options such as firewalls
4247 exist. For setups that require tcpwrap usage, please
4248 consider invoking your socket-activated service via tcpd,
4249 like on traditional inetd.
4251 * A new system.conf configuration option
4252 DefaultTimerAccuracySec= has been added that controls the
4253 default AccuracySec= setting of .timer units.
4255 * Timer units gained a new WakeSystem= switch. If enabled,
4256 timers configured this way will cause the system to resume
4257 from system suspend (if the system supports that, which most
4260 * Timer units gained a new Persistent= switch. If enabled,
4261 timers configured this way will save to disk when they have
4262 been last triggered. This information is then used on next
4263 reboot to possible execute overdue timer events, that
4264 could not take place because the system was powered off.
4265 This enables simple anacron-like behaviour for timer units.
4267 * systemctl's "list-timers" will now also list the time a
4268 timer unit was last triggered in addition to the next time
4269 it will be triggered.
4271 * systemd-networkd will now assign predictable IPv4LL
4272 addresses to its local interfaces.
4274 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4275 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4276 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4277 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4278 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4279 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4280 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4281 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4284 — Berlin, 2014-03-25
4288 * A new unit file setting RestrictAddressFamilies= has been
4289 added to restrict which socket address families unit
4290 processes gain access to. This takes address family names
4291 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4292 attack surface of services via exotic protocol stacks. This
4293 is built on seccomp system call filters.
4295 * Two new unit file settings RuntimeDirectory= and
4296 RuntimeDirectoryMode= have been added that may be used to
4297 manage a per-daemon runtime directories below /run. This is
4298 an alternative for setting up directory permissions with
4299 tmpfiles snippets, and has the advantage that the runtime
4300 directory's lifetime is bound to the daemon runtime and that
4301 the daemon starts up with an empty directory each time. This
4302 is particularly useful when writing services that drop
4303 privileges using the User= or Group= setting.
4305 * The DeviceAllow= unit setting now supports globbing for
4306 matching against device group names.
4308 * The systemd configuration file system.conf gained new
4309 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4310 DefaultMemoryAccounting= to globally turn on/off accounting
4311 for specific resources (cgroups) for all units. These
4312 settings may still be overridden individually in each unit
4315 * systemd-gpt-auto-generator is now able to discover /srv and
4316 root partitions in addition to /home and swap partitions. It
4317 also supports LUKS-encrypted partitions now. With this in
4318 place, automatic discovery of partitions to mount following
4319 the Discoverable Partitions Specification
4320 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4321 is now a lot more complete. This allows booting without
4322 /etc/fstab and without root= on the kernel command line on
4323 systems prepared appropriately.
4325 * systemd-nspawn gained a new --image= switch which allows
4326 booting up disk images and Linux installations on any block
4327 device that follow the Discoverable Partitions Specification
4328 (see above). This means that installations made with
4329 appropriately updated installers may now be started and
4330 deployed using container managers, completely
4331 unmodified. (We hope that libvirt-lxc will add support for
4332 this feature soon, too.)
4334 * systemd-nspawn gained a new --network-macvlan= setting to
4335 set up a private macvlan interface for the
4336 container. Similarly, systemd-networkd gained a new
4337 Kind=macvlan setting in .netdev files.
4339 * systemd-networkd now supports configuring local addresses
4342 * A new tool systemd-network-wait-online has been added to
4343 synchronously wait for network connectivity using
4346 * The sd-bus.h bus API gained a new sd_bus_track object for
4347 tracking the life-cycle of bus peers. Note that sd-bus.h is
4348 still not a public API though (unless you specify
4349 --enable-kdbus on the configure command line, which however
4350 voids your warranty and you get no API stability guarantee).
4352 * The $XDG_RUNTIME_DIR runtime directories for each user are
4353 now individual tmpfs instances, which has the benefit of
4354 introducing separate pools for each user, with individual
4355 size limits, and thus making sure that unprivileged clients
4356 can no longer negatively impact the system or other users by
4357 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4358 RuntimeDirectorySize= has been introduced that allows
4359 controlling the default size limit for all users. It
4360 defaults to 10% of the available physical memory. This is no
4361 replacement for quotas on tmpfs though (which the kernel
4362 still does not support), as /dev/shm and /tmp are still
4363 shared resources used by both the system and unprivileged
4366 * logind will now automatically turn off automatic suspending
4367 on laptop lid close when more than one display is
4368 connected. This was previously expected to be implemented
4369 individually in desktop environments (such as GNOME),
4370 however has been added to logind now, in order to fix a
4371 boot-time race where a desktop environment might not have
4372 been started yet and thus not been able to take an inhibitor
4373 lock at the time where logind already suspends the system
4374 due to a closed lid.
4376 * logind will now wait at least 30s after each system
4377 suspend/resume cycle, and 3min after system boot before
4378 suspending the system due to a closed laptop lid. This
4379 should give USB docking stations and similar enough time to
4380 be probed and configured after system resume and boot in
4381 order to then act as suspend blocker.
4383 * systemd-run gained a new --property= setting which allows
4384 initialization of resource control properties (and others)
4385 for the created scope or service unit. Example: "systemd-run
4386 --property=BlockIOWeight=10 updatedb" may be used to run
4387 updatedb at a low block IO scheduling weight.
4389 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4390 now also work in --scope mode.
4392 * When systemd is compiled with kdbus support, basic support
4393 for enforced policies is now in place. (Note that enabling
4394 kdbus still voids your warranty and no API compatibility
4397 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4398 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4399 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4400 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4401 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4402 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4403 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4404 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4405 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4406 Zbigniew Jędrzejewski-Szmek
4408 — Berlin, 2014-03-12
4412 * systemd will now relabel /dev after loading the SMACK policy
4413 according to SMACK rules.
4415 * A new unit file option AppArmorProfile= has been added to
4416 set the AppArmor profile for the processes of a unit.
4418 * A new condition check ConditionArchitecture= has been added
4419 to conditionalize units based on the system architecture, as
4420 reported by uname()'s "machine" field.
4422 * systemd-networkd now supports matching on the system
4423 virtualization, architecture, kernel command line, host name
4426 * logind is now a lot more aggressive when suspending the
4427 machine due to a closed laptop lid. Instead of acting only
4428 on the lid close action, it will continuously watch the lid
4429 status and act on it. This is useful for laptops where the
4430 power button is on the outside of the chassis so that it can
4431 be reached without opening the lid (such as the Lenovo
4432 Yoga). On those machines, logind will now immediately
4433 re-suspend the machine if the power button has been
4434 accidentally pressed while the laptop was suspended and in a
4435 backpack or similar.
4437 * logind will now watch SW_DOCK switches and inhibit reaction
4438 to the lid switch if it is pressed. This means that logind
4439 will not suspend the machine anymore if the lid is closed
4440 and the system is docked, if the laptop supports SW_DOCK
4441 notifications via the input layer. Note that ACPI docking
4442 stations do not generate this currently. Also note that this
4443 logic is usually not fully sufficient and Desktop
4444 Environments should take a lid switch inhibitor lock when an
4445 external display is connected, as systemd will not watch
4448 * nspawn will now make use of the devices cgroup controller by
4449 default, and only permit creation of and access to the usual
4450 API device nodes like /dev/null or /dev/random, as well as
4451 access to (but not creation of) the pty devices.
4453 * We will now ship a default .network file for
4454 systemd-networkd that automatically configures DHCP for
4455 network interfaces created by nspawn's --network-veth or
4456 --network-bridge= switches.
4458 * systemd will now understand the usual M, K, G, T suffixes
4459 according to SI conventions (i.e. to the base 1000) when
4460 referring to throughput and hardware metrics. It will stay
4461 with IEC conventions (i.e. to the base 1024) for software
4462 metrics, according to what is customary according to
4463 Wikipedia. We explicitly document which base applies for
4464 each configuration option.
4466 * The DeviceAllow= setting in unit files now supports a syntax
4467 to whitelist an entire group of devices node majors at once,
4468 based on the /proc/devices listing. For example, with the
4469 string "char-pts", it is now possible to whitelist all
4470 current and future pseudo-TTYs at once.
4472 * sd-event learned a new "post" event source. Event sources of
4473 this type are triggered by the dispatching of any event
4474 source of a type that is not "post". This is useful for
4475 implementing clean-up and check event sources that are
4476 triggered by other work being done in the program.
4478 * systemd-networkd is no longer statically enabled, but uses
4479 the usual [Install] sections so that it can be
4480 enabled/disabled using systemctl. It still is enabled by
4483 * When creating a veth interface pair with systemd-nspawn, the
4484 host side will now be prefixed with "vb-" if
4485 --network-bridge= is used, and with "ve-" if --network-veth
4486 is used. This way, it is easy to distinguish these cases on
4487 the host, for example to apply different configuration to
4488 them with systemd-networkd.
4490 * The compatibility libraries for libsystemd-journal.so,
4491 libsystem-id128.so, libsystemd-login.so and
4492 libsystemd-daemon.so do not make use of IFUNC
4493 anymore. Instead, we now build libsystemd.so multiple times
4494 under these alternative names. This means that the footprint
4495 is drastically increased, but given that these are
4496 transitional compatibility libraries, this should not matter
4497 much. This change has been made necessary to support the ARM
4498 platform for these compatibility libraries, as the ARM
4499 toolchain is not really at the same level as the toolchain
4500 for other architectures like x86 and does not support
4501 IFUNC. Please make sure to use --enable-compat-libs only
4502 during a transitional period!
4504 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4505 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4506 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4507 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4508 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4509 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4510 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4511 Zbigniew Jędrzejewski-Szmek
4513 — Berlin, 2014-02-24
4517 * A new component "systemd-networkd" has been added that can
4518 be used to configure local network interfaces statically or
4519 via DHCP. It is capable of bringing up bridges, VLANs, and
4520 bonding. Currently, no hook-ups for interactive network
4521 configuration are provided. Use this for your initrd,
4522 container, embedded, or server setup if you need a simple,
4523 yet powerful, network configuration solution. This
4524 configuration subsystem is quite nifty, as it allows wildcard
4525 hotplug matching in interfaces. For example, with a single
4526 configuration snippet, you can configure that all Ethernet
4527 interfaces showing up are automatically added to a bridge,
4528 or similar. It supports link-sensing and more.
4530 * A new tool "systemd-socket-proxyd" has been added which can
4531 act as a bidirectional proxy for TCP sockets. This is
4532 useful for adding socket activation support to services that
4533 do not actually support socket activation, including virtual
4534 machines and the like.
4536 * Add a new tool to save/restore rfkill state on
4539 * Save/restore state of keyboard backlights in addition to
4540 display backlights on shutdown/boot.
4542 * udev learned a new SECLABEL{} construct to label device
4543 nodes with a specific security label when they appear. For
4544 now, only SECLABEL{selinux} is supported, but the syntax is
4545 prepared for additional security frameworks.
4547 * udev gained a new scheme to configure link-level attributes
4548 from files in /etc/systemd/network/*.link. These files can
4549 match against MAC address, device path, driver name and type,
4550 and will apply attributes like the naming policy, link speed,
4551 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4552 address assignment policy (randomized, ...).
4554 * The configuration of network interface naming rules for
4555 "permanent interface names" has changed: a new NamePolicy=
4556 setting in the [Link] section of .link files determines the
4557 priority of possible naming schemes (onboard, slot, MAC,
4558 path). The default value of this setting is determined by
4559 /usr/lib/net/links/99-default.link. Old
4560 80-net-name-slot.rules udev configuration file has been
4561 removed, so local configuration overriding this file should
4562 be adapted to override 99-default.link instead.
4564 * When the User= switch is used in a unit file, also
4565 initialize $SHELL= based on the user database entry.
4567 * systemd no longer depends on libdbus. All communication is
4568 now done with sd-bus, systemd's low-level bus library
4571 * kdbus support has been added to PID 1 itself. When kdbus is
4572 enabled, this causes PID 1 to set up the system bus and
4573 enable support for a new ".busname" unit type that
4574 encapsulates bus name activation on kdbus. It works a little
4575 bit like ".socket" units, except for bus names. A new
4576 generator has been added that converts classic dbus1 service
4577 activation files automatically into native systemd .busname
4580 * sd-bus: add a light-weight vtable implementation that allows
4581 defining objects on the bus with a simple static const
4582 vtable array of its methods, signals and properties.
4584 * systemd will not generate or install static dbus
4585 introspection data anymore to /usr/share/dbus-1/interfaces,
4586 as the precise format of these files is unclear, and
4587 nothing makes use of it.
4589 * A proxy daemon is now provided to proxy clients connecting
4590 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4591 compatibility with classic D-Bus.
4593 * A bus driver implementation has been added that supports the
4594 classic D-Bus bus driver calls on kdbus, also for
4595 compatibility purposes.
4597 * A new API "sd-event.h" has been added that implements a
4598 minimal event loop API built around epoll. It provides a
4599 couple of features that direct epoll usage is lacking:
4600 prioritization of events, scales to large numbers of timer
4601 events, per-event timer slack (accuracy), system-wide
4602 coalescing of timer events, exit handlers, watchdog
4603 supervision support using systemd's sd_notify() API, child
4606 * A new API "sd-rntl.h" has been added that provides an API
4607 around the route netlink interface of the kernel, similar in
4608 style to "sd-bus.h".
4610 * A new API "sd-dhcp-client.h" has been added that provides a
4611 small DHCPv4 client-side implementation. This is used by
4614 * There is a new kernel command line option
4615 "systemd.restore_state=0|1". When set to "0", none of the
4616 systemd tools will restore saved runtime state to hardware
4617 devices. More specifically, the rfkill and backlight states
4620 * The FsckPassNo= compatibility option in mount/service units
4621 has been removed. The fstab generator will now add the
4622 necessary dependencies automatically, and does not require
4623 PID1's support for that anymore.
4625 * journalctl gained a new switch, --list-boots, that lists
4626 recent boots with their times and boot IDs.
4628 * The various tools like systemctl, loginctl, timedatectl,
4629 busctl, systemd-run, ... have gained a new switch "-M" to
4630 connect to a specific, local OS container (as direct
4631 connection, without requiring SSH). This works on any
4632 container that is registered with machined, such as those
4633 created by libvirt-lxc or nspawn.
4635 * systemd-run and systemd-analyze also gained support for "-H"
4636 to connect to remote hosts via SSH. This is particularly
4637 useful for systemd-run because it enables queuing of jobs
4638 onto remote systems.
4640 * machinectl gained a new command "login" to open a getty
4641 login in any local container. This works with any container
4642 that is registered with machined (such as those created by
4643 libvirt-lxc or nspawn), and which runs systemd inside.
4645 * machinectl gained a new "reboot" command that may be used to
4646 trigger a reboot on a specific container that is registered
4647 with machined. This works on any container that runs an init
4648 system of some kind.
4650 * systemctl gained a new "list-timers" command to print a nice
4651 listing of installed timer units with the times they elapse
4654 * Alternative reboot() parameters may now be specified on the
4655 "systemctl reboot" command line and are passed to the
4656 reboot() system call.
4658 * systemctl gained a new --job-mode= switch to configure the
4659 mode to queue a job with. This is a more generic version of
4660 --fail, --irreversible, and --ignore-dependencies, which are
4661 still available but not advertised anymore.
4663 * /etc/systemd/system.conf gained new settings to configure
4664 various default timeouts of units, as well as the default
4665 start limit interval and burst. These may still be overridden
4668 * PID1 will now export on the bus profile data of the security
4669 policy upload process (such as the SELinux policy upload to
4672 * journald: when forwarding logs to the console, include
4673 timestamps (following the setting in
4674 /sys/module/printk/parameters/time).
4676 * OnCalendar= in timer units now understands the special
4677 strings "yearly" and "annually". (Both are equivalent)
4679 * The accuracy of timer units is now configurable with the new
4680 AccuracySec= setting. It defaults to 1min.
4682 * A new dependency type JoinsNamespaceOf= has been added that
4683 allows running two services within the same /tmp and network
4684 namespace, if PrivateNetwork= or PrivateTmp= are used.
4686 * A new command "cat" has been added to systemctl. It outputs
4687 the original unit file of a unit, and concatenates the
4688 contents of additional "drop-in" unit file snippets, so that
4689 the full configuration is shown.
4691 * systemctl now supports globbing on the various "list-xyz"
4692 commands, like "list-units" or "list-sockets", as well as on
4693 those commands which take multiple unit names.
4695 * journalctl's --unit= switch gained support for globbing.
4697 * All systemd daemons now make use of the watchdog logic so
4698 that systemd automatically notices when they hang.
4700 * If the $container_ttys environment variable is set,
4701 getty-generator will automatically spawn a getty for each
4702 listed tty. This is useful for container managers to request
4703 login gettys to be spawned on as many ttys as needed.
4705 * %h, %s, %U specifier support is not available anymore when
4706 used in unit files for PID 1. This is because NSS calls are
4707 not safe from PID 1. They stay available for --user
4708 instances of systemd, and as special case for the root user.
4710 * loginctl gained a new "--no-legend" switch to turn off output
4713 * The "sd-login.h" API gained three new calls:
4714 sd_session_is_remote(), sd_session_get_remote_user(),
4715 sd_session_get_remote_host() to query information about
4718 * The udev hardware database now also carries vendor/product
4719 information of SDIO devices.
4721 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4722 determine whether watchdog notifications are requested by
4725 * Socket-activated per-connection services now include a
4726 short description of the connection parameters in the
4729 * tmpfiles gained a new "--boot" option. When this is not used,
4730 only lines where the command character is not suffixed with
4731 "!" are executed. When this option is specified, those
4732 options are executed too. This partitions tmpfiles
4733 directives into those that can be safely executed at any
4734 time, and those which should be run only at boot (for
4735 example, a line that creates /run/nologin).
4737 * A new API "sd-resolve.h" has been added which provides a simple
4738 asynchronous wrapper around glibc NSS host name resolution
4739 calls, such as getaddrinfo(). In contrast to glibc's
4740 getaddrinfo_a(), it does not use signals. In contrast to most
4741 other asynchronous name resolution libraries, this one does
4742 not reimplement DNS, but reuses NSS, so that alternate
4743 host name resolution systems continue to work, such as mDNS,
4744 LDAP, etc. This API is based on libasyncns, but it has been
4745 cleaned up for inclusion in systemd.
4747 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4748 "sd-daemon.h" are no longer found in individual libraries
4749 libsystemd-journal.so, libsystemd-login.so,
4750 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4751 merged them into a single library, libsystemd.so, which
4752 provides all symbols. The reason for this is cyclic
4753 dependencies, as these libraries tend to use each other's
4754 symbols. So far, we have managed to workaround that by linking
4755 a copy of a good part of our code into each of these
4756 libraries again and again, which, however, makes certain
4757 things hard to do, like sharing static variables. Also, it
4758 substantially increases footprint. With this change, there
4759 is only one library for the basic APIs systemd
4760 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4761 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4762 library as well, however are subject to the --enable-kdbus
4763 switch (see below). Note that "sd-dhcp-client.h" is not part
4764 of this library (this is because it only consumes, never
4765 provides, services of/to other APIs). To make the transition
4766 easy from the separate libraries to the unified one, we
4767 provide the --enable-compat-libs compile-time switch which
4768 will generate stub libraries that are compatible with the
4769 old ones but redirect all calls to the new one.
4771 * All of the kdbus logic and the new APIs "sd-bus.h",
4772 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4773 and "sd-utf8.h" are compile-time optional via the
4774 "--enable-kdbus" switch, and they are not compiled in by
4775 default. To make use of kdbus, you have to explicitly enable
4776 the switch. Note however, that neither the kernel nor the
4777 userspace API for all of this is considered stable yet. We
4778 want to maintain the freedom to still change the APIs for
4779 now. By specifying this build-time switch, you acknowledge
4780 that you are aware of the instability of the current
4783 * Also, note that while kdbus is pretty much complete,
4784 it lacks one thing: proper policy support. This means you
4785 can build a fully working system with all features; however,
4786 it will be highly insecure. Policy support will be added in
4787 one of the next releases, at the same time that we will
4788 declare the APIs stable.
4790 * When the kernel command line argument "kdbus" is specified,
4791 systemd will automatically load the kdbus.ko kernel module. At
4792 this stage of development, it is only useful for testing kdbus
4793 and should not be used in production. Note: if "--enable-kdbus"
4794 is specified, and the kdbus.ko kernel module is available, and
4795 "kdbus" is added to the kernel command line, the entire system
4796 runs with kdbus instead of dbus-daemon, with the above mentioned
4797 problem of missing the system policy enforcement. Also a future
4798 version of kdbus.ko or a newer systemd will not be compatible with
4799 each other, and will unlikely be able to boot the machine if only
4800 one of them is updated.
4802 * systemctl gained a new "import-environment" command which
4803 uploads the caller's environment (or parts thereof) into the
4804 service manager so that it is inherited by services started
4805 by the manager. This is useful to upload variables like
4806 $DISPLAY into the user service manager.
4808 * A new PrivateDevices= switch has been added to service units
4809 which allows running a service with a namespaced /dev
4810 directory that does not contain any device nodes for
4811 physical devices. More specifically, it only includes devices
4812 such as /dev/null, /dev/urandom, and /dev/zero which are API
4815 * logind has been extended to support behaviour like VT
4816 switching on seats that do not support a VT. This makes
4817 multi-session available on seats that are not the first seat
4818 (seat0), and on systems where kernel support for VTs has
4819 been disabled at compile-time.
4821 * If a process holds a delay lock for system sleep or shutdown
4822 and fails to release it in time, we will now log its
4823 identity. This makes it easier to identify processes that
4824 cause slow suspends or power-offs.
4826 * When parsing /etc/crypttab, support for a new key-slot=
4827 option as supported by Debian is added. It allows indicating
4828 which LUKS slot to use on disk, speeding up key loading.
4830 * The sd_journald_sendv() API call has been checked and
4831 officially declared to be async-signal-safe so that it may
4832 be invoked from signal handlers for logging purposes.
4834 * Boot-time status output is now enabled automatically after a
4835 short timeout if boot does not progress, in order to give
4836 the user an indication what she or he is waiting for.
4838 * The boot-time output has been improved to show how much time
4839 remains until jobs expire.
4841 * The KillMode= switch in service units gained a new possible
4842 value "mixed". If set, and the unit is shut down, then the
4843 initial SIGTERM signal is sent only to the main daemon
4844 process, while the following SIGKILL signal is sent to
4845 all remaining processes of the service.
4847 * When a scope unit is registered, a new property "Controller"
4848 may be set. If set to a valid bus name, systemd will send a
4849 RequestStop() signal to this name when it would like to shut
4850 down the scope. This may be used to hook manager logic into
4851 the shutdown logic of scope units. Also, scope units may now
4852 be put in a special "abandoned" state, in which case the
4853 manager process which created them takes no further
4854 responsibilities for it.
4856 * When reading unit files, systemd will now verify
4857 the access mode of these files, and warn about certain
4858 suspicious combinations. This has been added to make it
4859 easier to track down packaging bugs where unit files are
4860 marked executable or world-writable.
4862 * systemd-nspawn gained a new "--setenv=" switch to set
4863 container-wide environment variables. The similar option in
4864 systemd-activate was renamed from "--environment=" to
4865 "--setenv=" for consistency.
4867 * systemd-nspawn has been updated to create a new kdbus domain
4868 for each container that is invoked, thus allowing each
4869 container to have its own set of system and user buses,
4870 independent of the host.
4872 * systemd-nspawn gained a new --drop-capability= switch to run
4873 the container with less capabilities than the default. Both
4874 --drop-capability= and --capability= now take the special
4875 string "all" for dropping or keeping all capabilities.
4877 * systemd-nspawn gained new switches for executing containers
4878 with specific SELinux labels set.
4880 * systemd-nspawn gained a new --quiet switch to not generate
4881 any additional output but the container's own console
4884 * systemd-nspawn gained a new --share-system switch to run a
4885 container without PID namespacing enabled.
4887 * systemd-nspawn gained a new --register= switch to control
4888 whether the container is registered with systemd-machined or
4889 not. This is useful for containers that do not run full
4890 OS images, but only specific apps.
4892 * systemd-nspawn gained a new --keep-unit which may be used
4893 when invoked as the only program from a service unit, and
4894 results in registration of the unit service itself in
4895 systemd-machined, instead of a newly opened scope unit.
4897 * systemd-nspawn gained a new --network-interface= switch for
4898 moving arbitrary interfaces to the container. The new
4899 --network-veth switch creates a virtual Ethernet connection
4900 between host and container. The new --network-bridge=
4901 switch then allows assigning the host side of this virtual
4902 Ethernet connection to a bridge device.
4904 * systemd-nspawn gained a new --personality= switch for
4905 setting the kernel personality for the container. This is
4906 useful when running a 32-bit container on a 64-bit host. A
4907 similar option Personality= is now also available for service
4910 * logind will now also track a "Desktop" identifier for each
4911 session which encodes the desktop environment of it. This is
4912 useful for desktop environments that want to identify
4913 multiple running sessions of itself easily.
4915 * A new SELinuxContext= setting for service units has been
4916 added that allows setting a specific SELinux execution
4917 context for a service.
4919 * Most systemd client tools will now honour $SYSTEMD_LESS for
4920 settings of the "less" pager. By default, these tools will
4921 override $LESS to allow certain operations to work, such as
4922 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4923 influence this logic.
4925 * systemd's "seccomp" hook-up has been changed to make use of
4926 the libseccomp library instead of using its own
4927 implementation. This has benefits for portability among
4930 * For usage together with SystemCallFilter=, a new
4931 SystemCallErrorNumber= setting has been introduced that
4932 allows configuration of a system error number to be returned
4933 on filtered system calls, instead of immediately killing the
4934 process. Also, SystemCallArchitectures= has been added to
4935 limit access to system calls of a particular architecture
4936 (in order to turn off support for unused secondary
4937 architectures). There is also a global
4938 SystemCallArchitectures= setting in system.conf now to turn
4939 off support for non-native system calls system-wide.
4941 * systemd requires a kernel with a working name_to_handle_at(),
4942 please see the kernel config requirements in the README file.
4944 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4945 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4946 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4947 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4948 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4949 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4950 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4951 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4952 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4953 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4954 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4955 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4956 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4957 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4958 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4959 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4960 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4961 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4962 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4963 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4964 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4965 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4966 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4967 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4969 — Berlin, 2014-02-20
4973 * logind has gained support for facilitating privileged input
4974 and drm device access for unprivileged clients. This work is
4975 useful to allow Wayland display servers (and similar
4976 programs, such as kmscon) to run under the user's ID and
4977 access input and drm devices which are normally
4978 protected. When this is used (and the kernel is new enough)
4979 logind will "mute" IO on the file descriptors passed to
4980 Wayland as long as it is in the background and "unmute" it
4981 if it returns into the foreground. This allows secure
4982 session switching without allowing background sessions to
4983 eavesdrop on input and display data. This also introduces
4984 session switching support if VT support is turned off in the
4985 kernel, and on seats that are not seat0.
4987 * A new kernel command line option luks.options= is understood
4988 now which allows specifying LUKS options for usage for LUKS
4989 encrypted partitions specified with luks.uuid=.
4991 * tmpfiles.d(5) snippets may now use specifier expansion in
4992 path names. More specifically %m, %b, %H, %v, are now
4993 replaced by the local machine id, boot id, hostname, and
4994 kernel version number.
4996 * A new tmpfiles.d(5) command "m" has been introduced which
4997 may be used to change the owner/group/access mode of a file
4998 or directory if it exists, but do nothing if it does not.
5000 * This release removes high-level support for the
5001 MemorySoftLimit= cgroup setting. The underlying kernel
5002 cgroup attribute memory.soft_limit= is currently badly
5003 designed and likely to be removed from the kernel API in its
5004 current form, hence we should not expose it for now.
5006 * The memory.use_hierarchy cgroup attribute is now enabled for
5007 all cgroups systemd creates in the memory cgroup
5008 hierarchy. This option is likely to be come the built-in
5009 default in the kernel anyway, and the non-hierarchical mode
5010 never made much sense in the intrinsically hierarchical
5013 * A new field _SYSTEMD_SLICE= is logged along with all journal
5014 messages containing the slice a message was generated
5015 from. This is useful to allow easy per-customer filtering of
5016 logs among other things.
5018 * systemd-journald will no longer adjust the group of journal
5019 files it creates to the "systemd-journal" group. Instead we
5020 rely on the journal directory to be owned by the
5021 "systemd-journal" group, and its setgid bit set, so that the
5022 kernel file system layer will automatically enforce that
5023 journal files inherit this group assignment. The reason for
5024 this change is that we cannot allow NSS look-ups from
5025 journald which would be necessary to resolve
5026 "systemd-journal" to a numeric GID, because this might
5027 create deadlocks if NSS involves synchronous queries to
5028 other daemons (such as nscd, or sssd) which in turn are
5029 logging clients of journald and might block on it, which
5030 would then dead lock. A tmpfiles.d(5) snippet included in
5031 systemd will make sure the setgid bit and group are
5032 properly set on the journal directory if it exists on every
5033 boot. However, we recommend adjusting it manually after
5034 upgrades too (or from RPM scriptlets), so that the change is
5035 not delayed until next reboot.
5037 * Backlight and random seed files in /var/lib/ have moved into
5038 the /var/lib/systemd/ directory, in order to centralize all
5039 systemd generated files in one directory.
5041 * Boot time performance measurements (as displayed by
5042 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5043 performance information if that's available to determine how
5044 much time BIOS and boot loader initialization required. With
5045 a sufficiently new BIOS you hence no longer need to boot
5046 with Gummiboot to get access to such information.
5048 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5049 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5050 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5051 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5052 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5053 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5054 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5056 — Berlin, 2013-10-02
5060 * The Restart= option for services now understands a new
5061 on-watchdog setting, which will restart the service
5062 automatically if the service stops sending out watchdog keep
5063 alive messages (as configured with WatchdogSec=).
5065 * The getty generator (which is responsible for bringing up a
5066 getty on configured serial consoles) will no longer only
5067 start a getty on the primary kernel console but on all
5068 others, too. This makes the order in which console= is
5069 specified on the kernel command line less important.
5071 * libsystemd-logind gained a new sd_session_get_vt() call to
5072 retrieve the VT number of a session.
5074 * If the option "tries=0" is set for an entry of /etc/crypttab
5075 its passphrase is queried indefinitely instead of any
5076 maximum number of tries.
5078 * If a service with a configure PID file terminates its PID
5079 file will now be removed automatically if it still exists
5080 afterwards. This should put an end to stale PID files.
5082 * systemd-run will now also take relative binary path names
5083 for execution and no longer insists on absolute paths.
5085 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5086 paths that are optionally prefixed with "-" to indicate that
5087 it should not be considered a failure if they do not exist.
5089 * journalctl -o (and similar commands) now understands a new
5090 output mode "short-precise", it is similar to "short" but
5091 shows timestamps with usec accuracy.
5093 * The option "discard" (as known from Debian) is now
5094 synonymous to "allow-discards" in /etc/crypttab. In fact,
5095 "discard" is preferred now (since it is easier to remember
5098 * Some licensing clean-ups were made, so that more code is now
5099 LGPL-2.1 licensed than before.
5101 * A minimal tool to save/restore the display backlight
5102 brightness across reboots has been added. It will store the
5103 backlight setting as late as possible at shutdown, and
5104 restore it as early as possible during reboot.
5106 * A logic to automatically discover and enable home and swap
5107 partitions on GPT disks has been added. With this in place
5108 /etc/fstab becomes optional for many setups as systemd can
5109 discover certain partitions located on the root disk
5110 automatically. Home partitions are recognized under their
5111 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5112 partitions are recognized under their GPT type ID
5113 0657fd6da4ab43c484e50933c84b4f4f.
5115 * systemd will no longer pass any environment from the kernel
5116 or initrd to system services. If you want to set an
5117 environment for all services, do so via the kernel command
5118 line systemd.setenv= assignment.
5120 * The systemd-sysctl tool no longer natively reads the file
5121 /etc/sysctl.conf. If desired, the file should be symlinked
5122 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5123 legacy support by a symlink rather than built-in code, it
5124 also makes the otherwise hidden order of application of the
5125 different files visible. (Note that this partly reverts to a
5126 pre-198 application order of sysctl knobs!)
5128 * The "systemctl set-log-level" and "systemctl dump" commands
5129 have been moved to systemd-analyze.
5131 * systemd-run learned the new --remain-after-exit switch,
5132 which causes the scope unit not to be cleaned up
5133 automatically after the process terminated.
5135 * tmpfiles learned a new --exclude-prefix= switch to exclude
5136 certain paths from operation.
5138 * journald will now automatically flush all messages to disk
5139 as soon as a message at the log level CRIT, ALERT or EMERG
5142 Contributions from: Andrew Cook, Brandon Philips, Christian
5143 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5144 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5145 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5146 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5147 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5148 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5149 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5150 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5151 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5152 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5153 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5154 William Giokas, Zbigniew Jędrzejewski-Szmek
5156 — Berlin, 2013-09-13
5160 * The documentation has been updated to cover the various new
5161 concepts introduced with 205.
5163 * Unit files now understand the new %v specifier which
5164 resolves to the kernel version string as returned by "uname
5167 * systemctl now supports filtering the unit list output by
5168 load state, active state and sub state, using the new
5171 * "systemctl status" will now show the results of the
5172 condition checks (like ConditionPathExists= and similar) of
5173 the last start attempts of the unit. They are also logged to
5176 * "journalctl -b" may now be used to look for boot output of a
5177 specific boot. Try "journalctl -b -1" for the previous boot,
5178 but the syntax is substantially more powerful.
5180 * "journalctl --show-cursor" has been added which prints the
5181 cursor string the last shown log line. This may then be used
5182 with the new "journalctl --after-cursor=" switch to continue
5183 browsing logs from that point on.
5185 * "journalctl --force" may now be used to force regeneration
5188 * Creation of "dead" device nodes has been moved from udev
5189 into kmod and tmpfiles. Previously, udev would read the kmod
5190 databases to pre-generate dead device nodes based on meta
5191 information contained in kernel modules, so that these would
5192 be auto-loaded on access rather then at boot. As this
5193 does not really have much to do with the exposing actual
5194 kernel devices to userspace this has always been slightly
5195 alien in the udev codebase. Following the new scheme kmod
5196 will now generate a runtime snippet for tmpfiles from the
5197 module meta information and it now is tmpfiles' job to the
5198 create the nodes. This also allows overriding access and
5199 other parameters for the nodes using the usual tmpfiles
5200 facilities. As side effect this allows us to remove the
5201 CAP_SYS_MKNOD capability bit from udevd entirely.
5203 * logind's device ACLs may now be applied to these "dead"
5204 devices nodes too, thus finally allowing managed access to
5205 devices such as /dev/snd/sequencer without loading the
5206 backing module right-away.
5208 * A new RPM macro has been added that may be used to apply
5209 tmpfiles configuration during package installation.
5211 * systemd-detect-virt and ConditionVirtualization= now can
5212 detect User-Mode-Linux machines (UML).
5214 * journald will now implicitly log the effective capabilities
5215 set of processes in the message metadata.
5217 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5219 * The initrd interface has been simplified (more specifically,
5220 support for passing performance data via environment
5221 variables and fsck results via files in /run has been
5222 removed). These features were non-essential, and are
5223 nowadays available in a much nicer way by having systemd in
5224 the initrd serialize its state and have the hosts systemd
5225 deserialize it again.
5227 * The udev "keymap" data files and tools to apply keyboard
5228 specific mappings of scan to key codes, and force-release
5229 scan code lists have been entirely replaced by a udev
5230 "keyboard" builtin and a hwdb data file.
5232 * systemd will now honour the kernel's "quiet" command line
5233 argument also during late shutdown, resulting in a
5234 completely silent shutdown when used.
5236 * There's now an option to control the SO_REUSEPORT socket
5237 option in .socket units.
5239 * Instance units will now automatically get a per-template
5240 subslice of system.slice unless something else is explicitly
5241 configured. For example, instances of sshd@.service will now
5242 implicitly be placed in system-sshd.slice rather than
5243 system.slice as before.
5245 * Test coverage support may now be enabled at build time.
5247 Contributions from: Dave Reisner, Frederic Crozat, Harald
5248 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5249 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5250 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5251 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5252 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5253 Giokas, Zbigniew Jędrzejewski-Szmek
5255 — Berlin, 2013-07-23
5259 * Two new unit types have been introduced:
5261 Scope units are very similar to service units, however, are
5262 created out of pre-existing processes — instead of PID 1
5263 forking off the processes. By using scope units it is
5264 possible for system services and applications to group their
5265 own child processes (worker processes) in a powerful way
5266 which then maybe used to organize them, or kill them
5267 together, or apply resource limits on them.
5269 Slice units may be used to partition system resources in an
5270 hierarchical fashion and then assign other units to them. By
5271 default there are now three slices: system.slice (for all
5272 system services), user.slice (for all user sessions),
5273 machine.slice (for VMs and containers).
5275 Slices and scopes have been introduced primarily in
5276 context of the work to move cgroup handling to a
5277 single-writer scheme, where only PID 1
5278 creates/removes/manages cgroups.
5280 * There's a new concept of "transient" units. In contrast to
5281 normal units these units are created via an API at runtime,
5282 not from configuration from disk. More specifically this
5283 means it is now possible to run arbitrary programs as
5284 independent services, with all execution parameters passed
5285 in via bus APIs rather than read from disk. Transient units
5286 make systemd substantially more dynamic then it ever was,
5287 and useful as a general batch manager.
5289 * logind has been updated to make use of scope and slice units
5290 for managing user sessions. As a user logs in he will get
5291 his own private slice unit, to which all sessions are added
5292 as scope units. We also added support for automatically
5293 adding an instance of user@.service for the user into the
5294 slice. Effectively logind will no longer create cgroup
5295 hierarchies on its own now, it will defer entirely to PID 1
5296 for this by means of scope, service and slice units. Since
5297 user sessions this way become entities managed by PID 1
5298 the output of "systemctl" is now a lot more comprehensive.
5300 * A new mini-daemon "systemd-machined" has been added which
5301 may be used by virtualization managers to register local
5302 VMs/containers. nspawn has been updated accordingly, and
5303 libvirt will be updated shortly. machined will collect a bit
5304 of meta information about the VMs/containers, and assign
5305 them their own scope unit (see above). The collected
5306 meta-data is then made available via the "machinectl" tool,
5307 and exposed in "ps" and similar tools. machined/machinectl
5308 is compile-time optional.
5310 * As discussed earlier, the low-level cgroup configuration
5311 options ControlGroup=, ControlGroupModify=,
5312 ControlGroupPersistent=, ControlGroupAttribute= have been
5313 removed. Please use high-level attribute settings instead as
5314 well as slice units.
5316 * A new bus call SetUnitProperties() has been added to alter
5317 various runtime parameters of a unit. This is primarily
5318 useful to alter cgroup parameters dynamically in a nice way,
5319 but will be extended later on to make more properties
5320 modifiable at runtime. systemctl gained a new set-properties
5321 command that wraps this call.
5323 * A new tool "systemd-run" has been added which can be used to
5324 run arbitrary command lines as transient services or scopes,
5325 while configuring a number of settings via the command
5326 line. This tool is currently very basic, however already
5327 very useful. We plan to extend this tool to even allow
5328 queuing of execution jobs with time triggers from the
5329 command line, similar in fashion to "at".
5331 * nspawn will now inform the user explicitly that kernels with
5332 audit enabled break containers, and suggest the user to turn
5335 * Support for detecting the IMA and AppArmor security
5336 frameworks with ConditionSecurity= has been added.
5338 * journalctl gained a new "-k" switch for showing only kernel
5339 messages, mimicking dmesg output; in addition to "--user"
5340 and "--system" switches for showing only user's own logs
5343 * systemd-delta can now show information about drop-in
5344 snippets extending unit files.
5346 * libsystemd-bus has been substantially updated but is still
5347 not available as public API.
5349 * systemd will now look for the "debug" argument on the kernel
5350 command line and enable debug logging, similar to what
5351 "systemd.log_level=debug" already did before.
5353 * "systemctl set-default", "systemctl get-default" has been
5354 added to configure the default.target symlink, which
5355 controls what to boot into by default.
5357 * "systemctl set-log-level" has been added as a convenient
5358 way to raise and lower systemd logging threshold.
5360 * "systemd-analyze plot" will now show the time the various
5361 generators needed for execution, as well as information
5362 about the unit file loading.
5364 * libsystemd-journal gained a new sd_journal_open_files() call
5365 for opening specific journal files. journactl also gained a
5366 new switch to expose this new functionality. Previously we
5367 only supported opening all files from a directory, or all
5368 files from the system, as opening individual files only is
5369 racy due to journal file rotation.
5371 * systemd gained the new DefaultEnvironment= setting in
5372 /etc/systemd/system.conf to set environment variables for
5375 * If a privileged process logs a journal message with the
5376 OBJECT_PID= field set, then journald will automatically
5377 augment this with additional OBJECT_UID=, OBJECT_GID=,
5378 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5379 system services want to log events about specific client
5380 processes. journactl/systemctl has been updated to make use
5381 of this information if all log messages regarding a specific
5384 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5385 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5386 Reisner, David Coppa, David King, David Strauss, Eelco
5387 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5388 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5389 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5390 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5391 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5392 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5393 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5394 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5395 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5396 Łukasz Stelmach, 장동준
5400 * The Python bindings gained some minimal support for the APIs
5401 exposed by libsystemd-logind.
5403 * ConditionSecurity= gained support for detecting SMACK. Since
5404 this condition already supports SELinux and AppArmor we only
5405 miss IMA for this. Patches welcome!
5407 Contributions from: Karol Lewandowski, Lennart Poettering,
5408 Zbigniew Jędrzejewski-Szmek
5412 * systemd-nspawn will now create /etc/resolv.conf if
5413 necessary, before bind-mounting the host's file onto it.
5415 * systemd-nspawn will now store meta information about a
5416 container on the container's cgroup as extended attribute
5417 fields, including the root directory.
5419 * The cgroup hierarchy has been reworked in many ways. All
5420 objects any of the components systemd creates in the cgroup
5421 tree are now suffixed. More specifically, user sessions are
5422 now placed in cgroups suffixed with ".session", users in
5423 cgroups suffixed with ".user", and nspawn containers in
5424 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5425 names are now escaped in a simple scheme to avoid collision
5426 of userspace object names with kernel filenames. This work
5427 is preparation for making these objects relocatable in the
5428 cgroup tree, in order to allow easy resource partitioning of
5429 these objects without causing naming conflicts.
5431 * systemctl list-dependencies gained the new switches
5432 --plain, --reverse, --after and --before.
5434 * systemd-inhibit now shows the process name of processes that
5435 have taken an inhibitor lock.
5437 * nss-myhostname will now also resolve "localhost"
5438 implicitly. This makes /etc/hosts an optional file and
5439 nicely handles that on IPv6 ::1 maps to both "localhost" and
5442 * libsystemd-logind.so gained a new call
5443 sd_get_machine_names() to enumerate running containers and
5444 VMs (currently only supported by very new libvirt and
5445 nspawn). sd_login_monitor can now be used to watch
5446 VMs/containers coming and going.
5448 * .include is not allowed recursively anymore, and only in
5449 unit files. Usually it is better to use drop-in snippets in
5450 .d/*.conf anyway, as introduced with systemd 198.
5452 * systemd-analyze gained a new "critical-chain" command that
5453 determines the slowest chain of units run during system
5454 boot-up. It is very useful for tracking down where
5455 optimizing boot time is the most beneficial.
5457 * systemd will no longer allow manipulating service paths in
5458 the name=systemd:/system cgroup tree using ControlGroup= in
5459 units. (But is still fine with it in all other dirs.)
5461 * There's a new systemd-nspawn@.service service file that may
5462 be used to easily run nspawn containers as system
5463 services. With the container's root directory in
5464 /var/lib/container/foobar it is now sufficient to run
5465 "systemctl start systemd-nspawn@foobar.service" to boot it.
5467 * systemd-cgls gained a new parameter "--machine" to list only
5468 the processes within a certain container.
5470 * ConditionSecurity= now can check for "apparmor". We still
5471 are lacking checks for SMACK and IMA for this condition
5472 check though. Patches welcome!
5474 * A new configuration file /etc/systemd/sleep.conf has been
5475 added that may be used to configure which kernel operation
5476 systemd is supposed to execute when "suspend", "hibernate"
5477 or "hybrid-sleep" is requested. This makes the new kernel
5478 "freeze" state accessible to the user.
5480 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5481 the passed argument if applicable.
5483 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5484 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5485 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5486 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5487 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5488 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5489 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5494 * The output of 'systemctl list-jobs' got some polishing. The
5495 '--type=' argument may now be passed more than once. A new
5496 command 'systemctl list-sockets' has been added which shows
5497 a list of kernel sockets systemd is listening on with the
5498 socket units they belong to, plus the units these socket
5501 * The experimental libsystemd-bus library got substantial
5502 updates to work in conjunction with the (also experimental)
5503 kdbus kernel project. It works well enough to exchange
5504 messages with some sophistication. Note that kdbus is not
5505 ready yet, and the library is mostly an elaborate test case
5506 for now, and not installable.
5508 * systemd gained a new unit 'systemd-static-nodes.service'
5509 that generates static device nodes earlier during boot, and
5510 can run in conjunction with udev.
5512 * libsystemd-login gained a new call sd_pid_get_user_unit()
5513 to retrieve the user systemd unit a process is running
5514 in. This is useful for systems where systemd is used as
5517 * systemd-nspawn now places all containers in the new /machine
5518 top-level cgroup directory in the name=systemd
5519 hierarchy. libvirt will soon do the same, so that we get a
5520 uniform separation of /system, /user and /machine for system
5521 services, user processes and containers/virtual
5522 machines. This new cgroup hierarchy is also useful to stick
5523 stable names to specific container instances, which can be
5524 recognized later this way (this name may be controlled
5525 via systemd-nspawn's new -M switch). libsystemd-login also
5526 gained a new call sd_pid_get_machine_name() to retrieve the
5527 name of the container/VM a specific process belongs to.
5529 * bootchart can now store its data in the journal.
5531 * libsystemd-journal gained a new call
5532 sd_journal_add_conjunction() for AND expressions to the
5533 matching logic. This can be used to express more complex
5534 logical expressions.
5536 * journactl can now take multiple --unit= and --user-unit=
5539 * The cryptsetup logic now understands the "luks.key=" kernel
5540 command line switch for specifying a file to read the
5541 decryption key from. Also, if a configured key file is not
5542 found the tool will now automatically fall back to prompting
5545 * Python systemd.journal module was updated to wrap recently
5546 added functions from libsystemd-journal. The interface was
5547 changed to bring the low level interface in s.j._Reader
5548 closer to the C API, and the high level interface in
5549 s.j.Reader was updated to wrap and convert all data about
5552 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5553 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5554 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5555 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5556 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5557 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5561 * journalctl --update-catalog now understands a new --root=
5562 option to operate on catalogs found in a different root
5565 * During shutdown after systemd has terminated all running
5566 services a final killing loop kills all remaining left-over
5567 processes. We will now print the name of these processes
5568 when we send SIGKILL to them, since this usually indicates a
5571 * If /etc/crypttab refers to password files stored on
5572 configured mount points automatic dependencies will now be
5573 generated to ensure the specific mount is established first
5574 before the key file is attempted to be read.
5576 * 'systemctl status' will now show information about the
5577 network sockets a socket unit is listening on.
5579 * 'systemctl status' will also shown information about any
5580 drop-in configuration file for units. (Drop-In configuration
5581 files in this context are files such as
5582 /etc/systemd/systemd/foobar.service.d/*.conf)
5584 * systemd-cgtop now optionally shows summed up CPU times of
5585 cgroups. Press '%' while running cgtop to switch between
5586 percentage and absolute mode. This is useful to determine
5587 which cgroups use up the most CPU time over the entire
5588 runtime of the system. systemd-cgtop has also been updated
5589 to be 'pipeable' for processing with further shell tools.
5591 * 'hostnamectl set-hostname' will now allow setting of FQDN
5594 * The formatting and parsing of time span values has been
5595 changed. The parser now understands fractional expressions
5596 such as "5.5h". The formatter will now output fractional
5597 expressions for all time spans under 1min, i.e. "5.123456s"
5598 rather than "5s 123ms 456us". For time spans under 1s
5599 millisecond values are shown, for those under 1ms
5600 microsecond values are shown. This should greatly improve
5601 all time-related output of systemd.
5603 * libsystemd-login and libsystemd-journal gained new
5604 functions for querying the poll() events mask and poll()
5605 timeout value for integration into arbitrary event
5608 * localectl gained the ability to list available X11 keymaps
5609 (models, layouts, variants, options).
5611 * 'systemd-analyze dot' gained the ability to filter for
5612 specific units via shell-style globs, to create smaller,
5613 more useful graphs. I.e. it is now possible to create simple
5614 graphs of all the dependencies between only target units, or
5615 of all units that Avahi has dependencies with.
5617 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5618 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5619 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5620 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5621 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5622 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5623 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5627 * The boot-time readahead implementation for rotating media
5628 will now read the read-ahead data in multiple passes which
5629 consist of all read requests made in equidistant time
5630 intervals. This means instead of strictly reading read-ahead
5631 data in its physical order on disk we now try to find a
5632 middle ground between physical and access time order.
5634 * /etc/os-release files gained a new BUILD_ID= field for usage
5635 on operating systems that provide continuous builds of OS
5638 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5639 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5640 William Douglas, Zbigniew Jędrzejewski-Szmek
5644 * systemd-python gained an API exposing libsystemd-daemon.
5646 * The SMACK setup logic gained support for uploading CIPSO
5649 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5650 ReadOnlyDirectories= and InaccessibleDirectories= has
5651 changed. The private /tmp and /var/tmp directories are now
5652 shared by all processes of a service (which means
5653 ExecStartPre= may now leave data in /tmp that ExecStart= of
5654 the same service can still access). When a service is
5655 stopped its temporary directories are immediately deleted
5656 (normal clean-up with tmpfiles is still done in addition to
5659 * By default, systemd will now set a couple of sysctl
5660 variables in the kernel: the safe sysrq options are turned
5661 on, IP route verification is turned on, and source routing
5662 disabled. The recently added hardlink and softlink
5663 protection of the kernel is turned on. These settings should
5664 be reasonably safe, and good defaults for all new systems.
5666 * The predictable network naming logic may now be turned off
5667 with a new kernel command line switch: net.ifnames=0.
5669 * A new libsystemd-bus module has been added that implements a
5670 pretty complete D-Bus client library. For details see:
5672 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5674 * journald will now explicitly flush the journal files to disk
5675 at the latest 5min after each write. The file will then also
5676 be marked offline until the next write. This should increase
5677 reliability in case of a crash. The synchronization delay
5678 can be configured via SyncIntervalSec= in journald.conf.
5680 * There's a new remote-fs-setup.target unit that can be used
5681 to pull in specific services when at least one remote file
5682 system is to be mounted.
5684 * There are new targets timers.target and paths.target as
5685 canonical targets to pull user timer and path units in
5686 from. This complements sockets.target with a similar
5687 purpose for socket units.
5689 * libudev gained a new call udev_device_set_attribute_value()
5690 to set sysfs attributes of a device.
5692 * The udev daemon now sets the default number of worker
5693 processes executed in parallel based on the number of available
5694 CPUs instead of the amount of available RAM. This is supposed
5695 to provide a more reliable default and limit a too aggressive
5696 parallelism for setups with 1000s of devices connected.
5698 Contributions from: Auke Kok, Colin Walters, Cristian
5699 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5700 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5701 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5702 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5703 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5704 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5705 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5706 Zbigniew Jędrzejewski-Szmek
5710 * Configuration of unit files may now be extended via drop-in
5711 files without having to edit/override the unit files
5712 themselves. More specifically, if the administrator wants to
5713 change one value for a service file foobar.service he can
5714 now do so by dropping in a configuration snippet into
5715 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5716 will load all these snippets and apply them on top of the
5717 main unit configuration file, possibly extending or
5718 overriding its settings. Using these drop-in snippets is
5719 generally nicer than the two earlier options for changing
5720 unit files locally: copying the files from
5721 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5722 them there; or creating a new file in /etc/systemd/system/
5723 that incorporates the original one via ".include". Drop-in
5724 snippets into these .d/ directories can be placed in any
5725 directory systemd looks for units in, and the usual
5726 overriding semantics between /usr/lib, /etc and /run apply
5729 * Most unit file settings which take lists of items can now be
5730 reset by assigning the empty string to them. For example,
5731 normally, settings such as Environment=FOO=BAR append a new
5732 environment variable assignment to the environment block,
5733 each time they are used. By assigning Environment= the empty
5734 string the environment block can be reset to empty. This is
5735 particularly useful with the .d/*.conf drop-in snippets
5736 mentioned above, since this adds the ability to reset list
5737 settings from vendor unit files via these drop-ins.
5739 * systemctl gained a new "list-dependencies" command for
5740 listing the dependencies of a unit recursively.
5742 * Inhibitors are now honored and listed by "systemctl
5743 suspend", "systemctl poweroff" (and similar) too, not only
5744 GNOME. These commands will also list active sessions by
5747 * Resource limits (as exposed by the various control group
5748 controllers) can now be controlled dynamically at runtime
5749 for all units. More specifically, you can now use a command
5750 like "systemctl set-cgroup-attr foobar.service cpu.shares
5751 2000" to alter the CPU shares a specific service gets. These
5752 settings are stored persistently on disk, and thus allow the
5753 administrator to easily adjust the resource usage of
5754 services with a few simple commands. This dynamic resource
5755 management logic is also available to other programs via the
5756 bus. Almost any kernel cgroup attribute and controller is
5759 * systemd-vconsole-setup will now copy all font settings to
5760 all allocated VTs, where it previously applied them only to
5763 * libsystemd-login gained the new sd_session_get_tty() API
5766 * This release drops support for a few legacy or
5767 distribution-specific LSB facility names when parsing init
5768 scripts: $x-display-manager, $mail-transfer-agent,
5769 $mail-transport-agent, $mail-transfer-agent, $smtp,
5770 $null. Also, the mail-transfer-agent.target unit backing
5771 this has been removed. Distributions which want to retain
5772 compatibility with this should carry the burden for
5773 supporting this themselves and patch support for these back
5774 in, if they really need to. Also, the facilities $syslog and
5775 $local_fs are now ignored, since systemd does not support
5776 early-boot LSB init scripts anymore, and these facilities
5777 are implied anyway for normal services. syslog.target has
5780 * There are new bus calls on PID1's Manager object for
5781 cancelling jobs, and removing snapshot units. Previously,
5782 both calls were only available on the Job and Snapshot
5785 * systemd-journal-gatewayd gained SSL support.
5787 * The various "environment" files, such as /etc/locale.conf
5788 now support continuation lines with a backslash ("\") as
5789 last character in the line, similarly in style (but different)
5790 to how this is supported in shells.
5792 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5793 now implicitly appended to every log entry logged. systemctl
5794 has been updated to filter by this field when operating on a
5795 user systemd instance.
5797 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5798 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5799 the container. This makes it easier to boot unmodified
5800 Fedora systems in a container, which however still requires
5801 audit=0 to be passed on the kernel command line. Auditing in
5802 kernel and userspace is unfortunately still too broken in
5803 context of containers, hence we recommend compiling it out
5804 of the kernel or using audit=0. Hopefully this will be fixed
5805 one day for good in the kernel.
5807 * nspawn gained the new --bind= and --bind-ro= parameters to
5808 bind mount specific directories from the host into the
5811 * nspawn will now mount its own devpts file system instance
5812 into the container, in order not to leak pty devices from
5813 the host into the container.
5815 * systemd will now read the firmware boot time performance
5816 information from the EFI variables, if the used boot loader
5817 supports this, and takes it into account for boot performance
5818 analysis via "systemd-analyze". This is currently supported
5819 only in conjunction with Gummiboot, but could be supported
5820 by other boot loaders too. For details see:
5822 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5824 * A new generator has been added that automatically mounts the
5825 EFI System Partition (ESP) to /boot, if that directory
5826 exists, is empty, and no other file system has been
5827 configured to be mounted there.
5829 * logind will now send out PrepareForSleep(false) out
5830 unconditionally, after coming back from suspend. This may be
5831 used by applications as asynchronous notification for
5832 system resume events.
5834 * "systemctl unlock-sessions" has been added, that allows
5835 unlocking the screens of all user sessions at once, similar
5836 to how "systemctl lock-sessions" already locked all users
5837 sessions. This is backed by a new D-Bus call UnlockSessions().
5839 * "loginctl seat-status" will now show the master device of a
5840 seat. (i.e. the device of a seat that needs to be around for
5841 the seat to be considered available, usually the graphics
5844 * tmpfiles gained a new "X" line type, that allows
5845 configuration of files and directories (with wildcards) that
5846 shall be excluded from automatic cleanup ("aging").
5848 * udev default rules set the device node permissions now only
5849 at "add" events, and do not change them any longer with a
5850 later "change" event.
5852 * The log messages for lid events and power/sleep keypresses
5853 now carry a message ID.
5855 * We now have a substantially larger unit test suite, but this
5856 continues to be work in progress.
5858 * udevadm hwdb gained a new --root= parameter to change the
5859 root directory to operate relative to.
5861 * logind will now issue a background sync() request to the kernel
5862 early at shutdown, so that dirty buffers are flushed to disk early
5863 instead of at the last moment, in order to optimize shutdown
5866 * A new bootctl tool has been added that is an interface for
5867 certain boot loader operations. This is currently a preview
5868 and is likely to be extended into a small mechanism daemon
5869 like timedated, localed, hostnamed, and can be used by
5870 graphical UIs to enumerate available boot options, and
5871 request boot into firmware operations.
5873 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5874 the rest of the package. It also has been updated to work
5875 correctly in initrds.
5877 * Policykit previously has been runtime optional, and is now
5878 also compile time optional via a configure switch.
5880 * systemd-analyze has been reimplemented in C. Also "systemctl
5881 dot" has moved into systemd-analyze.
5883 * "systemctl status" with no further parameters will now print
5884 the status of all active or failed units.
5886 * Operations such as "systemctl start" can now be executed
5887 with a new mode "--irreversible" which may be used to queue
5888 operations that cannot accidentally be reversed by a later
5889 job queuing. This is by default used to make shutdown
5890 requests more robust.
5892 * The Python API of systemd now gained a new module for
5893 reading journal files.
5895 * A new tool kernel-install has been added that can install
5896 kernel images according to the Boot Loader Specification:
5898 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5900 * Boot time console output has been improved to provide
5901 animated boot time output for hanging jobs.
5903 * A new tool systemd-activate has been added which can be used
5904 to test socket activation with, directly from the command
5905 line. This should make it much easier to test and debug
5906 socket activation in daemons.
5908 * journalctl gained a new "--reverse" (or -r) option to show
5909 journal output in reverse order (i.e. newest line first).
5911 * journalctl gained a new "--pager-end" (or -e) option to jump
5912 to immediately jump to the end of the journal in the
5913 pager. This is only supported in conjunction with "less".
5915 * journalctl gained a new "--user-unit=" option, that works
5916 similarly to "--unit=" but filters for user units rather than
5919 * A number of unit files to ease adoption of systemd in
5920 initrds has been added. This moves some minimal logic from
5921 the various initrd implementations into systemd proper.
5923 * The journal files are now owned by a new group
5924 "systemd-journal", which exists specifically to allow access
5925 to the journal, and nothing else. Previously, we used the
5926 "adm" group for that, which however possibly covers more
5927 than just journal/log file access. This new group is now
5928 already used by systemd-journal-gatewayd to ensure this
5929 daemon gets access to the journal files and as little else
5930 as possible. Note that "make install" will also set FS ACLs
5931 up for /var/log/journal to give "adm" and "wheel" read
5932 access to it, in addition to "systemd-journal" which owns
5933 the journal files. We recommend that packaging scripts also
5934 add read access to "adm" + "wheel" to /var/log/journal, and
5935 all existing/future journal files. To normal users and
5936 administrators little changes, however packagers need to
5937 ensure to create the "systemd-journal" system group at
5938 package installation time.
5940 * The systemd-journal-gatewayd now runs as unprivileged user
5941 systemd-journal-gateway:systemd-journal-gateway. Packaging
5942 scripts need to create these system user/group at
5945 * timedated now exposes a new boolean property CanNTP that
5946 indicates whether a local NTP service is available or not.
5948 * systemd-detect-virt will now also detect xen PVs
5950 * The pstore file system is now mounted by default, if it is
5953 * In addition to the SELinux and IMA policies we will now also
5954 load SMACK policies at early boot.
5956 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5957 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5958 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5959 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5960 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5961 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5962 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5963 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5964 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5965 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5966 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5967 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5968 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5969 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5973 * Timer units now support calendar time events in addition to
5974 monotonic time events. That means you can now trigger a unit
5975 based on a calendar time specification such as "Thu,Fri
5976 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5977 or fifth day of any month of the year 2013, given that it is
5978 a thursday or friday. This brings timer event support
5979 considerably closer to cron's capabilities. For details on
5980 the supported calendar time specification language see
5983 * udev now supports a number of different naming policies for
5984 network interfaces for predictable names, and a combination
5985 of these policies is now the default. Please see this wiki
5986 document for details:
5988 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5990 * Auke Kok's bootchart implementation has been added to the
5991 systemd tree. It is an optional component that can graph the
5992 boot in quite some detail. It is one of the best bootchart
5993 implementations around and minimal in its code and
5996 * nss-myhostname has been integrated into the systemd source
5997 tree. nss-myhostname guarantees that the local hostname
5998 always stays resolvable via NSS. It has been a weak
5999 requirement of systemd-hostnamed since a long time, and
6000 since its code is actually trivial we decided to just
6001 include it in systemd's source tree. It can be turned off
6002 with a configure switch.
6004 * The read-ahead logic is now capable of properly detecting
6005 whether a btrfs file system is on SSD or rotating media, in
6006 order to optimize the read-ahead scheme. Previously, it was
6007 only capable of detecting this on traditional file systems
6010 * In udev, additional device properties are now read from the
6011 IAB in addition to the OUI database. Also, Bluetooth company
6012 identities are attached to the devices as well.
6014 * In service files %U may be used as specifier that is
6015 replaced by the configured user name of the service.
6017 * nspawn may now be invoked without a controlling TTY. This
6018 makes it suitable for invocation as its own service. This
6019 may be used to set up a simple containerized server system
6020 using only core OS tools.
6022 * systemd and nspawn can now accept socket file descriptors
6023 when they are started for socket activation. This enables
6024 implementation of socket activated nspawn
6025 containers. i.e. think about autospawning an entire OS image
6026 when the first SSH or HTTP connection is received. We expect
6027 that similar functionality will also be added to libvirt-lxc
6030 * journalctl will now suppress ANSI color codes when
6031 presenting log data.
6033 * systemctl will no longer show control group information for
6034 a unit if the control group is empty anyway.
6036 * logind can now automatically suspend/hibernate/shutdown the
6039 * /etc/machine-info and hostnamed now also expose the chassis
6040 type of the system. This can be used to determine whether
6041 the local system is a laptop, desktop, handset or
6042 tablet. This information may either be configured by the
6043 user/vendor or is automatically determined from ACPI and DMI
6044 information if possible.
6046 * A number of PolicyKit actions are now bound together with
6047 "imply" rules. This should simplify creating UIs because
6048 many actions will now authenticate similar ones as well.
6050 * Unit files learnt a new condition ConditionACPower= which
6051 may be used to conditionalize a unit depending on whether an
6052 AC power source is connected or not, of whether the system
6053 is running on battery power.
6055 * systemctl gained a new "is-failed" verb that may be used in
6056 shell scripts and suchlike to check whether a specific unit
6057 is in the "failed" state.
6059 * The EnvironmentFile= setting in unit files now supports file
6060 globbing, and can hence be used to easily read a number of
6061 environment files at once.
6063 * systemd will no longer detect and recognize specific
6064 distributions. All distribution-specific #ifdeffery has been
6065 removed, systemd is now fully generic and
6066 distribution-agnostic. Effectively, not too much is lost as
6067 a lot of the code is still accessible via explicit configure
6068 switches. However, support for some distribution specific
6069 legacy configuration file formats has been dropped. We
6070 recommend distributions to simply adopt the configuration
6071 files everybody else uses now and convert the old
6072 configuration from packaging scripts. Most distributions
6073 already did that. If that's not possible or desirable,
6074 distributions are welcome to forward port the specific
6075 pieces of code locally from the git history.
6077 * When logging a message about a unit systemd will now always
6078 log the unit name in the message meta data.
6080 * localectl will now also discover system locale data that is
6081 not stored in locale archives, but directly unpacked.
6083 * logind will no longer unconditionally use framebuffer
6084 devices as seat masters, i.e. as devices that are required
6085 to be existing before a seat is considered preset. Instead,
6086 it will now look for all devices that are tagged as
6087 "seat-master" in udev. By default, framebuffer devices will
6088 be marked as such, but depending on local systems, other
6089 devices might be marked as well. This may be used to
6090 integrate graphics cards using closed source drivers (such
6091 as NVidia ones) more nicely into logind. Note however, that
6092 we recommend using the open source NVidia drivers instead,
6093 and no udev rules for the closed-source drivers will be
6094 shipped from us upstream.
6096 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6097 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6098 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6099 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6100 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6101 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6102 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6103 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6104 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6105 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6106 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6107 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6112 * udev gained support for loading additional device properties
6113 from an indexed database that is keyed by vendor/product IDs
6114 and similar device identifiers. For the beginning this
6115 "hwdb" is populated with data from the well-known PCI and
6116 USB database, but also includes PNP, ACPI and OID data. In
6117 the longer run this indexed database shall grow into
6118 becoming the one central database for non-essential
6119 userspace device metadata. Previously, data from the PCI/USB
6120 database was only attached to select devices, since the
6121 lookup was a relatively expensive operation due to O(n) time
6122 complexity (with n being the number of entries in the
6123 database). Since this is now O(1), we decided to add in this
6124 data for all devices where this is available, by
6125 default. Note that the indexed database needs to be rebuilt
6126 when new data files are installed. To achieve this you need
6127 to update your packaging scripts to invoke "udevadm hwdb
6128 --update" after installation of hwdb data files. For
6129 RPM-based distributions we introduced the new
6130 %udev_hwdb_update macro for this purpose.
6132 * The Journal gained support for the "Message Catalog", an
6133 indexed database to link up additional information with
6134 journal entries. For further details please check:
6136 https://www.freedesktop.org/wiki/Software/systemd/catalog
6138 The indexed message catalog database also needs to be
6139 rebuilt after installation of message catalog files. Use
6140 "journalctl --update-catalog" for this. For RPM-based
6141 distributions we introduced the %journal_catalog_update
6142 macro for this purpose.
6144 * The Python Journal bindings gained support for the standard
6145 Python logging framework.
6147 * The Journal API gained new functions for checking whether
6148 the underlying file system of a journal file is capable of
6149 properly reporting file change notifications, or whether
6150 applications that want to reflect journal changes "live"
6151 need to recheck journal files continuously in appropriate
6154 * It is now possible to set the "age" field for tmpfiles
6155 entries to 0, indicating that files matching this entry
6156 shall always be removed when the directories are cleaned up.
6158 * coredumpctl gained a new "gdb" verb which invokes gdb
6159 right-away on the selected coredump.
6161 * There's now support for "hybrid sleep" on kernels that
6162 support this, in addition to "suspend" and "hibernate". Use
6163 "systemctl hybrid-sleep" to make use of this.
6165 * logind's HandleSuspendKey= setting (and related settings)
6166 now gained support for a new "lock" setting to simply
6167 request the screen lock on all local sessions, instead of
6168 actually executing a suspend or hibernation.
6170 * systemd will now mount the EFI variables file system by
6173 * Socket units now gained support for configuration of the
6174 SMACK security label.
6176 * timedatectl will now output the time of the last and next
6177 daylight saving change.
6179 * We dropped support for various legacy and distro-specific
6180 concepts, such as insserv, early-boot SysV services
6181 (i.e. those for non-standard runlevels such as 'b' or 'S')
6182 or ArchLinux /etc/rc.conf support. We recommend the
6183 distributions who still need support this to either continue
6184 to maintain the necessary patches downstream, or find a
6185 different solution. (Talk to us if you have questions!)
6187 * Various systemd components will now bypass PolicyKit checks
6188 for root and otherwise handle properly if PolicyKit is not
6189 found to be around. This should fix most issues for
6190 PolicyKit-less systems. Quite frankly this should have been
6191 this way since day one. It is absolutely our intention to
6192 make systemd work fine on PolicyKit-less systems, and we
6193 consider it a bug if something does not work as it should if
6194 PolicyKit is not around.
6196 * For embedded systems it is now possible to build udev and
6197 systemd without blkid and/or kmod support.
6199 * "systemctl switch-root" is now capable of switching root
6200 more than once. I.e. in addition to transitions from the
6201 initrd to the host OS it is now possible to transition to
6202 further OS images from the host. This is useful to implement
6203 offline updating tools.
6205 * Various other additions have been made to the RPM macros
6206 shipped with systemd. Use %udev_rules_update() after
6207 installing new udev rules files. %_udevhwdbdir,
6208 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6209 %_sysctldir are now available which resolve to the right
6210 directories for packages to place various data files in.
6212 * journalctl gained the new --full switch (in addition to
6213 --all, to disable ellipsation for long messages.
6215 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6216 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6217 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6218 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6219 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6220 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6221 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6222 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6223 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6227 * journalctl gained new --since= and --until= switches to
6228 filter by time. It also now supports nice filtering for
6229 units via --unit=/-u.
6231 * Type=oneshot services may use ExecReload= and do the
6234 * The journal daemon now supports time-based rotation and
6235 vacuuming, in addition to the usual disk-space based
6238 * The journal will now index the available field values for
6239 each field name. This enables clients to show pretty drop
6240 downs of available match values when filtering. The bash
6241 completion of journalctl has been updated
6242 accordingly. journalctl gained a new switch -F to list all
6243 values a certain field takes in the journal database.
6245 * More service events are now written as structured messages
6246 to the journal, and made recognizable via message IDs.
6248 * The timedated, localed and hostnamed mini-services which
6249 previously only provided support for changing time, locale
6250 and hostname settings from graphical DEs such as GNOME now
6251 also have a minimal (but very useful) text-based client
6252 utility each. This is probably the nicest way to changing
6253 these settings from the command line now, especially since
6254 it lists available options and is fully integrated with bash
6257 * There's now a new tool "systemd-coredumpctl" to list and
6258 extract coredumps from the journal.
6260 * We now install a README each in /var/log/ and
6261 /etc/rc.d/init.d explaining where the system logs and init
6262 scripts went. This hopefully should help folks who go to
6263 that dirs and look into the otherwise now empty void and
6264 scratch their heads.
6266 * When user-services are invoked (by systemd --user) the
6267 $MANAGERPID env var is set to the PID of systemd.
6269 * SIGRTMIN+24 when sent to a --user instance will now result
6270 in immediate termination of systemd.
6272 * gatewayd received numerous feature additions such as a
6273 "follow" mode, for live syncing and filtering.
6275 * browse.html now allows filtering and showing detailed
6276 information on specific entries. Keyboard navigation and
6277 mouse screen support has been added.
6279 * gatewayd/journalctl now supports HTML5/JSON
6280 Server-Sent-Events as output.
6282 * The SysV init script compatibility logic will now
6283 heuristically determine whether a script supports the
6284 "reload" verb, and only then make this available as
6287 * "systemctl status --follow" has been removed, use "journalctl
6290 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6291 have been removed since they are hardly useful to be
6294 * And I'd like to take the opportunity to specifically mention
6295 Zbigniew for his great contributions. Zbigniew, you rock!
6297 Contributions from: Andrew Eikum, Christian Hesse, Colin
6298 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6299 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6300 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6301 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6302 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6303 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6307 * If /etc/vconsole.conf is non-existent or empty we will no
6308 longer load any console font or key map at boot by
6309 default. Instead the kernel defaults will be left
6310 intact. This is definitely the right thing to do, as no
6311 configuration should mean no configuration, and hard-coding
6312 font names that are different on all archs is probably a bad
6313 idea. Also, the kernel default key map and font should be
6314 good enough for most cases anyway, and mostly identical to
6315 the userspace fonts/key maps we previously overloaded them
6316 with. If distributions want to continue to default to a
6317 non-kernel font or key map they should ship a default
6318 /etc/vconsole.conf with the appropriate contents.
6320 Contributions from: Colin Walters, Daniel J Walsh, Dave
6321 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6322 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6326 * journalctl gained a new --cursor= switch to show entries
6327 starting from the specified location in the journal.
6329 * We now enforce a size limit on journal entry fields exported
6330 with "-o json" in journalctl. Fields larger than 4K will be
6331 assigned null. This can be turned off with --all.
6333 * An (optional) journal gateway daemon is now available as
6334 "systemd-journal-gatewayd.service". This service provides
6335 access to the journal via HTTP and JSON. This functionality
6336 will be used to implement live log synchronization in both
6337 pull and push modes, but has various other users too, such
6338 as easy log access for debugging of embedded devices. Right
6339 now it is already useful to retrieve the journal via HTTP:
6341 # systemctl start systemd-journal-gatewayd.service
6342 # wget http://localhost:19531/entries
6344 This will download the journal contents in a
6345 /var/log/messages compatible format. The same as JSON:
6347 # curl -H"Accept: application/json" http://localhost:19531/entries
6349 This service is also accessible via a web browser where a
6350 single static HTML5 app is served that uses the JSON logic
6351 to enable the user to do some basic browsing of the
6352 journal. This will be extended later on. Here's an example
6353 screenshot of this app in its current state:
6355 http://0pointer.de/public/journal-gatewayd
6357 Contributions from: Kay Sievers, Lennart Poettering, Robert
6358 Milasan, Tom Gundersen
6362 * The bash completion logic is now available for journalctl
6365 * We do not mount the "cpuset" controller anymore together with
6366 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6367 started if no parameters are assigned to it. "cpuset" hence
6368 broke code that assumed it could create "cpu" groups and
6371 * journalctl -f will now subscribe to terminal size changes,
6372 and line break accordingly.
6374 Contributions from: Dave Reisner, Kay Sievers, Lennart
6375 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6379 * nspawn will now create a symlink /etc/localtime in the
6380 container environment, copying the host's timezone
6381 setting. Previously this has been done via a bind mount, but
6382 since symlinks cannot be bind mounted this has now been
6383 changed to create/update the appropriate symlink.
6385 * journalctl -n's line number argument is now optional, and
6386 will default to 10 if omitted.
6388 * journald will now log the maximum size the journal files may
6389 take up on disk. This is particularly useful if the default
6390 built-in logic of determining this parameter from the file
6391 system size is used. Use "systemctl status
6392 systemd-journald.service" to see this information.
6394 * The multi-seat X wrapper tool has been stripped down. As X
6395 is now capable of enumerating graphics devices via udev in a
6396 seat-aware way the wrapper is not strictly necessary
6397 anymore. A stripped down temporary stop-gap is still shipped
6398 until the upstream display managers have been updated to
6399 fully support the new X logic. Expect this wrapper to be
6400 removed entirely in one of the next releases.
6402 * HandleSleepKey= in logind.conf has been split up into
6403 HandleSuspendKey= and HandleHibernateKey=. The old setting
6404 is not available anymore. X11 and the kernel are
6405 distinguishing between these keys and we should too. This
6406 also means the inhibition lock for these keys has been split
6409 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6410 Poettering, Lukas Nykryn, Václav Pavlín
6414 * Whenever a unit changes state we will now log this to the
6415 journal and show along the unit's own log output in
6418 * ConditionPathIsMountPoint= can now properly detect bind
6419 mount points too. (Previously, a bind mount of one file
6420 system to another place in the same file system could not be
6421 detected as mount, since they shared struct stat's st_dev
6424 * We will now mount the cgroup controllers cpu, cpuacct,
6425 cpuset and the controllers net_cls, net_prio together by
6428 * nspawn containers will now have a virtualized boot
6429 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6430 over with a randomized ID at container initialization). This
6431 has the effect of making "journalctl -b" do the right thing
6434 * The JSON output journal serialization has been updated not
6435 to generate "endless" list objects anymore, but rather one
6436 JSON object per line. This is more in line how most JSON
6437 parsers expect JSON objects. The new output mode
6438 "json-pretty" has been added to provide similar output, but
6439 neatly aligned for readability by humans.
6441 * We dropped all explicit sync() invocations in the shutdown
6442 code. The kernel does this implicitly anyway in the kernel
6443 reboot() syscall. halt(8)'s -n option is now a compatibility
6446 * We now support virtualized reboot() in containers, as
6447 supported by newer kernels. We will fall back to exit() if
6448 CAP_SYS_REBOOT is not available to the container. Also,
6449 nspawn makes use of this now and will actually reboot the
6450 container if the containerized OS asks for that.
6452 * journalctl will only show local log output by default
6453 now. Use --merge (-m) to show remote log output, too.
6455 * libsystemd-journal gained the new sd_journal_get_usage()
6456 call to determine the current disk usage of all journal
6457 files. This is exposed in the new "journalctl --disk-usage"
6460 * journald gained a new configuration setting SplitMode= in
6461 journald.conf which may be used to control how user journals
6462 are split off. See journald.conf(5) for details.
6464 * A new condition type ConditionFileNotEmpty= has been added.
6466 * tmpfiles' "w" lines now support file globbing, to write
6467 multiple files at once.
6469 * We added Python bindings for the journal submission
6470 APIs. More Python APIs for a number of selected APIs will
6471 likely follow. Note that we intend to add native bindings
6472 only for the Python language, as we consider it common
6473 enough to deserve bindings shipped within systemd. There are
6474 various projects outside of systemd that provide bindings
6475 for languages such as PHP or Lua.
6477 * Many conditions will now resolve specifiers such as %i. In
6478 addition, PathChanged= and related directives of .path units
6479 now support specifiers as well.
6481 * There's now a new RPM macro definition for the system preset
6484 * journald will now warn if it ca not forward a message to the
6485 syslog daemon because its socket is full.
6487 * timedated will no longer write or process /etc/timezone,
6488 except on Debian. As we do not support late mounted /usr
6489 anymore /etc/localtime always being a symlink is now safe,
6490 and hence the information in /etc/timezone is not necessary
6493 * logind will now always reserve one VT for a text getty (VT6
6494 by default). Previously if more than 6 X sessions where
6495 started they took up all the VTs with auto-spawned gettys,
6496 so that no text gettys were available anymore.
6498 * udev will now automatically inform the btrfs kernel logic
6499 about btrfs RAID components showing up. This should make
6500 simple hotplug based btrfs RAID assembly work.
6502 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6503 (but not for its children which will stay at the kernel
6504 default). This should allow setups with a lot more listening
6507 * systemd will now always pass the configured timezone to the
6508 kernel at boot. timedated will do the same when the timezone
6511 * logind's inhibition logic has been updated. By default,
6512 logind will now handle the lid switch, the power and sleep
6513 keys all the time, even in graphical sessions. If DEs want
6514 to handle these events on their own they should take the new
6515 handle-power-key, handle-sleep-key and handle-lid-switch
6516 inhibitors during their runtime. A simple way to achieve
6517 that is to invoke the DE wrapped in an invocation of:
6519 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6521 * Access to unit operations is now checked via SELinux taking
6522 the unit file label and client process label into account.
6524 * systemd will now notify the administrator in the journal
6525 when he over-mounts a non-empty directory.
6527 * There are new specifiers that are resolved in unit files,
6528 for the host name (%H), the machine ID (%m) and the boot ID
6531 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6532 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6533 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6534 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6535 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6536 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6537 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6541 * Support for reading structured kernel messages from
6542 /dev/kmsg has now been added and is enabled by default.
6544 * Support for reading kernel messages from /proc/kmsg has now
6545 been removed. If you want kernel messages in the journal
6546 make sure to run a recent kernel (>= 3.5) that supports
6547 reading structured messages from /dev/kmsg (see
6548 above). /proc/kmsg is now exclusive property of classic
6549 syslog daemons again.
6551 * The libudev API gained the new
6552 udev_device_new_from_device_id() call.
6554 * The logic for file system namespace (ReadOnlyDirectory=,
6555 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6556 require pivot_root() anymore. This means fewer temporary
6557 directories are created below /tmp for this feature.
6559 * nspawn containers will now see and receive all submounts
6560 made on the host OS below the root file system of the
6563 * Forward Secure Sealing is now supported for Journal files,
6564 which provide cryptographical sealing of journal files so
6565 that attackers cannot alter log history anymore without this
6566 being detectable. Lennart will soon post a blog story about
6567 this explaining it in more detail.
6569 * There are two new service settings RestartPreventExitStatus=
6570 and SuccessExitStatus= which allow configuration of exit
6571 status (exit code or signal) which will be excepted from the
6572 restart logic, resp. consider successful.
6574 * journalctl gained the new --verify switch that can be used
6575 to check the integrity of the structure of journal files and
6576 (if Forward Secure Sealing is enabled) the contents of
6579 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6580 and similar symlinks pre-created. This makes running shells
6581 as container init process a lot more fun.
6583 * The fstab support can now handle PARTUUID= and PARTLABEL=
6586 * A new ConditionHost= condition has been added to match
6587 against the hostname (with globs) and machine ID. This is
6588 useful for clusters where a single OS image is used to
6589 provision a large number of hosts which shall run slightly
6590 different sets of services.
6592 * Services which hit the restart limit will now be placed in a
6595 Contributions from: Bertram Poettering, Dave Reisner, Huang
6596 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6597 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6601 * When running in --user mode systemd will now become a
6602 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6603 tree a lot more organized.
6605 * A new PartOf= unit dependency type has been introduced that
6606 may be used to group services in a natural way.
6608 * "systemctl enable" may now be used to enable instances of
6611 * journalctl now prints error log levels in red, and
6612 warning/notice log levels in bright white. It also supports
6613 filtering by log level now.
6615 * cgtop gained a new -n switch (similar to top), to configure
6616 the maximum number of iterations to run for. It also gained
6617 -b, to run in batch mode (accepting no input).
6619 * The suffix ".service" may now be omitted on most systemctl
6620 command lines involving service unit names.
6622 * There's a new bus call in logind to lock all sessions, as
6623 well as a loginctl verb for it "lock-sessions".
6625 * libsystemd-logind.so gained a new call sd_journal_perror()
6626 that works similar to libc perror() but logs to the journal
6627 and encodes structured information about the error number.
6629 * /etc/crypttab entries now understand the new keyfile-size=
6632 * shutdown(8) now can send a (configurable) wall message when
6633 a shutdown is cancelled.
6635 * The mount propagation mode for the root file system will now
6636 default to "shared", which is useful to make containers work
6637 nicely out-of-the-box so that they receive new mounts from
6638 the host. This can be undone locally by running "mount
6639 --make-rprivate /" if needed.
6641 * The prefdm.service file has been removed. Distributions
6642 should maintain this unit downstream if they intend to keep
6643 it around. However, we recommend writing normal unit files
6644 for display managers instead.
6646 * Since systemd is a crucial part of the OS we will now
6647 default to a number of compiler switches that improve
6648 security (hardening) such as read-only relocations, stack
6649 protection, and suchlike.
6651 * The TimeoutSec= setting for services is now split into
6652 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6653 of individual time outs for the start and the stop phase of
6656 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6657 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6658 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6659 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6660 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6661 Gundersen, Zbigniew Jędrzejewski-Szmek
6665 * The journal and id128 C APIs are now fully documented as man
6668 * Extra safety checks have been added when transitioning from
6669 the initial RAM disk to the main system to avoid accidental
6672 * /etc/crypttab entries now understand the new keyfile-offset=
6675 * systemctl -t can now be used to filter by unit load state.
6677 * The journal C API gained the new sd_journal_wait() call to
6678 make writing synchronous journal clients easier.
6680 * journalctl gained the new -D switch to show journals from a
6683 * journalctl now displays a special marker between log
6684 messages of two different boots.
6686 * The journal is now explicitly flushed to /var via a service
6687 systemd-journal-flush.service, rather than implicitly simply
6688 by seeing /var/log/journal to be writable.
6690 * journalctl (and the journal C APIs) can now match for much
6691 more complex expressions, with alternatives and
6694 * When transitioning from the initial RAM disk to the main
6695 system we will now kill all processes in a killing spree to
6696 ensure no processes stay around by accident.
6698 * Three new specifiers may be used in unit files: %u, %h, %s
6699 resolve to the user name, user home directory resp. user
6700 shell. This is useful for running systemd user instances.
6702 * We now automatically rotate journal files if their data
6703 object hash table gets a fill level > 75%. We also size the
6704 hash table based on the configured maximum file size. This
6705 together should lower hash collisions drastically and thus
6706 speed things up a bit.
6708 * journalctl gained the new "--header" switch to introspect
6709 header data of journal files.
6711 * A new setting SystemCallFilters= has been added to services
6712 which may be used to apply blacklists or whitelists to
6713 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6715 * nspawn gained a new --link-journal= switch (and quicker: -j)
6716 to link the container journal with the host. This makes it
6717 very easy to centralize log viewing on the host for all
6718 guests while still keeping the journal files separated.
6720 * Many bugfixes and optimizations
6722 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6723 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6724 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6729 * Several tools now understand kernel command line arguments,
6730 which are only read when run in an initial RAM disk. They
6731 usually follow closely their normal counterparts, but are
6734 * There's a new tool to analyze the readahead files that are
6735 automatically generated at boot. Use:
6737 /usr/lib/systemd/systemd-readahead analyze /.readahead
6739 * We now provide an early debug shell on tty9 if this enabled. Use:
6741 systemctl enable debug-shell.service
6743 * All plymouth related units have been moved into the Plymouth
6744 package. Please make sure to upgrade your Plymouth version
6747 * systemd-tmpfiles now supports getting passed the basename of
6748 a configuration file only, in which case it will look for it
6749 in all appropriate directories automatically.
6751 * udevadm info now takes a /dev or /sys path as argument, and
6752 does the right thing. Example:
6754 udevadm info /dev/sda
6755 udevadm info /sys/class/block/sda
6757 * systemctl now prints a warning if a unit is stopped but a
6758 unit that might trigger it continues to run. Example: a
6759 service is stopped but the socket that activates it is left
6762 * "systemctl status" will now mention if the log output was
6763 shortened due to rotation since a service has been started.
6765 * The journal API now exposes functions to determine the
6766 "cutoff" times due to rotation.
6768 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6769 immediately flushing of runtime logs to /var if possible,
6770 resp. for triggering immediate rotation of the journal
6773 * It is now considered an error if a service is attempted to
6774 be stopped that is not loaded.
6776 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6778 * systemd-analyze now supports Python 3
6780 * tmpfiles now supports cleaning up directories via aging
6781 where the first level dirs are always kept around but
6782 directories beneath it automatically aged. This is enabled
6783 by prefixing the age field with '~'.
6785 * Seat objects now expose CanGraphical, CanTTY properties
6786 which is required to deal with very fast bootups where the
6787 display manager might be running before the graphics drivers
6788 completed initialization.
6790 * Seat objects now expose a State property.
6792 * We now include RPM macros for service enabling/disabling
6793 based on the preset logic. We recommend RPM based
6794 distributions to make use of these macros if possible. This
6795 makes it simpler to reuse RPM spec files across
6798 * We now make sure that the collected systemd unit name is
6799 always valid when services log to the journal via
6802 * There's a new man page kernel-command-line(7) detailing all
6803 command line options we understand.
6805 * The fstab generator may now be disabled at boot by passing
6806 fstab=0 on the kernel command line.
6808 * A new kernel command line option modules-load= is now understood
6809 to load a specific kernel module statically, early at boot.
6811 * Unit names specified on the systemctl command line are now
6812 automatically escaped as needed. Also, if file system or
6813 device paths are specified they are automatically turned
6814 into the appropriate mount or device unit names. Example:
6816 systemctl status /home
6817 systemctl status /dev/sda
6819 * The SysVConsole= configuration option has been removed from
6820 system.conf parsing.
6822 * The SysV search path is no longer exported on the D-Bus
6825 * The Names= option has been removed from unit file parsing.
6827 * There's a new man page bootup(7) detailing the boot process.
6829 * Every unit and every generator we ship with systemd now
6830 comes with full documentation. The self-explanatory boot is
6833 * A couple of services gained "systemd-" prefixes in their
6834 name if they wrap systemd code, rather than only external
6835 code. Among them fsck@.service which is now
6836 systemd-fsck@.service.
6838 * The HaveWatchdog property has been removed from the D-Bus
6841 * systemd.confirm_spawn= on the kernel command line should now
6844 * There's a new man page crypttab(5) which details all options
6845 we actually understand.
6847 * systemd-nspawn gained a new --capability= switch to pass
6848 additional capabilities to the container.
6850 * timedated will now read known NTP implementation unit names
6851 from /usr/lib/systemd/ntp-units.d/*.list,
6852 systemd-timedated-ntp.target has been removed.
6854 * journalctl gained a new switch "-b" that lists log data of
6855 the current boot only.
6857 * The notify socket is in the abstract namespace again, in
6858 order to support daemons which chroot() at start-up.
6860 * There is a new Storage= configuration option for journald
6861 which allows configuration of where log data should go. This
6862 also provides a way to disable journal logging entirely, so
6863 that data collected is only forwarded to the console, the
6864 kernel log buffer or another syslog implementation.
6866 * Many bugfixes and optimizations
6868 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6869 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6870 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6871 Shawn Landden, Tom Gundersen
6875 * "systemctl help <unit>" now shows the man page if one is
6878 * Several new man pages have been added.
6880 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6881 MaxLevelConsole= can now be specified in
6882 journald.conf. These options allow reducing the amount of
6883 data stored on disk or forwarded by the log level.
6885 * TimerSlackNSec= can now be specified in system.conf for
6886 PID1. This allows system-wide power savings.
6888 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6889 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6894 * logind is now capable of (optionally) handling power and
6895 sleep keys as well as the lid switch.
6897 * journalctl now understands the syntax "journalctl
6898 /usr/bin/avahi-daemon" to get all log output of a specific
6901 * CapabilityBoundingSet= in system.conf now also influences
6902 the capability bound set of usermode helpers of the kernel.
6904 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6905 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6906 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6907 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6911 * Note that we skipped 139 releases here in order to set the
6912 new version to something that is greater than both udev's
6913 and systemd's most recent version number.
6915 * udev: all udev sources are merged into the systemd source tree now.
6916 All future udev development will happen in the systemd tree. It
6917 is still fully supported to use the udev daemon and tools without
6918 systemd running, like in initramfs or other init systems. Building
6919 udev though, will require the *build* of the systemd tree, but
6920 udev can be properly *run* without systemd.
6922 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6923 should be used to create dead device nodes as workarounds for broken
6926 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6927 no longer supported. udev_monitor_new_from_netlink() needs to be
6928 used to subscribe to events.
6930 * udev: when udevd is started by systemd, processes which are left
6931 behind by forking them off of udev rules, are unconditionally cleaned
6932 up and killed now after the event handling has finished. Services or
6933 daemons must be started as systemd services. Services can be
6934 pulled-in by udev to get started, but they can no longer be directly
6935 forked by udev rules.
6937 * udev: the daemon binary is called systemd-udevd now and installed
6938 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6939 to adapt to that, create symlink, or rename the binary after building
6942 * libudev no longer provides these symbols:
6943 udev_monitor_from_socket()
6944 udev_queue_get_failed_list_entry()
6945 udev_get_{dev,sys,run}_path()
6946 The versions number was bumped and symbol versioning introduced.
6948 * systemd-loginctl and systemd-journalctl have been renamed
6949 to loginctl and journalctl to match systemctl.
6951 * The config files: /etc/systemd/systemd-logind.conf and
6952 /etc/systemd/systemd-journald.conf have been renamed to
6953 logind.conf and journald.conf. Package updates should rename
6954 the files to the new names on upgrade.
6956 * For almost all files the license is now LGPL2.1+, changed
6957 from the previous GPL2.0+. Exceptions are some minor stuff
6958 of udev (which will be changed to LGPL2.1 eventually, too),
6959 and the MIT licensed sd-daemon.[ch] library that is suitable
6960 to be used as drop-in files.
6962 * systemd and logind now handle system sleep states, in
6963 particular suspending and hibernating.
6965 * logind now implements a sleep/shutdown/idle inhibiting logic
6966 suitable for a variety of uses. Soonishly Lennart will blog
6967 about this in more detail.
6969 * var-run.mount and var-lock.mount are no longer provided
6970 (which previously bind mounted these directories to their new
6971 places). Distributions which have not converted these
6972 directories to symlinks should consider stealing these files
6973 from git history and add them downstream.
6975 * We introduced the Documentation= field for units and added
6976 this to all our shipped units. This is useful to make it
6977 easier to explore the boot and the purpose of the various
6980 * All smaller setup units (such as
6981 systemd-vconsole-setup.service) now detect properly if they
6982 are run in a container and are skipped when
6983 appropriate. This guarantees an entirely noise-free boot in
6984 Linux container environments such as systemd-nspawn.
6986 * A framework for implementing offline system updates is now
6987 integrated, for details see:
6988 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
6990 * A new service type Type=idle is available now which helps us
6991 avoiding ugly interleaving of getty output and boot status
6994 * There's now a system-wide CapabilityBoundingSet= option to
6995 globally reduce the set of capabilities for the
6996 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6997 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6998 even CAP_NET_ADMIN system-wide for secure systems.
7000 * There are now system-wide DefaultLimitXXX= options to
7001 globally change the defaults of the various resource limits
7002 for all units started by PID 1.
7004 * Harald Hoyer's systemd test suite has been integrated into
7005 systemd which allows easy testing of systemd builds in qemu
7006 and nspawn. (This is really awesome! Ask us for details!)
7008 * The fstab parser is now implemented as generator, not inside
7011 * systemctl will now warn you if .mount units generated from
7012 /etc/fstab are out of date due to changes in fstab that
7013 have not been read by systemd yet.
7015 * systemd is now suitable for usage in initrds. Dracut has
7016 already been updated to make use of this. With this in place
7017 initrds get a slight bit faster but primarily are much
7018 easier to introspect and debug since "systemctl status" in
7019 the host system can be used to introspect initrd services,
7020 and the journal from the initrd is kept around too.
7022 * systemd-delta has been added, a tool to explore differences
7023 between user/admin configuration and vendor defaults.
7025 * PrivateTmp= now affects both /tmp and /var/tmp.
7027 * Boot time status messages are now much prettier and feature
7028 proper english language. Booting up systemd has never been
7031 * Read-ahead pack files now include the inode number of all
7032 files to pre-cache. When the inode changes the pre-caching
7033 is not attempted. This should be nicer to deal with updated
7034 packages which might result in changes of read-ahead
7037 * We now temporaritly lower the kernel's read_ahead_kb variable
7038 when collecting read-ahead data to ensure the kernel's
7039 built-in read-ahead does not add noise to our measurements
7040 of necessary blocks to pre-cache.
7042 * There's now RequiresMountsFor= to add automatic dependencies
7043 for all mounts necessary for a specific file system path.
7045 * MountAuto= and SwapAuto= have been removed from
7046 system.conf. Mounting file systems at boot has to take place
7049 * nspawn now learned a new switch --uuid= to set the machine
7050 ID on the command line.
7052 * nspawn now learned the -b switch to automatically search
7055 * vt102 is now the default TERM for serial TTYs, upgraded from
7058 * systemd-logind now works on VT-less systems.
7060 * The build tree has been reorganized. The individual
7061 components now have directories of their own.
7063 * A new condition type ConditionPathIsReadWrite= is now available.
7065 * nspawn learned the new -C switch to create cgroups for the
7066 container in other hierarchies.
7068 * We now have support for hardware watchdogs, configurable in
7071 * The scheduled shutdown logic now has a public API.
7073 * We now mount /tmp as tmpfs by default, but this can be
7074 masked and /etc/fstab can override it.
7076 * Since udisks does not make use of /media anymore we are not
7077 mounting a tmpfs on it anymore.
7079 * journalctl gained a new --local switch to only interleave
7080 locally generated journal files.
7082 * We can now load the IMA policy at boot automatically.
7084 * The GTK tools have been split off into a systemd-ui.
7086 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7087 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7088 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7089 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7090 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7091 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7092 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7093 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7094 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7099 * This is mostly a bugfix release
7101 * Support optional initialization of the machine ID from the
7102 KVM or container configured UUID.
7104 * Support immediate reboots with "systemctl reboot -ff"
7106 * Show /etc/os-release data in systemd-analyze output
7108 * Many bugfixes for the journal, including endianness fixes and
7109 ensuring that disk space enforcement works
7111 * sd-login.h is C++ compatible again
7113 * Extend the /etc/os-release format on request of the Debian
7116 * We now refuse non-UTF8 strings used in various configuration
7117 and unit files. This is done to ensure we do not pass invalid
7118 data over D-Bus or expose it elsewhere.
7120 * Register Mimo USB Screens as suitable for automatic seat
7123 * Read SELinux client context from journal clients in a race
7126 * Reorder configuration file lookup order. /etc now always
7127 overrides /run in order to allow the administrator to always
7128 and unconditionally override vendor-supplied or
7129 automatically generated data.
7131 * The various user visible bits of the journal now have man
7132 pages. We still lack man pages for the journal API calls
7135 * We now ship all man pages in HTML format again in the
7138 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7139 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7140 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7145 * This is mostly a bugfix release
7147 * systems lacking /etc/os-release are no longer supported.
7149 * Various functionality updates to libsystemd-login.so
7151 * Track class of PAM logins to distinguish greeters from
7154 Contributions from: Kay Sievers, Lennart Poettering, Michael
7159 * This is an important bugfix release for v41.
7161 * Building man pages is now optional which should be useful
7162 for those building systemd from git but unwilling to install
7165 * Watchdog support for supervising services is now usable. In
7166 a future release support for hardware watchdogs
7167 (i.e. /dev/watchdog) will be added building on this.
7169 * Service start rate limiting is now configurable and can be
7170 turned off per service. When a start rate limit is hit a
7171 reboot can automatically be triggered.
7173 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7175 Contributions from: Benjamin Franzke, Bill Nottingham,
7176 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7177 Schmidt, Michał Górny, Piotr Drąg
7181 * The systemd binary is installed /usr/lib/systemd/systemd now;
7182 An existing /sbin/init symlink needs to be adapted with the
7185 * The code that loads kernel modules has been ported to invoke
7186 libkmod directly, instead of modprobe. This means we do not
7187 support systems with module-init-tools anymore.
7189 * Watchdog support is now already useful, but still not
7192 * A new kernel command line option systemd.setenv= is
7193 understood to set system wide environment variables
7194 dynamically at boot.
7196 * We now limit the set of capabilities of systemd-journald.
7198 * We now set SIGPIPE to ignore by default, since it only is
7199 useful in shell pipelines, and has little use in general
7200 code. This can be disabled with IgnoreSIPIPE=no in unit
7203 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7204 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7209 * This is mostly a bugfix release
7211 * We now expose the reason why a service failed in the
7212 "Result" D-Bus property.
7214 * Rudimentary service watchdog support (will be completed over
7215 the next few releases.)
7217 * When systemd forks off in order execute some service we will
7218 now immediately changes its argv[0] to reflect which process
7219 it will execute. This is useful to minimize the time window
7220 with a generic argv[0], which makes bootcharts more useful
7222 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7223 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7224 Mike Kazantsev, Ray Strode
7228 * This is mostly a test release, but incorporates many
7231 * New systemd-cgtop tool to show control groups by their
7234 * Linking against libacl for ACLs is optional again. If
7235 disabled, support tracking device access for active logins
7236 goes becomes unavailable, and so does access to the user
7237 journals by the respective users.
7239 * If a group "adm" exists, journal files are automatically
7240 owned by them, thus allow members of this group full access
7241 to the system journal as well as all user journals.
7243 * The journal now stores the SELinux context of the logging
7244 client for all entries.
7246 * Add C++ inclusion guards to all public headers
7248 * New output mode "cat" in the journal to print only text
7249 messages, without any meta data like date or time.
7251 * Include tiny X server wrapper as a temporary stop-gap to
7252 teach XOrg udev display enumeration. This is used by display
7253 managers such as gdm, and will go away as soon as XOrg
7254 learned native udev hotplugging for display devices.
7256 * Add new systemd-cat tool for executing arbitrary programs
7257 with STDERR/STDOUT connected to the journal. Can also act as
7258 BSD logger replacement, and does so by default.
7260 * Optionally store all locally generated coredumps in the
7261 journal along with meta data.
7263 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7264 writing short strings to files (for usage for /sys), and for
7265 creating symlinks, character and block device nodes.
7267 * New unit file option ControlGroupPersistent= to make cgroups
7268 persistent, following the mechanisms outlined in
7269 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7271 * Support multiple local RTCs in a sane way
7273 * No longer monopolize IO when replaying readahead data on
7274 rotating disks, since we might starve non-file-system IO to
7275 death, since fanotify() will not see accesses done by blkid,
7278 * Do not show kernel threads in systemd-cgls anymore, unless
7279 requested with new -k switch.
7281 Contributions from: Dan Horák, Kay Sievers, Lennart
7282 Poettering, Michal Schmidt
7286 * This is mostly a test release, but incorporates many
7289 * The git repository moved to:
7290 git://anongit.freedesktop.org/systemd/systemd
7291 ssh://git.freedesktop.org/git/systemd/systemd
7293 * First release with the journal
7294 http://0pointer.de/blog/projects/the-journal.html
7296 * The journal replaces both systemd-kmsg-syslogd and
7297 systemd-stdout-bridge.
7299 * New sd_pid_get_unit() API call in libsystemd-logind
7301 * Many systemadm clean-ups
7303 * Introduce remote-fs-pre.target which is ordered before all
7304 remote mounts and may be used to start services before all
7307 * Added Mageia support
7309 * Add bash completion for systemd-loginctl
7311 * Actively monitor PID file creation for daemons which exit in
7312 the parent process before having finished writing the PID
7313 file in the daemon process. Daemons which do this need to be
7314 fixed (i.e. PID file creation must have finished before the
7315 parent exits), but we now react a bit more gracefully to them.
7317 * Add colourful boot output, mimicking the well-known output
7318 of existing distributions.
7320 * New option PassCredentials= for socket units, for
7321 compatibility with a recent kernel ABI breakage.
7323 * /etc/rc.local is now hooked in via a generator binary, and
7324 thus will no longer act as synchronization point during
7327 * systemctl list-unit-files now supports --root=.
7329 * systemd-tmpfiles now understands two new commands: z, Z for
7330 relabelling files according to the SELinux database. This is
7331 useful to apply SELinux labels to specific files in /sys,
7334 * Output of SysV services is now forwarded to both the console
7335 and the journal by default, not only just the console.
7337 * New man pages for all APIs from libsystemd-login.
7339 * The build tree got reorganized and the build system is a
7340 lot more modular allowing embedded setups to specifically
7341 select the components of systemd they are interested in.
7343 * Support for Linux systems lacking the kernel VT subsystem is
7346 * configure's --with-rootdir= got renamed to
7347 --with-rootprefix= to follow the naming used by udev and
7350 * Unless specified otherwise we will now install to /usr instead
7351 of /usr/local by default.
7353 * Processes with '@' in argv[0][0] are now excluded from the
7354 final shut-down killing spree, following the logic explained
7356 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7358 * All processes remaining in a service cgroup when we enter
7359 the START or START_PRE states are now killed with
7360 SIGKILL. That means it is no longer possible to spawn
7361 background processes from ExecStart= lines (which was never
7362 supported anyway, and bad style).
7364 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7365 reloading of units together.
7367 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7368 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7369 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7370 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7371 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob