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 ambiguity 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.
14 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
15 which print the logging level and target of the system manager,
16 respectively. They complement the existing "set-log-level" and
17 "set-log-target" verbs, which can be used to change those values.
19 * systemd-networkd .network DHCP setting UseMTU default has changed
20 from false to true. Meaning, DHCP server advertised mtu setting is
21 now applied by default. This resolves networking issues on low-mtu
26 * Meson is now supported as build system in addition to Automake. It is
27 our plan to remove Automake in one of our next releases, so that
28 Meson becomes our exclusive build system. Hence, please start using
29 the Meson build system in your downstream packaging. There's plenty
30 of documentation around how to use Meson, the extremely brief
33 ./autogen.sh && ./configure && make && sudo make install
37 meson build && ninja -C build && sudo ninja -C build install
39 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
40 which permits configuring a timeout on the time a job is
41 running. This is particularly useful for setting timeouts on jobs for
44 * Unit files gained two new options ConditionUser= and ConditionGroup=
45 for conditionalizing units based on the identity of the user/group
46 running a systemd user instance.
48 * systemd-networkd now understands a new FlowLabel= setting in the
49 [VXLAN] section of .network files, as well as a Priority= in
50 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
51 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
52 gained support for configuration of GENEVE links, and IPv6 address
53 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
55 * .link files now understand a new Port= setting.
57 * systemd-networkd's DHCP support gained support for DHCP option 119
60 * systemd-networkd gained support for serving IPv6 address ranges using
61 the Router Advertisment protocol. The new .network configuration
62 section [IPv6Prefix] may be used to configure the ranges to
63 serve. This is implemented based on a new, minimal, native server
66 * journalctl's --output= switch gained support for a new parameter
67 "short-iso-precise" for a mode where timestamps are shown as precise
70 * systemd-udevd's "net_id" builtin may now generate stable network
71 interface names from IBM PowerVM VIO devices as well as ACPI platform
74 * MulticastDNS support in systemd-resolved may now be explicitly
75 enabled/disabled using the new MulticastDNS= configuration file
78 * systemd-resolved may now optionally use libidn2 instead of the libidn
79 for processing internationalized domain names. Support for libidn2
80 should be considered experimental and should not be enabled by
83 * "machinectl pull-tar" and related call may now do verification of
84 downloaded images using SUSE-style .sha256 checksum files in addition
85 to the already existing support for validating using Ubuntu-style
88 * sd-bus gained support for a new sd_bus_message_appendv() call which
89 is va_list equivalent of sd_bus_message_append().
91 * sd-boot gained support for validating images using SHIM/MOK.
93 * The SMACK code learnt support for "onlycap".
95 * systemd-mount --umount is now much smarter in figuring out how to
96 properly unmount a device given its mount or device path.
98 * The code to call libnss_dns as a fallback from libnss_resolve when
99 the communication with systemd-resolved fails was removed. This
100 fallback was redundant and interfered with the [!UNAVAIL=return]
101 suffix. See nss-resolve(8) for the recommended configuration.
103 * systemd-logind may now be restarted without losing state. It stores
104 the file descriptors for devices it manages in the system manager
105 using the FDSTORE= mechanism. Please note that further changes in
106 other components may be required to make use of this (for example
107 Xorg has code to listen for stops of systemd-logind and terminate
108 itself when logind is stopped or restarted, in order to avoid using
109 stale file descriptors for graphical devices, which is now
110 counterproductive and must be reverted in order for restarts of
111 systemd-logind to be safe. See
112 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
114 * All kernel install plugins are called with the environment variable
115 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
116 /etc/machine-id. If the file is missing or empty, the variable is
117 empty and BOOT_DIR_ABS is the path of a temporary directory which is
118 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
119 is empty, all plugins should not put anything in BOOT_DIR_ABS.
121 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
122 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
123 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
124 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
125 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
126 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
127 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
128 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
129 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
130 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
131 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
132 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
133 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
134 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
135 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
136 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
137 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
138 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
139 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
140 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
141 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
142 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
143 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
144 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
145 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
146 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
147 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
148 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
149 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
156 * The "hybrid" control group mode has been modified to improve
157 compatibility with "legacy" cgroups-v1 setups. Specifically, the
158 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
159 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
160 cgroups-v1 hierarchy), the only externally visible change being that
161 the cgroups-v2 hierarchy is also mounted, to
162 /sys/fs/cgroup/unified. This should provide a large degree of
163 compatibility with "legacy" cgroups-v1, while taking benefit of the
164 better management capabilities of cgroups-v2.
166 * The default control group setup mode may be selected both a boot-time
167 via a set of kernel command line parameters (specifically:
168 systemd.unified_cgroup_hierarchy= and
169 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
170 default selected on the configure command line
171 (--with-default-hierarchy=). The upstream default is "hybrid"
172 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
173 this will change in a future systemd version to be "unified" (pure
174 cgroups-v2 mode). The third option for the compile time option is
175 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
176 distributions to default to "hybrid" mode for release distributions,
177 starting with v233. We recommend "unified" for development
178 distributions (specifically: distributions such as Fedora's rawhide)
179 as that's where things are headed in the long run. Use "legacy" for
180 greatest stability and compatibility only.
182 * Note one current limitation of "unified" and "hybrid" control group
183 setup modes: the kernel currently does not permit the systemd --user
184 instance (i.e. unprivileged code) to migrate processes between two
185 disconnected cgroup subtrees, even if both are managed and owned by
186 the user. This effectively means "systemd-run --user --scope" doesn't
187 work when invoked from outside of any "systemd --user" service or
188 scope. Specifically, it is not supported from session scopes. We are
189 working on fixing this in a future systemd version. (See #3388 for
190 further details about this.)
192 * DBus policy files are now installed into /usr rather than /etc. Make
193 sure your system has dbus >= 1.9.18 running before upgrading to this
194 version, or override the install path with --with-dbuspolicydir= .
196 * All python scripts shipped with systemd (specifically: the various
197 tests written in Python) now require Python 3.
199 * systemd unit tests can now run standalone (without the source or
200 build directories), and can be installed into /usr/lib/systemd/tests/
201 with 'make install-tests'.
203 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
204 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
207 * Support for the %c, %r, %R specifiers in unit files has been
208 removed. Specifiers are not supposed to be dependent on configuration
209 in the unit file itself (so that they resolve the same regardless
210 where used in the unit files), but these specifiers were influenced
211 by the Slice= option.
213 * The shell invoked by debug-shell.service now defaults to /bin/sh in
214 all cases. If distributions want to use a different shell for this
215 purpose (for example Fedora's /sbin/sushell) they need to specify
216 this explicitly at configure time using --with-debug-shell=.
218 * The confirmation spawn prompt has been reworked to offer the
221 (c)ontinue, proceed without asking anymore
222 (D)ump, show the state of the unit
223 (f)ail, don't execute the command and pretend it failed
225 (i)nfo, show a short summary of the unit
226 (j)obs, show jobs that are in progress
227 (s)kip, don't execute the command and pretend it succeeded
228 (y)es, execute the command
230 The 'n' choice for the confirmation spawn prompt has been removed,
231 because its meaning was confusing.
233 The prompt may now also be redirected to an alternative console by
234 specifying the console as parameter to systemd.confirm_spawn=.
236 * Services of Type=notify require a READY=1 notification to be sent
237 during startup. If no such message is sent, the service now fails,
238 even if the main process exited with a successful exit code.
240 * Services that fail to start up correctly now always have their
241 ExecStopPost= commands executed. Previously, they'd enter "failed"
242 state directly, without executing these commands.
244 * The option MulticastDNS= of network configuration files has acquired
245 an actual implementation. With MulticastDNS=yes a host can resolve
246 names of remote hosts and reply to mDNS A and AAAA requests.
248 * When units are about to be started an additional check is now done to
249 ensure that all dependencies of type BindsTo= (when used in
250 combination with After=) have been started.
252 * systemd-analyze gained a new verb "syscall-filter" which shows which
253 system call groups are defined for the SystemCallFilter= unit file
254 setting, and which system calls they contain.
256 * A new system call filter group "@filesystem" has been added,
257 consisting of various file system related system calls. Group
258 "@reboot" has been added, covering reboot, kexec and shutdown related
259 calls. Finally, group "@swap" has been added covering swap
260 configuration related calls.
262 * A new unit file option RestrictNamespaces= has been added that may be
263 used to restrict access to the various process namespace types the
264 Linux kernel provides. Specifically, it may be used to take away the
265 right for a service unit to create additional file system, network,
266 user, and other namespaces. This sandboxing option is particularly
267 relevant due to the high amount of recently discovered namespacing
268 related vulnerabilities in the kernel.
270 * systemd-udev's .link files gained support for a new AutoNegotiation=
271 setting for configuring Ethernet auto-negotiation.
273 * systemd-networkd's .network files gained support for a new
274 ListenPort= setting in the [DHCP] section to explicitly configure the
275 UDP client port the DHCP client shall listen on.
277 * .network files gained a new Unmanaged= boolean setting for explicitly
278 excluding one or more interfaces from management by systemd-networkd.
280 * The systemd-networkd ProxyARP= option has been renamed to
281 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
282 renamed to ReduceARPProxy=. The old names continue to be available
285 * systemd-networkd gained support for configuring IPv6 Proxy NDP
286 addresses via the new IPv6ProxyNDPAddress= .network file setting.
288 * systemd-networkd's bonding device support gained support for two new
289 configuration options ActiveSlave= and PrimarySlave=.
291 * The various options in the [Match] section of .network files gained
292 support for negative matching.
294 * New systemd-specific mount options are now understood in /etc/fstab:
296 x-systemd.mount-timeout= may be used to configure the maximum
297 permitted runtime of the mount command.
299 x-systemd.device-bound may be set to bind a mount point to its
300 backing device unit, in order to automatically remove a mount point
301 if its backing device is unplugged. This option may also be
302 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
303 on the block device, which is now automatically set for all CDROM
304 drives, so that mounted CDs are automatically unmounted when they are
305 removed from the drive.
307 x-systemd.after= and x-systemd.before= may be used to explicitly
308 order a mount after or before another unit or mount point.
310 * Enqueued start jobs for device units are now automatically garbage
311 collected if there are no jobs waiting for them anymore.
313 * systemctl list-jobs gained two new switches: with --after, for every
314 queued job the jobs it's waiting for are shown; with --before the
315 jobs which it's blocking are shown.
317 * systemd-nspawn gained support for ephemeral boots from disk images
318 (or in other words: --ephemeral and --image= may now be
319 combined). Moreover, ephemeral boots are now supported for normal
320 directories, even if the backing file system is not btrfs. Of course,
321 if the file system does not support file system snapshots or
322 reflinks, the initial copy operation will be relatively expensive, but
323 this should still be suitable for many use cases.
325 * Calendar time specifications in .timer units now support
326 specifications relative to the end of a month by using "~" instead of
327 "-" as separator between month and day. For example, "*-02~03" means
328 "the third last day in February". In addition a new syntax for
329 repeated events has been added using the "/" character. For example,
330 "9..17/2:00" means "every two hours from 9am to 5pm".
332 * systemd-socket-proxyd gained a new parameter --connections-max= for
333 configuring the maximum number of concurrent connections.
335 * sd-id128 gained a new API for generating unique IDs for the host in a
336 way that does not leak the machine ID. Specifically,
337 sd_id128_get_machine_app_specific() derives an ID based on the
338 machine ID a in well-defined, non-reversible, stable way. This is
339 useful whenever an identifier for the host is needed but where the
340 identifier shall not be useful to identify the system beyond the
341 scope of the application itself. (Internally this uses HMAC-SHA256 as
342 keyed hash function using the machine ID as input.)
344 * NotifyAccess= gained a new supported value "exec". When set
345 notifications are accepted from all processes systemd itself invoked,
346 including all control processes.
348 * .nspawn files gained support for defining overlay mounts using the
349 Overlay= and OverlayReadOnly= options. Previously this functionality
350 was only available on the systemd-nspawn command line.
352 * systemd-nspawn's --bind= and --overlay= options gained support for
353 bind/overlay mounts whose source lies within the container tree by
354 prefixing the source path with "+".
356 * systemd-nspawn's --bind= and --overlay= options gained support for
357 automatically allocating a temporary source directory in /var/tmp
358 that is removed when the container dies. Specifically, if the source
359 directory is specified as empty string this mechanism is selected. An
360 example usage is --overlay=+/var::/var, which creates an overlay
361 mount based on the original /var contained in the image, overlayed
362 with a temporary directory in the host's /var/tmp. This way changes
363 to /var are automatically flushed when the container shuts down.
365 * systemd-nspawn --image= option does now permit raw file system block
366 devices (in addition to images containing partition tables, as
369 * The disk image dissection logic in systemd-nspawn gained support for
370 automatically setting up LUKS encrypted as well as Verity protected
371 partitions. When a container is booted from an encrypted image the
372 passphrase is queried at start-up time. When a container with Verity
373 data is started, the root hash is search in a ".roothash" file
374 accompanying the disk image (alternatively, pass the root hash via
375 the new --root-hash= command line option).
377 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
378 be used to dissect disk images the same way as systemd-nspawn does
379 it, following the Bootable Partition Specification. It may even be
380 used to mount disk images with complex partition setups (including
381 LUKS and Verity partitions) to a local host directory, in order to
382 inspect them. This tool is not considered public API (yet), and is
383 thus not installed into /usr/bin. Please do not rely on its
384 existence, since it might go away or be changed in later systemd
387 * A new generator "systemd-verity-generator" has been added, similar in
388 style to "systemd-cryptsetup-generator", permitting automatic setup of
389 Verity root partitions when systemd boots up. In order to make use of
390 this your partition setup should follow the Discoverable Partitions
391 Specification, and the GPT partition ID of the root file system
392 partition should be identical to the upper 128bit of the Verity root
393 hash. The GPT partition ID of the Verity partition protecting it
394 should be the lower 128bit of the Verity root hash. If the partition
395 image follows this model it is sufficient to specify a single
396 "roothash=" kernel command line argument to both configure which root
397 image and verity partition to use as well as the root hash for
398 it. Note that systemd-nspawn's Verity support follows the same
399 semantics, meaning that disk images with proper Verity data in place
400 may be booted in containers with systemd-nspawn as well as on
401 physical systems via the verity generator. Also note that the "mkosi"
402 tool available at https://github.com/systemd/mkosi has been updated
403 to generate Verity protected disk images following this scheme. In
404 fact, it has been updated to generate disk images that optionally
405 implement a complete UEFI SecureBoot trust chain, involving a signed
406 kernel and initrd image that incorporates such a root hash as well as
407 a Verity-enabled root partition.
409 * The hardware database (hwdb) udev supports has been updated to carry
410 accelerometer quirks.
412 * All system services are now run with a fresh kernel keyring set up
413 for them. The invocation ID is stored by default in it, thus
414 providing a safe, non-overridable way to determine the invocation
417 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
418 options for bind mounting arbitrary paths in a service-specific
419 way. When these options are used, arbitrary host or service files and
420 directories may be mounted to arbitrary locations in the service's
423 * Documentation has been added that lists all of systemd's low-level
424 environment variables:
426 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
428 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
429 whether a specific socket file descriptor matches a specified socket
432 * systemd-firstboot has been updated to check for the
433 systemd.firstboot= kernel command line option. It accepts a boolean
434 and when set to false the first boot questions are skipped.
436 * systemd-fstab-generator has been updated to check for the
437 systemd.volatile= kernel command line option, which either takes an
438 optional boolean parameter or the special value "state". If used the
439 system may be booted in a "volatile" boot mode. Specifically,
440 "systemd.volatile" is used, the root directory will be mounted as
441 tmpfs, and only /usr is mounted from the actual root file system. If
442 "systemd.volatile=state" is used, the root directory will be mounted
443 as usual, but /var is mounted as tmpfs. This concept provides similar
444 functionality as systemd-nspawn's --volatile= option, but provides it
445 on physical boots. Use this option for implementing stateless
446 systems, or testing systems with all state and/or configuration reset
447 to the defaults. (Note though that many distributions are not
448 prepared to boot up without a populated /etc or /var, though.)
450 * systemd-gpt-auto-generator gained support for LUKS encrypted root
451 partitions. Previously it only supported LUKS encrypted partitions
452 for all other uses, except for the root partition itself.
454 * Socket units gained support for listening on AF_VSOCK sockets for
455 communication in virtualized QEMU environments.
457 * The "configure" script gained a new option --with-fallback-hostname=
458 for specifying the fallback hostname to use if none is configured in
459 /etc/hostname. For example, by specifying
460 --with-fallback-hostname=fedora it is possible to default to a
461 hostname of "fedora" on pristine installations.
463 * systemd-cgls gained support for a new --unit= switch for listing only
464 the control groups of a specific unit. Similar --user-unit= has been
465 added for listing only the control groups of a specific user unit.
467 * systemd-mount gained a new --umount switch for unmounting a mount or
468 automount point (and all mount/automount points below it).
470 * systemd will now refuse full configuration reloads (via systemctl
471 daemon-reload and related calls) unless at least 16MiB of free space
472 are available in /run. This is a safety precaution in order to ensure
473 that generators can safely operate after the reload completed.
475 * A new unit file option RootImage= has been added, which has a similar
476 effect as RootDirectory= but mounts the service's root directory from
477 a disk image instead of plain directory. This logic reuses the same
478 image dissection and mount logic that systemd-nspawn already uses,
479 and hence supports any disk images systemd-nspawn supports, including
480 those following the Discoverable Partition Specification, as well as
481 Verity enabled images. This option enables systemd to run system
482 services directly off disk images acting as resource bundles,
483 possibly even including full integrity data.
485 * A new MountAPIVFS= unit file option has been added, taking a boolean
486 argument. If enabled /proc, /sys and /dev (collectively called the
487 "API VFS") will be mounted for the service. This is only relevant if
488 RootDirectory= or RootImage= is used for the service, as these mounts
489 are of course in place in the host mount namespace anyway.
491 * systemd-nspawn gained support for a new --pivot-root= switch. If
492 specified the root directory within the container image is pivoted to
493 the specified mount point, while the original root disk is moved to a
494 different place. This option enables booting of ostree images
495 directly with systemd-nspawn.
497 * The systemd build scripts will no longer complain if the NTP server
498 addresses are not changed from the defaults. Google now supports
499 these NTP servers officially. We still recommend downstreams to
500 properly register an NTP pool with the NTP pool project though.
502 * coredumpctl gained a new "--reverse" option for printing the list
503 of coredumps in reverse order.
505 * coredumpctl will now show additional information about truncated and
506 inaccessible coredumps, as well as coredumps that are still being
507 processed. It also gained a new --quiet switch for suppressing
508 additional informational message in its output.
510 * coredumpctl gained support for only showing coredumps newer and/or
511 older than specific timestamps, using the new --since= and --until=
512 options, reminiscent of journalctl's options by the same name.
514 * The systemd-coredump logic has been improved so that it may be reused
515 to collect backtraces in non-compiled languages, for example in
516 scripting languages such as Python.
518 * machinectl will now show the UID shift of local containers, if user
519 namespacing is enabled for them.
521 * systemd will now optionally run "environment generator" binaries at
522 configuration load time. They may be used to add environment
523 variables to the environment block passed to services invoked. One
524 user environment generator is shipped by default that sets up
525 environment variables based on files dropped into /etc/environment.d
526 and ~/.config/environment.d/.
528 * systemd-resolved now includes the new, recently published 2017 DNSSEC
531 * hostnamed has been updated to report a new chassis type of
532 "convertible" to cover "foldable" laptops that can both act as a
533 tablet and as a laptop, such as various Lenovo Yoga devices.
535 Contributions from: Adrián López, Alexander Galanin, Alexander
536 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
537 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
538 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
539 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
540 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
541 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
542 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
543 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
544 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
545 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
546 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
547 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
548 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
549 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
550 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
551 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
552 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
553 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
554 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
555 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
556 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
557 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
558 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
559 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
560 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
561 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
562 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
569 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
570 RestrictAddressFamilies= enabled. These sandboxing options should
571 generally be compatible with the various external udev call-out
572 binaries we are aware of, however there may be exceptions, in
573 particular when exotic languages for these call-outs are used. In
574 this case, consider turning off these settings locally.
576 * The new RemoveIPC= option can be used to remove IPC objects owned by
577 the user or group of a service when that service exits.
579 * The new ProtectKernelModules= option can be used to disable explicit
580 load and unload operations of kernel modules by a service. In
581 addition access to /usr/lib/modules is removed if this option is set.
583 * ProtectSystem= option gained a new value "strict", which causes the
584 whole file system tree with the exception of /dev, /proc, and /sys,
585 to be remounted read-only for a service.
587 * The new ProtectKernelTunables= option can be used to disable
588 modification of configuration files in /sys and /proc by a service.
589 Various directories and files are remounted read-only, so access is
590 restricted even if the file permissions would allow it.
592 * The new ProtectControlGroups= option can be used to disable write
593 access by a service to /sys/fs/cgroup.
595 * Various systemd services have been hardened with
596 ProtectKernelTunables=yes, ProtectControlGroups=yes,
597 RestrictAddressFamilies=.
599 * Support for dynamically creating users for the lifetime of a service
600 has been added. If DynamicUser=yes is specified, user and group IDs
601 will be allocated from the range 61184..65519 for the lifetime of the
602 service. They can be resolved using the new nss-systemd.so NSS
603 module. The module must be enabled in /etc/nsswitch.conf. Services
604 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
605 any resources allocated by the service will be cleaned up when the
606 service exits. They also have ProtectHome=read-only and
607 ProtectSystem=strict enabled, so they are not able to make any
608 permanent modifications to the system.
610 * The nss-systemd module also always resolves root and nobody, making
611 it possible to have no /etc/passwd or /etc/group files in minimal
612 container or chroot environments.
614 * Services may be started with their own user namespace using the new
615 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
616 under which the service is running are mapped. All other users are
619 * Support for the cgroup namespace has been added to systemd-nspawn. If
620 supported by kernel, the container system started by systemd-nspawn
621 will have its own view of the cgroup hierarchy. This new behaviour
622 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
624 * The new MemorySwapMax= option can be used to limit the maximum swap
625 usage under the unified cgroup hierarchy.
627 * Support for the CPU controller in the unified cgroup hierarchy has
628 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
629 options. This controller requires out-of-tree patches for the kernel
630 and the support is provisional.
632 * Mount and automount units may now be created transiently
633 (i.e. dynamically at runtime via the bus API, instead of requiring
634 unit files in the file system).
636 * systemd-mount is a new tool which may mount file systems – much like
637 mount(8), optionally pulling in additional dependencies through
638 transient .mount and .automount units. For example, this tool
639 automatically runs fsck on a backing block device before mounting,
640 and allows the automount logic to be used dynamically from the
641 command line for establishing mount points. This tool is particularly
642 useful when dealing with removable media, as it will ensure fsck is
643 run – if necessary – before the first access and that the file system
644 is quickly unmounted after each access by utilizing the automount
645 logic. This maximizes the chance that the file system on the
646 removable media stays in a clean state, and if it isn't in a clean
647 state is fixed automatically.
649 * LazyUnmount=yes option for mount units has been added to expose the
650 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
653 * /efi will be used as the mount point of the EFI boot partition, if
654 the directory is present, and the mount point was not configured
655 through other means (e.g. fstab). If /efi directory does not exist,
656 /boot will be used as before. This makes it easier to automatically
657 mount the EFI partition on systems where /boot is used for something
660 * When operating on GPT disk images for containers, systemd-nspawn will
661 now mount the ESP to /boot or /efi according to the same rules as PID
662 1 running on a host. This allows tools like "bootctl" to operate
663 correctly within such containers, in order to make container images
664 bootable on physical systems.
666 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
668 * Two new user session targets have been added to support running
669 graphical sessions under the systemd --user instance:
670 graphical-session.target and graphical-session-pre.target. See
671 systemd.special(7) for a description of how those targets should be
674 * The vconsole initialization code has been significantly reworked to
675 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
676 support unicode keymaps. Font and keymap configuration will now be
677 copied to all allocated virtual consoles.
679 * FreeBSD's bhyve virtualization is now detected.
681 * Information recorded in the journal for core dumps now includes the
682 contents of /proc/mountinfo and the command line of the process at
683 the top of the process hierarchy (which is usually the init process
686 * systemd-journal-gatewayd learned the --directory= option to serve
687 files from the specified location.
689 * journalctl --root=… can be used to peruse the journal in the
690 /var/log/ directories inside of a container tree. This is similar to
691 the existing --machine= option, but does not require the container to
694 * The hardware database has been extended to support
695 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
698 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
699 specify the click rate for mice which include a horizontal wheel with
700 a click rate that is different than the one for the vertical wheel.
702 * systemd-run gained a new --wait option that makes service execution
703 synchronous. (Specifically, the command will not return until the
704 specified service binary exited.)
706 * systemctl gained a new --wait option that causes the start command to
707 wait until the units being started have terminated again.
709 * A new journal output mode "short-full" has been added which displays
710 timestamps with abbreviated English day names and adds a timezone
711 suffix. Those timestamps include more information than the default
712 "short" output mode, and can be passed directly to journalctl's
713 --since= and --until= options.
715 * /etc/resolv.conf will be bind-mounted into containers started by
716 systemd-nspawn, if possible, so any changes to resolv.conf contents
717 are automatically propagated to the container.
719 * The number of instances for socket-activated services originating
720 from a single IP address can be limited with
721 MaxConnectionsPerSource=, extending the existing setting of
724 * systemd-networkd gained support for vcan ("Virtual CAN") interface
727 * .netdev and .network configuration can now be extended through
730 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
731 Segmentation Offload, Generic Receive Offload, Large Receive Offload
732 can be enabled and disabled using the new UDPSegmentationOffload=,
733 TCPSegmentationOffload=, GenericSegmentationOffload=,
734 GenericReceiveOffload=, LargeReceiveOffload= options in the
735 [Link] section of .link files.
737 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
738 Port VLAN ID can be configured for bridge devices using the new STP=,
739 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
740 section of .netdev files.
742 * The route table to which routes received over DHCP or RA should be
743 added can be configured with the new RouteTable= option in the [DHCP]
744 and [IPv6AcceptRA] sections of .network files.
746 * The Address Resolution Protocol can be disabled on links managed by
747 systemd-networkd using the ARP=no setting in the [Link] section of
750 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
751 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
752 encode information about the result and exit codes of the current
753 service runtime cycle.
755 * systemd-sysctl will now configure kernel parameters in the order
756 they occur in the configuration files. This matches what sysctl
757 has been traditionally doing.
759 * kernel-install "plugins" that are executed to perform various
760 tasks after a new kernel is added and before an old one is removed
761 can now return a special value to terminate the procedure and
762 prevent any later plugins from running.
764 * Journald's SplitMode=login setting has been deprecated. It has been
765 removed from documentation, and its use is discouraged. In a future
766 release it will be completely removed, and made equivalent to current
767 default of SplitMode=uid.
769 * Storage=both option setting in /etc/systemd/coredump.conf has been
770 removed. With fast LZ4 compression storing the core dump twice is not
773 * The --share-system systemd-nspawn option has been replaced with an
774 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
775 this functionality is discouraged. In addition the variables
776 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
777 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
778 individual namespaces.
780 * "machinectl list" now shows the IP address of running containers in
781 the output, as well as OS release information.
783 * "loginctl list" now shows the TTY of each session in the output.
785 * sd-bus gained new API calls sd_bus_track_set_recursive(),
786 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
787 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
788 tracking objects in a "recursive" mode, where a single client can be
789 counted multiple times, if it takes multiple references.
791 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
792 sd_bus_get_exit_on_disconnect(). They may be used to to make a
793 process using sd-bus automatically exit if the bus connection is
796 * Bus clients of the service manager may now "pin" loaded units into
797 memory, by taking an explicit reference on them. This is useful to
798 ensure the client can retrieve runtime data about the service even
799 after the service completed execution. Taking such a reference is
800 available only for privileged clients and should be helpful to watch
801 running services in a race-free manner, and in particular collect
802 information about exit statuses and results.
804 * The nss-resolve module has been changed to strictly return UNAVAIL
805 when communication via D-Bus with resolved failed, and NOTFOUND when
806 a lookup completed but was negative. This means it is now possible to
807 neatly configure fallbacks using nsswitch.conf result checking
808 expressions. Taking benefit of this, the new recommended
809 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
811 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
813 * A new setting CtrlAltDelBurstAction= has been added to
814 /etc/systemd/system.conf which may be used to configure the precise
815 behaviour if the user on the console presses Ctrl-Alt-Del more often
816 than 7 times in 2s. Previously this would unconditionally result in
817 an expedited, immediate reboot. With this new setting the precise
818 operation may be configured in more detail, and also turned off
821 * In .netdev files two new settings RemoteChecksumTx= and
822 RemoteChecksumRx= are now understood that permit configuring the
823 remote checksumming logic for VXLAN networks.
825 * The service manager learnt a new "invocation ID" concept for invoked
826 services. Each runtime cycle of a service will get a new invocation
827 ID (a 128bit random UUID) assigned that identifies the current
828 run of the service uniquely and globally. A new invocation ID
829 is generated each time a service starts up. The journal will store
830 the invocation ID of a service along with any logged messages, thus
831 making the invocation ID useful for matching the online runtime of a
832 service with the offline log data it generated in a safe way without
833 relying on synchronized timestamps. In many ways this new service
834 invocation ID concept is similar to the kernel's boot ID concept that
835 uniquely and globally identifies the runtime of each boot. The
836 invocation ID of a service is passed to the service itself via an
837 environment variable ($INVOCATION_ID). A new bus call
838 GetUnitByInvocationID() has been added that is similar to GetUnit()
839 but instead of retrieving the bus path for a unit by its name
840 retrieves it by its invocation ID. The returned path is valid only as
841 long as the passed invocation ID is current.
843 * systemd-resolved gained a new "DNSStubListener" setting in
844 resolved.conf. It either takes a boolean value or the special values
845 "udp" and "tcp", and configures whether to enable the stub DNS
846 listener on 127.0.0.53:53.
848 * IP addresses configured via networkd may now carry additional
849 configuration settings supported by the kernel. New options include:
850 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
851 PrefixRoute=, AutoJoin=.
853 * The PAM configuration fragment file for "user@.service" shipped with
854 systemd (i.e. the --user instance of systemd) has been stripped to
855 the minimum necessary to make the system boot. Previously, it
856 contained Fedora-specific stanzas that did not apply to other
857 distributions. It is expected that downstream distributions add
858 additional configuration lines, matching their needs to this file,
859 using it only as rough template of what systemd itself needs. Note
860 that this reduced fragment does not even include an invocation of
861 pam_limits which most distributions probably want to add, even though
862 systemd itself does not need it. (There's also the new build time
863 option --with-pamconfdir=no to disable installation of the PAM
866 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
867 capability is now also dropped from its set (in addition to
868 CAP_SYS_MKNOD as before).
870 * In service unit files it is now possible to connect a specific named
871 file descriptor with stdin/stdout/stdout of an executed service. The
872 name may be specified in matching .socket units using the
873 FileDescriptorName= setting.
875 * A number of journal settings may now be configured on the kernel
876 command line. Specifically, the following options are now understood:
877 systemd.journald.max_level_console=,
878 systemd.journald.max_level_store=,
879 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
880 systemd.journald.max_level_wall=.
882 * "systemctl is-enabled --full" will now show by which symlinks a unit
883 file is enabled in the unit dependency tree.
885 * Support for VeraCrypt encrypted partitions has been added to the
886 "cryptsetup" logic and /etc/crypttab.
888 * systemd-detect-virt gained support for a new --private-users switch
889 that checks whether the invoking processes are running inside a user
890 namespace. Similar, a new special value "private-users" for the
891 existing ConditionVirtualization= setting has been added, permitting
892 skipping of specific units in user namespace environments.
894 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
895 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
896 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
897 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
898 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
899 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
900 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
901 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
902 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
903 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
904 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
905 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
906 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
907 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
908 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
909 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
910 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
911 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
912 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
913 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
914 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
915 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
916 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
917 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
918 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
919 Jędrzejewski-Szmek, Zeal Jagannatha
921 — Santa Fe, 2016-11-03
925 * In service units the various ExecXYZ= settings have been extended
926 with an additional special character as first argument of the
927 assigned value: if the character '+' is used the specified command
928 line it will be run with full privileges, regardless of User=,
929 Group=, CapabilityBoundingSet= and similar options. The effect is
930 similar to the existing PermissionsStartOnly= option, but allows
931 configuration of this concept for each executed command line
934 * Services may now alter the service watchdog timeout at runtime by
935 sending a WATCHDOG_USEC= message via sd_notify().
937 * MemoryLimit= and related unit settings now optionally take percentage
938 specifications. The percentage is taken relative to the amount of
939 physical memory in the system (or in case of containers, the assigned
940 amount of memory). This allows scaling service resources neatly with
941 the amount of RAM available on the system. Similarly, systemd-logind's
942 RuntimeDirectorySize= option now also optionally takes percentage
945 * In similar fashion TasksMax= takes percentage values now, too. The
946 value is taken relative to the configured maximum number of processes
947 on the system. The per-service task maximum has been changed to 15%
948 using this functionality. (Effectively this is an increase of 512 →
949 4915 for service units, given the kernel's default pid_max setting.)
951 * Calendar time specifications in .timer units now understand a ".."
952 syntax for time ranges. Example: "4..7:10" may now be used for
953 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
956 * The InaccessableDirectories=, ReadOnlyDirectories= and
957 ReadWriteDirectories= unit file settings have been renamed to
958 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
959 applied to all kinds of file nodes, and not just directories, with
960 the exception of symlinks. Specifically these settings may now be
961 used on block and character device nodes, UNIX sockets and FIFOS as
962 well as regular files. The old names of these settings remain
963 available for compatibility.
965 * systemd will now log about all service processes it kills forcibly
966 (using SIGKILL) because they remained after the clean shutdown phase
967 of the service completed. This should help identifying services that
968 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
969 systemd-logind's configuration a similar log message is generated for
970 processes killed at the end of each session due to this setting.
972 * systemd will now set the $JOURNAL_STREAM environment variable for all
973 services whose stdout/stderr are connected to the Journal (which
974 effectively means by default: all services). The variable contains
975 the device and inode number of the file descriptor used for
976 stdout/stderr. This may be used by invoked programs to detect whether
977 their stdout/stderr is connected to the Journal, in which case they
978 can switch over to direct Journal communication, thus being able to
979 pass extended, structured metadata along with their log messages. As
980 one example, this is now used by glib's logging primitives.
982 * When using systemd's default tmp.mount unit for /tmp, the mount point
983 will now be established with the "nosuid" and "nodev" options. This
984 avoids privilege escalation attacks that put traps and exploits into
985 /tmp. However, this might cause problems if you e. g. put container
986 images or overlays into /tmp; if you need this, override tmp.mount's
987 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
990 * systemd now supports the "memory" cgroup controller also on
993 * The systemd-cgtop tool now optionally takes a control group path as
994 command line argument. If specified, the control group list shown is
995 limited to subgroups of that group.
997 * The SystemCallFilter= unit file setting gained support for
998 pre-defined, named system call filter sets. For example
999 SystemCallFilter=@clock is now an effective way to make all clock
1000 changing-related system calls unavailable to a service. A number of
1001 similar pre-defined groups are defined. Writing system call filters
1002 for system services is simplified substantially with this new
1003 concept. Accordingly, all of systemd's own, long-running services now
1004 enable system call filtering based on this, by default.
1006 * A new service setting MemoryDenyWriteExecute= has been added, taking
1007 a boolean value. If turned on, a service may no longer create memory
1008 mappings that are writable and executable at the same time. This
1009 enhances security for services where this is enabled as it becomes
1010 harder to dynamically write and then execute memory in exploited
1011 service processes. This option has been enabled for all of systemd's
1012 own long-running services.
1014 * A new RestrictRealtime= service setting has been added, taking a
1015 boolean argument. If set the service's processes may no longer
1016 acquire realtime scheduling. This improves security as realtime
1017 scheduling may otherwise be used to easily freeze the system.
1019 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1020 value. This may be used for requesting that the system manager inside
1021 of the container reports start-up completion to nspawn which then
1022 propagates this notification further to the service manager
1023 supervising nspawn itself. A related option NotifyReady= in .nspawn
1024 files has been added too. This functionality allows ordering of the
1025 start-up of multiple containers using the usual systemd ordering
1028 * machinectl gained a new command "stop" that is an alias for
1031 * systemd-resolved gained support for contacting DNS servers on
1032 link-local IPv6 addresses.
1034 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1035 its caches. A method call for requesting the same operation has been
1036 added to the bus API too, and is made available via "systemd-resolve
1039 * systemd-resolve gained a new --status switch. If passed a brief
1040 summary of the used DNS configuration with per-interface information
1043 * resolved.conf gained a new Cache= boolean option, defaulting to
1044 on. If turned off local DNS caching is disabled. This comes with a
1045 performance penalty in particular when DNSSEC is enabled. Note that
1046 resolved disables its internal caching implicitly anyway, when the
1047 configured DNS server is on a host-local IP address such as ::1 or
1048 127.0.0.1, thus automatically avoiding double local caching.
1050 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1051 for DNS requests. This improves compatibility with local programs
1052 that do not use the libc NSS or systemd-resolved's bus APIs for name
1053 resolution. This minimal DNS service is only available to local
1054 programs and does not implement the full DNS protocol, but enough to
1055 cover local DNS clients. A new, static resolv.conf file, listing just
1056 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1057 now recommended to make /etc/resolv.conf a symlink to this file in
1058 order to route all DNS lookups to systemd-resolved, regardless if
1059 done via NSS, the bus API or raw DNS packets. Note that this local
1060 DNS service is not as fully featured as the libc NSS or
1061 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1062 used to deliver link-local address information (as this implies
1063 sending a local interface index along), LLMNR/mDNS support via this
1064 interface is severely restricted. It is thus strongly recommended for
1065 all applications to use the libc NSS API or native systemd-resolved
1068 * systemd-networkd's bridge support learned a new setting
1069 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1070 in .network files has been added for configuring VLAN bridging in
1071 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1073 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1074 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1075 now be acquired without relying on DHCPv6. Two new options
1076 UseDomains= and UseDNS= have been added to configure this behaviour.
1078 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1079 renamed IPv6AcceptRA=, without altering its behaviour. The old
1080 setting name remains available for compatibility reasons.
1082 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1083 Key=, InputKey= and OutputKey=.
1085 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1086 interface configuration.
1088 * "systemctl edit" may now be used to create new unit files by
1089 specifying the --force switch.
1091 * sd-event gained a new function sd_event_get_iteration() for
1092 requesting the current iteration counter of the event loop. It starts
1093 at zero and is increased by one with each event loop iteration.
1095 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1096 file. It can be used in lieu of %systemd_requires in packages which
1097 don't use any systemd functionality and are intended to be installed
1098 in minimal containers without systemd present. This macro provides
1099 ordering dependencies to ensure that if the package is installed in
1100 the same rpm transaction as systemd, systemd will be installed before
1101 the scriptlets for the package are executed, allowing unit presets
1104 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1105 been added to simplify packaging of generators.
1107 * The os-release file gained VERSION_CODENAME field for the
1108 distribution nickname (e.g. VERSION_CODENAME=woody).
1110 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1111 can be set to disable parsing of metadata and the creation
1112 of persistent symlinks for that device.
1114 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1115 to make them available to logged-in users has been reverted.
1117 * Much of the common code of the various systemd components is now
1118 built into an internal shared library libsystemd-shared-231.so
1119 (incorporating the systemd version number in the name, to be updated
1120 with future releases) that the components link to. This should
1121 decrease systemd footprint both in memory during runtime and on
1122 disk. Note that the shared library is not for public use, and is
1123 neither API not ABI stable, but is likely to change with every new
1124 released update. Packagers need to make sure that binaries
1125 linking to libsystemd-shared.so are updated in step with the
1128 * Configuration for "mkosi" is now part of the systemd
1129 repository. mkosi is a tool to easily build legacy-free OS images,
1130 and is available on github: https://github.com/systemd/mkosi. If
1131 "mkosi" is invoked in the build tree a new raw OS image is generated
1132 incorporating the systemd sources currently being worked on and a
1133 clean, fresh distribution installation. The generated OS image may be
1134 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1135 UEFI PC. This functionality is particularly useful to easily test
1136 local changes made to systemd in a pristine, defined environment. See
1137 HACKING for details.
1139 * configure learned the --with-support-url= option to specify the
1140 distribution's bugtracker.
1142 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1143 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1144 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1145 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1146 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1147 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1148 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1149 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1150 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1151 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1152 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1153 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1154 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1155 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1156 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1157 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1158 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1159 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1160 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1162 — Berlin, 2016-07-25
1166 * DNSSEC is now turned on by default in systemd-resolved (in
1167 "allow-downgrade" mode), but may be turned off during compile time by
1168 passing "--with-default-dnssec=no" to "configure" (and of course,
1169 during runtime with DNSSEC= in resolved.conf). We recommend
1170 downstreams to leave this on at least during development cycles and
1171 report any issues with the DNSSEC logic upstream. We are very
1172 interested in collecting feedback about the DNSSEC validator and its
1173 limitations in the wild. Note however, that DNSSEC support is
1174 probably nothing downstreams should turn on in stable distros just
1175 yet, as it might create incompatibilities with a few DNS servers and
1176 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1177 automatically whenever we detect such incompatible setups, but there
1178 might be systems we do not cover yet. Hence: please help us testing
1179 the DNSSEC code, leave this on where you can, report back, but then
1180 again don't consider turning this on in your stable, LTS or
1181 production release just yet. (Note that you have to enable
1182 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1183 and its DNSSEC mode for host name resolution from local
1186 * systemd-resolve conveniently resolves DANE records with the --tlsa
1187 option and OPENPGPKEY records with the --openpgp option. It also
1188 supports dumping raw DNS record data via the new --raw= switch.
1190 * systemd-logind will now by default terminate user processes that are
1191 part of the user session scope unit (session-XX.scope) when the user
1192 logs out. This behavior is controlled by the KillUserProcesses=
1193 setting in logind.conf, and the previous default of "no" is now
1194 changed to "yes". This means that user sessions will be properly
1195 cleaned up after, but additional steps are necessary to allow
1196 intentionally long-running processes to survive logout.
1198 While the user is logged in at least once, user@.service is running,
1199 and any service that should survive the end of any individual login
1200 session can be started at a user service or scope using systemd-run.
1201 systemd-run(1) man page has been extended with an example which shows
1202 how to run screen in a scope unit underneath user@.service. The same
1203 command works for tmux.
1205 After the user logs out of all sessions, user@.service will be
1206 terminated too, by default, unless the user has "lingering" enabled.
1207 To effectively allow users to run long-term tasks even if they are
1208 logged out, lingering must be enabled for them. See loginctl(1) for
1209 details. The default polkit policy was modified to allow users to
1210 set lingering for themselves without authentication.
1212 Previous defaults can be restored at compile time by the
1213 --without-kill-user-processes option to "configure".
1215 * systemd-logind gained new configuration settings SessionsMax= and
1216 InhibitorsMax=, both with a default of 8192. It will not register new
1217 user sessions or inhibitors above this limit.
1219 * systemd-logind will now reload configuration on SIGHUP.
1221 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1222 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1223 enable. Also, support for the "io" cgroup controller in the unified
1224 hierarchy has been added, so that the "memory", "pids" and "io" are
1225 now the controllers that are supported on the unified hierarchy.
1227 WARNING: it is not possible to use previous systemd versions with
1228 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1229 is necessary to also update systemd in the initramfs if using the
1230 unified hierarchy. An updated SELinux policy is also required.
1232 * LLDP support has been extended, and both passive (receive-only) and
1233 active (sender) modes are supported. Passive mode ("routers-only") is
1234 enabled by default in systemd-networkd. Active LLDP mode is enabled
1235 by default for containers on the internal network. The "networkctl
1236 lldp" command may be used to list information gathered. "networkctl
1237 status" will also show basic LLDP information on connected peers now.
1239 * The IAID and DUID unique identifier sent in DHCP requests may now be
1240 configured for the system and each .network file managed by
1241 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1243 * systemd-networkd gained support for configuring proxy ARP support for
1244 each interface, via the ProxyArp= setting in .network files. It also
1245 gained support for configuring the multicast querier feature of
1246 bridge devices, via the new MulticastQuerier= setting in .netdev
1247 files. Similarly, snooping on the IGMP traffic can be controlled
1248 via the new setting MulticastSnooping=.
1250 A new setting PreferredLifetime= has been added for addresses
1251 configured in .network file to configure the lifetime intended for an
1254 The systemd-networkd DHCP server gained the option EmitRouter=, which
1255 defaults to yes, to configure whether the DHCP Option 3 (Router)
1258 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1259 systemd-socket-activate and installed into /usr/bin. It is now fully
1262 * systemd-journald now uses separate threads to flush changes to disk
1263 when closing journal files, thus reducing impact of slow disk I/O on
1264 logging performance.
1266 * The sd-journal API gained two new calls
1267 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1268 can be used to open journal files using file descriptors instead of
1269 file or directory paths. sd_journal_open_container() has been
1270 deprecated, sd_journal_open_directory_fd() should be used instead
1271 with the flag SD_JOURNAL_OS_ROOT.
1273 * journalctl learned a new output mode "-o short-unix" that outputs log
1274 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1275 UTC). It also gained support for a new --no-hostname setting to
1276 suppress the hostname column in the family of "short" output modes.
1278 * systemd-ask-password now optionally skips printing of the password to
1279 stdout with --no-output which can be useful in scripts.
1281 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1282 (devices tagged with ID_MAKER_TOOL) are now tagged with
1283 "uaccess" and are available to logged in users.
1285 * The DeviceAllow= unit setting now supports specifiers (with "%").
1287 * "systemctl show" gained a new --value switch, which allows print a
1288 only the contents of a specific unit property, without also printing
1289 the property's name. Similar support was added to "show*" verbs
1290 of loginctl and machinectl that output "key=value" lists.
1292 * A new unit type "generated" was added for files dynamically generated
1293 by generator tools. Similarly, a new unit type "transient" is used
1294 for unit files created using the runtime API. "systemctl enable" will
1295 refuse to operate on such files.
1297 * A new command "systemctl revert" has been added that may be used to
1298 revert to the vendor version of a unit file, in case local changes
1299 have been made by adding drop-ins or overriding the unit file.
1301 * "machinectl clean" gained a new verb to automatically remove all or
1302 just hidden container images.
1304 * systemd-tmpfiles gained support for a new line type "e" for emptying
1305 directories, if they exist, without creating them if they don't.
1307 * systemd-nspawn gained support for automatically patching the UID/GIDs
1308 of the owners and the ACLs of all files and directories in a
1309 container tree to match the UID/GID user namespacing range selected
1310 for the container invocation. This mode is enabled via the new
1311 --private-users-chown switch. It also gained support for
1312 automatically choosing a free, previously unused UID/GID range when
1313 starting a container, via the new --private-users=pick setting (which
1314 implies --private-users-chown). Together, these options for the first
1315 time make user namespacing for nspawn containers fully automatic and
1316 thus deployable. The systemd-nspawn@.service template unit file has
1317 been changed to use this functionality by default.
1319 * systemd-nspawn gained a new --network-zone= switch, that allows
1320 creating ad-hoc virtual Ethernet links between multiple containers,
1321 that only exist as long as at least one container referencing them is
1322 running. This allows easy connecting of multiple containers with a
1323 common link that implements an Ethernet broadcast domain. Each of
1324 these network "zones" may be named relatively freely by the user, and
1325 may be referenced by any number of containers, but each container may
1326 only reference one of these "zones". On the lower level, this is
1327 implemented by an automatically managed bridge network interface for
1328 each zone, that is created when the first container referencing its
1329 zone is created and removed when the last one referencing its zone
1332 * The default start timeout may now be configured on the kernel command
1333 line via systemd.default_timeout_start_sec=. It was already
1334 configurable via the DefaultTimeoutStartSec= option in
1335 /etc/systemd/system.conf.
1337 * Socket units gained a new TriggerLimitIntervalSec= and
1338 TriggerLimitBurst= setting to configure a limit on the activation
1339 rate of the socket unit.
1341 * The LimitNICE= setting now optionally takes normal UNIX nice values
1342 in addition to the raw integer limit value. If the specified
1343 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1344 value is understood as UNIX nice value. If not prefixed like this it
1345 is understood as raw RLIMIT_NICE limit.
1347 * Note that the effect of the PrivateDevices= unit file setting changed
1348 slightly with this release: the per-device /dev file system will be
1349 mounted read-only from this version on, and will have "noexec"
1350 set. This (minor) change of behavior might cause some (exceptional)
1351 legacy software to break, when PrivateDevices=yes is set for its
1352 service. Please leave PrivateDevices= off if you run into problems
1355 * systemd-bootchart has been split out to a separate repository:
1356 https://github.com/systemd/systemd-bootchart
1358 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1359 merged into the kernel in its current form.
1361 * The compatibility libraries libsystemd-daemon.so,
1362 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1363 which have been deprecated since systemd-209 have been removed along
1364 with the corresponding pkg-config files. All symbols provided by
1365 those libraries are provided by libsystemd.so.
1367 * The Capabilities= unit file setting has been removed (it is ignored
1368 for backwards compatibility). AmbientCapabilities= and
1369 CapabilityBoundingSet= should be used instead.
1371 * A new special target has been added, initrd-root-device.target,
1372 which creates a synchronization point for dependencies of the root
1373 device in early userspace. Initramfs builders must ensure that this
1374 target is now included in early userspace.
1376 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1377 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1378 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1379 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1380 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1381 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1382 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1383 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1384 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1385 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1386 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1387 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1388 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1389 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1390 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1391 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1392 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1393 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1394 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1395 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1396 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1397 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1398 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1399 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1400 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1403 — Fairfax, 2016-05-21
1407 * The systemd-resolved DNS resolver service has gained a substantial
1408 set of new features, most prominently it may now act as a DNSSEC
1409 validating stub resolver. DNSSEC mode is currently turned off by
1410 default, but is expected to be turned on by default in one of the
1411 next releases. For now, we invite everybody to test the DNSSEC logic
1412 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1413 service also gained a full set of D-Bus interfaces, including calls
1414 to configure DNS and DNSSEC settings per link (for use by external
1415 network management software). systemd-resolved and systemd-networkd
1416 now distinguish between "search" and "routing" domains. The former
1417 are used to qualify single-label names, the latter are used purely
1418 for routing lookups within certain domains to specific links.
1419 resolved now also synthesizes RRs for all entries from /etc/hosts.
1421 * The systemd-resolve tool (which is a client utility for
1422 systemd-resolved) has been improved considerably and is now fully
1423 supported and documented. Hence it has moved from /usr/lib/systemd to
1426 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1429 * The coredump collection logic has been reworked: when a coredump is
1430 collected it is now written to disk, compressed and processed
1431 (including stacktrace extraction) from a new instantiated service
1432 systemd-coredump@.service, instead of directly from the
1433 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1434 processing large coredumps can take up a substantial amount of
1435 resources and time, and this previously happened entirely outside of
1436 systemd's service supervision. With the new logic the core_pattern
1437 hook only does minimal metadata collection before passing off control
1438 to the new instantiated service, which is configured with a time
1439 limit, a nice level and other settings to minimize negative impact on
1440 the rest of the system. Also note that the new logic will honour the
1441 RLIMIT_CORE setting of the crashed process, which now allows users
1442 and processes to turn off coredumping for their processes by setting
1445 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1446 and all forked processes by default. Previously, PID 1 would leave
1447 the setting at "0" for all processes, as set by the kernel. Note that
1448 the resource limit traditionally has no effect on the generated
1449 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1450 logic is used. Since the limit is now honoured (see above) its
1451 default has been changed so that the coredumping logic is enabled by
1452 default for all processes, while allowing specific opt-out.
1454 * When the stacktrace is extracted from processes of system users, this
1455 is now done as "systemd-coredump" user, in order to sandbox this
1456 potentially security sensitive parsing operation. (Note that when
1457 processing coredumps of normal users this is done under the user ID
1458 of process that crashed, as before.) Packagers should take notice
1459 that it is now necessary to create the "systemd-coredump" system user
1460 and group at package installation time.
1462 * The systemd-activate socket activation testing tool gained support
1463 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1464 and --seqpacket switches. It also has been extended to support both
1465 new-style and inetd-style file descriptor passing. Use the new
1466 --inetd switch to request inetd-style file descriptor passing.
1468 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1469 variable, which takes a boolean value. If set to false, ANSI color
1470 output is disabled in the tools even when run on a terminal that
1473 * The VXLAN support in networkd now supports two new settings
1474 DestinationPort= and PortRange=.
1476 * A new systemd.machine_id= kernel command line switch has been added,
1477 that may be used to set the machine ID in /etc/machine-id if it is
1478 not initialized yet. This command line option has no effect if the
1479 file is already initialized.
1481 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1482 specified command line as PID 2 rather than PID 1 in the
1483 container. In this mode PID 1 is a minimal stub init process that
1484 implements the special POSIX and Linux semantics of PID 1 regarding
1485 signal and child process management. Note that this stub init process
1486 is implemented in nspawn itself and requires no support from the
1487 container image. This new logic is useful to support running
1488 arbitrary commands in the container, as normal processes are
1489 generally not prepared to run as PID 1.
1491 * systemd-nspawn gained a new --chdir= switch for setting the current
1492 working directory for the process started in the container.
1494 * "journalctl /dev/sda" will now output all kernel log messages for
1495 specified device from the current boot, in addition to all devices
1496 that are parents of it. This should make log output about devices
1497 pretty useful, as long as kernel drivers attach enough metadata to
1498 the log messages. (The usual SATA drivers do.)
1500 * The sd-journal API gained two new calls
1501 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1502 that report whether log data from /run or /var has been found.
1504 * journalctl gained a new switch "--fields" that prints all journal
1505 record field names currently in use in the journal. This is backed
1506 by two new sd-journal API calls sd_journal_enumerate_fields() and
1507 sd_journal_restart_fields().
1509 * Most configurable timeouts in systemd now expect an argument of
1510 "infinity" to turn them off, instead of "0" as before. The semantics
1511 from now on is that a timeout of "0" means "now", and "infinity"
1512 means "never". To maintain backwards compatibility, "0" continues to
1513 turn off previously existing timeout settings.
1515 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1516 try-reload-or-restart" to clarify what it actually does: the "try"
1517 logic applies to both reloading and restarting, not just restarting.
1518 The old name continues to be accepted for compatibility.
1520 * On boot-up, when PID 1 detects that the system clock is behind the
1521 release date of the systemd version in use, the clock is now set
1522 to the latter. Previously, this was already done in timesyncd, in order
1523 to avoid running with clocks set to the various clock epochs such as
1524 1902, 1938 or 1970. With this change the logic is now done in PID 1
1525 in addition to timesyncd during early boot-up, so that it is enforced
1526 before the first process is spawned by systemd. Note that the logic
1527 in timesyncd remains, as it is more comprehensive and ensures
1528 clock monotonicity by maintaining a persistent timestamp file in
1529 /var. Since /var is generally not available in earliest boot or the
1530 initrd, this part of the logic remains in timesyncd, and is not done
1533 * Support for tweaking details in net_cls.class_id through the
1534 NetClass= configuration directive has been removed, as the kernel
1535 people have decided to deprecate that controller in cgroup v2.
1536 Userspace tools such as nftables are moving over to setting rules
1537 that are specific to the full cgroup path of a task, which obsoletes
1538 these controllers anyway. The NetClass= directive is kept around for
1539 legacy compatibility reasons. For a more in-depth description of the
1540 kernel change, please refer to the respective upstream commit:
1542 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1544 * A new service setting RuntimeMaxSec= has been added that may be used
1545 to specify a maximum runtime for a service. If the timeout is hit, the
1546 service is terminated and put into a failure state.
1548 * A new service setting AmbientCapabilities= has been added. It allows
1549 configuration of additional Linux process capabilities that are
1550 passed to the activated processes. This is only available on very
1553 * The process resource limit settings in service units may now be used
1554 to configure hard and soft limits individually.
1556 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1557 expose support for gcc's __attribute__((cleanup())) C extension.
1558 Specifically, for many object destructor functions alternative
1559 versions have been added that have names suffixed with "p" and take a
1560 pointer to a pointer to the object to destroy, instead of just a
1561 pointer to the object itself. This is useful because these destructor
1562 functions may be used directly as parameters to the cleanup
1563 construct. Internally, systemd has been a heavy user of this GCC
1564 extension for a long time, and with this change similar support is
1565 now available to consumers of the library outside of systemd. Note
1566 that by using this extension in your sources compatibility with old
1567 and strictly ANSI compatible C compilers is lost. However, all gcc or
1568 LLVM versions of recent years support this extension.
1570 * Timer units gained support for a new setting RandomizedDelaySec= that
1571 allows configuring some additional randomized delay to the configured
1572 time. This is useful to spread out timer events to avoid load peaks in
1573 clusters or larger setups.
1575 * Calendar time specifications now support sub-second accuracy.
1577 * Socket units now support listening on SCTP and UDP-lite protocol
1580 * The sd-event API now comes with a full set of man pages.
1582 * Older versions of systemd contained experimental support for
1583 compressing journal files and coredumps with the LZ4 compressor that
1584 was not compatible with the lz4 binary (due to API limitations of the
1585 lz4 library). This support has been removed; only support for files
1586 compatible with the lz4 binary remains. This LZ4 logic is now
1587 officially supported and no longer considered experimental.
1589 * The dkr image import logic has been removed again from importd. dkr's
1590 micro-services focus doesn't fit into the machine image focus of
1591 importd, and quickly got out of date with the upstream dkr API.
1593 * Creation of the /run/lock/lockdev/ directory was dropped from
1594 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1595 been available for many years. If you still need this, you need to
1596 create your own tmpfiles.d config file with:
1598 d /run/lock/lockdev 0775 root lock -
1600 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1601 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1602 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1603 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1604 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1605 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1606 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1607 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1608 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1609 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1610 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1611 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1612 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1613 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1614 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1615 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1616 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1617 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1618 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1620 — Berlin, 2016-02-11
1624 * A number of properties previously only settable in unit
1625 files are now also available as properties to set when
1626 creating transient units programmatically via the bus, as it
1627 is exposed with systemd-run's --property=
1628 setting. Specifically, these are: SyslogIdentifier=,
1629 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1630 EnvironmentFile=, ReadWriteDirectories=,
1631 ReadOnlyDirectories=, InaccessibleDirectories=,
1632 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1634 * When creating transient services via the bus API it is now
1635 possible to pass in a set of file descriptors to use as
1636 STDIN/STDOUT/STDERR for the invoked process.
1638 * Slice units may now be created transiently via the bus APIs,
1639 similar to the way service and scope units may already be
1640 created transiently.
1642 * Wherever systemd expects a calendar timestamp specification
1643 (like in journalctl's --since= and --until= switches) UTC
1644 timestamps are now supported. Timestamps suffixed with "UTC"
1645 are now considered to be in Universal Time Coordinated
1646 instead of the local timezone. Also, timestamps may now
1647 optionally be specified with sub-second accuracy. Both of
1648 these additions also apply to recurring calendar event
1649 specification, such as OnCalendar= in timer units.
1651 * journalctl gained a new "--sync" switch that asks the
1652 journal daemon to write all so far unwritten log messages to
1653 disk and sync the files, before returning.
1655 * systemd-tmpfiles learned two new line types "q" and "Q" that
1656 operate like "v", but also set up a basic btrfs quota
1657 hierarchy when used on a btrfs file system with quota
1660 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1661 instead of a subvolume (even on a btrfs file system) if the
1662 root directory is a plain directory, and not a
1663 subvolume. This should simplify things with certain chroot()
1664 environments which are not aware of the concept of btrfs
1667 * systemd-detect-virt gained a new --chroot switch to detect
1668 whether execution takes place in a chroot() environment.
1670 * CPUAffinity= now takes CPU index ranges in addition to
1673 * The various memory-related resource limit settings (such as
1674 LimitAS=) now understand the usual K, M, G, ... suffixes to
1675 the base of 1024 (IEC). Similar, the time-related resource
1676 limit settings understand the usual min, h, day, ...
1679 * There's a new system.conf setting DefaultTasksMax= to
1680 control the default TasksMax= setting for services and
1681 scopes running on the system. (TasksMax= is the primary
1682 setting that exposes the "pids" cgroup controller on systemd
1683 and was introduced in the previous systemd release.) The
1684 setting now defaults to 512, which means services that are
1685 not explicitly configured otherwise will only be able to
1686 create 512 processes or threads at maximum, from this
1687 version on. Note that this means that thread- or
1688 process-heavy services might need to be reconfigured to set
1689 TasksMax= to a higher value. It is sufficient to set
1690 TasksMax= in these specific unit files to a higher value, or
1691 even "infinity". Similar, there's now a logind.conf setting
1692 UserTasksMax= that defaults to 4096 and limits the total
1693 number of processes or tasks each user may own
1694 concurrently. nspawn containers also have the TasksMax=
1695 value set by default now, to 8192. Note that all of this
1696 only has an effect if the "pids" cgroup controller is
1697 enabled in the kernel. The general benefit of these changes
1698 should be a more robust and safer system, that provides a
1699 certain amount of per-service fork() bomb protection.
1701 * systemd-nspawn gained the new --network-veth-extra= switch
1702 to define additional and arbitrarily-named virtual Ethernet
1703 links between the host and the container.
1705 * A new service execution setting PassEnvironment= has been
1706 added that allows importing select environment variables
1707 from PID1's environment block into the environment block of
1710 * Timer units gained support for a new RemainAfterElapse=
1711 setting which takes a boolean argument. It defaults to on,
1712 exposing behaviour unchanged to previous releases. If set to
1713 off, timer units are unloaded after they elapsed if they
1714 cannot elapse again. This is particularly useful for
1715 transient timer units, which shall not stay around longer
1716 than until they first elapse.
1718 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1719 default now (the kernel default is 16). This is beneficial
1720 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1721 allows substantially larger numbers of queued
1722 datagrams. This should increase the capability of systemd to
1723 parallelize boot-up, as logging and sd_notify() are unlikely
1724 to stall execution anymore. If you need to change the value
1725 from the new defaults, use the usual sysctl.d/ snippets.
1727 * The compression framing format used by the journal or
1728 coredump processing has changed to be in line with what the
1729 official LZ4 tools generate. LZ4 compression support in
1730 systemd was considered unsupported previously, as the format
1731 was not compatible with the normal tools. With this release
1732 this has changed now, and it is hence safe for downstream
1733 distributions to turn it on. While not compressing as well
1734 as the XZ, LZ4 is substantially faster, which makes
1735 it a good default choice for the compression logic in the
1736 journal and in coredump handling.
1738 * Any reference to /etc/mtab has been dropped from
1739 systemd. The file has been obsolete since a while, but
1740 systemd refused to work on systems where it was incorrectly
1741 set up (it should be a symlink or non-existent). Please make
1742 sure to update to util-linux 2.27.1 or newer in conjunction
1743 with this systemd release, which also drops any reference to
1744 /etc/mtab. If you maintain a distribution make sure that no
1745 software you package still references it, as this is a
1746 likely source of bugs. There's also a glibc bug pending,
1747 asking for removal of any reference to this obsolete file:
1749 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1751 Note that only util-linux versions built with
1752 --enable-libmount-force-mountinfo are supported.
1754 * Support for the ".snapshot" unit type has been removed. This
1755 feature turned out to be little useful and little used, and
1756 has now been removed from the core and from systemctl.
1758 * The dependency types RequiresOverridable= and
1759 RequisiteOverridable= have been removed from systemd. They
1760 have been used only very sparingly to our knowledge and
1761 other options that provide a similar effect (such as
1762 systemctl --mode=ignore-dependencies) are much more useful
1763 and commonly used. Moreover, they were only half-way
1764 implemented as the option to control behaviour regarding
1765 these dependencies was never added to systemctl. By removing
1766 these dependency types the execution engine becomes a bit
1767 simpler. Unit files that use these dependencies should be
1768 changed to use the non-Overridable dependency types
1769 instead. In fact, when parsing unit files with these
1770 options, that's what systemd will automatically convert them
1771 too, but it will also warn, asking users to fix the unit
1772 files accordingly. Removal of these dependency types should
1773 only affect a negligible number of unit files in the wild.
1775 * Behaviour of networkd's IPForward= option changed
1776 (again). It will no longer maintain a per-interface setting,
1777 but propagate one way from interfaces where this is enabled
1778 to the global kernel setting. The global setting will be
1779 enabled when requested by a network that is set up, but
1780 never be disabled again. This change was made to make sure
1781 IPv4 and IPv6 behaviour regarding packet forwarding is
1782 similar (as the Linux IPv6 stack does not support
1783 per-interface control of this setting) and to minimize
1786 * In unit files the behaviour of %u, %U, %h, %s has
1787 changed. These specifiers will now unconditionally resolve
1788 to the various user database fields of the user that the
1789 systemd instance is running as, instead of the user
1790 configured in the specific unit via User=. Note that this
1791 effectively doesn't change much, as resolving of these
1792 specifiers was already turned off in the --system instance
1793 of systemd, as we cannot do NSS lookups from PID 1. In the
1794 --user instance of systemd these specifiers where correctly
1795 resolved, but hardly made any sense, since the user instance
1796 lacks privileges to do user switches anyway, and User= is
1797 hence useless. Moreover, even in the --user instance of
1798 systemd behaviour was awkward as it would only take settings
1799 from User= assignment placed before the specifier into
1800 account. In order to unify and simplify the logic around
1801 this the specifiers will now always resolve to the
1802 credentials of the user invoking the manager (which in case
1803 of PID 1 is the root user).
1805 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1806 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1807 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1808 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1809 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1810 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1811 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1812 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1813 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1814 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1815 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1816 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1817 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1818 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1821 — Berlin, 2015-11-18
1825 * systemd now depends on util-linux v2.27. More specifically,
1826 the newly added mount monitor feature in libmount now
1827 replaces systemd's former own implementation.
1829 * libmount mandates /etc/mtab not to be regular file, and
1830 systemd now enforces this condition at early boot.
1831 /etc/mtab has been deprecated and warned about for a very
1832 long time, so systems running systemd should already have
1833 stopped having this file around as anything else than a
1834 symlink to /proc/self/mounts.
1836 * Support for the "pids" cgroup controller has been added. It
1837 allows accounting the number of tasks in a cgroup and
1838 enforcing limits on it. This adds two new setting
1839 TasksAccounting= and TasksMax= to each unit, as well as a
1840 global option DefaultTasksAccounting=.
1842 * Support for the "net_cls" cgroup controller has been added.
1843 It allows assigning a net class ID to each task in the
1844 cgroup, which can then be used in firewall rules and traffic
1845 shaping configurations. Note that the kernel netfilter net
1846 class code does not currently work reliably for ingress
1847 packets on unestablished sockets.
1849 This adds a new config directive called NetClass= to CGroup
1850 enabled units. Allowed values are positive numbers for fixed
1851 assignments and "auto" for picking a free value
1854 * 'systemctl is-system-running' now returns 'offline' if the
1855 system is not booted with systemd. This command can now be
1856 used as a substitute for 'systemd-notify --booted'.
1858 * Watchdog timeouts have been increased to 3 minutes for all
1859 in-tree service files. Apparently, disk IO issues are more
1860 frequent than we hoped, and user reported >1 minute waiting
1863 * 'machine-id-commit' functionality has been merged into
1864 'machine-id-setup --commit'. The separate binary has been
1867 * The WorkingDirectory= directive in unit files may now be set
1868 to the special value '~'. In this case, the working
1869 directory is set to the home directory of the user
1870 configured in User=.
1872 * "machinectl shell" will now open the shell in the home
1873 directory of the selected user by default.
1875 * The CrashChVT= configuration file setting is renamed to
1876 CrashChangeVT=, following our usual logic of not
1877 abbreviating unnecessarily. The old directive is still
1878 supported for compat reasons. Also, this directive now takes
1879 an integer value between 1 and 63, or a boolean value. The
1880 formerly supported '-1' value for disabling stays around for
1883 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1884 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1885 RootDirectory= properties can now be set for transient
1888 * The systemd-analyze tool gained a new "set-log-target" verb
1889 to change the logging target the system manager logs to
1890 dynamically during runtime. This is similar to how
1891 "systemd-analyze set-log-level" already changes the log
1894 * In nspawn /sys is now mounted as tmpfs, with only a selected
1895 set of subdirectories mounted in from the real sysfs. This
1896 enhances security slightly, and is useful for ensuring user
1897 namespaces work correctly.
1899 * Support for USB FunctionFS activation has been added. This
1900 allows implementation of USB gadget services that are
1901 activated as soon as they are requested, so that they don't
1902 have to run continuously, similar to classic socket
1905 * The "systemctl exit" command now optionally takes an
1906 additional parameter that sets the exit code to return from
1907 the systemd manager when exiting. This is only relevant when
1908 running the systemd user instance, or when running the
1909 system instance in a container.
1911 * sd-bus gained the new API calls sd_bus_path_encode_many()
1912 and sd_bus_path_decode_many() that allow easy encoding and
1913 decoding of multiple identifier strings inside a D-Bus
1914 object path. Another new call sd_bus_default_flush_close()
1915 has been added to flush and close per-thread default
1918 * systemd-cgtop gained support for a -M/--machine= switch to
1919 show the control groups within a certain container only.
1921 * "systemctl kill" gained support for an optional --fail
1922 switch. If specified the requested operation will fail of no
1923 processes have been killed, because the unit had no
1924 processes attached, or similar.
1926 * A new systemd.crash_reboot=1 kernel command line option has
1927 been added that triggers a reboot after crashing. This can
1928 also be set through CrashReboot= in systemd.conf.
1930 * The RuntimeDirectory= setting now understands unit
1931 specifiers like %i or %f.
1933 * A new (still internal) library API sd-ipv4acd has been added,
1934 that implements address conflict detection for IPv4. It's
1935 based on code from sd-ipv4ll, and will be useful for
1936 detecting DHCP address conflicts.
1938 * File descriptors passed during socket activation may now be
1939 named. A new API sd_listen_fds_with_names() is added to
1940 access the names. The default names may be overridden,
1941 either in the .socket file using the FileDescriptorName=
1942 parameter, or by passing FDNAME= when storing the file
1943 descriptors using sd_notify().
1945 * systemd-networkd gained support for:
1947 - Setting the IPv6 Router Advertisement settings via
1948 IPv6AcceptRouterAdvertisements= in .network files.
1950 - Configuring the HelloTimeSec=, MaxAgeSec= and
1951 ForwardDelaySec= bridge parameters in .netdev files.
1953 - Configuring PreferredSource= for static routes in
1956 * The "ask-password" framework used to query for LUKS harddisk
1957 passwords or SSL passwords during boot gained support for
1958 caching passwords in the kernel keyring, if it is
1959 available. This makes sure that the user only has to type in
1960 a passphrase once if there are multiple objects to unlock
1961 with the same one. Previously, such password caching was
1962 available only when Plymouth was used; this moves the
1963 caching logic into the systemd codebase itself. The
1964 "systemd-ask-password" utility gained a new --keyname=
1965 switch to control which kernel keyring key to use for
1966 caching a password in. This functionality is also useful for
1967 enabling display managers such as gdm to automatically
1968 unlock the user's GNOME keyring if its passphrase, the
1969 user's password and the harddisk password are the same, if
1970 gdm-autologin is used.
1972 * When downloading tar or raw images using "machinectl
1973 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1974 file is now also downloaded, if it is available and stored
1975 next to the image file.
1977 * Units of type ".socket" gained a new boolean setting
1978 Writable= which is only useful in conjunction with
1979 ListenSpecial=. If true, enables opening the specified
1980 special file in O_RDWR mode rather than O_RDONLY mode.
1982 * systemd-rfkill has been reworked to become a singleton
1983 service that is activated through /dev/rfkill on each rfkill
1984 state change and saves the settings to disk. This way,
1985 systemd-rfkill is now compatible with devices that exist
1986 only intermittendly, and even restores state if the previous
1987 system shutdown was abrupt rather than clean.
1989 * The journal daemon gained support for vacuuming old journal
1990 files controlled by the number of files that shall remain,
1991 in addition to the already existing control by size and by
1992 date. This is useful as journal interleaving performance
1993 degrades with too many separate journal files, and allows
1994 putting an effective limit on them. The new setting defaults
1995 to 100, but this may be changed by setting SystemMaxFiles=
1996 and RuntimeMaxFiles= in journald.conf. Also, the
1997 "journalctl" tool gained the new --vacuum-files= switch to
1998 manually vacuum journal files to leave only the specified
1999 number of files in place.
2001 * udev will now create /dev/disk/by-path links for ATA devices
2002 on kernels where that is supported.
2004 * Galician, Serbian, Turkish and Korean translations were added.
2006 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2007 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2008 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2009 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2010 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2011 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2012 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2013 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2014 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2015 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2016 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2017 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2018 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2019 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2020 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2021 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2022 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2023 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2025 — Berlin, 2015-10-07
2029 * The DHCP implementation of systemd-networkd gained a set of
2032 - The DHCP server now supports emitting DNS and NTP
2033 information. It may be enabled and configured via
2034 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2035 and NTP information is enabled, but no servers are
2036 configured, the corresponding uplink information (if there
2037 is any) is propagated.
2039 - Server and client now support transmission and reception
2040 of timezone information. It can be configured via the
2041 newly introduced network options UseTimezone=,
2042 EmitTimezone=, and Timezone=. Transmission of timezone
2043 information is enabled between host and containers by
2044 default now: the container will change its local timezone
2045 to what the host has set.
2047 - Lease timeouts can now be configured via
2048 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2050 - The DHCP server improved on the stability of
2051 leases. Clients are more likely to get the same lease
2052 information back, even if the server loses state.
2054 - The DHCP server supports two new configuration options to
2055 control the lease address pool metrics, PoolOffset= and
2058 * The encapsulation limit of tunnels in systemd-networkd may
2059 now be configured via 'EncapsulationLimit='. It allows
2060 modifying the maximum additional levels of encapsulation
2061 that are permitted to be prepended to a packet.
2063 * systemd now supports the concept of user buses replacing
2064 session buses, if used with dbus-1.10 (and enabled via dbus
2065 --enable-user-session). It previously only supported this on
2066 kdbus-enabled systems, and this release expands this to
2067 'dbus-daemon' systems.
2069 * systemd-networkd now supports predictable interface names
2072 * systemd now optionally supports the new Linux kernel
2073 "unified" control group hierarchy. If enabled via the kernel
2074 command-line option 'systemd.unified_cgroup_hierarchy=1',
2075 systemd will try to mount the unified cgroup hierarchy
2076 directly on /sys/fs/cgroup. If not enabled, or not
2077 available, systemd will fall back to the legacy cgroup
2078 hierarchy setup, as before. Host system and containers can
2079 mix and match legacy and unified hierarchies as they
2080 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2081 environment variable to individually select the hierarchy to
2082 use for executed containers. By default, nspawn will use the
2083 unified hierarchy for the containers if the host uses the
2084 unified hierarchy, and the legacy hierarchy otherwise.
2085 Please note that at this point the unified hierarchy is an
2086 experimental kernel feature and is likely to change in one
2087 of the next kernel releases. Therefore, it should not be
2088 enabled by default in downstream distributions yet. The
2089 minimum required kernel version for the unified hierarchy to
2090 work is 4.2. Note that when the unified hierarchy is used
2091 for the first time delegated access to controllers is
2092 safe. Because of this systemd-nspawn containers will get
2093 access to controllers now, as will systemd user
2094 sessions. This means containers and user sessions may now
2095 manage their own resources, partitioning up what the system
2098 * A new special scope unit "init.scope" has been introduced
2099 that encapsulates PID 1 of the system. It may be used to
2100 determine resource usage and enforce resource limits on PID
2101 1 itself. PID 1 hence moved out of the root of the control
2104 * The cgtop tool gained support for filtering out kernel
2105 threads when counting tasks in a control group. Also, the
2106 count of processes is now recursively summed up by
2107 default. Two options -k and --recursive= have been added to
2108 revert to old behaviour. The tool has also been updated to
2109 work correctly in containers now.
2111 * systemd-nspawn's --bind= and --bind-ro= options have been
2112 extended to allow creation of non-recursive bind mounts.
2114 * libsystemd gained two new calls sd_pid_get_cgroup() and
2115 sd_peer_get_cgroup() which return the control group path of
2116 a process or peer of a connected AF_UNIX socket. This
2117 function call is particularly useful when implementing
2118 delegated subtrees support in the control group hierarchy.
2120 * The "sd-event" event loop API of libsystemd now supports
2121 correct dequeuing of real-time signals, without losing
2124 * When systemd requests a PolicyKit decision when managing
2125 units it will now add additional fields to the request,
2126 including unit name and desired operation. This enables more
2127 powerful PolicyKit policies, that make decisions depending
2128 on these parameters.
2130 * nspawn learnt support for .nspawn settings files, that may
2131 accompany the image files or directories of containers, and
2132 may contain additional settings for the container. This is
2133 an alternative to configuring container parameters via the
2134 nspawn command line.
2136 Contributions from: Cristian Rodríguez, Daniel Mack, David
2137 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2138 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2139 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2140 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2141 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2142 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2143 Andersen, Tom Gundersen, Torstein Husebø
2145 — Berlin, 2015-09-08
2149 * machinectl gained a new verb 'shell' which opens a fresh
2150 shell on the target container or the host. It is similar to
2151 the existing 'login' command of machinectl, but spawns the
2152 shell directly without prompting for username or
2153 password. The pseudo machine '.host' now refers to the local
2154 host and is used by default. Hence, 'machinectl shell' can
2155 be used as replacement for 'su -' which spawns a session as
2156 a fresh systemd unit in a way that is fully isolated from
2157 the originating session.
2159 * systemd-networkd learned to cope with private-zone DHCP
2160 options and allows other programs to query the values.
2162 * SELinux access control when enabling/disabling units is no
2163 longer enforced with this release. The previous
2164 implementation was incorrect, and a new corrected
2165 implementation is not yet available. As unit file operations
2166 are still protected via PolicyKit and D-Bus policy this is
2167 not a security problem. Yet, distributions which care about
2168 optimal SELinux support should probably not stabilize on
2171 * sd-bus gained support for matches of type "arg0has=", that
2172 test for membership of strings in string arrays sent in bus
2175 * systemd-resolved now dumps the contents of its DNS and LLMNR
2176 caches to the logs on reception of the SIGUSR1 signal. This
2177 is useful to debug DNS behaviour.
2179 * The coredumpctl tool gained a new --directory= option to
2180 operate on journal files in a specific directory.
2182 * "systemctl reboot" and related commands gained a new
2183 "--message=" option which may be used to set a free-text
2184 wall message when shutting down or rebooting the
2185 system. This message is also logged, which is useful for
2186 figuring out the reason for a reboot or shutdown a
2189 * The "systemd-resolve-host" tool's -i switch now takes
2190 network interface numbers as alternative to interface names.
2192 * A new unit file setting for services has been introduced:
2193 UtmpMode= allows configuration of how precisely systemd
2194 handles utmp and wtmp entries for the service if this is
2195 enabled. This allows writing services that appear similar to
2196 user sessions in the output of the "w", "who", "last" and
2199 * systemd-resolved will now locally synthesize DNS resource
2200 records for the "localhost" and "gateway" domains as well as
2201 the local hostname. This should ensure that clients querying
2202 RRs via resolved will get similar results as those going via
2203 NSS, if nss-myhostname is enabled.
2205 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2206 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2207 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2208 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2209 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2210 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2211 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2212 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2213 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2214 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2215 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2216 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2218 — Berlin, 2015-08-27
2222 * The systemd-efi-boot-generator functionality was merged into
2223 systemd-gpt-auto-generator.
2225 * systemd-networkd now supports Group Policy for vxlan
2226 devices. It can be enabled via the new boolean configuration
2227 option called 'GroupPolicyExtension='.
2229 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2230 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2231 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2233 — Berlin, 2015-07-31
2237 * The python-systemd code has been removed from the systemd repository.
2238 A new repository has been created which accommodates the code from
2239 now on, and we kindly ask distributions to create a separate package
2240 for this: https://github.com/systemd/python-systemd
2242 * The systemd daemon will now reload its main configuration
2243 (/etc/systemd/system.conf) on daemon-reload.
2245 * sd-dhcp now exposes vendor specific extensions via
2246 sd_dhcp_lease_get_vendor_specific().
2248 * systemd-networkd gained a number of new configuration options.
2250 - A new boolean configuration option for TAP devices called
2251 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2252 device, thus allowing to send and receive GSO packets.
2254 - A new tunnel configuration option called 'CopyDSCP='.
2255 If enabled, the DSCP field of ip6 tunnels is copied into the
2256 decapsulated packet.
2258 - A set of boolean bridge configuration options were added.
2259 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2260 and 'UnicastFlood=' are now parsed by networkd and applied to the
2261 respective bridge link device via the respective IFLA_BRPORT_*
2264 - A new string configuration option to override the hostname sent
2265 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2266 is true, networkd will use the configured hostname instead of the
2267 system hostname when sending DHCP requests.
2269 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2270 networkd will configure the IPv6 flow-label of the tunnel device
2271 according to RFC2460.
2273 - The 'macvtap' virtual network devices are now supported, similar to
2274 the already supported 'macvlan' devices.
2276 * systemd-resolved now implements RFC5452 to improve resilience against
2277 cache poisoning. Additionally, source port randomization is enabled
2278 by default to further protect against DNS spoofing attacks.
2280 * nss-mymachines now supports translating UIDs and GIDs of running
2281 containers with user-namespaces enabled. If a container 'foo'
2282 translates a host uid 'UID' to the container uid 'TUID', then
2283 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2284 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2285 mapped as 'vg-foo-TGID'.
2287 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2288 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2289 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2290 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2291 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2292 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2293 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2294 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2295 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2296 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2298 — Berlin, 2015-07-29
2302 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2303 There are no known issues with current sysfs, and udev does not need
2304 or should be used to work around such bugs.
2306 * udev does no longer enable USB HID power management. Several reports
2307 indicate, that some devices cannot handle that setting.
2309 * The udev accelerometer helper was removed. The functionality
2310 is now fully included in iio-sensor-proxy. But this means,
2311 older iio-sensor-proxy versions will no longer provide
2312 accelerometer/orientation data with this systemd version.
2313 Please upgrade iio-sensor-proxy to version 1.0.
2315 * networkd gained a new configuration option IPv6PrivacyExtensions=
2316 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2317 for Stateless Address") on selected networks.
2319 * For the sake of fewer build-time dependencies and less code in the
2320 main repository, the python bindings are about to be removed in the
2321 next release. A new repository has been created which accommodates
2322 the code from now on, and we kindly ask distributions to create a
2323 separate package for this. The removal will take place in v223.
2325 https://github.com/systemd/python-systemd
2327 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2328 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2329 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2330 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2331 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2332 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2333 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2334 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2335 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2336 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2338 — Berlin, 2015-07-07
2342 * The sd-bus.h and sd-event.h APIs have now been declared
2343 stable and have been added to the official interface of
2344 libsystemd.so. sd-bus implements an alternative D-Bus client
2345 library, that is relatively easy to use, very efficient and
2346 supports both classic D-Bus as well as kdbus as transport
2347 backend. sd-event is a generic event loop abstraction that
2348 is built around Linux epoll, but adds features such as event
2349 prioritization or efficient timer handling. Both APIs are good
2350 choices for C programs looking for a bus and/or event loop
2351 implementation that is minimal and does not have to be
2352 portable to other kernels.
2354 * kdbus support is no longer compile-time optional. It is now
2355 always built-in. However, it can still be disabled at
2356 runtime using the kdbus=0 kernel command line setting, and
2357 that setting may be changed to default to off, by specifying
2358 --disable-kdbus at build-time. Note though that the kernel
2359 command line setting has no effect if the kdbus.ko kernel
2360 module is not installed, in which case kdbus is (obviously)
2361 also disabled. We encourage all downstream distributions to
2362 begin testing kdbus by adding it to the kernel images in the
2363 development distributions, and leaving kdbus support in
2366 * The minimal required util-linux version has been bumped to
2369 * Support for chkconfig (--enable-chkconfig) was removed in
2370 favor of calling an abstraction tool
2371 /lib/systemd/systemd-sysv-install. This needs to be
2372 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2373 in README for details.
2375 * If there's a systemd unit and a SysV init script for the
2376 same service name, and the user executes "systemctl enable"
2377 for it (or a related call), then this will now enable both
2378 (or execute the related operation on both), not just the
2381 * The libudev API documentation has been converted from gtkdoc
2384 * gudev has been removed from the systemd tree, it is now an
2387 * The systemd-cgtop tool learnt a new --raw switch to generate
2388 "raw" (machine parsable) output.
2390 * networkd's IPForwarding= .network file setting learnt the
2391 new setting "kernel", which ensures that networkd does not
2392 change the IP forwarding sysctl from the default kernel
2395 * The systemd-logind bus API now exposes a new boolean
2396 property "Docked" that reports whether logind considers the
2397 system "docked", i.e. connected to a docking station or not.
2399 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2400 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2401 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2402 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2403 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2404 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2405 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2406 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2407 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2408 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2409 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2410 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2411 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2412 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2413 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2414 Fink, Zbigniew Jędrzejewski-Szmek
2416 — Berlin, 2015-06-19
2420 * The gudev library has been extracted into a separate repository
2421 available at: https://git.gnome.org/browse/libgudev/
2422 It is now managed as part of the Gnome project. Distributions
2423 are recommended to pass --disable-gudev to systemd and use
2424 gudev from the Gnome project instead. gudev is still included
2425 in systemd, for now. It will be removed soon, though. Please
2426 also see the announcement-thread on systemd-devel:
2427 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2429 * systemd now exposes a CPUUsageNSec= property for each
2430 service unit on the bus, that contains the overall consumed
2431 CPU time of a service (the sum of what each process of the
2432 service consumed). This value is only available if
2433 CPUAccounting= is turned on for a service, and is then shown
2434 in the "systemctl status" output.
2436 * Support for configuring alternative mappings of the old SysV
2437 runlevels to systemd targets has been removed. They are now
2438 hardcoded in a way that runlevels 2, 3, 4 all map to
2439 multi-user.target and 5 to graphical.target (which
2440 previously was already the default behaviour).
2442 * The auto-mounter logic gained support for mount point
2443 expiry, using a new TimeoutIdleSec= setting in .automount
2444 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2446 * The EFI System Partition (ESP) as mounted to /boot by
2447 systemd-efi-boot-generator will now be unmounted
2448 automatically after 2 minutes of not being used. This should
2449 minimize the risk of ESP corruptions.
2451 * New /etc/fstab options x-systemd.requires= and
2452 x-systemd.requires-mounts-for= are now supported to express
2453 additional dependencies for mounts. This is useful for
2454 journalling file systems that support external journal
2455 devices or overlay file systems that require underlying file
2456 systems to be mounted.
2458 * systemd does not support direct live-upgrades (via systemctl
2459 daemon-reexec) from versions older than v44 anymore. As no
2460 distribution we are aware of shipped such old versions in a
2461 stable release this should not be problematic.
2463 * When systemd forks off a new per-connection service instance
2464 it will now set the $REMOTE_ADDR environment variable to the
2465 remote IP address, and $REMOTE_PORT environment variable to
2466 the remote IP port. This behaviour is similar to the
2467 corresponding environment variables defined by CGI.
2469 * systemd-networkd gained support for uplink failure
2470 detection. The BindCarrier= option allows binding interface
2471 configuration dynamically to the link sense of other
2472 interfaces. This is useful to achieve behaviour like in
2475 * systemd-networkd gained support for configuring the DHCP
2476 client identifier to use when requesting leases.
2478 * systemd-networkd now has a per-network UseNTP= option to
2479 configure whether NTP server information acquired via DHCP
2480 is passed on to services like systemd-timesyncd.
2482 * systemd-networkd gained support for vti6 tunnels.
2484 * Note that systemd-networkd manages the sysctl variable
2485 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2486 it is configured for since v219. The variable controls IP
2487 forwarding, and is a per-interface alternative to the global
2488 /proc/sys/net/ipv[46]/ip_forward. This setting is
2489 configurable in the IPForward= option, which defaults to
2490 "no". This means if networkd is used for an interface it is
2491 no longer sufficient to set the global sysctl option to turn
2492 on IP forwarding! Instead, the .network file option
2493 IPForward= needs to be turned on! Note that the
2494 implementation of this behaviour was broken in v219 and has
2497 * Many bonding and vxlan options are now configurable in
2500 * systemd-nspawn gained a new --property= setting to set unit
2501 properties for the container scope. This is useful for
2502 setting resource parameters (e.g. "CPUShares=500") on
2503 containers started from the command line.
2505 * systemd-nspawn gained a new --private-users= switch to make
2506 use of user namespacing available on recent Linux kernels.
2508 * systemd-nspawn may now be called as part of a shell pipeline
2509 in which case the pipes used for stdin and stdout are passed
2510 directly to the process invoked in the container, without
2511 indirection via a pseudo tty.
2513 * systemd-nspawn gained a new switch to control the UNIX
2514 signal to use when killing the init process of the container
2517 * systemd-nspawn gained a new --overlay= switch for mounting
2518 overlay file systems into the container using the new kernel
2521 * When a container image is imported via systemd-importd and
2522 the host file system is not btrfs, a loopback block device
2523 file is created in /var/lib/machines.raw with a btrfs file
2524 system inside. It is then mounted to /var/lib/machines to
2525 enable btrfs features for container management. The loopback
2526 file and btrfs file system is grown as needed when container
2527 images are imported via systemd-importd.
2529 * systemd-machined/systemd-importd gained support for btrfs
2530 quota, to enforce container disk space limits on disk. This
2531 is exposed in "machinectl set-limit".
2533 * systemd-importd now can import containers from local .tar,
2534 .raw and .qcow2 images, and export them to .tar and .raw. It
2535 can also import dkr v2 images now from the network (on top
2538 * systemd-importd gained support for verifying downloaded
2539 images with gpg2 (previously only gpg1 was supported).
2541 * systemd-machined, systemd-logind, systemd: most bus calls
2542 are now accessible to unprivileged processes via
2543 PolicyKit. Also, systemd-logind will now allow users to kill
2544 their own sessions without further privileges or
2547 * systemd-shutdownd has been removed. This service was
2548 previously responsible for implementing scheduled shutdowns
2549 as exposed in /usr/bin/shutdown's time parameter. This
2550 functionality has now been moved into systemd-logind and is
2551 accessible via a bus interface.
2553 * "systemctl reboot" gained a new switch --firmware-setup that
2554 can be used to reboot into the EFI firmware setup, if that
2555 is available. systemd-logind now exposes an API on the bus
2556 to trigger such reboots, in case graphical desktop UIs want
2557 to cover this functionality.
2559 * "systemctl enable", "systemctl disable" and "systemctl mask"
2560 now support a new "--now" switch. If specified the units
2561 that are enabled will also be started, and the ones
2562 disabled/masked also stopped.
2564 * The Gummiboot EFI boot loader tool has been merged into
2565 systemd, and renamed to "systemd-boot". The bootctl tool has been
2566 updated to support systemd-boot.
2568 * An EFI kernel stub has been added that may be used to create
2569 kernel EFI binaries that contain not only the actual kernel,
2570 but also an initrd, boot splash, command line and OS release
2571 information. This combined binary can then be signed as a
2572 single image, so that the firmware can verify it all in one
2573 step. systemd-boot has special support for EFI binaries created
2574 like this and can extract OS release information from them
2575 and show them in the boot menu. This functionality is useful
2576 to implement cryptographically verified boot schemes.
2578 * Optional support has been added to systemd-fsck to pass
2579 fsck's progress report to an AF_UNIX socket in the file
2582 * udev will no longer create device symlinks for all block
2583 devices by default. A blacklist for excluding special block
2584 devices from this logic has been turned into a whitelist
2585 that requires picking block devices explicitly that require
2588 * A new (currently still internal) API sd-device.h has been
2589 added to libsystemd. This modernized API is supposed to
2590 replace libudev eventually. In fact, already much of libudev
2591 is now just a wrapper around sd-device.h.
2593 * A new hwdb database for storing metadata about pointing
2594 stick devices has been added.
2596 * systemd-tmpfiles gained support for setting file attributes
2597 similar to the "chattr" tool with new 'h' and 'H' lines.
2599 * systemd-journald will no longer unconditionally set the
2600 btrfs NOCOW flag on new journal files. This is instead done
2601 with tmpfiles snippet using the new 'h' line type. This
2602 allows easy disabling of this logic, by masking the
2603 journal-nocow.conf tmpfiles file.
2605 * systemd-journald will now translate audit message types to
2606 human readable identifiers when writing them to the
2607 journal. This should improve readability of audit messages.
2609 * The LUKS logic gained support for the offset= and skip=
2610 options in /etc/crypttab, as previously implemented by
2613 * /usr/lib/os-release gained a new optional field VARIANT= for
2614 distributions that support multiple variants (such as a
2615 desktop edition, a server edition, ...)
2617 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2618 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2619 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2620 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2621 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2622 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2623 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2624 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2625 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2626 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2627 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2628 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2629 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2630 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2631 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2632 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2633 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2634 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2635 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2636 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2637 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2638 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2639 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2640 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2641 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2642 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2643 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2645 — Berlin, 2015-05-22
2649 * Introduce a new API "sd-hwdb.h" for querying the hardware
2650 metadata database. With this minimal interface one can query
2651 and enumerate the udev hwdb, decoupled from the old libudev
2652 library. libudev's interface for this is now only a wrapper
2653 around sd-hwdb. A new tool systemd-hwdb has been added to
2654 interface with and update the database.
2656 * When any of systemd's tools copies files (for example due to
2657 tmpfiles' C lines) a btrfs reflink will attempted first,
2658 before bytewise copying is done.
2660 * systemd-nspawn gained a new --ephemeral switch. When
2661 specified a btrfs snapshot is taken of the container's root
2662 directory, and immediately removed when the container
2663 terminates again. Thus, a container can be started whose
2664 changes never alter the container's root directory, and are
2665 lost on container termination. This switch can also be used
2666 for starting a container off the root file system of the
2667 host without affecting the host OS. This switch is only
2668 available on btrfs file systems.
2670 * systemd-nspawn gained a new --template= switch. It takes the
2671 path to a container tree to use as template for the tree
2672 specified via --directory=, should that directory be
2673 missing. This allows instantiating containers dynamically,
2674 on first run. This switch is only available on btrfs file
2677 * When a .mount unit refers to a mount point on which multiple
2678 mounts are stacked, and the .mount unit is stopped all of
2679 the stacked mount points will now be unmounted until no
2680 mount point remains.
2682 * systemd now has an explicit notion of supported and
2683 unsupported unit types. Jobs enqueued for unsupported unit
2684 types will now fail with an "unsupported" error code. More
2685 specifically .swap, .automount and .device units are not
2686 supported in containers, .busname units are not supported on
2687 non-kdbus systems. .swap and .automount are also not
2688 supported if their respective kernel compile time options
2691 * machinectl gained support for two new "copy-from" and
2692 "copy-to" commands for copying files from a running
2693 container to the host or vice versa.
2695 * machinectl gained support for a new "bind" command to bind
2696 mount host directories into local containers. This is
2697 currently only supported for nspawn containers.
2699 * networkd gained support for configuring bridge forwarding
2700 database entries (fdb) from .network files.
2702 * A new tiny daemon "systemd-importd" has been added that can
2703 download container images in tar, raw, qcow2 or dkr formats,
2704 and make them available locally in /var/lib/machines, so
2705 that they can run as nspawn containers. The daemon can GPG
2706 verify the downloads (not supported for dkr, since it has no
2707 provisions for verifying downloads). It will transparently
2708 decompress bz2, xz, gzip compressed downloads if necessary,
2709 and restore sparse files on disk. The daemon uses privilege
2710 separation to ensure the actual download logic runs with
2711 fewer privileges than the daemon itself. machinectl has
2712 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2713 make the functionality of importd available to the
2714 user. With this in place the Fedora and Ubuntu "Cloud"
2715 images can be downloaded and booted as containers unmodified
2716 (the Fedora images lack the appropriate GPG signature files
2717 currently, so they cannot be verified, but this will change
2718 soon, hopefully). Note that downloading images is currently
2719 only fully supported on btrfs.
2721 * machinectl is now able to list container images found in
2722 /var/lib/machines, along with some metadata about sizes of
2723 disk and similar. If the directory is located on btrfs and
2724 quota is enabled, this includes quota display. A new command
2725 "image-status" has been added that shows additional
2726 information about images.
2728 * machinectl is now able to clone container images
2729 efficiently, if the underlying file system (btrfs) supports
2730 it, with the new "machinectl clone" command. It also
2731 gained commands for renaming and removing images, as well as
2732 marking them read-only or read-write (supported also on
2733 legacy file systems).
2735 * networkd gained support for collecting LLDP network
2736 announcements, from hardware that supports this. This is
2737 shown in networkctl output.
2739 * systemd-run gained support for a new -t (--pty) switch for
2740 invoking a binary on a pty whose input and output is
2741 connected to the invoking terminal. This allows executing
2742 processes as system services while interactively
2743 communicating with them via the terminal. Most interestingly
2744 this is supported across container boundaries. Invoking
2745 "systemd-run -t /bin/bash" is an alternative to running a
2746 full login session, the difference being that the former
2747 will not register a session, nor go through the PAM session
2750 * tmpfiles gained support for a new "v" line type for creating
2751 btrfs subvolumes. If the underlying file system is a legacy
2752 file system, this automatically degrades to creating a
2753 normal directory. Among others /var/lib/machines is now
2754 created like this at boot, should it be missing.
2756 * The directory /var/lib/containers/ has been deprecated and
2757 been replaced by /var/lib/machines. The term "machines" has
2758 been used in the systemd context as generic term for both
2759 VMs and containers, and hence appears more appropriate for
2760 this, as the directory can also contain raw images bootable
2763 * systemd-nspawn when invoked with -M but without --directory=
2764 or --image= is now capable of searching for the container
2765 root directory, subvolume or disk image automatically, in
2766 /var/lib/machines. systemd-nspawn@.service has been updated
2767 to make use of this, thus allowing it to be used for raw
2770 * A new machines.target unit has been introduced that is
2771 supposed to group all containers/VMs invoked as services on
2772 the system. systemd-nspawn@.service has been updated to
2773 integrate with that.
2775 * machinectl gained a new "start" command, for invoking a
2776 container as a service. "machinectl start foo" is mostly
2777 equivalent to "systemctl start systemd-nspawn@foo.service",
2778 but handles escaping in a nicer way.
2780 * systemd-nspawn will now mount most of the cgroupfs tree
2781 read-only into each container, with the exception of the
2782 container's own subtree in the name=systemd hierarchy.
2784 * journald now sets the special FS_NOCOW file flag for its
2785 journal files. This should improve performance on btrfs, by
2786 avoiding heavy fragmentation when journald's write-pattern
2787 is used on COW file systems. It degrades btrfs' data
2788 integrity guarantees for the files to the same levels as for
2789 ext3/ext4 however. This should be OK though as journald does
2790 its own data integrity checks and all its objects are
2791 checksummed on disk. Also, journald should handle btrfs disk
2792 full events a lot more gracefully now, by processing SIGBUS
2793 errors, and not relying on fallocate() anymore.
2795 * When journald detects that journal files it is writing to
2796 have been deleted it will immediately start new journal
2799 * systemd now provides a way to store file descriptors
2800 per-service in PID 1. This is useful for daemons to ensure
2801 that fds they require are not lost during a daemon
2802 restart. The fds are passed to the daemon on the next
2803 invocation in the same way socket activation fds are
2804 passed. This is now used by journald to ensure that the
2805 various sockets connected to all the system's stdout/stderr
2806 are not lost when journald is restarted. File descriptors
2807 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2808 an extension to sd_notify(). Note that a limit is enforced
2809 on the number of fds a service can store in PID 1, and it
2810 defaults to 0, so that no fds may be stored, unless this is
2811 explicitly turned on.
2813 * The default TERM variable to use for units connected to a
2814 terminal, when no other value is explicitly is set is now
2815 vt220 rather than vt102. This should be fairly safe still,
2816 but allows PgUp/PgDn work.
2818 * The /etc/crypttab option header= as known from Debian is now
2821 * "loginctl user-status" and "loginctl session-status" will
2822 now show the last 10 lines of log messages of the
2823 user/session following the status output. Similar,
2824 "machinectl status" will show the last 10 log lines
2825 associated with a virtual machine or container
2826 service. (Note that this is usually not the log messages
2827 done in the VM/container itself, but simply what the
2828 container manager logs. For nspawn this includes all console
2831 * "loginctl session-status" without further argument will now
2832 show the status of the session of the caller. Similar,
2833 "lock-session", "unlock-session", "activate",
2834 "enable-linger", "disable-linger" may now be called without
2835 session/user parameter in which case they apply to the
2836 caller's session/user.
2838 * An X11 session scriptlet is now shipped that uploads
2839 $DISPLAY and $XAUTHORITY into the environment of the systemd
2840 --user daemon if a session begins. This should improve
2841 compatibility with X11 enabled applications run as systemd
2844 * Generators are now subject to masking via /etc and /run, the
2845 same way as unit files.
2847 * networkd .network files gained support for configuring
2848 per-link IPv4/IPv6 packet forwarding as well as IPv4
2849 masquerading. This is by default turned on for veth links to
2850 containers, as registered by systemd-nspawn. This means that
2851 nspawn containers run with --network-veth will now get
2852 automatic routed access to the host's networks without any
2853 further configuration or setup, as long as networkd runs on
2856 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2857 or UDP posts of a container on the host. With this in place
2858 it is possible to run containers with private veth links
2859 (--network-veth), and have their functionality exposed on
2860 the host as if their services were running directly on the
2863 * systemd-nspawn's --network-veth switch now gained a short
2864 version "-n", since with the changes above it is now truly
2865 useful out-of-the-box. The systemd-nspawn@.service has been
2866 updated to make use of it too by default.
2868 * systemd-nspawn will now maintain a per-image R/W lock, to
2869 ensure that the same image is not started more than once
2870 writable. (It's OK to run an image multiple times
2871 simultaneously in read-only mode.)
2873 * systemd-nspawn's --image= option is now capable of
2874 dissecting and booting MBR and GPT disk images that contain
2875 only a single active Linux partition. Previously it
2876 supported only GPT disk images with proper GPT type
2877 IDs. This allows running cloud images from major
2878 distributions directly with systemd-nspawn, without
2881 * In addition to collecting mouse dpi data in the udev
2882 hardware database, there's now support for collecting angle
2883 information for mouse scroll wheels. The database is
2884 supposed to guarantee similar scrolling behavior on mice
2885 that it knows about. There's also support for collecting
2886 information about Touchpad types.
2888 * udev's input_id built-in will now also collect touch screen
2889 dimension data and attach it to probed devices.
2891 * /etc/os-release gained support for a Distribution Privacy
2894 * networkd gained support for creating "ipvlan", "gretap",
2895 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2897 * systemd-tmpfiles gained support for "a" lines for setting
2900 * systemd-nspawn will now mount /tmp in the container to
2901 tmpfs, automatically.
2903 * systemd now exposes the memory.usage_in_bytes cgroup
2904 attribute and shows it for each service in the "systemctl
2905 status" output, if available.
2907 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2908 immediate reboot is triggered. This useful if shutdown is
2909 hung and is unable to complete, to expedite the
2910 operation. Note that this kind of reboot will still unmount
2911 all file systems, and hence should not result in fsck being
2914 * A .device unit for an optical block device will now be
2915 considered active only when a medium is in the drive. Also,
2916 mount units are now bound to their backing devices thus
2917 triggering automatic unmounting when devices become
2918 unavailable. With this in place systemd will now
2919 automatically unmount left-over mounts when a CD-ROM is
2920 ejected or an USB stick is yanked from the system.
2922 * networkd-wait-online now has support for waiting for
2923 specific interfaces only (with globbing), and for giving up
2924 after a configurable timeout.
2926 * networkd now exits when idle. It will be automatically
2927 restarted as soon as interfaces show up, are removed or
2928 change state. networkd will stay around as long as there is
2929 at least one DHCP state machine or similar around, that keep
2932 * networkd may now configure IPv6 link-local addressing in
2933 addition to IPv4 link-local addressing.
2935 * The IPv6 "token" for use in SLAAC may now be configured for
2936 each .network interface in networkd.
2938 * Routes configured with networkd may now be assigned a scope
2941 * networkd's [Match] sections now support globbing and lists
2942 of multiple space-separated matches per item.
2944 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2945 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2946 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2947 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2948 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2949 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2950 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2951 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2952 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2953 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2954 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2955 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2956 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2957 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2958 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2959 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2960 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2961 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2962 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2963 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2964 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2965 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2966 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2967 Hoffmann, Zbigniew Jędrzejewski-Szmek
2969 — Berlin, 2015-02-16
2973 * When querying unit file enablement status (for example via
2974 "systemctl is-enabled"), a new state "indirect" is now known
2975 which indicates that a unit might not be enabled itself, but
2976 another unit listed in its Also= setting might be.
2978 * Similar to the various existing ConditionXYZ= settings for
2979 units, there are now matching AssertXYZ= settings. While
2980 failing conditions cause a unit to be skipped, but its job
2981 to succeed, failing assertions declared like this will cause
2982 a unit start operation and its job to fail.
2984 * hostnamed now knows a new chassis type "embedded".
2986 * systemctl gained a new "edit" command. When used on a unit
2987 file, this allows extending unit files with .d/ drop-in
2988 configuration snippets or editing the full file (after
2989 copying it from /usr/lib to /etc). This will invoke the
2990 user's editor (as configured with $EDITOR), and reload the
2991 modified configuration after editing.
2993 * "systemctl status" now shows the suggested enablement state
2994 for a unit, as declared in the (usually vendor-supplied)
2995 system preset files.
2997 * nss-myhostname will now resolve the single-label host name
2998 "gateway" to the locally configured default IP routing
2999 gateways, ordered by their metrics. This assigns a stable
3000 name to the used gateways, regardless which ones are
3001 currently configured. Note that the name will only be
3002 resolved after all other name sources (if nss-myhostname is
3003 configured properly) and should hence not negatively impact
3004 systems that use the single-label host name "gateway" in
3007 * systemd-inhibit now allows filtering by mode when listing
3010 * Scope and service units gained a new "Delegate" boolean
3011 property, which, when set, allows processes running inside the
3012 unit to further partition resources. This is primarily
3013 useful for systemd user instances as well as container
3016 * journald will now pick up audit messages directly from
3017 the kernel, and log them like any other log message. The
3018 audit fields are split up and fully indexed. This means that
3019 journalctl in many ways is now a (nicer!) alternative to
3020 ausearch, the traditional audit client. Note that this
3021 implements only a minimal audit client. If you want the
3022 special audit modes like reboot-on-log-overflow, please use
3023 the traditional auditd instead, which can be used in
3024 parallel to journald.
3026 * The ConditionSecurity= unit file option now understands the
3027 special string "audit" to check whether auditing is
3030 * journalctl gained two new commands --vacuum-size= and
3031 --vacuum-time= to delete old journal files until the
3032 remaining ones take up no more than the specified size on disk,
3033 or are not older than the specified time.
3035 * A new, native PPPoE library has been added to sd-network,
3036 systemd's library of light-weight networking protocols. This
3037 library will be used in a future version of networkd to
3038 enable PPPoE communication without an external pppd daemon.
3040 * The busctl tool now understands a new "capture" verb that
3041 works similar to "monitor", but writes a packet capture
3042 trace to STDOUT that can be redirected to a file which is
3043 compatible with libcap's capture file format. This can then
3044 be loaded in Wireshark and similar tools to inspect bus
3047 * The busctl tool now understands a new "tree" verb that shows
3048 the object trees of a specific service on the bus, or of all
3051 * The busctl tool now understands a new "introspect" verb that
3052 shows all interfaces and members of objects on the bus,
3053 including their signature and values. This is particularly
3054 useful to get more information about bus objects shown by
3055 the new "busctl tree" command.
3057 * The busctl tool now understands new verbs "call",
3058 "set-property" and "get-property" for invoking bus method
3059 calls, setting and getting bus object properties in a
3062 * busctl gained a new --augment-creds= argument that controls
3063 whether the tool shall augment credential information it
3064 gets from the bus with data from /proc, in a possibly
3067 * nspawn's --link-journal= switch gained two new values
3068 "try-guest" and "try-host" that work like "guest" and
3069 "host", but do not fail if the host has no persistent
3070 journalling enabled. -j is now equivalent to
3071 --link-journal=try-guest.
3073 * macvlan network devices created by nspawn will now have
3074 stable MAC addresses.
3076 * A new SmackProcessLabel= unit setting has been added, which
3077 controls the SMACK security label processes forked off by
3078 the respective unit shall use.
3080 * If compiled with --enable-xkbcommon, systemd-localed will
3081 verify x11 keymap settings by compiling the given keymap. It
3082 will spew out warnings if the compilation fails. This
3083 requires libxkbcommon to be installed.
3085 * When a coredump is collected, a larger number of metadata
3086 fields is now collected and included in the journal records
3087 created for it. More specifically, control group membership,
3088 environment variables, memory maps, working directory,
3089 chroot directory, /proc/$PID/status, and a list of open file
3090 descriptors is now stored in the log entry.
3092 * The udev hwdb now contains DPI information for mice. For
3095 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3097 * All systemd programs that read standalone configuration
3098 files in /etc now also support a corresponding series of
3099 .conf.d configuration directories in /etc/, /run/,
3100 /usr/local/lib/, /usr/lib/, and (if configured with
3101 --enable-split-usr) /lib/. In particular, the following
3102 configuration files now have corresponding configuration
3103 directories: system.conf user.conf, logind.conf,
3104 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3105 resolved.conf, timesyncd.conf, journal-remote.conf, and
3106 journal-upload.conf. Note that distributions should use the
3107 configuration directories in /usr/lib/; the directories in
3108 /etc/ are reserved for the system administrator.
3110 * systemd-rfkill will no longer take the rfkill device name
3111 into account when storing rfkill state on disk, as the name
3112 might be dynamically assigned and not stable. Instead, the
3113 ID_PATH udev variable combined with the rfkill type (wlan,
3114 bluetooth, ...) is used.
3116 * A new service systemd-machine-id-commit.service has been
3117 added. When used on systems where /etc is read-only during
3118 boot, and /etc/machine-id is not initialized (but an empty
3119 file), this service will copy the temporary machine ID
3120 created as replacement into /etc after the system is fully
3121 booted up. This is useful for systems that are freshly
3122 installed with a non-initialized machine ID, but should get
3123 a fixed machine ID for subsequent boots.
3125 * networkd's .netdev files now provide a large set of
3126 configuration parameters for VXLAN devices. Similarly, the
3127 bridge port cost parameter is now configurable in .network
3128 files. There's also new support for configuring IP source
3129 routing. networkd .link files gained support for a new
3130 OriginalName= match that is useful to match against the
3131 original interface name the kernel assigned. .network files
3132 may include MTU= and MACAddress= fields for altering the MTU
3133 and MAC address while being connected to a specific network
3136 * The LUKS logic gained supported for configuring
3137 UUID-specific key files. There's also new support for naming
3138 LUKS device from the kernel command line, using the new
3139 luks.name= argument.
3141 * Timer units may now be transiently created via the bus API
3142 (this was previously already available for scope and service
3143 units). In addition it is now possible to create multiple
3144 transient units at the same time with a single bus call. The
3145 "systemd-run" tool has been updated to make use of this for
3146 running commands on a specified time, in at(1)-style.
3148 * tmpfiles gained support for "t" lines, for assigning
3149 extended attributes to files. Among other uses this may be
3150 used to assign SMACK labels to files.
3152 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3153 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3154 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3155 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3156 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3157 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3158 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3159 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3160 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3161 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3162 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3163 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3164 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3165 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3166 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3167 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3168 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3169 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3171 — Berlin, 2014-12-10
3175 * journalctl gained the new options -t/--identifier= to match
3176 on the syslog identifier (aka "tag"), as well as --utc to
3177 show log timestamps in the UTC timezone. journalctl now also
3178 accepts -n/--lines=all to disable line capping in a pager.
3180 * journalctl gained a new switch, --flush, that synchronously
3181 flushes logs from /run/log/journal to /var/log/journal if
3182 persistent storage is enabled. systemd-journal-flush.service
3183 now waits until the operation is complete.
3185 * Services can notify the manager before they start a reload
3186 (by sending RELOADING=1) or shutdown (by sending
3187 STOPPING=1). This allows the manager to track and show the
3188 internal state of daemons and closes a race condition when
3189 the process is still running but has closed its D-Bus
3192 * Services with Type=oneshot do not have to have any ExecStart
3195 * User units are now loaded also from
3196 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3197 /run/systemd/user directory that was already previously
3198 supported, but is under the control of the user.
3200 * Job timeouts (i.e. time-outs on the time a job that is
3201 queued stays in the run queue) can now optionally result in
3202 immediate reboot or power-off actions (JobTimeoutAction= and
3203 JobTimeoutRebootArgument=). This is useful on ".target"
3204 units, to limit the maximum time a target remains
3205 undispatched in the run queue, and to trigger an emergency
3206 operation in such a case. This is now used by default to
3207 turn off the system if boot-up (as defined by everything in
3208 basic.target) hangs and does not complete for at least
3209 15min. Also, if power-off or reboot hang for at least 30min
3210 an immediate power-off/reboot operation is triggered. This
3211 functionality is particularly useful to increase reliability
3212 on embedded devices, but also on laptops which might
3213 accidentally get powered on when carried in a backpack and
3214 whose boot stays stuck in a hard disk encryption passphrase
3217 * systemd-logind can be configured to also handle lid switch
3218 events even when the machine is docked or multiple displays
3219 are attached (HandleLidSwitchDocked= option).
3221 * A helper binary and a service have been added which can be
3222 used to resume from hibernation in the initramfs. A
3223 generator will parse the resume= option on the kernel
3224 command line to trigger resume.
3226 * A user console daemon systemd-consoled has been
3227 added. Currently, it is a preview, and will so far open a
3228 single terminal on each session of the user marked as
3229 Desktop=systemd-console.
3231 * Route metrics can be specified for DHCP routes added by
3234 * The SELinux context of socket-activated services can be set
3235 from the information provided by the networking stack
3236 (SELinuxContextFromNet= option).
3238 * Userspace firmware loading support has been removed and
3239 the minimum supported kernel version is thus bumped to 3.7.
3241 * Timeout for udev workers has been increased from 1 to 3
3242 minutes, but a warning will be printed after 1 minute to
3243 help diagnose kernel modules that take a long time to load.
3245 * Udev rules can now remove tags on devices with TAG-="foobar".
3247 * systemd's readahead implementation has been removed. In many
3248 circumstances it didn't give expected benefits even for
3249 rotational disk drives and was becoming less relevant in the
3250 age of SSDs. As none of the developers has been using
3251 rotating media anymore, and nobody stepped up to actively
3252 maintain this component of systemd it has now been removed.
3254 * Swap units can use Options= to specify discard options.
3255 Discard options specified for swaps in /etc/fstab are now
3258 * Docker containers are now detected as a separate type of
3261 * The Password Agent protocol gained support for queries where
3262 the user input is shown, useful e.g. for user names.
3263 systemd-ask-password gained a new --echo option to turn that
3266 * The default sysctl.d/ snippets will now set:
3268 net.core.default_qdisc = fq_codel
3270 This selects Fair Queuing Controlled Delay as the default
3271 queuing discipline for network interfaces. fq_codel helps
3272 fight the network bufferbloat problem. It is believed to be
3273 a good default with no tuning required for most workloads.
3274 Downstream distributions may override this choice. On 10Gbit
3275 servers that do not do forwarding, "fq" may perform better.
3276 Systems without a good clocksource should use "pfifo_fast".
3278 * If kdbus is enabled during build a new option BusPolicy= is
3279 available for service units, that allows locking all service
3280 processes into a stricter bus policy, in order to limit
3281 access to various bus services, or even hide most of them
3282 from the service's view entirely.
3284 * networkctl will now show the .network and .link file
3285 networkd has applied to a specific interface.
3287 * sd-login gained a new API call sd_session_get_desktop() to
3288 query which desktop environment has been selected for a
3291 * UNIX utmp support is now compile-time optional to support
3292 legacy-free systems.
3294 * systemctl gained two new commands "add-wants" and
3295 "add-requires" for pulling in units from specific targets
3298 * If the word "rescue" is specified on the kernel command line
3299 the system will now boot into rescue mode (aka
3300 rescue.target), which was previously available only by
3301 specifying "1" or "systemd.unit=rescue.target" on the kernel
3302 command line. This new kernel command line option nicely
3303 mirrors the already existing "emergency" kernel command line
3306 * New kernel command line options mount.usr=, mount.usrflags=,
3307 mount.usrfstype= have been added that match root=, rootflags=,
3308 rootfstype= but allow mounting a specific file system to
3311 * The $NOTIFY_SOCKET is now also passed to control processes of
3312 services, not only the main process.
3314 * This version reenables support for fsck's -l switch. This
3315 means at least version v2.25 of util-linux is required for
3316 operation, otherwise dead-locks on device nodes may
3317 occur. Again: you need to update util-linux to at least
3318 v2.25 when updating systemd to v217.
3320 * The "multi-seat-x" tool has been removed from systemd, as
3321 its functionality has been integrated into X servers 1.16,
3322 and the tool is hence redundant. It is recommended to update
3323 display managers invoking this tool to simply invoke X
3324 directly from now on, again.
3326 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3327 message flag has been added for all of systemd's PolicyKit
3328 authenticated method calls has been added. In particular
3329 this now allows optional interactive authorization via
3330 PolicyKit for many of PID1's privileged operations such as
3331 unit file enabling and disabling.
3333 * "udevadm hwdb --update" learnt a new switch "--usr" for
3334 placing the rebuilt hardware database in /usr instead of
3335 /etc. When used only hardware database entries stored in
3336 /usr will be used, and any user database entries in /etc are
3337 ignored. This functionality is useful for vendors to ship a
3338 pre-built database on systems where local configuration is
3339 unnecessary or unlikely.
3341 * Calendar time specifications in .timer units now also
3342 understand the strings "semi-annually", "quarterly" and
3343 "minutely" as shortcuts (in addition to the preexisting
3344 "anually", "hourly", ...).
3346 * systemd-tmpfiles will now correctly create files in /dev
3347 at boot which are marked for creation only at boot. It is
3348 recommended to always create static device nodes with 'c!'
3349 and 'b!', so that they are created only at boot and not
3350 overwritten at runtime.
3352 * When the watchdog logic is used for a service (WatchdogSec=)
3353 and the watchdog timeout is hit the service will now be
3354 terminated with SIGABRT (instead of just SIGTERM), in order
3355 to make sure a proper coredump and backtrace is
3356 generated. This ensures that hanging services will result in
3357 similar coredump/backtrace behaviour as services that hit a
3360 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3361 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3362 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3363 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3364 Herrmann, David Sommerseth, David Strauss, Emil Renner
3365 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3366 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3367 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3368 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3369 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3370 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3371 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3372 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3373 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3374 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3375 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3376 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3377 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3378 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3379 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3380 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3383 — Berlin, 2014-10-28
3387 * timedated no longer reads NTP implementation unit names from
3388 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3389 implementations should add a
3391 Conflicts=systemd-timesyncd.service
3393 to their unit files to take over and replace systemd's NTP
3394 default functionality.
3396 * systemd-sysusers gained a new line type "r" for configuring
3397 which UID/GID ranges to allocate system users/groups
3398 from. Lines of type "u" may now add an additional column
3399 that specifies the home directory for the system user to be
3400 created. Also, systemd-sysusers may now optionally read user
3401 information from STDIN instead of a file. This is useful for
3402 invoking it from RPM preinst scriptlets that need to create
3403 users before the first RPM file is installed since these
3404 files might need to be owned by them. A new
3405 %sysusers_create_inline RPM macro has been introduced to do
3406 just that. systemd-sysusers now updates the shadow files as
3407 well as the user/group databases, which should enhance
3408 compatibility with certain tools like grpck.
3410 * A number of bus APIs of PID 1 now optionally consult
3411 PolicyKit to permit access for otherwise unprivileged
3412 clients under certain conditions. Note that this currently
3413 doesn't support interactive authentication yet, but this is
3414 expected to be added eventually, too.
3416 * /etc/machine-info now has new fields for configuring the
3417 deployment environment of the machine, as well as the
3418 location of the machine. hostnamectl has been updated with
3419 new command to update these fields.
3421 * systemd-timesyncd has been updated to automatically acquire
3422 NTP server information from systemd-networkd, which might
3423 have been discovered via DHCP.
3425 * systemd-resolved now includes a caching DNS stub resolver
3426 and a complete LLMNR name resolution implementation. A new
3427 NSS module "nss-resolve" has been added which can be used
3428 instead of glibc's own "nss-dns" to resolve hostnames via
3429 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3430 be resolved via systemd-resolved D-Bus APIs. In contrast to
3431 the glibc internal resolver systemd-resolved is aware of
3432 multi-homed system, and keeps DNS server and caches separate
3433 and per-interface. Queries are sent simultaneously on all
3434 interfaces that have DNS servers configured, in order to
3435 properly handle VPNs and local LANs which might resolve
3436 separate sets of domain names. systemd-resolved may acquire
3437 DNS server information from systemd-networkd automatically,
3438 which in turn might have discovered them via DHCP. A tool
3439 "systemd-resolve-host" has been added that may be used to
3440 query the DNS logic in resolved. systemd-resolved implements
3441 IDNA and automatically uses IDNA or UTF-8 encoding depending
3442 on whether classic DNS or LLMNR is used as transport. In the
3443 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3444 implementation to systemd-resolved.
3446 * A new NSS module nss-mymachines has been added, that
3447 automatically resolves the names of all local registered
3448 containers to their respective IP addresses.
3450 * A new client tool "networkctl" for systemd-networkd has been
3451 added. It currently is entirely passive and will query
3452 networking configuration from udev, rtnetlink and networkd,
3453 and present it to the user in a very friendly
3454 way. Eventually, we hope to extend it to become a full
3455 control utility for networkd.
3457 * .socket units gained a new DeferAcceptSec= setting that
3458 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3459 TCP. Similarly, support for controlling TCP keep-alive
3460 settings has been added (KeepAliveTimeSec=,
3461 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3462 turning off Nagle's algorithm on TCP has been added
3465 * logind learned a new session type "web", for use in projects
3466 like Cockpit which register web clients as PAM sessions.
3468 * timer units with at least one OnCalendar= setting will now
3469 be started only after time-sync.target has been
3470 reached. This way they will not elapse before the system
3471 clock has been corrected by a local NTP client or
3472 similar. This is particular useful on RTC-less embedded
3473 machines, that come up with an invalid system clock.
3475 * systemd-nspawn's --network-veth= switch should now result in
3476 stable MAC addresses for both the outer and the inner side
3479 * systemd-nspawn gained a new --volatile= switch for running
3480 container instances with /etc or /var unpopulated.
3482 * The kdbus client code has been updated to use the new Linux
3483 3.17 memfd subsystem instead of the old kdbus-specific one.
3485 * systemd-networkd's DHCP client and server now support
3486 FORCERENEW. There are also new configuration options to
3487 configure the vendor client identifier and broadcast mode
3490 * systemd will no longer inform the kernel about the current
3491 timezone, as this is necessarily incorrect and racy as the
3492 kernel has no understanding of DST and similar
3493 concepts. This hence means FAT timestamps will be always
3494 considered UTC, similar to what Android is already
3495 doing. Also, when the RTC is configured to the local time
3496 (rather than UTC) systemd will never synchronize back to it,
3497 as this might confuse Windows at a later boot.
3499 * systemd-analyze gained a new command "verify" for offline
3500 validation of unit files.
3502 * systemd-networkd gained support for a couple of additional
3503 settings for bonding networking setups. Also, the metric for
3504 statically configured routes may now be configured. For
3505 network interfaces where this is appropriate the peer IP
3506 address may now be configured.
3508 * systemd-networkd's DHCP client will no longer request
3509 broadcasting by default, as this tripped up some networks.
3510 For hardware where broadcast is required the feature should
3511 be switched back on using RequestBroadcast=yes.
3513 * systemd-networkd will now set up IPv4LL addresses (when
3514 enabled) even if DHCP is configured successfully.
3516 * udev will now default to respect network device names given
3517 by the kernel when the kernel indicates that these are
3518 predictable. This behavior can be tweaked by changing
3519 NamePolicy= in the relevant .link file.
3521 * A new library systemd-terminal has been added that
3522 implements full TTY stream parsing and rendering. This
3523 library is supposed to be used later on for implementing a
3524 full userspace VT subsystem, replacing the current kernel
3527 * A new tool systemd-journal-upload has been added to push
3528 journal data to a remote system running
3529 systemd-journal-remote.
3531 * journald will no longer forward all local data to another
3532 running syslog daemon. This change has been made because
3533 rsyslog (which appears to be the most commonly used syslog
3534 implementation these days) no longer makes use of this, and
3535 instead pulls the data out of the journal on its own. Since
3536 forwarding the messages to a non-existent syslog server is
3537 more expensive than we assumed we have now turned this
3538 off. If you run a syslog server that is not a recent rsyslog
3539 version, you have to turn this option on again
3540 (ForwardToSyslog= in journald.conf).
3542 * journald now optionally supports the LZ4 compressor for
3543 larger journal fields. This compressor should perform much
3544 better than XZ which was the previous default.
3546 * machinectl now shows the IP addresses of local containers,
3547 if it knows them, plus the interface name of the container.
3549 * A new tool "systemd-escape" has been added that makes it
3550 easy to escape strings to build unit names and similar.
3552 * sd_notify() messages may now include a new ERRNO= field
3553 which is parsed and collected by systemd and shown among the
3554 "systemctl status" output for a service.
3556 * A new component "systemd-firstboot" has been added that
3557 queries the most basic systemd information (timezone,
3558 hostname, root password) interactively on first
3559 boot. Alternatively it may also be used to provision these
3560 things offline on OS images installed into directories.
3562 * The default sysctl.d/ snippets will now set
3564 net.ipv4.conf.default.promote_secondaries=1
3566 This has the benefit of no flushing secondary IP addresses
3567 when primary addresses are removed.
3569 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3570 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3571 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3572 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3573 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3574 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3575 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3576 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3577 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3578 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3579 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3580 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3581 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3582 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3583 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3585 — Berlin, 2014-08-19
3589 * A new tool systemd-sysusers has been added. This tool
3590 creates system users and groups in /etc/passwd and
3591 /etc/group, based on static declarative system user/group
3592 definitions in /usr/lib/sysusers.d/. This is useful to
3593 enable factory resets and volatile systems that boot up with
3594 an empty /etc directory, and thus need system users and
3595 groups created during early boot. systemd now also ships
3596 with two default sysusers.d/ files for the most basic
3597 users and groups systemd and the core operating system
3600 * A new tmpfiles snippet has been added that rebuilds the
3601 essential files in /etc on boot, should they be missing.
3603 * A directive for ensuring automatic clean-up of
3604 /var/cache/man/ has been removed from the default
3605 configuration. This line should now be shipped by the man
3606 implementation. The necessary change has been made to the
3607 man-db implementation. Note that you need to update your man
3608 implementation to one that ships this line, otherwise no
3609 automatic clean-up of /var/cache/man will take place.
3611 * A new condition ConditionNeedsUpdate= has been added that
3612 may conditionalize services to only run when /etc or /var
3613 are "older" than the vendor operating system resources in
3614 /usr. This is useful for reconstructing or updating /etc
3615 after an offline update of /usr or a factory reset, on the
3616 next reboot. Services that want to run once after such an
3617 update or reset should use this condition and order
3618 themselves before the new systemd-update-done.service, which
3619 will mark the two directories as fully updated. A number of
3620 service files have been added making use of this, to rebuild
3621 the udev hardware database, the journald message catalog and
3622 dynamic loader cache (ldconfig). The systemd-sysusers tool
3623 described above also makes use of this now. With this in
3624 place it is now possible to start up a minimal operating
3625 system with /etc empty cleanly. For more information on the
3626 concepts involved see this recent blog story:
3628 http://0pointer.de/blog/projects/stateless.html
3630 * A new system group "input" has been introduced, and all
3631 input device nodes get this group assigned. This is useful
3632 for system-level software to get access to input devices. It
3633 complements what is already done for "audio" and "video".
3635 * systemd-networkd learnt minimal DHCPv4 server support in
3636 addition to the existing DHCPv4 client support. It also
3637 learnt DHCPv6 client and IPv6 Router Solicitation client
3638 support. The DHCPv4 client gained support for static routes
3639 passed in from the server. Note that the [DHCPv4] section
3640 known in older systemd-networkd versions has been renamed to
3641 [DHCP] and is now also used by the DHCPv6 client. Existing
3642 .network files using settings of this section should be
3643 updated, though compatibility is maintained. Optionally, the
3644 client hostname may now be sent to the DHCP server.
3646 * networkd gained support for vxlan virtual networks as well
3647 as tun/tap and dummy devices.
3649 * networkd gained support for automatic allocation of address
3650 ranges for interfaces from a system-wide pool of
3651 addresses. This is useful for dynamically managing a large
3652 number of interfaces with a single network configuration
3653 file. In particular this is useful to easily assign
3654 appropriate IP addresses to the veth links of a large number
3655 of nspawn instances.
3657 * RPM macros for processing sysusers, sysctl and binfmt
3658 drop-in snippets at package installation time have been
3661 * The /etc/os-release file should now be placed in
3662 /usr/lib/os-release. The old location is automatically
3663 created as symlink. /usr/lib is the more appropriate
3664 location of this file, since it shall actually describe the
3665 vendor operating system shipped in /usr, and not the
3666 configuration stored in /etc.
3668 * .mount units gained a new boolean SloppyOptions= setting
3669 that maps to mount(8)'s -s option which enables permissive
3670 parsing of unknown mount options.
3672 * tmpfiles learnt a new "L+" directive which creates a symlink
3673 but (unlike "L") deletes a pre-existing file first, should
3674 it already exist and not already be the correct
3675 symlink. Similarly, "b+", "c+" and "p+" directives have been
3676 added as well, which create block and character devices, as
3677 well as fifos in the filesystem, possibly removing any
3678 pre-existing files of different types.
3680 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3681 'argument' field (which so far specified the source to
3682 symlink/copy the files from) is now optional. If omitted the
3683 same file os copied from /usr/share/factory/ suffixed by the
3684 full destination path. This is useful for populating /etc
3685 with essential files, by copying them from vendor defaults
3686 shipped in /usr/share/factory/etc.
3688 * A new command "systemctl preset-all" has been added that
3689 applies the service preset settings to all installed unit
3690 files. A new switch --preset-mode= has been added that
3691 controls whether only enable or only disable operations
3694 * A new command "systemctl is-system-running" has been added
3695 that allows checking the overall state of the system, for
3696 example whether it is fully up and running.
3698 * When the system boots up with an empty /etc, the equivalent
3699 to "systemctl preset-all" is executed during early boot, to
3700 make sure all default services are enabled after a factory
3703 * systemd now contains a minimal preset file that enables the
3704 most basic services systemd ships by default.
3706 * Unit files' [Install] section gained a new DefaultInstance=
3707 field for defining the default instance to create if a
3708 template unit is enabled with no instance specified.
3710 * A new passive target cryptsetup-pre.target has been added
3711 that may be used by services that need to make they run and
3712 finish before the first LUKS cryptographic device is set up.
3714 * The /dev/loop-control and /dev/btrfs-control device nodes
3715 are now owned by the "disk" group by default, opening up
3716 access to this group.
3718 * systemd-coredump will now automatically generate a
3719 stack trace of all core dumps taking place on the system,
3720 based on elfutils' libdw library. This stack trace is logged
3723 * systemd-coredump may now optionally store coredumps directly
3724 on disk (in /var/lib/systemd/coredump, possibly compressed),
3725 instead of storing them unconditionally in the journal. This
3726 mode is the new default. A new configuration file
3727 /etc/systemd/coredump.conf has been added to configure this
3728 and other parameters of systemd-coredump.
3730 * coredumpctl gained a new "info" verb to show details about a
3731 specific coredump. A new switch "-1" has also been added
3732 that makes sure to only show information about the most
3733 recent entry instead of all entries. Also, as the tool is
3734 generally useful now the "systemd-" prefix of the binary
3735 name has been removed. Distributions that want to maintain
3736 compatibility with the old name should add a symlink from
3737 the old name to the new name.
3739 * journald's SplitMode= now defaults to "uid". This makes sure
3740 that unprivileged users can access their own coredumps with
3741 coredumpctl without restrictions.
3743 * New kernel command line options "systemd.wants=" (for
3744 pulling an additional unit during boot), "systemd.mask="
3745 (for masking a specific unit for the boot), and
3746 "systemd.debug-shell" (for enabling the debug shell on tty9)
3747 have been added. This is implemented in the new generator
3748 "systemd-debug-generator".
3750 * systemd-nspawn will now by default filter a couple of
3751 syscalls for containers, among them those required for
3752 kernel module loading, direct x86 IO port access, swap
3753 management, and kexec. Most importantly though
3754 open_by_handle_at() is now prohibited for containers,
3755 closing a hole similar to a recently discussed vulnerability
3756 in docker regarding access to files on file hierarchies the
3757 container should normally not have access to. Note that, for
3758 nspawn, we generally make no security claims anyway (and
3759 this is explicitly documented in the man page), so this is
3760 just a fix for one of the most obvious problems.
3762 * A new man page file-hierarchy(7) has been added that
3763 contains a minimized, modernized version of the file system
3764 layout systemd expects, similar in style to the FHS
3765 specification or hier(5). A new tool systemd-path(1) has
3766 been added to query many of these paths for the local
3769 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3770 longer done. Since the directory now has a per-user size
3771 limit, and is cleaned on logout this appears unnecessary,
3772 in particular since this now brings the lifecycle of this
3773 directory closer in line with how IPC objects are handled.
3775 * systemd.pc now exports a number of additional directories,
3776 including $libdir (which is useful to identify the library
3777 path for the primary architecture of the system), and a
3778 couple of drop-in directories.
3780 * udev's predictable network interface names now use the dev_port
3781 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3782 distinguish between ports of the same PCI function. dev_id should
3783 only be used for ports using the same HW address, hence the need
3786 * machined has been updated to export the OS version of a
3787 container (read from /etc/os-release and
3788 /usr/lib/os-release) on the bus. This is now shown in
3789 "machinectl status" for a machine.
3791 * A new service setting RestartForceExitStatus= has been
3792 added. If configured to a set of exit signals or process
3793 return values, the service will be restarted when the main
3794 daemon process exits with any of them, regardless of the
3797 * systemctl's -H switch for connecting to remote systemd
3798 machines has been extended so that it may be used to
3799 directly connect to a specific container on the
3800 host. "systemctl -H root@foobar:waldi" will now connect as
3801 user "root" to host "foobar", and then proceed directly to
3802 the container named "waldi". Note that currently you have to
3803 authenticate as user "root" for this to work, as entering
3804 containers is a privileged operation.
3806 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3807 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3808 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3809 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3810 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3811 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3812 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3813 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3814 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3815 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3816 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3817 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3819 — Berlin, 2014-07-03
3823 * As an experimental feature, udev now tries to lock the
3824 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3825 executes events for the disk or any of its partitions.
3826 Applications like partitioning programs can lock the
3827 disk device node (flock(LOCK_EX)) and claim temporary
3828 device ownership that way; udev will entirely skip all event
3829 handling for this disk and its partitions. If the disk
3830 was opened for writing, the close will trigger a partition
3831 table rescan in udev's "watch" facility, and if needed
3832 synthesize "change" events for the disk and all its partitions.
3833 This is now unconditionally enabled, and if it turns out to
3834 cause major problems, we might turn it on only for specific
3835 devices, or might need to disable it entirely. Device Mapper
3836 devices are excluded from this logic.
3838 * We temporarily dropped the "-l" switch for fsck invocations,
3839 since they collide with the flock() logic above. util-linux
3840 upstream has been changed already to avoid this conflict,
3841 and we will readd "-l" as soon as util-linux with this
3842 change has been released.
3844 * The dependency on libattr has been removed. Since a long
3845 time, the extended attribute calls have moved to glibc, and
3846 libattr is thus unnecessary.
3848 * Virtualization detection works without privileges now. This
3849 means the systemd-detect-virt binary no longer requires
3850 CAP_SYS_PTRACE file capabilities, and our daemons can run
3851 with fewer privileges.
3853 * systemd-networkd now runs under its own "systemd-network"
3854 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3855 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3856 loses the ability to write to files owned by root this way.
3858 * Similarly, systemd-resolved now runs under its own
3859 "systemd-resolve" user with no capabilities remaining.
3861 * Similarly, systemd-bus-proxyd now runs under its own
3862 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3864 * systemd-networkd gained support for setting up "veth"
3865 virtual Ethernet devices for container connectivity, as well
3866 as GRE and VTI tunnels.
3868 * systemd-networkd will no longer automatically attempt to
3869 manually load kernel modules necessary for certain tunnel
3870 transports. Instead, it is assumed the kernel loads them
3871 automatically when required. This only works correctly on
3872 very new kernels. On older kernels, please consider adding
3873 the kernel modules to /etc/modules-load.d/ as a work-around.
3875 * The resolv.conf file systemd-resolved generates has been
3876 moved to /run/systemd/resolve/. If you have a symlink from
3877 /etc/resolv.conf, it might be necessary to correct it.
3879 * Two new service settings, ProtectHome= and ProtectSystem=,
3880 have been added. When enabled, they will make the user data
3881 (such as /home) inaccessible or read-only and the system
3882 (such as /usr) read-only, for specific services. This allows
3883 very light-weight per-service sandboxing to avoid
3884 modifications of user data or system files from
3885 services. These two new switches have been enabled for all
3886 of systemd's long-running services, where appropriate.
3888 * Socket units gained new SocketUser= and SocketGroup=
3889 settings to set the owner user and group of AF_UNIX sockets
3890 and FIFOs in the file system.
3892 * Socket units gained a new RemoveOnStop= setting. If enabled,
3893 all FIFOS and sockets in the file system will be removed
3894 when the specific socket unit is stopped.
3896 * Socket units gained a new Symlinks= setting. It takes a list
3897 of symlinks to create to file system sockets or FIFOs
3898 created by the specific Unix sockets. This is useful to
3899 manage symlinks to socket nodes with the same life-cycle as
3902 * The /dev/log socket and /dev/initctl FIFO have been moved to
3903 /run, and have been replaced by symlinks. This allows
3904 connecting to these facilities even if PrivateDevices=yes is
3905 used for a service (which makes /dev/log itself unavailable,
3906 but /run is left). This also has the benefit of ensuring
3907 that /dev only contains device nodes, directories and
3908 symlinks, and nothing else.
3910 * sd-daemon gained two new calls sd_pid_notify() and
3911 sd_pid_notifyf(). They are similar to sd_notify() and
3912 sd_notifyf(), but allow overriding of the source PID of
3913 notification messages if permissions permit this. This is
3914 useful to send notify messages on behalf of a different
3915 process (for example, the parent process). The
3916 systemd-notify tool has been updated to make use of this
3917 when sending messages (so that notification messages now
3918 originate from the shell script invoking systemd-notify and
3919 not the systemd-notify process itself. This should minimize
3920 a race where systemd fails to associate notification
3921 messages to services when the originating process already
3924 * A new "on-abnormal" setting for Restart= has been added. If
3925 set, it will result in automatic restarts on all "abnormal"
3926 reasons for a process to exit, which includes unclean
3927 signals, core dumps, timeouts and watchdog timeouts, but
3928 does not include clean and unclean exit codes or clean
3929 signals. Restart=on-abnormal is an alternative for
3930 Restart=on-failure for services that shall be able to
3931 terminate and avoid restarts on certain errors, by
3932 indicating so with an unclean exit code. Restart=on-failure
3933 or Restart=on-abnormal is now the recommended setting for
3934 all long-running services.
3936 * If the InaccessibleDirectories= service setting points to a
3937 mount point (or if there are any submounts contained within
3938 it), it is now attempted to completely unmount it, to make
3939 the file systems truly unavailable for the respective
3942 * The ReadOnlyDirectories= service setting and
3943 systemd-nspawn's --read-only parameter are now recursively
3944 applied to all submounts, too.
3946 * Mount units may now be created transiently via the bus APIs.
3948 * The support for SysV and LSB init scripts has been removed
3949 from the systemd daemon itself. Instead, it is now
3950 implemented as a generator that creates native systemd units
3951 from these scripts when needed. This enables us to remove a
3952 substantial amount of legacy code from PID 1, following the
3953 fact that many distributions only ship a very small number
3954 of LSB/SysV init scripts nowadays.
3956 * Privileged Xen (dom0) domains are not considered
3957 virtualization anymore by the virtualization detection
3958 logic. After all, they generally have unrestricted access to
3959 the hardware and usually are used to manage the unprivileged
3962 * systemd-tmpfiles gained a new "C" line type, for copying
3963 files or entire directories.
3965 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3966 lines. So far, they have been non-globbing versions of the
3967 latter, and have thus been redundant. In future, it is
3968 recommended to only use "z". "m" has hence been removed
3969 from the documentation, even though it stays supported.
3971 * A tmpfiles snippet to recreate the most basic structure in
3972 /var has been added. This is enough to create the /var/run →
3973 /run symlink and create a couple of structural
3974 directories. This allows systems to boot up with an empty or
3975 volatile /var. Of course, while with this change, the core OS
3976 now is capable with dealing with a volatile /var, not all
3977 user services are ready for it. However, we hope that sooner
3978 or later, many service daemons will be changed upstream so
3979 that they are able to automatically create their necessary
3980 directories in /var at boot, should they be missing. This is
3981 the first step to allow state-less systems that only require
3982 the vendor image for /usr to boot.
3984 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3985 empty tmpfs instance to a specific directory. This is
3986 particularly useful for making use of the automatic
3987 reconstruction of /var (see above), by passing --tmpfs=/var.
3989 * Access modes specified in tmpfiles snippets may now be
3990 prefixed with "~", which indicates that they shall be masked
3991 by whether the existing file or directory is currently
3992 writable, readable or executable at all. Also, if specified,
3993 the sgid/suid/sticky bits will be masked for all
3996 * A new passive target unit "network-pre.target" has been
3997 added which is useful for services that shall run before any
3998 network is configured, for example firewall scripts.
4000 * The "floppy" group that previously owned the /dev/fd*
4001 devices is no longer used. The "disk" group is now used
4002 instead. Distributions should probably deprecate usage of
4005 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4006 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4007 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4008 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4009 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4010 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4013 — Berlin, 2014-06-11
4017 * A new "systemd-timesyncd" daemon has been added for
4018 synchronizing the system clock across the network. It
4019 implements an SNTP client. In contrast to NTP
4020 implementations such as chrony or the NTP reference server,
4021 this only implements a client side, and does not bother with
4022 the full NTP complexity, focusing only on querying time from
4023 one remote server and synchronizing the local clock to
4024 it. Unless you intend to serve NTP to networked clients or
4025 want to connect to local hardware clocks, this simple NTP
4026 client should be more than appropriate for most
4027 installations. The daemon runs with minimal privileges, and
4028 has been hooked up with networkd to only operate when
4029 network connectivity is available. The daemon saves the
4030 current clock to disk every time a new NTP sync has been
4031 acquired, and uses this to possibly correct the system clock
4032 early at bootup, in order to accommodate for systems that
4033 lack an RTC such as the Raspberry Pi and embedded devices,
4034 and to make sure that time monotonically progresses on these
4035 systems, even if it is not always correct. To make use of
4036 this daemon, a new system user and group "systemd-timesync"
4037 needs to be created on installation of systemd.
4039 * The queue "seqnum" interface of libudev has been disabled, as
4040 it was generally incompatible with device namespacing as
4041 sequence numbers of devices go "missing" if the devices are
4042 part of a different namespace.
4044 * "systemctl list-timers" and "systemctl list-sockets" gained
4045 a --recursive switch for showing units of these types also
4046 for all local containers, similar in style to the already
4047 supported --recursive switch for "systemctl list-units".
4049 * A new RebootArgument= setting has been added for service
4050 units, which may be used to specify a kernel reboot argument
4051 to use when triggering reboots with StartLimitAction=.
4053 * A new FailureAction= setting has been added for service
4054 units which may be used to specify an operation to trigger
4055 when a service fails. This works similarly to
4056 StartLimitAction=, but unlike it, controls what is done
4057 immediately rather than only after several attempts to
4058 restart the service in question.
4060 * hostnamed got updated to also expose the kernel name,
4061 release, and version on the bus. This is useful for
4062 executing commands like hostnamectl with the -H switch.
4063 systemd-analyze makes use of this to properly display
4064 details when running non-locally.
4066 * The bootchart tool can now show cgroup information in the
4067 graphs it generates.
4069 * The CFS CPU quota cgroup attribute is now exposed for
4070 services. The new CPUQuota= switch has been added for this
4071 which takes a percentage value. Setting this will have the
4072 result that a service may never get more CPU time than the
4073 specified percentage, even if the machine is otherwise idle.
4075 * systemd-networkd learned IPIP and SIT tunnel support.
4077 * LSB init scripts exposing a dependency on $network will now
4078 get a dependency on network-online.target rather than simply
4079 network.target. This should bring LSB handling closer to
4080 what it was on SysV systems.
4082 * A new fsck.repair= kernel option has been added to control
4083 how fsck shall deal with unclean file systems at boot.
4085 * The (.ini) configuration file parser will now silently
4086 ignore sections whose name begins with "X-". This may be
4087 used to maintain application-specific extension sections in unit
4090 * machined gained a new API to query the IP addresses of
4091 registered containers. "machinectl status" has been updated
4092 to show these addresses in its output.
4094 * A new call sd_uid_get_display() has been added to the
4095 sd-login APIs for querying the "primary" session of a
4096 user. The "primary" session of the user is elected from the
4097 user's sessions and generally a graphical session is
4098 preferred over a text one.
4100 * A minimal systemd-resolved daemon has been added. It
4101 currently simply acts as a companion to systemd-networkd and
4102 manages resolv.conf based on per-interface DNS
4103 configuration, possibly supplied via DHCP. In the long run
4104 we hope to extend this into a local DNSSEC enabled DNS and
4107 * The systemd-networkd-wait-online tool is now enabled by
4108 default. It will delay network-online.target until a network
4109 connection has been configured. The tool primarily integrates
4110 with networkd, but will also make a best effort to make sense
4111 of network configuration performed in some other way.
4113 * Two new service options StartupCPUShares= and
4114 StartupBlockIOWeight= have been added that work similarly to
4115 CPUShares= and BlockIOWeight= however only apply during
4116 system startup. This is useful to prioritize certain services
4117 differently during bootup than during normal runtime.
4119 * hostnamed has been changed to prefer the statically
4120 configured hostname in /etc/hostname (unless set to
4121 'localhost' or empty) over any dynamic one supplied by
4122 dhcp. With this change, the rules for picking the hostname
4123 match more closely the rules of other configuration settings
4124 where the local administrator's configuration in /etc always
4125 overrides any other settings.
4127 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4128 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4129 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4130 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4131 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4132 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4133 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4134 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4135 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4136 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4137 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4138 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4139 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4140 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4141 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4142 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4145 — Beijing, 2014-05-28
4149 * When restoring the screen brightness at boot, stay away from
4150 the darkest setting or from the lowest 5% of the available
4151 range, depending on which is the larger value of both. This
4152 should effectively protect the user from rebooting into a
4153 black screen, should the brightness have been set to minimum
4156 * sd-login gained a new sd_machine_get_class() call to
4157 determine the class ("vm" or "container") of a machine
4158 registered with machined.
4160 * sd-login gained new calls
4161 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4162 to query the identity of the peer of a local AF_UNIX
4163 connection. They operate similarly to their sd_pid_get_xyz()
4166 * PID 1 will now maintain a system-wide system state engine
4167 with the states "starting", "running", "degraded",
4168 "maintenance", "stopping". These states are bound to system
4169 startup, normal runtime, runtime with at least one failed
4170 service, rescue/emergency mode and system shutdown. This
4171 state is shown in the "systemctl status" output when no unit
4172 name is passed. It is useful to determine system state, in
4173 particularly when doing so for many systems or containers at
4176 * A new command "list-machines" has been added to "systemctl"
4177 that lists all local OS containers and shows their system
4178 state (see above), if systemd runs inside of them.
4180 * systemctl gained a new "-r" switch to recursively enumerate
4181 units on all local containers, when used with the
4182 "list-unit" command (which is the default one that is
4183 executed when no parameters are specified).
4185 * The GPT automatic partition discovery logic will now honour
4186 two GPT partition flags: one may be set on a partition to
4187 cause it to be mounted read-only, and the other may be set
4188 on a partition to ignore it during automatic discovery.
4190 * Two new GPT type UUIDs have been added for automatic root
4191 partition discovery, for 32-bit and 64-bit ARM. This is not
4192 particularly useful for discovering the root directory on
4193 these architectures during bare-metal boots (since UEFI is
4194 not common there), but still very useful to allow booting of
4195 ARM disk images in nspawn with the -i option.
4197 * MAC addresses of interfaces created with nspawn's
4198 --network-interface= switch will now be generated from the
4199 machine name, and thus be stable between multiple invocations
4202 * logind will now automatically remove all IPC objects owned
4203 by a user if she or he fully logs out. This makes sure that
4204 users who are logged out cannot continue to consume IPC
4205 resources. This covers SysV memory, semaphores and message
4206 queues as well as POSIX shared memory and message
4207 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4208 limits. With this functionality, that is corrected. This may
4209 be turned off by using the RemoveIPC= switch of logind.conf.
4211 * The systemd-machine-id-setup and tmpfiles tools gained a
4212 --root= switch to operate on a specific root directory,
4215 * journald can now forward logged messages to the TTYs of all
4216 logged in users ("wall"). This is the default for all
4217 emergency messages now.
4219 * A new tool systemd-journal-remote has been added to stream
4220 journal log messages across the network.
4222 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4223 controller trees are mounted into it. Note that the
4224 directories mounted beneath it are not read-only. This is a
4225 security measure and is particularly useful because glibc
4226 actually includes a search logic to pick any tmpfs it can
4227 find to implement shm_open() if /dev/shm is not available
4228 (which it might very well be in namespaced setups).
4230 * machinectl gained a new "poweroff" command to cleanly power
4231 down a local OS container.
4233 * The PrivateDevices= unit file setting will now also drop the
4234 CAP_MKNOD capability from the capability bound set, and
4235 imply DevicePolicy=closed.
4237 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4238 comprehensively on all long-running systemd services where
4239 this is appropriate.
4241 * systemd-udevd will now run in a disassociated mount
4242 namespace. To mount directories from udev rules, make sure to
4243 pull in mount units via SYSTEMD_WANTS properties.
4245 * The kdbus support gained support for uploading policy into
4246 the kernel. sd-bus gained support for creating "monitoring"
4247 connections that can eavesdrop into all bus communication
4248 for debugging purposes.
4250 * Timestamps may now be specified in seconds since the UNIX
4251 epoch Jan 1st, 1970 by specifying "@" followed by the value
4254 * Native tcpwrap support in systemd has been removed. tcpwrap
4255 is old code, not really maintained anymore and has serious
4256 shortcomings, and better options such as firewalls
4257 exist. For setups that require tcpwrap usage, please
4258 consider invoking your socket-activated service via tcpd,
4259 like on traditional inetd.
4261 * A new system.conf configuration option
4262 DefaultTimerAccuracySec= has been added that controls the
4263 default AccuracySec= setting of .timer units.
4265 * Timer units gained a new WakeSystem= switch. If enabled,
4266 timers configured this way will cause the system to resume
4267 from system suspend (if the system supports that, which most
4270 * Timer units gained a new Persistent= switch. If enabled,
4271 timers configured this way will save to disk when they have
4272 been last triggered. This information is then used on next
4273 reboot to possible execute overdue timer events, that
4274 could not take place because the system was powered off.
4275 This enables simple anacron-like behaviour for timer units.
4277 * systemctl's "list-timers" will now also list the time a
4278 timer unit was last triggered in addition to the next time
4279 it will be triggered.
4281 * systemd-networkd will now assign predictable IPv4LL
4282 addresses to its local interfaces.
4284 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4285 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4286 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4287 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4288 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4289 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4290 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4291 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4294 — Berlin, 2014-03-25
4298 * A new unit file setting RestrictAddressFamilies= has been
4299 added to restrict which socket address families unit
4300 processes gain access to. This takes address family names
4301 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4302 attack surface of services via exotic protocol stacks. This
4303 is built on seccomp system call filters.
4305 * Two new unit file settings RuntimeDirectory= and
4306 RuntimeDirectoryMode= have been added that may be used to
4307 manage a per-daemon runtime directories below /run. This is
4308 an alternative for setting up directory permissions with
4309 tmpfiles snippets, and has the advantage that the runtime
4310 directory's lifetime is bound to the daemon runtime and that
4311 the daemon starts up with an empty directory each time. This
4312 is particularly useful when writing services that drop
4313 privileges using the User= or Group= setting.
4315 * The DeviceAllow= unit setting now supports globbing for
4316 matching against device group names.
4318 * The systemd configuration file system.conf gained new
4319 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4320 DefaultMemoryAccounting= to globally turn on/off accounting
4321 for specific resources (cgroups) for all units. These
4322 settings may still be overridden individually in each unit
4325 * systemd-gpt-auto-generator is now able to discover /srv and
4326 root partitions in addition to /home and swap partitions. It
4327 also supports LUKS-encrypted partitions now. With this in
4328 place, automatic discovery of partitions to mount following
4329 the Discoverable Partitions Specification
4330 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4331 is now a lot more complete. This allows booting without
4332 /etc/fstab and without root= on the kernel command line on
4333 systems prepared appropriately.
4335 * systemd-nspawn gained a new --image= switch which allows
4336 booting up disk images and Linux installations on any block
4337 device that follow the Discoverable Partitions Specification
4338 (see above). This means that installations made with
4339 appropriately updated installers may now be started and
4340 deployed using container managers, completely
4341 unmodified. (We hope that libvirt-lxc will add support for
4342 this feature soon, too.)
4344 * systemd-nspawn gained a new --network-macvlan= setting to
4345 set up a private macvlan interface for the
4346 container. Similarly, systemd-networkd gained a new
4347 Kind=macvlan setting in .netdev files.
4349 * systemd-networkd now supports configuring local addresses
4352 * A new tool systemd-network-wait-online has been added to
4353 synchronously wait for network connectivity using
4356 * The sd-bus.h bus API gained a new sd_bus_track object for
4357 tracking the life-cycle of bus peers. Note that sd-bus.h is
4358 still not a public API though (unless you specify
4359 --enable-kdbus on the configure command line, which however
4360 voids your warranty and you get no API stability guarantee).
4362 * The $XDG_RUNTIME_DIR runtime directories for each user are
4363 now individual tmpfs instances, which has the benefit of
4364 introducing separate pools for each user, with individual
4365 size limits, and thus making sure that unprivileged clients
4366 can no longer negatively impact the system or other users by
4367 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4368 RuntimeDirectorySize= has been introduced that allows
4369 controlling the default size limit for all users. It
4370 defaults to 10% of the available physical memory. This is no
4371 replacement for quotas on tmpfs though (which the kernel
4372 still does not support), as /dev/shm and /tmp are still
4373 shared resources used by both the system and unprivileged
4376 * logind will now automatically turn off automatic suspending
4377 on laptop lid close when more than one display is
4378 connected. This was previously expected to be implemented
4379 individually in desktop environments (such as GNOME),
4380 however has been added to logind now, in order to fix a
4381 boot-time race where a desktop environment might not have
4382 been started yet and thus not been able to take an inhibitor
4383 lock at the time where logind already suspends the system
4384 due to a closed lid.
4386 * logind will now wait at least 30s after each system
4387 suspend/resume cycle, and 3min after system boot before
4388 suspending the system due to a closed laptop lid. This
4389 should give USB docking stations and similar enough time to
4390 be probed and configured after system resume and boot in
4391 order to then act as suspend blocker.
4393 * systemd-run gained a new --property= setting which allows
4394 initialization of resource control properties (and others)
4395 for the created scope or service unit. Example: "systemd-run
4396 --property=BlockIOWeight=10 updatedb" may be used to run
4397 updatedb at a low block IO scheduling weight.
4399 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4400 now also work in --scope mode.
4402 * When systemd is compiled with kdbus support, basic support
4403 for enforced policies is now in place. (Note that enabling
4404 kdbus still voids your warranty and no API compatibility
4407 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4408 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4409 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4410 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4411 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4412 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4413 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4414 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4415 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4416 Zbigniew Jędrzejewski-Szmek
4418 — Berlin, 2014-03-12
4422 * systemd will now relabel /dev after loading the SMACK policy
4423 according to SMACK rules.
4425 * A new unit file option AppArmorProfile= has been added to
4426 set the AppArmor profile for the processes of a unit.
4428 * A new condition check ConditionArchitecture= has been added
4429 to conditionalize units based on the system architecture, as
4430 reported by uname()'s "machine" field.
4432 * systemd-networkd now supports matching on the system
4433 virtualization, architecture, kernel command line, host name
4436 * logind is now a lot more aggressive when suspending the
4437 machine due to a closed laptop lid. Instead of acting only
4438 on the lid close action, it will continuously watch the lid
4439 status and act on it. This is useful for laptops where the
4440 power button is on the outside of the chassis so that it can
4441 be reached without opening the lid (such as the Lenovo
4442 Yoga). On those machines, logind will now immediately
4443 re-suspend the machine if the power button has been
4444 accidentally pressed while the laptop was suspended and in a
4445 backpack or similar.
4447 * logind will now watch SW_DOCK switches and inhibit reaction
4448 to the lid switch if it is pressed. This means that logind
4449 will not suspend the machine anymore if the lid is closed
4450 and the system is docked, if the laptop supports SW_DOCK
4451 notifications via the input layer. Note that ACPI docking
4452 stations do not generate this currently. Also note that this
4453 logic is usually not fully sufficient and Desktop
4454 Environments should take a lid switch inhibitor lock when an
4455 external display is connected, as systemd will not watch
4458 * nspawn will now make use of the devices cgroup controller by
4459 default, and only permit creation of and access to the usual
4460 API device nodes like /dev/null or /dev/random, as well as
4461 access to (but not creation of) the pty devices.
4463 * We will now ship a default .network file for
4464 systemd-networkd that automatically configures DHCP for
4465 network interfaces created by nspawn's --network-veth or
4466 --network-bridge= switches.
4468 * systemd will now understand the usual M, K, G, T suffixes
4469 according to SI conventions (i.e. to the base 1000) when
4470 referring to throughput and hardware metrics. It will stay
4471 with IEC conventions (i.e. to the base 1024) for software
4472 metrics, according to what is customary according to
4473 Wikipedia. We explicitly document which base applies for
4474 each configuration option.
4476 * The DeviceAllow= setting in unit files now supports a syntax
4477 to whitelist an entire group of devices node majors at once,
4478 based on the /proc/devices listing. For example, with the
4479 string "char-pts", it is now possible to whitelist all
4480 current and future pseudo-TTYs at once.
4482 * sd-event learned a new "post" event source. Event sources of
4483 this type are triggered by the dispatching of any event
4484 source of a type that is not "post". This is useful for
4485 implementing clean-up and check event sources that are
4486 triggered by other work being done in the program.
4488 * systemd-networkd is no longer statically enabled, but uses
4489 the usual [Install] sections so that it can be
4490 enabled/disabled using systemctl. It still is enabled by
4493 * When creating a veth interface pair with systemd-nspawn, the
4494 host side will now be prefixed with "vb-" if
4495 --network-bridge= is used, and with "ve-" if --network-veth
4496 is used. This way, it is easy to distinguish these cases on
4497 the host, for example to apply different configuration to
4498 them with systemd-networkd.
4500 * The compatibility libraries for libsystemd-journal.so,
4501 libsystem-id128.so, libsystemd-login.so and
4502 libsystemd-daemon.so do not make use of IFUNC
4503 anymore. Instead, we now build libsystemd.so multiple times
4504 under these alternative names. This means that the footprint
4505 is drastically increased, but given that these are
4506 transitional compatibility libraries, this should not matter
4507 much. This change has been made necessary to support the ARM
4508 platform for these compatibility libraries, as the ARM
4509 toolchain is not really at the same level as the toolchain
4510 for other architectures like x86 and does not support
4511 IFUNC. Please make sure to use --enable-compat-libs only
4512 during a transitional period!
4514 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4515 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4516 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4517 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4518 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4519 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4520 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4521 Zbigniew Jędrzejewski-Szmek
4523 — Berlin, 2014-02-24
4527 * A new component "systemd-networkd" has been added that can
4528 be used to configure local network interfaces statically or
4529 via DHCP. It is capable of bringing up bridges, VLANs, and
4530 bonding. Currently, no hook-ups for interactive network
4531 configuration are provided. Use this for your initrd,
4532 container, embedded, or server setup if you need a simple,
4533 yet powerful, network configuration solution. This
4534 configuration subsystem is quite nifty, as it allows wildcard
4535 hotplug matching in interfaces. For example, with a single
4536 configuration snippet, you can configure that all Ethernet
4537 interfaces showing up are automatically added to a bridge,
4538 or similar. It supports link-sensing and more.
4540 * A new tool "systemd-socket-proxyd" has been added which can
4541 act as a bidirectional proxy for TCP sockets. This is
4542 useful for adding socket activation support to services that
4543 do not actually support socket activation, including virtual
4544 machines and the like.
4546 * Add a new tool to save/restore rfkill state on
4549 * Save/restore state of keyboard backlights in addition to
4550 display backlights on shutdown/boot.
4552 * udev learned a new SECLABEL{} construct to label device
4553 nodes with a specific security label when they appear. For
4554 now, only SECLABEL{selinux} is supported, but the syntax is
4555 prepared for additional security frameworks.
4557 * udev gained a new scheme to configure link-level attributes
4558 from files in /etc/systemd/network/*.link. These files can
4559 match against MAC address, device path, driver name and type,
4560 and will apply attributes like the naming policy, link speed,
4561 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4562 address assignment policy (randomized, ...).
4564 * The configuration of network interface naming rules for
4565 "permanent interface names" has changed: a new NamePolicy=
4566 setting in the [Link] section of .link files determines the
4567 priority of possible naming schemes (onboard, slot, MAC,
4568 path). The default value of this setting is determined by
4569 /usr/lib/net/links/99-default.link. Old
4570 80-net-name-slot.rules udev configuration file has been
4571 removed, so local configuration overriding this file should
4572 be adapted to override 99-default.link instead.
4574 * When the User= switch is used in a unit file, also
4575 initialize $SHELL= based on the user database entry.
4577 * systemd no longer depends on libdbus. All communication is
4578 now done with sd-bus, systemd's low-level bus library
4581 * kdbus support has been added to PID 1 itself. When kdbus is
4582 enabled, this causes PID 1 to set up the system bus and
4583 enable support for a new ".busname" unit type that
4584 encapsulates bus name activation on kdbus. It works a little
4585 bit like ".socket" units, except for bus names. A new
4586 generator has been added that converts classic dbus1 service
4587 activation files automatically into native systemd .busname
4590 * sd-bus: add a light-weight vtable implementation that allows
4591 defining objects on the bus with a simple static const
4592 vtable array of its methods, signals and properties.
4594 * systemd will not generate or install static dbus
4595 introspection data anymore to /usr/share/dbus-1/interfaces,
4596 as the precise format of these files is unclear, and
4597 nothing makes use of it.
4599 * A proxy daemon is now provided to proxy clients connecting
4600 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4601 compatibility with classic D-Bus.
4603 * A bus driver implementation has been added that supports the
4604 classic D-Bus bus driver calls on kdbus, also for
4605 compatibility purposes.
4607 * A new API "sd-event.h" has been added that implements a
4608 minimal event loop API built around epoll. It provides a
4609 couple of features that direct epoll usage is lacking:
4610 prioritization of events, scales to large numbers of timer
4611 events, per-event timer slack (accuracy), system-wide
4612 coalescing of timer events, exit handlers, watchdog
4613 supervision support using systemd's sd_notify() API, child
4616 * A new API "sd-rntl.h" has been added that provides an API
4617 around the route netlink interface of the kernel, similar in
4618 style to "sd-bus.h".
4620 * A new API "sd-dhcp-client.h" has been added that provides a
4621 small DHCPv4 client-side implementation. This is used by
4624 * There is a new kernel command line option
4625 "systemd.restore_state=0|1". When set to "0", none of the
4626 systemd tools will restore saved runtime state to hardware
4627 devices. More specifically, the rfkill and backlight states
4630 * The FsckPassNo= compatibility option in mount/service units
4631 has been removed. The fstab generator will now add the
4632 necessary dependencies automatically, and does not require
4633 PID1's support for that anymore.
4635 * journalctl gained a new switch, --list-boots, that lists
4636 recent boots with their times and boot IDs.
4638 * The various tools like systemctl, loginctl, timedatectl,
4639 busctl, systemd-run, ... have gained a new switch "-M" to
4640 connect to a specific, local OS container (as direct
4641 connection, without requiring SSH). This works on any
4642 container that is registered with machined, such as those
4643 created by libvirt-lxc or nspawn.
4645 * systemd-run and systemd-analyze also gained support for "-H"
4646 to connect to remote hosts via SSH. This is particularly
4647 useful for systemd-run because it enables queuing of jobs
4648 onto remote systems.
4650 * machinectl gained a new command "login" to open a getty
4651 login in any local container. This works with any container
4652 that is registered with machined (such as those created by
4653 libvirt-lxc or nspawn), and which runs systemd inside.
4655 * machinectl gained a new "reboot" command that may be used to
4656 trigger a reboot on a specific container that is registered
4657 with machined. This works on any container that runs an init
4658 system of some kind.
4660 * systemctl gained a new "list-timers" command to print a nice
4661 listing of installed timer units with the times they elapse
4664 * Alternative reboot() parameters may now be specified on the
4665 "systemctl reboot" command line and are passed to the
4666 reboot() system call.
4668 * systemctl gained a new --job-mode= switch to configure the
4669 mode to queue a job with. This is a more generic version of
4670 --fail, --irreversible, and --ignore-dependencies, which are
4671 still available but not advertised anymore.
4673 * /etc/systemd/system.conf gained new settings to configure
4674 various default timeouts of units, as well as the default
4675 start limit interval and burst. These may still be overridden
4678 * PID1 will now export on the bus profile data of the security
4679 policy upload process (such as the SELinux policy upload to
4682 * journald: when forwarding logs to the console, include
4683 timestamps (following the setting in
4684 /sys/module/printk/parameters/time).
4686 * OnCalendar= in timer units now understands the special
4687 strings "yearly" and "annually". (Both are equivalent)
4689 * The accuracy of timer units is now configurable with the new
4690 AccuracySec= setting. It defaults to 1min.
4692 * A new dependency type JoinsNamespaceOf= has been added that
4693 allows running two services within the same /tmp and network
4694 namespace, if PrivateNetwork= or PrivateTmp= are used.
4696 * A new command "cat" has been added to systemctl. It outputs
4697 the original unit file of a unit, and concatenates the
4698 contents of additional "drop-in" unit file snippets, so that
4699 the full configuration is shown.
4701 * systemctl now supports globbing on the various "list-xyz"
4702 commands, like "list-units" or "list-sockets", as well as on
4703 those commands which take multiple unit names.
4705 * journalctl's --unit= switch gained support for globbing.
4707 * All systemd daemons now make use of the watchdog logic so
4708 that systemd automatically notices when they hang.
4710 * If the $container_ttys environment variable is set,
4711 getty-generator will automatically spawn a getty for each
4712 listed tty. This is useful for container managers to request
4713 login gettys to be spawned on as many ttys as needed.
4715 * %h, %s, %U specifier support is not available anymore when
4716 used in unit files for PID 1. This is because NSS calls are
4717 not safe from PID 1. They stay available for --user
4718 instances of systemd, and as special case for the root user.
4720 * loginctl gained a new "--no-legend" switch to turn off output
4723 * The "sd-login.h" API gained three new calls:
4724 sd_session_is_remote(), sd_session_get_remote_user(),
4725 sd_session_get_remote_host() to query information about
4728 * The udev hardware database now also carries vendor/product
4729 information of SDIO devices.
4731 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4732 determine whether watchdog notifications are requested by
4735 * Socket-activated per-connection services now include a
4736 short description of the connection parameters in the
4739 * tmpfiles gained a new "--boot" option. When this is not used,
4740 only lines where the command character is not suffixed with
4741 "!" are executed. When this option is specified, those
4742 options are executed too. This partitions tmpfiles
4743 directives into those that can be safely executed at any
4744 time, and those which should be run only at boot (for
4745 example, a line that creates /run/nologin).
4747 * A new API "sd-resolve.h" has been added which provides a simple
4748 asynchronous wrapper around glibc NSS host name resolution
4749 calls, such as getaddrinfo(). In contrast to glibc's
4750 getaddrinfo_a(), it does not use signals. In contrast to most
4751 other asynchronous name resolution libraries, this one does
4752 not reimplement DNS, but reuses NSS, so that alternate
4753 host name resolution systems continue to work, such as mDNS,
4754 LDAP, etc. This API is based on libasyncns, but it has been
4755 cleaned up for inclusion in systemd.
4757 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4758 "sd-daemon.h" are no longer found in individual libraries
4759 libsystemd-journal.so, libsystemd-login.so,
4760 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4761 merged them into a single library, libsystemd.so, which
4762 provides all symbols. The reason for this is cyclic
4763 dependencies, as these libraries tend to use each other's
4764 symbols. So far, we have managed to workaround that by linking
4765 a copy of a good part of our code into each of these
4766 libraries again and again, which, however, makes certain
4767 things hard to do, like sharing static variables. Also, it
4768 substantially increases footprint. With this change, there
4769 is only one library for the basic APIs systemd
4770 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4771 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4772 library as well, however are subject to the --enable-kdbus
4773 switch (see below). Note that "sd-dhcp-client.h" is not part
4774 of this library (this is because it only consumes, never
4775 provides, services of/to other APIs). To make the transition
4776 easy from the separate libraries to the unified one, we
4777 provide the --enable-compat-libs compile-time switch which
4778 will generate stub libraries that are compatible with the
4779 old ones but redirect all calls to the new one.
4781 * All of the kdbus logic and the new APIs "sd-bus.h",
4782 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4783 and "sd-utf8.h" are compile-time optional via the
4784 "--enable-kdbus" switch, and they are not compiled in by
4785 default. To make use of kdbus, you have to explicitly enable
4786 the switch. Note however, that neither the kernel nor the
4787 userspace API for all of this is considered stable yet. We
4788 want to maintain the freedom to still change the APIs for
4789 now. By specifying this build-time switch, you acknowledge
4790 that you are aware of the instability of the current
4793 * Also, note that while kdbus is pretty much complete,
4794 it lacks one thing: proper policy support. This means you
4795 can build a fully working system with all features; however,
4796 it will be highly insecure. Policy support will be added in
4797 one of the next releases, at the same time that we will
4798 declare the APIs stable.
4800 * When the kernel command line argument "kdbus" is specified,
4801 systemd will automatically load the kdbus.ko kernel module. At
4802 this stage of development, it is only useful for testing kdbus
4803 and should not be used in production. Note: if "--enable-kdbus"
4804 is specified, and the kdbus.ko kernel module is available, and
4805 "kdbus" is added to the kernel command line, the entire system
4806 runs with kdbus instead of dbus-daemon, with the above mentioned
4807 problem of missing the system policy enforcement. Also a future
4808 version of kdbus.ko or a newer systemd will not be compatible with
4809 each other, and will unlikely be able to boot the machine if only
4810 one of them is updated.
4812 * systemctl gained a new "import-environment" command which
4813 uploads the caller's environment (or parts thereof) into the
4814 service manager so that it is inherited by services started
4815 by the manager. This is useful to upload variables like
4816 $DISPLAY into the user service manager.
4818 * A new PrivateDevices= switch has been added to service units
4819 which allows running a service with a namespaced /dev
4820 directory that does not contain any device nodes for
4821 physical devices. More specifically, it only includes devices
4822 such as /dev/null, /dev/urandom, and /dev/zero which are API
4825 * logind has been extended to support behaviour like VT
4826 switching on seats that do not support a VT. This makes
4827 multi-session available on seats that are not the first seat
4828 (seat0), and on systems where kernel support for VTs has
4829 been disabled at compile-time.
4831 * If a process holds a delay lock for system sleep or shutdown
4832 and fails to release it in time, we will now log its
4833 identity. This makes it easier to identify processes that
4834 cause slow suspends or power-offs.
4836 * When parsing /etc/crypttab, support for a new key-slot=
4837 option as supported by Debian is added. It allows indicating
4838 which LUKS slot to use on disk, speeding up key loading.
4840 * The sd_journald_sendv() API call has been checked and
4841 officially declared to be async-signal-safe so that it may
4842 be invoked from signal handlers for logging purposes.
4844 * Boot-time status output is now enabled automatically after a
4845 short timeout if boot does not progress, in order to give
4846 the user an indication what she or he is waiting for.
4848 * The boot-time output has been improved to show how much time
4849 remains until jobs expire.
4851 * The KillMode= switch in service units gained a new possible
4852 value "mixed". If set, and the unit is shut down, then the
4853 initial SIGTERM signal is sent only to the main daemon
4854 process, while the following SIGKILL signal is sent to
4855 all remaining processes of the service.
4857 * When a scope unit is registered, a new property "Controller"
4858 may be set. If set to a valid bus name, systemd will send a
4859 RequestStop() signal to this name when it would like to shut
4860 down the scope. This may be used to hook manager logic into
4861 the shutdown logic of scope units. Also, scope units may now
4862 be put in a special "abandoned" state, in which case the
4863 manager process which created them takes no further
4864 responsibilities for it.
4866 * When reading unit files, systemd will now verify
4867 the access mode of these files, and warn about certain
4868 suspicious combinations. This has been added to make it
4869 easier to track down packaging bugs where unit files are
4870 marked executable or world-writable.
4872 * systemd-nspawn gained a new "--setenv=" switch to set
4873 container-wide environment variables. The similar option in
4874 systemd-activate was renamed from "--environment=" to
4875 "--setenv=" for consistency.
4877 * systemd-nspawn has been updated to create a new kdbus domain
4878 for each container that is invoked, thus allowing each
4879 container to have its own set of system and user buses,
4880 independent of the host.
4882 * systemd-nspawn gained a new --drop-capability= switch to run
4883 the container with less capabilities than the default. Both
4884 --drop-capability= and --capability= now take the special
4885 string "all" for dropping or keeping all capabilities.
4887 * systemd-nspawn gained new switches for executing containers
4888 with specific SELinux labels set.
4890 * systemd-nspawn gained a new --quiet switch to not generate
4891 any additional output but the container's own console
4894 * systemd-nspawn gained a new --share-system switch to run a
4895 container without PID namespacing enabled.
4897 * systemd-nspawn gained a new --register= switch to control
4898 whether the container is registered with systemd-machined or
4899 not. This is useful for containers that do not run full
4900 OS images, but only specific apps.
4902 * systemd-nspawn gained a new --keep-unit which may be used
4903 when invoked as the only program from a service unit, and
4904 results in registration of the unit service itself in
4905 systemd-machined, instead of a newly opened scope unit.
4907 * systemd-nspawn gained a new --network-interface= switch for
4908 moving arbitrary interfaces to the container. The new
4909 --network-veth switch creates a virtual Ethernet connection
4910 between host and container. The new --network-bridge=
4911 switch then allows assigning the host side of this virtual
4912 Ethernet connection to a bridge device.
4914 * systemd-nspawn gained a new --personality= switch for
4915 setting the kernel personality for the container. This is
4916 useful when running a 32-bit container on a 64-bit host. A
4917 similar option Personality= is now also available for service
4920 * logind will now also track a "Desktop" identifier for each
4921 session which encodes the desktop environment of it. This is
4922 useful for desktop environments that want to identify
4923 multiple running sessions of itself easily.
4925 * A new SELinuxContext= setting for service units has been
4926 added that allows setting a specific SELinux execution
4927 context for a service.
4929 * Most systemd client tools will now honour $SYSTEMD_LESS for
4930 settings of the "less" pager. By default, these tools will
4931 override $LESS to allow certain operations to work, such as
4932 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4933 influence this logic.
4935 * systemd's "seccomp" hook-up has been changed to make use of
4936 the libseccomp library instead of using its own
4937 implementation. This has benefits for portability among
4940 * For usage together with SystemCallFilter=, a new
4941 SystemCallErrorNumber= setting has been introduced that
4942 allows configuration of a system error number to be returned
4943 on filtered system calls, instead of immediately killing the
4944 process. Also, SystemCallArchitectures= has been added to
4945 limit access to system calls of a particular architecture
4946 (in order to turn off support for unused secondary
4947 architectures). There is also a global
4948 SystemCallArchitectures= setting in system.conf now to turn
4949 off support for non-native system calls system-wide.
4951 * systemd requires a kernel with a working name_to_handle_at(),
4952 please see the kernel config requirements in the README file.
4954 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4955 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4956 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4957 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4958 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4959 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4960 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4961 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4962 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4963 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4964 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4965 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4966 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4967 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4968 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4969 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4970 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4971 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4972 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4973 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4974 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4975 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4976 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4977 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4979 — Berlin, 2014-02-20
4983 * logind has gained support for facilitating privileged input
4984 and drm device access for unprivileged clients. This work is
4985 useful to allow Wayland display servers (and similar
4986 programs, such as kmscon) to run under the user's ID and
4987 access input and drm devices which are normally
4988 protected. When this is used (and the kernel is new enough)
4989 logind will "mute" IO on the file descriptors passed to
4990 Wayland as long as it is in the background and "unmute" it
4991 if it returns into the foreground. This allows secure
4992 session switching without allowing background sessions to
4993 eavesdrop on input and display data. This also introduces
4994 session switching support if VT support is turned off in the
4995 kernel, and on seats that are not seat0.
4997 * A new kernel command line option luks.options= is understood
4998 now which allows specifying LUKS options for usage for LUKS
4999 encrypted partitions specified with luks.uuid=.
5001 * tmpfiles.d(5) snippets may now use specifier expansion in
5002 path names. More specifically %m, %b, %H, %v, are now
5003 replaced by the local machine id, boot id, hostname, and
5004 kernel version number.
5006 * A new tmpfiles.d(5) command "m" has been introduced which
5007 may be used to change the owner/group/access mode of a file
5008 or directory if it exists, but do nothing if it does not.
5010 * This release removes high-level support for the
5011 MemorySoftLimit= cgroup setting. The underlying kernel
5012 cgroup attribute memory.soft_limit= is currently badly
5013 designed and likely to be removed from the kernel API in its
5014 current form, hence we should not expose it for now.
5016 * The memory.use_hierarchy cgroup attribute is now enabled for
5017 all cgroups systemd creates in the memory cgroup
5018 hierarchy. This option is likely to be come the built-in
5019 default in the kernel anyway, and the non-hierarchical mode
5020 never made much sense in the intrinsically hierarchical
5023 * A new field _SYSTEMD_SLICE= is logged along with all journal
5024 messages containing the slice a message was generated
5025 from. This is useful to allow easy per-customer filtering of
5026 logs among other things.
5028 * systemd-journald will no longer adjust the group of journal
5029 files it creates to the "systemd-journal" group. Instead we
5030 rely on the journal directory to be owned by the
5031 "systemd-journal" group, and its setgid bit set, so that the
5032 kernel file system layer will automatically enforce that
5033 journal files inherit this group assignment. The reason for
5034 this change is that we cannot allow NSS look-ups from
5035 journald which would be necessary to resolve
5036 "systemd-journal" to a numeric GID, because this might
5037 create deadlocks if NSS involves synchronous queries to
5038 other daemons (such as nscd, or sssd) which in turn are
5039 logging clients of journald and might block on it, which
5040 would then dead lock. A tmpfiles.d(5) snippet included in
5041 systemd will make sure the setgid bit and group are
5042 properly set on the journal directory if it exists on every
5043 boot. However, we recommend adjusting it manually after
5044 upgrades too (or from RPM scriptlets), so that the change is
5045 not delayed until next reboot.
5047 * Backlight and random seed files in /var/lib/ have moved into
5048 the /var/lib/systemd/ directory, in order to centralize all
5049 systemd generated files in one directory.
5051 * Boot time performance measurements (as displayed by
5052 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5053 performance information if that's available to determine how
5054 much time BIOS and boot loader initialization required. With
5055 a sufficiently new BIOS you hence no longer need to boot
5056 with Gummiboot to get access to such information.
5058 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5059 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5060 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5061 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5062 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5063 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5064 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5066 — Berlin, 2013-10-02
5070 * The Restart= option for services now understands a new
5071 on-watchdog setting, which will restart the service
5072 automatically if the service stops sending out watchdog keep
5073 alive messages (as configured with WatchdogSec=).
5075 * The getty generator (which is responsible for bringing up a
5076 getty on configured serial consoles) will no longer only
5077 start a getty on the primary kernel console but on all
5078 others, too. This makes the order in which console= is
5079 specified on the kernel command line less important.
5081 * libsystemd-logind gained a new sd_session_get_vt() call to
5082 retrieve the VT number of a session.
5084 * If the option "tries=0" is set for an entry of /etc/crypttab
5085 its passphrase is queried indefinitely instead of any
5086 maximum number of tries.
5088 * If a service with a configure PID file terminates its PID
5089 file will now be removed automatically if it still exists
5090 afterwards. This should put an end to stale PID files.
5092 * systemd-run will now also take relative binary path names
5093 for execution and no longer insists on absolute paths.
5095 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5096 paths that are optionally prefixed with "-" to indicate that
5097 it should not be considered a failure if they do not exist.
5099 * journalctl -o (and similar commands) now understands a new
5100 output mode "short-precise", it is similar to "short" but
5101 shows timestamps with usec accuracy.
5103 * The option "discard" (as known from Debian) is now
5104 synonymous to "allow-discards" in /etc/crypttab. In fact,
5105 "discard" is preferred now (since it is easier to remember
5108 * Some licensing clean-ups were made, so that more code is now
5109 LGPL-2.1 licensed than before.
5111 * A minimal tool to save/restore the display backlight
5112 brightness across reboots has been added. It will store the
5113 backlight setting as late as possible at shutdown, and
5114 restore it as early as possible during reboot.
5116 * A logic to automatically discover and enable home and swap
5117 partitions on GPT disks has been added. With this in place
5118 /etc/fstab becomes optional for many setups as systemd can
5119 discover certain partitions located on the root disk
5120 automatically. Home partitions are recognized under their
5121 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5122 partitions are recognized under their GPT type ID
5123 0657fd6da4ab43c484e50933c84b4f4f.
5125 * systemd will no longer pass any environment from the kernel
5126 or initrd to system services. If you want to set an
5127 environment for all services, do so via the kernel command
5128 line systemd.setenv= assignment.
5130 * The systemd-sysctl tool no longer natively reads the file
5131 /etc/sysctl.conf. If desired, the file should be symlinked
5132 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5133 legacy support by a symlink rather than built-in code, it
5134 also makes the otherwise hidden order of application of the
5135 different files visible. (Note that this partly reverts to a
5136 pre-198 application order of sysctl knobs!)
5138 * The "systemctl set-log-level" and "systemctl dump" commands
5139 have been moved to systemd-analyze.
5141 * systemd-run learned the new --remain-after-exit switch,
5142 which causes the scope unit not to be cleaned up
5143 automatically after the process terminated.
5145 * tmpfiles learned a new --exclude-prefix= switch to exclude
5146 certain paths from operation.
5148 * journald will now automatically flush all messages to disk
5149 as soon as a message at the log level CRIT, ALERT or EMERG
5152 Contributions from: Andrew Cook, Brandon Philips, Christian
5153 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5154 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5155 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5156 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5157 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5158 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5159 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5160 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5161 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5162 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5163 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5164 William Giokas, Zbigniew Jędrzejewski-Szmek
5166 — Berlin, 2013-09-13
5170 * The documentation has been updated to cover the various new
5171 concepts introduced with 205.
5173 * Unit files now understand the new %v specifier which
5174 resolves to the kernel version string as returned by "uname
5177 * systemctl now supports filtering the unit list output by
5178 load state, active state and sub state, using the new
5181 * "systemctl status" will now show the results of the
5182 condition checks (like ConditionPathExists= and similar) of
5183 the last start attempts of the unit. They are also logged to
5186 * "journalctl -b" may now be used to look for boot output of a
5187 specific boot. Try "journalctl -b -1" for the previous boot,
5188 but the syntax is substantially more powerful.
5190 * "journalctl --show-cursor" has been added which prints the
5191 cursor string the last shown log line. This may then be used
5192 with the new "journalctl --after-cursor=" switch to continue
5193 browsing logs from that point on.
5195 * "journalctl --force" may now be used to force regeneration
5198 * Creation of "dead" device nodes has been moved from udev
5199 into kmod and tmpfiles. Previously, udev would read the kmod
5200 databases to pre-generate dead device nodes based on meta
5201 information contained in kernel modules, so that these would
5202 be auto-loaded on access rather then at boot. As this
5203 does not really have much to do with the exposing actual
5204 kernel devices to userspace this has always been slightly
5205 alien in the udev codebase. Following the new scheme kmod
5206 will now generate a runtime snippet for tmpfiles from the
5207 module meta information and it now is tmpfiles' job to the
5208 create the nodes. This also allows overriding access and
5209 other parameters for the nodes using the usual tmpfiles
5210 facilities. As side effect this allows us to remove the
5211 CAP_SYS_MKNOD capability bit from udevd entirely.
5213 * logind's device ACLs may now be applied to these "dead"
5214 devices nodes too, thus finally allowing managed access to
5215 devices such as /dev/snd/sequencer without loading the
5216 backing module right-away.
5218 * A new RPM macro has been added that may be used to apply
5219 tmpfiles configuration during package installation.
5221 * systemd-detect-virt and ConditionVirtualization= now can
5222 detect User-Mode-Linux machines (UML).
5224 * journald will now implicitly log the effective capabilities
5225 set of processes in the message metadata.
5227 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5229 * The initrd interface has been simplified (more specifically,
5230 support for passing performance data via environment
5231 variables and fsck results via files in /run has been
5232 removed). These features were non-essential, and are
5233 nowadays available in a much nicer way by having systemd in
5234 the initrd serialize its state and have the hosts systemd
5235 deserialize it again.
5237 * The udev "keymap" data files and tools to apply keyboard
5238 specific mappings of scan to key codes, and force-release
5239 scan code lists have been entirely replaced by a udev
5240 "keyboard" builtin and a hwdb data file.
5242 * systemd will now honour the kernel's "quiet" command line
5243 argument also during late shutdown, resulting in a
5244 completely silent shutdown when used.
5246 * There's now an option to control the SO_REUSEPORT socket
5247 option in .socket units.
5249 * Instance units will now automatically get a per-template
5250 subslice of system.slice unless something else is explicitly
5251 configured. For example, instances of sshd@.service will now
5252 implicitly be placed in system-sshd.slice rather than
5253 system.slice as before.
5255 * Test coverage support may now be enabled at build time.
5257 Contributions from: Dave Reisner, Frederic Crozat, Harald
5258 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5259 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5260 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5261 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5262 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5263 Giokas, Zbigniew Jędrzejewski-Szmek
5265 — Berlin, 2013-07-23
5269 * Two new unit types have been introduced:
5271 Scope units are very similar to service units, however, are
5272 created out of pre-existing processes — instead of PID 1
5273 forking off the processes. By using scope units it is
5274 possible for system services and applications to group their
5275 own child processes (worker processes) in a powerful way
5276 which then maybe used to organize them, or kill them
5277 together, or apply resource limits on them.
5279 Slice units may be used to partition system resources in an
5280 hierarchical fashion and then assign other units to them. By
5281 default there are now three slices: system.slice (for all
5282 system services), user.slice (for all user sessions),
5283 machine.slice (for VMs and containers).
5285 Slices and scopes have been introduced primarily in
5286 context of the work to move cgroup handling to a
5287 single-writer scheme, where only PID 1
5288 creates/removes/manages cgroups.
5290 * There's a new concept of "transient" units. In contrast to
5291 normal units these units are created via an API at runtime,
5292 not from configuration from disk. More specifically this
5293 means it is now possible to run arbitrary programs as
5294 independent services, with all execution parameters passed
5295 in via bus APIs rather than read from disk. Transient units
5296 make systemd substantially more dynamic then it ever was,
5297 and useful as a general batch manager.
5299 * logind has been updated to make use of scope and slice units
5300 for managing user sessions. As a user logs in he will get
5301 his own private slice unit, to which all sessions are added
5302 as scope units. We also added support for automatically
5303 adding an instance of user@.service for the user into the
5304 slice. Effectively logind will no longer create cgroup
5305 hierarchies on its own now, it will defer entirely to PID 1
5306 for this by means of scope, service and slice units. Since
5307 user sessions this way become entities managed by PID 1
5308 the output of "systemctl" is now a lot more comprehensive.
5310 * A new mini-daemon "systemd-machined" has been added which
5311 may be used by virtualization managers to register local
5312 VMs/containers. nspawn has been updated accordingly, and
5313 libvirt will be updated shortly. machined will collect a bit
5314 of meta information about the VMs/containers, and assign
5315 them their own scope unit (see above). The collected
5316 meta-data is then made available via the "machinectl" tool,
5317 and exposed in "ps" and similar tools. machined/machinectl
5318 is compile-time optional.
5320 * As discussed earlier, the low-level cgroup configuration
5321 options ControlGroup=, ControlGroupModify=,
5322 ControlGroupPersistent=, ControlGroupAttribute= have been
5323 removed. Please use high-level attribute settings instead as
5324 well as slice units.
5326 * A new bus call SetUnitProperties() has been added to alter
5327 various runtime parameters of a unit. This is primarily
5328 useful to alter cgroup parameters dynamically in a nice way,
5329 but will be extended later on to make more properties
5330 modifiable at runtime. systemctl gained a new set-properties
5331 command that wraps this call.
5333 * A new tool "systemd-run" has been added which can be used to
5334 run arbitrary command lines as transient services or scopes,
5335 while configuring a number of settings via the command
5336 line. This tool is currently very basic, however already
5337 very useful. We plan to extend this tool to even allow
5338 queuing of execution jobs with time triggers from the
5339 command line, similar in fashion to "at".
5341 * nspawn will now inform the user explicitly that kernels with
5342 audit enabled break containers, and suggest the user to turn
5345 * Support for detecting the IMA and AppArmor security
5346 frameworks with ConditionSecurity= has been added.
5348 * journalctl gained a new "-k" switch for showing only kernel
5349 messages, mimicking dmesg output; in addition to "--user"
5350 and "--system" switches for showing only user's own logs
5353 * systemd-delta can now show information about drop-in
5354 snippets extending unit files.
5356 * libsystemd-bus has been substantially updated but is still
5357 not available as public API.
5359 * systemd will now look for the "debug" argument on the kernel
5360 command line and enable debug logging, similar to what
5361 "systemd.log_level=debug" already did before.
5363 * "systemctl set-default", "systemctl get-default" has been
5364 added to configure the default.target symlink, which
5365 controls what to boot into by default.
5367 * "systemctl set-log-level" has been added as a convenient
5368 way to raise and lower systemd logging threshold.
5370 * "systemd-analyze plot" will now show the time the various
5371 generators needed for execution, as well as information
5372 about the unit file loading.
5374 * libsystemd-journal gained a new sd_journal_open_files() call
5375 for opening specific journal files. journactl also gained a
5376 new switch to expose this new functionality. Previously we
5377 only supported opening all files from a directory, or all
5378 files from the system, as opening individual files only is
5379 racy due to journal file rotation.
5381 * systemd gained the new DefaultEnvironment= setting in
5382 /etc/systemd/system.conf to set environment variables for
5385 * If a privileged process logs a journal message with the
5386 OBJECT_PID= field set, then journald will automatically
5387 augment this with additional OBJECT_UID=, OBJECT_GID=,
5388 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5389 system services want to log events about specific client
5390 processes. journactl/systemctl has been updated to make use
5391 of this information if all log messages regarding a specific
5394 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5395 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5396 Reisner, David Coppa, David King, David Strauss, Eelco
5397 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5398 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5399 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5400 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5401 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5402 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5403 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5404 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5405 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5406 Łukasz Stelmach, 장동준
5410 * The Python bindings gained some minimal support for the APIs
5411 exposed by libsystemd-logind.
5413 * ConditionSecurity= gained support for detecting SMACK. Since
5414 this condition already supports SELinux and AppArmor we only
5415 miss IMA for this. Patches welcome!
5417 Contributions from: Karol Lewandowski, Lennart Poettering,
5418 Zbigniew Jędrzejewski-Szmek
5422 * systemd-nspawn will now create /etc/resolv.conf if
5423 necessary, before bind-mounting the host's file onto it.
5425 * systemd-nspawn will now store meta information about a
5426 container on the container's cgroup as extended attribute
5427 fields, including the root directory.
5429 * The cgroup hierarchy has been reworked in many ways. All
5430 objects any of the components systemd creates in the cgroup
5431 tree are now suffixed. More specifically, user sessions are
5432 now placed in cgroups suffixed with ".session", users in
5433 cgroups suffixed with ".user", and nspawn containers in
5434 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5435 names are now escaped in a simple scheme to avoid collision
5436 of userspace object names with kernel filenames. This work
5437 is preparation for making these objects relocatable in the
5438 cgroup tree, in order to allow easy resource partitioning of
5439 these objects without causing naming conflicts.
5441 * systemctl list-dependencies gained the new switches
5442 --plain, --reverse, --after and --before.
5444 * systemd-inhibit now shows the process name of processes that
5445 have taken an inhibitor lock.
5447 * nss-myhostname will now also resolve "localhost"
5448 implicitly. This makes /etc/hosts an optional file and
5449 nicely handles that on IPv6 ::1 maps to both "localhost" and
5452 * libsystemd-logind.so gained a new call
5453 sd_get_machine_names() to enumerate running containers and
5454 VMs (currently only supported by very new libvirt and
5455 nspawn). sd_login_monitor can now be used to watch
5456 VMs/containers coming and going.
5458 * .include is not allowed recursively anymore, and only in
5459 unit files. Usually it is better to use drop-in snippets in
5460 .d/*.conf anyway, as introduced with systemd 198.
5462 * systemd-analyze gained a new "critical-chain" command that
5463 determines the slowest chain of units run during system
5464 boot-up. It is very useful for tracking down where
5465 optimizing boot time is the most beneficial.
5467 * systemd will no longer allow manipulating service paths in
5468 the name=systemd:/system cgroup tree using ControlGroup= in
5469 units. (But is still fine with it in all other dirs.)
5471 * There's a new systemd-nspawn@.service service file that may
5472 be used to easily run nspawn containers as system
5473 services. With the container's root directory in
5474 /var/lib/container/foobar it is now sufficient to run
5475 "systemctl start systemd-nspawn@foobar.service" to boot it.
5477 * systemd-cgls gained a new parameter "--machine" to list only
5478 the processes within a certain container.
5480 * ConditionSecurity= now can check for "apparmor". We still
5481 are lacking checks for SMACK and IMA for this condition
5482 check though. Patches welcome!
5484 * A new configuration file /etc/systemd/sleep.conf has been
5485 added that may be used to configure which kernel operation
5486 systemd is supposed to execute when "suspend", "hibernate"
5487 or "hybrid-sleep" is requested. This makes the new kernel
5488 "freeze" state accessible to the user.
5490 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5491 the passed argument if applicable.
5493 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5494 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5495 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5496 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5497 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5498 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5499 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5504 * The output of 'systemctl list-jobs' got some polishing. The
5505 '--type=' argument may now be passed more than once. A new
5506 command 'systemctl list-sockets' has been added which shows
5507 a list of kernel sockets systemd is listening on with the
5508 socket units they belong to, plus the units these socket
5511 * The experimental libsystemd-bus library got substantial
5512 updates to work in conjunction with the (also experimental)
5513 kdbus kernel project. It works well enough to exchange
5514 messages with some sophistication. Note that kdbus is not
5515 ready yet, and the library is mostly an elaborate test case
5516 for now, and not installable.
5518 * systemd gained a new unit 'systemd-static-nodes.service'
5519 that generates static device nodes earlier during boot, and
5520 can run in conjunction with udev.
5522 * libsystemd-login gained a new call sd_pid_get_user_unit()
5523 to retrieve the user systemd unit a process is running
5524 in. This is useful for systems where systemd is used as
5527 * systemd-nspawn now places all containers in the new /machine
5528 top-level cgroup directory in the name=systemd
5529 hierarchy. libvirt will soon do the same, so that we get a
5530 uniform separation of /system, /user and /machine for system
5531 services, user processes and containers/virtual
5532 machines. This new cgroup hierarchy is also useful to stick
5533 stable names to specific container instances, which can be
5534 recognized later this way (this name may be controlled
5535 via systemd-nspawn's new -M switch). libsystemd-login also
5536 gained a new call sd_pid_get_machine_name() to retrieve the
5537 name of the container/VM a specific process belongs to.
5539 * bootchart can now store its data in the journal.
5541 * libsystemd-journal gained a new call
5542 sd_journal_add_conjunction() for AND expressions to the
5543 matching logic. This can be used to express more complex
5544 logical expressions.
5546 * journactl can now take multiple --unit= and --user-unit=
5549 * The cryptsetup logic now understands the "luks.key=" kernel
5550 command line switch for specifying a file to read the
5551 decryption key from. Also, if a configured key file is not
5552 found the tool will now automatically fall back to prompting
5555 * Python systemd.journal module was updated to wrap recently
5556 added functions from libsystemd-journal. The interface was
5557 changed to bring the low level interface in s.j._Reader
5558 closer to the C API, and the high level interface in
5559 s.j.Reader was updated to wrap and convert all data about
5562 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5563 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5564 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5565 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5566 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5567 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5571 * journalctl --update-catalog now understands a new --root=
5572 option to operate on catalogs found in a different root
5575 * During shutdown after systemd has terminated all running
5576 services a final killing loop kills all remaining left-over
5577 processes. We will now print the name of these processes
5578 when we send SIGKILL to them, since this usually indicates a
5581 * If /etc/crypttab refers to password files stored on
5582 configured mount points automatic dependencies will now be
5583 generated to ensure the specific mount is established first
5584 before the key file is attempted to be read.
5586 * 'systemctl status' will now show information about the
5587 network sockets a socket unit is listening on.
5589 * 'systemctl status' will also shown information about any
5590 drop-in configuration file for units. (Drop-In configuration
5591 files in this context are files such as
5592 /etc/systemd/systemd/foobar.service.d/*.conf)
5594 * systemd-cgtop now optionally shows summed up CPU times of
5595 cgroups. Press '%' while running cgtop to switch between
5596 percentage and absolute mode. This is useful to determine
5597 which cgroups use up the most CPU time over the entire
5598 runtime of the system. systemd-cgtop has also been updated
5599 to be 'pipeable' for processing with further shell tools.
5601 * 'hostnamectl set-hostname' will now allow setting of FQDN
5604 * The formatting and parsing of time span values has been
5605 changed. The parser now understands fractional expressions
5606 such as "5.5h". The formatter will now output fractional
5607 expressions for all time spans under 1min, i.e. "5.123456s"
5608 rather than "5s 123ms 456us". For time spans under 1s
5609 millisecond values are shown, for those under 1ms
5610 microsecond values are shown. This should greatly improve
5611 all time-related output of systemd.
5613 * libsystemd-login and libsystemd-journal gained new
5614 functions for querying the poll() events mask and poll()
5615 timeout value for integration into arbitrary event
5618 * localectl gained the ability to list available X11 keymaps
5619 (models, layouts, variants, options).
5621 * 'systemd-analyze dot' gained the ability to filter for
5622 specific units via shell-style globs, to create smaller,
5623 more useful graphs. I.e. it is now possible to create simple
5624 graphs of all the dependencies between only target units, or
5625 of all units that Avahi has dependencies with.
5627 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5628 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5629 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5630 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5631 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5632 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5633 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5637 * The boot-time readahead implementation for rotating media
5638 will now read the read-ahead data in multiple passes which
5639 consist of all read requests made in equidistant time
5640 intervals. This means instead of strictly reading read-ahead
5641 data in its physical order on disk we now try to find a
5642 middle ground between physical and access time order.
5644 * /etc/os-release files gained a new BUILD_ID= field for usage
5645 on operating systems that provide continuous builds of OS
5648 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5649 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5650 William Douglas, Zbigniew Jędrzejewski-Szmek
5654 * systemd-python gained an API exposing libsystemd-daemon.
5656 * The SMACK setup logic gained support for uploading CIPSO
5659 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5660 ReadOnlyDirectories= and InaccessibleDirectories= has
5661 changed. The private /tmp and /var/tmp directories are now
5662 shared by all processes of a service (which means
5663 ExecStartPre= may now leave data in /tmp that ExecStart= of
5664 the same service can still access). When a service is
5665 stopped its temporary directories are immediately deleted
5666 (normal clean-up with tmpfiles is still done in addition to
5669 * By default, systemd will now set a couple of sysctl
5670 variables in the kernel: the safe sysrq options are turned
5671 on, IP route verification is turned on, and source routing
5672 disabled. The recently added hardlink and softlink
5673 protection of the kernel is turned on. These settings should
5674 be reasonably safe, and good defaults for all new systems.
5676 * The predictable network naming logic may now be turned off
5677 with a new kernel command line switch: net.ifnames=0.
5679 * A new libsystemd-bus module has been added that implements a
5680 pretty complete D-Bus client library. For details see:
5682 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5684 * journald will now explicitly flush the journal files to disk
5685 at the latest 5min after each write. The file will then also
5686 be marked offline until the next write. This should increase
5687 reliability in case of a crash. The synchronization delay
5688 can be configured via SyncIntervalSec= in journald.conf.
5690 * There's a new remote-fs-setup.target unit that can be used
5691 to pull in specific services when at least one remote file
5692 system is to be mounted.
5694 * There are new targets timers.target and paths.target as
5695 canonical targets to pull user timer and path units in
5696 from. This complements sockets.target with a similar
5697 purpose for socket units.
5699 * libudev gained a new call udev_device_set_attribute_value()
5700 to set sysfs attributes of a device.
5702 * The udev daemon now sets the default number of worker
5703 processes executed in parallel based on the number of available
5704 CPUs instead of the amount of available RAM. This is supposed
5705 to provide a more reliable default and limit a too aggressive
5706 parallelism for setups with 1000s of devices connected.
5708 Contributions from: Auke Kok, Colin Walters, Cristian
5709 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5710 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5711 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5712 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5713 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5714 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5715 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5716 Zbigniew Jędrzejewski-Szmek
5720 * Configuration of unit files may now be extended via drop-in
5721 files without having to edit/override the unit files
5722 themselves. More specifically, if the administrator wants to
5723 change one value for a service file foobar.service he can
5724 now do so by dropping in a configuration snippet into
5725 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5726 will load all these snippets and apply them on top of the
5727 main unit configuration file, possibly extending or
5728 overriding its settings. Using these drop-in snippets is
5729 generally nicer than the two earlier options for changing
5730 unit files locally: copying the files from
5731 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5732 them there; or creating a new file in /etc/systemd/system/
5733 that incorporates the original one via ".include". Drop-in
5734 snippets into these .d/ directories can be placed in any
5735 directory systemd looks for units in, and the usual
5736 overriding semantics between /usr/lib, /etc and /run apply
5739 * Most unit file settings which take lists of items can now be
5740 reset by assigning the empty string to them. For example,
5741 normally, settings such as Environment=FOO=BAR append a new
5742 environment variable assignment to the environment block,
5743 each time they are used. By assigning Environment= the empty
5744 string the environment block can be reset to empty. This is
5745 particularly useful with the .d/*.conf drop-in snippets
5746 mentioned above, since this adds the ability to reset list
5747 settings from vendor unit files via these drop-ins.
5749 * systemctl gained a new "list-dependencies" command for
5750 listing the dependencies of a unit recursively.
5752 * Inhibitors are now honored and listed by "systemctl
5753 suspend", "systemctl poweroff" (and similar) too, not only
5754 GNOME. These commands will also list active sessions by
5757 * Resource limits (as exposed by the various control group
5758 controllers) can now be controlled dynamically at runtime
5759 for all units. More specifically, you can now use a command
5760 like "systemctl set-cgroup-attr foobar.service cpu.shares
5761 2000" to alter the CPU shares a specific service gets. These
5762 settings are stored persistently on disk, and thus allow the
5763 administrator to easily adjust the resource usage of
5764 services with a few simple commands. This dynamic resource
5765 management logic is also available to other programs via the
5766 bus. Almost any kernel cgroup attribute and controller is
5769 * systemd-vconsole-setup will now copy all font settings to
5770 all allocated VTs, where it previously applied them only to
5773 * libsystemd-login gained the new sd_session_get_tty() API
5776 * This release drops support for a few legacy or
5777 distribution-specific LSB facility names when parsing init
5778 scripts: $x-display-manager, $mail-transfer-agent,
5779 $mail-transport-agent, $mail-transfer-agent, $smtp,
5780 $null. Also, the mail-transfer-agent.target unit backing
5781 this has been removed. Distributions which want to retain
5782 compatibility with this should carry the burden for
5783 supporting this themselves and patch support for these back
5784 in, if they really need to. Also, the facilities $syslog and
5785 $local_fs are now ignored, since systemd does not support
5786 early-boot LSB init scripts anymore, and these facilities
5787 are implied anyway for normal services. syslog.target has
5790 * There are new bus calls on PID1's Manager object for
5791 cancelling jobs, and removing snapshot units. Previously,
5792 both calls were only available on the Job and Snapshot
5795 * systemd-journal-gatewayd gained SSL support.
5797 * The various "environment" files, such as /etc/locale.conf
5798 now support continuation lines with a backslash ("\") as
5799 last character in the line, similarly in style (but different)
5800 to how this is supported in shells.
5802 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5803 now implicitly appended to every log entry logged. systemctl
5804 has been updated to filter by this field when operating on a
5805 user systemd instance.
5807 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5808 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5809 the container. This makes it easier to boot unmodified
5810 Fedora systems in a container, which however still requires
5811 audit=0 to be passed on the kernel command line. Auditing in
5812 kernel and userspace is unfortunately still too broken in
5813 context of containers, hence we recommend compiling it out
5814 of the kernel or using audit=0. Hopefully this will be fixed
5815 one day for good in the kernel.
5817 * nspawn gained the new --bind= and --bind-ro= parameters to
5818 bind mount specific directories from the host into the
5821 * nspawn will now mount its own devpts file system instance
5822 into the container, in order not to leak pty devices from
5823 the host into the container.
5825 * systemd will now read the firmware boot time performance
5826 information from the EFI variables, if the used boot loader
5827 supports this, and takes it into account for boot performance
5828 analysis via "systemd-analyze". This is currently supported
5829 only in conjunction with Gummiboot, but could be supported
5830 by other boot loaders too. For details see:
5832 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5834 * A new generator has been added that automatically mounts the
5835 EFI System Partition (ESP) to /boot, if that directory
5836 exists, is empty, and no other file system has been
5837 configured to be mounted there.
5839 * logind will now send out PrepareForSleep(false) out
5840 unconditionally, after coming back from suspend. This may be
5841 used by applications as asynchronous notification for
5842 system resume events.
5844 * "systemctl unlock-sessions" has been added, that allows
5845 unlocking the screens of all user sessions at once, similar
5846 to how "systemctl lock-sessions" already locked all users
5847 sessions. This is backed by a new D-Bus call UnlockSessions().
5849 * "loginctl seat-status" will now show the master device of a
5850 seat. (i.e. the device of a seat that needs to be around for
5851 the seat to be considered available, usually the graphics
5854 * tmpfiles gained a new "X" line type, that allows
5855 configuration of files and directories (with wildcards) that
5856 shall be excluded from automatic cleanup ("aging").
5858 * udev default rules set the device node permissions now only
5859 at "add" events, and do not change them any longer with a
5860 later "change" event.
5862 * The log messages for lid events and power/sleep keypresses
5863 now carry a message ID.
5865 * We now have a substantially larger unit test suite, but this
5866 continues to be work in progress.
5868 * udevadm hwdb gained a new --root= parameter to change the
5869 root directory to operate relative to.
5871 * logind will now issue a background sync() request to the kernel
5872 early at shutdown, so that dirty buffers are flushed to disk early
5873 instead of at the last moment, in order to optimize shutdown
5876 * A new bootctl tool has been added that is an interface for
5877 certain boot loader operations. This is currently a preview
5878 and is likely to be extended into a small mechanism daemon
5879 like timedated, localed, hostnamed, and can be used by
5880 graphical UIs to enumerate available boot options, and
5881 request boot into firmware operations.
5883 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5884 the rest of the package. It also has been updated to work
5885 correctly in initrds.
5887 * Policykit previously has been runtime optional, and is now
5888 also compile time optional via a configure switch.
5890 * systemd-analyze has been reimplemented in C. Also "systemctl
5891 dot" has moved into systemd-analyze.
5893 * "systemctl status" with no further parameters will now print
5894 the status of all active or failed units.
5896 * Operations such as "systemctl start" can now be executed
5897 with a new mode "--irreversible" which may be used to queue
5898 operations that cannot accidentally be reversed by a later
5899 job queuing. This is by default used to make shutdown
5900 requests more robust.
5902 * The Python API of systemd now gained a new module for
5903 reading journal files.
5905 * A new tool kernel-install has been added that can install
5906 kernel images according to the Boot Loader Specification:
5908 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5910 * Boot time console output has been improved to provide
5911 animated boot time output for hanging jobs.
5913 * A new tool systemd-activate has been added which can be used
5914 to test socket activation with, directly from the command
5915 line. This should make it much easier to test and debug
5916 socket activation in daemons.
5918 * journalctl gained a new "--reverse" (or -r) option to show
5919 journal output in reverse order (i.e. newest line first).
5921 * journalctl gained a new "--pager-end" (or -e) option to jump
5922 to immediately jump to the end of the journal in the
5923 pager. This is only supported in conjunction with "less".
5925 * journalctl gained a new "--user-unit=" option, that works
5926 similarly to "--unit=" but filters for user units rather than
5929 * A number of unit files to ease adoption of systemd in
5930 initrds has been added. This moves some minimal logic from
5931 the various initrd implementations into systemd proper.
5933 * The journal files are now owned by a new group
5934 "systemd-journal", which exists specifically to allow access
5935 to the journal, and nothing else. Previously, we used the
5936 "adm" group for that, which however possibly covers more
5937 than just journal/log file access. This new group is now
5938 already used by systemd-journal-gatewayd to ensure this
5939 daemon gets access to the journal files and as little else
5940 as possible. Note that "make install" will also set FS ACLs
5941 up for /var/log/journal to give "adm" and "wheel" read
5942 access to it, in addition to "systemd-journal" which owns
5943 the journal files. We recommend that packaging scripts also
5944 add read access to "adm" + "wheel" to /var/log/journal, and
5945 all existing/future journal files. To normal users and
5946 administrators little changes, however packagers need to
5947 ensure to create the "systemd-journal" system group at
5948 package installation time.
5950 * The systemd-journal-gatewayd now runs as unprivileged user
5951 systemd-journal-gateway:systemd-journal-gateway. Packaging
5952 scripts need to create these system user/group at
5955 * timedated now exposes a new boolean property CanNTP that
5956 indicates whether a local NTP service is available or not.
5958 * systemd-detect-virt will now also detect xen PVs
5960 * The pstore file system is now mounted by default, if it is
5963 * In addition to the SELinux and IMA policies we will now also
5964 load SMACK policies at early boot.
5966 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5967 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5968 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5969 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5970 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5971 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5972 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5973 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5974 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5975 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5976 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5977 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5978 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5979 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5983 * Timer units now support calendar time events in addition to
5984 monotonic time events. That means you can now trigger a unit
5985 based on a calendar time specification such as "Thu,Fri
5986 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5987 or fifth day of any month of the year 2013, given that it is
5988 a thursday or friday. This brings timer event support
5989 considerably closer to cron's capabilities. For details on
5990 the supported calendar time specification language see
5993 * udev now supports a number of different naming policies for
5994 network interfaces for predictable names, and a combination
5995 of these policies is now the default. Please see this wiki
5996 document for details:
5998 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6000 * Auke Kok's bootchart implementation has been added to the
6001 systemd tree. It is an optional component that can graph the
6002 boot in quite some detail. It is one of the best bootchart
6003 implementations around and minimal in its code and
6006 * nss-myhostname has been integrated into the systemd source
6007 tree. nss-myhostname guarantees that the local hostname
6008 always stays resolvable via NSS. It has been a weak
6009 requirement of systemd-hostnamed since a long time, and
6010 since its code is actually trivial we decided to just
6011 include it in systemd's source tree. It can be turned off
6012 with a configure switch.
6014 * The read-ahead logic is now capable of properly detecting
6015 whether a btrfs file system is on SSD or rotating media, in
6016 order to optimize the read-ahead scheme. Previously, it was
6017 only capable of detecting this on traditional file systems
6020 * In udev, additional device properties are now read from the
6021 IAB in addition to the OUI database. Also, Bluetooth company
6022 identities are attached to the devices as well.
6024 * In service files %U may be used as specifier that is
6025 replaced by the configured user name of the service.
6027 * nspawn may now be invoked without a controlling TTY. This
6028 makes it suitable for invocation as its own service. This
6029 may be used to set up a simple containerized server system
6030 using only core OS tools.
6032 * systemd and nspawn can now accept socket file descriptors
6033 when they are started for socket activation. This enables
6034 implementation of socket activated nspawn
6035 containers. i.e. think about autospawning an entire OS image
6036 when the first SSH or HTTP connection is received. We expect
6037 that similar functionality will also be added to libvirt-lxc
6040 * journalctl will now suppress ANSI color codes when
6041 presenting log data.
6043 * systemctl will no longer show control group information for
6044 a unit if the control group is empty anyway.
6046 * logind can now automatically suspend/hibernate/shutdown the
6049 * /etc/machine-info and hostnamed now also expose the chassis
6050 type of the system. This can be used to determine whether
6051 the local system is a laptop, desktop, handset or
6052 tablet. This information may either be configured by the
6053 user/vendor or is automatically determined from ACPI and DMI
6054 information if possible.
6056 * A number of PolicyKit actions are now bound together with
6057 "imply" rules. This should simplify creating UIs because
6058 many actions will now authenticate similar ones as well.
6060 * Unit files learnt a new condition ConditionACPower= which
6061 may be used to conditionalize a unit depending on whether an
6062 AC power source is connected or not, of whether the system
6063 is running on battery power.
6065 * systemctl gained a new "is-failed" verb that may be used in
6066 shell scripts and suchlike to check whether a specific unit
6067 is in the "failed" state.
6069 * The EnvironmentFile= setting in unit files now supports file
6070 globbing, and can hence be used to easily read a number of
6071 environment files at once.
6073 * systemd will no longer detect and recognize specific
6074 distributions. All distribution-specific #ifdeffery has been
6075 removed, systemd is now fully generic and
6076 distribution-agnostic. Effectively, not too much is lost as
6077 a lot of the code is still accessible via explicit configure
6078 switches. However, support for some distribution specific
6079 legacy configuration file formats has been dropped. We
6080 recommend distributions to simply adopt the configuration
6081 files everybody else uses now and convert the old
6082 configuration from packaging scripts. Most distributions
6083 already did that. If that's not possible or desirable,
6084 distributions are welcome to forward port the specific
6085 pieces of code locally from the git history.
6087 * When logging a message about a unit systemd will now always
6088 log the unit name in the message meta data.
6090 * localectl will now also discover system locale data that is
6091 not stored in locale archives, but directly unpacked.
6093 * logind will no longer unconditionally use framebuffer
6094 devices as seat masters, i.e. as devices that are required
6095 to be existing before a seat is considered preset. Instead,
6096 it will now look for all devices that are tagged as
6097 "seat-master" in udev. By default, framebuffer devices will
6098 be marked as such, but depending on local systems, other
6099 devices might be marked as well. This may be used to
6100 integrate graphics cards using closed source drivers (such
6101 as NVidia ones) more nicely into logind. Note however, that
6102 we recommend using the open source NVidia drivers instead,
6103 and no udev rules for the closed-source drivers will be
6104 shipped from us upstream.
6106 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6107 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6108 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6109 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6110 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6111 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6112 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6113 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6114 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6115 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6116 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6117 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6122 * udev gained support for loading additional device properties
6123 from an indexed database that is keyed by vendor/product IDs
6124 and similar device identifiers. For the beginning this
6125 "hwdb" is populated with data from the well-known PCI and
6126 USB database, but also includes PNP, ACPI and OID data. In
6127 the longer run this indexed database shall grow into
6128 becoming the one central database for non-essential
6129 userspace device metadata. Previously, data from the PCI/USB
6130 database was only attached to select devices, since the
6131 lookup was a relatively expensive operation due to O(n) time
6132 complexity (with n being the number of entries in the
6133 database). Since this is now O(1), we decided to add in this
6134 data for all devices where this is available, by
6135 default. Note that the indexed database needs to be rebuilt
6136 when new data files are installed. To achieve this you need
6137 to update your packaging scripts to invoke "udevadm hwdb
6138 --update" after installation of hwdb data files. For
6139 RPM-based distributions we introduced the new
6140 %udev_hwdb_update macro for this purpose.
6142 * The Journal gained support for the "Message Catalog", an
6143 indexed database to link up additional information with
6144 journal entries. For further details please check:
6146 https://www.freedesktop.org/wiki/Software/systemd/catalog
6148 The indexed message catalog database also needs to be
6149 rebuilt after installation of message catalog files. Use
6150 "journalctl --update-catalog" for this. For RPM-based
6151 distributions we introduced the %journal_catalog_update
6152 macro for this purpose.
6154 * The Python Journal bindings gained support for the standard
6155 Python logging framework.
6157 * The Journal API gained new functions for checking whether
6158 the underlying file system of a journal file is capable of
6159 properly reporting file change notifications, or whether
6160 applications that want to reflect journal changes "live"
6161 need to recheck journal files continuously in appropriate
6164 * It is now possible to set the "age" field for tmpfiles
6165 entries to 0, indicating that files matching this entry
6166 shall always be removed when the directories are cleaned up.
6168 * coredumpctl gained a new "gdb" verb which invokes gdb
6169 right-away on the selected coredump.
6171 * There's now support for "hybrid sleep" on kernels that
6172 support this, in addition to "suspend" and "hibernate". Use
6173 "systemctl hybrid-sleep" to make use of this.
6175 * logind's HandleSuspendKey= setting (and related settings)
6176 now gained support for a new "lock" setting to simply
6177 request the screen lock on all local sessions, instead of
6178 actually executing a suspend or hibernation.
6180 * systemd will now mount the EFI variables file system by
6183 * Socket units now gained support for configuration of the
6184 SMACK security label.
6186 * timedatectl will now output the time of the last and next
6187 daylight saving change.
6189 * We dropped support for various legacy and distro-specific
6190 concepts, such as insserv, early-boot SysV services
6191 (i.e. those for non-standard runlevels such as 'b' or 'S')
6192 or ArchLinux /etc/rc.conf support. We recommend the
6193 distributions who still need support this to either continue
6194 to maintain the necessary patches downstream, or find a
6195 different solution. (Talk to us if you have questions!)
6197 * Various systemd components will now bypass PolicyKit checks
6198 for root and otherwise handle properly if PolicyKit is not
6199 found to be around. This should fix most issues for
6200 PolicyKit-less systems. Quite frankly this should have been
6201 this way since day one. It is absolutely our intention to
6202 make systemd work fine on PolicyKit-less systems, and we
6203 consider it a bug if something does not work as it should if
6204 PolicyKit is not around.
6206 * For embedded systems it is now possible to build udev and
6207 systemd without blkid and/or kmod support.
6209 * "systemctl switch-root" is now capable of switching root
6210 more than once. I.e. in addition to transitions from the
6211 initrd to the host OS it is now possible to transition to
6212 further OS images from the host. This is useful to implement
6213 offline updating tools.
6215 * Various other additions have been made to the RPM macros
6216 shipped with systemd. Use %udev_rules_update() after
6217 installing new udev rules files. %_udevhwdbdir,
6218 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6219 %_sysctldir are now available which resolve to the right
6220 directories for packages to place various data files in.
6222 * journalctl gained the new --full switch (in addition to
6223 --all, to disable ellipsation for long messages.
6225 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6226 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6227 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6228 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6229 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6230 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6231 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6232 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6233 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6237 * journalctl gained new --since= and --until= switches to
6238 filter by time. It also now supports nice filtering for
6239 units via --unit=/-u.
6241 * Type=oneshot services may use ExecReload= and do the
6244 * The journal daemon now supports time-based rotation and
6245 vacuuming, in addition to the usual disk-space based
6248 * The journal will now index the available field values for
6249 each field name. This enables clients to show pretty drop
6250 downs of available match values when filtering. The bash
6251 completion of journalctl has been updated
6252 accordingly. journalctl gained a new switch -F to list all
6253 values a certain field takes in the journal database.
6255 * More service events are now written as structured messages
6256 to the journal, and made recognizable via message IDs.
6258 * The timedated, localed and hostnamed mini-services which
6259 previously only provided support for changing time, locale
6260 and hostname settings from graphical DEs such as GNOME now
6261 also have a minimal (but very useful) text-based client
6262 utility each. This is probably the nicest way to changing
6263 these settings from the command line now, especially since
6264 it lists available options and is fully integrated with bash
6267 * There's now a new tool "systemd-coredumpctl" to list and
6268 extract coredumps from the journal.
6270 * We now install a README each in /var/log/ and
6271 /etc/rc.d/init.d explaining where the system logs and init
6272 scripts went. This hopefully should help folks who go to
6273 that dirs and look into the otherwise now empty void and
6274 scratch their heads.
6276 * When user-services are invoked (by systemd --user) the
6277 $MANAGERPID env var is set to the PID of systemd.
6279 * SIGRTMIN+24 when sent to a --user instance will now result
6280 in immediate termination of systemd.
6282 * gatewayd received numerous feature additions such as a
6283 "follow" mode, for live syncing and filtering.
6285 * browse.html now allows filtering and showing detailed
6286 information on specific entries. Keyboard navigation and
6287 mouse screen support has been added.
6289 * gatewayd/journalctl now supports HTML5/JSON
6290 Server-Sent-Events as output.
6292 * The SysV init script compatibility logic will now
6293 heuristically determine whether a script supports the
6294 "reload" verb, and only then make this available as
6297 * "systemctl status --follow" has been removed, use "journalctl
6300 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6301 have been removed since they are hardly useful to be
6304 * And I'd like to take the opportunity to specifically mention
6305 Zbigniew for his great contributions. Zbigniew, you rock!
6307 Contributions from: Andrew Eikum, Christian Hesse, Colin
6308 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6309 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6310 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6311 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6312 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6313 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6317 * If /etc/vconsole.conf is non-existent or empty we will no
6318 longer load any console font or key map at boot by
6319 default. Instead the kernel defaults will be left
6320 intact. This is definitely the right thing to do, as no
6321 configuration should mean no configuration, and hard-coding
6322 font names that are different on all archs is probably a bad
6323 idea. Also, the kernel default key map and font should be
6324 good enough for most cases anyway, and mostly identical to
6325 the userspace fonts/key maps we previously overloaded them
6326 with. If distributions want to continue to default to a
6327 non-kernel font or key map they should ship a default
6328 /etc/vconsole.conf with the appropriate contents.
6330 Contributions from: Colin Walters, Daniel J Walsh, Dave
6331 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6332 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6336 * journalctl gained a new --cursor= switch to show entries
6337 starting from the specified location in the journal.
6339 * We now enforce a size limit on journal entry fields exported
6340 with "-o json" in journalctl. Fields larger than 4K will be
6341 assigned null. This can be turned off with --all.
6343 * An (optional) journal gateway daemon is now available as
6344 "systemd-journal-gatewayd.service". This service provides
6345 access to the journal via HTTP and JSON. This functionality
6346 will be used to implement live log synchronization in both
6347 pull and push modes, but has various other users too, such
6348 as easy log access for debugging of embedded devices. Right
6349 now it is already useful to retrieve the journal via HTTP:
6351 # systemctl start systemd-journal-gatewayd.service
6352 # wget http://localhost:19531/entries
6354 This will download the journal contents in a
6355 /var/log/messages compatible format. The same as JSON:
6357 # curl -H"Accept: application/json" http://localhost:19531/entries
6359 This service is also accessible via a web browser where a
6360 single static HTML5 app is served that uses the JSON logic
6361 to enable the user to do some basic browsing of the
6362 journal. This will be extended later on. Here's an example
6363 screenshot of this app in its current state:
6365 http://0pointer.de/public/journal-gatewayd
6367 Contributions from: Kay Sievers, Lennart Poettering, Robert
6368 Milasan, Tom Gundersen
6372 * The bash completion logic is now available for journalctl
6375 * We do not mount the "cpuset" controller anymore together with
6376 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6377 started if no parameters are assigned to it. "cpuset" hence
6378 broke code that assumed it could create "cpu" groups and
6381 * journalctl -f will now subscribe to terminal size changes,
6382 and line break accordingly.
6384 Contributions from: Dave Reisner, Kay Sievers, Lennart
6385 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6389 * nspawn will now create a symlink /etc/localtime in the
6390 container environment, copying the host's timezone
6391 setting. Previously this has been done via a bind mount, but
6392 since symlinks cannot be bind mounted this has now been
6393 changed to create/update the appropriate symlink.
6395 * journalctl -n's line number argument is now optional, and
6396 will default to 10 if omitted.
6398 * journald will now log the maximum size the journal files may
6399 take up on disk. This is particularly useful if the default
6400 built-in logic of determining this parameter from the file
6401 system size is used. Use "systemctl status
6402 systemd-journald.service" to see this information.
6404 * The multi-seat X wrapper tool has been stripped down. As X
6405 is now capable of enumerating graphics devices via udev in a
6406 seat-aware way the wrapper is not strictly necessary
6407 anymore. A stripped down temporary stop-gap is still shipped
6408 until the upstream display managers have been updated to
6409 fully support the new X logic. Expect this wrapper to be
6410 removed entirely in one of the next releases.
6412 * HandleSleepKey= in logind.conf has been split up into
6413 HandleSuspendKey= and HandleHibernateKey=. The old setting
6414 is not available anymore. X11 and the kernel are
6415 distinguishing between these keys and we should too. This
6416 also means the inhibition lock for these keys has been split
6419 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6420 Poettering, Lukas Nykryn, Václav Pavlín
6424 * Whenever a unit changes state we will now log this to the
6425 journal and show along the unit's own log output in
6428 * ConditionPathIsMountPoint= can now properly detect bind
6429 mount points too. (Previously, a bind mount of one file
6430 system to another place in the same file system could not be
6431 detected as mount, since they shared struct stat's st_dev
6434 * We will now mount the cgroup controllers cpu, cpuacct,
6435 cpuset and the controllers net_cls, net_prio together by
6438 * nspawn containers will now have a virtualized boot
6439 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6440 over with a randomized ID at container initialization). This
6441 has the effect of making "journalctl -b" do the right thing
6444 * The JSON output journal serialization has been updated not
6445 to generate "endless" list objects anymore, but rather one
6446 JSON object per line. This is more in line how most JSON
6447 parsers expect JSON objects. The new output mode
6448 "json-pretty" has been added to provide similar output, but
6449 neatly aligned for readability by humans.
6451 * We dropped all explicit sync() invocations in the shutdown
6452 code. The kernel does this implicitly anyway in the kernel
6453 reboot() syscall. halt(8)'s -n option is now a compatibility
6456 * We now support virtualized reboot() in containers, as
6457 supported by newer kernels. We will fall back to exit() if
6458 CAP_SYS_REBOOT is not available to the container. Also,
6459 nspawn makes use of this now and will actually reboot the
6460 container if the containerized OS asks for that.
6462 * journalctl will only show local log output by default
6463 now. Use --merge (-m) to show remote log output, too.
6465 * libsystemd-journal gained the new sd_journal_get_usage()
6466 call to determine the current disk usage of all journal
6467 files. This is exposed in the new "journalctl --disk-usage"
6470 * journald gained a new configuration setting SplitMode= in
6471 journald.conf which may be used to control how user journals
6472 are split off. See journald.conf(5) for details.
6474 * A new condition type ConditionFileNotEmpty= has been added.
6476 * tmpfiles' "w" lines now support file globbing, to write
6477 multiple files at once.
6479 * We added Python bindings for the journal submission
6480 APIs. More Python APIs for a number of selected APIs will
6481 likely follow. Note that we intend to add native bindings
6482 only for the Python language, as we consider it common
6483 enough to deserve bindings shipped within systemd. There are
6484 various projects outside of systemd that provide bindings
6485 for languages such as PHP or Lua.
6487 * Many conditions will now resolve specifiers such as %i. In
6488 addition, PathChanged= and related directives of .path units
6489 now support specifiers as well.
6491 * There's now a new RPM macro definition for the system preset
6494 * journald will now warn if it ca not forward a message to the
6495 syslog daemon because its socket is full.
6497 * timedated will no longer write or process /etc/timezone,
6498 except on Debian. As we do not support late mounted /usr
6499 anymore /etc/localtime always being a symlink is now safe,
6500 and hence the information in /etc/timezone is not necessary
6503 * logind will now always reserve one VT for a text getty (VT6
6504 by default). Previously if more than 6 X sessions where
6505 started they took up all the VTs with auto-spawned gettys,
6506 so that no text gettys were available anymore.
6508 * udev will now automatically inform the btrfs kernel logic
6509 about btrfs RAID components showing up. This should make
6510 simple hotplug based btrfs RAID assembly work.
6512 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6513 (but not for its children which will stay at the kernel
6514 default). This should allow setups with a lot more listening
6517 * systemd will now always pass the configured timezone to the
6518 kernel at boot. timedated will do the same when the timezone
6521 * logind's inhibition logic has been updated. By default,
6522 logind will now handle the lid switch, the power and sleep
6523 keys all the time, even in graphical sessions. If DEs want
6524 to handle these events on their own they should take the new
6525 handle-power-key, handle-sleep-key and handle-lid-switch
6526 inhibitors during their runtime. A simple way to achieve
6527 that is to invoke the DE wrapped in an invocation of:
6529 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6531 * Access to unit operations is now checked via SELinux taking
6532 the unit file label and client process label into account.
6534 * systemd will now notify the administrator in the journal
6535 when he over-mounts a non-empty directory.
6537 * There are new specifiers that are resolved in unit files,
6538 for the host name (%H), the machine ID (%m) and the boot ID
6541 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6542 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6543 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6544 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6545 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6546 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6547 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6551 * Support for reading structured kernel messages from
6552 /dev/kmsg has now been added and is enabled by default.
6554 * Support for reading kernel messages from /proc/kmsg has now
6555 been removed. If you want kernel messages in the journal
6556 make sure to run a recent kernel (>= 3.5) that supports
6557 reading structured messages from /dev/kmsg (see
6558 above). /proc/kmsg is now exclusive property of classic
6559 syslog daemons again.
6561 * The libudev API gained the new
6562 udev_device_new_from_device_id() call.
6564 * The logic for file system namespace (ReadOnlyDirectory=,
6565 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6566 require pivot_root() anymore. This means fewer temporary
6567 directories are created below /tmp for this feature.
6569 * nspawn containers will now see and receive all submounts
6570 made on the host OS below the root file system of the
6573 * Forward Secure Sealing is now supported for Journal files,
6574 which provide cryptographical sealing of journal files so
6575 that attackers cannot alter log history anymore without this
6576 being detectable. Lennart will soon post a blog story about
6577 this explaining it in more detail.
6579 * There are two new service settings RestartPreventExitStatus=
6580 and SuccessExitStatus= which allow configuration of exit
6581 status (exit code or signal) which will be excepted from the
6582 restart logic, resp. consider successful.
6584 * journalctl gained the new --verify switch that can be used
6585 to check the integrity of the structure of journal files and
6586 (if Forward Secure Sealing is enabled) the contents of
6589 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6590 and similar symlinks pre-created. This makes running shells
6591 as container init process a lot more fun.
6593 * The fstab support can now handle PARTUUID= and PARTLABEL=
6596 * A new ConditionHost= condition has been added to match
6597 against the hostname (with globs) and machine ID. This is
6598 useful for clusters where a single OS image is used to
6599 provision a large number of hosts which shall run slightly
6600 different sets of services.
6602 * Services which hit the restart limit will now be placed in a
6605 Contributions from: Bertram Poettering, Dave Reisner, Huang
6606 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6607 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6611 * When running in --user mode systemd will now become a
6612 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6613 tree a lot more organized.
6615 * A new PartOf= unit dependency type has been introduced that
6616 may be used to group services in a natural way.
6618 * "systemctl enable" may now be used to enable instances of
6621 * journalctl now prints error log levels in red, and
6622 warning/notice log levels in bright white. It also supports
6623 filtering by log level now.
6625 * cgtop gained a new -n switch (similar to top), to configure
6626 the maximum number of iterations to run for. It also gained
6627 -b, to run in batch mode (accepting no input).
6629 * The suffix ".service" may now be omitted on most systemctl
6630 command lines involving service unit names.
6632 * There's a new bus call in logind to lock all sessions, as
6633 well as a loginctl verb for it "lock-sessions".
6635 * libsystemd-logind.so gained a new call sd_journal_perror()
6636 that works similar to libc perror() but logs to the journal
6637 and encodes structured information about the error number.
6639 * /etc/crypttab entries now understand the new keyfile-size=
6642 * shutdown(8) now can send a (configurable) wall message when
6643 a shutdown is cancelled.
6645 * The mount propagation mode for the root file system will now
6646 default to "shared", which is useful to make containers work
6647 nicely out-of-the-box so that they receive new mounts from
6648 the host. This can be undone locally by running "mount
6649 --make-rprivate /" if needed.
6651 * The prefdm.service file has been removed. Distributions
6652 should maintain this unit downstream if they intend to keep
6653 it around. However, we recommend writing normal unit files
6654 for display managers instead.
6656 * Since systemd is a crucial part of the OS we will now
6657 default to a number of compiler switches that improve
6658 security (hardening) such as read-only relocations, stack
6659 protection, and suchlike.
6661 * The TimeoutSec= setting for services is now split into
6662 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6663 of individual time outs for the start and the stop phase of
6666 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6667 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6668 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6669 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6670 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6671 Gundersen, Zbigniew Jędrzejewski-Szmek
6675 * The journal and id128 C APIs are now fully documented as man
6678 * Extra safety checks have been added when transitioning from
6679 the initial RAM disk to the main system to avoid accidental
6682 * /etc/crypttab entries now understand the new keyfile-offset=
6685 * systemctl -t can now be used to filter by unit load state.
6687 * The journal C API gained the new sd_journal_wait() call to
6688 make writing synchronous journal clients easier.
6690 * journalctl gained the new -D switch to show journals from a
6693 * journalctl now displays a special marker between log
6694 messages of two different boots.
6696 * The journal is now explicitly flushed to /var via a service
6697 systemd-journal-flush.service, rather than implicitly simply
6698 by seeing /var/log/journal to be writable.
6700 * journalctl (and the journal C APIs) can now match for much
6701 more complex expressions, with alternatives and
6704 * When transitioning from the initial RAM disk to the main
6705 system we will now kill all processes in a killing spree to
6706 ensure no processes stay around by accident.
6708 * Three new specifiers may be used in unit files: %u, %h, %s
6709 resolve to the user name, user home directory resp. user
6710 shell. This is useful for running systemd user instances.
6712 * We now automatically rotate journal files if their data
6713 object hash table gets a fill level > 75%. We also size the
6714 hash table based on the configured maximum file size. This
6715 together should lower hash collisions drastically and thus
6716 speed things up a bit.
6718 * journalctl gained the new "--header" switch to introspect
6719 header data of journal files.
6721 * A new setting SystemCallFilters= has been added to services
6722 which may be used to apply blacklists or whitelists to
6723 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6725 * nspawn gained a new --link-journal= switch (and quicker: -j)
6726 to link the container journal with the host. This makes it
6727 very easy to centralize log viewing on the host for all
6728 guests while still keeping the journal files separated.
6730 * Many bugfixes and optimizations
6732 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6733 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6734 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6739 * Several tools now understand kernel command line arguments,
6740 which are only read when run in an initial RAM disk. They
6741 usually follow closely their normal counterparts, but are
6744 * There's a new tool to analyze the readahead files that are
6745 automatically generated at boot. Use:
6747 /usr/lib/systemd/systemd-readahead analyze /.readahead
6749 * We now provide an early debug shell on tty9 if this enabled. Use:
6751 systemctl enable debug-shell.service
6753 * All plymouth related units have been moved into the Plymouth
6754 package. Please make sure to upgrade your Plymouth version
6757 * systemd-tmpfiles now supports getting passed the basename of
6758 a configuration file only, in which case it will look for it
6759 in all appropriate directories automatically.
6761 * udevadm info now takes a /dev or /sys path as argument, and
6762 does the right thing. Example:
6764 udevadm info /dev/sda
6765 udevadm info /sys/class/block/sda
6767 * systemctl now prints a warning if a unit is stopped but a
6768 unit that might trigger it continues to run. Example: a
6769 service is stopped but the socket that activates it is left
6772 * "systemctl status" will now mention if the log output was
6773 shortened due to rotation since a service has been started.
6775 * The journal API now exposes functions to determine the
6776 "cutoff" times due to rotation.
6778 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6779 immediately flushing of runtime logs to /var if possible,
6780 resp. for triggering immediate rotation of the journal
6783 * It is now considered an error if a service is attempted to
6784 be stopped that is not loaded.
6786 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6788 * systemd-analyze now supports Python 3
6790 * tmpfiles now supports cleaning up directories via aging
6791 where the first level dirs are always kept around but
6792 directories beneath it automatically aged. This is enabled
6793 by prefixing the age field with '~'.
6795 * Seat objects now expose CanGraphical, CanTTY properties
6796 which is required to deal with very fast bootups where the
6797 display manager might be running before the graphics drivers
6798 completed initialization.
6800 * Seat objects now expose a State property.
6802 * We now include RPM macros for service enabling/disabling
6803 based on the preset logic. We recommend RPM based
6804 distributions to make use of these macros if possible. This
6805 makes it simpler to reuse RPM spec files across
6808 * We now make sure that the collected systemd unit name is
6809 always valid when services log to the journal via
6812 * There's a new man page kernel-command-line(7) detailing all
6813 command line options we understand.
6815 * The fstab generator may now be disabled at boot by passing
6816 fstab=0 on the kernel command line.
6818 * A new kernel command line option modules-load= is now understood
6819 to load a specific kernel module statically, early at boot.
6821 * Unit names specified on the systemctl command line are now
6822 automatically escaped as needed. Also, if file system or
6823 device paths are specified they are automatically turned
6824 into the appropriate mount or device unit names. Example:
6826 systemctl status /home
6827 systemctl status /dev/sda
6829 * The SysVConsole= configuration option has been removed from
6830 system.conf parsing.
6832 * The SysV search path is no longer exported on the D-Bus
6835 * The Names= option has been removed from unit file parsing.
6837 * There's a new man page bootup(7) detailing the boot process.
6839 * Every unit and every generator we ship with systemd now
6840 comes with full documentation. The self-explanatory boot is
6843 * A couple of services gained "systemd-" prefixes in their
6844 name if they wrap systemd code, rather than only external
6845 code. Among them fsck@.service which is now
6846 systemd-fsck@.service.
6848 * The HaveWatchdog property has been removed from the D-Bus
6851 * systemd.confirm_spawn= on the kernel command line should now
6854 * There's a new man page crypttab(5) which details all options
6855 we actually understand.
6857 * systemd-nspawn gained a new --capability= switch to pass
6858 additional capabilities to the container.
6860 * timedated will now read known NTP implementation unit names
6861 from /usr/lib/systemd/ntp-units.d/*.list,
6862 systemd-timedated-ntp.target has been removed.
6864 * journalctl gained a new switch "-b" that lists log data of
6865 the current boot only.
6867 * The notify socket is in the abstract namespace again, in
6868 order to support daemons which chroot() at start-up.
6870 * There is a new Storage= configuration option for journald
6871 which allows configuration of where log data should go. This
6872 also provides a way to disable journal logging entirely, so
6873 that data collected is only forwarded to the console, the
6874 kernel log buffer or another syslog implementation.
6876 * Many bugfixes and optimizations
6878 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6879 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6880 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6881 Shawn Landden, Tom Gundersen
6885 * "systemctl help <unit>" now shows the man page if one is
6888 * Several new man pages have been added.
6890 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6891 MaxLevelConsole= can now be specified in
6892 journald.conf. These options allow reducing the amount of
6893 data stored on disk or forwarded by the log level.
6895 * TimerSlackNSec= can now be specified in system.conf for
6896 PID1. This allows system-wide power savings.
6898 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6899 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6904 * logind is now capable of (optionally) handling power and
6905 sleep keys as well as the lid switch.
6907 * journalctl now understands the syntax "journalctl
6908 /usr/bin/avahi-daemon" to get all log output of a specific
6911 * CapabilityBoundingSet= in system.conf now also influences
6912 the capability bound set of usermode helpers of the kernel.
6914 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6915 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6916 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6917 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6921 * Note that we skipped 139 releases here in order to set the
6922 new version to something that is greater than both udev's
6923 and systemd's most recent version number.
6925 * udev: all udev sources are merged into the systemd source tree now.
6926 All future udev development will happen in the systemd tree. It
6927 is still fully supported to use the udev daemon and tools without
6928 systemd running, like in initramfs or other init systems. Building
6929 udev though, will require the *build* of the systemd tree, but
6930 udev can be properly *run* without systemd.
6932 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6933 should be used to create dead device nodes as workarounds for broken
6936 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6937 no longer supported. udev_monitor_new_from_netlink() needs to be
6938 used to subscribe to events.
6940 * udev: when udevd is started by systemd, processes which are left
6941 behind by forking them off of udev rules, are unconditionally cleaned
6942 up and killed now after the event handling has finished. Services or
6943 daemons must be started as systemd services. Services can be
6944 pulled-in by udev to get started, but they can no longer be directly
6945 forked by udev rules.
6947 * udev: the daemon binary is called systemd-udevd now and installed
6948 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6949 to adapt to that, create symlink, or rename the binary after building
6952 * libudev no longer provides these symbols:
6953 udev_monitor_from_socket()
6954 udev_queue_get_failed_list_entry()
6955 udev_get_{dev,sys,run}_path()
6956 The versions number was bumped and symbol versioning introduced.
6958 * systemd-loginctl and systemd-journalctl have been renamed
6959 to loginctl and journalctl to match systemctl.
6961 * The config files: /etc/systemd/systemd-logind.conf and
6962 /etc/systemd/systemd-journald.conf have been renamed to
6963 logind.conf and journald.conf. Package updates should rename
6964 the files to the new names on upgrade.
6966 * For almost all files the license is now LGPL2.1+, changed
6967 from the previous GPL2.0+. Exceptions are some minor stuff
6968 of udev (which will be changed to LGPL2.1 eventually, too),
6969 and the MIT licensed sd-daemon.[ch] library that is suitable
6970 to be used as drop-in files.
6972 * systemd and logind now handle system sleep states, in
6973 particular suspending and hibernating.
6975 * logind now implements a sleep/shutdown/idle inhibiting logic
6976 suitable for a variety of uses. Soonishly Lennart will blog
6977 about this in more detail.
6979 * var-run.mount and var-lock.mount are no longer provided
6980 (which previously bind mounted these directories to their new
6981 places). Distributions which have not converted these
6982 directories to symlinks should consider stealing these files
6983 from git history and add them downstream.
6985 * We introduced the Documentation= field for units and added
6986 this to all our shipped units. This is useful to make it
6987 easier to explore the boot and the purpose of the various
6990 * All smaller setup units (such as
6991 systemd-vconsole-setup.service) now detect properly if they
6992 are run in a container and are skipped when
6993 appropriate. This guarantees an entirely noise-free boot in
6994 Linux container environments such as systemd-nspawn.
6996 * A framework for implementing offline system updates is now
6997 integrated, for details see:
6998 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7000 * A new service type Type=idle is available now which helps us
7001 avoiding ugly interleaving of getty output and boot status
7004 * There's now a system-wide CapabilityBoundingSet= option to
7005 globally reduce the set of capabilities for the
7006 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7007 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7008 even CAP_NET_ADMIN system-wide for secure systems.
7010 * There are now system-wide DefaultLimitXXX= options to
7011 globally change the defaults of the various resource limits
7012 for all units started by PID 1.
7014 * Harald Hoyer's systemd test suite has been integrated into
7015 systemd which allows easy testing of systemd builds in qemu
7016 and nspawn. (This is really awesome! Ask us for details!)
7018 * The fstab parser is now implemented as generator, not inside
7021 * systemctl will now warn you if .mount units generated from
7022 /etc/fstab are out of date due to changes in fstab that
7023 have not been read by systemd yet.
7025 * systemd is now suitable for usage in initrds. Dracut has
7026 already been updated to make use of this. With this in place
7027 initrds get a slight bit faster but primarily are much
7028 easier to introspect and debug since "systemctl status" in
7029 the host system can be used to introspect initrd services,
7030 and the journal from the initrd is kept around too.
7032 * systemd-delta has been added, a tool to explore differences
7033 between user/admin configuration and vendor defaults.
7035 * PrivateTmp= now affects both /tmp and /var/tmp.
7037 * Boot time status messages are now much prettier and feature
7038 proper english language. Booting up systemd has never been
7041 * Read-ahead pack files now include the inode number of all
7042 files to pre-cache. When the inode changes the pre-caching
7043 is not attempted. This should be nicer to deal with updated
7044 packages which might result in changes of read-ahead
7047 * We now temporaritly lower the kernel's read_ahead_kb variable
7048 when collecting read-ahead data to ensure the kernel's
7049 built-in read-ahead does not add noise to our measurements
7050 of necessary blocks to pre-cache.
7052 * There's now RequiresMountsFor= to add automatic dependencies
7053 for all mounts necessary for a specific file system path.
7055 * MountAuto= and SwapAuto= have been removed from
7056 system.conf. Mounting file systems at boot has to take place
7059 * nspawn now learned a new switch --uuid= to set the machine
7060 ID on the command line.
7062 * nspawn now learned the -b switch to automatically search
7065 * vt102 is now the default TERM for serial TTYs, upgraded from
7068 * systemd-logind now works on VT-less systems.
7070 * The build tree has been reorganized. The individual
7071 components now have directories of their own.
7073 * A new condition type ConditionPathIsReadWrite= is now available.
7075 * nspawn learned the new -C switch to create cgroups for the
7076 container in other hierarchies.
7078 * We now have support for hardware watchdogs, configurable in
7081 * The scheduled shutdown logic now has a public API.
7083 * We now mount /tmp as tmpfs by default, but this can be
7084 masked and /etc/fstab can override it.
7086 * Since udisks does not make use of /media anymore we are not
7087 mounting a tmpfs on it anymore.
7089 * journalctl gained a new --local switch to only interleave
7090 locally generated journal files.
7092 * We can now load the IMA policy at boot automatically.
7094 * The GTK tools have been split off into a systemd-ui.
7096 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7097 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7098 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7099 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7100 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7101 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7102 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7103 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7104 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7109 * This is mostly a bugfix release
7111 * Support optional initialization of the machine ID from the
7112 KVM or container configured UUID.
7114 * Support immediate reboots with "systemctl reboot -ff"
7116 * Show /etc/os-release data in systemd-analyze output
7118 * Many bugfixes for the journal, including endianness fixes and
7119 ensuring that disk space enforcement works
7121 * sd-login.h is C++ compatible again
7123 * Extend the /etc/os-release format on request of the Debian
7126 * We now refuse non-UTF8 strings used in various configuration
7127 and unit files. This is done to ensure we do not pass invalid
7128 data over D-Bus or expose it elsewhere.
7130 * Register Mimo USB Screens as suitable for automatic seat
7133 * Read SELinux client context from journal clients in a race
7136 * Reorder configuration file lookup order. /etc now always
7137 overrides /run in order to allow the administrator to always
7138 and unconditionally override vendor-supplied or
7139 automatically generated data.
7141 * The various user visible bits of the journal now have man
7142 pages. We still lack man pages for the journal API calls
7145 * We now ship all man pages in HTML format again in the
7148 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7149 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7150 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7155 * This is mostly a bugfix release
7157 * systems lacking /etc/os-release are no longer supported.
7159 * Various functionality updates to libsystemd-login.so
7161 * Track class of PAM logins to distinguish greeters from
7164 Contributions from: Kay Sievers, Lennart Poettering, Michael
7169 * This is an important bugfix release for v41.
7171 * Building man pages is now optional which should be useful
7172 for those building systemd from git but unwilling to install
7175 * Watchdog support for supervising services is now usable. In
7176 a future release support for hardware watchdogs
7177 (i.e. /dev/watchdog) will be added building on this.
7179 * Service start rate limiting is now configurable and can be
7180 turned off per service. When a start rate limit is hit a
7181 reboot can automatically be triggered.
7183 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7185 Contributions from: Benjamin Franzke, Bill Nottingham,
7186 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7187 Schmidt, Michał Górny, Piotr Drąg
7191 * The systemd binary is installed /usr/lib/systemd/systemd now;
7192 An existing /sbin/init symlink needs to be adapted with the
7195 * The code that loads kernel modules has been ported to invoke
7196 libkmod directly, instead of modprobe. This means we do not
7197 support systems with module-init-tools anymore.
7199 * Watchdog support is now already useful, but still not
7202 * A new kernel command line option systemd.setenv= is
7203 understood to set system wide environment variables
7204 dynamically at boot.
7206 * We now limit the set of capabilities of systemd-journald.
7208 * We now set SIGPIPE to ignore by default, since it only is
7209 useful in shell pipelines, and has little use in general
7210 code. This can be disabled with IgnoreSIPIPE=no in unit
7213 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7214 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7219 * This is mostly a bugfix release
7221 * We now expose the reason why a service failed in the
7222 "Result" D-Bus property.
7224 * Rudimentary service watchdog support (will be completed over
7225 the next few releases.)
7227 * When systemd forks off in order execute some service we will
7228 now immediately changes its argv[0] to reflect which process
7229 it will execute. This is useful to minimize the time window
7230 with a generic argv[0], which makes bootcharts more useful
7232 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7233 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7234 Mike Kazantsev, Ray Strode
7238 * This is mostly a test release, but incorporates many
7241 * New systemd-cgtop tool to show control groups by their
7244 * Linking against libacl for ACLs is optional again. If
7245 disabled, support tracking device access for active logins
7246 goes becomes unavailable, and so does access to the user
7247 journals by the respective users.
7249 * If a group "adm" exists, journal files are automatically
7250 owned by them, thus allow members of this group full access
7251 to the system journal as well as all user journals.
7253 * The journal now stores the SELinux context of the logging
7254 client for all entries.
7256 * Add C++ inclusion guards to all public headers
7258 * New output mode "cat" in the journal to print only text
7259 messages, without any meta data like date or time.
7261 * Include tiny X server wrapper as a temporary stop-gap to
7262 teach XOrg udev display enumeration. This is used by display
7263 managers such as gdm, and will go away as soon as XOrg
7264 learned native udev hotplugging for display devices.
7266 * Add new systemd-cat tool for executing arbitrary programs
7267 with STDERR/STDOUT connected to the journal. Can also act as
7268 BSD logger replacement, and does so by default.
7270 * Optionally store all locally generated coredumps in the
7271 journal along with meta data.
7273 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7274 writing short strings to files (for usage for /sys), and for
7275 creating symlinks, character and block device nodes.
7277 * New unit file option ControlGroupPersistent= to make cgroups
7278 persistent, following the mechanisms outlined in
7279 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7281 * Support multiple local RTCs in a sane way
7283 * No longer monopolize IO when replaying readahead data on
7284 rotating disks, since we might starve non-file-system IO to
7285 death, since fanotify() will not see accesses done by blkid,
7288 * Do not show kernel threads in systemd-cgls anymore, unless
7289 requested with new -k switch.
7291 Contributions from: Dan Horák, Kay Sievers, Lennart
7292 Poettering, Michal Schmidt
7296 * This is mostly a test release, but incorporates many
7299 * The git repository moved to:
7300 git://anongit.freedesktop.org/systemd/systemd
7301 ssh://git.freedesktop.org/git/systemd/systemd
7303 * First release with the journal
7304 http://0pointer.de/blog/projects/the-journal.html
7306 * The journal replaces both systemd-kmsg-syslogd and
7307 systemd-stdout-bridge.
7309 * New sd_pid_get_unit() API call in libsystemd-logind
7311 * Many systemadm clean-ups
7313 * Introduce remote-fs-pre.target which is ordered before all
7314 remote mounts and may be used to start services before all
7317 * Added Mageia support
7319 * Add bash completion for systemd-loginctl
7321 * Actively monitor PID file creation for daemons which exit in
7322 the parent process before having finished writing the PID
7323 file in the daemon process. Daemons which do this need to be
7324 fixed (i.e. PID file creation must have finished before the
7325 parent exits), but we now react a bit more gracefully to them.
7327 * Add colourful boot output, mimicking the well-known output
7328 of existing distributions.
7330 * New option PassCredentials= for socket units, for
7331 compatibility with a recent kernel ABI breakage.
7333 * /etc/rc.local is now hooked in via a generator binary, and
7334 thus will no longer act as synchronization point during
7337 * systemctl list-unit-files now supports --root=.
7339 * systemd-tmpfiles now understands two new commands: z, Z for
7340 relabelling files according to the SELinux database. This is
7341 useful to apply SELinux labels to specific files in /sys,
7344 * Output of SysV services is now forwarded to both the console
7345 and the journal by default, not only just the console.
7347 * New man pages for all APIs from libsystemd-login.
7349 * The build tree got reorganized and the build system is a
7350 lot more modular allowing embedded setups to specifically
7351 select the components of systemd they are interested in.
7353 * Support for Linux systems lacking the kernel VT subsystem is
7356 * configure's --with-rootdir= got renamed to
7357 --with-rootprefix= to follow the naming used by udev and
7360 * Unless specified otherwise we will now install to /usr instead
7361 of /usr/local by default.
7363 * Processes with '@' in argv[0][0] are now excluded from the
7364 final shut-down killing spree, following the logic explained
7366 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7368 * All processes remaining in a service cgroup when we enter
7369 the START or START_PRE states are now killed with
7370 SIGKILL. That means it is no longer possible to spawn
7371 background processes from ExecStart= lines (which was never
7372 supported anyway, and bad style).
7374 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7375 reloading of units together.
7377 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7378 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7379 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7380 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7381 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob