1 systemd System and Service Manager
5 * Meson is now supported as build system in addition to Automake. It is
6 our plan to remove Automake in one of our next releases, so that
7 Meson becomes our exclusive build system. Hence, please start using
8 the Meson build system in your downstream packaging. There's plenty
9 of documentation around how to use Meson, the extremely brief
12 ./autogen.sh && ./configure && make && sudo make install
16 meson build && ninja -C build && sudo ninja -C build install
18 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
19 which permits configuring a timeout on the time a job is
20 running. This is particularly useful for setting timeouts on jobs for
23 * Unit files gained two new options ConditionUser= and ConditionGroup=
24 for conditionalizing units based on the identity of the user/group
25 running a systemd user instance.
27 * systemd-networkd now understands a new FlowLabel= setting in the
28 [VXLAN] section of .network files, as well as a Priority= in
29 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
30 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
31 gained support for configuration of GENEVE links, and IPv6 address
32 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
34 * .link files now understand a new Port= setting.
36 * systemd-networkd's DHCP support gained support for DHCP option 119
39 * systemd-networkd gained support for serving IPv6 address ranges using
40 the Router Advertisment protocol. The new .network configuration
41 section [IPv6Prefix] may be used to configure the ranges to
42 serve. This is implemented based on a new, minimal, native server
45 * journalctl's --output= switch gained support for a new parameter
46 "short-iso-precise" for a mode where timestamps are shown as precise
49 * systemd-udevd's "net_id" builtin may now generate stable network
50 interface names from IBM PowerVM VIO devices as well as ACPI platform
53 * MulticastDNS support in systemd-resolved may now be explicitly
54 enabled/disabled using the new MulticastDNS= configuration file
57 * systemd-resolved may now optionally use libidn2 instead of the libidn
58 for processing internationalized domain names. Support for libidn2
59 should be considered experimental and should not be enabled by
62 * "machinectl pull-tar" and related call may now do verification of
63 downloaded images using SUSE-style .sha256 checksum files in addition
64 to the already existing support for validating using Ubuntu-style
67 * sd-bus gained support for a new sd_bus_message_appendv() call which
68 is va_list equivalent of sd_bus_message_append().
70 * sd-boot gained support for validating images using SHIM/MOK.
72 * The SMACK code learnt support for "onlycap".
74 * systemd-mount --umount is now much smarter in figuring out how to
75 properly unmount a device given its mount or device path.
77 * The code to call libnss_dns as a fallback from libnss_resolve when
78 the communication with systemd-resolved fails was removed. This
79 fallback was redundant and interfered with the [!UNAVAIL=return]
80 suffix. See nss-resolve(8) for the recommended configuration.
82 * systemd-logind may now be restarted without losing state. It stores
83 the file descriptors for devices it manages in the system manager
84 using the FDSTORE= mechanism. Please note that further changes in
85 other components may be required to make use of this (for example
86 Xorg has code to listen for stops of systemd-logind and terminate
87 itself when logind is stopped or restarted, in order to avoid using
88 stale file descriptors for graphical devices, which is now
89 counterproductive and must be reverted in order for restarts of
90 systemd-logind to be safe. See
91 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
93 * All kernel install plugins are called with the environment variable
94 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
95 /etc/machine-id. If the file is missing or empty, the variable is
96 empty and BOOT_DIR_ABS is the path of a temporary directory which is
97 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
98 is empty, all plugins should not put anything in BOOT_DIR_ABS.
100 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
101 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
102 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
103 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
104 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
105 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
106 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
107 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
108 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
109 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
110 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
111 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
112 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
113 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
114 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
115 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
116 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
117 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
118 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
119 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
120 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
121 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
122 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
123 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
124 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
125 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
126 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
127 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
128 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
135 * This version requires at least gperf 3.1 for building, 3.0 is not
138 * The "hybrid" control group mode has been modified to improve
139 compatibility with "legacy" cgroups-v1 setups. Specifically, the
140 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
141 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
142 cgroups-v1 hierarchy), the only externally visible change being that
143 the cgroups-v2 hierarchy is also mounted, to
144 /sys/fs/cgroup/unified. This should provide a large degree of
145 compatibility with "legacy" cgroups-v1, while taking benefit of the
146 better management capabilities of cgroups-v2.
148 * The default control group setup mode may be selected both a boot-time
149 via a set of kernel command line parameters (specifically:
150 systemd.unified_cgroup_hierarchy= and
151 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
152 default selected on the configure command line
153 (--with-default-hierarchy=). The upstream default is "hybrid"
154 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
155 this will change in a future systemd version to be "unified" (pure
156 cgroups-v2 mode). The third option for the compile time option is
157 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
158 distributions to default to "hybrid" mode for release distributions,
159 starting with v233. We recommend "unified" for development
160 distributions (specifically: distributions such as Fedora's rawhide)
161 as that's where things are headed in the long run. Use "legacy" for
162 greatest stability and compatibility only.
164 * Note one current limitation of "unified" and "hybrid" control group
165 setup modes: the kernel currently does not permit the systemd --user
166 instance (i.e. unprivileged code) to migrate processes between two
167 disconnected cgroup subtrees, even if both are managed and owned by
168 the user. This effectively means "systemd-run --user --scope" doesn't
169 work when invoked from outside of any "systemd --user" service or
170 scope. Specifically, it is not supported from session scopes. We are
171 working on fixing this in a future systemd version. (See #3388 for
172 further details about this.)
174 * DBus policy files are now installed into /usr rather than /etc. Make
175 sure your system has dbus >= 1.9.18 running before upgrading to this
176 version, or override the install path with --with-dbuspolicydir= .
178 * All python scripts shipped with systemd (specifically: the various
179 tests written in Python) now require Python 3.
181 * systemd unit tests can now run standalone (without the source or
182 build directories), and can be installed into /usr/lib/systemd/tests/
183 with 'make install-tests'.
185 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
186 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
189 * Support for the %c, %r, %R specifiers in unit files has been
190 removed. Specifiers are not supposed to be dependent on configuration
191 in the unit file itself (so that they resolve the same regardless
192 where used in the unit files), but these specifiers were influenced
193 by the Slice= option.
195 * The shell invoked by debug-shell.service now defaults to /bin/sh in
196 all cases. If distributions want to use a different shell for this
197 purpose (for example Fedora's /sbin/sushell) they need to specify
198 this explicitly at configure time using --with-debug-shell=.
200 * The confirmation spawn prompt has been reworked to offer the
203 (c)ontinue, proceed without asking anymore
204 (D)ump, show the state of the unit
205 (f)ail, don't execute the command and pretend it failed
207 (i)nfo, show a short summary of the unit
208 (j)obs, show jobs that are in progress
209 (s)kip, don't execute the command and pretend it succeeded
210 (y)es, execute the command
212 The 'n' choice for the confirmation spawn prompt has been removed,
213 because its meaning was confusing.
215 The prompt may now also be redirected to an alternative console by
216 specifying the console as parameter to systemd.confirm_spawn=.
218 * Services of Type=notify require a READY=1 notification to be sent
219 during startup. If no such message is sent, the service now fails,
220 even if the main process exited with a successful exit code.
222 * Services that fail to start up correctly now always have their
223 ExecStopPost= commands executed. Previously, they'd enter "failed"
224 state directly, without executing these commands.
226 * The option MulticastDNS= of network configuration files has acquired
227 an actual implementation. With MulticastDNS=yes a host can resolve
228 names of remote hosts and reply to mDNS A and AAAA requests.
230 * When units are about to be started an additional check is now done to
231 ensure that all dependencies of type BindsTo= (when used in
232 combination with After=) have been started.
234 * systemd-analyze gained a new verb "syscall-filter" which shows which
235 system call groups are defined for the SystemCallFilter= unit file
236 setting, and which system calls they contain.
238 * A new system call filter group "@filesystem" has been added,
239 consisting of various file system related system calls. Group
240 "@reboot" has been added, covering reboot, kexec and shutdown related
241 calls. Finally, group "@swap" has been added covering swap
242 configuration related calls.
244 * A new unit file option RestrictNamespaces= has been added that may be
245 used to restrict access to the various process namespace types the
246 Linux kernel provides. Specifically, it may be used to take away the
247 right for a service unit to create additional file system, network,
248 user, and other namespaces. This sandboxing option is particularly
249 relevant due to the high amount of recently discovered namespacing
250 related vulnerabilities in the kernel.
252 * systemd-udev's .link files gained support for a new AutoNegotiation=
253 setting for configuring Ethernet auto-negotiation.
255 * systemd-networkd's .network files gained support for a new
256 ListenPort= setting in the [DHCP] section to explicitly configure the
257 UDP client port the DHCP client shall listen on.
259 * .network files gained a new Unmanaged= boolean setting for explicitly
260 excluding one or more interfaces from management by systemd-networkd.
262 * The systemd-networkd ProxyARP= option has been renamed to
263 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
264 renamed to ReduceARPProxy=. The old names continue to be available
267 * systemd-networkd gained support for configuring IPv6 Proxy NDP
268 addresses via the new IPv6ProxyNDPAddress= .network file setting.
270 * systemd-networkd's bonding device support gained support for two new
271 configuration options ActiveSlave= and PrimarySlave=.
273 * The various options in the [Match] section of .network files gained
274 support for negative matching.
276 * New systemd-specific mount options are now understood in /etc/fstab:
278 x-systemd.mount-timeout= may be used to configure the maximum
279 permitted runtime of the mount command.
281 x-systemd.device-bound may be set to bind a mount point to its
282 backing device unit, in order to automatically remove a mount point
283 if its backing device is unplugged. This option may also be
284 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
285 on the block device, which is now automatically set for all CDROM
286 drives, so that mounted CDs are automatically unmounted when they are
287 removed from the drive.
289 x-systemd.after= and x-systemd.before= may be used to explicitly
290 order a mount after or before another unit or mount point.
292 * Enqueued start jobs for device units are now automatically garbage
293 collected if there are no jobs waiting for them anymore.
295 * systemctl list-jobs gained two new switches: with --after, for every
296 queued job the jobs it's waiting for are shown; with --before the
297 jobs which it's blocking are shown.
299 * systemd-nspawn gained support for ephemeral boots from disk images
300 (or in other words: --ephemeral and --image= may now be
301 combined). Moreover, ephemeral boots are now supported for normal
302 directories, even if the backing file system is not btrfs. Of course,
303 if the file system does not support file system snapshots or
304 reflinks, the initial copy operation will be relatively expensive, but
305 this should still be suitable for many use cases.
307 * Calendar time specifications in .timer units now support
308 specifications relative to the end of a month by using "~" instead of
309 "-" as separator between month and day. For example, "*-02~03" means
310 "the third last day in February". In addition a new syntax for
311 repeated events has been added using the "/" character. For example,
312 "9..17/2:00" means "every two hours from 9am to 5pm".
314 * systemd-socket-proxyd gained a new parameter --connections-max= for
315 configuring the maximum number of concurrent connections.
317 * sd-id128 gained a new API for generating unique IDs for the host in a
318 way that does not leak the machine ID. Specifically,
319 sd_id128_get_machine_app_specific() derives an ID based on the
320 machine ID a in well-defined, non-reversible, stable way. This is
321 useful whenever an identifier for the host is needed but where the
322 identifier shall not be useful to identify the system beyond the
323 scope of the application itself. (Internally this uses HMAC-SHA256 as
324 keyed hash function using the machine ID as input.)
326 * NotifyAccess= gained a new supported value "exec". When set
327 notifications are accepted from all processes systemd itself invoked,
328 including all control processes.
330 * .nspawn files gained support for defining overlay mounts using the
331 Overlay= and OverlayReadOnly= options. Previously this functionality
332 was only available on the systemd-nspawn command line.
334 * systemd-nspawn's --bind= and --overlay= options gained support for
335 bind/overlay mounts whose source lies within the container tree by
336 prefixing the source path with "+".
338 * systemd-nspawn's --bind= and --overlay= options gained support for
339 automatically allocating a temporary source directory in /var/tmp
340 that is removed when the container dies. Specifically, if the source
341 directory is specified as empty string this mechanism is selected. An
342 example usage is --overlay=+/var::/var, which creates an overlay
343 mount based on the original /var contained in the image, overlayed
344 with a temporary directory in the host's /var/tmp. This way changes
345 to /var are automatically flushed when the container shuts down.
347 * systemd-nspawn --image= option does now permit raw file system block
348 devices (in addition to images containing partition tables, as
351 * The disk image dissection logic in systemd-nspawn gained support for
352 automatically setting up LUKS encrypted as well as Verity protected
353 partitions. When a container is booted from an encrypted image the
354 passphrase is queried at start-up time. When a container with Verity
355 data is started, the root hash is search in a ".roothash" file
356 accompanying the disk image (alternatively, pass the root hash via
357 the new --root-hash= command line option).
359 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
360 be used to dissect disk images the same way as systemd-nspawn does
361 it, following the Bootable Partition Specification. It may even be
362 used to mount disk images with complex partition setups (including
363 LUKS and Verity partitions) to a local host directory, in order to
364 inspect them. This tool is not considered public API (yet), and is
365 thus not installed into /usr/bin. Please do not rely on its
366 existence, since it might go away or be changed in later systemd
369 * A new generator "systemd-verity-generator" has been added, similar in
370 style to "systemd-cryptsetup-generator", permitting automatic setup of
371 Verity root partitions when systemd boots up. In order to make use of
372 this your partition setup should follow the Discoverable Partitions
373 Specification, and the GPT partition ID of the root file system
374 partition should be identical to the upper 128bit of the Verity root
375 hash. The GPT partition ID of the Verity partition protecting it
376 should be the lower 128bit of the Verity root hash. If the partition
377 image follows this model it is sufficient to specify a single
378 "roothash=" kernel command line argument to both configure which root
379 image and verity partition to use as well as the root hash for
380 it. Note that systemd-nspawn's Verity support follows the same
381 semantics, meaning that disk images with proper Verity data in place
382 may be booted in containers with systemd-nspawn as well as on
383 physical systems via the verity generator. Also note that the "mkosi"
384 tool available at https://github.com/systemd/mkosi has been updated
385 to generate Verity protected disk images following this scheme. In
386 fact, it has been updated to generate disk images that optionally
387 implement a complete UEFI SecureBoot trust chain, involving a signed
388 kernel and initrd image that incorporates such a root hash as well as
389 a Verity-enabled root partition.
391 * The hardware database (hwdb) udev supports has been updated to carry
392 accelerometer quirks.
394 * All system services are now run with a fresh kernel keyring set up
395 for them. The invocation ID is stored by default in it, thus
396 providing a safe, non-overridable way to determine the invocation
399 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
400 options for bind mounting arbitrary paths in a service-specific
401 way. When these options are used, arbitrary host or service files and
402 directories may be mounted to arbitrary locations in the service's
405 * Documentation has been added that lists all of systemd's low-level
406 environment variables:
408 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
410 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
411 whether a specific socket file descriptor matches a specified socket
414 * systemd-firstboot has been updated to check for the
415 systemd.firstboot= kernel command line option. It accepts a boolean
416 and when set to false the first boot questions are skipped.
418 * systemd-fstab-generator has been updated to check for the
419 systemd.volatile= kernel command line option, which either takes an
420 optional boolean parameter or the special value "state". If used the
421 system may be booted in a "volatile" boot mode. Specifically,
422 "systemd.volatile" is used, the root directory will be mounted as
423 tmpfs, and only /usr is mounted from the actual root file system. If
424 "systemd.volatile=state" is used, the root directory will be mounted
425 as usual, but /var is mounted as tmpfs. This concept provides similar
426 functionality as systemd-nspawn's --volatile= option, but provides it
427 on physical boots. Use this option for implementing stateless
428 systems, or testing systems with all state and/or configuration reset
429 to the defaults. (Note though that many distributions are not
430 prepared to boot up without a populated /etc or /var, though.)
432 * systemd-gpt-auto-generator gained support for LUKS encrypted root
433 partitions. Previously it only supported LUKS encrypted partitions
434 for all other uses, except for the root partition itself.
436 * Socket units gained support for listening on AF_VSOCK sockets for
437 communication in virtualized QEMU environments.
439 * The "configure" script gained a new option --with-fallback-hostname=
440 for specifying the fallback hostname to use if none is configured in
441 /etc/hostname. For example, by specifying
442 --with-fallback-hostname=fedora it is possible to default to a
443 hostname of "fedora" on pristine installations.
445 * systemd-cgls gained support for a new --unit= switch for listing only
446 the control groups of a specific unit. Similar --user-unit= has been
447 added for listing only the control groups of a specific user unit.
449 * systemd-mount gained a new --umount switch for unmounting a mount or
450 automount point (and all mount/automount points below it).
452 * systemd will now refuse full configuration reloads (via systemctl
453 daemon-reload and related calls) unless at least 16MiB of free space
454 are available in /run. This is a safety precaution in order to ensure
455 that generators can safely operate after the reload completed.
457 * A new unit file option RootImage= has been added, which has a similar
458 effect as RootDirectory= but mounts the service's root directory from
459 a disk image instead of plain directory. This logic reuses the same
460 image dissection and mount logic that systemd-nspawn already uses,
461 and hence supports any disk images systemd-nspawn supports, including
462 those following the Discoverable Partition Specification, as well as
463 Verity enabled images. This option enables systemd to run system
464 services directly off disk images acting as resource bundles,
465 possibly even including full integrity data.
467 * A new MountAPIVFS= unit file option has been added, taking a boolean
468 argument. If enabled /proc, /sys and /dev (collectively called the
469 "API VFS") will be mounted for the service. This is only relevant if
470 RootDirectory= or RootImage= is used for the service, as these mounts
471 are of course in place in the host mount namespace anyway.
473 * systemd-nspawn gained support for a new --pivot-root= switch. If
474 specified the root directory within the container image is pivoted to
475 the specified mount point, while the original root disk is moved to a
476 different place. This option enables booting of ostree images
477 directly with systemd-nspawn.
479 * The systemd build scripts will no longer complain if the NTP server
480 addresses are not changed from the defaults. Google now supports
481 these NTP servers officially. We still recommend downstreams to
482 properly register an NTP pool with the NTP pool project though.
484 * coredumpctl gained a new "--reverse" option for printing the list
485 of coredumps in reverse order.
487 * coredumpctl will now show additional information about truncated and
488 inaccessible coredumps, as well as coredumps that are still being
489 processed. It also gained a new --quiet switch for suppressing
490 additional informational message in its output.
492 * coredumpctl gained support for only showing coredumps newer and/or
493 older than specific timestamps, using the new --since= and --until=
494 options, reminiscent of journalctl's options by the same name.
496 * The systemd-coredump logic has been improved so that it may be reused
497 to collect backtraces in non-compiled languages, for example in
498 scripting languages such as Python.
500 * machinectl will now show the UID shift of local containers, if user
501 namespacing is enabled for them.
503 * systemd will now optionally run "environment generator" binaries at
504 configuration load time. They may be used to add environment
505 variables to the environment block passed to services invoked. One
506 user environment generator is shipped by default that sets up
507 environment variables based on files dropped into /etc/environment.d
508 and ~/.config/environment.d/.
510 * systemd-resolved now includes the new, recently published 2017 DNSSEC
513 * hostnamed has been updated to report a new chassis type of
514 "convertible" to cover "foldable" laptops that can both act as a
515 tablet and as a laptop, such as various Lenovo Yoga devices.
517 Contributions from: Adrián López, Alexander Galanin, Alexander
518 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
519 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
520 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
521 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
522 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
523 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
524 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
525 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
526 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
527 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
528 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
529 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
530 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
531 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
532 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
533 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
534 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
535 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
536 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
537 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
538 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
539 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
540 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
541 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
542 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
543 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
544 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
551 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
552 RestrictAddressFamilies= enabled. These sandboxing options should
553 generally be compatible with the various external udev call-out
554 binaries we are aware of, however there may be exceptions, in
555 particular when exotic languages for these call-outs are used. In
556 this case, consider turning off these settings locally.
558 * The new RemoveIPC= option can be used to remove IPC objects owned by
559 the user or group of a service when that service exits.
561 * The new ProtectKernelModules= option can be used to disable explicit
562 load and unload operations of kernel modules by a service. In
563 addition access to /usr/lib/modules is removed if this option is set.
565 * ProtectSystem= option gained a new value "strict", which causes the
566 whole file system tree with the exception of /dev, /proc, and /sys,
567 to be remounted read-only for a service.
569 * The new ProtectKernelTunables= option can be used to disable
570 modification of configuration files in /sys and /proc by a service.
571 Various directories and files are remounted read-only, so access is
572 restricted even if the file permissions would allow it.
574 * The new ProtectControlGroups= option can be used to disable write
575 access by a service to /sys/fs/cgroup.
577 * Various systemd services have been hardened with
578 ProtectKernelTunables=yes, ProtectControlGroups=yes,
579 RestrictAddressFamilies=.
581 * Support for dynamically creating users for the lifetime of a service
582 has been added. If DynamicUser=yes is specified, user and group IDs
583 will be allocated from the range 61184..65519 for the lifetime of the
584 service. They can be resolved using the new nss-systemd.so NSS
585 module. The module must be enabled in /etc/nsswitch.conf. Services
586 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
587 any resources allocated by the service will be cleaned up when the
588 service exits. They also have ProtectHome=read-only and
589 ProtectSystem=strict enabled, so they are not able to make any
590 permanent modifications to the system.
592 * The nss-systemd module also always resolves root and nobody, making
593 it possible to have no /etc/passwd or /etc/group files in minimal
594 container or chroot environments.
596 * Services may be started with their own user namespace using the new
597 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
598 under which the service is running are mapped. All other users are
601 * Support for the cgroup namespace has been added to systemd-nspawn. If
602 supported by kernel, the container system started by systemd-nspawn
603 will have its own view of the cgroup hierarchy. This new behaviour
604 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
606 * The new MemorySwapMax= option can be used to limit the maximum swap
607 usage under the unified cgroup hierarchy.
609 * Support for the CPU controller in the unified cgroup hierarchy has
610 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
611 options. This controller requires out-of-tree patches for the kernel
612 and the support is provisional.
614 * Mount and automount units may now be created transiently
615 (i.e. dynamically at runtime via the bus API, instead of requiring
616 unit files in the file system).
618 * systemd-mount is a new tool which may mount file systems – much like
619 mount(8), optionally pulling in additional dependencies through
620 transient .mount and .automount units. For example, this tool
621 automatically runs fsck on a backing block device before mounting,
622 and allows the automount logic to be used dynamically from the
623 command line for establishing mount points. This tool is particularly
624 useful when dealing with removable media, as it will ensure fsck is
625 run – if necessary – before the first access and that the file system
626 is quickly unmounted after each access by utilizing the automount
627 logic. This maximizes the chance that the file system on the
628 removable media stays in a clean state, and if it isn't in a clean
629 state is fixed automatically.
631 * LazyUnmount=yes option for mount units has been added to expose the
632 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
635 * /efi will be used as the mount point of the EFI boot partition, if
636 the directory is present, and the mount point was not configured
637 through other means (e.g. fstab). If /efi directory does not exist,
638 /boot will be used as before. This makes it easier to automatically
639 mount the EFI partition on systems where /boot is used for something
642 * When operating on GPT disk images for containers, systemd-nspawn will
643 now mount the ESP to /boot or /efi according to the same rules as PID
644 1 running on a host. This allows tools like "bootctl" to operate
645 correctly within such containers, in order to make container images
646 bootable on physical systems.
648 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
650 * Two new user session targets have been added to support running
651 graphical sessions under the systemd --user instance:
652 graphical-session.target and graphical-session-pre.target. See
653 systemd.special(7) for a description of how those targets should be
656 * The vconsole initialization code has been significantly reworked to
657 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
658 support unicode keymaps. Font and keymap configuration will now be
659 copied to all allocated virtual consoles.
661 * FreeBSD's bhyve virtualization is now detected.
663 * Information recorded in the journal for core dumps now includes the
664 contents of /proc/mountinfo and the command line of the process at
665 the top of the process hierarchy (which is usually the init process
668 * systemd-journal-gatewayd learned the --directory= option to serve
669 files from the specified location.
671 * journalctl --root=… can be used to peruse the journal in the
672 /var/log/ directories inside of a container tree. This is similar to
673 the existing --machine= option, but does not require the container to
676 * The hardware database has been extended to support
677 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
680 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
681 specify the click rate for mice which include a horizontal wheel with
682 a click rate that is different than the one for the vertical wheel.
684 * systemd-run gained a new --wait option that makes service execution
685 synchronous. (Specifically, the command will not return until the
686 specified service binary exited.)
688 * systemctl gained a new --wait option that causes the start command to
689 wait until the units being started have terminated again.
691 * A new journal output mode "short-full" has been added which displays
692 timestamps with abbreviated English day names and adds a timezone
693 suffix. Those timestamps include more information than the default
694 "short" output mode, and can be passed directly to journalctl's
695 --since= and --until= options.
697 * /etc/resolv.conf will be bind-mounted into containers started by
698 systemd-nspawn, if possible, so any changes to resolv.conf contents
699 are automatically propagated to the container.
701 * The number of instances for socket-activated services originating
702 from a single IP address can be limited with
703 MaxConnectionsPerSource=, extending the existing setting of
706 * systemd-networkd gained support for vcan ("Virtual CAN") interface
709 * .netdev and .network configuration can now be extended through
712 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
713 Segmentation Offload, Generic Receive Offload, Large Receive Offload
714 can be enabled and disabled using the new UDPSegmentationOffload=,
715 TCPSegmentationOffload=, GenericSegmentationOffload=,
716 GenericReceiveOffload=, LargeReceiveOffload= options in the
717 [Link] section of .link files.
719 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
720 Port VLAN ID can be configured for bridge devices using the new STP=,
721 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
722 section of .netdev files.
724 * The route table to which routes received over DHCP or RA should be
725 added can be configured with the new RouteTable= option in the [DHCP]
726 and [IPv6AcceptRA] sections of .network files.
728 * The Address Resolution Protocol can be disabled on links managed by
729 systemd-networkd using the ARP=no setting in the [Link] section of
732 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
733 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
734 encode information about the result and exit codes of the current
735 service runtime cycle.
737 * systemd-sysctl will now configure kernel parameters in the order
738 they occur in the configuration files. This matches what sysctl
739 has been traditionally doing.
741 * kernel-install "plugins" that are executed to perform various
742 tasks after a new kernel is added and before an old one is removed
743 can now return a special value to terminate the procedure and
744 prevent any later plugins from running.
746 * Journald's SplitMode=login setting has been deprecated. It has been
747 removed from documentation, and its use is discouraged. In a future
748 release it will be completely removed, and made equivalent to current
749 default of SplitMode=uid.
751 * Storage=both option setting in /etc/systemd/coredump.conf has been
752 removed. With fast LZ4 compression storing the core dump twice is not
755 * The --share-system systemd-nspawn option has been replaced with an
756 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
757 this functionality is discouraged. In addition the variables
758 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
759 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
760 individual namespaces.
762 * "machinectl list" now shows the IP address of running containers in
763 the output, as well as OS release information.
765 * "loginctl list" now shows the TTY of each session in the output.
767 * sd-bus gained new API calls sd_bus_track_set_recursive(),
768 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
769 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
770 tracking objects in a "recursive" mode, where a single client can be
771 counted multiple times, if it takes multiple references.
773 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
774 sd_bus_get_exit_on_disconnect(). They may be used to to make a
775 process using sd-bus automatically exit if the bus connection is
778 * Bus clients of the service manager may now "pin" loaded units into
779 memory, by taking an explicit reference on them. This is useful to
780 ensure the client can retrieve runtime data about the service even
781 after the service completed execution. Taking such a reference is
782 available only for privileged clients and should be helpful to watch
783 running services in a race-free manner, and in particular collect
784 information about exit statuses and results.
786 * The nss-resolve module has been changed to strictly return UNAVAIL
787 when communication via D-Bus with resolved failed, and NOTFOUND when
788 a lookup completed but was negative. This means it is now possible to
789 neatly configure fallbacks using nsswitch.conf result checking
790 expressions. Taking benefit of this, the new recommended
791 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
793 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
795 * A new setting CtrlAltDelBurstAction= has been added to
796 /etc/systemd/system.conf which may be used to configure the precise
797 behaviour if the user on the console presses Ctrl-Alt-Del more often
798 than 7 times in 2s. Previously this would unconditionally result in
799 an expedited, immediate reboot. With this new setting the precise
800 operation may be configured in more detail, and also turned off
803 * In .netdev files two new settings RemoteChecksumTx= and
804 RemoteChecksumRx= are now understood that permit configuring the
805 remote checksumming logic for VXLAN networks.
807 * The service manager learnt a new "invocation ID" concept for invoked
808 services. Each runtime cycle of a service will get a new invocation
809 ID (a 128bit random UUID) assigned that identifies the current
810 run of the service uniquely and globally. A new invocation ID
811 is generated each time a service starts up. The journal will store
812 the invocation ID of a service along with any logged messages, thus
813 making the invocation ID useful for matching the online runtime of a
814 service with the offline log data it generated in a safe way without
815 relying on synchronized timestamps. In many ways this new service
816 invocation ID concept is similar to the kernel's boot ID concept that
817 uniquely and globally identifies the runtime of each boot. The
818 invocation ID of a service is passed to the service itself via an
819 environment variable ($INVOCATION_ID). A new bus call
820 GetUnitByInvocationID() has been added that is similar to GetUnit()
821 but instead of retrieving the bus path for a unit by its name
822 retrieves it by its invocation ID. The returned path is valid only as
823 long as the passed invocation ID is current.
825 * systemd-resolved gained a new "DNSStubListener" setting in
826 resolved.conf. It either takes a boolean value or the special values
827 "udp" and "tcp", and configures whether to enable the stub DNS
828 listener on 127.0.0.53:53.
830 * IP addresses configured via networkd may now carry additional
831 configuration settings supported by the kernel. New options include:
832 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
833 PrefixRoute=, AutoJoin=.
835 * The PAM configuration fragment file for "user@.service" shipped with
836 systemd (i.e. the --user instance of systemd) has been stripped to
837 the minimum necessary to make the system boot. Previously, it
838 contained Fedora-specific stanzas that did not apply to other
839 distributions. It is expected that downstream distributions add
840 additional configuration lines, matching their needs to this file,
841 using it only as rough template of what systemd itself needs. Note
842 that this reduced fragment does not even include an invocation of
843 pam_limits which most distributions probably want to add, even though
844 systemd itself does not need it. (There's also the new build time
845 option --with-pamconfdir=no to disable installation of the PAM
848 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
849 capability is now also dropped from its set (in addition to
850 CAP_SYS_MKNOD as before).
852 * In service unit files it is now possible to connect a specific named
853 file descriptor with stdin/stdout/stdout of an executed service. The
854 name may be specified in matching .socket units using the
855 FileDescriptorName= setting.
857 * A number of journal settings may now be configured on the kernel
858 command line. Specifically, the following options are now understood:
859 systemd.journald.max_level_console=,
860 systemd.journald.max_level_store=,
861 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
862 systemd.journald.max_level_wall=.
864 * "systemctl is-enabled --full" will now show by which symlinks a unit
865 file is enabled in the unit dependency tree.
867 * Support for VeraCrypt encrypted partitions has been added to the
868 "cryptsetup" logic and /etc/crypttab.
870 * systemd-detect-virt gained support for a new --private-users switch
871 that checks whether the invoking processes are running inside a user
872 namespace. Similar, a new special value "private-users" for the
873 existing ConditionVirtualization= setting has been added, permitting
874 skipping of specific units in user namespace environments.
876 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
877 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
878 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
879 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
880 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
881 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
882 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
883 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
884 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
885 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
886 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
887 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
888 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
889 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
890 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
891 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
892 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
893 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
894 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
895 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
896 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
897 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
898 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
899 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
900 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
901 Jędrzejewski-Szmek, Zeal Jagannatha
903 — Santa Fe, 2016-11-03
907 * In service units the various ExecXYZ= settings have been extended
908 with an additional special character as first argument of the
909 assigned value: if the character '+' is used the specified command
910 line it will be run with full privileges, regardless of User=,
911 Group=, CapabilityBoundingSet= and similar options. The effect is
912 similar to the existing PermissionsStartOnly= option, but allows
913 configuration of this concept for each executed command line
916 * Services may now alter the service watchdog timeout at runtime by
917 sending a WATCHDOG_USEC= message via sd_notify().
919 * MemoryLimit= and related unit settings now optionally take percentage
920 specifications. The percentage is taken relative to the amount of
921 physical memory in the system (or in case of containers, the assigned
922 amount of memory). This allows scaling service resources neatly with
923 the amount of RAM available on the system. Similarly, systemd-logind's
924 RuntimeDirectorySize= option now also optionally takes percentage
927 * In similar fashion TasksMax= takes percentage values now, too. The
928 value is taken relative to the configured maximum number of processes
929 on the system. The per-service task maximum has been changed to 15%
930 using this functionality. (Effectively this is an increase of 512 →
931 4915 for service units, given the kernel's default pid_max setting.)
933 * Calendar time specifications in .timer units now understand a ".."
934 syntax for time ranges. Example: "4..7:10" may now be used for
935 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
938 * The InaccessableDirectories=, ReadOnlyDirectories= and
939 ReadWriteDirectories= unit file settings have been renamed to
940 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
941 applied to all kinds of file nodes, and not just directories, with
942 the exception of symlinks. Specifically these settings may now be
943 used on block and character device nodes, UNIX sockets and FIFOS as
944 well as regular files. The old names of these settings remain
945 available for compatibility.
947 * systemd will now log about all service processes it kills forcibly
948 (using SIGKILL) because they remained after the clean shutdown phase
949 of the service completed. This should help identifying services that
950 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
951 systemd-logind's configuration a similar log message is generated for
952 processes killed at the end of each session due to this setting.
954 * systemd will now set the $JOURNAL_STREAM environment variable for all
955 services whose stdout/stderr are connected to the Journal (which
956 effectively means by default: all services). The variable contains
957 the device and inode number of the file descriptor used for
958 stdout/stderr. This may be used by invoked programs to detect whether
959 their stdout/stderr is connected to the Journal, in which case they
960 can switch over to direct Journal communication, thus being able to
961 pass extended, structured metadata along with their log messages. As
962 one example, this is now used by glib's logging primitives.
964 * When using systemd's default tmp.mount unit for /tmp, the mount point
965 will now be established with the "nosuid" and "nodev" options. This
966 avoids privilege escalation attacks that put traps and exploits into
967 /tmp. However, this might cause problems if you e. g. put container
968 images or overlays into /tmp; if you need this, override tmp.mount's
969 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
972 * systemd now supports the "memory" cgroup controller also on
975 * The systemd-cgtop tool now optionally takes a control group path as
976 command line argument. If specified, the control group list shown is
977 limited to subgroups of that group.
979 * The SystemCallFilter= unit file setting gained support for
980 pre-defined, named system call filter sets. For example
981 SystemCallFilter=@clock is now an effective way to make all clock
982 changing-related system calls unavailable to a service. A number of
983 similar pre-defined groups are defined. Writing system call filters
984 for system services is simplified substantially with this new
985 concept. Accordingly, all of systemd's own, long-running services now
986 enable system call filtering based on this, by default.
988 * A new service setting MemoryDenyWriteExecute= has been added, taking
989 a boolean value. If turned on, a service may no longer create memory
990 mappings that are writable and executable at the same time. This
991 enhances security for services where this is enabled as it becomes
992 harder to dynamically write and then execute memory in exploited
993 service processes. This option has been enabled for all of systemd's
994 own long-running services.
996 * A new RestrictRealtime= service setting has been added, taking a
997 boolean argument. If set the service's processes may no longer
998 acquire realtime scheduling. This improves security as realtime
999 scheduling may otherwise be used to easily freeze the system.
1001 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1002 value. This may be used for requesting that the system manager inside
1003 of the container reports start-up completion to nspawn which then
1004 propagates this notification further to the service manager
1005 supervising nspawn itself. A related option NotifyReady= in .nspawn
1006 files has been added too. This functionality allows ordering of the
1007 start-up of multiple containers using the usual systemd ordering
1010 * machinectl gained a new command "stop" that is an alias for
1013 * systemd-resolved gained support for contacting DNS servers on
1014 link-local IPv6 addresses.
1016 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1017 its caches. A method call for requesting the same operation has been
1018 added to the bus API too, and is made available via "systemd-resolve
1021 * systemd-resolve gained a new --status switch. If passed a brief
1022 summary of the used DNS configuration with per-interface information
1025 * resolved.conf gained a new Cache= boolean option, defaulting to
1026 on. If turned off local DNS caching is disabled. This comes with a
1027 performance penalty in particular when DNSSEC is enabled. Note that
1028 resolved disables its internal caching implicitly anyway, when the
1029 configured DNS server is on a host-local IP address such as ::1 or
1030 127.0.0.1, thus automatically avoiding double local caching.
1032 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1033 for DNS requests. This improves compatibility with local programs
1034 that do not use the libc NSS or systemd-resolved's bus APIs for name
1035 resolution. This minimal DNS service is only available to local
1036 programs and does not implement the full DNS protocol, but enough to
1037 cover local DNS clients. A new, static resolv.conf file, listing just
1038 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1039 now recommended to make /etc/resolv.conf a symlink to this file in
1040 order to route all DNS lookups to systemd-resolved, regardless if
1041 done via NSS, the bus API or raw DNS packets. Note that this local
1042 DNS service is not as fully featured as the libc NSS or
1043 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1044 used to deliver link-local address information (as this implies
1045 sending a local interface index along), LLMNR/mDNS support via this
1046 interface is severely restricted. It is thus strongly recommended for
1047 all applications to use the libc NSS API or native systemd-resolved
1050 * systemd-networkd's bridge support learned a new setting
1051 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1052 in .network files has been added for configuring VLAN bridging in
1053 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1055 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1056 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1057 now be acquired without relying on DHCPv6. Two new options
1058 UseDomains= and UseDNS= have been added to configure this behaviour.
1060 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1061 renamed IPv6AcceptRA=, without altering its behaviour. The old
1062 setting name remains available for compatibility reasons.
1064 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1065 Key=, InputKey= and OutputKey=.
1067 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1068 interface configuration.
1070 * "systemctl edit" may now be used to create new unit files by
1071 specifying the --force switch.
1073 * sd-event gained a new function sd_event_get_iteration() for
1074 requesting the current iteration counter of the event loop. It starts
1075 at zero and is increased by one with each event loop iteration.
1077 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1078 file. It can be used in lieu of %systemd_requires in packages which
1079 don't use any systemd functionality and are intended to be installed
1080 in minimal containers without systemd present. This macro provides
1081 ordering dependencies to ensure that if the package is installed in
1082 the same rpm transaction as systemd, systemd will be installed before
1083 the scriptlets for the package are executed, allowing unit presets
1086 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1087 been added to simplify packaging of generators.
1089 * The os-release file gained VERSION_CODENAME field for the
1090 distribution nickname (e.g. VERSION_CODENAME=woody).
1092 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1093 can be set to disable parsing of metadata and the creation
1094 of persistent symlinks for that device.
1096 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1097 to make them available to logged-in users has been reverted.
1099 * Much of the common code of the various systemd components is now
1100 built into an internal shared library libsystemd-shared-231.so
1101 (incorporating the systemd version number in the name, to be updated
1102 with future releases) that the components link to. This should
1103 decrease systemd footprint both in memory during runtime and on
1104 disk. Note that the shared library is not for public use, and is
1105 neither API not ABI stable, but is likely to change with every new
1106 released update. Packagers need to make sure that binaries
1107 linking to libsystemd-shared.so are updated in step with the
1110 * Configuration for "mkosi" is now part of the systemd
1111 repository. mkosi is a tool to easily build legacy-free OS images,
1112 and is available on github: https://github.com/systemd/mkosi. If
1113 "mkosi" is invoked in the build tree a new raw OS image is generated
1114 incorporating the systemd sources currently being worked on and a
1115 clean, fresh distribution installation. The generated OS image may be
1116 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1117 UEFI PC. This functionality is particularly useful to easily test
1118 local changes made to systemd in a pristine, defined environment. See
1119 HACKING for details.
1121 * configure learned the --with-support-url= option to specify the
1122 distribution's bugtracker.
1124 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1125 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1126 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1127 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1128 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1129 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1130 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1131 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1132 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1133 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1134 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1135 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1136 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1137 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1138 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1139 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1140 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1141 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1142 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1144 — Berlin, 2016-07-25
1148 * DNSSEC is now turned on by default in systemd-resolved (in
1149 "allow-downgrade" mode), but may be turned off during compile time by
1150 passing "--with-default-dnssec=no" to "configure" (and of course,
1151 during runtime with DNSSEC= in resolved.conf). We recommend
1152 downstreams to leave this on at least during development cycles and
1153 report any issues with the DNSSEC logic upstream. We are very
1154 interested in collecting feedback about the DNSSEC validator and its
1155 limitations in the wild. Note however, that DNSSEC support is
1156 probably nothing downstreams should turn on in stable distros just
1157 yet, as it might create incompatibilities with a few DNS servers and
1158 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1159 automatically whenever we detect such incompatible setups, but there
1160 might be systems we do not cover yet. Hence: please help us testing
1161 the DNSSEC code, leave this on where you can, report back, but then
1162 again don't consider turning this on in your stable, LTS or
1163 production release just yet. (Note that you have to enable
1164 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1165 and its DNSSEC mode for host name resolution from local
1168 * systemd-resolve conveniently resolves DANE records with the --tlsa
1169 option and OPENPGPKEY records with the --openpgp option. It also
1170 supports dumping raw DNS record data via the new --raw= switch.
1172 * systemd-logind will now by default terminate user processes that are
1173 part of the user session scope unit (session-XX.scope) when the user
1174 logs out. This behavior is controlled by the KillUserProcesses=
1175 setting in logind.conf, and the previous default of "no" is now
1176 changed to "yes". This means that user sessions will be properly
1177 cleaned up after, but additional steps are necessary to allow
1178 intentionally long-running processes to survive logout.
1180 While the user is logged in at least once, user@.service is running,
1181 and any service that should survive the end of any individual login
1182 session can be started at a user service or scope using systemd-run.
1183 systemd-run(1) man page has been extended with an example which shows
1184 how to run screen in a scope unit underneath user@.service. The same
1185 command works for tmux.
1187 After the user logs out of all sessions, user@.service will be
1188 terminated too, by default, unless the user has "lingering" enabled.
1189 To effectively allow users to run long-term tasks even if they are
1190 logged out, lingering must be enabled for them. See loginctl(1) for
1191 details. The default polkit policy was modified to allow users to
1192 set lingering for themselves without authentication.
1194 Previous defaults can be restored at compile time by the
1195 --without-kill-user-processes option to "configure".
1197 * systemd-logind gained new configuration settings SessionsMax= and
1198 InhibitorsMax=, both with a default of 8192. It will not register new
1199 user sessions or inhibitors above this limit.
1201 * systemd-logind will now reload configuration on SIGHUP.
1203 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1204 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1205 enable. Also, support for the "io" cgroup controller in the unified
1206 hierarchy has been added, so that the "memory", "pids" and "io" are
1207 now the controllers that are supported on the unified hierarchy.
1209 WARNING: it is not possible to use previous systemd versions with
1210 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1211 is necessary to also update systemd in the initramfs if using the
1212 unified hierarchy. An updated SELinux policy is also required.
1214 * LLDP support has been extended, and both passive (receive-only) and
1215 active (sender) modes are supported. Passive mode ("routers-only") is
1216 enabled by default in systemd-networkd. Active LLDP mode is enabled
1217 by default for containers on the internal network. The "networkctl
1218 lldp" command may be used to list information gathered. "networkctl
1219 status" will also show basic LLDP information on connected peers now.
1221 * The IAID and DUID unique identifier sent in DHCP requests may now be
1222 configured for the system and each .network file managed by
1223 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1225 * systemd-networkd gained support for configuring proxy ARP support for
1226 each interface, via the ProxyArp= setting in .network files. It also
1227 gained support for configuring the multicast querier feature of
1228 bridge devices, via the new MulticastQuerier= setting in .netdev
1229 files. Similarly, snooping on the IGMP traffic can be controlled
1230 via the new setting MulticastSnooping=.
1232 A new setting PreferredLifetime= has been added for addresses
1233 configured in .network file to configure the lifetime intended for an
1236 The systemd-networkd DHCP server gained the option EmitRouter=, which
1237 defaults to yes, to configure whether the DHCP Option 3 (Router)
1240 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1241 systemd-socket-activate and installed into /usr/bin. It is now fully
1244 * systemd-journald now uses separate threads to flush changes to disk
1245 when closing journal files, thus reducing impact of slow disk I/O on
1246 logging performance.
1248 * The sd-journal API gained two new calls
1249 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1250 can be used to open journal files using file descriptors instead of
1251 file or directory paths. sd_journal_open_container() has been
1252 deprecated, sd_journal_open_directory_fd() should be used instead
1253 with the flag SD_JOURNAL_OS_ROOT.
1255 * journalctl learned a new output mode "-o short-unix" that outputs log
1256 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1257 UTC). It also gained support for a new --no-hostname setting to
1258 suppress the hostname column in the family of "short" output modes.
1260 * systemd-ask-password now optionally skips printing of the password to
1261 stdout with --no-output which can be useful in scripts.
1263 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1264 (devices tagged with ID_MAKER_TOOL) are now tagged with
1265 "uaccess" and are available to logged in users.
1267 * The DeviceAllow= unit setting now supports specifiers (with "%").
1269 * "systemctl show" gained a new --value switch, which allows print a
1270 only the contents of a specific unit property, without also printing
1271 the property's name. Similar support was added to "show*" verbs
1272 of loginctl and machinectl that output "key=value" lists.
1274 * A new unit type "generated" was added for files dynamically generated
1275 by generator tools. Similarly, a new unit type "transient" is used
1276 for unit files created using the runtime API. "systemctl enable" will
1277 refuse to operate on such files.
1279 * A new command "systemctl revert" has been added that may be used to
1280 revert to the vendor version of a unit file, in case local changes
1281 have been made by adding drop-ins or overriding the unit file.
1283 * "machinectl clean" gained a new verb to automatically remove all or
1284 just hidden container images.
1286 * systemd-tmpfiles gained support for a new line type "e" for emptying
1287 directories, if they exist, without creating them if they don't.
1289 * systemd-nspawn gained support for automatically patching the UID/GIDs
1290 of the owners and the ACLs of all files and directories in a
1291 container tree to match the UID/GID user namespacing range selected
1292 for the container invocation. This mode is enabled via the new
1293 --private-users-chown switch. It also gained support for
1294 automatically choosing a free, previously unused UID/GID range when
1295 starting a container, via the new --private-users=pick setting (which
1296 implies --private-users-chown). Together, these options for the first
1297 time make user namespacing for nspawn containers fully automatic and
1298 thus deployable. The systemd-nspawn@.service template unit file has
1299 been changed to use this functionality by default.
1301 * systemd-nspawn gained a new --network-zone= switch, that allows
1302 creating ad-hoc virtual Ethernet links between multiple containers,
1303 that only exist as long as at least one container referencing them is
1304 running. This allows easy connecting of multiple containers with a
1305 common link that implements an Ethernet broadcast domain. Each of
1306 these network "zones" may be named relatively freely by the user, and
1307 may be referenced by any number of containers, but each container may
1308 only reference one of these "zones". On the lower level, this is
1309 implemented by an automatically managed bridge network interface for
1310 each zone, that is created when the first container referencing its
1311 zone is created and removed when the last one referencing its zone
1314 * The default start timeout may now be configured on the kernel command
1315 line via systemd.default_timeout_start_sec=. It was already
1316 configurable via the DefaultTimeoutStartSec= option in
1317 /etc/systemd/system.conf.
1319 * Socket units gained a new TriggerLimitIntervalSec= and
1320 TriggerLimitBurst= setting to configure a limit on the activation
1321 rate of the socket unit.
1323 * The LimitNICE= setting now optionally takes normal UNIX nice values
1324 in addition to the raw integer limit value. If the specified
1325 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1326 value is understood as UNIX nice value. If not prefixed like this it
1327 is understood as raw RLIMIT_NICE limit.
1329 * Note that the effect of the PrivateDevices= unit file setting changed
1330 slightly with this release: the per-device /dev file system will be
1331 mounted read-only from this version on, and will have "noexec"
1332 set. This (minor) change of behavior might cause some (exceptional)
1333 legacy software to break, when PrivateDevices=yes is set for its
1334 service. Please leave PrivateDevices= off if you run into problems
1337 * systemd-bootchart has been split out to a separate repository:
1338 https://github.com/systemd/systemd-bootchart
1340 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1341 merged into the kernel in its current form.
1343 * The compatibility libraries libsystemd-daemon.so,
1344 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1345 which have been deprecated since systemd-209 have been removed along
1346 with the corresponding pkg-config files. All symbols provided by
1347 those libraries are provided by libsystemd.so.
1349 * The Capabilities= unit file setting has been removed (it is ignored
1350 for backwards compatibility). AmbientCapabilities= and
1351 CapabilityBoundingSet= should be used instead.
1353 * A new special target has been added, initrd-root-device.target,
1354 which creates a synchronization point for dependencies of the root
1355 device in early userspace. Initramfs builders must ensure that this
1356 target is now included in early userspace.
1358 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1359 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1360 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1361 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1362 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1363 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1364 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1365 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1366 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1367 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1368 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1369 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1370 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1371 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1372 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1373 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1374 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1375 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1376 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1377 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1378 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1379 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1380 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1381 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1382 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1385 — Fairfax, 2016-05-21
1389 * The systemd-resolved DNS resolver service has gained a substantial
1390 set of new features, most prominently it may now act as a DNSSEC
1391 validating stub resolver. DNSSEC mode is currently turned off by
1392 default, but is expected to be turned on by default in one of the
1393 next releases. For now, we invite everybody to test the DNSSEC logic
1394 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1395 service also gained a full set of D-Bus interfaces, including calls
1396 to configure DNS and DNSSEC settings per link (for use by external
1397 network management software). systemd-resolved and systemd-networkd
1398 now distinguish between "search" and "routing" domains. The former
1399 are used to qualify single-label names, the latter are used purely
1400 for routing lookups within certain domains to specific links.
1401 resolved now also synthesizes RRs for all entries from /etc/hosts.
1403 * The systemd-resolve tool (which is a client utility for
1404 systemd-resolved) has been improved considerably and is now fully
1405 supported and documented. Hence it has moved from /usr/lib/systemd to
1408 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1411 * The coredump collection logic has been reworked: when a coredump is
1412 collected it is now written to disk, compressed and processed
1413 (including stacktrace extraction) from a new instantiated service
1414 systemd-coredump@.service, instead of directly from the
1415 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1416 processing large coredumps can take up a substantial amount of
1417 resources and time, and this previously happened entirely outside of
1418 systemd's service supervision. With the new logic the core_pattern
1419 hook only does minimal metadata collection before passing off control
1420 to the new instantiated service, which is configured with a time
1421 limit, a nice level and other settings to minimize negative impact on
1422 the rest of the system. Also note that the new logic will honour the
1423 RLIMIT_CORE setting of the crashed process, which now allows users
1424 and processes to turn off coredumping for their processes by setting
1427 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1428 and all forked processes by default. Previously, PID 1 would leave
1429 the setting at "0" for all processes, as set by the kernel. Note that
1430 the resource limit traditionally has no effect on the generated
1431 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1432 logic is used. Since the limit is now honoured (see above) its
1433 default has been changed so that the coredumping logic is enabled by
1434 default for all processes, while allowing specific opt-out.
1436 * When the stacktrace is extracted from processes of system users, this
1437 is now done as "systemd-coredump" user, in order to sandbox this
1438 potentially security sensitive parsing operation. (Note that when
1439 processing coredumps of normal users this is done under the user ID
1440 of process that crashed, as before.) Packagers should take notice
1441 that it is now necessary to create the "systemd-coredump" system user
1442 and group at package installation time.
1444 * The systemd-activate socket activation testing tool gained support
1445 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1446 and --seqpacket switches. It also has been extended to support both
1447 new-style and inetd-style file descriptor passing. Use the new
1448 --inetd switch to request inetd-style file descriptor passing.
1450 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1451 variable, which takes a boolean value. If set to false, ANSI color
1452 output is disabled in the tools even when run on a terminal that
1455 * The VXLAN support in networkd now supports two new settings
1456 DestinationPort= and PortRange=.
1458 * A new systemd.machine_id= kernel command line switch has been added,
1459 that may be used to set the machine ID in /etc/machine-id if it is
1460 not initialized yet. This command line option has no effect if the
1461 file is already initialized.
1463 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1464 specified command line as PID 2 rather than PID 1 in the
1465 container. In this mode PID 1 is a minimal stub init process that
1466 implements the special POSIX and Linux semantics of PID 1 regarding
1467 signal and child process management. Note that this stub init process
1468 is implemented in nspawn itself and requires no support from the
1469 container image. This new logic is useful to support running
1470 arbitrary commands in the container, as normal processes are
1471 generally not prepared to run as PID 1.
1473 * systemd-nspawn gained a new --chdir= switch for setting the current
1474 working directory for the process started in the container.
1476 * "journalctl /dev/sda" will now output all kernel log messages for
1477 specified device from the current boot, in addition to all devices
1478 that are parents of it. This should make log output about devices
1479 pretty useful, as long as kernel drivers attach enough metadata to
1480 the log messages. (The usual SATA drivers do.)
1482 * The sd-journal API gained two new calls
1483 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1484 that report whether log data from /run or /var has been found.
1486 * journalctl gained a new switch "--fields" that prints all journal
1487 record field names currently in use in the journal. This is backed
1488 by two new sd-journal API calls sd_journal_enumerate_fields() and
1489 sd_journal_restart_fields().
1491 * Most configurable timeouts in systemd now expect an argument of
1492 "infinity" to turn them off, instead of "0" as before. The semantics
1493 from now on is that a timeout of "0" means "now", and "infinity"
1494 means "never". To maintain backwards compatibility, "0" continues to
1495 turn off previously existing timeout settings.
1497 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1498 try-reload-or-restart" to clarify what it actually does: the "try"
1499 logic applies to both reloading and restarting, not just restarting.
1500 The old name continues to be accepted for compatibility.
1502 * On boot-up, when PID 1 detects that the system clock is behind the
1503 release date of the systemd version in use, the clock is now set
1504 to the latter. Previously, this was already done in timesyncd, in order
1505 to avoid running with clocks set to the various clock epochs such as
1506 1902, 1938 or 1970. With this change the logic is now done in PID 1
1507 in addition to timesyncd during early boot-up, so that it is enforced
1508 before the first process is spawned by systemd. Note that the logic
1509 in timesyncd remains, as it is more comprehensive and ensures
1510 clock monotonicity by maintaining a persistent timestamp file in
1511 /var. Since /var is generally not available in earliest boot or the
1512 initrd, this part of the logic remains in timesyncd, and is not done
1515 * Support for tweaking details in net_cls.class_id through the
1516 NetClass= configuration directive has been removed, as the kernel
1517 people have decided to deprecate that controller in cgroup v2.
1518 Userspace tools such as nftables are moving over to setting rules
1519 that are specific to the full cgroup path of a task, which obsoletes
1520 these controllers anyway. The NetClass= directive is kept around for
1521 legacy compatibility reasons. For a more in-depth description of the
1522 kernel change, please refer to the respective upstream commit:
1524 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1526 * A new service setting RuntimeMaxSec= has been added that may be used
1527 to specify a maximum runtime for a service. If the timeout is hit, the
1528 service is terminated and put into a failure state.
1530 * A new service setting AmbientCapabilities= has been added. It allows
1531 configuration of additional Linux process capabilities that are
1532 passed to the activated processes. This is only available on very
1535 * The process resource limit settings in service units may now be used
1536 to configure hard and soft limits individually.
1538 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1539 expose support for gcc's __attribute__((cleanup())) C extension.
1540 Specifically, for many object destructor functions alternative
1541 versions have been added that have names suffixed with "p" and take a
1542 pointer to a pointer to the object to destroy, instead of just a
1543 pointer to the object itself. This is useful because these destructor
1544 functions may be used directly as parameters to the cleanup
1545 construct. Internally, systemd has been a heavy user of this GCC
1546 extension for a long time, and with this change similar support is
1547 now available to consumers of the library outside of systemd. Note
1548 that by using this extension in your sources compatibility with old
1549 and strictly ANSI compatible C compilers is lost. However, all gcc or
1550 LLVM versions of recent years support this extension.
1552 * Timer units gained support for a new setting RandomizedDelaySec= that
1553 allows configuring some additional randomized delay to the configured
1554 time. This is useful to spread out timer events to avoid load peaks in
1555 clusters or larger setups.
1557 * Calendar time specifications now support sub-second accuracy.
1559 * Socket units now support listening on SCTP and UDP-lite protocol
1562 * The sd-event API now comes with a full set of man pages.
1564 * Older versions of systemd contained experimental support for
1565 compressing journal files and coredumps with the LZ4 compressor that
1566 was not compatible with the lz4 binary (due to API limitations of the
1567 lz4 library). This support has been removed; only support for files
1568 compatible with the lz4 binary remains. This LZ4 logic is now
1569 officially supported and no longer considered experimental.
1571 * The dkr image import logic has been removed again from importd. dkr's
1572 micro-services focus doesn't fit into the machine image focus of
1573 importd, and quickly got out of date with the upstream dkr API.
1575 * Creation of the /run/lock/lockdev/ directory was dropped from
1576 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1577 been available for many years. If you still need this, you need to
1578 create your own tmpfiles.d config file with:
1580 d /run/lock/lockdev 0775 root lock -
1582 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1583 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1584 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1585 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1586 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1587 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1588 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1589 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1590 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1591 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1592 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1593 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1594 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1595 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1596 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1597 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1598 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1599 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1600 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1602 — Berlin, 2016-02-11
1606 * A number of properties previously only settable in unit
1607 files are now also available as properties to set when
1608 creating transient units programmatically via the bus, as it
1609 is exposed with systemd-run's --property=
1610 setting. Specifically, these are: SyslogIdentifier=,
1611 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1612 EnvironmentFile=, ReadWriteDirectories=,
1613 ReadOnlyDirectories=, InaccessibleDirectories=,
1614 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1616 * When creating transient services via the bus API it is now
1617 possible to pass in a set of file descriptors to use as
1618 STDIN/STDOUT/STDERR for the invoked process.
1620 * Slice units may now be created transiently via the bus APIs,
1621 similar to the way service and scope units may already be
1622 created transiently.
1624 * Wherever systemd expects a calendar timestamp specification
1625 (like in journalctl's --since= and --until= switches) UTC
1626 timestamps are now supported. Timestamps suffixed with "UTC"
1627 are now considered to be in Universal Time Coordinated
1628 instead of the local timezone. Also, timestamps may now
1629 optionally be specified with sub-second accuracy. Both of
1630 these additions also apply to recurring calendar event
1631 specification, such as OnCalendar= in timer units.
1633 * journalctl gained a new "--sync" switch that asks the
1634 journal daemon to write all so far unwritten log messages to
1635 disk and sync the files, before returning.
1637 * systemd-tmpfiles learned two new line types "q" and "Q" that
1638 operate like "v", but also set up a basic btrfs quota
1639 hierarchy when used on a btrfs file system with quota
1642 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1643 instead of a subvolume (even on a btrfs file system) if the
1644 root directory is a plain directory, and not a
1645 subvolume. This should simplify things with certain chroot()
1646 environments which are not aware of the concept of btrfs
1649 * systemd-detect-virt gained a new --chroot switch to detect
1650 whether execution takes place in a chroot() environment.
1652 * CPUAffinity= now takes CPU index ranges in addition to
1655 * The various memory-related resource limit settings (such as
1656 LimitAS=) now understand the usual K, M, G, ... suffixes to
1657 the base of 1024 (IEC). Similar, the time-related resource
1658 limit settings understand the usual min, h, day, ...
1661 * There's a new system.conf setting DefaultTasksMax= to
1662 control the default TasksMax= setting for services and
1663 scopes running on the system. (TasksMax= is the primary
1664 setting that exposes the "pids" cgroup controller on systemd
1665 and was introduced in the previous systemd release.) The
1666 setting now defaults to 512, which means services that are
1667 not explicitly configured otherwise will only be able to
1668 create 512 processes or threads at maximum, from this
1669 version on. Note that this means that thread- or
1670 process-heavy services might need to be reconfigured to set
1671 TasksMax= to a higher value. It is sufficient to set
1672 TasksMax= in these specific unit files to a higher value, or
1673 even "infinity". Similar, there's now a logind.conf setting
1674 UserTasksMax= that defaults to 4096 and limits the total
1675 number of processes or tasks each user may own
1676 concurrently. nspawn containers also have the TasksMax=
1677 value set by default now, to 8192. Note that all of this
1678 only has an effect if the "pids" cgroup controller is
1679 enabled in the kernel. The general benefit of these changes
1680 should be a more robust and safer system, that provides a
1681 certain amount of per-service fork() bomb protection.
1683 * systemd-nspawn gained the new --network-veth-extra= switch
1684 to define additional and arbitrarily-named virtual Ethernet
1685 links between the host and the container.
1687 * A new service execution setting PassEnvironment= has been
1688 added that allows importing select environment variables
1689 from PID1's environment block into the environment block of
1692 * Timer units gained support for a new RemainAfterElapse=
1693 setting which takes a boolean argument. It defaults to on,
1694 exposing behaviour unchanged to previous releases. If set to
1695 off, timer units are unloaded after they elapsed if they
1696 cannot elapse again. This is particularly useful for
1697 transient timer units, which shall not stay around longer
1698 than until they first elapse.
1700 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1701 default now (the kernel default is 16). This is beneficial
1702 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1703 allows substantially larger numbers of queued
1704 datagrams. This should increase the capability of systemd to
1705 parallelize boot-up, as logging and sd_notify() are unlikely
1706 to stall execution anymore. If you need to change the value
1707 from the new defaults, use the usual sysctl.d/ snippets.
1709 * The compression framing format used by the journal or
1710 coredump processing has changed to be in line with what the
1711 official LZ4 tools generate. LZ4 compression support in
1712 systemd was considered unsupported previously, as the format
1713 was not compatible with the normal tools. With this release
1714 this has changed now, and it is hence safe for downstream
1715 distributions to turn it on. While not compressing as well
1716 as the XZ, LZ4 is substantially faster, which makes
1717 it a good default choice for the compression logic in the
1718 journal and in coredump handling.
1720 * Any reference to /etc/mtab has been dropped from
1721 systemd. The file has been obsolete since a while, but
1722 systemd refused to work on systems where it was incorrectly
1723 set up (it should be a symlink or non-existent). Please make
1724 sure to update to util-linux 2.27.1 or newer in conjunction
1725 with this systemd release, which also drops any reference to
1726 /etc/mtab. If you maintain a distribution make sure that no
1727 software you package still references it, as this is a
1728 likely source of bugs. There's also a glibc bug pending,
1729 asking for removal of any reference to this obsolete file:
1731 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1733 Note that only util-linux versions built with
1734 --enable-libmount-force-mountinfo are supported.
1736 * Support for the ".snapshot" unit type has been removed. This
1737 feature turned out to be little useful and little used, and
1738 has now been removed from the core and from systemctl.
1740 * The dependency types RequiresOverridable= and
1741 RequisiteOverridable= have been removed from systemd. They
1742 have been used only very sparingly to our knowledge and
1743 other options that provide a similar effect (such as
1744 systemctl --mode=ignore-dependencies) are much more useful
1745 and commonly used. Moreover, they were only half-way
1746 implemented as the option to control behaviour regarding
1747 these dependencies was never added to systemctl. By removing
1748 these dependency types the execution engine becomes a bit
1749 simpler. Unit files that use these dependencies should be
1750 changed to use the non-Overridable dependency types
1751 instead. In fact, when parsing unit files with these
1752 options, that's what systemd will automatically convert them
1753 too, but it will also warn, asking users to fix the unit
1754 files accordingly. Removal of these dependency types should
1755 only affect a negligible number of unit files in the wild.
1757 * Behaviour of networkd's IPForward= option changed
1758 (again). It will no longer maintain a per-interface setting,
1759 but propagate one way from interfaces where this is enabled
1760 to the global kernel setting. The global setting will be
1761 enabled when requested by a network that is set up, but
1762 never be disabled again. This change was made to make sure
1763 IPv4 and IPv6 behaviour regarding packet forwarding is
1764 similar (as the Linux IPv6 stack does not support
1765 per-interface control of this setting) and to minimize
1768 * In unit files the behaviour of %u, %U, %h, %s has
1769 changed. These specifiers will now unconditionally resolve
1770 to the various user database fields of the user that the
1771 systemd instance is running as, instead of the user
1772 configured in the specific unit via User=. Note that this
1773 effectively doesn't change much, as resolving of these
1774 specifiers was already turned off in the --system instance
1775 of systemd, as we cannot do NSS lookups from PID 1. In the
1776 --user instance of systemd these specifiers where correctly
1777 resolved, but hardly made any sense, since the user instance
1778 lacks privileges to do user switches anyway, and User= is
1779 hence useless. Moreover, even in the --user instance of
1780 systemd behaviour was awkward as it would only take settings
1781 from User= assignment placed before the specifier into
1782 account. In order to unify and simplify the logic around
1783 this the specifiers will now always resolve to the
1784 credentials of the user invoking the manager (which in case
1785 of PID 1 is the root user).
1787 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1788 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1789 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1790 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1791 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1792 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1793 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1794 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1795 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1796 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1797 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1798 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1799 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1800 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1803 — Berlin, 2015-11-18
1807 * systemd now depends on util-linux v2.27. More specifically,
1808 the newly added mount monitor feature in libmount now
1809 replaces systemd's former own implementation.
1811 * libmount mandates /etc/mtab not to be regular file, and
1812 systemd now enforces this condition at early boot.
1813 /etc/mtab has been deprecated and warned about for a very
1814 long time, so systems running systemd should already have
1815 stopped having this file around as anything else than a
1816 symlink to /proc/self/mounts.
1818 * Support for the "pids" cgroup controller has been added. It
1819 allows accounting the number of tasks in a cgroup and
1820 enforcing limits on it. This adds two new setting
1821 TasksAccounting= and TasksMax= to each unit, as well as a
1822 global option DefaultTasksAccounting=.
1824 * Support for the "net_cls" cgroup controller has been added.
1825 It allows assigning a net class ID to each task in the
1826 cgroup, which can then be used in firewall rules and traffic
1827 shaping configurations. Note that the kernel netfilter net
1828 class code does not currently work reliably for ingress
1829 packets on unestablished sockets.
1831 This adds a new config directive called NetClass= to CGroup
1832 enabled units. Allowed values are positive numbers for fixed
1833 assignments and "auto" for picking a free value
1836 * 'systemctl is-system-running' now returns 'offline' if the
1837 system is not booted with systemd. This command can now be
1838 used as a substitute for 'systemd-notify --booted'.
1840 * Watchdog timeouts have been increased to 3 minutes for all
1841 in-tree service files. Apparently, disk IO issues are more
1842 frequent than we hoped, and user reported >1 minute waiting
1845 * 'machine-id-commit' functionality has been merged into
1846 'machine-id-setup --commit'. The separate binary has been
1849 * The WorkingDirectory= directive in unit files may now be set
1850 to the special value '~'. In this case, the working
1851 directory is set to the home directory of the user
1852 configured in User=.
1854 * "machinectl shell" will now open the shell in the home
1855 directory of the selected user by default.
1857 * The CrashChVT= configuration file setting is renamed to
1858 CrashChangeVT=, following our usual logic of not
1859 abbreviating unnecessarily. The old directive is still
1860 supported for compat reasons. Also, this directive now takes
1861 an integer value between 1 and 63, or a boolean value. The
1862 formerly supported '-1' value for disabling stays around for
1865 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1866 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1867 RootDirectory= properties can now be set for transient
1870 * The systemd-analyze tool gained a new "set-log-target" verb
1871 to change the logging target the system manager logs to
1872 dynamically during runtime. This is similar to how
1873 "systemd-analyze set-log-level" already changes the log
1876 * In nspawn /sys is now mounted as tmpfs, with only a selected
1877 set of subdirectories mounted in from the real sysfs. This
1878 enhances security slightly, and is useful for ensuring user
1879 namespaces work correctly.
1881 * Support for USB FunctionFS activation has been added. This
1882 allows implementation of USB gadget services that are
1883 activated as soon as they are requested, so that they don't
1884 have to run continuously, similar to classic socket
1887 * The "systemctl exit" command now optionally takes an
1888 additional parameter that sets the exit code to return from
1889 the systemd manager when exiting. This is only relevant when
1890 running the systemd user instance, or when running the
1891 system instance in a container.
1893 * sd-bus gained the new API calls sd_bus_path_encode_many()
1894 and sd_bus_path_decode_many() that allow easy encoding and
1895 decoding of multiple identifier strings inside a D-Bus
1896 object path. Another new call sd_bus_default_flush_close()
1897 has been added to flush and close per-thread default
1900 * systemd-cgtop gained support for a -M/--machine= switch to
1901 show the control groups within a certain container only.
1903 * "systemctl kill" gained support for an optional --fail
1904 switch. If specified the requested operation will fail of no
1905 processes have been killed, because the unit had no
1906 processes attached, or similar.
1908 * A new systemd.crash_reboot=1 kernel command line option has
1909 been added that triggers a reboot after crashing. This can
1910 also be set through CrashReboot= in systemd.conf.
1912 * The RuntimeDirectory= setting now understands unit
1913 specifiers like %i or %f.
1915 * A new (still internal) library API sd-ipv4acd has been added,
1916 that implements address conflict detection for IPv4. It's
1917 based on code from sd-ipv4ll, and will be useful for
1918 detecting DHCP address conflicts.
1920 * File descriptors passed during socket activation may now be
1921 named. A new API sd_listen_fds_with_names() is added to
1922 access the names. The default names may be overridden,
1923 either in the .socket file using the FileDescriptorName=
1924 parameter, or by passing FDNAME= when storing the file
1925 descriptors using sd_notify().
1927 * systemd-networkd gained support for:
1929 - Setting the IPv6 Router Advertisement settings via
1930 IPv6AcceptRouterAdvertisements= in .network files.
1932 - Configuring the HelloTimeSec=, MaxAgeSec= and
1933 ForwardDelaySec= bridge parameters in .netdev files.
1935 - Configuring PreferredSource= for static routes in
1938 * The "ask-password" framework used to query for LUKS harddisk
1939 passwords or SSL passwords during boot gained support for
1940 caching passwords in the kernel keyring, if it is
1941 available. This makes sure that the user only has to type in
1942 a passphrase once if there are multiple objects to unlock
1943 with the same one. Previously, such password caching was
1944 available only when Plymouth was used; this moves the
1945 caching logic into the systemd codebase itself. The
1946 "systemd-ask-password" utility gained a new --keyname=
1947 switch to control which kernel keyring key to use for
1948 caching a password in. This functionality is also useful for
1949 enabling display managers such as gdm to automatically
1950 unlock the user's GNOME keyring if its passphrase, the
1951 user's password and the harddisk password are the same, if
1952 gdm-autologin is used.
1954 * When downloading tar or raw images using "machinectl
1955 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1956 file is now also downloaded, if it is available and stored
1957 next to the image file.
1959 * Units of type ".socket" gained a new boolean setting
1960 Writable= which is only useful in conjunction with
1961 ListenSpecial=. If true, enables opening the specified
1962 special file in O_RDWR mode rather than O_RDONLY mode.
1964 * systemd-rfkill has been reworked to become a singleton
1965 service that is activated through /dev/rfkill on each rfkill
1966 state change and saves the settings to disk. This way,
1967 systemd-rfkill is now compatible with devices that exist
1968 only intermittendly, and even restores state if the previous
1969 system shutdown was abrupt rather than clean.
1971 * The journal daemon gained support for vacuuming old journal
1972 files controlled by the number of files that shall remain,
1973 in addition to the already existing control by size and by
1974 date. This is useful as journal interleaving performance
1975 degrades with too many separate journal files, and allows
1976 putting an effective limit on them. The new setting defaults
1977 to 100, but this may be changed by setting SystemMaxFiles=
1978 and RuntimeMaxFiles= in journald.conf. Also, the
1979 "journalctl" tool gained the new --vacuum-files= switch to
1980 manually vacuum journal files to leave only the specified
1981 number of files in place.
1983 * udev will now create /dev/disk/by-path links for ATA devices
1984 on kernels where that is supported.
1986 * Galician, Serbian, Turkish and Korean translations were added.
1988 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1989 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1990 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1991 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1992 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1993 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1994 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1995 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1996 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1997 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1998 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1999 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2000 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2001 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2002 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2003 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2004 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2005 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2007 — Berlin, 2015-10-07
2011 * The DHCP implementation of systemd-networkd gained a set of
2014 - The DHCP server now supports emitting DNS and NTP
2015 information. It may be enabled and configured via
2016 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2017 and NTP information is enabled, but no servers are
2018 configured, the corresponding uplink information (if there
2019 is any) is propagated.
2021 - Server and client now support transmission and reception
2022 of timezone information. It can be configured via the
2023 newly introduced network options UseTimezone=,
2024 EmitTimezone=, and Timezone=. Transmission of timezone
2025 information is enabled between host and containers by
2026 default now: the container will change its local timezone
2027 to what the host has set.
2029 - Lease timeouts can now be configured via
2030 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2032 - The DHCP server improved on the stability of
2033 leases. Clients are more likely to get the same lease
2034 information back, even if the server loses state.
2036 - The DHCP server supports two new configuration options to
2037 control the lease address pool metrics, PoolOffset= and
2040 * The encapsulation limit of tunnels in systemd-networkd may
2041 now be configured via 'EncapsulationLimit='. It allows
2042 modifying the maximum additional levels of encapsulation
2043 that are permitted to be prepended to a packet.
2045 * systemd now supports the concept of user buses replacing
2046 session buses, if used with dbus-1.10 (and enabled via dbus
2047 --enable-user-session). It previously only supported this on
2048 kdbus-enabled systems, and this release expands this to
2049 'dbus-daemon' systems.
2051 * systemd-networkd now supports predictable interface names
2054 * systemd now optionally supports the new Linux kernel
2055 "unified" control group hierarchy. If enabled via the kernel
2056 command-line option 'systemd.unified_cgroup_hierarchy=1',
2057 systemd will try to mount the unified cgroup hierarchy
2058 directly on /sys/fs/cgroup. If not enabled, or not
2059 available, systemd will fall back to the legacy cgroup
2060 hierarchy setup, as before. Host system and containers can
2061 mix and match legacy and unified hierarchies as they
2062 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2063 environment variable to individually select the hierarchy to
2064 use for executed containers. By default, nspawn will use the
2065 unified hierarchy for the containers if the host uses the
2066 unified hierarchy, and the legacy hierarchy otherwise.
2067 Please note that at this point the unified hierarchy is an
2068 experimental kernel feature and is likely to change in one
2069 of the next kernel releases. Therefore, it should not be
2070 enabled by default in downstream distributions yet. The
2071 minimum required kernel version for the unified hierarchy to
2072 work is 4.2. Note that when the unified hierarchy is used
2073 for the first time delegated access to controllers is
2074 safe. Because of this systemd-nspawn containers will get
2075 access to controllers now, as will systemd user
2076 sessions. This means containers and user sessions may now
2077 manage their own resources, partitioning up what the system
2080 * A new special scope unit "init.scope" has been introduced
2081 that encapsulates PID 1 of the system. It may be used to
2082 determine resource usage and enforce resource limits on PID
2083 1 itself. PID 1 hence moved out of the root of the control
2086 * The cgtop tool gained support for filtering out kernel
2087 threads when counting tasks in a control group. Also, the
2088 count of processes is now recursively summed up by
2089 default. Two options -k and --recursive= have been added to
2090 revert to old behaviour. The tool has also been updated to
2091 work correctly in containers now.
2093 * systemd-nspawn's --bind= and --bind-ro= options have been
2094 extended to allow creation of non-recursive bind mounts.
2096 * libsystemd gained two new calls sd_pid_get_cgroup() and
2097 sd_peer_get_cgroup() which return the control group path of
2098 a process or peer of a connected AF_UNIX socket. This
2099 function call is particularly useful when implementing
2100 delegated subtrees support in the control group hierarchy.
2102 * The "sd-event" event loop API of libsystemd now supports
2103 correct dequeuing of real-time signals, without losing
2106 * When systemd requests a PolicyKit decision when managing
2107 units it will now add additional fields to the request,
2108 including unit name and desired operation. This enables more
2109 powerful PolicyKit policies, that make decisions depending
2110 on these parameters.
2112 * nspawn learnt support for .nspawn settings files, that may
2113 accompany the image files or directories of containers, and
2114 may contain additional settings for the container. This is
2115 an alternative to configuring container parameters via the
2116 nspawn command line.
2118 Contributions from: Cristian Rodríguez, Daniel Mack, David
2119 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2120 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2121 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2122 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2123 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2124 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2125 Andersen, Tom Gundersen, Torstein Husebø
2127 — Berlin, 2015-09-08
2131 * machinectl gained a new verb 'shell' which opens a fresh
2132 shell on the target container or the host. It is similar to
2133 the existing 'login' command of machinectl, but spawns the
2134 shell directly without prompting for username or
2135 password. The pseudo machine '.host' now refers to the local
2136 host and is used by default. Hence, 'machinectl shell' can
2137 be used as replacement for 'su -' which spawns a session as
2138 a fresh systemd unit in a way that is fully isolated from
2139 the originating session.
2141 * systemd-networkd learned to cope with private-zone DHCP
2142 options and allows other programs to query the values.
2144 * SELinux access control when enabling/disabling units is no
2145 longer enforced with this release. The previous
2146 implementation was incorrect, and a new corrected
2147 implementation is not yet available. As unit file operations
2148 are still protected via PolicyKit and D-Bus policy this is
2149 not a security problem. Yet, distributions which care about
2150 optimal SELinux support should probably not stabilize on
2153 * sd-bus gained support for matches of type "arg0has=", that
2154 test for membership of strings in string arrays sent in bus
2157 * systemd-resolved now dumps the contents of its DNS and LLMNR
2158 caches to the logs on reception of the SIGUSR1 signal. This
2159 is useful to debug DNS behaviour.
2161 * The coredumpctl tool gained a new --directory= option to
2162 operate on journal files in a specific directory.
2164 * "systemctl reboot" and related commands gained a new
2165 "--message=" option which may be used to set a free-text
2166 wall message when shutting down or rebooting the
2167 system. This message is also logged, which is useful for
2168 figuring out the reason for a reboot or shutdown a
2171 * The "systemd-resolve-host" tool's -i switch now takes
2172 network interface numbers as alternative to interface names.
2174 * A new unit file setting for services has been introduced:
2175 UtmpMode= allows configuration of how precisely systemd
2176 handles utmp and wtmp entries for the service if this is
2177 enabled. This allows writing services that appear similar to
2178 user sessions in the output of the "w", "who", "last" and
2181 * systemd-resolved will now locally synthesize DNS resource
2182 records for the "localhost" and "gateway" domains as well as
2183 the local hostname. This should ensure that clients querying
2184 RRs via resolved will get similar results as those going via
2185 NSS, if nss-myhostname is enabled.
2187 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2188 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2189 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2190 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2191 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2192 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2193 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2194 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2195 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2196 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2197 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2198 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2200 — Berlin, 2015-08-27
2204 * The systemd-efi-boot-generator functionality was merged into
2205 systemd-gpt-auto-generator.
2207 * systemd-networkd now supports Group Policy for vxlan
2208 devices. It can be enabled via the new boolean configuration
2209 option called 'GroupPolicyExtension='.
2211 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2212 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2213 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2215 — Berlin, 2015-07-31
2219 * The python-systemd code has been removed from the systemd repository.
2220 A new repository has been created which accommodates the code from
2221 now on, and we kindly ask distributions to create a separate package
2222 for this: https://github.com/systemd/python-systemd
2224 * The systemd daemon will now reload its main configuration
2225 (/etc/systemd/system.conf) on daemon-reload.
2227 * sd-dhcp now exposes vendor specific extensions via
2228 sd_dhcp_lease_get_vendor_specific().
2230 * systemd-networkd gained a number of new configuration options.
2232 - A new boolean configuration option for TAP devices called
2233 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2234 device, thus allowing to send and receive GSO packets.
2236 - A new tunnel configuration option called 'CopyDSCP='.
2237 If enabled, the DSCP field of ip6 tunnels is copied into the
2238 decapsulated packet.
2240 - A set of boolean bridge configuration options were added.
2241 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2242 and 'UnicastFlood=' are now parsed by networkd and applied to the
2243 respective bridge link device via the respective IFLA_BRPORT_*
2246 - A new string configuration option to override the hostname sent
2247 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2248 is true, networkd will use the configured hostname instead of the
2249 system hostname when sending DHCP requests.
2251 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2252 networkd will configure the IPv6 flow-label of the tunnel device
2253 according to RFC2460.
2255 - The 'macvtap' virtual network devices are now supported, similar to
2256 the already supported 'macvlan' devices.
2258 * systemd-resolved now implements RFC5452 to improve resilience against
2259 cache poisoning. Additionally, source port randomization is enabled
2260 by default to further protect against DNS spoofing attacks.
2262 * nss-mymachines now supports translating UIDs and GIDs of running
2263 containers with user-namespaces enabled. If a container 'foo'
2264 translates a host uid 'UID' to the container uid 'TUID', then
2265 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2266 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2267 mapped as 'vg-foo-TGID'.
2269 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2270 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2271 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2272 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2273 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2274 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2275 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2276 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2277 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2278 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2280 — Berlin, 2015-07-29
2284 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2285 There are no known issues with current sysfs, and udev does not need
2286 or should be used to work around such bugs.
2288 * udev does no longer enable USB HID power management. Several reports
2289 indicate, that some devices cannot handle that setting.
2291 * The udev accelerometer helper was removed. The functionality
2292 is now fully included in iio-sensor-proxy. But this means,
2293 older iio-sensor-proxy versions will no longer provide
2294 accelerometer/orientation data with this systemd version.
2295 Please upgrade iio-sensor-proxy to version 1.0.
2297 * networkd gained a new configuration option IPv6PrivacyExtensions=
2298 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2299 for Stateless Address") on selected networks.
2301 * For the sake of fewer build-time dependencies and less code in the
2302 main repository, the python bindings are about to be removed in the
2303 next release. A new repository has been created which accommodates
2304 the code from now on, and we kindly ask distributions to create a
2305 separate package for this. The removal will take place in v223.
2307 https://github.com/systemd/python-systemd
2309 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2310 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2311 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2312 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2313 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2314 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2315 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2316 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2317 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2318 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2320 — Berlin, 2015-07-07
2324 * The sd-bus.h and sd-event.h APIs have now been declared
2325 stable and have been added to the official interface of
2326 libsystemd.so. sd-bus implements an alternative D-Bus client
2327 library, that is relatively easy to use, very efficient and
2328 supports both classic D-Bus as well as kdbus as transport
2329 backend. sd-event is a generic event loop abstraction that
2330 is built around Linux epoll, but adds features such as event
2331 prioritization or efficient timer handling. Both APIs are good
2332 choices for C programs looking for a bus and/or event loop
2333 implementation that is minimal and does not have to be
2334 portable to other kernels.
2336 * kdbus support is no longer compile-time optional. It is now
2337 always built-in. However, it can still be disabled at
2338 runtime using the kdbus=0 kernel command line setting, and
2339 that setting may be changed to default to off, by specifying
2340 --disable-kdbus at build-time. Note though that the kernel
2341 command line setting has no effect if the kdbus.ko kernel
2342 module is not installed, in which case kdbus is (obviously)
2343 also disabled. We encourage all downstream distributions to
2344 begin testing kdbus by adding it to the kernel images in the
2345 development distributions, and leaving kdbus support in
2348 * The minimal required util-linux version has been bumped to
2351 * Support for chkconfig (--enable-chkconfig) was removed in
2352 favor of calling an abstraction tool
2353 /lib/systemd/systemd-sysv-install. This needs to be
2354 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2355 in README for details.
2357 * If there's a systemd unit and a SysV init script for the
2358 same service name, and the user executes "systemctl enable"
2359 for it (or a related call), then this will now enable both
2360 (or execute the related operation on both), not just the
2363 * The libudev API documentation has been converted from gtkdoc
2366 * gudev has been removed from the systemd tree, it is now an
2369 * The systemd-cgtop tool learnt a new --raw switch to generate
2370 "raw" (machine parsable) output.
2372 * networkd's IPForwarding= .network file setting learnt the
2373 new setting "kernel", which ensures that networkd does not
2374 change the IP forwarding sysctl from the default kernel
2377 * The systemd-logind bus API now exposes a new boolean
2378 property "Docked" that reports whether logind considers the
2379 system "docked", i.e. connected to a docking station or not.
2381 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2382 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2383 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2384 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2385 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2386 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2387 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2388 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2389 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2390 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2391 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2392 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2393 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2394 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2395 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2396 Fink, Zbigniew Jędrzejewski-Szmek
2398 — Berlin, 2015-06-19
2402 * The gudev library has been extracted into a separate repository
2403 available at: https://git.gnome.org/browse/libgudev/
2404 It is now managed as part of the Gnome project. Distributions
2405 are recommended to pass --disable-gudev to systemd and use
2406 gudev from the Gnome project instead. gudev is still included
2407 in systemd, for now. It will be removed soon, though. Please
2408 also see the announcement-thread on systemd-devel:
2409 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2411 * systemd now exposes a CPUUsageNSec= property for each
2412 service unit on the bus, that contains the overall consumed
2413 CPU time of a service (the sum of what each process of the
2414 service consumed). This value is only available if
2415 CPUAccounting= is turned on for a service, and is then shown
2416 in the "systemctl status" output.
2418 * Support for configuring alternative mappings of the old SysV
2419 runlevels to systemd targets has been removed. They are now
2420 hardcoded in a way that runlevels 2, 3, 4 all map to
2421 multi-user.target and 5 to graphical.target (which
2422 previously was already the default behaviour).
2424 * The auto-mounter logic gained support for mount point
2425 expiry, using a new TimeoutIdleSec= setting in .automount
2426 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2428 * The EFI System Partition (ESP) as mounted to /boot by
2429 systemd-efi-boot-generator will now be unmounted
2430 automatically after 2 minutes of not being used. This should
2431 minimize the risk of ESP corruptions.
2433 * New /etc/fstab options x-systemd.requires= and
2434 x-systemd.requires-mounts-for= are now supported to express
2435 additional dependencies for mounts. This is useful for
2436 journalling file systems that support external journal
2437 devices or overlay file systems that require underlying file
2438 systems to be mounted.
2440 * systemd does not support direct live-upgrades (via systemctl
2441 daemon-reexec) from versions older than v44 anymore. As no
2442 distribution we are aware of shipped such old versions in a
2443 stable release this should not be problematic.
2445 * When systemd forks off a new per-connection service instance
2446 it will now set the $REMOTE_ADDR environment variable to the
2447 remote IP address, and $REMOTE_PORT environment variable to
2448 the remote IP port. This behaviour is similar to the
2449 corresponding environment variables defined by CGI.
2451 * systemd-networkd gained support for uplink failure
2452 detection. The BindCarrier= option allows binding interface
2453 configuration dynamically to the link sense of other
2454 interfaces. This is useful to achieve behaviour like in
2457 * systemd-networkd gained support for configuring the DHCP
2458 client identifier to use when requesting leases.
2460 * systemd-networkd now has a per-network UseNTP= option to
2461 configure whether NTP server information acquired via DHCP
2462 is passed on to services like systemd-timesyncd.
2464 * systemd-networkd gained support for vti6 tunnels.
2466 * Note that systemd-networkd manages the sysctl variable
2467 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2468 it is configured for since v219. The variable controls IP
2469 forwarding, and is a per-interface alternative to the global
2470 /proc/sys/net/ipv[46]/ip_forward. This setting is
2471 configurable in the IPForward= option, which defaults to
2472 "no". This means if networkd is used for an interface it is
2473 no longer sufficient to set the global sysctl option to turn
2474 on IP forwarding! Instead, the .network file option
2475 IPForward= needs to be turned on! Note that the
2476 implementation of this behaviour was broken in v219 and has
2479 * Many bonding and vxlan options are now configurable in
2482 * systemd-nspawn gained a new --property= setting to set unit
2483 properties for the container scope. This is useful for
2484 setting resource parameters (e.g. "CPUShares=500") on
2485 containers started from the command line.
2487 * systemd-nspawn gained a new --private-users= switch to make
2488 use of user namespacing available on recent Linux kernels.
2490 * systemd-nspawn may now be called as part of a shell pipeline
2491 in which case the pipes used for stdin and stdout are passed
2492 directly to the process invoked in the container, without
2493 indirection via a pseudo tty.
2495 * systemd-nspawn gained a new switch to control the UNIX
2496 signal to use when killing the init process of the container
2499 * systemd-nspawn gained a new --overlay= switch for mounting
2500 overlay file systems into the container using the new kernel
2503 * When a container image is imported via systemd-importd and
2504 the host file system is not btrfs, a loopback block device
2505 file is created in /var/lib/machines.raw with a btrfs file
2506 system inside. It is then mounted to /var/lib/machines to
2507 enable btrfs features for container management. The loopback
2508 file and btrfs file system is grown as needed when container
2509 images are imported via systemd-importd.
2511 * systemd-machined/systemd-importd gained support for btrfs
2512 quota, to enforce container disk space limits on disk. This
2513 is exposed in "machinectl set-limit".
2515 * systemd-importd now can import containers from local .tar,
2516 .raw and .qcow2 images, and export them to .tar and .raw. It
2517 can also import dkr v2 images now from the network (on top
2520 * systemd-importd gained support for verifying downloaded
2521 images with gpg2 (previously only gpg1 was supported).
2523 * systemd-machined, systemd-logind, systemd: most bus calls
2524 are now accessible to unprivileged processes via
2525 PolicyKit. Also, systemd-logind will now allow users to kill
2526 their own sessions without further privileges or
2529 * systemd-shutdownd has been removed. This service was
2530 previously responsible for implementing scheduled shutdowns
2531 as exposed in /usr/bin/shutdown's time parameter. This
2532 functionality has now been moved into systemd-logind and is
2533 accessible via a bus interface.
2535 * "systemctl reboot" gained a new switch --firmware-setup that
2536 can be used to reboot into the EFI firmware setup, if that
2537 is available. systemd-logind now exposes an API on the bus
2538 to trigger such reboots, in case graphical desktop UIs want
2539 to cover this functionality.
2541 * "systemctl enable", "systemctl disable" and "systemctl mask"
2542 now support a new "--now" switch. If specified the units
2543 that are enabled will also be started, and the ones
2544 disabled/masked also stopped.
2546 * The Gummiboot EFI boot loader tool has been merged into
2547 systemd, and renamed to "systemd-boot". The bootctl tool has been
2548 updated to support systemd-boot.
2550 * An EFI kernel stub has been added that may be used to create
2551 kernel EFI binaries that contain not only the actual kernel,
2552 but also an initrd, boot splash, command line and OS release
2553 information. This combined binary can then be signed as a
2554 single image, so that the firmware can verify it all in one
2555 step. systemd-boot has special support for EFI binaries created
2556 like this and can extract OS release information from them
2557 and show them in the boot menu. This functionality is useful
2558 to implement cryptographically verified boot schemes.
2560 * Optional support has been added to systemd-fsck to pass
2561 fsck's progress report to an AF_UNIX socket in the file
2564 * udev will no longer create device symlinks for all block
2565 devices by default. A blacklist for excluding special block
2566 devices from this logic has been turned into a whitelist
2567 that requires picking block devices explicitly that require
2570 * A new (currently still internal) API sd-device.h has been
2571 added to libsystemd. This modernized API is supposed to
2572 replace libudev eventually. In fact, already much of libudev
2573 is now just a wrapper around sd-device.h.
2575 * A new hwdb database for storing metadata about pointing
2576 stick devices has been added.
2578 * systemd-tmpfiles gained support for setting file attributes
2579 similar to the "chattr" tool with new 'h' and 'H' lines.
2581 * systemd-journald will no longer unconditionally set the
2582 btrfs NOCOW flag on new journal files. This is instead done
2583 with tmpfiles snippet using the new 'h' line type. This
2584 allows easy disabling of this logic, by masking the
2585 journal-nocow.conf tmpfiles file.
2587 * systemd-journald will now translate audit message types to
2588 human readable identifiers when writing them to the
2589 journal. This should improve readability of audit messages.
2591 * The LUKS logic gained support for the offset= and skip=
2592 options in /etc/crypttab, as previously implemented by
2595 * /usr/lib/os-release gained a new optional field VARIANT= for
2596 distributions that support multiple variants (such as a
2597 desktop edition, a server edition, ...)
2599 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2600 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2601 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2602 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2603 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2604 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2605 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2606 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2607 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2608 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2609 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2610 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2611 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2612 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2613 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2614 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2615 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2616 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2617 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2618 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2619 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2620 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2621 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2622 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2623 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2624 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2625 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2627 — Berlin, 2015-05-22
2631 * Introduce a new API "sd-hwdb.h" for querying the hardware
2632 metadata database. With this minimal interface one can query
2633 and enumerate the udev hwdb, decoupled from the old libudev
2634 library. libudev's interface for this is now only a wrapper
2635 around sd-hwdb. A new tool systemd-hwdb has been added to
2636 interface with and update the database.
2638 * When any of systemd's tools copies files (for example due to
2639 tmpfiles' C lines) a btrfs reflink will attempted first,
2640 before bytewise copying is done.
2642 * systemd-nspawn gained a new --ephemeral switch. When
2643 specified a btrfs snapshot is taken of the container's root
2644 directory, and immediately removed when the container
2645 terminates again. Thus, a container can be started whose
2646 changes never alter the container's root directory, and are
2647 lost on container termination. This switch can also be used
2648 for starting a container off the root file system of the
2649 host without affecting the host OS. This switch is only
2650 available on btrfs file systems.
2652 * systemd-nspawn gained a new --template= switch. It takes the
2653 path to a container tree to use as template for the tree
2654 specified via --directory=, should that directory be
2655 missing. This allows instantiating containers dynamically,
2656 on first run. This switch is only available on btrfs file
2659 * When a .mount unit refers to a mount point on which multiple
2660 mounts are stacked, and the .mount unit is stopped all of
2661 the stacked mount points will now be unmounted until no
2662 mount point remains.
2664 * systemd now has an explicit notion of supported and
2665 unsupported unit types. Jobs enqueued for unsupported unit
2666 types will now fail with an "unsupported" error code. More
2667 specifically .swap, .automount and .device units are not
2668 supported in containers, .busname units are not supported on
2669 non-kdbus systems. .swap and .automount are also not
2670 supported if their respective kernel compile time options
2673 * machinectl gained support for two new "copy-from" and
2674 "copy-to" commands for copying files from a running
2675 container to the host or vice versa.
2677 * machinectl gained support for a new "bind" command to bind
2678 mount host directories into local containers. This is
2679 currently only supported for nspawn containers.
2681 * networkd gained support for configuring bridge forwarding
2682 database entries (fdb) from .network files.
2684 * A new tiny daemon "systemd-importd" has been added that can
2685 download container images in tar, raw, qcow2 or dkr formats,
2686 and make them available locally in /var/lib/machines, so
2687 that they can run as nspawn containers. The daemon can GPG
2688 verify the downloads (not supported for dkr, since it has no
2689 provisions for verifying downloads). It will transparently
2690 decompress bz2, xz, gzip compressed downloads if necessary,
2691 and restore sparse files on disk. The daemon uses privilege
2692 separation to ensure the actual download logic runs with
2693 fewer privileges than the daemon itself. machinectl has
2694 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2695 make the functionality of importd available to the
2696 user. With this in place the Fedora and Ubuntu "Cloud"
2697 images can be downloaded and booted as containers unmodified
2698 (the Fedora images lack the appropriate GPG signature files
2699 currently, so they cannot be verified, but this will change
2700 soon, hopefully). Note that downloading images is currently
2701 only fully supported on btrfs.
2703 * machinectl is now able to list container images found in
2704 /var/lib/machines, along with some metadata about sizes of
2705 disk and similar. If the directory is located on btrfs and
2706 quota is enabled, this includes quota display. A new command
2707 "image-status" has been added that shows additional
2708 information about images.
2710 * machinectl is now able to clone container images
2711 efficiently, if the underlying file system (btrfs) supports
2712 it, with the new "machinectl clone" command. It also
2713 gained commands for renaming and removing images, as well as
2714 marking them read-only or read-write (supported also on
2715 legacy file systems).
2717 * networkd gained support for collecting LLDP network
2718 announcements, from hardware that supports this. This is
2719 shown in networkctl output.
2721 * systemd-run gained support for a new -t (--pty) switch for
2722 invoking a binary on a pty whose input and output is
2723 connected to the invoking terminal. This allows executing
2724 processes as system services while interactively
2725 communicating with them via the terminal. Most interestingly
2726 this is supported across container boundaries. Invoking
2727 "systemd-run -t /bin/bash" is an alternative to running a
2728 full login session, the difference being that the former
2729 will not register a session, nor go through the PAM session
2732 * tmpfiles gained support for a new "v" line type for creating
2733 btrfs subvolumes. If the underlying file system is a legacy
2734 file system, this automatically degrades to creating a
2735 normal directory. Among others /var/lib/machines is now
2736 created like this at boot, should it be missing.
2738 * The directory /var/lib/containers/ has been deprecated and
2739 been replaced by /var/lib/machines. The term "machines" has
2740 been used in the systemd context as generic term for both
2741 VMs and containers, and hence appears more appropriate for
2742 this, as the directory can also contain raw images bootable
2745 * systemd-nspawn when invoked with -M but without --directory=
2746 or --image= is now capable of searching for the container
2747 root directory, subvolume or disk image automatically, in
2748 /var/lib/machines. systemd-nspawn@.service has been updated
2749 to make use of this, thus allowing it to be used for raw
2752 * A new machines.target unit has been introduced that is
2753 supposed to group all containers/VMs invoked as services on
2754 the system. systemd-nspawn@.service has been updated to
2755 integrate with that.
2757 * machinectl gained a new "start" command, for invoking a
2758 container as a service. "machinectl start foo" is mostly
2759 equivalent to "systemctl start systemd-nspawn@foo.service",
2760 but handles escaping in a nicer way.
2762 * systemd-nspawn will now mount most of the cgroupfs tree
2763 read-only into each container, with the exception of the
2764 container's own subtree in the name=systemd hierarchy.
2766 * journald now sets the special FS_NOCOW file flag for its
2767 journal files. This should improve performance on btrfs, by
2768 avoiding heavy fragmentation when journald's write-pattern
2769 is used on COW file systems. It degrades btrfs' data
2770 integrity guarantees for the files to the same levels as for
2771 ext3/ext4 however. This should be OK though as journald does
2772 its own data integrity checks and all its objects are
2773 checksummed on disk. Also, journald should handle btrfs disk
2774 full events a lot more gracefully now, by processing SIGBUS
2775 errors, and not relying on fallocate() anymore.
2777 * When journald detects that journal files it is writing to
2778 have been deleted it will immediately start new journal
2781 * systemd now provides a way to store file descriptors
2782 per-service in PID 1. This is useful for daemons to ensure
2783 that fds they require are not lost during a daemon
2784 restart. The fds are passed to the daemon on the next
2785 invocation in the same way socket activation fds are
2786 passed. This is now used by journald to ensure that the
2787 various sockets connected to all the system's stdout/stderr
2788 are not lost when journald is restarted. File descriptors
2789 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2790 an extension to sd_notify(). Note that a limit is enforced
2791 on the number of fds a service can store in PID 1, and it
2792 defaults to 0, so that no fds may be stored, unless this is
2793 explicitly turned on.
2795 * The default TERM variable to use for units connected to a
2796 terminal, when no other value is explicitly is set is now
2797 vt220 rather than vt102. This should be fairly safe still,
2798 but allows PgUp/PgDn work.
2800 * The /etc/crypttab option header= as known from Debian is now
2803 * "loginctl user-status" and "loginctl session-status" will
2804 now show the last 10 lines of log messages of the
2805 user/session following the status output. Similar,
2806 "machinectl status" will show the last 10 log lines
2807 associated with a virtual machine or container
2808 service. (Note that this is usually not the log messages
2809 done in the VM/container itself, but simply what the
2810 container manager logs. For nspawn this includes all console
2813 * "loginctl session-status" without further argument will now
2814 show the status of the session of the caller. Similar,
2815 "lock-session", "unlock-session", "activate",
2816 "enable-linger", "disable-linger" may now be called without
2817 session/user parameter in which case they apply to the
2818 caller's session/user.
2820 * An X11 session scriptlet is now shipped that uploads
2821 $DISPLAY and $XAUTHORITY into the environment of the systemd
2822 --user daemon if a session begins. This should improve
2823 compatibility with X11 enabled applications run as systemd
2826 * Generators are now subject to masking via /etc and /run, the
2827 same way as unit files.
2829 * networkd .network files gained support for configuring
2830 per-link IPv4/IPv6 packet forwarding as well as IPv4
2831 masquerading. This is by default turned on for veth links to
2832 containers, as registered by systemd-nspawn. This means that
2833 nspawn containers run with --network-veth will now get
2834 automatic routed access to the host's networks without any
2835 further configuration or setup, as long as networkd runs on
2838 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2839 or UDP posts of a container on the host. With this in place
2840 it is possible to run containers with private veth links
2841 (--network-veth), and have their functionality exposed on
2842 the host as if their services were running directly on the
2845 * systemd-nspawn's --network-veth switch now gained a short
2846 version "-n", since with the changes above it is now truly
2847 useful out-of-the-box. The systemd-nspawn@.service has been
2848 updated to make use of it too by default.
2850 * systemd-nspawn will now maintain a per-image R/W lock, to
2851 ensure that the same image is not started more than once
2852 writable. (It's OK to run an image multiple times
2853 simultaneously in read-only mode.)
2855 * systemd-nspawn's --image= option is now capable of
2856 dissecting and booting MBR and GPT disk images that contain
2857 only a single active Linux partition. Previously it
2858 supported only GPT disk images with proper GPT type
2859 IDs. This allows running cloud images from major
2860 distributions directly with systemd-nspawn, without
2863 * In addition to collecting mouse dpi data in the udev
2864 hardware database, there's now support for collecting angle
2865 information for mouse scroll wheels. The database is
2866 supposed to guarantee similar scrolling behavior on mice
2867 that it knows about. There's also support for collecting
2868 information about Touchpad types.
2870 * udev's input_id built-in will now also collect touch screen
2871 dimension data and attach it to probed devices.
2873 * /etc/os-release gained support for a Distribution Privacy
2876 * networkd gained support for creating "ipvlan", "gretap",
2877 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2879 * systemd-tmpfiles gained support for "a" lines for setting
2882 * systemd-nspawn will now mount /tmp in the container to
2883 tmpfs, automatically.
2885 * systemd now exposes the memory.usage_in_bytes cgroup
2886 attribute and shows it for each service in the "systemctl
2887 status" output, if available.
2889 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2890 immediate reboot is triggered. This useful if shutdown is
2891 hung and is unable to complete, to expedite the
2892 operation. Note that this kind of reboot will still unmount
2893 all file systems, and hence should not result in fsck being
2896 * A .device unit for an optical block device will now be
2897 considered active only when a medium is in the drive. Also,
2898 mount units are now bound to their backing devices thus
2899 triggering automatic unmounting when devices become
2900 unavailable. With this in place systemd will now
2901 automatically unmount left-over mounts when a CD-ROM is
2902 ejected or an USB stick is yanked from the system.
2904 * networkd-wait-online now has support for waiting for
2905 specific interfaces only (with globbing), and for giving up
2906 after a configurable timeout.
2908 * networkd now exits when idle. It will be automatically
2909 restarted as soon as interfaces show up, are removed or
2910 change state. networkd will stay around as long as there is
2911 at least one DHCP state machine or similar around, that keep
2914 * networkd may now configure IPv6 link-local addressing in
2915 addition to IPv4 link-local addressing.
2917 * The IPv6 "token" for use in SLAAC may now be configured for
2918 each .network interface in networkd.
2920 * Routes configured with networkd may now be assigned a scope
2923 * networkd's [Match] sections now support globbing and lists
2924 of multiple space-separated matches per item.
2926 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2927 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2928 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2929 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2930 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2931 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2932 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2933 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2934 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2935 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2936 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2937 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2938 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2939 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2940 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2941 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2942 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2943 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2944 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2945 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2946 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2947 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2948 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2949 Hoffmann, Zbigniew Jędrzejewski-Szmek
2951 — Berlin, 2015-02-16
2955 * When querying unit file enablement status (for example via
2956 "systemctl is-enabled"), a new state "indirect" is now known
2957 which indicates that a unit might not be enabled itself, but
2958 another unit listed in its Also= setting might be.
2960 * Similar to the various existing ConditionXYZ= settings for
2961 units, there are now matching AssertXYZ= settings. While
2962 failing conditions cause a unit to be skipped, but its job
2963 to succeed, failing assertions declared like this will cause
2964 a unit start operation and its job to fail.
2966 * hostnamed now knows a new chassis type "embedded".
2968 * systemctl gained a new "edit" command. When used on a unit
2969 file, this allows extending unit files with .d/ drop-in
2970 configuration snippets or editing the full file (after
2971 copying it from /usr/lib to /etc). This will invoke the
2972 user's editor (as configured with $EDITOR), and reload the
2973 modified configuration after editing.
2975 * "systemctl status" now shows the suggested enablement state
2976 for a unit, as declared in the (usually vendor-supplied)
2977 system preset files.
2979 * nss-myhostname will now resolve the single-label host name
2980 "gateway" to the locally configured default IP routing
2981 gateways, ordered by their metrics. This assigns a stable
2982 name to the used gateways, regardless which ones are
2983 currently configured. Note that the name will only be
2984 resolved after all other name sources (if nss-myhostname is
2985 configured properly) and should hence not negatively impact
2986 systems that use the single-label host name "gateway" in
2989 * systemd-inhibit now allows filtering by mode when listing
2992 * Scope and service units gained a new "Delegate" boolean
2993 property, which, when set, allows processes running inside the
2994 unit to further partition resources. This is primarily
2995 useful for systemd user instances as well as container
2998 * journald will now pick up audit messages directly from
2999 the kernel, and log them like any other log message. The
3000 audit fields are split up and fully indexed. This means that
3001 journalctl in many ways is now a (nicer!) alternative to
3002 ausearch, the traditional audit client. Note that this
3003 implements only a minimal audit client. If you want the
3004 special audit modes like reboot-on-log-overflow, please use
3005 the traditional auditd instead, which can be used in
3006 parallel to journald.
3008 * The ConditionSecurity= unit file option now understands the
3009 special string "audit" to check whether auditing is
3012 * journalctl gained two new commands --vacuum-size= and
3013 --vacuum-time= to delete old journal files until the
3014 remaining ones take up no more than the specified size on disk,
3015 or are not older than the specified time.
3017 * A new, native PPPoE library has been added to sd-network,
3018 systemd's library of light-weight networking protocols. This
3019 library will be used in a future version of networkd to
3020 enable PPPoE communication without an external pppd daemon.
3022 * The busctl tool now understands a new "capture" verb that
3023 works similar to "monitor", but writes a packet capture
3024 trace to STDOUT that can be redirected to a file which is
3025 compatible with libcap's capture file format. This can then
3026 be loaded in Wireshark and similar tools to inspect bus
3029 * The busctl tool now understands a new "tree" verb that shows
3030 the object trees of a specific service on the bus, or of all
3033 * The busctl tool now understands a new "introspect" verb that
3034 shows all interfaces and members of objects on the bus,
3035 including their signature and values. This is particularly
3036 useful to get more information about bus objects shown by
3037 the new "busctl tree" command.
3039 * The busctl tool now understands new verbs "call",
3040 "set-property" and "get-property" for invoking bus method
3041 calls, setting and getting bus object properties in a
3044 * busctl gained a new --augment-creds= argument that controls
3045 whether the tool shall augment credential information it
3046 gets from the bus with data from /proc, in a possibly
3049 * nspawn's --link-journal= switch gained two new values
3050 "try-guest" and "try-host" that work like "guest" and
3051 "host", but do not fail if the host has no persistent
3052 journalling enabled. -j is now equivalent to
3053 --link-journal=try-guest.
3055 * macvlan network devices created by nspawn will now have
3056 stable MAC addresses.
3058 * A new SmackProcessLabel= unit setting has been added, which
3059 controls the SMACK security label processes forked off by
3060 the respective unit shall use.
3062 * If compiled with --enable-xkbcommon, systemd-localed will
3063 verify x11 keymap settings by compiling the given keymap. It
3064 will spew out warnings if the compilation fails. This
3065 requires libxkbcommon to be installed.
3067 * When a coredump is collected, a larger number of metadata
3068 fields is now collected and included in the journal records
3069 created for it. More specifically, control group membership,
3070 environment variables, memory maps, working directory,
3071 chroot directory, /proc/$PID/status, and a list of open file
3072 descriptors is now stored in the log entry.
3074 * The udev hwdb now contains DPI information for mice. For
3077 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3079 * All systemd programs that read standalone configuration
3080 files in /etc now also support a corresponding series of
3081 .conf.d configuration directories in /etc/, /run/,
3082 /usr/local/lib/, /usr/lib/, and (if configured with
3083 --enable-split-usr) /lib/. In particular, the following
3084 configuration files now have corresponding configuration
3085 directories: system.conf user.conf, logind.conf,
3086 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3087 resolved.conf, timesyncd.conf, journal-remote.conf, and
3088 journal-upload.conf. Note that distributions should use the
3089 configuration directories in /usr/lib/; the directories in
3090 /etc/ are reserved for the system administrator.
3092 * systemd-rfkill will no longer take the rfkill device name
3093 into account when storing rfkill state on disk, as the name
3094 might be dynamically assigned and not stable. Instead, the
3095 ID_PATH udev variable combined with the rfkill type (wlan,
3096 bluetooth, ...) is used.
3098 * A new service systemd-machine-id-commit.service has been
3099 added. When used on systems where /etc is read-only during
3100 boot, and /etc/machine-id is not initialized (but an empty
3101 file), this service will copy the temporary machine ID
3102 created as replacement into /etc after the system is fully
3103 booted up. This is useful for systems that are freshly
3104 installed with a non-initialized machine ID, but should get
3105 a fixed machine ID for subsequent boots.
3107 * networkd's .netdev files now provide a large set of
3108 configuration parameters for VXLAN devices. Similarly, the
3109 bridge port cost parameter is now configurable in .network
3110 files. There's also new support for configuring IP source
3111 routing. networkd .link files gained support for a new
3112 OriginalName= match that is useful to match against the
3113 original interface name the kernel assigned. .network files
3114 may include MTU= and MACAddress= fields for altering the MTU
3115 and MAC address while being connected to a specific network
3118 * The LUKS logic gained supported for configuring
3119 UUID-specific key files. There's also new support for naming
3120 LUKS device from the kernel command line, using the new
3121 luks.name= argument.
3123 * Timer units may now be transiently created via the bus API
3124 (this was previously already available for scope and service
3125 units). In addition it is now possible to create multiple
3126 transient units at the same time with a single bus call. The
3127 "systemd-run" tool has been updated to make use of this for
3128 running commands on a specified time, in at(1)-style.
3130 * tmpfiles gained support for "t" lines, for assigning
3131 extended attributes to files. Among other uses this may be
3132 used to assign SMACK labels to files.
3134 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3135 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3136 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3137 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3138 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3139 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3140 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3141 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3142 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3143 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3144 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3145 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3146 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3147 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3148 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3149 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3150 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3151 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3153 — Berlin, 2014-12-10
3157 * journalctl gained the new options -t/--identifier= to match
3158 on the syslog identifier (aka "tag"), as well as --utc to
3159 show log timestamps in the UTC timezone. journalctl now also
3160 accepts -n/--lines=all to disable line capping in a pager.
3162 * journalctl gained a new switch, --flush, that synchronously
3163 flushes logs from /run/log/journal to /var/log/journal if
3164 persistent storage is enabled. systemd-journal-flush.service
3165 now waits until the operation is complete.
3167 * Services can notify the manager before they start a reload
3168 (by sending RELOADING=1) or shutdown (by sending
3169 STOPPING=1). This allows the manager to track and show the
3170 internal state of daemons and closes a race condition when
3171 the process is still running but has closed its D-Bus
3174 * Services with Type=oneshot do not have to have any ExecStart
3177 * User units are now loaded also from
3178 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3179 /run/systemd/user directory that was already previously
3180 supported, but is under the control of the user.
3182 * Job timeouts (i.e. time-outs on the time a job that is
3183 queued stays in the run queue) can now optionally result in
3184 immediate reboot or power-off actions (JobTimeoutAction= and
3185 JobTimeoutRebootArgument=). This is useful on ".target"
3186 units, to limit the maximum time a target remains
3187 undispatched in the run queue, and to trigger an emergency
3188 operation in such a case. This is now used by default to
3189 turn off the system if boot-up (as defined by everything in
3190 basic.target) hangs and does not complete for at least
3191 15min. Also, if power-off or reboot hang for at least 30min
3192 an immediate power-off/reboot operation is triggered. This
3193 functionality is particularly useful to increase reliability
3194 on embedded devices, but also on laptops which might
3195 accidentally get powered on when carried in a backpack and
3196 whose boot stays stuck in a hard disk encryption passphrase
3199 * systemd-logind can be configured to also handle lid switch
3200 events even when the machine is docked or multiple displays
3201 are attached (HandleLidSwitchDocked= option).
3203 * A helper binary and a service have been added which can be
3204 used to resume from hibernation in the initramfs. A
3205 generator will parse the resume= option on the kernel
3206 command line to trigger resume.
3208 * A user console daemon systemd-consoled has been
3209 added. Currently, it is a preview, and will so far open a
3210 single terminal on each session of the user marked as
3211 Desktop=systemd-console.
3213 * Route metrics can be specified for DHCP routes added by
3216 * The SELinux context of socket-activated services can be set
3217 from the information provided by the networking stack
3218 (SELinuxContextFromNet= option).
3220 * Userspace firmware loading support has been removed and
3221 the minimum supported kernel version is thus bumped to 3.7.
3223 * Timeout for udev workers has been increased from 1 to 3
3224 minutes, but a warning will be printed after 1 minute to
3225 help diagnose kernel modules that take a long time to load.
3227 * Udev rules can now remove tags on devices with TAG-="foobar".
3229 * systemd's readahead implementation has been removed. In many
3230 circumstances it didn't give expected benefits even for
3231 rotational disk drives and was becoming less relevant in the
3232 age of SSDs. As none of the developers has been using
3233 rotating media anymore, and nobody stepped up to actively
3234 maintain this component of systemd it has now been removed.
3236 * Swap units can use Options= to specify discard options.
3237 Discard options specified for swaps in /etc/fstab are now
3240 * Docker containers are now detected as a separate type of
3243 * The Password Agent protocol gained support for queries where
3244 the user input is shown, useful e.g. for user names.
3245 systemd-ask-password gained a new --echo option to turn that
3248 * The default sysctl.d/ snippets will now set:
3250 net.core.default_qdisc = fq_codel
3252 This selects Fair Queuing Controlled Delay as the default
3253 queuing discipline for network interfaces. fq_codel helps
3254 fight the network bufferbloat problem. It is believed to be
3255 a good default with no tuning required for most workloads.
3256 Downstream distributions may override this choice. On 10Gbit
3257 servers that do not do forwarding, "fq" may perform better.
3258 Systems without a good clocksource should use "pfifo_fast".
3260 * If kdbus is enabled during build a new option BusPolicy= is
3261 available for service units, that allows locking all service
3262 processes into a stricter bus policy, in order to limit
3263 access to various bus services, or even hide most of them
3264 from the service's view entirely.
3266 * networkctl will now show the .network and .link file
3267 networkd has applied to a specific interface.
3269 * sd-login gained a new API call sd_session_get_desktop() to
3270 query which desktop environment has been selected for a
3273 * UNIX utmp support is now compile-time optional to support
3274 legacy-free systems.
3276 * systemctl gained two new commands "add-wants" and
3277 "add-requires" for pulling in units from specific targets
3280 * If the word "rescue" is specified on the kernel command line
3281 the system will now boot into rescue mode (aka
3282 rescue.target), which was previously available only by
3283 specifying "1" or "systemd.unit=rescue.target" on the kernel
3284 command line. This new kernel command line option nicely
3285 mirrors the already existing "emergency" kernel command line
3288 * New kernel command line options mount.usr=, mount.usrflags=,
3289 mount.usrfstype= have been added that match root=, rootflags=,
3290 rootfstype= but allow mounting a specific file system to
3293 * The $NOTIFY_SOCKET is now also passed to control processes of
3294 services, not only the main process.
3296 * This version reenables support for fsck's -l switch. This
3297 means at least version v2.25 of util-linux is required for
3298 operation, otherwise dead-locks on device nodes may
3299 occur. Again: you need to update util-linux to at least
3300 v2.25 when updating systemd to v217.
3302 * The "multi-seat-x" tool has been removed from systemd, as
3303 its functionality has been integrated into X servers 1.16,
3304 and the tool is hence redundant. It is recommended to update
3305 display managers invoking this tool to simply invoke X
3306 directly from now on, again.
3308 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3309 message flag has been added for all of systemd's PolicyKit
3310 authenticated method calls has been added. In particular
3311 this now allows optional interactive authorization via
3312 PolicyKit for many of PID1's privileged operations such as
3313 unit file enabling and disabling.
3315 * "udevadm hwdb --update" learnt a new switch "--usr" for
3316 placing the rebuilt hardware database in /usr instead of
3317 /etc. When used only hardware database entries stored in
3318 /usr will be used, and any user database entries in /etc are
3319 ignored. This functionality is useful for vendors to ship a
3320 pre-built database on systems where local configuration is
3321 unnecessary or unlikely.
3323 * Calendar time specifications in .timer units now also
3324 understand the strings "semi-annually", "quarterly" and
3325 "minutely" as shortcuts (in addition to the preexisting
3326 "anually", "hourly", ...).
3328 * systemd-tmpfiles will now correctly create files in /dev
3329 at boot which are marked for creation only at boot. It is
3330 recommended to always create static device nodes with 'c!'
3331 and 'b!', so that they are created only at boot and not
3332 overwritten at runtime.
3334 * When the watchdog logic is used for a service (WatchdogSec=)
3335 and the watchdog timeout is hit the service will now be
3336 terminated with SIGABRT (instead of just SIGTERM), in order
3337 to make sure a proper coredump and backtrace is
3338 generated. This ensures that hanging services will result in
3339 similar coredump/backtrace behaviour as services that hit a
3342 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3343 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3344 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3345 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3346 Herrmann, David Sommerseth, David Strauss, Emil Renner
3347 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3348 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3349 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3350 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3351 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3352 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3353 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3354 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3355 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3356 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3357 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3358 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3359 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3360 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3361 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3362 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3365 — Berlin, 2014-10-28
3369 * timedated no longer reads NTP implementation unit names from
3370 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3371 implementations should add a
3373 Conflicts=systemd-timesyncd.service
3375 to their unit files to take over and replace systemd's NTP
3376 default functionality.
3378 * systemd-sysusers gained a new line type "r" for configuring
3379 which UID/GID ranges to allocate system users/groups
3380 from. Lines of type "u" may now add an additional column
3381 that specifies the home directory for the system user to be
3382 created. Also, systemd-sysusers may now optionally read user
3383 information from STDIN instead of a file. This is useful for
3384 invoking it from RPM preinst scriptlets that need to create
3385 users before the first RPM file is installed since these
3386 files might need to be owned by them. A new
3387 %sysusers_create_inline RPM macro has been introduced to do
3388 just that. systemd-sysusers now updates the shadow files as
3389 well as the user/group databases, which should enhance
3390 compatibility with certain tools like grpck.
3392 * A number of bus APIs of PID 1 now optionally consult
3393 PolicyKit to permit access for otherwise unprivileged
3394 clients under certain conditions. Note that this currently
3395 doesn't support interactive authentication yet, but this is
3396 expected to be added eventually, too.
3398 * /etc/machine-info now has new fields for configuring the
3399 deployment environment of the machine, as well as the
3400 location of the machine. hostnamectl has been updated with
3401 new command to update these fields.
3403 * systemd-timesyncd has been updated to automatically acquire
3404 NTP server information from systemd-networkd, which might
3405 have been discovered via DHCP.
3407 * systemd-resolved now includes a caching DNS stub resolver
3408 and a complete LLMNR name resolution implementation. A new
3409 NSS module "nss-resolve" has been added which can be used
3410 instead of glibc's own "nss-dns" to resolve hostnames via
3411 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3412 be resolved via systemd-resolved D-Bus APIs. In contrast to
3413 the glibc internal resolver systemd-resolved is aware of
3414 multi-homed system, and keeps DNS server and caches separate
3415 and per-interface. Queries are sent simultaneously on all
3416 interfaces that have DNS servers configured, in order to
3417 properly handle VPNs and local LANs which might resolve
3418 separate sets of domain names. systemd-resolved may acquire
3419 DNS server information from systemd-networkd automatically,
3420 which in turn might have discovered them via DHCP. A tool
3421 "systemd-resolve-host" has been added that may be used to
3422 query the DNS logic in resolved. systemd-resolved implements
3423 IDNA and automatically uses IDNA or UTF-8 encoding depending
3424 on whether classic DNS or LLMNR is used as transport. In the
3425 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3426 implementation to systemd-resolved.
3428 * A new NSS module nss-mymachines has been added, that
3429 automatically resolves the names of all local registered
3430 containers to their respective IP addresses.
3432 * A new client tool "networkctl" for systemd-networkd has been
3433 added. It currently is entirely passive and will query
3434 networking configuration from udev, rtnetlink and networkd,
3435 and present it to the user in a very friendly
3436 way. Eventually, we hope to extend it to become a full
3437 control utility for networkd.
3439 * .socket units gained a new DeferAcceptSec= setting that
3440 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3441 TCP. Similarly, support for controlling TCP keep-alive
3442 settings has been added (KeepAliveTimeSec=,
3443 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3444 turning off Nagle's algorithm on TCP has been added
3447 * logind learned a new session type "web", for use in projects
3448 like Cockpit which register web clients as PAM sessions.
3450 * timer units with at least one OnCalendar= setting will now
3451 be started only after time-sync.target has been
3452 reached. This way they will not elapse before the system
3453 clock has been corrected by a local NTP client or
3454 similar. This is particular useful on RTC-less embedded
3455 machines, that come up with an invalid system clock.
3457 * systemd-nspawn's --network-veth= switch should now result in
3458 stable MAC addresses for both the outer and the inner side
3461 * systemd-nspawn gained a new --volatile= switch for running
3462 container instances with /etc or /var unpopulated.
3464 * The kdbus client code has been updated to use the new Linux
3465 3.17 memfd subsystem instead of the old kdbus-specific one.
3467 * systemd-networkd's DHCP client and server now support
3468 FORCERENEW. There are also new configuration options to
3469 configure the vendor client identifier and broadcast mode
3472 * systemd will no longer inform the kernel about the current
3473 timezone, as this is necessarily incorrect and racy as the
3474 kernel has no understanding of DST and similar
3475 concepts. This hence means FAT timestamps will be always
3476 considered UTC, similar to what Android is already
3477 doing. Also, when the RTC is configured to the local time
3478 (rather than UTC) systemd will never synchronize back to it,
3479 as this might confuse Windows at a later boot.
3481 * systemd-analyze gained a new command "verify" for offline
3482 validation of unit files.
3484 * systemd-networkd gained support for a couple of additional
3485 settings for bonding networking setups. Also, the metric for
3486 statically configured routes may now be configured. For
3487 network interfaces where this is appropriate the peer IP
3488 address may now be configured.
3490 * systemd-networkd's DHCP client will no longer request
3491 broadcasting by default, as this tripped up some networks.
3492 For hardware where broadcast is required the feature should
3493 be switched back on using RequestBroadcast=yes.
3495 * systemd-networkd will now set up IPv4LL addresses (when
3496 enabled) even if DHCP is configured successfully.
3498 * udev will now default to respect network device names given
3499 by the kernel when the kernel indicates that these are
3500 predictable. This behavior can be tweaked by changing
3501 NamePolicy= in the relevant .link file.
3503 * A new library systemd-terminal has been added that
3504 implements full TTY stream parsing and rendering. This
3505 library is supposed to be used later on for implementing a
3506 full userspace VT subsystem, replacing the current kernel
3509 * A new tool systemd-journal-upload has been added to push
3510 journal data to a remote system running
3511 systemd-journal-remote.
3513 * journald will no longer forward all local data to another
3514 running syslog daemon. This change has been made because
3515 rsyslog (which appears to be the most commonly used syslog
3516 implementation these days) no longer makes use of this, and
3517 instead pulls the data out of the journal on its own. Since
3518 forwarding the messages to a non-existent syslog server is
3519 more expensive than we assumed we have now turned this
3520 off. If you run a syslog server that is not a recent rsyslog
3521 version, you have to turn this option on again
3522 (ForwardToSyslog= in journald.conf).
3524 * journald now optionally supports the LZ4 compressor for
3525 larger journal fields. This compressor should perform much
3526 better than XZ which was the previous default.
3528 * machinectl now shows the IP addresses of local containers,
3529 if it knows them, plus the interface name of the container.
3531 * A new tool "systemd-escape" has been added that makes it
3532 easy to escape strings to build unit names and similar.
3534 * sd_notify() messages may now include a new ERRNO= field
3535 which is parsed and collected by systemd and shown among the
3536 "systemctl status" output for a service.
3538 * A new component "systemd-firstboot" has been added that
3539 queries the most basic systemd information (timezone,
3540 hostname, root password) interactively on first
3541 boot. Alternatively it may also be used to provision these
3542 things offline on OS images installed into directories.
3544 * The default sysctl.d/ snippets will now set
3546 net.ipv4.conf.default.promote_secondaries=1
3548 This has the benefit of no flushing secondary IP addresses
3549 when primary addresses are removed.
3551 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3552 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3553 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3554 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3555 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3556 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3557 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3558 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3559 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3560 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3561 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3562 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3563 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3564 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3565 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3567 — Berlin, 2014-08-19
3571 * A new tool systemd-sysusers has been added. This tool
3572 creates system users and groups in /etc/passwd and
3573 /etc/group, based on static declarative system user/group
3574 definitions in /usr/lib/sysusers.d/. This is useful to
3575 enable factory resets and volatile systems that boot up with
3576 an empty /etc directory, and thus need system users and
3577 groups created during early boot. systemd now also ships
3578 with two default sysusers.d/ files for the most basic
3579 users and groups systemd and the core operating system
3582 * A new tmpfiles snippet has been added that rebuilds the
3583 essential files in /etc on boot, should they be missing.
3585 * A directive for ensuring automatic clean-up of
3586 /var/cache/man/ has been removed from the default
3587 configuration. This line should now be shipped by the man
3588 implementation. The necessary change has been made to the
3589 man-db implementation. Note that you need to update your man
3590 implementation to one that ships this line, otherwise no
3591 automatic clean-up of /var/cache/man will take place.
3593 * A new condition ConditionNeedsUpdate= has been added that
3594 may conditionalize services to only run when /etc or /var
3595 are "older" than the vendor operating system resources in
3596 /usr. This is useful for reconstructing or updating /etc
3597 after an offline update of /usr or a factory reset, on the
3598 next reboot. Services that want to run once after such an
3599 update or reset should use this condition and order
3600 themselves before the new systemd-update-done.service, which
3601 will mark the two directories as fully updated. A number of
3602 service files have been added making use of this, to rebuild
3603 the udev hardware database, the journald message catalog and
3604 dynamic loader cache (ldconfig). The systemd-sysusers tool
3605 described above also makes use of this now. With this in
3606 place it is now possible to start up a minimal operating
3607 system with /etc empty cleanly. For more information on the
3608 concepts involved see this recent blog story:
3610 http://0pointer.de/blog/projects/stateless.html
3612 * A new system group "input" has been introduced, and all
3613 input device nodes get this group assigned. This is useful
3614 for system-level software to get access to input devices. It
3615 complements what is already done for "audio" and "video".
3617 * systemd-networkd learnt minimal DHCPv4 server support in
3618 addition to the existing DHCPv4 client support. It also
3619 learnt DHCPv6 client and IPv6 Router Solicitation client
3620 support. The DHCPv4 client gained support for static routes
3621 passed in from the server. Note that the [DHCPv4] section
3622 known in older systemd-networkd versions has been renamed to
3623 [DHCP] and is now also used by the DHCPv6 client. Existing
3624 .network files using settings of this section should be
3625 updated, though compatibility is maintained. Optionally, the
3626 client hostname may now be sent to the DHCP server.
3628 * networkd gained support for vxlan virtual networks as well
3629 as tun/tap and dummy devices.
3631 * networkd gained support for automatic allocation of address
3632 ranges for interfaces from a system-wide pool of
3633 addresses. This is useful for dynamically managing a large
3634 number of interfaces with a single network configuration
3635 file. In particular this is useful to easily assign
3636 appropriate IP addresses to the veth links of a large number
3637 of nspawn instances.
3639 * RPM macros for processing sysusers, sysctl and binfmt
3640 drop-in snippets at package installation time have been
3643 * The /etc/os-release file should now be placed in
3644 /usr/lib/os-release. The old location is automatically
3645 created as symlink. /usr/lib is the more appropriate
3646 location of this file, since it shall actually describe the
3647 vendor operating system shipped in /usr, and not the
3648 configuration stored in /etc.
3650 * .mount units gained a new boolean SloppyOptions= setting
3651 that maps to mount(8)'s -s option which enables permissive
3652 parsing of unknown mount options.
3654 * tmpfiles learnt a new "L+" directive which creates a symlink
3655 but (unlike "L") deletes a pre-existing file first, should
3656 it already exist and not already be the correct
3657 symlink. Similarly, "b+", "c+" and "p+" directives have been
3658 added as well, which create block and character devices, as
3659 well as fifos in the filesystem, possibly removing any
3660 pre-existing files of different types.
3662 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3663 'argument' field (which so far specified the source to
3664 symlink/copy the files from) is now optional. If omitted the
3665 same file os copied from /usr/share/factory/ suffixed by the
3666 full destination path. This is useful for populating /etc
3667 with essential files, by copying them from vendor defaults
3668 shipped in /usr/share/factory/etc.
3670 * A new command "systemctl preset-all" has been added that
3671 applies the service preset settings to all installed unit
3672 files. A new switch --preset-mode= has been added that
3673 controls whether only enable or only disable operations
3676 * A new command "systemctl is-system-running" has been added
3677 that allows checking the overall state of the system, for
3678 example whether it is fully up and running.
3680 * When the system boots up with an empty /etc, the equivalent
3681 to "systemctl preset-all" is executed during early boot, to
3682 make sure all default services are enabled after a factory
3685 * systemd now contains a minimal preset file that enables the
3686 most basic services systemd ships by default.
3688 * Unit files' [Install] section gained a new DefaultInstance=
3689 field for defining the default instance to create if a
3690 template unit is enabled with no instance specified.
3692 * A new passive target cryptsetup-pre.target has been added
3693 that may be used by services that need to make they run and
3694 finish before the first LUKS cryptographic device is set up.
3696 * The /dev/loop-control and /dev/btrfs-control device nodes
3697 are now owned by the "disk" group by default, opening up
3698 access to this group.
3700 * systemd-coredump will now automatically generate a
3701 stack trace of all core dumps taking place on the system,
3702 based on elfutils' libdw library. This stack trace is logged
3705 * systemd-coredump may now optionally store coredumps directly
3706 on disk (in /var/lib/systemd/coredump, possibly compressed),
3707 instead of storing them unconditionally in the journal. This
3708 mode is the new default. A new configuration file
3709 /etc/systemd/coredump.conf has been added to configure this
3710 and other parameters of systemd-coredump.
3712 * coredumpctl gained a new "info" verb to show details about a
3713 specific coredump. A new switch "-1" has also been added
3714 that makes sure to only show information about the most
3715 recent entry instead of all entries. Also, as the tool is
3716 generally useful now the "systemd-" prefix of the binary
3717 name has been removed. Distributions that want to maintain
3718 compatibility with the old name should add a symlink from
3719 the old name to the new name.
3721 * journald's SplitMode= now defaults to "uid". This makes sure
3722 that unprivileged users can access their own coredumps with
3723 coredumpctl without restrictions.
3725 * New kernel command line options "systemd.wants=" (for
3726 pulling an additional unit during boot), "systemd.mask="
3727 (for masking a specific unit for the boot), and
3728 "systemd.debug-shell" (for enabling the debug shell on tty9)
3729 have been added. This is implemented in the new generator
3730 "systemd-debug-generator".
3732 * systemd-nspawn will now by default filter a couple of
3733 syscalls for containers, among them those required for
3734 kernel module loading, direct x86 IO port access, swap
3735 management, and kexec. Most importantly though
3736 open_by_handle_at() is now prohibited for containers,
3737 closing a hole similar to a recently discussed vulnerability
3738 in docker regarding access to files on file hierarchies the
3739 container should normally not have access to. Note that, for
3740 nspawn, we generally make no security claims anyway (and
3741 this is explicitly documented in the man page), so this is
3742 just a fix for one of the most obvious problems.
3744 * A new man page file-hierarchy(7) has been added that
3745 contains a minimized, modernized version of the file system
3746 layout systemd expects, similar in style to the FHS
3747 specification or hier(5). A new tool systemd-path(1) has
3748 been added to query many of these paths for the local
3751 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3752 longer done. Since the directory now has a per-user size
3753 limit, and is cleaned on logout this appears unnecessary,
3754 in particular since this now brings the lifecycle of this
3755 directory closer in line with how IPC objects are handled.
3757 * systemd.pc now exports a number of additional directories,
3758 including $libdir (which is useful to identify the library
3759 path for the primary architecture of the system), and a
3760 couple of drop-in directories.
3762 * udev's predictable network interface names now use the dev_port
3763 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3764 distinguish between ports of the same PCI function. dev_id should
3765 only be used for ports using the same HW address, hence the need
3768 * machined has been updated to export the OS version of a
3769 container (read from /etc/os-release and
3770 /usr/lib/os-release) on the bus. This is now shown in
3771 "machinectl status" for a machine.
3773 * A new service setting RestartForceExitStatus= has been
3774 added. If configured to a set of exit signals or process
3775 return values, the service will be restarted when the main
3776 daemon process exits with any of them, regardless of the
3779 * systemctl's -H switch for connecting to remote systemd
3780 machines has been extended so that it may be used to
3781 directly connect to a specific container on the
3782 host. "systemctl -H root@foobar:waldi" will now connect as
3783 user "root" to host "foobar", and then proceed directly to
3784 the container named "waldi". Note that currently you have to
3785 authenticate as user "root" for this to work, as entering
3786 containers is a privileged operation.
3788 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3789 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3790 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3791 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3792 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3793 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3794 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3795 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3796 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3797 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3798 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3799 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3801 — Berlin, 2014-07-03
3805 * As an experimental feature, udev now tries to lock the
3806 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3807 executes events for the disk or any of its partitions.
3808 Applications like partitioning programs can lock the
3809 disk device node (flock(LOCK_EX)) and claim temporary
3810 device ownership that way; udev will entirely skip all event
3811 handling for this disk and its partitions. If the disk
3812 was opened for writing, the close will trigger a partition
3813 table rescan in udev's "watch" facility, and if needed
3814 synthesize "change" events for the disk and all its partitions.
3815 This is now unconditionally enabled, and if it turns out to
3816 cause major problems, we might turn it on only for specific
3817 devices, or might need to disable it entirely. Device Mapper
3818 devices are excluded from this logic.
3820 * We temporarily dropped the "-l" switch for fsck invocations,
3821 since they collide with the flock() logic above. util-linux
3822 upstream has been changed already to avoid this conflict,
3823 and we will readd "-l" as soon as util-linux with this
3824 change has been released.
3826 * The dependency on libattr has been removed. Since a long
3827 time, the extended attribute calls have moved to glibc, and
3828 libattr is thus unnecessary.
3830 * Virtualization detection works without privileges now. This
3831 means the systemd-detect-virt binary no longer requires
3832 CAP_SYS_PTRACE file capabilities, and our daemons can run
3833 with fewer privileges.
3835 * systemd-networkd now runs under its own "systemd-network"
3836 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3837 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3838 loses the ability to write to files owned by root this way.
3840 * Similarly, systemd-resolved now runs under its own
3841 "systemd-resolve" user with no capabilities remaining.
3843 * Similarly, systemd-bus-proxyd now runs under its own
3844 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3846 * systemd-networkd gained support for setting up "veth"
3847 virtual Ethernet devices for container connectivity, as well
3848 as GRE and VTI tunnels.
3850 * systemd-networkd will no longer automatically attempt to
3851 manually load kernel modules necessary for certain tunnel
3852 transports. Instead, it is assumed the kernel loads them
3853 automatically when required. This only works correctly on
3854 very new kernels. On older kernels, please consider adding
3855 the kernel modules to /etc/modules-load.d/ as a work-around.
3857 * The resolv.conf file systemd-resolved generates has been
3858 moved to /run/systemd/resolve/. If you have a symlink from
3859 /etc/resolv.conf, it might be necessary to correct it.
3861 * Two new service settings, ProtectHome= and ProtectSystem=,
3862 have been added. When enabled, they will make the user data
3863 (such as /home) inaccessible or read-only and the system
3864 (such as /usr) read-only, for specific services. This allows
3865 very light-weight per-service sandboxing to avoid
3866 modifications of user data or system files from
3867 services. These two new switches have been enabled for all
3868 of systemd's long-running services, where appropriate.
3870 * Socket units gained new SocketUser= and SocketGroup=
3871 settings to set the owner user and group of AF_UNIX sockets
3872 and FIFOs in the file system.
3874 * Socket units gained a new RemoveOnStop= setting. If enabled,
3875 all FIFOS and sockets in the file system will be removed
3876 when the specific socket unit is stopped.
3878 * Socket units gained a new Symlinks= setting. It takes a list
3879 of symlinks to create to file system sockets or FIFOs
3880 created by the specific Unix sockets. This is useful to
3881 manage symlinks to socket nodes with the same life-cycle as
3884 * The /dev/log socket and /dev/initctl FIFO have been moved to
3885 /run, and have been replaced by symlinks. This allows
3886 connecting to these facilities even if PrivateDevices=yes is
3887 used for a service (which makes /dev/log itself unavailable,
3888 but /run is left). This also has the benefit of ensuring
3889 that /dev only contains device nodes, directories and
3890 symlinks, and nothing else.
3892 * sd-daemon gained two new calls sd_pid_notify() and
3893 sd_pid_notifyf(). They are similar to sd_notify() and
3894 sd_notifyf(), but allow overriding of the source PID of
3895 notification messages if permissions permit this. This is
3896 useful to send notify messages on behalf of a different
3897 process (for example, the parent process). The
3898 systemd-notify tool has been updated to make use of this
3899 when sending messages (so that notification messages now
3900 originate from the shell script invoking systemd-notify and
3901 not the systemd-notify process itself. This should minimize
3902 a race where systemd fails to associate notification
3903 messages to services when the originating process already
3906 * A new "on-abnormal" setting for Restart= has been added. If
3907 set, it will result in automatic restarts on all "abnormal"
3908 reasons for a process to exit, which includes unclean
3909 signals, core dumps, timeouts and watchdog timeouts, but
3910 does not include clean and unclean exit codes or clean
3911 signals. Restart=on-abnormal is an alternative for
3912 Restart=on-failure for services that shall be able to
3913 terminate and avoid restarts on certain errors, by
3914 indicating so with an unclean exit code. Restart=on-failure
3915 or Restart=on-abnormal is now the recommended setting for
3916 all long-running services.
3918 * If the InaccessibleDirectories= service setting points to a
3919 mount point (or if there are any submounts contained within
3920 it), it is now attempted to completely unmount it, to make
3921 the file systems truly unavailable for the respective
3924 * The ReadOnlyDirectories= service setting and
3925 systemd-nspawn's --read-only parameter are now recursively
3926 applied to all submounts, too.
3928 * Mount units may now be created transiently via the bus APIs.
3930 * The support for SysV and LSB init scripts has been removed
3931 from the systemd daemon itself. Instead, it is now
3932 implemented as a generator that creates native systemd units
3933 from these scripts when needed. This enables us to remove a
3934 substantial amount of legacy code from PID 1, following the
3935 fact that many distributions only ship a very small number
3936 of LSB/SysV init scripts nowadays.
3938 * Privileged Xen (dom0) domains are not considered
3939 virtualization anymore by the virtualization detection
3940 logic. After all, they generally have unrestricted access to
3941 the hardware and usually are used to manage the unprivileged
3944 * systemd-tmpfiles gained a new "C" line type, for copying
3945 files or entire directories.
3947 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3948 lines. So far, they have been non-globbing versions of the
3949 latter, and have thus been redundant. In future, it is
3950 recommended to only use "z". "m" has hence been removed
3951 from the documentation, even though it stays supported.
3953 * A tmpfiles snippet to recreate the most basic structure in
3954 /var has been added. This is enough to create the /var/run →
3955 /run symlink and create a couple of structural
3956 directories. This allows systems to boot up with an empty or
3957 volatile /var. Of course, while with this change, the core OS
3958 now is capable with dealing with a volatile /var, not all
3959 user services are ready for it. However, we hope that sooner
3960 or later, many service daemons will be changed upstream so
3961 that they are able to automatically create their necessary
3962 directories in /var at boot, should they be missing. This is
3963 the first step to allow state-less systems that only require
3964 the vendor image for /usr to boot.
3966 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3967 empty tmpfs instance to a specific directory. This is
3968 particularly useful for making use of the automatic
3969 reconstruction of /var (see above), by passing --tmpfs=/var.
3971 * Access modes specified in tmpfiles snippets may now be
3972 prefixed with "~", which indicates that they shall be masked
3973 by whether the existing file or directory is currently
3974 writable, readable or executable at all. Also, if specified,
3975 the sgid/suid/sticky bits will be masked for all
3978 * A new passive target unit "network-pre.target" has been
3979 added which is useful for services that shall run before any
3980 network is configured, for example firewall scripts.
3982 * The "floppy" group that previously owned the /dev/fd*
3983 devices is no longer used. The "disk" group is now used
3984 instead. Distributions should probably deprecate usage of
3987 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3988 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3989 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3990 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3991 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3992 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3995 — Berlin, 2014-06-11
3999 * A new "systemd-timesyncd" daemon has been added for
4000 synchronizing the system clock across the network. It
4001 implements an SNTP client. In contrast to NTP
4002 implementations such as chrony or the NTP reference server,
4003 this only implements a client side, and does not bother with
4004 the full NTP complexity, focusing only on querying time from
4005 one remote server and synchronizing the local clock to
4006 it. Unless you intend to serve NTP to networked clients or
4007 want to connect to local hardware clocks, this simple NTP
4008 client should be more than appropriate for most
4009 installations. The daemon runs with minimal privileges, and
4010 has been hooked up with networkd to only operate when
4011 network connectivity is available. The daemon saves the
4012 current clock to disk every time a new NTP sync has been
4013 acquired, and uses this to possibly correct the system clock
4014 early at bootup, in order to accommodate for systems that
4015 lack an RTC such as the Raspberry Pi and embedded devices,
4016 and to make sure that time monotonically progresses on these
4017 systems, even if it is not always correct. To make use of
4018 this daemon, a new system user and group "systemd-timesync"
4019 needs to be created on installation of systemd.
4021 * The queue "seqnum" interface of libudev has been disabled, as
4022 it was generally incompatible with device namespacing as
4023 sequence numbers of devices go "missing" if the devices are
4024 part of a different namespace.
4026 * "systemctl list-timers" and "systemctl list-sockets" gained
4027 a --recursive switch for showing units of these types also
4028 for all local containers, similar in style to the already
4029 supported --recursive switch for "systemctl list-units".
4031 * A new RebootArgument= setting has been added for service
4032 units, which may be used to specify a kernel reboot argument
4033 to use when triggering reboots with StartLimitAction=.
4035 * A new FailureAction= setting has been added for service
4036 units which may be used to specify an operation to trigger
4037 when a service fails. This works similarly to
4038 StartLimitAction=, but unlike it, controls what is done
4039 immediately rather than only after several attempts to
4040 restart the service in question.
4042 * hostnamed got updated to also expose the kernel name,
4043 release, and version on the bus. This is useful for
4044 executing commands like hostnamectl with the -H switch.
4045 systemd-analyze makes use of this to properly display
4046 details when running non-locally.
4048 * The bootchart tool can now show cgroup information in the
4049 graphs it generates.
4051 * The CFS CPU quota cgroup attribute is now exposed for
4052 services. The new CPUQuota= switch has been added for this
4053 which takes a percentage value. Setting this will have the
4054 result that a service may never get more CPU time than the
4055 specified percentage, even if the machine is otherwise idle.
4057 * systemd-networkd learned IPIP and SIT tunnel support.
4059 * LSB init scripts exposing a dependency on $network will now
4060 get a dependency on network-online.target rather than simply
4061 network.target. This should bring LSB handling closer to
4062 what it was on SysV systems.
4064 * A new fsck.repair= kernel option has been added to control
4065 how fsck shall deal with unclean file systems at boot.
4067 * The (.ini) configuration file parser will now silently
4068 ignore sections whose name begins with "X-". This may be
4069 used to maintain application-specific extension sections in unit
4072 * machined gained a new API to query the IP addresses of
4073 registered containers. "machinectl status" has been updated
4074 to show these addresses in its output.
4076 * A new call sd_uid_get_display() has been added to the
4077 sd-login APIs for querying the "primary" session of a
4078 user. The "primary" session of the user is elected from the
4079 user's sessions and generally a graphical session is
4080 preferred over a text one.
4082 * A minimal systemd-resolved daemon has been added. It
4083 currently simply acts as a companion to systemd-networkd and
4084 manages resolv.conf based on per-interface DNS
4085 configuration, possibly supplied via DHCP. In the long run
4086 we hope to extend this into a local DNSSEC enabled DNS and
4089 * The systemd-networkd-wait-online tool is now enabled by
4090 default. It will delay network-online.target until a network
4091 connection has been configured. The tool primarily integrates
4092 with networkd, but will also make a best effort to make sense
4093 of network configuration performed in some other way.
4095 * Two new service options StartupCPUShares= and
4096 StartupBlockIOWeight= have been added that work similarly to
4097 CPUShares= and BlockIOWeight= however only apply during
4098 system startup. This is useful to prioritize certain services
4099 differently during bootup than during normal runtime.
4101 * hostnamed has been changed to prefer the statically
4102 configured hostname in /etc/hostname (unless set to
4103 'localhost' or empty) over any dynamic one supplied by
4104 dhcp. With this change, the rules for picking the hostname
4105 match more closely the rules of other configuration settings
4106 where the local administrator's configuration in /etc always
4107 overrides any other settings.
4109 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4110 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4111 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4112 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4113 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4114 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4115 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4116 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4117 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4118 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4119 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4120 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4121 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4122 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4123 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4124 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4127 — Beijing, 2014-05-28
4131 * When restoring the screen brightness at boot, stay away from
4132 the darkest setting or from the lowest 5% of the available
4133 range, depending on which is the larger value of both. This
4134 should effectively protect the user from rebooting into a
4135 black screen, should the brightness have been set to minimum
4138 * sd-login gained a new sd_machine_get_class() call to
4139 determine the class ("vm" or "container") of a machine
4140 registered with machined.
4142 * sd-login gained new calls
4143 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4144 to query the identity of the peer of a local AF_UNIX
4145 connection. They operate similarly to their sd_pid_get_xyz()
4148 * PID 1 will now maintain a system-wide system state engine
4149 with the states "starting", "running", "degraded",
4150 "maintenance", "stopping". These states are bound to system
4151 startup, normal runtime, runtime with at least one failed
4152 service, rescue/emergency mode and system shutdown. This
4153 state is shown in the "systemctl status" output when no unit
4154 name is passed. It is useful to determine system state, in
4155 particularly when doing so for many systems or containers at
4158 * A new command "list-machines" has been added to "systemctl"
4159 that lists all local OS containers and shows their system
4160 state (see above), if systemd runs inside of them.
4162 * systemctl gained a new "-r" switch to recursively enumerate
4163 units on all local containers, when used with the
4164 "list-unit" command (which is the default one that is
4165 executed when no parameters are specified).
4167 * The GPT automatic partition discovery logic will now honour
4168 two GPT partition flags: one may be set on a partition to
4169 cause it to be mounted read-only, and the other may be set
4170 on a partition to ignore it during automatic discovery.
4172 * Two new GPT type UUIDs have been added for automatic root
4173 partition discovery, for 32-bit and 64-bit ARM. This is not
4174 particularly useful for discovering the root directory on
4175 these architectures during bare-metal boots (since UEFI is
4176 not common there), but still very useful to allow booting of
4177 ARM disk images in nspawn with the -i option.
4179 * MAC addresses of interfaces created with nspawn's
4180 --network-interface= switch will now be generated from the
4181 machine name, and thus be stable between multiple invocations
4184 * logind will now automatically remove all IPC objects owned
4185 by a user if she or he fully logs out. This makes sure that
4186 users who are logged out cannot continue to consume IPC
4187 resources. This covers SysV memory, semaphores and message
4188 queues as well as POSIX shared memory and message
4189 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4190 limits. With this functionality, that is corrected. This may
4191 be turned off by using the RemoveIPC= switch of logind.conf.
4193 * The systemd-machine-id-setup and tmpfiles tools gained a
4194 --root= switch to operate on a specific root directory,
4197 * journald can now forward logged messages to the TTYs of all
4198 logged in users ("wall"). This is the default for all
4199 emergency messages now.
4201 * A new tool systemd-journal-remote has been added to stream
4202 journal log messages across the network.
4204 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4205 controller trees are mounted into it. Note that the
4206 directories mounted beneath it are not read-only. This is a
4207 security measure and is particularly useful because glibc
4208 actually includes a search logic to pick any tmpfs it can
4209 find to implement shm_open() if /dev/shm is not available
4210 (which it might very well be in namespaced setups).
4212 * machinectl gained a new "poweroff" command to cleanly power
4213 down a local OS container.
4215 * The PrivateDevices= unit file setting will now also drop the
4216 CAP_MKNOD capability from the capability bound set, and
4217 imply DevicePolicy=closed.
4219 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4220 comprehensively on all long-running systemd services where
4221 this is appropriate.
4223 * systemd-udevd will now run in a disassociated mount
4224 namespace. To mount directories from udev rules, make sure to
4225 pull in mount units via SYSTEMD_WANTS properties.
4227 * The kdbus support gained support for uploading policy into
4228 the kernel. sd-bus gained support for creating "monitoring"
4229 connections that can eavesdrop into all bus communication
4230 for debugging purposes.
4232 * Timestamps may now be specified in seconds since the UNIX
4233 epoch Jan 1st, 1970 by specifying "@" followed by the value
4236 * Native tcpwrap support in systemd has been removed. tcpwrap
4237 is old code, not really maintained anymore and has serious
4238 shortcomings, and better options such as firewalls
4239 exist. For setups that require tcpwrap usage, please
4240 consider invoking your socket-activated service via tcpd,
4241 like on traditional inetd.
4243 * A new system.conf configuration option
4244 DefaultTimerAccuracySec= has been added that controls the
4245 default AccuracySec= setting of .timer units.
4247 * Timer units gained a new WakeSystem= switch. If enabled,
4248 timers configured this way will cause the system to resume
4249 from system suspend (if the system supports that, which most
4252 * Timer units gained a new Persistent= switch. If enabled,
4253 timers configured this way will save to disk when they have
4254 been last triggered. This information is then used on next
4255 reboot to possible execute overdue timer events, that
4256 could not take place because the system was powered off.
4257 This enables simple anacron-like behaviour for timer units.
4259 * systemctl's "list-timers" will now also list the time a
4260 timer unit was last triggered in addition to the next time
4261 it will be triggered.
4263 * systemd-networkd will now assign predictable IPv4LL
4264 addresses to its local interfaces.
4266 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4267 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4268 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4269 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4270 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4271 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4272 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4273 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4276 — Berlin, 2014-03-25
4280 * A new unit file setting RestrictAddressFamilies= has been
4281 added to restrict which socket address families unit
4282 processes gain access to. This takes address family names
4283 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4284 attack surface of services via exotic protocol stacks. This
4285 is built on seccomp system call filters.
4287 * Two new unit file settings RuntimeDirectory= and
4288 RuntimeDirectoryMode= have been added that may be used to
4289 manage a per-daemon runtime directories below /run. This is
4290 an alternative for setting up directory permissions with
4291 tmpfiles snippets, and has the advantage that the runtime
4292 directory's lifetime is bound to the daemon runtime and that
4293 the daemon starts up with an empty directory each time. This
4294 is particularly useful when writing services that drop
4295 privileges using the User= or Group= setting.
4297 * The DeviceAllow= unit setting now supports globbing for
4298 matching against device group names.
4300 * The systemd configuration file system.conf gained new
4301 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4302 DefaultMemoryAccounting= to globally turn on/off accounting
4303 for specific resources (cgroups) for all units. These
4304 settings may still be overridden individually in each unit
4307 * systemd-gpt-auto-generator is now able to discover /srv and
4308 root partitions in addition to /home and swap partitions. It
4309 also supports LUKS-encrypted partitions now. With this in
4310 place, automatic discovery of partitions to mount following
4311 the Discoverable Partitions Specification
4312 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4313 is now a lot more complete. This allows booting without
4314 /etc/fstab and without root= on the kernel command line on
4315 systems prepared appropriately.
4317 * systemd-nspawn gained a new --image= switch which allows
4318 booting up disk images and Linux installations on any block
4319 device that follow the Discoverable Partitions Specification
4320 (see above). This means that installations made with
4321 appropriately updated installers may now be started and
4322 deployed using container managers, completely
4323 unmodified. (We hope that libvirt-lxc will add support for
4324 this feature soon, too.)
4326 * systemd-nspawn gained a new --network-macvlan= setting to
4327 set up a private macvlan interface for the
4328 container. Similarly, systemd-networkd gained a new
4329 Kind=macvlan setting in .netdev files.
4331 * systemd-networkd now supports configuring local addresses
4334 * A new tool systemd-network-wait-online has been added to
4335 synchronously wait for network connectivity using
4338 * The sd-bus.h bus API gained a new sd_bus_track object for
4339 tracking the life-cycle of bus peers. Note that sd-bus.h is
4340 still not a public API though (unless you specify
4341 --enable-kdbus on the configure command line, which however
4342 voids your warranty and you get no API stability guarantee).
4344 * The $XDG_RUNTIME_DIR runtime directories for each user are
4345 now individual tmpfs instances, which has the benefit of
4346 introducing separate pools for each user, with individual
4347 size limits, and thus making sure that unprivileged clients
4348 can no longer negatively impact the system or other users by
4349 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4350 RuntimeDirectorySize= has been introduced that allows
4351 controlling the default size limit for all users. It
4352 defaults to 10% of the available physical memory. This is no
4353 replacement for quotas on tmpfs though (which the kernel
4354 still does not support), as /dev/shm and /tmp are still
4355 shared resources used by both the system and unprivileged
4358 * logind will now automatically turn off automatic suspending
4359 on laptop lid close when more than one display is
4360 connected. This was previously expected to be implemented
4361 individually in desktop environments (such as GNOME),
4362 however has been added to logind now, in order to fix a
4363 boot-time race where a desktop environment might not have
4364 been started yet and thus not been able to take an inhibitor
4365 lock at the time where logind already suspends the system
4366 due to a closed lid.
4368 * logind will now wait at least 30s after each system
4369 suspend/resume cycle, and 3min after system boot before
4370 suspending the system due to a closed laptop lid. This
4371 should give USB docking stations and similar enough time to
4372 be probed and configured after system resume and boot in
4373 order to then act as suspend blocker.
4375 * systemd-run gained a new --property= setting which allows
4376 initialization of resource control properties (and others)
4377 for the created scope or service unit. Example: "systemd-run
4378 --property=BlockIOWeight=10 updatedb" may be used to run
4379 updatedb at a low block IO scheduling weight.
4381 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4382 now also work in --scope mode.
4384 * When systemd is compiled with kdbus support, basic support
4385 for enforced policies is now in place. (Note that enabling
4386 kdbus still voids your warranty and no API compatibility
4389 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4390 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4391 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4392 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4393 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4394 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4395 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4396 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4397 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4398 Zbigniew Jędrzejewski-Szmek
4400 — Berlin, 2014-03-12
4404 * systemd will now relabel /dev after loading the SMACK policy
4405 according to SMACK rules.
4407 * A new unit file option AppArmorProfile= has been added to
4408 set the AppArmor profile for the processes of a unit.
4410 * A new condition check ConditionArchitecture= has been added
4411 to conditionalize units based on the system architecture, as
4412 reported by uname()'s "machine" field.
4414 * systemd-networkd now supports matching on the system
4415 virtualization, architecture, kernel command line, host name
4418 * logind is now a lot more aggressive when suspending the
4419 machine due to a closed laptop lid. Instead of acting only
4420 on the lid close action, it will continuously watch the lid
4421 status and act on it. This is useful for laptops where the
4422 power button is on the outside of the chassis so that it can
4423 be reached without opening the lid (such as the Lenovo
4424 Yoga). On those machines, logind will now immediately
4425 re-suspend the machine if the power button has been
4426 accidentally pressed while the laptop was suspended and in a
4427 backpack or similar.
4429 * logind will now watch SW_DOCK switches and inhibit reaction
4430 to the lid switch if it is pressed. This means that logind
4431 will not suspend the machine anymore if the lid is closed
4432 and the system is docked, if the laptop supports SW_DOCK
4433 notifications via the input layer. Note that ACPI docking
4434 stations do not generate this currently. Also note that this
4435 logic is usually not fully sufficient and Desktop
4436 Environments should take a lid switch inhibitor lock when an
4437 external display is connected, as systemd will not watch
4440 * nspawn will now make use of the devices cgroup controller by
4441 default, and only permit creation of and access to the usual
4442 API device nodes like /dev/null or /dev/random, as well as
4443 access to (but not creation of) the pty devices.
4445 * We will now ship a default .network file for
4446 systemd-networkd that automatically configures DHCP for
4447 network interfaces created by nspawn's --network-veth or
4448 --network-bridge= switches.
4450 * systemd will now understand the usual M, K, G, T suffixes
4451 according to SI conventions (i.e. to the base 1000) when
4452 referring to throughput and hardware metrics. It will stay
4453 with IEC conventions (i.e. to the base 1024) for software
4454 metrics, according to what is customary according to
4455 Wikipedia. We explicitly document which base applies for
4456 each configuration option.
4458 * The DeviceAllow= setting in unit files now supports a syntax
4459 to whitelist an entire group of devices node majors at once,
4460 based on the /proc/devices listing. For example, with the
4461 string "char-pts", it is now possible to whitelist all
4462 current and future pseudo-TTYs at once.
4464 * sd-event learned a new "post" event source. Event sources of
4465 this type are triggered by the dispatching of any event
4466 source of a type that is not "post". This is useful for
4467 implementing clean-up and check event sources that are
4468 triggered by other work being done in the program.
4470 * systemd-networkd is no longer statically enabled, but uses
4471 the usual [Install] sections so that it can be
4472 enabled/disabled using systemctl. It still is enabled by
4475 * When creating a veth interface pair with systemd-nspawn, the
4476 host side will now be prefixed with "vb-" if
4477 --network-bridge= is used, and with "ve-" if --network-veth
4478 is used. This way, it is easy to distinguish these cases on
4479 the host, for example to apply different configuration to
4480 them with systemd-networkd.
4482 * The compatibility libraries for libsystemd-journal.so,
4483 libsystem-id128.so, libsystemd-login.so and
4484 libsystemd-daemon.so do not make use of IFUNC
4485 anymore. Instead, we now build libsystemd.so multiple times
4486 under these alternative names. This means that the footprint
4487 is drastically increased, but given that these are
4488 transitional compatibility libraries, this should not matter
4489 much. This change has been made necessary to support the ARM
4490 platform for these compatibility libraries, as the ARM
4491 toolchain is not really at the same level as the toolchain
4492 for other architectures like x86 and does not support
4493 IFUNC. Please make sure to use --enable-compat-libs only
4494 during a transitional period!
4496 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4497 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4498 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4499 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4500 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4501 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4502 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4503 Zbigniew Jędrzejewski-Szmek
4505 — Berlin, 2014-02-24
4509 * A new component "systemd-networkd" has been added that can
4510 be used to configure local network interfaces statically or
4511 via DHCP. It is capable of bringing up bridges, VLANs, and
4512 bonding. Currently, no hook-ups for interactive network
4513 configuration are provided. Use this for your initrd,
4514 container, embedded, or server setup if you need a simple,
4515 yet powerful, network configuration solution. This
4516 configuration subsystem is quite nifty, as it allows wildcard
4517 hotplug matching in interfaces. For example, with a single
4518 configuration snippet, you can configure that all Ethernet
4519 interfaces showing up are automatically added to a bridge,
4520 or similar. It supports link-sensing and more.
4522 * A new tool "systemd-socket-proxyd" has been added which can
4523 act as a bidirectional proxy for TCP sockets. This is
4524 useful for adding socket activation support to services that
4525 do not actually support socket activation, including virtual
4526 machines and the like.
4528 * Add a new tool to save/restore rfkill state on
4531 * Save/restore state of keyboard backlights in addition to
4532 display backlights on shutdown/boot.
4534 * udev learned a new SECLABEL{} construct to label device
4535 nodes with a specific security label when they appear. For
4536 now, only SECLABEL{selinux} is supported, but the syntax is
4537 prepared for additional security frameworks.
4539 * udev gained a new scheme to configure link-level attributes
4540 from files in /etc/systemd/network/*.link. These files can
4541 match against MAC address, device path, driver name and type,
4542 and will apply attributes like the naming policy, link speed,
4543 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4544 address assignment policy (randomized, ...).
4546 * The configuration of network interface naming rules for
4547 "permanent interface names" has changed: a new NamePolicy=
4548 setting in the [Link] section of .link files determines the
4549 priority of possible naming schemes (onboard, slot, MAC,
4550 path). The default value of this setting is determined by
4551 /usr/lib/net/links/99-default.link. Old
4552 80-net-name-slot.rules udev configuration file has been
4553 removed, so local configuration overriding this file should
4554 be adapted to override 99-default.link instead.
4556 * When the User= switch is used in a unit file, also
4557 initialize $SHELL= based on the user database entry.
4559 * systemd no longer depends on libdbus. All communication is
4560 now done with sd-bus, systemd's low-level bus library
4563 * kdbus support has been added to PID 1 itself. When kdbus is
4564 enabled, this causes PID 1 to set up the system bus and
4565 enable support for a new ".busname" unit type that
4566 encapsulates bus name activation on kdbus. It works a little
4567 bit like ".socket" units, except for bus names. A new
4568 generator has been added that converts classic dbus1 service
4569 activation files automatically into native systemd .busname
4572 * sd-bus: add a light-weight vtable implementation that allows
4573 defining objects on the bus with a simple static const
4574 vtable array of its methods, signals and properties.
4576 * systemd will not generate or install static dbus
4577 introspection data anymore to /usr/share/dbus-1/interfaces,
4578 as the precise format of these files is unclear, and
4579 nothing makes use of it.
4581 * A proxy daemon is now provided to proxy clients connecting
4582 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4583 compatibility with classic D-Bus.
4585 * A bus driver implementation has been added that supports the
4586 classic D-Bus bus driver calls on kdbus, also for
4587 compatibility purposes.
4589 * A new API "sd-event.h" has been added that implements a
4590 minimal event loop API built around epoll. It provides a
4591 couple of features that direct epoll usage is lacking:
4592 prioritization of events, scales to large numbers of timer
4593 events, per-event timer slack (accuracy), system-wide
4594 coalescing of timer events, exit handlers, watchdog
4595 supervision support using systemd's sd_notify() API, child
4598 * A new API "sd-rntl.h" has been added that provides an API
4599 around the route netlink interface of the kernel, similar in
4600 style to "sd-bus.h".
4602 * A new API "sd-dhcp-client.h" has been added that provides a
4603 small DHCPv4 client-side implementation. This is used by
4606 * There is a new kernel command line option
4607 "systemd.restore_state=0|1". When set to "0", none of the
4608 systemd tools will restore saved runtime state to hardware
4609 devices. More specifically, the rfkill and backlight states
4612 * The FsckPassNo= compatibility option in mount/service units
4613 has been removed. The fstab generator will now add the
4614 necessary dependencies automatically, and does not require
4615 PID1's support for that anymore.
4617 * journalctl gained a new switch, --list-boots, that lists
4618 recent boots with their times and boot IDs.
4620 * The various tools like systemctl, loginctl, timedatectl,
4621 busctl, systemd-run, ... have gained a new switch "-M" to
4622 connect to a specific, local OS container (as direct
4623 connection, without requiring SSH). This works on any
4624 container that is registered with machined, such as those
4625 created by libvirt-lxc or nspawn.
4627 * systemd-run and systemd-analyze also gained support for "-H"
4628 to connect to remote hosts via SSH. This is particularly
4629 useful for systemd-run because it enables queuing of jobs
4630 onto remote systems.
4632 * machinectl gained a new command "login" to open a getty
4633 login in any local container. This works with any container
4634 that is registered with machined (such as those created by
4635 libvirt-lxc or nspawn), and which runs systemd inside.
4637 * machinectl gained a new "reboot" command that may be used to
4638 trigger a reboot on a specific container that is registered
4639 with machined. This works on any container that runs an init
4640 system of some kind.
4642 * systemctl gained a new "list-timers" command to print a nice
4643 listing of installed timer units with the times they elapse
4646 * Alternative reboot() parameters may now be specified on the
4647 "systemctl reboot" command line and are passed to the
4648 reboot() system call.
4650 * systemctl gained a new --job-mode= switch to configure the
4651 mode to queue a job with. This is a more generic version of
4652 --fail, --irreversible, and --ignore-dependencies, which are
4653 still available but not advertised anymore.
4655 * /etc/systemd/system.conf gained new settings to configure
4656 various default timeouts of units, as well as the default
4657 start limit interval and burst. These may still be overridden
4660 * PID1 will now export on the bus profile data of the security
4661 policy upload process (such as the SELinux policy upload to
4664 * journald: when forwarding logs to the console, include
4665 timestamps (following the setting in
4666 /sys/module/printk/parameters/time).
4668 * OnCalendar= in timer units now understands the special
4669 strings "yearly" and "annually". (Both are equivalent)
4671 * The accuracy of timer units is now configurable with the new
4672 AccuracySec= setting. It defaults to 1min.
4674 * A new dependency type JoinsNamespaceOf= has been added that
4675 allows running two services within the same /tmp and network
4676 namespace, if PrivateNetwork= or PrivateTmp= are used.
4678 * A new command "cat" has been added to systemctl. It outputs
4679 the original unit file of a unit, and concatenates the
4680 contents of additional "drop-in" unit file snippets, so that
4681 the full configuration is shown.
4683 * systemctl now supports globbing on the various "list-xyz"
4684 commands, like "list-units" or "list-sockets", as well as on
4685 those commands which take multiple unit names.
4687 * journalctl's --unit= switch gained support for globbing.
4689 * All systemd daemons now make use of the watchdog logic so
4690 that systemd automatically notices when they hang.
4692 * If the $container_ttys environment variable is set,
4693 getty-generator will automatically spawn a getty for each
4694 listed tty. This is useful for container managers to request
4695 login gettys to be spawned on as many ttys as needed.
4697 * %h, %s, %U specifier support is not available anymore when
4698 used in unit files for PID 1. This is because NSS calls are
4699 not safe from PID 1. They stay available for --user
4700 instances of systemd, and as special case for the root user.
4702 * loginctl gained a new "--no-legend" switch to turn off output
4705 * The "sd-login.h" API gained three new calls:
4706 sd_session_is_remote(), sd_session_get_remote_user(),
4707 sd_session_get_remote_host() to query information about
4710 * The udev hardware database now also carries vendor/product
4711 information of SDIO devices.
4713 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4714 determine whether watchdog notifications are requested by
4717 * Socket-activated per-connection services now include a
4718 short description of the connection parameters in the
4721 * tmpfiles gained a new "--boot" option. When this is not used,
4722 only lines where the command character is not suffixed with
4723 "!" are executed. When this option is specified, those
4724 options are executed too. This partitions tmpfiles
4725 directives into those that can be safely executed at any
4726 time, and those which should be run only at boot (for
4727 example, a line that creates /run/nologin).
4729 * A new API "sd-resolve.h" has been added which provides a simple
4730 asynchronous wrapper around glibc NSS host name resolution
4731 calls, such as getaddrinfo(). In contrast to glibc's
4732 getaddrinfo_a(), it does not use signals. In contrast to most
4733 other asynchronous name resolution libraries, this one does
4734 not reimplement DNS, but reuses NSS, so that alternate
4735 host name resolution systems continue to work, such as mDNS,
4736 LDAP, etc. This API is based on libasyncns, but it has been
4737 cleaned up for inclusion in systemd.
4739 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4740 "sd-daemon.h" are no longer found in individual libraries
4741 libsystemd-journal.so, libsystemd-login.so,
4742 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4743 merged them into a single library, libsystemd.so, which
4744 provides all symbols. The reason for this is cyclic
4745 dependencies, as these libraries tend to use each other's
4746 symbols. So far, we have managed to workaround that by linking
4747 a copy of a good part of our code into each of these
4748 libraries again and again, which, however, makes certain
4749 things hard to do, like sharing static variables. Also, it
4750 substantially increases footprint. With this change, there
4751 is only one library for the basic APIs systemd
4752 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4753 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4754 library as well, however are subject to the --enable-kdbus
4755 switch (see below). Note that "sd-dhcp-client.h" is not part
4756 of this library (this is because it only consumes, never
4757 provides, services of/to other APIs). To make the transition
4758 easy from the separate libraries to the unified one, we
4759 provide the --enable-compat-libs compile-time switch which
4760 will generate stub libraries that are compatible with the
4761 old ones but redirect all calls to the new one.
4763 * All of the kdbus logic and the new APIs "sd-bus.h",
4764 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4765 and "sd-utf8.h" are compile-time optional via the
4766 "--enable-kdbus" switch, and they are not compiled in by
4767 default. To make use of kdbus, you have to explicitly enable
4768 the switch. Note however, that neither the kernel nor the
4769 userspace API for all of this is considered stable yet. We
4770 want to maintain the freedom to still change the APIs for
4771 now. By specifying this build-time switch, you acknowledge
4772 that you are aware of the instability of the current
4775 * Also, note that while kdbus is pretty much complete,
4776 it lacks one thing: proper policy support. This means you
4777 can build a fully working system with all features; however,
4778 it will be highly insecure. Policy support will be added in
4779 one of the next releases, at the same time that we will
4780 declare the APIs stable.
4782 * When the kernel command line argument "kdbus" is specified,
4783 systemd will automatically load the kdbus.ko kernel module. At
4784 this stage of development, it is only useful for testing kdbus
4785 and should not be used in production. Note: if "--enable-kdbus"
4786 is specified, and the kdbus.ko kernel module is available, and
4787 "kdbus" is added to the kernel command line, the entire system
4788 runs with kdbus instead of dbus-daemon, with the above mentioned
4789 problem of missing the system policy enforcement. Also a future
4790 version of kdbus.ko or a newer systemd will not be compatible with
4791 each other, and will unlikely be able to boot the machine if only
4792 one of them is updated.
4794 * systemctl gained a new "import-environment" command which
4795 uploads the caller's environment (or parts thereof) into the
4796 service manager so that it is inherited by services started
4797 by the manager. This is useful to upload variables like
4798 $DISPLAY into the user service manager.
4800 * A new PrivateDevices= switch has been added to service units
4801 which allows running a service with a namespaced /dev
4802 directory that does not contain any device nodes for
4803 physical devices. More specifically, it only includes devices
4804 such as /dev/null, /dev/urandom, and /dev/zero which are API
4807 * logind has been extended to support behaviour like VT
4808 switching on seats that do not support a VT. This makes
4809 multi-session available on seats that are not the first seat
4810 (seat0), and on systems where kernel support for VTs has
4811 been disabled at compile-time.
4813 * If a process holds a delay lock for system sleep or shutdown
4814 and fails to release it in time, we will now log its
4815 identity. This makes it easier to identify processes that
4816 cause slow suspends or power-offs.
4818 * When parsing /etc/crypttab, support for a new key-slot=
4819 option as supported by Debian is added. It allows indicating
4820 which LUKS slot to use on disk, speeding up key loading.
4822 * The sd_journald_sendv() API call has been checked and
4823 officially declared to be async-signal-safe so that it may
4824 be invoked from signal handlers for logging purposes.
4826 * Boot-time status output is now enabled automatically after a
4827 short timeout if boot does not progress, in order to give
4828 the user an indication what she or he is waiting for.
4830 * The boot-time output has been improved to show how much time
4831 remains until jobs expire.
4833 * The KillMode= switch in service units gained a new possible
4834 value "mixed". If set, and the unit is shut down, then the
4835 initial SIGTERM signal is sent only to the main daemon
4836 process, while the following SIGKILL signal is sent to
4837 all remaining processes of the service.
4839 * When a scope unit is registered, a new property "Controller"
4840 may be set. If set to a valid bus name, systemd will send a
4841 RequestStop() signal to this name when it would like to shut
4842 down the scope. This may be used to hook manager logic into
4843 the shutdown logic of scope units. Also, scope units may now
4844 be put in a special "abandoned" state, in which case the
4845 manager process which created them takes no further
4846 responsibilities for it.
4848 * When reading unit files, systemd will now verify
4849 the access mode of these files, and warn about certain
4850 suspicious combinations. This has been added to make it
4851 easier to track down packaging bugs where unit files are
4852 marked executable or world-writable.
4854 * systemd-nspawn gained a new "--setenv=" switch to set
4855 container-wide environment variables. The similar option in
4856 systemd-activate was renamed from "--environment=" to
4857 "--setenv=" for consistency.
4859 * systemd-nspawn has been updated to create a new kdbus domain
4860 for each container that is invoked, thus allowing each
4861 container to have its own set of system and user buses,
4862 independent of the host.
4864 * systemd-nspawn gained a new --drop-capability= switch to run
4865 the container with less capabilities than the default. Both
4866 --drop-capability= and --capability= now take the special
4867 string "all" for dropping or keeping all capabilities.
4869 * systemd-nspawn gained new switches for executing containers
4870 with specific SELinux labels set.
4872 * systemd-nspawn gained a new --quiet switch to not generate
4873 any additional output but the container's own console
4876 * systemd-nspawn gained a new --share-system switch to run a
4877 container without PID namespacing enabled.
4879 * systemd-nspawn gained a new --register= switch to control
4880 whether the container is registered with systemd-machined or
4881 not. This is useful for containers that do not run full
4882 OS images, but only specific apps.
4884 * systemd-nspawn gained a new --keep-unit which may be used
4885 when invoked as the only program from a service unit, and
4886 results in registration of the unit service itself in
4887 systemd-machined, instead of a newly opened scope unit.
4889 * systemd-nspawn gained a new --network-interface= switch for
4890 moving arbitrary interfaces to the container. The new
4891 --network-veth switch creates a virtual Ethernet connection
4892 between host and container. The new --network-bridge=
4893 switch then allows assigning the host side of this virtual
4894 Ethernet connection to a bridge device.
4896 * systemd-nspawn gained a new --personality= switch for
4897 setting the kernel personality for the container. This is
4898 useful when running a 32-bit container on a 64-bit host. A
4899 similar option Personality= is now also available for service
4902 * logind will now also track a "Desktop" identifier for each
4903 session which encodes the desktop environment of it. This is
4904 useful for desktop environments that want to identify
4905 multiple running sessions of itself easily.
4907 * A new SELinuxContext= setting for service units has been
4908 added that allows setting a specific SELinux execution
4909 context for a service.
4911 * Most systemd client tools will now honour $SYSTEMD_LESS for
4912 settings of the "less" pager. By default, these tools will
4913 override $LESS to allow certain operations to work, such as
4914 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4915 influence this logic.
4917 * systemd's "seccomp" hook-up has been changed to make use of
4918 the libseccomp library instead of using its own
4919 implementation. This has benefits for portability among
4922 * For usage together with SystemCallFilter=, a new
4923 SystemCallErrorNumber= setting has been introduced that
4924 allows configuration of a system error number to be returned
4925 on filtered system calls, instead of immediately killing the
4926 process. Also, SystemCallArchitectures= has been added to
4927 limit access to system calls of a particular architecture
4928 (in order to turn off support for unused secondary
4929 architectures). There is also a global
4930 SystemCallArchitectures= setting in system.conf now to turn
4931 off support for non-native system calls system-wide.
4933 * systemd requires a kernel with a working name_to_handle_at(),
4934 please see the kernel config requirements in the README file.
4936 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4937 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4938 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4939 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4940 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4941 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4942 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4943 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4944 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4945 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4946 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4947 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4948 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4949 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4950 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4951 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4952 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4953 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4954 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4955 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4956 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4957 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4958 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4959 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4961 — Berlin, 2014-02-20
4965 * logind has gained support for facilitating privileged input
4966 and drm device access for unprivileged clients. This work is
4967 useful to allow Wayland display servers (and similar
4968 programs, such as kmscon) to run under the user's ID and
4969 access input and drm devices which are normally
4970 protected. When this is used (and the kernel is new enough)
4971 logind will "mute" IO on the file descriptors passed to
4972 Wayland as long as it is in the background and "unmute" it
4973 if it returns into the foreground. This allows secure
4974 session switching without allowing background sessions to
4975 eavesdrop on input and display data. This also introduces
4976 session switching support if VT support is turned off in the
4977 kernel, and on seats that are not seat0.
4979 * A new kernel command line option luks.options= is understood
4980 now which allows specifying LUKS options for usage for LUKS
4981 encrypted partitions specified with luks.uuid=.
4983 * tmpfiles.d(5) snippets may now use specifier expansion in
4984 path names. More specifically %m, %b, %H, %v, are now
4985 replaced by the local machine id, boot id, hostname, and
4986 kernel version number.
4988 * A new tmpfiles.d(5) command "m" has been introduced which
4989 may be used to change the owner/group/access mode of a file
4990 or directory if it exists, but do nothing if it does not.
4992 * This release removes high-level support for the
4993 MemorySoftLimit= cgroup setting. The underlying kernel
4994 cgroup attribute memory.soft_limit= is currently badly
4995 designed and likely to be removed from the kernel API in its
4996 current form, hence we should not expose it for now.
4998 * The memory.use_hierarchy cgroup attribute is now enabled for
4999 all cgroups systemd creates in the memory cgroup
5000 hierarchy. This option is likely to be come the built-in
5001 default in the kernel anyway, and the non-hierarchical mode
5002 never made much sense in the intrinsically hierarchical
5005 * A new field _SYSTEMD_SLICE= is logged along with all journal
5006 messages containing the slice a message was generated
5007 from. This is useful to allow easy per-customer filtering of
5008 logs among other things.
5010 * systemd-journald will no longer adjust the group of journal
5011 files it creates to the "systemd-journal" group. Instead we
5012 rely on the journal directory to be owned by the
5013 "systemd-journal" group, and its setgid bit set, so that the
5014 kernel file system layer will automatically enforce that
5015 journal files inherit this group assignment. The reason for
5016 this change is that we cannot allow NSS look-ups from
5017 journald which would be necessary to resolve
5018 "systemd-journal" to a numeric GID, because this might
5019 create deadlocks if NSS involves synchronous queries to
5020 other daemons (such as nscd, or sssd) which in turn are
5021 logging clients of journald and might block on it, which
5022 would then dead lock. A tmpfiles.d(5) snippet included in
5023 systemd will make sure the setgid bit and group are
5024 properly set on the journal directory if it exists on every
5025 boot. However, we recommend adjusting it manually after
5026 upgrades too (or from RPM scriptlets), so that the change is
5027 not delayed until next reboot.
5029 * Backlight and random seed files in /var/lib/ have moved into
5030 the /var/lib/systemd/ directory, in order to centralize all
5031 systemd generated files in one directory.
5033 * Boot time performance measurements (as displayed by
5034 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5035 performance information if that's available to determine how
5036 much time BIOS and boot loader initialization required. With
5037 a sufficiently new BIOS you hence no longer need to boot
5038 with Gummiboot to get access to such information.
5040 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5041 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5042 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5043 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5044 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5045 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5046 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5048 — Berlin, 2013-10-02
5052 * The Restart= option for services now understands a new
5053 on-watchdog setting, which will restart the service
5054 automatically if the service stops sending out watchdog keep
5055 alive messages (as configured with WatchdogSec=).
5057 * The getty generator (which is responsible for bringing up a
5058 getty on configured serial consoles) will no longer only
5059 start a getty on the primary kernel console but on all
5060 others, too. This makes the order in which console= is
5061 specified on the kernel command line less important.
5063 * libsystemd-logind gained a new sd_session_get_vt() call to
5064 retrieve the VT number of a session.
5066 * If the option "tries=0" is set for an entry of /etc/crypttab
5067 its passphrase is queried indefinitely instead of any
5068 maximum number of tries.
5070 * If a service with a configure PID file terminates its PID
5071 file will now be removed automatically if it still exists
5072 afterwards. This should put an end to stale PID files.
5074 * systemd-run will now also take relative binary path names
5075 for execution and no longer insists on absolute paths.
5077 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5078 paths that are optionally prefixed with "-" to indicate that
5079 it should not be considered a failure if they do not exist.
5081 * journalctl -o (and similar commands) now understands a new
5082 output mode "short-precise", it is similar to "short" but
5083 shows timestamps with usec accuracy.
5085 * The option "discard" (as known from Debian) is now
5086 synonymous to "allow-discards" in /etc/crypttab. In fact,
5087 "discard" is preferred now (since it is easier to remember
5090 * Some licensing clean-ups were made, so that more code is now
5091 LGPL-2.1 licensed than before.
5093 * A minimal tool to save/restore the display backlight
5094 brightness across reboots has been added. It will store the
5095 backlight setting as late as possible at shutdown, and
5096 restore it as early as possible during reboot.
5098 * A logic to automatically discover and enable home and swap
5099 partitions on GPT disks has been added. With this in place
5100 /etc/fstab becomes optional for many setups as systemd can
5101 discover certain partitions located on the root disk
5102 automatically. Home partitions are recognized under their
5103 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5104 partitions are recognized under their GPT type ID
5105 0657fd6da4ab43c484e50933c84b4f4f.
5107 * systemd will no longer pass any environment from the kernel
5108 or initrd to system services. If you want to set an
5109 environment for all services, do so via the kernel command
5110 line systemd.setenv= assignment.
5112 * The systemd-sysctl tool no longer natively reads the file
5113 /etc/sysctl.conf. If desired, the file should be symlinked
5114 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5115 legacy support by a symlink rather than built-in code, it
5116 also makes the otherwise hidden order of application of the
5117 different files visible. (Note that this partly reverts to a
5118 pre-198 application order of sysctl knobs!)
5120 * The "systemctl set-log-level" and "systemctl dump" commands
5121 have been moved to systemd-analyze.
5123 * systemd-run learned the new --remain-after-exit switch,
5124 which causes the scope unit not to be cleaned up
5125 automatically after the process terminated.
5127 * tmpfiles learned a new --exclude-prefix= switch to exclude
5128 certain paths from operation.
5130 * journald will now automatically flush all messages to disk
5131 as soon as a message at the log level CRIT, ALERT or EMERG
5134 Contributions from: Andrew Cook, Brandon Philips, Christian
5135 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5136 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5137 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5138 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5139 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5140 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5141 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5142 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5143 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5144 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5145 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5146 William Giokas, Zbigniew Jędrzejewski-Szmek
5148 — Berlin, 2013-09-13
5152 * The documentation has been updated to cover the various new
5153 concepts introduced with 205.
5155 * Unit files now understand the new %v specifier which
5156 resolves to the kernel version string as returned by "uname
5159 * systemctl now supports filtering the unit list output by
5160 load state, active state and sub state, using the new
5163 * "systemctl status" will now show the results of the
5164 condition checks (like ConditionPathExists= and similar) of
5165 the last start attempts of the unit. They are also logged to
5168 * "journalctl -b" may now be used to look for boot output of a
5169 specific boot. Try "journalctl -b -1" for the previous boot,
5170 but the syntax is substantially more powerful.
5172 * "journalctl --show-cursor" has been added which prints the
5173 cursor string the last shown log line. This may then be used
5174 with the new "journalctl --after-cursor=" switch to continue
5175 browsing logs from that point on.
5177 * "journalctl --force" may now be used to force regeneration
5180 * Creation of "dead" device nodes has been moved from udev
5181 into kmod and tmpfiles. Previously, udev would read the kmod
5182 databases to pre-generate dead device nodes based on meta
5183 information contained in kernel modules, so that these would
5184 be auto-loaded on access rather then at boot. As this
5185 does not really have much to do with the exposing actual
5186 kernel devices to userspace this has always been slightly
5187 alien in the udev codebase. Following the new scheme kmod
5188 will now generate a runtime snippet for tmpfiles from the
5189 module meta information and it now is tmpfiles' job to the
5190 create the nodes. This also allows overriding access and
5191 other parameters for the nodes using the usual tmpfiles
5192 facilities. As side effect this allows us to remove the
5193 CAP_SYS_MKNOD capability bit from udevd entirely.
5195 * logind's device ACLs may now be applied to these "dead"
5196 devices nodes too, thus finally allowing managed access to
5197 devices such as /dev/snd/sequencer without loading the
5198 backing module right-away.
5200 * A new RPM macro has been added that may be used to apply
5201 tmpfiles configuration during package installation.
5203 * systemd-detect-virt and ConditionVirtualization= now can
5204 detect User-Mode-Linux machines (UML).
5206 * journald will now implicitly log the effective capabilities
5207 set of processes in the message metadata.
5209 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5211 * The initrd interface has been simplified (more specifically,
5212 support for passing performance data via environment
5213 variables and fsck results via files in /run has been
5214 removed). These features were non-essential, and are
5215 nowadays available in a much nicer way by having systemd in
5216 the initrd serialize its state and have the hosts systemd
5217 deserialize it again.
5219 * The udev "keymap" data files and tools to apply keyboard
5220 specific mappings of scan to key codes, and force-release
5221 scan code lists have been entirely replaced by a udev
5222 "keyboard" builtin and a hwdb data file.
5224 * systemd will now honour the kernel's "quiet" command line
5225 argument also during late shutdown, resulting in a
5226 completely silent shutdown when used.
5228 * There's now an option to control the SO_REUSEPORT socket
5229 option in .socket units.
5231 * Instance units will now automatically get a per-template
5232 subslice of system.slice unless something else is explicitly
5233 configured. For example, instances of sshd@.service will now
5234 implicitly be placed in system-sshd.slice rather than
5235 system.slice as before.
5237 * Test coverage support may now be enabled at build time.
5239 Contributions from: Dave Reisner, Frederic Crozat, Harald
5240 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5241 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5242 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5243 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5244 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5245 Giokas, Zbigniew Jędrzejewski-Szmek
5247 — Berlin, 2013-07-23
5251 * Two new unit types have been introduced:
5253 Scope units are very similar to service units, however, are
5254 created out of pre-existing processes — instead of PID 1
5255 forking off the processes. By using scope units it is
5256 possible for system services and applications to group their
5257 own child processes (worker processes) in a powerful way
5258 which then maybe used to organize them, or kill them
5259 together, or apply resource limits on them.
5261 Slice units may be used to partition system resources in an
5262 hierarchical fashion and then assign other units to them. By
5263 default there are now three slices: system.slice (for all
5264 system services), user.slice (for all user sessions),
5265 machine.slice (for VMs and containers).
5267 Slices and scopes have been introduced primarily in
5268 context of the work to move cgroup handling to a
5269 single-writer scheme, where only PID 1
5270 creates/removes/manages cgroups.
5272 * There's a new concept of "transient" units. In contrast to
5273 normal units these units are created via an API at runtime,
5274 not from configuration from disk. More specifically this
5275 means it is now possible to run arbitrary programs as
5276 independent services, with all execution parameters passed
5277 in via bus APIs rather than read from disk. Transient units
5278 make systemd substantially more dynamic then it ever was,
5279 and useful as a general batch manager.
5281 * logind has been updated to make use of scope and slice units
5282 for managing user sessions. As a user logs in he will get
5283 his own private slice unit, to which all sessions are added
5284 as scope units. We also added support for automatically
5285 adding an instance of user@.service for the user into the
5286 slice. Effectively logind will no longer create cgroup
5287 hierarchies on its own now, it will defer entirely to PID 1
5288 for this by means of scope, service and slice units. Since
5289 user sessions this way become entities managed by PID 1
5290 the output of "systemctl" is now a lot more comprehensive.
5292 * A new mini-daemon "systemd-machined" has been added which
5293 may be used by virtualization managers to register local
5294 VMs/containers. nspawn has been updated accordingly, and
5295 libvirt will be updated shortly. machined will collect a bit
5296 of meta information about the VMs/containers, and assign
5297 them their own scope unit (see above). The collected
5298 meta-data is then made available via the "machinectl" tool,
5299 and exposed in "ps" and similar tools. machined/machinectl
5300 is compile-time optional.
5302 * As discussed earlier, the low-level cgroup configuration
5303 options ControlGroup=, ControlGroupModify=,
5304 ControlGroupPersistent=, ControlGroupAttribute= have been
5305 removed. Please use high-level attribute settings instead as
5306 well as slice units.
5308 * A new bus call SetUnitProperties() has been added to alter
5309 various runtime parameters of a unit. This is primarily
5310 useful to alter cgroup parameters dynamically in a nice way,
5311 but will be extended later on to make more properties
5312 modifiable at runtime. systemctl gained a new set-properties
5313 command that wraps this call.
5315 * A new tool "systemd-run" has been added which can be used to
5316 run arbitrary command lines as transient services or scopes,
5317 while configuring a number of settings via the command
5318 line. This tool is currently very basic, however already
5319 very useful. We plan to extend this tool to even allow
5320 queuing of execution jobs with time triggers from the
5321 command line, similar in fashion to "at".
5323 * nspawn will now inform the user explicitly that kernels with
5324 audit enabled break containers, and suggest the user to turn
5327 * Support for detecting the IMA and AppArmor security
5328 frameworks with ConditionSecurity= has been added.
5330 * journalctl gained a new "-k" switch for showing only kernel
5331 messages, mimicking dmesg output; in addition to "--user"
5332 and "--system" switches for showing only user's own logs
5335 * systemd-delta can now show information about drop-in
5336 snippets extending unit files.
5338 * libsystemd-bus has been substantially updated but is still
5339 not available as public API.
5341 * systemd will now look for the "debug" argument on the kernel
5342 command line and enable debug logging, similar to what
5343 "systemd.log_level=debug" already did before.
5345 * "systemctl set-default", "systemctl get-default" has been
5346 added to configure the default.target symlink, which
5347 controls what to boot into by default.
5349 * "systemctl set-log-level" has been added as a convenient
5350 way to raise and lower systemd logging threshold.
5352 * "systemd-analyze plot" will now show the time the various
5353 generators needed for execution, as well as information
5354 about the unit file loading.
5356 * libsystemd-journal gained a new sd_journal_open_files() call
5357 for opening specific journal files. journactl also gained a
5358 new switch to expose this new functionality. Previously we
5359 only supported opening all files from a directory, or all
5360 files from the system, as opening individual files only is
5361 racy due to journal file rotation.
5363 * systemd gained the new DefaultEnvironment= setting in
5364 /etc/systemd/system.conf to set environment variables for
5367 * If a privileged process logs a journal message with the
5368 OBJECT_PID= field set, then journald will automatically
5369 augment this with additional OBJECT_UID=, OBJECT_GID=,
5370 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5371 system services want to log events about specific client
5372 processes. journactl/systemctl has been updated to make use
5373 of this information if all log messages regarding a specific
5376 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5377 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5378 Reisner, David Coppa, David King, David Strauss, Eelco
5379 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5380 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5381 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5382 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5383 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5384 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5385 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5386 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5387 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5388 Łukasz Stelmach, 장동준
5392 * The Python bindings gained some minimal support for the APIs
5393 exposed by libsystemd-logind.
5395 * ConditionSecurity= gained support for detecting SMACK. Since
5396 this condition already supports SELinux and AppArmor we only
5397 miss IMA for this. Patches welcome!
5399 Contributions from: Karol Lewandowski, Lennart Poettering,
5400 Zbigniew Jędrzejewski-Szmek
5404 * systemd-nspawn will now create /etc/resolv.conf if
5405 necessary, before bind-mounting the host's file onto it.
5407 * systemd-nspawn will now store meta information about a
5408 container on the container's cgroup as extended attribute
5409 fields, including the root directory.
5411 * The cgroup hierarchy has been reworked in many ways. All
5412 objects any of the components systemd creates in the cgroup
5413 tree are now suffixed. More specifically, user sessions are
5414 now placed in cgroups suffixed with ".session", users in
5415 cgroups suffixed with ".user", and nspawn containers in
5416 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5417 names are now escaped in a simple scheme to avoid collision
5418 of userspace object names with kernel filenames. This work
5419 is preparation for making these objects relocatable in the
5420 cgroup tree, in order to allow easy resource partitioning of
5421 these objects without causing naming conflicts.
5423 * systemctl list-dependencies gained the new switches
5424 --plain, --reverse, --after and --before.
5426 * systemd-inhibit now shows the process name of processes that
5427 have taken an inhibitor lock.
5429 * nss-myhostname will now also resolve "localhost"
5430 implicitly. This makes /etc/hosts an optional file and
5431 nicely handles that on IPv6 ::1 maps to both "localhost" and
5434 * libsystemd-logind.so gained a new call
5435 sd_get_machine_names() to enumerate running containers and
5436 VMs (currently only supported by very new libvirt and
5437 nspawn). sd_login_monitor can now be used to watch
5438 VMs/containers coming and going.
5440 * .include is not allowed recursively anymore, and only in
5441 unit files. Usually it is better to use drop-in snippets in
5442 .d/*.conf anyway, as introduced with systemd 198.
5444 * systemd-analyze gained a new "critical-chain" command that
5445 determines the slowest chain of units run during system
5446 boot-up. It is very useful for tracking down where
5447 optimizing boot time is the most beneficial.
5449 * systemd will no longer allow manipulating service paths in
5450 the name=systemd:/system cgroup tree using ControlGroup= in
5451 units. (But is still fine with it in all other dirs.)
5453 * There's a new systemd-nspawn@.service service file that may
5454 be used to easily run nspawn containers as system
5455 services. With the container's root directory in
5456 /var/lib/container/foobar it is now sufficient to run
5457 "systemctl start systemd-nspawn@foobar.service" to boot it.
5459 * systemd-cgls gained a new parameter "--machine" to list only
5460 the processes within a certain container.
5462 * ConditionSecurity= now can check for "apparmor". We still
5463 are lacking checks for SMACK and IMA for this condition
5464 check though. Patches welcome!
5466 * A new configuration file /etc/systemd/sleep.conf has been
5467 added that may be used to configure which kernel operation
5468 systemd is supposed to execute when "suspend", "hibernate"
5469 or "hybrid-sleep" is requested. This makes the new kernel
5470 "freeze" state accessible to the user.
5472 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5473 the passed argument if applicable.
5475 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5476 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5477 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5478 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5479 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5480 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5481 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5486 * The output of 'systemctl list-jobs' got some polishing. The
5487 '--type=' argument may now be passed more than once. A new
5488 command 'systemctl list-sockets' has been added which shows
5489 a list of kernel sockets systemd is listening on with the
5490 socket units they belong to, plus the units these socket
5493 * The experimental libsystemd-bus library got substantial
5494 updates to work in conjunction with the (also experimental)
5495 kdbus kernel project. It works well enough to exchange
5496 messages with some sophistication. Note that kdbus is not
5497 ready yet, and the library is mostly an elaborate test case
5498 for now, and not installable.
5500 * systemd gained a new unit 'systemd-static-nodes.service'
5501 that generates static device nodes earlier during boot, and
5502 can run in conjunction with udev.
5504 * libsystemd-login gained a new call sd_pid_get_user_unit()
5505 to retrieve the user systemd unit a process is running
5506 in. This is useful for systems where systemd is used as
5509 * systemd-nspawn now places all containers in the new /machine
5510 top-level cgroup directory in the name=systemd
5511 hierarchy. libvirt will soon do the same, so that we get a
5512 uniform separation of /system, /user and /machine for system
5513 services, user processes and containers/virtual
5514 machines. This new cgroup hierarchy is also useful to stick
5515 stable names to specific container instances, which can be
5516 recognized later this way (this name may be controlled
5517 via systemd-nspawn's new -M switch). libsystemd-login also
5518 gained a new call sd_pid_get_machine_name() to retrieve the
5519 name of the container/VM a specific process belongs to.
5521 * bootchart can now store its data in the journal.
5523 * libsystemd-journal gained a new call
5524 sd_journal_add_conjunction() for AND expressions to the
5525 matching logic. This can be used to express more complex
5526 logical expressions.
5528 * journactl can now take multiple --unit= and --user-unit=
5531 * The cryptsetup logic now understands the "luks.key=" kernel
5532 command line switch for specifying a file to read the
5533 decryption key from. Also, if a configured key file is not
5534 found the tool will now automatically fall back to prompting
5537 * Python systemd.journal module was updated to wrap recently
5538 added functions from libsystemd-journal. The interface was
5539 changed to bring the low level interface in s.j._Reader
5540 closer to the C API, and the high level interface in
5541 s.j.Reader was updated to wrap and convert all data about
5544 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5545 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5546 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5547 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5548 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5549 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5553 * journalctl --update-catalog now understands a new --root=
5554 option to operate on catalogs found in a different root
5557 * During shutdown after systemd has terminated all running
5558 services a final killing loop kills all remaining left-over
5559 processes. We will now print the name of these processes
5560 when we send SIGKILL to them, since this usually indicates a
5563 * If /etc/crypttab refers to password files stored on
5564 configured mount points automatic dependencies will now be
5565 generated to ensure the specific mount is established first
5566 before the key file is attempted to be read.
5568 * 'systemctl status' will now show information about the
5569 network sockets a socket unit is listening on.
5571 * 'systemctl status' will also shown information about any
5572 drop-in configuration file for units. (Drop-In configuration
5573 files in this context are files such as
5574 /etc/systemd/systemd/foobar.service.d/*.conf)
5576 * systemd-cgtop now optionally shows summed up CPU times of
5577 cgroups. Press '%' while running cgtop to switch between
5578 percentage and absolute mode. This is useful to determine
5579 which cgroups use up the most CPU time over the entire
5580 runtime of the system. systemd-cgtop has also been updated
5581 to be 'pipeable' for processing with further shell tools.
5583 * 'hostnamectl set-hostname' will now allow setting of FQDN
5586 * The formatting and parsing of time span values has been
5587 changed. The parser now understands fractional expressions
5588 such as "5.5h". The formatter will now output fractional
5589 expressions for all time spans under 1min, i.e. "5.123456s"
5590 rather than "5s 123ms 456us". For time spans under 1s
5591 millisecond values are shown, for those under 1ms
5592 microsecond values are shown. This should greatly improve
5593 all time-related output of systemd.
5595 * libsystemd-login and libsystemd-journal gained new
5596 functions for querying the poll() events mask and poll()
5597 timeout value for integration into arbitrary event
5600 * localectl gained the ability to list available X11 keymaps
5601 (models, layouts, variants, options).
5603 * 'systemd-analyze dot' gained the ability to filter for
5604 specific units via shell-style globs, to create smaller,
5605 more useful graphs. I.e. it is now possible to create simple
5606 graphs of all the dependencies between only target units, or
5607 of all units that Avahi has dependencies with.
5609 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5610 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5611 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5612 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5613 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5614 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5615 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5619 * The boot-time readahead implementation for rotating media
5620 will now read the read-ahead data in multiple passes which
5621 consist of all read requests made in equidistant time
5622 intervals. This means instead of strictly reading read-ahead
5623 data in its physical order on disk we now try to find a
5624 middle ground between physical and access time order.
5626 * /etc/os-release files gained a new BUILD_ID= field for usage
5627 on operating systems that provide continuous builds of OS
5630 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5631 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5632 William Douglas, Zbigniew Jędrzejewski-Szmek
5636 * systemd-python gained an API exposing libsystemd-daemon.
5638 * The SMACK setup logic gained support for uploading CIPSO
5641 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5642 ReadOnlyDirectories= and InaccessibleDirectories= has
5643 changed. The private /tmp and /var/tmp directories are now
5644 shared by all processes of a service (which means
5645 ExecStartPre= may now leave data in /tmp that ExecStart= of
5646 the same service can still access). When a service is
5647 stopped its temporary directories are immediately deleted
5648 (normal clean-up with tmpfiles is still done in addition to
5651 * By default, systemd will now set a couple of sysctl
5652 variables in the kernel: the safe sysrq options are turned
5653 on, IP route verification is turned on, and source routing
5654 disabled. The recently added hardlink and softlink
5655 protection of the kernel is turned on. These settings should
5656 be reasonably safe, and good defaults for all new systems.
5658 * The predictable network naming logic may now be turned off
5659 with a new kernel command line switch: net.ifnames=0.
5661 * A new libsystemd-bus module has been added that implements a
5662 pretty complete D-Bus client library. For details see:
5664 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5666 * journald will now explicitly flush the journal files to disk
5667 at the latest 5min after each write. The file will then also
5668 be marked offline until the next write. This should increase
5669 reliability in case of a crash. The synchronization delay
5670 can be configured via SyncIntervalSec= in journald.conf.
5672 * There's a new remote-fs-setup.target unit that can be used
5673 to pull in specific services when at least one remote file
5674 system is to be mounted.
5676 * There are new targets timers.target and paths.target as
5677 canonical targets to pull user timer and path units in
5678 from. This complements sockets.target with a similar
5679 purpose for socket units.
5681 * libudev gained a new call udev_device_set_attribute_value()
5682 to set sysfs attributes of a device.
5684 * The udev daemon now sets the default number of worker
5685 processes executed in parallel based on the number of available
5686 CPUs instead of the amount of available RAM. This is supposed
5687 to provide a more reliable default and limit a too aggressive
5688 parallelism for setups with 1000s of devices connected.
5690 Contributions from: Auke Kok, Colin Walters, Cristian
5691 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5692 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5693 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5694 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5695 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5696 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5697 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5698 Zbigniew Jędrzejewski-Szmek
5702 * Configuration of unit files may now be extended via drop-in
5703 files without having to edit/override the unit files
5704 themselves. More specifically, if the administrator wants to
5705 change one value for a service file foobar.service he can
5706 now do so by dropping in a configuration snippet into
5707 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5708 will load all these snippets and apply them on top of the
5709 main unit configuration file, possibly extending or
5710 overriding its settings. Using these drop-in snippets is
5711 generally nicer than the two earlier options for changing
5712 unit files locally: copying the files from
5713 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5714 them there; or creating a new file in /etc/systemd/system/
5715 that incorporates the original one via ".include". Drop-in
5716 snippets into these .d/ directories can be placed in any
5717 directory systemd looks for units in, and the usual
5718 overriding semantics between /usr/lib, /etc and /run apply
5721 * Most unit file settings which take lists of items can now be
5722 reset by assigning the empty string to them. For example,
5723 normally, settings such as Environment=FOO=BAR append a new
5724 environment variable assignment to the environment block,
5725 each time they are used. By assigning Environment= the empty
5726 string the environment block can be reset to empty. This is
5727 particularly useful with the .d/*.conf drop-in snippets
5728 mentioned above, since this adds the ability to reset list
5729 settings from vendor unit files via these drop-ins.
5731 * systemctl gained a new "list-dependencies" command for
5732 listing the dependencies of a unit recursively.
5734 * Inhibitors are now honored and listed by "systemctl
5735 suspend", "systemctl poweroff" (and similar) too, not only
5736 GNOME. These commands will also list active sessions by
5739 * Resource limits (as exposed by the various control group
5740 controllers) can now be controlled dynamically at runtime
5741 for all units. More specifically, you can now use a command
5742 like "systemctl set-cgroup-attr foobar.service cpu.shares
5743 2000" to alter the CPU shares a specific service gets. These
5744 settings are stored persistently on disk, and thus allow the
5745 administrator to easily adjust the resource usage of
5746 services with a few simple commands. This dynamic resource
5747 management logic is also available to other programs via the
5748 bus. Almost any kernel cgroup attribute and controller is
5751 * systemd-vconsole-setup will now copy all font settings to
5752 all allocated VTs, where it previously applied them only to
5755 * libsystemd-login gained the new sd_session_get_tty() API
5758 * This release drops support for a few legacy or
5759 distribution-specific LSB facility names when parsing init
5760 scripts: $x-display-manager, $mail-transfer-agent,
5761 $mail-transport-agent, $mail-transfer-agent, $smtp,
5762 $null. Also, the mail-transfer-agent.target unit backing
5763 this has been removed. Distributions which want to retain
5764 compatibility with this should carry the burden for
5765 supporting this themselves and patch support for these back
5766 in, if they really need to. Also, the facilities $syslog and
5767 $local_fs are now ignored, since systemd does not support
5768 early-boot LSB init scripts anymore, and these facilities
5769 are implied anyway for normal services. syslog.target has
5772 * There are new bus calls on PID1's Manager object for
5773 cancelling jobs, and removing snapshot units. Previously,
5774 both calls were only available on the Job and Snapshot
5777 * systemd-journal-gatewayd gained SSL support.
5779 * The various "environment" files, such as /etc/locale.conf
5780 now support continuation lines with a backslash ("\") as
5781 last character in the line, similarly in style (but different)
5782 to how this is supported in shells.
5784 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5785 now implicitly appended to every log entry logged. systemctl
5786 has been updated to filter by this field when operating on a
5787 user systemd instance.
5789 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5790 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5791 the container. This makes it easier to boot unmodified
5792 Fedora systems in a container, which however still requires
5793 audit=0 to be passed on the kernel command line. Auditing in
5794 kernel and userspace is unfortunately still too broken in
5795 context of containers, hence we recommend compiling it out
5796 of the kernel or using audit=0. Hopefully this will be fixed
5797 one day for good in the kernel.
5799 * nspawn gained the new --bind= and --bind-ro= parameters to
5800 bind mount specific directories from the host into the
5803 * nspawn will now mount its own devpts file system instance
5804 into the container, in order not to leak pty devices from
5805 the host into the container.
5807 * systemd will now read the firmware boot time performance
5808 information from the EFI variables, if the used boot loader
5809 supports this, and takes it into account for boot performance
5810 analysis via "systemd-analyze". This is currently supported
5811 only in conjunction with Gummiboot, but could be supported
5812 by other boot loaders too. For details see:
5814 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5816 * A new generator has been added that automatically mounts the
5817 EFI System Partition (ESP) to /boot, if that directory
5818 exists, is empty, and no other file system has been
5819 configured to be mounted there.
5821 * logind will now send out PrepareForSleep(false) out
5822 unconditionally, after coming back from suspend. This may be
5823 used by applications as asynchronous notification for
5824 system resume events.
5826 * "systemctl unlock-sessions" has been added, that allows
5827 unlocking the screens of all user sessions at once, similar
5828 to how "systemctl lock-sessions" already locked all users
5829 sessions. This is backed by a new D-Bus call UnlockSessions().
5831 * "loginctl seat-status" will now show the master device of a
5832 seat. (i.e. the device of a seat that needs to be around for
5833 the seat to be considered available, usually the graphics
5836 * tmpfiles gained a new "X" line type, that allows
5837 configuration of files and directories (with wildcards) that
5838 shall be excluded from automatic cleanup ("aging").
5840 * udev default rules set the device node permissions now only
5841 at "add" events, and do not change them any longer with a
5842 later "change" event.
5844 * The log messages for lid events and power/sleep keypresses
5845 now carry a message ID.
5847 * We now have a substantially larger unit test suite, but this
5848 continues to be work in progress.
5850 * udevadm hwdb gained a new --root= parameter to change the
5851 root directory to operate relative to.
5853 * logind will now issue a background sync() request to the kernel
5854 early at shutdown, so that dirty buffers are flushed to disk early
5855 instead of at the last moment, in order to optimize shutdown
5858 * A new bootctl tool has been added that is an interface for
5859 certain boot loader operations. This is currently a preview
5860 and is likely to be extended into a small mechanism daemon
5861 like timedated, localed, hostnamed, and can be used by
5862 graphical UIs to enumerate available boot options, and
5863 request boot into firmware operations.
5865 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5866 the rest of the package. It also has been updated to work
5867 correctly in initrds.
5869 * Policykit previously has been runtime optional, and is now
5870 also compile time optional via a configure switch.
5872 * systemd-analyze has been reimplemented in C. Also "systemctl
5873 dot" has moved into systemd-analyze.
5875 * "systemctl status" with no further parameters will now print
5876 the status of all active or failed units.
5878 * Operations such as "systemctl start" can now be executed
5879 with a new mode "--irreversible" which may be used to queue
5880 operations that cannot accidentally be reversed by a later
5881 job queuing. This is by default used to make shutdown
5882 requests more robust.
5884 * The Python API of systemd now gained a new module for
5885 reading journal files.
5887 * A new tool kernel-install has been added that can install
5888 kernel images according to the Boot Loader Specification:
5890 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5892 * Boot time console output has been improved to provide
5893 animated boot time output for hanging jobs.
5895 * A new tool systemd-activate has been added which can be used
5896 to test socket activation with, directly from the command
5897 line. This should make it much easier to test and debug
5898 socket activation in daemons.
5900 * journalctl gained a new "--reverse" (or -r) option to show
5901 journal output in reverse order (i.e. newest line first).
5903 * journalctl gained a new "--pager-end" (or -e) option to jump
5904 to immediately jump to the end of the journal in the
5905 pager. This is only supported in conjunction with "less".
5907 * journalctl gained a new "--user-unit=" option, that works
5908 similarly to "--unit=" but filters for user units rather than
5911 * A number of unit files to ease adoption of systemd in
5912 initrds has been added. This moves some minimal logic from
5913 the various initrd implementations into systemd proper.
5915 * The journal files are now owned by a new group
5916 "systemd-journal", which exists specifically to allow access
5917 to the journal, and nothing else. Previously, we used the
5918 "adm" group for that, which however possibly covers more
5919 than just journal/log file access. This new group is now
5920 already used by systemd-journal-gatewayd to ensure this
5921 daemon gets access to the journal files and as little else
5922 as possible. Note that "make install" will also set FS ACLs
5923 up for /var/log/journal to give "adm" and "wheel" read
5924 access to it, in addition to "systemd-journal" which owns
5925 the journal files. We recommend that packaging scripts also
5926 add read access to "adm" + "wheel" to /var/log/journal, and
5927 all existing/future journal files. To normal users and
5928 administrators little changes, however packagers need to
5929 ensure to create the "systemd-journal" system group at
5930 package installation time.
5932 * The systemd-journal-gatewayd now runs as unprivileged user
5933 systemd-journal-gateway:systemd-journal-gateway. Packaging
5934 scripts need to create these system user/group at
5937 * timedated now exposes a new boolean property CanNTP that
5938 indicates whether a local NTP service is available or not.
5940 * systemd-detect-virt will now also detect xen PVs
5942 * The pstore file system is now mounted by default, if it is
5945 * In addition to the SELinux and IMA policies we will now also
5946 load SMACK policies at early boot.
5948 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5949 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5950 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5951 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5952 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5953 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5954 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5955 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5956 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5957 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5958 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5959 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5960 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5961 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5965 * Timer units now support calendar time events in addition to
5966 monotonic time events. That means you can now trigger a unit
5967 based on a calendar time specification such as "Thu,Fri
5968 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5969 or fifth day of any month of the year 2013, given that it is
5970 a thursday or friday. This brings timer event support
5971 considerably closer to cron's capabilities. For details on
5972 the supported calendar time specification language see
5975 * udev now supports a number of different naming policies for
5976 network interfaces for predictable names, and a combination
5977 of these policies is now the default. Please see this wiki
5978 document for details:
5980 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5982 * Auke Kok's bootchart implementation has been added to the
5983 systemd tree. It is an optional component that can graph the
5984 boot in quite some detail. It is one of the best bootchart
5985 implementations around and minimal in its code and
5988 * nss-myhostname has been integrated into the systemd source
5989 tree. nss-myhostname guarantees that the local hostname
5990 always stays resolvable via NSS. It has been a weak
5991 requirement of systemd-hostnamed since a long time, and
5992 since its code is actually trivial we decided to just
5993 include it in systemd's source tree. It can be turned off
5994 with a configure switch.
5996 * The read-ahead logic is now capable of properly detecting
5997 whether a btrfs file system is on SSD or rotating media, in
5998 order to optimize the read-ahead scheme. Previously, it was
5999 only capable of detecting this on traditional file systems
6002 * In udev, additional device properties are now read from the
6003 IAB in addition to the OUI database. Also, Bluetooth company
6004 identities are attached to the devices as well.
6006 * In service files %U may be used as specifier that is
6007 replaced by the configured user name of the service.
6009 * nspawn may now be invoked without a controlling TTY. This
6010 makes it suitable for invocation as its own service. This
6011 may be used to set up a simple containerized server system
6012 using only core OS tools.
6014 * systemd and nspawn can now accept socket file descriptors
6015 when they are started for socket activation. This enables
6016 implementation of socket activated nspawn
6017 containers. i.e. think about autospawning an entire OS image
6018 when the first SSH or HTTP connection is received. We expect
6019 that similar functionality will also be added to libvirt-lxc
6022 * journalctl will now suppress ANSI color codes when
6023 presenting log data.
6025 * systemctl will no longer show control group information for
6026 a unit if the control group is empty anyway.
6028 * logind can now automatically suspend/hibernate/shutdown the
6031 * /etc/machine-info and hostnamed now also expose the chassis
6032 type of the system. This can be used to determine whether
6033 the local system is a laptop, desktop, handset or
6034 tablet. This information may either be configured by the
6035 user/vendor or is automatically determined from ACPI and DMI
6036 information if possible.
6038 * A number of PolicyKit actions are now bound together with
6039 "imply" rules. This should simplify creating UIs because
6040 many actions will now authenticate similar ones as well.
6042 * Unit files learnt a new condition ConditionACPower= which
6043 may be used to conditionalize a unit depending on whether an
6044 AC power source is connected or not, of whether the system
6045 is running on battery power.
6047 * systemctl gained a new "is-failed" verb that may be used in
6048 shell scripts and suchlike to check whether a specific unit
6049 is in the "failed" state.
6051 * The EnvironmentFile= setting in unit files now supports file
6052 globbing, and can hence be used to easily read a number of
6053 environment files at once.
6055 * systemd will no longer detect and recognize specific
6056 distributions. All distribution-specific #ifdeffery has been
6057 removed, systemd is now fully generic and
6058 distribution-agnostic. Effectively, not too much is lost as
6059 a lot of the code is still accessible via explicit configure
6060 switches. However, support for some distribution specific
6061 legacy configuration file formats has been dropped. We
6062 recommend distributions to simply adopt the configuration
6063 files everybody else uses now and convert the old
6064 configuration from packaging scripts. Most distributions
6065 already did that. If that's not possible or desirable,
6066 distributions are welcome to forward port the specific
6067 pieces of code locally from the git history.
6069 * When logging a message about a unit systemd will now always
6070 log the unit name in the message meta data.
6072 * localectl will now also discover system locale data that is
6073 not stored in locale archives, but directly unpacked.
6075 * logind will no longer unconditionally use framebuffer
6076 devices as seat masters, i.e. as devices that are required
6077 to be existing before a seat is considered preset. Instead,
6078 it will now look for all devices that are tagged as
6079 "seat-master" in udev. By default, framebuffer devices will
6080 be marked as such, but depending on local systems, other
6081 devices might be marked as well. This may be used to
6082 integrate graphics cards using closed source drivers (such
6083 as NVidia ones) more nicely into logind. Note however, that
6084 we recommend using the open source NVidia drivers instead,
6085 and no udev rules for the closed-source drivers will be
6086 shipped from us upstream.
6088 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6089 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6090 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6091 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6092 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6093 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6094 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6095 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6096 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6097 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6098 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6099 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6104 * udev gained support for loading additional device properties
6105 from an indexed database that is keyed by vendor/product IDs
6106 and similar device identifiers. For the beginning this
6107 "hwdb" is populated with data from the well-known PCI and
6108 USB database, but also includes PNP, ACPI and OID data. In
6109 the longer run this indexed database shall grow into
6110 becoming the one central database for non-essential
6111 userspace device metadata. Previously, data from the PCI/USB
6112 database was only attached to select devices, since the
6113 lookup was a relatively expensive operation due to O(n) time
6114 complexity (with n being the number of entries in the
6115 database). Since this is now O(1), we decided to add in this
6116 data for all devices where this is available, by
6117 default. Note that the indexed database needs to be rebuilt
6118 when new data files are installed. To achieve this you need
6119 to update your packaging scripts to invoke "udevadm hwdb
6120 --update" after installation of hwdb data files. For
6121 RPM-based distributions we introduced the new
6122 %udev_hwdb_update macro for this purpose.
6124 * The Journal gained support for the "Message Catalog", an
6125 indexed database to link up additional information with
6126 journal entries. For further details please check:
6128 https://www.freedesktop.org/wiki/Software/systemd/catalog
6130 The indexed message catalog database also needs to be
6131 rebuilt after installation of message catalog files. Use
6132 "journalctl --update-catalog" for this. For RPM-based
6133 distributions we introduced the %journal_catalog_update
6134 macro for this purpose.
6136 * The Python Journal bindings gained support for the standard
6137 Python logging framework.
6139 * The Journal API gained new functions for checking whether
6140 the underlying file system of a journal file is capable of
6141 properly reporting file change notifications, or whether
6142 applications that want to reflect journal changes "live"
6143 need to recheck journal files continuously in appropriate
6146 * It is now possible to set the "age" field for tmpfiles
6147 entries to 0, indicating that files matching this entry
6148 shall always be removed when the directories are cleaned up.
6150 * coredumpctl gained a new "gdb" verb which invokes gdb
6151 right-away on the selected coredump.
6153 * There's now support for "hybrid sleep" on kernels that
6154 support this, in addition to "suspend" and "hibernate". Use
6155 "systemctl hybrid-sleep" to make use of this.
6157 * logind's HandleSuspendKey= setting (and related settings)
6158 now gained support for a new "lock" setting to simply
6159 request the screen lock on all local sessions, instead of
6160 actually executing a suspend or hibernation.
6162 * systemd will now mount the EFI variables file system by
6165 * Socket units now gained support for configuration of the
6166 SMACK security label.
6168 * timedatectl will now output the time of the last and next
6169 daylight saving change.
6171 * We dropped support for various legacy and distro-specific
6172 concepts, such as insserv, early-boot SysV services
6173 (i.e. those for non-standard runlevels such as 'b' or 'S')
6174 or ArchLinux /etc/rc.conf support. We recommend the
6175 distributions who still need support this to either continue
6176 to maintain the necessary patches downstream, or find a
6177 different solution. (Talk to us if you have questions!)
6179 * Various systemd components will now bypass PolicyKit checks
6180 for root and otherwise handle properly if PolicyKit is not
6181 found to be around. This should fix most issues for
6182 PolicyKit-less systems. Quite frankly this should have been
6183 this way since day one. It is absolutely our intention to
6184 make systemd work fine on PolicyKit-less systems, and we
6185 consider it a bug if something does not work as it should if
6186 PolicyKit is not around.
6188 * For embedded systems it is now possible to build udev and
6189 systemd without blkid and/or kmod support.
6191 * "systemctl switch-root" is now capable of switching root
6192 more than once. I.e. in addition to transitions from the
6193 initrd to the host OS it is now possible to transition to
6194 further OS images from the host. This is useful to implement
6195 offline updating tools.
6197 * Various other additions have been made to the RPM macros
6198 shipped with systemd. Use %udev_rules_update() after
6199 installing new udev rules files. %_udevhwdbdir,
6200 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6201 %_sysctldir are now available which resolve to the right
6202 directories for packages to place various data files in.
6204 * journalctl gained the new --full switch (in addition to
6205 --all, to disable ellipsation for long messages.
6207 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6208 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6209 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6210 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6211 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6212 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6213 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6214 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6215 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6219 * journalctl gained new --since= and --until= switches to
6220 filter by time. It also now supports nice filtering for
6221 units via --unit=/-u.
6223 * Type=oneshot services may use ExecReload= and do the
6226 * The journal daemon now supports time-based rotation and
6227 vacuuming, in addition to the usual disk-space based
6230 * The journal will now index the available field values for
6231 each field name. This enables clients to show pretty drop
6232 downs of available match values when filtering. The bash
6233 completion of journalctl has been updated
6234 accordingly. journalctl gained a new switch -F to list all
6235 values a certain field takes in the journal database.
6237 * More service events are now written as structured messages
6238 to the journal, and made recognizable via message IDs.
6240 * The timedated, localed and hostnamed mini-services which
6241 previously only provided support for changing time, locale
6242 and hostname settings from graphical DEs such as GNOME now
6243 also have a minimal (but very useful) text-based client
6244 utility each. This is probably the nicest way to changing
6245 these settings from the command line now, especially since
6246 it lists available options and is fully integrated with bash
6249 * There's now a new tool "systemd-coredumpctl" to list and
6250 extract coredumps from the journal.
6252 * We now install a README each in /var/log/ and
6253 /etc/rc.d/init.d explaining where the system logs and init
6254 scripts went. This hopefully should help folks who go to
6255 that dirs and look into the otherwise now empty void and
6256 scratch their heads.
6258 * When user-services are invoked (by systemd --user) the
6259 $MANAGERPID env var is set to the PID of systemd.
6261 * SIGRTMIN+24 when sent to a --user instance will now result
6262 in immediate termination of systemd.
6264 * gatewayd received numerous feature additions such as a
6265 "follow" mode, for live syncing and filtering.
6267 * browse.html now allows filtering and showing detailed
6268 information on specific entries. Keyboard navigation and
6269 mouse screen support has been added.
6271 * gatewayd/journalctl now supports HTML5/JSON
6272 Server-Sent-Events as output.
6274 * The SysV init script compatibility logic will now
6275 heuristically determine whether a script supports the
6276 "reload" verb, and only then make this available as
6279 * "systemctl status --follow" has been removed, use "journalctl
6282 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6283 have been removed since they are hardly useful to be
6286 * And I'd like to take the opportunity to specifically mention
6287 Zbigniew for his great contributions. Zbigniew, you rock!
6289 Contributions from: Andrew Eikum, Christian Hesse, Colin
6290 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6291 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6292 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6293 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6294 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6295 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6299 * If /etc/vconsole.conf is non-existent or empty we will no
6300 longer load any console font or key map at boot by
6301 default. Instead the kernel defaults will be left
6302 intact. This is definitely the right thing to do, as no
6303 configuration should mean no configuration, and hard-coding
6304 font names that are different on all archs is probably a bad
6305 idea. Also, the kernel default key map and font should be
6306 good enough for most cases anyway, and mostly identical to
6307 the userspace fonts/key maps we previously overloaded them
6308 with. If distributions want to continue to default to a
6309 non-kernel font or key map they should ship a default
6310 /etc/vconsole.conf with the appropriate contents.
6312 Contributions from: Colin Walters, Daniel J Walsh, Dave
6313 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6314 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6318 * journalctl gained a new --cursor= switch to show entries
6319 starting from the specified location in the journal.
6321 * We now enforce a size limit on journal entry fields exported
6322 with "-o json" in journalctl. Fields larger than 4K will be
6323 assigned null. This can be turned off with --all.
6325 * An (optional) journal gateway daemon is now available as
6326 "systemd-journal-gatewayd.service". This service provides
6327 access to the journal via HTTP and JSON. This functionality
6328 will be used to implement live log synchronization in both
6329 pull and push modes, but has various other users too, such
6330 as easy log access for debugging of embedded devices. Right
6331 now it is already useful to retrieve the journal via HTTP:
6333 # systemctl start systemd-journal-gatewayd.service
6334 # wget http://localhost:19531/entries
6336 This will download the journal contents in a
6337 /var/log/messages compatible format. The same as JSON:
6339 # curl -H"Accept: application/json" http://localhost:19531/entries
6341 This service is also accessible via a web browser where a
6342 single static HTML5 app is served that uses the JSON logic
6343 to enable the user to do some basic browsing of the
6344 journal. This will be extended later on. Here's an example
6345 screenshot of this app in its current state:
6347 http://0pointer.de/public/journal-gatewayd
6349 Contributions from: Kay Sievers, Lennart Poettering, Robert
6350 Milasan, Tom Gundersen
6354 * The bash completion logic is now available for journalctl
6357 * We do not mount the "cpuset" controller anymore together with
6358 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6359 started if no parameters are assigned to it. "cpuset" hence
6360 broke code that assumed it could create "cpu" groups and
6363 * journalctl -f will now subscribe to terminal size changes,
6364 and line break accordingly.
6366 Contributions from: Dave Reisner, Kay Sievers, Lennart
6367 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6371 * nspawn will now create a symlink /etc/localtime in the
6372 container environment, copying the host's timezone
6373 setting. Previously this has been done via a bind mount, but
6374 since symlinks cannot be bind mounted this has now been
6375 changed to create/update the appropriate symlink.
6377 * journalctl -n's line number argument is now optional, and
6378 will default to 10 if omitted.
6380 * journald will now log the maximum size the journal files may
6381 take up on disk. This is particularly useful if the default
6382 built-in logic of determining this parameter from the file
6383 system size is used. Use "systemctl status
6384 systemd-journald.service" to see this information.
6386 * The multi-seat X wrapper tool has been stripped down. As X
6387 is now capable of enumerating graphics devices via udev in a
6388 seat-aware way the wrapper is not strictly necessary
6389 anymore. A stripped down temporary stop-gap is still shipped
6390 until the upstream display managers have been updated to
6391 fully support the new X logic. Expect this wrapper to be
6392 removed entirely in one of the next releases.
6394 * HandleSleepKey= in logind.conf has been split up into
6395 HandleSuspendKey= and HandleHibernateKey=. The old setting
6396 is not available anymore. X11 and the kernel are
6397 distinguishing between these keys and we should too. This
6398 also means the inhibition lock for these keys has been split
6401 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6402 Poettering, Lukas Nykryn, Václav Pavlín
6406 * Whenever a unit changes state we will now log this to the
6407 journal and show along the unit's own log output in
6410 * ConditionPathIsMountPoint= can now properly detect bind
6411 mount points too. (Previously, a bind mount of one file
6412 system to another place in the same file system could not be
6413 detected as mount, since they shared struct stat's st_dev
6416 * We will now mount the cgroup controllers cpu, cpuacct,
6417 cpuset and the controllers net_cls, net_prio together by
6420 * nspawn containers will now have a virtualized boot
6421 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6422 over with a randomized ID at container initialization). This
6423 has the effect of making "journalctl -b" do the right thing
6426 * The JSON output journal serialization has been updated not
6427 to generate "endless" list objects anymore, but rather one
6428 JSON object per line. This is more in line how most JSON
6429 parsers expect JSON objects. The new output mode
6430 "json-pretty" has been added to provide similar output, but
6431 neatly aligned for readability by humans.
6433 * We dropped all explicit sync() invocations in the shutdown
6434 code. The kernel does this implicitly anyway in the kernel
6435 reboot() syscall. halt(8)'s -n option is now a compatibility
6438 * We now support virtualized reboot() in containers, as
6439 supported by newer kernels. We will fall back to exit() if
6440 CAP_SYS_REBOOT is not available to the container. Also,
6441 nspawn makes use of this now and will actually reboot the
6442 container if the containerized OS asks for that.
6444 * journalctl will only show local log output by default
6445 now. Use --merge (-m) to show remote log output, too.
6447 * libsystemd-journal gained the new sd_journal_get_usage()
6448 call to determine the current disk usage of all journal
6449 files. This is exposed in the new "journalctl --disk-usage"
6452 * journald gained a new configuration setting SplitMode= in
6453 journald.conf which may be used to control how user journals
6454 are split off. See journald.conf(5) for details.
6456 * A new condition type ConditionFileNotEmpty= has been added.
6458 * tmpfiles' "w" lines now support file globbing, to write
6459 multiple files at once.
6461 * We added Python bindings for the journal submission
6462 APIs. More Python APIs for a number of selected APIs will
6463 likely follow. Note that we intend to add native bindings
6464 only for the Python language, as we consider it common
6465 enough to deserve bindings shipped within systemd. There are
6466 various projects outside of systemd that provide bindings
6467 for languages such as PHP or Lua.
6469 * Many conditions will now resolve specifiers such as %i. In
6470 addition, PathChanged= and related directives of .path units
6471 now support specifiers as well.
6473 * There's now a new RPM macro definition for the system preset
6476 * journald will now warn if it ca not forward a message to the
6477 syslog daemon because its socket is full.
6479 * timedated will no longer write or process /etc/timezone,
6480 except on Debian. As we do not support late mounted /usr
6481 anymore /etc/localtime always being a symlink is now safe,
6482 and hence the information in /etc/timezone is not necessary
6485 * logind will now always reserve one VT for a text getty (VT6
6486 by default). Previously if more than 6 X sessions where
6487 started they took up all the VTs with auto-spawned gettys,
6488 so that no text gettys were available anymore.
6490 * udev will now automatically inform the btrfs kernel logic
6491 about btrfs RAID components showing up. This should make
6492 simple hotplug based btrfs RAID assembly work.
6494 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6495 (but not for its children which will stay at the kernel
6496 default). This should allow setups with a lot more listening
6499 * systemd will now always pass the configured timezone to the
6500 kernel at boot. timedated will do the same when the timezone
6503 * logind's inhibition logic has been updated. By default,
6504 logind will now handle the lid switch, the power and sleep
6505 keys all the time, even in graphical sessions. If DEs want
6506 to handle these events on their own they should take the new
6507 handle-power-key, handle-sleep-key and handle-lid-switch
6508 inhibitors during their runtime. A simple way to achieve
6509 that is to invoke the DE wrapped in an invocation of:
6511 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6513 * Access to unit operations is now checked via SELinux taking
6514 the unit file label and client process label into account.
6516 * systemd will now notify the administrator in the journal
6517 when he over-mounts a non-empty directory.
6519 * There are new specifiers that are resolved in unit files,
6520 for the host name (%H), the machine ID (%m) and the boot ID
6523 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6524 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6525 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6526 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6527 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6528 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6529 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6533 * Support for reading structured kernel messages from
6534 /dev/kmsg has now been added and is enabled by default.
6536 * Support for reading kernel messages from /proc/kmsg has now
6537 been removed. If you want kernel messages in the journal
6538 make sure to run a recent kernel (>= 3.5) that supports
6539 reading structured messages from /dev/kmsg (see
6540 above). /proc/kmsg is now exclusive property of classic
6541 syslog daemons again.
6543 * The libudev API gained the new
6544 udev_device_new_from_device_id() call.
6546 * The logic for file system namespace (ReadOnlyDirectory=,
6547 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6548 require pivot_root() anymore. This means fewer temporary
6549 directories are created below /tmp for this feature.
6551 * nspawn containers will now see and receive all submounts
6552 made on the host OS below the root file system of the
6555 * Forward Secure Sealing is now supported for Journal files,
6556 which provide cryptographical sealing of journal files so
6557 that attackers cannot alter log history anymore without this
6558 being detectable. Lennart will soon post a blog story about
6559 this explaining it in more detail.
6561 * There are two new service settings RestartPreventExitStatus=
6562 and SuccessExitStatus= which allow configuration of exit
6563 status (exit code or signal) which will be excepted from the
6564 restart logic, resp. consider successful.
6566 * journalctl gained the new --verify switch that can be used
6567 to check the integrity of the structure of journal files and
6568 (if Forward Secure Sealing is enabled) the contents of
6571 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6572 and similar symlinks pre-created. This makes running shells
6573 as container init process a lot more fun.
6575 * The fstab support can now handle PARTUUID= and PARTLABEL=
6578 * A new ConditionHost= condition has been added to match
6579 against the hostname (with globs) and machine ID. This is
6580 useful for clusters where a single OS image is used to
6581 provision a large number of hosts which shall run slightly
6582 different sets of services.
6584 * Services which hit the restart limit will now be placed in a
6587 Contributions from: Bertram Poettering, Dave Reisner, Huang
6588 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6589 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6593 * When running in --user mode systemd will now become a
6594 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6595 tree a lot more organized.
6597 * A new PartOf= unit dependency type has been introduced that
6598 may be used to group services in a natural way.
6600 * "systemctl enable" may now be used to enable instances of
6603 * journalctl now prints error log levels in red, and
6604 warning/notice log levels in bright white. It also supports
6605 filtering by log level now.
6607 * cgtop gained a new -n switch (similar to top), to configure
6608 the maximum number of iterations to run for. It also gained
6609 -b, to run in batch mode (accepting no input).
6611 * The suffix ".service" may now be omitted on most systemctl
6612 command lines involving service unit names.
6614 * There's a new bus call in logind to lock all sessions, as
6615 well as a loginctl verb for it "lock-sessions".
6617 * libsystemd-logind.so gained a new call sd_journal_perror()
6618 that works similar to libc perror() but logs to the journal
6619 and encodes structured information about the error number.
6621 * /etc/crypttab entries now understand the new keyfile-size=
6624 * shutdown(8) now can send a (configurable) wall message when
6625 a shutdown is cancelled.
6627 * The mount propagation mode for the root file system will now
6628 default to "shared", which is useful to make containers work
6629 nicely out-of-the-box so that they receive new mounts from
6630 the host. This can be undone locally by running "mount
6631 --make-rprivate /" if needed.
6633 * The prefdm.service file has been removed. Distributions
6634 should maintain this unit downstream if they intend to keep
6635 it around. However, we recommend writing normal unit files
6636 for display managers instead.
6638 * Since systemd is a crucial part of the OS we will now
6639 default to a number of compiler switches that improve
6640 security (hardening) such as read-only relocations, stack
6641 protection, and suchlike.
6643 * The TimeoutSec= setting for services is now split into
6644 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6645 of individual time outs for the start and the stop phase of
6648 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6649 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6650 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6651 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6652 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6653 Gundersen, Zbigniew Jędrzejewski-Szmek
6657 * The journal and id128 C APIs are now fully documented as man
6660 * Extra safety checks have been added when transitioning from
6661 the initial RAM disk to the main system to avoid accidental
6664 * /etc/crypttab entries now understand the new keyfile-offset=
6667 * systemctl -t can now be used to filter by unit load state.
6669 * The journal C API gained the new sd_journal_wait() call to
6670 make writing synchronous journal clients easier.
6672 * journalctl gained the new -D switch to show journals from a
6675 * journalctl now displays a special marker between log
6676 messages of two different boots.
6678 * The journal is now explicitly flushed to /var via a service
6679 systemd-journal-flush.service, rather than implicitly simply
6680 by seeing /var/log/journal to be writable.
6682 * journalctl (and the journal C APIs) can now match for much
6683 more complex expressions, with alternatives and
6686 * When transitioning from the initial RAM disk to the main
6687 system we will now kill all processes in a killing spree to
6688 ensure no processes stay around by accident.
6690 * Three new specifiers may be used in unit files: %u, %h, %s
6691 resolve to the user name, user home directory resp. user
6692 shell. This is useful for running systemd user instances.
6694 * We now automatically rotate journal files if their data
6695 object hash table gets a fill level > 75%. We also size the
6696 hash table based on the configured maximum file size. This
6697 together should lower hash collisions drastically and thus
6698 speed things up a bit.
6700 * journalctl gained the new "--header" switch to introspect
6701 header data of journal files.
6703 * A new setting SystemCallFilters= has been added to services
6704 which may be used to apply blacklists or whitelists to
6705 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6707 * nspawn gained a new --link-journal= switch (and quicker: -j)
6708 to link the container journal with the host. This makes it
6709 very easy to centralize log viewing on the host for all
6710 guests while still keeping the journal files separated.
6712 * Many bugfixes and optimizations
6714 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6715 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6716 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6721 * Several tools now understand kernel command line arguments,
6722 which are only read when run in an initial RAM disk. They
6723 usually follow closely their normal counterparts, but are
6726 * There's a new tool to analyze the readahead files that are
6727 automatically generated at boot. Use:
6729 /usr/lib/systemd/systemd-readahead analyze /.readahead
6731 * We now provide an early debug shell on tty9 if this enabled. Use:
6733 systemctl enable debug-shell.service
6735 * All plymouth related units have been moved into the Plymouth
6736 package. Please make sure to upgrade your Plymouth version
6739 * systemd-tmpfiles now supports getting passed the basename of
6740 a configuration file only, in which case it will look for it
6741 in all appropriate directories automatically.
6743 * udevadm info now takes a /dev or /sys path as argument, and
6744 does the right thing. Example:
6746 udevadm info /dev/sda
6747 udevadm info /sys/class/block/sda
6749 * systemctl now prints a warning if a unit is stopped but a
6750 unit that might trigger it continues to run. Example: a
6751 service is stopped but the socket that activates it is left
6754 * "systemctl status" will now mention if the log output was
6755 shortened due to rotation since a service has been started.
6757 * The journal API now exposes functions to determine the
6758 "cutoff" times due to rotation.
6760 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6761 immediately flushing of runtime logs to /var if possible,
6762 resp. for triggering immediate rotation of the journal
6765 * It is now considered an error if a service is attempted to
6766 be stopped that is not loaded.
6768 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6770 * systemd-analyze now supports Python 3
6772 * tmpfiles now supports cleaning up directories via aging
6773 where the first level dirs are always kept around but
6774 directories beneath it automatically aged. This is enabled
6775 by prefixing the age field with '~'.
6777 * Seat objects now expose CanGraphical, CanTTY properties
6778 which is required to deal with very fast bootups where the
6779 display manager might be running before the graphics drivers
6780 completed initialization.
6782 * Seat objects now expose a State property.
6784 * We now include RPM macros for service enabling/disabling
6785 based on the preset logic. We recommend RPM based
6786 distributions to make use of these macros if possible. This
6787 makes it simpler to reuse RPM spec files across
6790 * We now make sure that the collected systemd unit name is
6791 always valid when services log to the journal via
6794 * There's a new man page kernel-command-line(7) detailing all
6795 command line options we understand.
6797 * The fstab generator may now be disabled at boot by passing
6798 fstab=0 on the kernel command line.
6800 * A new kernel command line option modules-load= is now understood
6801 to load a specific kernel module statically, early at boot.
6803 * Unit names specified on the systemctl command line are now
6804 automatically escaped as needed. Also, if file system or
6805 device paths are specified they are automatically turned
6806 into the appropriate mount or device unit names. Example:
6808 systemctl status /home
6809 systemctl status /dev/sda
6811 * The SysVConsole= configuration option has been removed from
6812 system.conf parsing.
6814 * The SysV search path is no longer exported on the D-Bus
6817 * The Names= option has been removed from unit file parsing.
6819 * There's a new man page bootup(7) detailing the boot process.
6821 * Every unit and every generator we ship with systemd now
6822 comes with full documentation. The self-explanatory boot is
6825 * A couple of services gained "systemd-" prefixes in their
6826 name if they wrap systemd code, rather than only external
6827 code. Among them fsck@.service which is now
6828 systemd-fsck@.service.
6830 * The HaveWatchdog property has been removed from the D-Bus
6833 * systemd.confirm_spawn= on the kernel command line should now
6836 * There's a new man page crypttab(5) which details all options
6837 we actually understand.
6839 * systemd-nspawn gained a new --capability= switch to pass
6840 additional capabilities to the container.
6842 * timedated will now read known NTP implementation unit names
6843 from /usr/lib/systemd/ntp-units.d/*.list,
6844 systemd-timedated-ntp.target has been removed.
6846 * journalctl gained a new switch "-b" that lists log data of
6847 the current boot only.
6849 * The notify socket is in the abstract namespace again, in
6850 order to support daemons which chroot() at start-up.
6852 * There is a new Storage= configuration option for journald
6853 which allows configuration of where log data should go. This
6854 also provides a way to disable journal logging entirely, so
6855 that data collected is only forwarded to the console, the
6856 kernel log buffer or another syslog implementation.
6858 * Many bugfixes and optimizations
6860 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6861 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6862 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6863 Shawn Landden, Tom Gundersen
6867 * "systemctl help <unit>" now shows the man page if one is
6870 * Several new man pages have been added.
6872 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6873 MaxLevelConsole= can now be specified in
6874 journald.conf. These options allow reducing the amount of
6875 data stored on disk or forwarded by the log level.
6877 * TimerSlackNSec= can now be specified in system.conf for
6878 PID1. This allows system-wide power savings.
6880 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6881 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6886 * logind is now capable of (optionally) handling power and
6887 sleep keys as well as the lid switch.
6889 * journalctl now understands the syntax "journalctl
6890 /usr/bin/avahi-daemon" to get all log output of a specific
6893 * CapabilityBoundingSet= in system.conf now also influences
6894 the capability bound set of usermode helpers of the kernel.
6896 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6897 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6898 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6899 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6903 * Note that we skipped 139 releases here in order to set the
6904 new version to something that is greater than both udev's
6905 and systemd's most recent version number.
6907 * udev: all udev sources are merged into the systemd source tree now.
6908 All future udev development will happen in the systemd tree. It
6909 is still fully supported to use the udev daemon and tools without
6910 systemd running, like in initramfs or other init systems. Building
6911 udev though, will require the *build* of the systemd tree, but
6912 udev can be properly *run* without systemd.
6914 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6915 should be used to create dead device nodes as workarounds for broken
6918 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6919 no longer supported. udev_monitor_new_from_netlink() needs to be
6920 used to subscribe to events.
6922 * udev: when udevd is started by systemd, processes which are left
6923 behind by forking them off of udev rules, are unconditionally cleaned
6924 up and killed now after the event handling has finished. Services or
6925 daemons must be started as systemd services. Services can be
6926 pulled-in by udev to get started, but they can no longer be directly
6927 forked by udev rules.
6929 * udev: the daemon binary is called systemd-udevd now and installed
6930 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6931 to adapt to that, create symlink, or rename the binary after building
6934 * libudev no longer provides these symbols:
6935 udev_monitor_from_socket()
6936 udev_queue_get_failed_list_entry()
6937 udev_get_{dev,sys,run}_path()
6938 The versions number was bumped and symbol versioning introduced.
6940 * systemd-loginctl and systemd-journalctl have been renamed
6941 to loginctl and journalctl to match systemctl.
6943 * The config files: /etc/systemd/systemd-logind.conf and
6944 /etc/systemd/systemd-journald.conf have been renamed to
6945 logind.conf and journald.conf. Package updates should rename
6946 the files to the new names on upgrade.
6948 * For almost all files the license is now LGPL2.1+, changed
6949 from the previous GPL2.0+. Exceptions are some minor stuff
6950 of udev (which will be changed to LGPL2.1 eventually, too),
6951 and the MIT licensed sd-daemon.[ch] library that is suitable
6952 to be used as drop-in files.
6954 * systemd and logind now handle system sleep states, in
6955 particular suspending and hibernating.
6957 * logind now implements a sleep/shutdown/idle inhibiting logic
6958 suitable for a variety of uses. Soonishly Lennart will blog
6959 about this in more detail.
6961 * var-run.mount and var-lock.mount are no longer provided
6962 (which previously bind mounted these directories to their new
6963 places). Distributions which have not converted these
6964 directories to symlinks should consider stealing these files
6965 from git history and add them downstream.
6967 * We introduced the Documentation= field for units and added
6968 this to all our shipped units. This is useful to make it
6969 easier to explore the boot and the purpose of the various
6972 * All smaller setup units (such as
6973 systemd-vconsole-setup.service) now detect properly if they
6974 are run in a container and are skipped when
6975 appropriate. This guarantees an entirely noise-free boot in
6976 Linux container environments such as systemd-nspawn.
6978 * A framework for implementing offline system updates is now
6979 integrated, for details see:
6980 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
6982 * A new service type Type=idle is available now which helps us
6983 avoiding ugly interleaving of getty output and boot status
6986 * There's now a system-wide CapabilityBoundingSet= option to
6987 globally reduce the set of capabilities for the
6988 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6989 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6990 even CAP_NET_ADMIN system-wide for secure systems.
6992 * There are now system-wide DefaultLimitXXX= options to
6993 globally change the defaults of the various resource limits
6994 for all units started by PID 1.
6996 * Harald Hoyer's systemd test suite has been integrated into
6997 systemd which allows easy testing of systemd builds in qemu
6998 and nspawn. (This is really awesome! Ask us for details!)
7000 * The fstab parser is now implemented as generator, not inside
7003 * systemctl will now warn you if .mount units generated from
7004 /etc/fstab are out of date due to changes in fstab that
7005 have not been read by systemd yet.
7007 * systemd is now suitable for usage in initrds. Dracut has
7008 already been updated to make use of this. With this in place
7009 initrds get a slight bit faster but primarily are much
7010 easier to introspect and debug since "systemctl status" in
7011 the host system can be used to introspect initrd services,
7012 and the journal from the initrd is kept around too.
7014 * systemd-delta has been added, a tool to explore differences
7015 between user/admin configuration and vendor defaults.
7017 * PrivateTmp= now affects both /tmp and /var/tmp.
7019 * Boot time status messages are now much prettier and feature
7020 proper english language. Booting up systemd has never been
7023 * Read-ahead pack files now include the inode number of all
7024 files to pre-cache. When the inode changes the pre-caching
7025 is not attempted. This should be nicer to deal with updated
7026 packages which might result in changes of read-ahead
7029 * We now temporaritly lower the kernel's read_ahead_kb variable
7030 when collecting read-ahead data to ensure the kernel's
7031 built-in read-ahead does not add noise to our measurements
7032 of necessary blocks to pre-cache.
7034 * There's now RequiresMountsFor= to add automatic dependencies
7035 for all mounts necessary for a specific file system path.
7037 * MountAuto= and SwapAuto= have been removed from
7038 system.conf. Mounting file systems at boot has to take place
7041 * nspawn now learned a new switch --uuid= to set the machine
7042 ID on the command line.
7044 * nspawn now learned the -b switch to automatically search
7047 * vt102 is now the default TERM for serial TTYs, upgraded from
7050 * systemd-logind now works on VT-less systems.
7052 * The build tree has been reorganized. The individual
7053 components now have directories of their own.
7055 * A new condition type ConditionPathIsReadWrite= is now available.
7057 * nspawn learned the new -C switch to create cgroups for the
7058 container in other hierarchies.
7060 * We now have support for hardware watchdogs, configurable in
7063 * The scheduled shutdown logic now has a public API.
7065 * We now mount /tmp as tmpfs by default, but this can be
7066 masked and /etc/fstab can override it.
7068 * Since udisks does not make use of /media anymore we are not
7069 mounting a tmpfs on it anymore.
7071 * journalctl gained a new --local switch to only interleave
7072 locally generated journal files.
7074 * We can now load the IMA policy at boot automatically.
7076 * The GTK tools have been split off into a systemd-ui.
7078 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7079 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7080 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7081 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7082 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7083 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7084 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7085 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7086 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7091 * This is mostly a bugfix release
7093 * Support optional initialization of the machine ID from the
7094 KVM or container configured UUID.
7096 * Support immediate reboots with "systemctl reboot -ff"
7098 * Show /etc/os-release data in systemd-analyze output
7100 * Many bugfixes for the journal, including endianness fixes and
7101 ensuring that disk space enforcement works
7103 * sd-login.h is C++ compatible again
7105 * Extend the /etc/os-release format on request of the Debian
7108 * We now refuse non-UTF8 strings used in various configuration
7109 and unit files. This is done to ensure we do not pass invalid
7110 data over D-Bus or expose it elsewhere.
7112 * Register Mimo USB Screens as suitable for automatic seat
7115 * Read SELinux client context from journal clients in a race
7118 * Reorder configuration file lookup order. /etc now always
7119 overrides /run in order to allow the administrator to always
7120 and unconditionally override vendor-supplied or
7121 automatically generated data.
7123 * The various user visible bits of the journal now have man
7124 pages. We still lack man pages for the journal API calls
7127 * We now ship all man pages in HTML format again in the
7130 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7131 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7132 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7137 * This is mostly a bugfix release
7139 * systems lacking /etc/os-release are no longer supported.
7141 * Various functionality updates to libsystemd-login.so
7143 * Track class of PAM logins to distinguish greeters from
7146 Contributions from: Kay Sievers, Lennart Poettering, Michael
7151 * This is an important bugfix release for v41.
7153 * Building man pages is now optional which should be useful
7154 for those building systemd from git but unwilling to install
7157 * Watchdog support for supervising services is now usable. In
7158 a future release support for hardware watchdogs
7159 (i.e. /dev/watchdog) will be added building on this.
7161 * Service start rate limiting is now configurable and can be
7162 turned off per service. When a start rate limit is hit a
7163 reboot can automatically be triggered.
7165 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7167 Contributions from: Benjamin Franzke, Bill Nottingham,
7168 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7169 Schmidt, Michał Górny, Piotr Drąg
7173 * The systemd binary is installed /usr/lib/systemd/systemd now;
7174 An existing /sbin/init symlink needs to be adapted with the
7177 * The code that loads kernel modules has been ported to invoke
7178 libkmod directly, instead of modprobe. This means we do not
7179 support systems with module-init-tools anymore.
7181 * Watchdog support is now already useful, but still not
7184 * A new kernel command line option systemd.setenv= is
7185 understood to set system wide environment variables
7186 dynamically at boot.
7188 * We now limit the set of capabilities of systemd-journald.
7190 * We now set SIGPIPE to ignore by default, since it only is
7191 useful in shell pipelines, and has little use in general
7192 code. This can be disabled with IgnoreSIPIPE=no in unit
7195 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7196 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7201 * This is mostly a bugfix release
7203 * We now expose the reason why a service failed in the
7204 "Result" D-Bus property.
7206 * Rudimentary service watchdog support (will be completed over
7207 the next few releases.)
7209 * When systemd forks off in order execute some service we will
7210 now immediately changes its argv[0] to reflect which process
7211 it will execute. This is useful to minimize the time window
7212 with a generic argv[0], which makes bootcharts more useful
7214 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7215 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7216 Mike Kazantsev, Ray Strode
7220 * This is mostly a test release, but incorporates many
7223 * New systemd-cgtop tool to show control groups by their
7226 * Linking against libacl for ACLs is optional again. If
7227 disabled, support tracking device access for active logins
7228 goes becomes unavailable, and so does access to the user
7229 journals by the respective users.
7231 * If a group "adm" exists, journal files are automatically
7232 owned by them, thus allow members of this group full access
7233 to the system journal as well as all user journals.
7235 * The journal now stores the SELinux context of the logging
7236 client for all entries.
7238 * Add C++ inclusion guards to all public headers
7240 * New output mode "cat" in the journal to print only text
7241 messages, without any meta data like date or time.
7243 * Include tiny X server wrapper as a temporary stop-gap to
7244 teach XOrg udev display enumeration. This is used by display
7245 managers such as gdm, and will go away as soon as XOrg
7246 learned native udev hotplugging for display devices.
7248 * Add new systemd-cat tool for executing arbitrary programs
7249 with STDERR/STDOUT connected to the journal. Can also act as
7250 BSD logger replacement, and does so by default.
7252 * Optionally store all locally generated coredumps in the
7253 journal along with meta data.
7255 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7256 writing short strings to files (for usage for /sys), and for
7257 creating symlinks, character and block device nodes.
7259 * New unit file option ControlGroupPersistent= to make cgroups
7260 persistent, following the mechanisms outlined in
7261 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7263 * Support multiple local RTCs in a sane way
7265 * No longer monopolize IO when replaying readahead data on
7266 rotating disks, since we might starve non-file-system IO to
7267 death, since fanotify() will not see accesses done by blkid,
7270 * Do not show kernel threads in systemd-cgls anymore, unless
7271 requested with new -k switch.
7273 Contributions from: Dan Horák, Kay Sievers, Lennart
7274 Poettering, Michal Schmidt
7278 * This is mostly a test release, but incorporates many
7281 * The git repository moved to:
7282 git://anongit.freedesktop.org/systemd/systemd
7283 ssh://git.freedesktop.org/git/systemd/systemd
7285 * First release with the journal
7286 http://0pointer.de/blog/projects/the-journal.html
7288 * The journal replaces both systemd-kmsg-syslogd and
7289 systemd-stdout-bridge.
7291 * New sd_pid_get_unit() API call in libsystemd-logind
7293 * Many systemadm clean-ups
7295 * Introduce remote-fs-pre.target which is ordered before all
7296 remote mounts and may be used to start services before all
7299 * Added Mageia support
7301 * Add bash completion for systemd-loginctl
7303 * Actively monitor PID file creation for daemons which exit in
7304 the parent process before having finished writing the PID
7305 file in the daemon process. Daemons which do this need to be
7306 fixed (i.e. PID file creation must have finished before the
7307 parent exits), but we now react a bit more gracefully to them.
7309 * Add colourful boot output, mimicking the well-known output
7310 of existing distributions.
7312 * New option PassCredentials= for socket units, for
7313 compatibility with a recent kernel ABI breakage.
7315 * /etc/rc.local is now hooked in via a generator binary, and
7316 thus will no longer act as synchronization point during
7319 * systemctl list-unit-files now supports --root=.
7321 * systemd-tmpfiles now understands two new commands: z, Z for
7322 relabelling files according to the SELinux database. This is
7323 useful to apply SELinux labels to specific files in /sys,
7326 * Output of SysV services is now forwarded to both the console
7327 and the journal by default, not only just the console.
7329 * New man pages for all APIs from libsystemd-login.
7331 * The build tree got reorganized and the build system is a
7332 lot more modular allowing embedded setups to specifically
7333 select the components of systemd they are interested in.
7335 * Support for Linux systems lacking the kernel VT subsystem is
7338 * configure's --with-rootdir= got renamed to
7339 --with-rootprefix= to follow the naming used by udev and
7342 * Unless specified otherwise we will now install to /usr instead
7343 of /usr/local by default.
7345 * Processes with '@' in argv[0][0] are now excluded from the
7346 final shut-down killing spree, following the logic explained
7348 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7350 * All processes remaining in a service cgroup when we enter
7351 the START or START_PRE states are now killed with
7352 SIGKILL. That means it is no longer possible to spawn
7353 background processes from ExecStart= lines (which was never
7354 supported anyway, and bad style).
7356 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7357 reloading of units together.
7359 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7360 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7361 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7362 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7363 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek